eBooks

Corpus · Culture · Discourse

0723
2008
978-3-8233-7426-8
978-3-8233-6426-9
Gunter Narr Verlag 
Tamsin Sanderson

Corpus, Culture and Discourse is a groundbreaking new study of academic discourse across cultures, languages and disciplines that will be of interest to researchers and practitioners in language for specific purposes, corpus linguistics and intercultural communication/pragmatics. Based on a large-scale parallel corpus, the study examines social interaction, identity construction and metadiscourse in English- and Germanlanguage research articles across five disciplines. The innovative combination of quantitative corpus and statistical analyses with detailed qualitative analysis delivers some surprising and interesting results. Corpus, Culture and Discourse examines the ways in which people from different cultures, writing in different languages and disciplines, use language in an academic context. The study rests upon an approximately 1 million word parallel corpus of research articles from five humanities disciplines, written by scholars from Great Britain, the United States and Germany.The innovative methodology combines quantitative corpus and statistical analyses with detailed qualitative analysis of the texts. The study demonstrates that the synthesis of discourse analysis with corpus linguistics two traditionally-opposed approaches to the study of language is both practicable and rewarding, and can provide a powerful tool for investigating linguistic behaviour. The study analyses academic discourse with respect to its social, cultural, institutional and disciplinary context. It sheds new light on the myriad ways in which academics use their native language in research articles to construct and express social identity, interact with their readers and the wider academic community, and manage textual interaction through metadiscourse. The results show that there are empirically verifiable, statistically significant differences across cultures, languages and disciplines in the management of social interaction in academic writing. Interestingly, it emerges that author age, status and gender also exercise a significant influence on specific aspects of language use.

<?page no="0"?> Gunter Narr Verlag Tübingen Tamsin Sanderson Corpus · Culture · Discourse Language in Performance LiP <?page no="1"?> Corpus · Culture · Discourse <?page no="2"?> 39 Edited by Werner Hüllen† and Rainer Schulze Advisory Board: Thomas Herbst (Erlangen), Andreas Jucker (Zürich), Manfred Krug (Bamberg), Christian Mair (Freiburg i. Br.), Ute Römer (Hannover), Andrea Sand (Trier), Hans-Jörg Schmid (München), Josef Schmied (Chemnitz) and Edgar W. Schneider (Regensburg) <?page no="3"?> Tamsin Sanderson Corpus · Culture · Discourse Gunter Narr Verlag Tübingen <?page no="4"?> Bibliografische Information der Deutschen Nationalbibliothek Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über <http: / / dnb.d-nb.de> abrufbar. © 2008 · Narr Francke Attempto Verlag GmbH + Co. KG Dischingerweg 5 · D-72070 Tübingen Das Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt. Jede Verwertung außerhalb der engen Grenzen des Urheberrechtsgesetzes ist ohne Zustimmung des Verlages unzulässig und strafbar. Das gilt insbesondere für Vervielfältigungen, Übersetzungen, Mikroverfilmungen und die Einspeicherung und Verarbeitung in elektronischen Systemen. Gedruckt auf säurefreiem und alterungsbeständigem Werkdruckpapier. Internet: www.narr.de E-Mail: info@narr.de Druck und Bindung: Ilmprint, Langewiesen Printed in Germany ISSN 0939-9399 ISBN 978-3-8233-6426-9 D 25 <?page no="5"?> meiner Oma <?page no="7"?> 7 Table of contents Acknowledgements ........................................................................... 9 Abbreviations ...................................................................................... 10 Figures and tables .............................................................................. 11 1. Introduction .................................................................................. 15 1.1 What this study is about............................................................... 15 1.2 Focus of the study........................................................................ 17 1.3 Science speaks English ................................................................ 18 1.4 The situation for non-native speaker researchers ........................ 23 1.5 The sting in the tail ...................................................................... 28 1.6 A question of culture or a question of language? ........................ 31 2. Researching cultural trends in academic discourse ..... 37 2.1 Research into academic discourse produced by German and English-speakers ................................................ 37 2.2 A discipline with dozens of methodologies ................................. 48 2.3 Combining corpus and discourse................................................. 59 2.4 Aims of the present study ............................................................ 64 2.5 The SCEGAD corpus .................................................................. 65 2.6 Analysing social interaction and identity construction in academic texts ......................................................................... 73 2.7 Method of analysis....................................................................... 81 3. Me, you and us: fashioning authorial, reader and social identities ............................................................................ 91 3.1 Constructing and expressing identity in academic texts .............. 92 3.2 Authorial presence and hedging .................................................. 96 3.3 The taboo and reality of authorial presence in English and German-language academic texts .......................100 <?page no="8"?> 8 3.4 Person reference according to language, variety, culture, discipline, gender and other variables .......................... 110 3.5 Person reference and discourse strategies.................................. 135 3.6 Multiple identification and multiple allegiances in academic prose ...................................................................... 145 3.7 Relationships between grammatical form and communicative function ............................................................ 151 3.8 Conclusion ................................................................................. 160 4. Writing about writing: author-reader interaction through textual metadiscourse ........................................... 165 4.1 Author, reader and text .............................................................. 166 4.2 Defining metadiscourse ............................................................. 169 4.3 Metadiscourse, evaluation, appraisal and co. ............................ 173 4.4 Textual and interpersonal metadiscourse................................... 175 4.5 From textual metadiscourse to text comments .......................... 178 4.6 Types of text comments............................................................. 182 4.7 Text comments, discourse deixis, advance organisers and co. .. 187 4.8 Text comments from a cross-cultural perspective .................... 191 4.9 Text comments according to language, variety, culture, discipline, text position and other variables ................. 199 4.10 Relating to the reader in academic prose .................................. 226 4.11 Metadiscourse and readability ................................................... 264 4.12 Conclusion ................................................................................. 267 5. Implications and applications of the study .................... 273 5.1 Summary of findings ................................................................. 274 5.2 Anglo-Saxon empiricism and Teutonic theorisation: beyond the stereotypes............................................................... 279 5.3 Implications for EAP teaching and learning.............................. 287 5.4 Approaches to EAP in a lingua franca world ............................ 295 5.5 Improving the situation of non-native speaker researchers ...... 299 5.6 The situation for scholarly languages other than English .......... 301 5.7 Directions for future research .................................................... 306 References .......................................................................................... 311 Appendix: corpus texts ................................................................. 349 <?page no="9"?> 9 Acknowledgements This book is based upon my PhD thesis, Writing science in English and German: a corpus-based study of academic discourse, which was accepted by the Philologische Fakultät of the University of Freiburg in February 2006. I would like to thank my supervisor, Prof. Dr. Dr. h.c. Christian Mair, for his good humour and patience, and gratefully acknowledge the financial support of the Landesgraduiertenförderung Baden-Württemberg through its PhD scholarship programme. Numerous current and former colleagues generously commented on earlier versions of my work. My sincere thanks to Nuria Hernandez, Daniela Kolbe, Bianca Kossmann, Iman Laversuch, Michelle Miles, Nadja Nesselhauf and Ursula Weinberger. Helpful comments and much-appreciated encouragement were also offered by Kirsten Adamzik, Annelie Ädel, Harald Bassler, Anna Breitkopf, Michael Clyne, Anje Müller Gjesdal, Maureen Hourigan, Marianne Hundt, Moshe Koppel, Elena Seaone Posses, John Swales, Winfried Thielemann, Christiane Thim-Mabrey, Peter Trudgill and Elisabeth Venohr. For reformatting and sending me material in a form accessible to my computerised analysis, I am most grateful to Wolfgang Adam and Diana Stört from Euphorion, David Hunt, Chris Payne and Fiona Willis of the Oxford University Press, Beate Plugge and Sabine Verk-Lindner from the Waxmann Verlag, Eckhardt Treichel from the Historische Zeitschrift and Anne Wulf from Griebsch & Rochol-Druck GmbH, who supplied me with articles from the Zeitschrift für Volkskunde. Finally, I wish to thank my partner Lars, who suffered me through many a research-related predicament, and, when my scholarship money ran out, became my financial support too. <?page no="10"?> 10 Abbreviations BNC British National Corpus BROWN Brown Corpus of Standard American English DaF Deutsch als Fremdsprache [German as a Foreign Language] EAP English for Academic Purposes EFL English as a Foreign Language EGAP English for General Academic Purposes ESAP English for Specific Academic Purposes ESL English as a Second Language ESP English for Specific Purposes FLOB Freiburg-Lancaster-Oslo/ Bergen Corpus FROWN Freiburg-Brown Corpus of Standard American English ICE International Corpus of English IMRD Introduction-Method-Results-Discussion L1 First-language L2 Second-language LOB Lancaster-Oslo/ Bergen Corpus LSP Language for Specific Purposes MICASE Michigan Corpus of Academic Spoken English NS Native Speaker(s) NNS Non-Native Speaker(s) RA Research Article SCEGAD Synchronic Corpus of English and German Academic Discourse <?page no="11"?> 11 Figures and tables Figure 1.1. Selected languages of publication for ISSN 19 journals in 1991 and 2003 Figure 2.1. Various ways of categorising Englishand 55 German-language (academic) discourse Figure 2.2. SCEGAD composition according to variables 68 Figure 2.3. Journals in the SCEGAD corpus 70 Figure 2.4. Coding of independent variables in frequency 84 tables Figure 2.5. Example of SPSS data input for 10 English- 85 language folklore articles Figure 3.1 English and German-language pronoun forms 110 analysed in the corpus Figure 3.2. Person reference: English and German-speaking 112 authors compared Figure 3.3. Person reference: prototypical profiles for 115 English and German-speaking authors compared Figure 3.4. Person reference: comparative difference in 115 frequency for Englishand German-speaking authors Figure 3.5. Person reference: British and US-American 116 authors compared Figure 3.6. Person reference: FLOB and FROWN corpora 118 compared <?page no="12"?> 12 Figure 3.7. Person reference: prototypical profiles for British, 119 US-American and German authors compared Figure 3.8. Person reference: British, US-American and 120 German-speaking authors compared Figure 3.9. Person reference according to discipline: bar graph 121 Figure 3.10. Person reference by discipline: comparative 121 difference in frequency Figure 3.11. Person reference by gender: bar graph 128 Figure 3.12. Person reference by gender: comparative 128 difference in frequency Figure 3.13. Person reference by gender: cultural profiles 129 Figure 3.14. Which variables best predict person reference 134 frequency: stepwise multiple regression model summary Figure 3.15. Person reference by discourse purpose: English 137 and German-language texts Figure 4.1. Taxonomy of text comments developed for the 183 present study Figure 4.2. Various terms for text comments and closely- 188 related phenomena in previous research Figure 4.3. Taxonomy of metacommunicative utterances 195 according to Hutz (1997) Figure 4.4. Taxonomy of text-commenting devices according 198 to Fandrych and Graefen (2002) Figure 4.5. Text comments: English and German-speaking 202 authors compared <?page no="13"?> 13 Figure 4.6. Text comments: prototypical profiles for English 204 and German-speaking authors compared Figure 4.7. Text comments: comparative difference in frequency 204 for English and German-speaking authors Figure 4.8. Text comments: British and US-American authors 205 compared Figure 4.9. Text comments: British, US-American and 206 German authors compared: bar graph Figure 4.10. Text comments: British, US-American and 207 German-speaking authors compared Figure 4.11. Text comments by discipline: bar graph 208 Figure 4.12. Text comments by discipline: comparative 208 difference in frequency Figure 4.13. Overarching text comment placement in SCEGAD: 219 cultural comparison Figure 4.14. Advance organiser placement according to 221 Clyne (1987: 229): percentage of all instances Figure 4.15. Overarching text comment placement in SCEGAD: 221 percentage of all instances Figure 4.16. Text comments and text length: Spearman Rho 223 correlation analysis Figure 4.17. Which variables best predict text comment 225 frequency: stepwise multiple regression model summary <?page no="15"?> 15 That English is today’s dominant language of science is stating what would be called a Binsenweisheit in German, a trivially obvious insight. _______________________________________ (A MMON 2001: v) …academics cannot help transporting into their professional exchanges knowledge of genuine interaction principles within their native speech communities. _______________________________________ (D USZAK 1997: 16) Chapter 1 Introduction 1.1 What this study is about This study is about how people use language in an academic context. In particular, it is about how scholars from Great Britain, the USA and Germany use their native language in research articles to construct and express various identities within a defined social group, how they interact with their readers and the wider academic community, and how they explicitly refer to the structure of their texts. Academic texts reflect their authors’ assumptions about the functioning of scientific exchange, their place within this exchange, and the audiences they address. As this study will show, these assumptions differ across cultures, languages and disciplines. Although research writing has traditionally been characterised as the decontextualised and disconnected transmission of universally-valid facts, it is in fact a personalised presentation of subjective arguments, particularly in the humanities texts examined in this study. Academic authors of different cultural backgrounds react differently to the tension between ideal-type <?page no="16"?> 16 scientific objectivity and the reality of individual subjectivity, allowing themselves varying degrees of visibility within their texts. Cultural background also exerts a significant influence on the kind of relationship academic authors establish with putative readers. Even in this typically distanced written genre, there are nonetheless sites where the communication between author and reader(s) is constructed as a dialogue, and the participants in the dialogue construed as partners in a cooperative textual enterprise. Such interpersonal content constitutes a meta-level of suppositions and disclosures, which runs parallel to the informative academic content cast as the primary purpose of such texts. Though the initial focus of this study was on possible cultural differences, it was clear even from the beginning that culture is only one of many variables which influence how academic authors write, and these other variables too are considered. Discipline in particular emerged as an influential indicator of linguistic choices, and the study compares and contrasts texts from five disciplines: philosophy, history, folklore, literary studies and linguistics. Other influential variables included author gender and academic status. Thus, the study has much to offer even to those readers not solely or even primarily interested in the role of culture in academic writing, as well as much that will broaden the perspective of readers whose main interest does concern cultural differences. This first chapter is intended to provide some background to the study. It critically examines the rise of English as an international scholarly language and suggests a role for linguistic research in assuaging some of the difficulties which have resulted from the current dominance of English in the scientific domain. The following chapter uses selected previous research to illustrate the methodological challenges facing cross-cultural research on academic discourse, and suggests possible ways of confronting these challenges. The chapter then introduces the present study in more detail, including the corpus upon which it is based. Following an examination of different approaches to studying social interaction and identity construction in academic texts, the second chapter also elucidates the objects of study and methodology chosen. The following two chapters present the results of the study. Chapter three is concerned with identity construction and interaction through person reference. It provides both a theoretical discussion and an empirical analysis of the communicative strategies academic authors use to position and represent themselves and others via language. It emerges that culture in particular, but also discipline, author gender, age and academic status, all exercise a significant influence on the frequency and use of person reference in the corpus texts. The fourth chapter focuses on author-reader interaction through textual metadiscourse. It offers a multifaceted view of the devices academic authors use to man- <?page no="17"?> 17 age the interface between writer, reader and text, and shows that authors from different cultures and disciplines use metadiscourse to create, structure and maintain different kinds of relationships, in different ways, with their imagined readers. The influence of other variables, such as author gender, age and academic status, on the frequency of the type of metadiscourse examined in the study is, however, not shown to be significant. The fifth and final chapter sums up the findings of the study and makes the point that well-constructed contrastive studies can not only uncover or confirm posited differences, but also question previously accepted dichotomies or stereotypes across cultures, disciplines and genders. This recognition can lead to a more nuanced understanding of writing in its social and cultural context. The final chapter also considers the implications of the study, and others like it, for EAP teaching and learning, and evaluates approaches to EAP in an (academic) world dominated by English. It proposes specific, concrete ways of improving the situation of nonnative speaker academics and also assesses more generally the situation of scholarly languages other than English. The study concludes with some suggestions for future research. In studying a complex phenomenon such as culture through the medium of language, it is vital to treat each of these variables separately. No matter which definition of language one subscribes to, it is clear that each language provides a different set of linguistic resources for achieving similar communicative aims. This study shows that there are considerable differences among typologically similar languages in the way these resources are employed within a defined communicative context. In the case of British and US-American English, formally identical items of the linguistic inventory are used differently in the same situation. The same is found to be true for male and female native speakers of the same language, and for authors with the same native language and gender, but of different academic status. This suggests that these differences in language use are due primarily to the influence of extra-linguistic variables, such as culture, rather than to differences in the linguistic resources or categories available to any one author. 1.2 Focus of the study The study concentrates on Englishand German-speaking academics for a variety of geo-political, practical and personal reasons. At a geo-political level, English is of enormous and continued research interest because of its present status as the global lingua franca, in academia and countless other domains. German too is of interest for geo-political reasons, as a major <?page no="18"?> 18 European language and former scholarly lingua franca. At a practical level, since English is at present the foremost language of international scholarly communication, it is hoped that a study of discourse norms underpinning two major varieties of English will be of use to NNS researchers now writing in English and those who teach them. The practical consideration in choosing German is that German academics are currently adjusting to the much-reduced status of their language on the international stage, while trying to maintain a long tradition of high-quality scholarship, now to be produced largely in a foreign tongue. The personal reason for the concentration on Englishand German-speaking academics concerns my own background as an English native speaker, who has worked and studied also in a German university environment, and publishes research in both languages. This background furnished me with an in-depth knowledge of the languages and (academic) cultures under examination, which proved to be of great use when conducting the study. 1.3 Science speaks English English is currently the main language of scholarly exchange, particularly in written domains and international contexts. The rise of English as a language of scholarly communication is of course inseparable from its rise as a world language, and its current prevalence in many other communicative domains alongside academia, such as the internet, business and finance, entertainment, politics and diplomacy, law, medicine and aviation. The background to this development - British colonialism, the Industrial Revolution, the post WWII geo-political dominance of the USA and its vast investment in research and development, the information technology revolution and so on - has been discussed in numerous studies and therefore need not be recounted here (see instead Fishman, Cooper & Conrad 1977; Bailey & Görlach 1984; Bailey 1992; Kachru 1992; Crystal 1997; Graddol 1997; King 2004). The particularly dominant position of English in published academic writing is not restricted to specific disciplines, but almost equally apparent in the natural sciences, social sciences and humanities (Ammon 1998). Comparative analyses of the languages of scholarly publication, as reflected by international databases of discipline-specific abstracts and citations databases, show that English has no competitors in this function (Applequist 1981; Tsunoda 1983; Garfield 1990; Ammon 1998; Sano 2002). A similar picture is obtained by examining the statistics from the ISSN (International Standard Serial Number), a standard database of over one million scientific records, two thirds of which are periodicals, published in <?page no="19"?> 19 some 372 different languages. Figures for selected languages over a twelve-year period are shown in Figure 1, along with the ranking of each language (unchanged over the period for all languages except Chinese): Figure 1.1. Selected languages of publication for ISSN journals in 1991 and 2003 (earliest and latest available figures) ______________________________________________________________________ Language(s) of 1991 2005 publication ________________________ ________________________ Records listed ISSN ranking Records listed ISSN ranking ______________________________________________________________________ English 244, 568 1 448, 157 1 French 125, 810 2 255, 692 2 Multilingual 46, 495 3 92, 994 3 German 33, 146 4 57, 997 4 Spanish 16, 429 5 48, 222 5 Russian 6, 050 14 16, 653 14 Chinese 3, 056 17 15, 624 16 ______________________________________________________________________ These data demonstrate the popularity of English in periodicals. 1 Frenchlanguage periodicals are a distant second to those published in English, and show no sign of catching up. Though the number of Chinese-language periodicals assigned an ISSN has increased five times in the space of 14 years, Chinese-language publications would need to increase roughly 30fold in order even to approach the number of English-language publications. 2 This makes Graddol’s recent comment “in the next decade the new ‘must learn’ language is likely to be Mandarin” (2004: 1330) appear rather premature, at least for academia (though it may well prove accurate for business). In recognition of English-language dominance in academia, numerous countries have introduced English as an additional medium of instruction for higher education. In Europe, these countries include Denmark, Germany, the Netherlands, Norway and Sweden (Ainsworth 2001; Ammon 2001; Gunnarsson 2001). In Eastand Southeast-Asia, McConnell counts that 9 of the 18 countries he examined use English as a language of teaching at the tertiary level; in Brunei, the Philippines and Singapore, English is the sole language of post-secondary instruction (2001: 118-120). 3 In the Middle East, English-medium classes are common at tertiary level in many 1 No similar records exist for monographs or other forms of scholarly publication. 2 This does not necessarily represent any actual increase in the number of Chineselanguage publications, which may have existed, without an ISSN number, for many years before inclusion in the ISSN system. 3 The nine countries McConnell examined were Brunei, Hong Kong, Taiwan, Malaysia, the Philippines, Singapore, China, Macau and Thailand (2001: 119-121). <?page no="20"?> 20 countries, with public universities in Oman, Qatar and the United Arab Emirates teaching almost exclusively in English (Zurghoul 2003: 19). English is the sole language of higher education in Eritrea and Ethopia (Gordon 2005), and is widely used at post-secondary level in many other African countries, such as Nigeria, South Africa and Zimbabwe (Read & Ambrose 1998; Nationmaster 2005). The prevalence of English as the international language of academic communication is widely recognised, not just by English-speakers, but also by researchers belonging to formerly dominant language groups (Godin 1995; Truchot 1997; Ammon 1998, 2001; Frath 2001; Mühleisen 2003). Some German linguists still choose to downplay the situation, noting that “Englisch hat beste Aussichten, sich als internationale Wissenschaftssprache durchzusetzen” [English has the best of prospects for asserting itself as the international language of science] (Adamzik 1998: 103). However, Ammon, who has undertaken extensive original research on the subject, maintains that the dominance of English as a scholarly language is nothing more than a “Binsenweisheit”, a German word for a trivially obvious insight (2001: v). To emphasise this point, one of Ammon’s popular-science publications on the topic is illustrated by a photograph of a gorilla skeleton bearing the caption “Fast ausgestorben: Deutsch als Wissenschaftssprache” [Nearly extinct: German as a scientific language] (1993: 21). In France, the President of the association Avenir de la langue française [future of the French language], an organisation for the promotion of French, had the following to say: La régression de la publication scientifique en français continue cependant dans une indifférence apparemment résignée, au point que l'activité intellectuelle en français risque bientôt de dépérir. [Nonetheless there seems to be an apparently resigned indifference to the decline in (the number of) scientific publications in French, which has reached the point where intellectual activity in French may well soon disappear altogether.] (Salon 2005) This was not always so; at the beginning of the 20th century, for example, there were three main languages of international scholarly publication: English, French and German (Ammon 2001: 343-5). Tsunoda’s longitudinal (1900-1980) study of abstracts databases attests to the erstwhile importance of French and German - alongside English - as languages of publication in medicine, mathematics, physics, biology, chemistry and geology (Tsudona 1983). Tsudona dates the emergence of English dominance in scientific publications to the period after 1960 (1983: 78). However, since we are dealing with a process which began much earlier and is ongoing, set dates are in a sense misleading. In the case of German, Ammon dates the beginning of its decline as an international scholarly lan- <?page no="21"?> 21 guage to the 1920s, and the rift in the international scholarly community caused by WWI (1998: 11; see also Swales 2004b: 33-6). Tsunoda dates the decline of French as a scholarly lingua franca somewhat earlier, tying it to the increasing dominance of the British Empire, the concurrent loss of French colonial influence, and the economic and scientific rise of the USA (1983: 78). Certainly, already in 1925 Sapir noted that English had become “de facto the international language of modern times” (1925: 65). From this position, it was a short step to dominance as the international research language. It is not only the case that the majority of publications is in English: the majority of leading scholarly publications is in English, too. Surveys of NNS have found that work published in the global lingua franca is perceived as being both more prestigious and more widely-read than is work in other languages. It is also cited more often. In Germany, where English enjoys high status in many domains, there has long been a feeling that English-language research is not only quantitatively dominant but also qualitatively superior. Twenty years ago, the then President of the Max Planck Gesellschaft, Professor Hubert Markl, coined a phrase now wellknown in Germany: “Die Spitzenforschung spricht Englisch” [top research speaks English] (Markl 1986). 4 The same situation - English-medium equals high-quality - has been recorded for the Netherlands (Vandenbroucke 1989: 1461). As regards the attitude of NNS researchers towards English-language publications, Ammon’s survey of German academics for example revealed that, for the publication of results they considered especially important, 78% of the informants would choose English and only 16% their mother tongue (1998: 106). One of the reasons why so many of the world’s top NNS academics decide to publish in English is their belief that publications written in their mother tongue would simply be ignored, or rather, not read the by the audience they desire. In a large-scale study undertaken by Stickel, fully 81.1% of the 2000 German researchers surveyed were of the opinion that, in their particular discipline, German-language publications were “seldom or definitely not regarded internationally” (2000: 135-6). The situation is similar for other languages: Sano for example estimates that “the number of readers of papers written in English will be tento twenty-fold larger than that of papers written in Japanese” (2002: 49). Wood notes that: 4 The Max Planck Gesellschaft [Max Planck Society for the Advancement of Science] is a non-government and non-profit organisation, which funds a great deal of high quality research at 80 independent institutes throughout Germany, as well as in Italy and the Netherlands. <?page no="22"?> 22 for a scientist to publish in a language other than English therefore is increasingly to cut herself off from the worldwide community of scientists who publish in English. The work may then be ignored simply because it is published in a language unknown to the rest of the world. (1997) 5 If NNS academics publish in their mother tongue, they will be read by other NNS with the same native language. It seems, however, that NNS do not particularly value the attention they would receive from native colleagues. Instead, they seem to covet the attention of a different audience - one that is not only larger, but also more influential. As Vandenbroucke observes for Dutch, “those scientists aspiring for international recognition write directly in English” (1989: 1461). Since fellow-NNS academic colleagues may be assumed to speak English anyway, the strategy appears to have no drawbacks. Subjective impressions regarding the greater prestige and international reception of English-language publications are lent empirical support by citation analyses. Garfield (1990) attests that: […] more source items are published in English by both native and nonnative [sic] speakers than any other language, and they have the highest impact. Also, most major scientific nations, regardless of their native language or languages, cite the English-language literature almost exclusively. (1990: 294) Similarly, Jaffe’s work on citation statistics found that “of the 40 most cited papers of the last five years in the database of the Institute for Scientific Information (ISI), not a single one was written in another language [than English]” (2003: 44). The work of Garfield and Jaffe therefore confirms the assessments of NNS scholars, and the statements by Wood (1997), that research by non-native speakers will not be internationally recognised unless it is published in English. The result, as a respondent in Schiewe’s study of academics from Freiburg university noted, is that “man damit gezwungen wird, um gelesen und bekannt bzw. zitiert zu werden, sich nur noch in dieser Sprache [Englisch] auszudrücken” [one is forced to express oneself solely in this language (English) in order to be read and become known, that is cited] (1991: 37). For an academic, the potential advantages of being read by the ‘right’ - i.e. widest possible and most influential - audience are significant, since citation brings recognition and prestige within the international community of academics who work within 5 In his book Words of the World, Abram de Swaan refers to movements that advocate the right of all people to speak the language of their choice: “alas, what decides is not the right of human beings to speak whatever language they wish, but the freedom of everyone else to ignore what they say in the language of their choice” (de Swaan 2001: 52). As in academic publication, wishful thinking is sometimes painfully confronted by reality. <?page no="23"?> 23 the same specialised field. Scholars whose work is well-known may be invited as key-note speakers at international conferences, receive pay rises, increased external funding and quicker promotion, or perhaps even awards such as honorary degrees. Publication which does not result in citation can be pointless not only for the individual scholar, but also for the community as a whole. Though the research may be of great worth and potential usefulness, it will be of little actual use if not read or cited, since academic findings, and indeed the progress of research as a cooperative and productive social activity, rely on dissemination and wide discussion among fellow researchers. NNS academics hope that, if they write in English, they will be read, discussed and also cited more often, and by more influential colleagues. This maximises the impact and usefulness of their research, while at the same time strengthening their position within the academic community. No wonder NNS academics publish in English. 1.4 The situation for non-native speaker researchers Many scholars welcome the popularity of English for academic communication, pointing out the advantages of international exchange in a common language. Montgomery writes of “recovering the ancient dream of a single language for the wisdom of nations” (2004: 1333). He points out that, in this function, English aids and encourages international cooperation and exchange, increases career options for academics, and advances technical progress in general (2004: 1334; see also de Swaan 2001). Sano notes that lingua franca English guarantees a larger audience for (English-language) publications, which is a “matter of great satisfaction” (2002: 49) for scientists, since it is through reader feedback that researchers develop their ideas and create new ones. The vast majority of the 92 German academics surveyed by Schiewe (1991) also saw nothing disadvantageous in the advent of English as the scholarly language. On the other hand, there is a large and vocal group of researchers who consider this development to be almost exclusively negative. Alexander problematises the “way English-based communication is talked about in glowing terms, as a contribution to world peace, to cooperation between nations, as a vehicle of educational resources” (2003: 89-90). The problem, he writes, is that “this discourse distracts from what happens when national resources in many countries are diverted towards learning a foreign language and away from self-development, both linguistic and cultural” (2003: 90). This comment suggests that learning foreign languages, and English in particular, is somehow incompatible with or even injurious to self-development, which is paradoxical when one considers that many ESL <?page no="24"?> 24 and EFL learners report that they are learning English precisely in order to develop themselves and open up opportunities otherwise closed to them (see for example surveys by Niiya 1998; Jamsen, Knutson, Scepanski & Strain 2004; Al-Khatib 2005; Rahman 2005). Other criticisms of international academic English are similar to those levelled, more generally, at English as a world language. Mühleisen for example is of the opinion that: […] possible negative effects [of English as the almost exclusive language used in the sciences and the humanities] are drastic and include the loss of ever more languages as well as the loss of knowledge encoded in those languages. The possible positive effects sound rather mundane in comparison. (2003: 117) This echoes influential work by researchers such as Phillipson (1992), Pennycook (1994) and Phillipson and Skutnabb-Kangas (1995), which talks of ‘linguicide’ and marks English a ‘killer language’. Here, a rather simplistic (mono)causal link is proposed between English-language acquisition and the extinction of other languages. Such contributions maintain, in a somewhat paranoid fashion, that English will automatically and necessarily be learnt at the cost of other languages, which need not be the case. There is absolutely no reason to assume that English alone cannot be simply one part of a multilingual speaker’s repertoire, if there is a need for this. The world is full of bilingual and multilingual people, hundreds of millions of whom speak English in addition to, not instead of, other tongues. Graddol therefore considers the idea that English will become the world language “to the exclusion of all others” to be “past its sell-by date” (2004: 1330; see also House 2003b). Its major impact, he writes, will be in “creating new generations of bilingual and multilingual speakers across the world” (2004: 1330). This harks back to Fishman’s assessment, nearly 20 years ago, that “the spread of English is likely to establish stable diglossia or even triglossia patterns rather than to intensify linguistic antagonisms” (1977a: 309). Skutnabb-Kangas is certainly justified insofar as she asserts the right of minority language groups to mother-tongue education (Skutnabb-Kangas 2004); what and how these people decide to speak thereafter, however, is their choice and their business only. Put simply, should there be a need for English only, people will speak English only. This is due not to any intrinsic homicidal characteristic or intent in the English language, but to each person’s assessment of their communicative needs. Language, after all, is not about killing other tongues, but about getting the message across. While recognising that English as a global scholarly language does indeed have advantages, on the international stage many non-native speakers feel themselves at an eternal disadvantage to native speaker researchers. <?page no="25"?> 25 Ammon (1989: 259) bluntly calls those who regard the expansion of English as purely negative “sehr engstirnig” [very narrow-minded], yet he is well aware of the negative effects its expansion can have. Duszak states plainly that “in an international discourse community founded on English, a nonnative [sic] speaker of that language is disadvantaged from the very beginning” (1997: 21). The paradox here is that, though English as a lingua franca is supposed to build bridges, for some academics, it can also lead to isolation. Based on a small-scale survey of 29 German university academics, Ammon describes how the respondents often avoided publishing in English-language periodicals, sometimes avoided personal contact with colleagues if it meant they had to speak English, and sometimes avoided going to conferences if they were to be held solely in English (Ammon 1989: 268-270). In these cases, the internationalisation of scholarship has led to the marginalisation and exclusion of those who lack sufficient English-language skills. Willemyns writes of the “undeniable and enviable advantage to native speakers of that language [English]” (2001: 341) when it comes to academic publication, and discussions along these lines surface in high-profile scientific journals such as the New Scientist, Science and Nature (Vandenbroucke 1989; Montgomery 2004; see also Wood 1997 and Arunachalam 1998). Jaffe (2003) quotes NNS researchers with similar concerns, and remarks that “almost any scientist will say that if you lack English fluency, your career will go nowhere.” For non-native speaker scholars, English is one more thing they have to worry about. Without doubt, from the perspective of the individual NNS academic, it requires much effort to learn a foreign language to a level high enough to publish in first-rate journals. Professional academics, like other Englishusing professionals worldwide, are elite language users and learners, who have usually learnt English in a structured environment, probably from primary school on. This costs much time and effort. On top of this comes the material expense of additional tuition and learning materials (textbooks, dictionaries, style guides etc.). For those academics whose English is not sufficiently advanced, or who are self-conscious about any possibly remaining errors, proof-reading or translation services may also be necessary. In addition, for a budding academic, it is advantageous to spend time in an English-speaking country as part of an undergraduate or postgraduate exchange. The (German) researchers surveyed by Skudlik, for example, regarded such an exchange as crucial to success when publishing in English (1990: 144, 146). Exchanges too cost time and a not inconsiderable amount of money. These requirements create a situation of English haves and have-nots. Nevertheless, even those wealthy or lucky enough to enjoy these opportunities remain at a disadvantage to native speakers of English. <?page no="26"?> 26 The disadvantage stems from the additional resources which English NNS academics must expend in order to compete with their NS colleagues at an international level. Pennycook (1994: 155) estimates that, in 1988, the amount of money private individuals spent on learning English was £6.25 billion (US$11.3 billion). Privately-spent resources (time, effort, money) can be added to the resources expended by nations to educate their citizens in English, which is perceived as a key skill, not just for academia (Fishman 1992; Crystal 1997; de Swaan 2001; Buck 2002; Zurghoul 2003). The amount of money governments spend on English-language training is difficult to quantify, since education budget statements are not generally divided up according to subjects taught. However, the recent discussion surrounding the introduction of English-language tuition in German and Swiss primary schools gives some indication: the Swiss canton Schwyz, with 10,684 primary-school students, 6 calculated costs of Fr1.9 million (ÚS$1.49 million) for additional teacher training alone (Huwyler 2002: 7). The German state of Bremen, with 24,228 primary-school students, 7 estimated that the introduction of English from the third grade on would cost an extra €663,260 (US$800, 000) in teaching costs for 2003 (Vogel & Buck 2002: 3). For the sake of comparison, consider the following: according to Chinese Ministry of Education figures, in China in 1999 there were 135.48 million primary-school students, all of whom learn English from grade one onwards (chinatoday.com 2004). Even if one accounts for the fact that wages are far lower in China than in Germany or Switzerland, it is nonetheless safe to assume that hundreds of millions, perhaps even billions, of dollars are spent on English-language instruction around the world. At the same time as nations with English as a native language are saving vast amounts of money through not having to teach English to their own citizens (as noted by Templer 2002), they are making equally vast amounts of money out of teaching English to others. In Australia, for example, international education was the fourth largest export industry in 2005, worth A$7.5 billion (some US$6.0) according to the latest official figures from the Department of Foreign Affairs and Trade (2006: 15). Even this huge sum is only a fraction of the amount of money made by countries such as the USA and the United Kingdom. In a recent speech in Beijing, 6 These are the figures from the Canton Department of Education’s Schulstatistik Kanton Schwyz 2002/ 2003 [school statistics Canton Schwyz], online under: http: / / www.sz.ch / rv/ index.html 7 These are the 2003 figures from the Statistisches Landesamt Bremen [State Office of Statistics Bremen], online under: http: / / www2.bremen.de/ info/ statistik/ aktuelle_ statistiken/ 09c.htm <?page no="27"?> 27 the British Chancellor of the Exchequer, the Rt Hon Gordon Brown MP, had the following to say: Britain is determined to lead the world in the provision of educational services. In future years we see it as one of our greatest export earners. In just five years the value of British education as an export has almost doubled, from £6.5 billion [US$12] to £10.3 [US$19] billion. Education and education-related services are our fastest growing export earner and have already eclipsed food, tobacco and drink exports, insurance, and ships and aircraft. Indeed, I believe that if we continue to make the right decisions, by 2020 education exports could contribute over £20 billion [US$36] a year to the UK economy. (Brown 2005) The latest available figures for the U.S. International Transactions Accounts show that US-American education exports totalled US$14.10 billion in 2005 (Bach 2006: 34). The people spending all this money, of course, are the NNS. Despite the huge numbers of NNS the world over who either already speak or are currently learning English, there are still important issues of access to English, particularly for NNS researchers from developing countries, who have limited opportunities to learn English to a sufficiently high level and also have little access to international scientific literature in either print or electronic form (Arunachalam 1998). 8 These researchers therefore reap scant benefit from the internationalisation of scholarship which lingua franca English brings. The exclusion of large numbers of people from the English-speaking international academic community reduces the usefulness of English as the scholarly lingua franca. It ought to be noted, however, that such inequality is part of a deeper global problem, which is not caused by English-language dominance, though for many people English symbolises it. The disadvantage felt by non-native speaker academics when they compete for international recognition and publication is certainly real: it is usually easier to write in one’s native language, as English native speakers may still do. Few scientific journals issue ‘Instructions to Authors’ specifically aimed at NNS (Gosden 2003: 98), or, in the case of the humanities, issue any stylistic advice at all. In addition, studies have found that the language advice English NS reviewers offer NNS authors on articles submitted for publication is often impractical, vague and unsubstantiated (Kourilová 1996; Gosden 2003). This frustrating situation - disadvantage without adequate help - is compounded by the enormous pressure non- 8 There are also researchers from countries who, though they have access to English, do not have access to the particular varieties of English required (mainly by NS) for scholarly communication. Academics from countries such as Jamaica, Nigeria or Singapore may have to ‘re-learn’ their own native language for academic purposes. <?page no="28"?> 28 native speakers are under to publish in English rather than in their mother tongue. The German scholars surveyed in the studies quoted above are acutely conscious of this pressure: for them, and for other NNS academics, the situation seems to be one of publish in English or perish. 9 1.5 The sting in the tail It is here that an unpleasant, and unexpected, situation emerges; there may be serious drawbacks even for those NNS who manage, after much cost, time and effort, to become fluent in both general and academic English. Many German academics, for example, are highly competent in both general purpose and academic English, and indeed often in other languages besides (Clyne 1984: 92; Kirk & Speier 1998; Ainsworth 2001: 4). In addition, and in contrast to their colleagues from developing countries, German academics have access to English-language publications, and generally can afford to participate in international conferences. Although some level of disadvantage of course remains, assuming German scholars are able to speak and write English well enough, one would think that they at least have the chance to be internationally competitive. When it comes to research publication, however, it appears that it is not sufficient for NNS researchers to write in excellent English. The reason for this is that, as numerous studies have shown, there is more to a successful research paper than the language it is written in (Kaplan 1966; Clyne 1987; Taylor & Chen 1991; Connor 1996; Meierkord 1996; Flowerdew 2001; Thompson 2001; Hyland 2002b; Ventola 2003). The central truth, as I have argued elsewhere (Sanderson 2006a), is that successful academic writing is not solely about content, but also requires an awareness of the accepted form that a piece of scholarly writing should take. By this I mean not just superficial matters such as footnote format or how tables are to be presented. Academic writing differs fundamentally from other types of written communication in the way authors conceive of their role in the text, the way they construct an authorial identity for themselves and how they interact with the reader. Learning how to write like an academic (academic literacy) is a crucial part of becoming an academic. The important part of this, for NNS academics, is that this form - the complex set of discourse conventions governing who writes in which way to whom - is culture-specific. A much-discussed contribution to the study of scientific discourse is Widdowson’s 1979 paper entitled “The Description of Scientific Lan- 9 Phillipson comes to the same conclusion, though he phrases this “publish or perish in English” (2003: 80). <?page no="29"?> 29 guage” (1979: 51-62; see also various critiques in Fix, Habscheid & Klein 2001). In this paper, Widdowson proposed what has become known as the theory of the “universality of scientific discourse” (1979: 54), in which he maintained that scientific discourse forms a secondary cultural system independent of the primary cultural system of each linguistic community. The appropriate extract from the essay reads as follows: I assume that the concepts and procedures of scientific inquiry constitute a secondary cultural system which is independent of primary cultural systems associated with difference societies. So although for example, a Japanese, and a Frenchman, have very different ways of life, beliefs, preconceptions, and so on deriving from the primary cultures of the societies they are members of, as scientists they have a common culture. In the same way, I take it that the discourse conventions which are used to communicate this common culture are independent of the particular linguistic means which are used to realise them. (1979: 51) This is true only at a very superficial level, in the sense that there are recognisable scientific text types - such a research articles - defined by their formal structural commonalities. As journal editors and educators note (Day 1979; Meier 1992; Anderson 2002; Kemp 2002), in the natural sciences the typical structure is roughly title, authors and affiliation, abstract, introduction, (materials and) methods, results, discussion, (conclusion), (acknowledgements), references, (appendices), (tables) and (figures and captions). This standardised structure appears to hold across cultures (Taylor & Chen 1991), though there is still a great deal of variation on the basic theme. It is, however, a mistake to assume, as does Widdowson (1979), that all areas of science function in the same way: in the humanities for example no such standard structure exists. Admittedly, even in the humanities there are some superficial similarities between academic texts written across various cultures. For example, both English and German-speaking historians produce articles with references and footnotes, and often include an introduction and a conclusion. But is this proof of a “common culture” (Widdowson 1979: 51)? Though Widdowson does not define what is meant by this term, surface-level similarities in reporting practices across some disciplines and some text types surely do not suffice. It is also extremely doubtful whether a partially shared way of presenting findings in some branches of the academy is enough to override culturally specific modes of interaction, beliefs and preconceptions. At a deeper textual level, there are such large differences between cultures, even within the same discipline, that one can no longer speak of all scientists as having a “common culture”. From an ethnological perspective, the presentation of scientific activity as immune to its cultural and social context itself seems <?page no="30"?> 30 culturally-bound, reflecting ideas about knowledge gathering specific to a particular culture and society (Beck 2004: 10). Widdowson’s assumption was just that, merely an assumption, and he offered no empirical support for what he took to be true. As soon as the assumption was tested, however, it was shown to be false. It is now accepted, thanks to the evidence of dozens of studies, that academic discourse is not universal, but fundamentally and substantially influenced by a range of factors, not least the cultural background of the author (see for example Clyne 1987; Oldenburg 1992; Mauranen 1993b; Connor 1996; Duszak 1997; Graefen 1997; Trumpp 1998). Studies have revealed differences at all levels of texts, from discourse organisation and macro-structure (Clyne 1987; Taylor & Chen 1991; Sachtleber 1993; Mauranen 1993b, 1996; Hutz 1997; Busch-Lauer 2001; Bassler 2003), to micro-level phenomena such as hedging (Clyne 1991; Markkanen & Schröder 1997), speech-act verbs (Graefen 2000b; Fandrych 2001), personal pronouns (Martín 2003; Suau 2004; Dahl 2004) and sentence structure in summaries (Oldenburg 1992). Differences have been discovered across many different academic text types or genres. Understandably, much work has been carried out on research articles, 10 or parts thereof such as abstracts (Hyland 2000; Huckin 2001; Samraj 2002; Bassler 2003; Martín 2003), introductions (Dudley- Evans & Henderson 1990; Taylor & Chen 1991; Oldenburg 1992; Thielmann 1999a, 1999b; Busch-Lauer 2001), titles (Busch-Lauer 2000; Dietz 2001, Haggan 2004) or acknowledgements sections (Cronin 1995; Giannoni 2002; Sanderson 2005). However, there have also been cross-cultural studies of other academic genres, such as peer reviews of research articles (Kretzenbacher 1995; Kourilová 1996; Gosden 2003), academic book reviews (Hyland 2000; Hutz 2001; Piitulainen 2001; Römer 2005), student essays (Kaplan 1966; Myers 2001; Thompson 2001; Helms-Park & Stapleton 2003), scientific letters (Hyland 2000), editorial letters (Flowerdew & Dudley-Evans 2002), textbooks (Myers 1992, 1995; Kearsey & Turner 1999; Hyland 1999c, 2000) and prefaces to historical companions to literature (Adamzik & Pieth 1999). Some researchers have also worked with mixed-genre corpora, consisting of published articles, textbooks and L2 student essays (Hyland 2002a), or of research articles, textbooks, grant proposals, prospectuses, reviews and popularisations (Myers 1996). Although at present the majority of previous studies has been concerned with written rather than spoken academic discourse, newer work has begun to fill this gap (Ventola, Shalom & Thompson 2002; Biber, Conrad, Reppen, Byrd & Helt 2002; Fortanet 2004b; Mauranen 2004; Simpson 10 Swales (2004b: 207) refers to the research article as “that master narrative of our time”, quoting Montgomery (1996). <?page no="31"?> 31 2004; Swales 2004b; Rowley-Jolivet & Carter-Thomas 2005). Such research has been greatly aided by the emergence of specialised corpora of academic speech such as the Michigan Corpus of Academic Spoken English (MICASE), the British Academic Spoken English (BASE) corpus and the corpus of English as a Lingua Franca in Academic Settings (ELFA). Studies have examined academic texts written by authors from a wide variety of cultural backgrounds, generally comparing English native speakers with native speakers of other languages, such as Arabic (Ostler 1987); Chinese (Taylor and Chen 1991); Czech (Cmerjková 1996); Estonian (Rummel 2005); Finnish (Mauranen 1993b); Finnish and Russian (Vanhala-Aniszewski 2004), French and Norwegian (Breivega, Dahl & Fløttum 2002; Fløttum, Dahl & Kinn 2006); German (Clyne 1987; Busch- Lauer 2001); Italian (Giannoni 2002); Italian and French (Rentel 2005); Japanese, Korean, Chinese and Thai (Hinds 1980, 1990; Eggington 1987; Kubota 1998); Polish (Duszak 1997); Russian and German (Bassler 2003); Slovenian (Peterlin 2005); Spanish (Simpson 2000; Neff, Diez, Prieto & Chaudron 2004; Martín 2005) and Ukranian (Zhabotynska 2001). These studies rest upon, and at the same time provide evidence for, a social constructivist view of knowledge, which regards scholarship not as a “rational, deductive affair, confined to the realm of ideas, but a human, institutionalised process” (Young 1996: 198). This of course goes against positivistic views of science (as will be discussed in chapters two and three), and against the assumption that scholarly findings are independent of the “linguistic means” used to communicate them (Widdowson 1979: 51). The contrastive cross-cultural studies cited above have established the culturally relative nature of academic literacy as an acquired way of communicating within a specific domain. This seems logical once it is accepted that, as Duszak notes, the skills researchers require for successful academic communication are “extensions of, and elaborations on, their skills in their native language, criss-crossing the areas of social, pragmatic, cultural, interactive and textual competence” (1997: 20). 1.6 A question of culture or a question of language? It is important to realise that the differences noted by language-contrastive studies are primarily due to culture and not language. If the differences were solely attributable to differing linguistic systems, once NNS of English wrote grammatically correct papers, all differences between their work and that of NS would presumably disappear. Numerous studies have shown that this is not at all the case (e.g. Kaplan 1966; Clyne 1987; Taylor & Chen 1991; Connor 1996; Meierkord 1996; Flowerdew 2001; Hyland <?page no="32"?> 32 2002a; Ventola 2003), and have suggested that the differences are due to transfer from native-language discourse norms. Such differences, these studies suggest, are at a deeper level than surface linguistic form. The idea that cultural background is more influential than linguistic background is also supported by the results of this study, which show that, even in cases where two (or possibly three) languages offer very similar lexicogrammatical possibilities to the writer, as is the case for English (British and US-American) and German, the use made of them in the same communicative context can be very different. Further, this study also demonstrates significant differences between the behaviour of British and US- American authors, who have even more similar linguistic inventories at their disposal, yet choose to employ them in different ways. At the most basic level, each language provides a linguistic repertoire, or set of tools, from which authors may choose. In this sense, some of the formal differences in how authors from different cultural backgrounds achieve the same discourse purpose are indeed assignable to the language system a culture uses. For example, authors writing in a language with no passive voice would not produce sentences such as ‘It has been shown that’. However, should they wish to avoid writing ‘I have shown’, they would certainly find some other way of doing this, without need of the passive. In any case, explanation through reference to different linguistic systems is only really useful for typologically highly dissimilar languages, which is not the case here. In addition, most languages provide more than one method for achieving the same discourse purpose, and it is here that something other than language begins to play a role. Assuming that each language generally has at its disposal any number of possible ways of structuring and expressing a given communicative meaning, idea or intention, some kind of extra-linguistic conditioning factors must help authors decide which option from the linguistic and stylistic repertoire to choose. As this study shows, one of the most influential extralinguistic conditioning factors on language use in academic writing is culture. It consists of a set of norms, both consciously learnt and subconsciously absorbed, which in any one context sanctions some choices and disapproves of others. Cultural norms may apply to all people sharing a culture, or only to specific subgroups of people within a specific cultural group (‘subcultures’). There is also a great deal of variation within cultures: as Scollon and Scollon note, “no individual member of a group embodies all of his or her groups’ characteristics” (2001: 170). Nonetheless, in an academic context, it appears that German, British and US-American cultural norms for the text type ‘research article’ differ. Culture does not operate in a deterministic fashion, but rather influences - whether consciously or subconsciously is difficult to determine - <?page no="33"?> 33 the choices made by individual authors. Culture also influences the other important party in academic writing, the reader. Academic authors write for a very specific, specialised audience. The expectations these readers have of an academic text are shaped in part by their cultural background, as are the authors’ assumptions about the expectations of their intended audience. As this study shows, and others quoted above have already shown, the way in which academic authors choose to present their findings is in fact influenced by many different factors, not just by language and culture. These include the intended audience, communicative situation, genre and academic discipline, as well as the author’s gender, academic status and age. The same is most likely true for reader expectations, though this cannot be examined directly in a corpus-based study. None of these factors dictates behaviour. Rather they provide or sanction opportunities for particular linguistic behaviours and neglect or marginalise others. In theory at least it is possible for an author to behave contrary to all expectations and defy all influences: one could make up verb suffixes not provided by the mother tongue, write a history research article as a limerick, and completely ignore all previous research on one’s topic. This rarely happens, because such an article would not be accepted for publication, but the fact that it is possible establishes an important theoretical distinction. We are dealing here with socially-motivated choices, not mathematically related causes and effects. It therefore follows that we are concerned with the boundaries of individual freedom within various, more or less voluntarilyaccepted, constraints (or rather necessarily accepted constraints, since nonconformity leads to censure). The concentration in previous studies on the variables of culture and language is understandable, because of the context in which this research has taken place. It is vitally important, for NNS, for there to be a better understanding of which norms underpin the target language’s culture(s). For most NNS academics today, that target language is English, and the target culture is most likely US-American or British, or possibly some other native-English variety culture such as Canadian or Australian. However, as the results will show, a narrow focus on language and culture to the exclusion of all other potentially relevant variables will lead to an inaccurate, fragmentary and ultimately unsatisfactory picture of language use in academic contexts. Since the sets of choices offered by different cultures are not the same, it is important to consider where this leaves NNS researchers publishing in English. There is evidence to suggest that English native speakers are not particularly tolerant when they encounter texts which do not conform to the discourse styles or patterns they expect. Ammon (1989) provides circumstantial evidence of this, based on his own and colleagues’ experiences as German academics now publishing in English. Skudlik (1990: 142-146) <?page no="34"?> 34 reports, based on surveys of German academics, that the transition to English-language publishing is not regarded as problematic. Based on his own observations, Ammon however describes this view as an “allerdings erstaunliche Selbsttäuschung” [quite astounding case of self-deception] (1989: 267) on the part of these colleagues, suggesting that they are simply in denial about the serious difficulties and disadvantages he describes. Clyne (1981, 1991) turned his attention book reviews, and concluded that English-language publications by respected German scholars had been criticised for a perceived lack of readability, without the theory or argument being in any way deficient: >Schwerfällig<, >konfus<, >weitschweifig< und >chaotisch< sind Ausdrücke, die häufig vorkommen. Andererseits wird Englischsprachigen von Deutschen oft vorgeworfen, daß sich ihre Texte oberflächlich, essayistisch und unwissenschaftlich lesen. [‘laboured’, ‘muddled’, ‘long-winded’ and ‘chaotic’ are expressions which often appear. On the other hand, Germans often accuse English-speakers of producing texts that are superficial, essayistic and unscientific.] (1991: 376) While there are obviously cases in which work written by English NS is also laboured or muddled, and NS reviewers can and do note this, Clyne argues that in the cases he examined, the texts would not have attracted these criticisms had the writers and reviewers been from the same cultural background. No study appears to have been carried out yet of stereotypes held by scholars towards other academic cultures. However, from my own experience, the assumptions about English-language academic writing which Clyne mentions do appear to be common among German academics. It seems possible that the closer we are culturally, the less we expect cultural differences, and therefore the less sensitive and forgiving we are of deviations from our expected norms. If this were the case, German academics publishing in English would be in a particularly difficult position. My own examination of some negative reviews of academic work by NNS supports Clyne’s impression that intolerant reviewers sometimes condemn features that may well be not poor academic writing, but evidence of foreign discourse norms in L2 production (Sanderson 2006a). However, I also showed that it is not just English native speaker reviewers who are sometimes culturally ignorant, insensitive and intolerant. Clyne’s suggestion that English-speaking scholars in particular are intolerant of cultural variation in academic texts is therefore questionable. So too is his association of Anglophone intolerance with the idea that ‘Anglo-Saxon’ academic style is supposedly ‘formally-oriented’, and as such less forgiving of formal deviations than are cultures which are putatively ‘contentoriented’ (Clyne 1981: 62, 1984: 471, 1993: 9). It seems to me that reviewers in general judge contributions, by both NS and NNS, according to their <?page no="35"?> 35 conformity to culturally specific discourse norms, and find it difficult to distinguish between inadequate content and inadequate methods of presentation, or rather, methods of presentation which do not conform to their expectations. The centrality of adherence to discourse conventions in the world of international scholarship is suggested by an interesting comparison. On the one hand, there are many cases of scientific fakes and spoofs, where nonsensical, worthless content has been accepted as serious scholarship because it observes the correct form. 11 On the other hand, there are indications that articles by NNS researchers are criticised or even rejected for publication, even if the content is valuable, because they do not abide by the expected discourse norms (Graefen 1994; Kourilová 1996; Kaplan 2000; Flowerdew 2001; Gosden 2003). This leaves NNS of English in an awkward position. Quite aside from the question of whether they should (or even wish to) conform to target-culture discourse patterns, they may lack the knowledge even of what these patterns are. It is here that linguistic research promises to help. By comparing texts by NS of English with those by NS of another language (or languages), it is hoped that studies will be able to elucidate differences in discourse norms which can eventually be translated into pedagogy. Non-contrastive studies of English-language academic texts written by native speakers seek to deepen our understanding of academic writing in practice and ensure that teaching recommendations accurately reflect the reality of actual texts. The need for, and interest in, such findings is great, since large numbers of NNS language-users stand to benefit from their practical application in advanced language teaching. Unfortunately, the research already carried out is not yet in a position to meet this need reliably. 11 Famous recent cases are the Sokal hoax of 1996, the ongoing controversy surrounding the inscrutable work of the Bogdanoff twins, and the SCIgen Rooter hoax of 2005. <?page no="37"?> 37 We still know surprisingly little about discourse Similarities or differences across texts and registers. _______________________________________ (B IBER , C ONRAD AND R EPPEN 1998: 106) Writing is social action. _______________________________________ (B AZERMAN 1998: 22) Chapter 2 Researching cultural trends in academic discourse 2.1 Research into academic discourse produced by German and English-speakers Early research into cultural differences in academic discourse was of an impressionistic, rather than stringently analytical or empirical, character. Researchers formulated vague and potentially offensive diagrams in an attempt to represent different cultures’ academic styles (Kaplan 1966; Galtung 1981) and the results of examinations of academic discourse across cultures were interspersed with personal observations about the format for addressing envelopes, use of office doors, procedure at business meetings and behaviour at parties, in queues and at university seminars in different countries (Clyne 1981, 1987). One of the earliest studies to consider cultural differences specifically between English-speaking and German-speaking academics was that of Galtung (1981) 12 in an “essay” comparing Saxonic (British and US-American), Teutonic (German), Gallic (French) and Nipponic (Japanese) approaches. As Galtung notes on the 12 Kaplan (1966) was not concerned with German academic discourse, and is thus not discussed in detail here. <?page no="38"?> 38 first page, the hypotheses contained within it are not based on any systematic analysis or evidence: Dear reader - what you have in front of you is very much an essay. It is based on impressions and intuitions, written down on paper and in my memory during many years of travels and stays in various intellectual climates around the world. (817) In fact, many features of the “intellectual styles” Galtung proposes appear to be little more than warmed-up ethnic stereotypes. He proposes that the Saxonic “intellectual style” is weak in paradigm analysis and theory formation (particularly in its US version), and very strong in description. The Teutonic intellectual style, in contrast, is strong in paradigm analysis and very strong in theory formation, but weak in description (823, 837). Both cultures are characterised as strong in their commentary on other intellectuals’ paradigms, propositions and theories (823). In addition, while the Saxonic style “fosters and encourages debate and discourse” in a cooperative, consensus-oriented spirit (823, 827), in the Teutonic style opinions are less diverse and criticism is less often qualified by initial praise. Discussants are more interested in immediately exposing the weakest point (824-5), whereafter apparently: […] there will be few if any soothing comments towards the end to put the defendant together as a human being; no attempt will be made to mop up the blood and put wounded egos back together. (825) Instead of indicating how these theories could be tested, Galtung states that no such experimental data are necessary: “how can all this be justified? There can be no reference to particular sources, it is all part of the general culture” (828). Such an approach contrasts with standard procedures in academic research, which are based on the scientific method. If there is no reference to any sources, the research cannot be critically examined and the theories cannot be refined or revised. One must accept Galtung’s theories on trust. Though Galtung’s work is engaging and at times thought-provoking, its method cannot suffice. His paper thus illustrates some possible pitfalls of comparing and characterising cultures without a sufficient data basis, and could have served as a warning for much later research. The studies that followed can be divided into three main groups according to the methodological approach: research into textual macrostructure, research into isolated textual phenomena, and attempts at comprehensive research examining assorted textual and pragmatic phenomena. The first, and largest, group of studies aimed at discerning stable macrostructural patterns across cultures by analysing sections and subsections of texts. This approach was based on the (structuralist) notion that <?page no="39"?> 39 articles could be portioned into Teiltexte [text sections] (Hutz 1997: 84) and Textteilsegmente [text-section segments] (Oldenburg 1992: 61), and that the arrangement of these sections and subsections varied across cultures according to Textbaupläne [text construction plans] (Oldenburg 1992: 61-2; Hutz 1997: 86-7; see also Gläser 1990: 55-56; Busch-Lauer 1992b; Minogue & Weber 1992). By studying the linguistic realisation and position in the text of these segments, researchers therefore endeavoured to identify culturally specific features. The approach was inspired by van Dijk’s (1980) work on textual macrostructure, and sought to determine which facets of macrostructure are fixed, and which are (culturally) variable. The system has much in common with that of Swales (1990), who analysed research articles according to a four-part introduction-methodresults-discussion (IMRD) structure, which is further divided up into ‘moves’ (Swales 1990; Hutz 1997: 102-3). Although Swales initially proposed models of standard moves, for example in research article introductions (Swales 1990), in later work he has warned against overly rigid application of his model (Swales 2002a). This suggests some ambivalence regarding this method of analysis even on the part of its creator. The next step in such analyses, the further division of each section into ‘moves’ subsections, must be carried out ‘intuitively’ (Hutz 1997: 103). Intuition can be an important guide in developing research questions, but not in delivering the answers to these questions: this is where some sort of scientific method is supposed to play a role. The methodology of macrostructure studies is therefore suspect, since it relies on the researcher’s personal intuition. The fact that intuition is influenced by cultural assumptions makes the method particularly unsuitable for cross-cultural studies. The exclusive focus on (assumed) macro-level patterns in information packaging is too restricted a way of looking at academic discourse: it ignores the social dimensions of texts and suggests that writing is simply a mechanical matter of putting the right bits in the right place. Further, the approach encourages a normative approach to writing, which neglects individual freedom and self-reflection as important facets of acquiring academic literacy, whether in the native or a foreign tongue. In practice, the rigid division of research articles into standardised and discrete IMDR sections is only feasible, if at all, in (some) natural and social sciences disciplines (Banks 1998; Al Ali & Holme 1999; Samraj & Swales 2000). Humanities texts seem to evince far more structural variation than those in the natural sciences, which means that a purely macrostructural approach to these disciplines is inappropriate. In any case, the prescriptive pedagogical worth of such a model is questionable: either target-culture texts in a specific discipline follow the IMRD structure, in which case they are likely to follow it across cultures, or they do not: in both cases, there is no need to teach the model. <?page no="40"?> 40 In a large study of textual macrostructure, Oldenburg (1992) examined the conclusions of 180 research articles published in engineering, economics and linguistics journals in the years 1985-1987: one German and one US-American journal were selected per discipline. The authors were “mit hoher Wahrscheinlichkeit” [most likely] all native-speakers of US- American English and German, or, at the very least, at least one of the coauthors per article was a NS, though it is not related how this was established (1992: 85). My own examination of some of the articles included in the corpus reveals that the US-American subcorpus contains first-named co-authors 13 or sole authors who are in fact of British, Canadian, Dutch, Greek, Indian, Irish and Spanish nationality. This undermines the validity of the study as a whole, and as a contrastive study of US-American and German academic styles. In any case, Oldenburg could find no systematic cultural differences in the macrostructure of the ‘conclusion’ text segment (1992: 161-163). He found that the conclusion was a highly variable text section, and distinguished four main structural patterns, which were mainly influenced by communicative situation and authorial choice (1992: 161). In the same study (1992), Oldenburg proposed an ‘integrative’ model of analysis for academic texts, which consists of a standard analysis of textual macrostructure according to the researcher’s intuitive identification of text sections and text-section segments (1992: 61-66). As noted, the model was not applied to whole texts. It does not consider levels of analysis other than macrostructure, nor does it integrate other perspectives on academic discourse into the analytical model (pragmatic, socio-linguistic, discoursal, communicative etc.). It is therefore difficult to regard the model as integrative in any real sense. In a previous study, Gnutzmann and Oldenburg (1991) had analysed the textual macrostructure of 20 introductions and 20 conclusions taken from articles published in one US-American and one German linguistics journal. As in Oldenburg (1992), the authors concluded that the structure of these text segments was highly variable, with no consistent patterns discernable either within or across languages, apart from the fact that the German-language conclusions were even more idiosyncratic in structure than the English-language ones (Gnutzmann & Oldenburg 1991: 214). In a later study informed by a similar approach, Hutz (1997) examined 60 RAs published in two social-psychology journals, one US-American and one German, in the years 1985-1993. The cultural background of the authors was not determined (1997: 79-80), though the articles in the US- American journal were “zum größten Teil” [in large part] from the USA (1997: 79). Hutz selected empirical articles only, since the theoretical or 13 The first-named author in natural sciences publications is generally the main author, and the one who actually wrote the paper, which in this case is significant. <?page no="41"?> 41 methodological articles demonstrated a “stark idiosynkratisch geprägten und somit variablen Textbauplan” [highly idiosyncratic and thus variable text structure] (1997: 80). The data collection was thus skewed; the 60 articles cannot be taken to represent the RA genre, and certainly not academic discourse as whole. Despite this fact, the explicitly stated aim of the study was to test whether academic discourse is universal or culturallybound (1997: 239). This momentous question cannot be answered - if it can be answered at all - on the basis of the data collected by Hutz. The author distinguished between the communicative, propositional and structural dimensions of texts, and - like Oldenburg before him - proposed an ‘integrative’ model of analysis in order to capture the reciprocal relationship between the three dimensions (1997: 82-7). However, since Hutz considered the corpus of 60 RAs to be too large for all dimensions to be examined, the study focused on the function of text segments only (1997: 85, 86-7). 14 By examining the subheadings already present in the texts, Hutz found that the empirical research articles selected from both journals followed the same IMRD macrostructure (1997: 98-99). Since editorial policy, coupled with the researcher’s pre-selection of the data, may have played a large part here, it is not clear that these ‘cultural similarities’ are genuine. Hutz then compared the number of sentences in different text sections, and found that while the English NS authors had the highest number of sentences in the method section, the German academics wrote the most sentences in the results section (1997: 101). It was also found that the German authors wrote a great number of sentences relating to the theoretical background of their work, while the US-Americans wrote more sentences presenting their own hypotheses (1997: 127). These results are not necessarily valid, because the length of the sentences - which can vary considerably within and across English and German - was not considered. The US-American authors may indeed have written more sentences relating to their own hypotheses, but if these sentences were half as long as those of the German authors, then we are dealing with a difference in syntax, not in discourse style. One cannot legitimately extrapolate the relative importance of discourse purposes from sentence length across languages. As regards text subsections, although Hutz intuitively identified the same moves in the respective sections, the sequence of these moves was idiosyncratic, particularly in the German texts, and followed no set pattern in either language (1997: 130-151). Therefore, no reliable cultural differences in text construction plans could be established. The cultural differences identified by Hutz - longer results sections and more theory in the German texts - were directly contradicted by 14 The 60 RAs were of unspecified word length, in total 546 pages. <?page no="42"?> 42 Busch-Lauer (2001), who reported exactly contrary findings upon examining text-section segments in the introductions to medicine and linguistics research articles written by British, US-American and German academics. This situation may be due to the fact that Hutz (1997) and Busch-Lauer (2001) drew upon different data, or it may be due to instability in the methodology common to both studies. As far as move patterns were concerned, Busch-Lauer came to the same conclusion as the previous studies: text organisation is influenced by communicative intention and publication guidelines (which dictated the IMRD structure followed in the medicine articles), but also is essentially idiosyncratic (2001: 64). In similar vein, Trumpp (1998) examined a mixed-genre corpus of English, French and German-language sport-science texts and found few significant cross-linguistic differences in the structure of the introduction and conclusion of research articles (1988: 174-5), though there were significant differences for other linguistic features, such as authorial self-reference and the use of topic sentences (both more common in English-language than in French or German RAs). Thus, while Trumpp found some differences across cultures in the linguistic realisation of particular communicative intentions, which will be discussed in the following chapters, no consistent cultural patterns of move types, combinations or sequences could be established. In conclusion, if one test of a methodology is in the results it provides, one may note the following: studies of sections of academic texts written by Englishand German-speaking scholars 15 have so far not found any systematic or consistent differences in textual macrostructure across cultures or languages. It is a moot point whether the absence of cultural tendencies is because stable differences in (intuitively-discerned) macrostructural patterns do not exist, or because the method is unable to detect them. The second group of studies on cultural tendencies in academic discourse was concerned not with textual macrostructure, but with specific, isolated phenomena in whole texts. Hedging for example was studied by Kreutz and Harres (1997), using a sample of 12 of the 52 texts contained in Clyne’s corpus (Clyne 1987). The authors examined phenomena such as agentless passives, passive infinitives, reflexives, modals and particles (Kreutz & Harres 1997: 187). The study found that the German texts contained “very few hedges overall” (1997: 189), though in the absence of any actual quantitative results it is unclear what this means. In any case, the authors could not confirm the posited connection between hedging and linearity/ digressiveness (1997: 196). Based on Clyne’s work, Kreutz and Harres assume that German academic writing is “more author-oriented and 15 The vast majority of these studies distinguish between authors solely on the basis of native language, not cultural background. As we have seen, some studies do not even ensure that the authors selected are native speakers of the language in which they write. <?page no="43"?> 43 less cooperative” (1997: 182). Further, they state that “it would seem natural to assume that in English texts in particular the reader plays some sort of participant role, that the texts are more interactive” (1997: 185). The researchers therefore interpret hedging as part of the “interactional aspect of discourse”, since “by modifying claims and statements, an argument becomes more open to discussion and is subjected to constructive criticism, as it draws in, or involves, the reader” (1997: 186). This seems to be the wrong order in which to conduct research: the phenomena are interpreted ad hoc in such a way as to fit a foregone conclusion. It is debatable whether the phenomena Kreutz and Harres examined can all be taken as evidence of reader involvement exclusively: rather it seems clear that the devices the researchers examined could serve very many different rhetorical functions. The fact that the German authors sampled employ fewer hedges therefore does not allow any general conclusion to be made about particular characteristics of German academic discourse. Another individual phenomenon, metadiscourse, was the subject of a number of studies by Fandrych and Graefen, who drew on a corpus of 36 research articles, 17 English-language and 19 German-language, taken from approximately 16 different disciplines (Graefen 2000b; Fandrych 2001, 2003; Fandrych & Graefen 2002; Fandrych 2005). The researchers state that the texts were “not chosen systematically” (Fandrych & Graefen 2002: 20), and that the authors examined in the study were not necessarily native-speakers (2002: 20, 39). The research focused on text-commenting devices, which are defined as those phrases in which the author comments on the organisation of the text (Graefen 2000b: 113). The researchers found that text-commenting devices were similarly frequent in both subcorpora (Graefen 2000b: 114; Fandrych 2001: 375; Fandrych & Graefen 2002: 21, 26), which contradicts Clyne’s (1987) findings. Fandrych and Graefen also discern differences across languages in the linguistic realisation of text-commenting devices: German authors employ more modal verbs (2002: 29) and prefer to structure their texts chronologically (text as process) rather than spatially (text as product) (2002: 33). However, these conclusions are to be regarded with scepticism, since the quantitative results - where offered - are presented in terms of absolute numbers (Fandrych 2001: 378; Fandrych & Graefen 2002: 28; Fandrych 2003: 6), and no total word counts are given for the two subcorpora. It would have been more helpful to adopt a consistent quantitative measure, for example per 10,000 words. Possibly, if a reliable quantitative measure had been adopted, the results could have been precisely the opposite: perhaps it is the English and not the German authors who employ more modal verbs and prefer to structure their texts chronologically. Unfortunately, it is not possible to test this without access to the data used in the study. The re- <?page no="44"?> 44 searchers do not consider whether the findings are due to differences in the language systems or to cultural preferences (Fandrych & Graefen 2002: 30-33, 35). Of course, it is possible that the results are due to neither variable, but in fact attributable to the intervening influence of discipline or composition date (the texts examined cover some twenty years of academic publication). The findings of these studies are therefore questionable and inconclusive. The last study of a specific phenomenon to be discussed here, by Hutz (2001), examined the formulation of criticism in 30 Englishand 30 German-language scholarly book reviews from linguistics and psychology. Here, Hutz departs partially from the methodology of his earlier work on macrostructure (1997) to consider discourse concerns as well. The study found that the German-language reviewers devoted slightly more space to critical evaluations, and formulated criticism in a more direct and confrontational fashion, than did the English-speakers (2001: 110-111, 119, 122-3). However, since the authors were not necessarily native-speakers, the observed differences may in fact be spurious. It is also possible that variables other than author native culture may be responsible for the observed results. The texts examined by Hutz span a publication period of fifty years, and it cannot be ruled out that changes in academic discourse norms over this time - rather than differences across cultures - may be responsible for some of the observed patterns. In addition, it is unclear what role disciplinary variation, another uncontrolled intervening variable, plays in explaining the results (2001: 119). There are statistical measures which could have clarified the relative influence of multiple independent variables (step-wise multiple regression). However, the author did not apply such measures, but rather attributed all differences to one variable only (culture). Returning to macrostructure, Hutz did not detect any consistent differences across languages in the text-construction plans of this particular text type. Considering the inconclusive results of previous studies of macrostructure, including that by Hutz himself (1997), this finding is as expected. In sum, while the methodology of the second group of cross-cultural studies of academic discourse seems more promising than that applied in the work on macrostructure, so far the results have been disappointing. The third and last group of previous studies on culturally specific tendencies in scholarly writing analysed assorted textual and pragmatic phenomena. These studies attempted the most difficult task of all: to provide a comprehensive picture of cultural differences in academic discourse written by Englishand German-speaking scholars. Some of the first systematic research projects with this aim were carried out by Clyne in the early 1980s. Following an initial contrastive investigation of examiner’s reports and essay-writing manuals (1980, 1981), Clyne (1987) analysed 52 texts <?page no="45"?> 45 chosen from a variety of academic genres (among them working papers, book chapters and published research articles) in the disciplines of linguistics and sociology. Of the 52 texts, 26 were written by US-American, British and Australian native-speakers of English (all in English) and 26 by German native-speakers from West and East Germany (of which 17 were written in German and 9 in English). Each German-authored text was matched to an English-authored example according to the author’s gender, text type, topic, length, purpose and intended audience. Clyne examined several features belonging to textual macrostructure, such as propositional hierarchy, symmetry and uniformity. He also analysed the thematic development in each text, as well as various lower-level textual features, in the form of advance organisers (metadiscourse), the definition of technical terms, and integration of additional data such as examples, statistics and quotations. These phenomena were assessed as contributing to either linearity or digressiveness in the text. The approach later rightly drew criticism for not clearly defining terms such as ‘linearity’ and ‘digressiveness’, and for proposing a model that could be neither replicated nor tested (Sachtleber 1990: 105, 1993: 65; Oldenburg 1992: 34; Graefen 1994: 141-49; Adamzik 1998: 104). Clyne’s general conclusion was that the texts in his corpus by German-educated scholars were likely to be structurally digressive (1987: 225), asymmetrical (226) and discontinuous (228), whereas the structure of those by the English-speaking academics tended to be linear, symmetrical and continuous. This conclusion was based on the observation that the German-authored texts contained propositions which were not dependent on the overarching proposition of the particular text section and did not follow the macro-propositions on which they depended. In addition, German scholars inserted text segments inside sections devoted to a different topic (225). The German academics sampled were also less likely to use advance organisers or define recurrent technical terms (229). Clyne speculated that knowledge is idealised in German culture, and is regarded as more important than the form used to present it (see also Clyne 1994: 187). Consequently, less effort is made to present material in a comprehensible fashion, it being the reader’s responsibility to understand what an author writes. In contrast, in ‘Anglo-Saxon’ culture, “knowledge is no more important than the way in which it is presented”, and the writer is responsible for communicating effectively (1987: 238). The notion of writer and reader responsible cultures, which Clyne adopts, had been introduced by Hinds (1980, 1987) following an examination of English and Japanese-language newspaper writing. Though the observations made in Clyne’s study may have been valid, the way in which they were interpreted was unsupportable (see chapter five for further discussion). <?page no="46"?> 46 A serious weakness with Clyne’s study was that it consistently characterised German academic discourse with emotive, negative language, for example: “digressive” (218), “repetitive” (212), “unrelated propositions” (219), “arguments not followed up” (219), “asymmetrical” (226), “not adequately integrated” (228) and “discontinuous” (228). In an earlier article, Clyne had described the non-linear structure of German academic discourse as best represented “by cooked spaghetti” (1981: 63). German academics in particular were understandably irritated by the characterisation of English-language discourse patterns as the norm, and German norms as deficient, objecting to Clyne’s “ethnocentricism” (Pöckl 1995: 103; Bassler 2002: 3; see also Graefen 1994: 141-49, 2000b: 123; Hutz 1997: 65). In fact, the inherent value-judgement in describing only one’s own culture as linear and all others as somehow digressive, deviant or deficient had already created problems for Kaplan (1966, 2000). German researchers stated that Clyne’s findings were not compatible with either their self-image or their own experience of reading German-language texts (Graefen 1997: 71), and asserted, without citing any supporting evidence, that Clyne’s findings were “heutzutage” [nowadays] regarded as too “pauschal und unzuverlässig” [sweeping and unreliable] (Adamzik 1998: 104). German scholars were personally, and justifiably, offended. On balance, however, the study remains one of the more methodologically sound contributions to research in this area, since it was one of the few that took care to construct a corpus of matched texts, controlled for extra-linguistic variables. Other studies attempted to provide a comprehensive picture of cultural differences by examining various other idiosyncratic assortments of textual or pragmatic phenomena. Baumann (1998) conducted qualitative and quantitative analyses of four phenomena: thematic progression (presence or absence of excursus sections), inter-textuality, text-structuring devices and the relationship between explicitly explained and implicitly assumed knowledge. The researcher examined three English-language and three German-language texts (total approximately 25,000 words) by British, US- American and German authors. The texts included one interview with a journalist, one semiotics article from a pharmaceutical company in-house magazine, one computing article from the magazine of the German doctor’s association, and three linguistics research articles. 16 The motivation for choosing these disciplines and text types is not revealed. Baumann proposed an interdisciplinary approach, which comprises “v.a. induktiv- 16 This is based on the information given in the appendix on the texts: it contradicts the information given in the body of the article, where it is stated that the ‘corpus’ contains single-discipline and interdisciplinary research articles from the disciplines Ancient History, Modern History, Communication Science and Management (1998: 306). <?page no="47"?> 47 empirische, deskriptive, formale, kognitive, sukzessive, simultane, statistische, strukturelle und funktionale Methoden” [mostly inductive-empirical, descriptive, formal, cognitive, successive, simultaneous, statistical, structural and functional methods] (1998: 306). This approach is recommended as a model for the complete analysis of all levels of text. However, since the model was demonstrated only partially, and this on only six texts, it is doubtful whether future studies could benefit from this example. The study found that the three German-language texts contained more excursus sections (how these were defined is not clear), more references to the work of other scholars, fewer text-structuring devices and a less clear connection between given and new information (1998: 310-313). The author interprets these highly subjective findings as evidence of the cultural differences between German and ‘Angloamerican’ specialised discourse. However, the data sampled are not representative of any of the cultures, text types or disciplines examined and the findings therefore cannot be generalised in this fashion. Buhl (1999) tested the hypothesis that English-language academic texts are generally reader-oriented and German-language academic texts authoror content-oriented by analysing four texts about the theory of relativity: one popular science book in English by Bertrand Russell and its German translation, and two texts in German by Albert Einstein (one research article and one popular science monograph). Only the scientific article was analysed in its entirety. The study found that the Germanlanguage original texts evinced greater terminological variation than the two English-language texts did, but were less likely to define technical terms and contained fewer instances of meta-communication and fewer redundancies. 17 All of these findings were interpreted as evidence of a greater degree of content-orientation in the German-language texts (1999: 138). While the methodological approach in this study, which owes much to Gerzymisch-Arbogast (1997b), is promising, since the data sample in fact contains only one specialised academic text, it is too small for any general conclusions to be drawn on this basis about British or German academic discourse. It is questionable whether Einstein can be taken as a model for German academic discourse, since he grew up in Germany, Italy and Switzerland, where he completed his secondary and tertiary education. In addition, it is lamentable that the texts are taken from a variety of text types, span a composition period of some 65 years, and are written by only two authors. It cannot be ruled out that genre variation, historical changes 17 The article was based on an unpublished graduate thesis, which had also examined textual features such as the title, beginning and conclusion of the text, thematic progression, concretisation, transphrasal theme-rheme structure, comparisons, examples and elements of colloquial language (1999: 125-133). <?page no="48"?> 48 or idiosyncratic writing styles are responsible for the observed results rather than cultural differences. Therefore, for this third group of studies too, the goal has proved elusive. None of the research projects yet completed has been able to provide a comprehensive description of the differences between academic prose written by Englishand German-speaking scholars. Interestingly, the studies in this third group suffer from the same methodological problems as those which examined only isolated phenomena, as discussed above. In other words, the methodological problems are not specific to a particular study aim. Neither are these methodological problems specific to studies of Englishand German-language texts; rather, they are evidence of wider issues in research into academic discourse as a whole. For this reason, the discussion in the following section will be somewhat broader in focus. There are two final comments to be made on the current situation of research into cultural differences in academic discourse. First, it is doubtful whether any one study is capable of examining or revealing all cultural differences in discourse norms. In most cases, these investigations have failed to offer an empirically-sound analysis of even one phenomenon among the many selected. The underlying realisation, which appears to be missing in previous research, is that cultural proclivities will, theoretically at least, be reflected in all aspects of language use, and no study can hope to achieve the mammoth task of looking at everything. Second, it is questionable whether this should even be the central goal: surely the most important consideration is that a study, no matter what its aim, should be empirically sound. At present, very few studies can claim this. At the very least, there are basic methodological considerations that need to be addressed before another attempt should be made at a comprehensive analysis of cultural differences in academic texts. 2.2 A discipline with dozens of methodologies From its very outset, research into academic discourse has been beset by methodological problems (noted by Taylor & Chen 1991; Oldenburg 1995; Kaplan 2000; Mauranen 2001), which, if the research discussed above is any indication, show little sign of imminent resolution. Authors of early studies later openly admitted the methodological weakness of their work: Kaplan for example commented that his seminal 1966 article was “based on an admittedly relatively poor research design” (2000: 83), and Clyne too acknowledged criticisms of his 1987 study, noting that “early attempts in a relatively new field are often methodologically unsophisticated” <?page no="49"?> 49 (1994: 167). Today, the field is no longer new, but the methodological unsophistication remains. Academic discourse analysis is a dynamic area of research, attracting contributions from linguists involved in many different fields. These include contrastive rhetoric (Hinds 1987, 1990; Kaplan 1966, 1987, 2000; Connor 1996, 2004), translation studies (Enkvist 1991; Gerzymisch- Arbogast 1993, 1997a, 1997b), ESP/ EAP (J. Flowerdew 2001, 2002c; Harwood 2005a, 2005b; Hyland 2000, 2002a, 2005), Fachsprachenforschung (Ammon 1989, 1998, 2001; Ehlich 1992, 2003; Baumann & Kalverkämper 1992; Graefen 1997; Trumpp 1998), genre analysis (Bhatia 1993, 2002; L. Flowerdew 2000, 2005; Swales 1990, 2004b), text linguistics/ Textsortenforschung (Wüest 1988; Kretzenbacher 1990; Mauranen 1993b, 1996; Adamzik 1998) and discourse analysis and intercultural pragmatics (Duszak 1997; House 1999, 2003a; Ventola 1996, 2003). These, and many other researchers, bring to the field a great diversity in approach. The absence of a single accepted methodology is characteristic of any number of disciplines, and need not necessarily be problematic. Indeed, in a positive sense, it enriches the work undertaken and provides a wealth of different impulses for future research. The drawback of a discipline without any one suitable model of analysis is that even studies which analyse the same phenomenon, in the same languages, and define this phenomenon in a similar fashion, are still usually not comparable. In some cases, the results of such studies are even contradictory (contrast for example Clyne 1987 with Fandrych & Graefen 2002, or Hutz 1997 with Busch-Lauer 2001). This makes it extremely difficult to survey findings on any one topic. The great variety of different approaches also precludes an overview of any academic text type as a whole. While these limitations exist in many fields, in this case they are serious, since they frustrate the process of translating the findings into pedagogical practice, which is an implicit if not explicit aim of most research in this area. Myriad viewpoints and backgrounds also complicate the task of identifying fundamental methodological issues, which must be addressed if the field is to progress. Research into academic discourse is a discipline with dozens of methodologies. Unfortunately, even among the abundance of suggestions, there is no individual model which convinces, and neither do they convince as a group. The central point is that previous studies are not empirically sound. My suggestion is that corpus linguistics could provide important impulses for increasing the methodological reliability and sophistication of future (contrastive) research on academic discourse. The individual issues may be addressed as follows. <?page no="50"?> 50 First, most previous studies of academic discourse are based on data samples of insufficient size. The stated aims are not achievable, and the posited conclusions not supportable, on the basis of the data analysed. In a potentially sensitive area of research such as contrastive text analysis, the question of what constitutes adequate data is potentially explosive, and it is simply not acceptable for studies to generalise about academic discourse styles based on four (Buhl 1999) or six (Baumann 1998) texts. Even 12 research articles (Suau 2004), 15 research articles (Mauranen 1993b), or 120 paragraphs of text (Mauranen 1996) seem scant basis upon which to draw conclusions about all academic writers from two different cultures. There are some encouraging exceptions to this criticism. One is Hyland and Tse’s work (2004) on doctoral and masters dissertations written in English by L2 postgraduate students at five Hong Kong universities. The authors collected 240 dissertations from six disciplines, totalling 4 million words. Assuming that the corpus is controlled for variables such as gender, discipline, composition date etc., this is probably a reasonable cross-section of Hong Kong student work from these disciplines in this period. Another promising study is Martín (2003), which examined 160 research article abstracts written in English and Spanish from leading journals in two closely-related disciplines. It is unfortunate though that the corpus was not controlled for author native language: it cannot be assumed that all writers, whether native speakers or not, “totally conformed to the rhetorical practices of the international English-speaking community”, simply because the papers had been accepted for publication (2003: 3). The absolute size of any text collection is meaningful only when judged in relation to the total number of items of which any given corpus is a sample (in statistical terms, the ‘population’). The one million word Brown corpus of written US-American English, for example, is therefore rightly considered small by corpus linguists, since it constitutes a tiny proportion of the total written output of US-American-English native speakers in the year 1961. In contrast, a million word sample of Queen Elizabeth II’s Christmas broadcasts to the Commonwealth, pageant speeches by Miss Indy Beauty Contest winners, or 20th century German constitutions, could well be representative, since there are so few examples of these text types in existence. When one considers academic text types such as monographs or journal articles, it immediately becomes clear that these are vastly more common than Christmas broadcasts by Queen Elizabeth II: Swales (1990) estimates the global output of research articles alone to be some five million per year. Therefore, it is patently obvious - to a corpus linguist at least - that a very large corpus indeed would be needed if one were to attempt to characterise such a text type as a whole. Ignoring this fact, many recent contrastive studies of academic discourse have nevertheless per- <?page no="51"?> 51 formed a quantitative analysis on less than a dozen texts (in total at most 50,000 words) and, based on this analysis, have made broad, often unqualified, claims about the discourse norms of particular cultures. Some of the claims made about Englishand German-language academic discourse for example are that ‘Anglo-Saxon’ authors follow the “obviously comprehensibility enhancing” principle of “writer responsibility” (Baumann 1998: 314), or that ‘Anglo-Saxon’ authors are reader-oriented and that German authors are content-oriented (Buhl 1999: 137-8). This may or may not be true, but it certainly cannot be discerned, nor tested, based on such minute text samples. The issue is not simply the absolute size, but the purpose of a text collection, since a corpus must be matched to the questions the researcher wants to ask. A huge corpus that is badly compiled is not superior to a well-compiled small corpus, which has been created to answer precisely those questions the researcher wishes to pose. The second issue of concern in research on academic discourse is that the samples selected are often not representative of the group(s) being analysed. As explained above, this is not a function of size. In order for a sample to be representative, it must contain all of the characteristics present in the wider population, in roughly the same proportions as in the wider population. For work on academic discourse, this means that a study of academics from one particular culture should include for example both men and women, of different academic levels and ages, tenured and untenured. Yet I know of not a single study on academic discourse which has even considered, let alone controlled for, these variables, apart from Römer’s Book Reviews in Linguistics Corpus (BRILC), which is controlled for one variable (gender), but only because gender was the subject of the study - all other variables, however, are left unaccounted for. 18 As we shall see below, this kind of methodological error is far less common in corpus linguistic research, though admittedly it can be a problem even there, as Römer (2005) demonstrates. Thus, contrastive studies which do not consider such factors are not representative of intellectual styles, gender tendencies, discourse norms or cultural traditions in academic 18 The fact that Römer (2005) does not account for variables such as author native language, native-culture and academic status in a corpus of academic book reviews is highly problematic. For example, in a study of how the corpus authors express criticism of the books they are reviewing, I should think it highly likely that author academic status within the community would be an important influence on how sharply one criticises colleagues’ work. It is also well-known that criticism is formulated differently across cultures. Since the corpus disregards all variables other than the one being examined (gender), the influence of variables such as these cannot be investigated. It is possible that the results are not due to gender at all, but to the influence of other powerful intervening variables, in which case the results are spurious. See also footnote 17 below. <?page no="52"?> 52 discourse. Similarly, a study of research articles is not representative of academic discourse as a whole. This may sound obvious, but only a very few studies that claim to describe academic discourse as a whole in a particular culture or cultures in fact work with mixed-genre corpora in order to support general claims (noteworthy are Myers 1996; Hyland 2000, 2002a). Likewise, a study of research articles from one or two disciplines is not representative of the whole genre (Mauranen 1993b; Martín 2003; Suau 2004), and this should be acknowledged in the study. The data collected for these studies are not in any way deficient - indeed, as mentioned above, Martín’s (2003) study design for example is in other respects relatively sophisticated. However, the data do not represent what the authors claim. They may very well be adequately representative of a subpopulation, and this is what the study should focus on. It is worth noting that this problem is not solved by constructing a corpus with one example each of texts from the largest possible assortment of disciplines (Fandrych & Graefen 2002). 19 This ensures only that the corpus is representative of neither any one discipline, nor of the genre as a whole, since there can be no guarantee in such a small sample that the sole archaeology paper sampled, for example, is not highly anomalous. There needs to be a balance between size and representativity, and this can come only from informed reflection when collecting data. In addition, a representative sample must be random, that is, the researcher must not pre-select the data to fit forgone conclusions, as in Hutz (1997), for example, who pre-selected the type of RA most likely to conform to a desired structure. If a sample is large enough, and the population has a normal distribution, the sample will probably be approximately representative, with a margin of error. However, it is extremely unlikely that any study of a main academic genre will ever be sufficiently large to ensure representativity on the basis of sample size alone. The implication is that researchers must consider which characteristics are present in their population, and try to ensure that the texts they select represent a broad and balanced (in the sense of being reflective of the population) cross-section of this population. This may necessitate scaling-down the scope of the 19 Fandrych and Graefen are very misleading when they criticise Clyne as follows: “Clyne’s empirical results may reflect certain writing traditions in the disciplines he looked at: (socio)linguistics and sociology. As our corpora are not restricted to certain disciplines, the supposed cultural differences seem to fade away.” (2002, 35). Precisely because he examined such closely-related disciplines, Clyne was in a position to examine whether the differences he found were due to cultural or disciplinary preferences. With an unmatched bilingual text collection taken from disciplines ranging from Neuropathology to US-American Literature, Fandrych and Graefen are in fact not in a position to make statements regarding the influence of either culture or discipline on textual production. <?page no="53"?> 53 study: rather than attempting to characterise academic discourse as a whole, it may be both more appropriate and more realistic to openly state that one studies male authors only, or physical science research articles only (Taylor & Chen 1991), or deliberately focuses on graduate theses alone (Hyland & Tse 2004), and if this subpopulation is of interest, such a study can be most rewarding. In corpus linguistic terms, the solution is to construct specialised corpora, and research into academic discourse could benefit from an awareness of this aspect of corpus linguistic methodology. Corpus linguists have long been constructing carefully controlled specialised corpora and drawing cautious but substantive conclusions on the basis of these corpora. Such a method offers an alternative to the unhealthy pattern of trying to characterise the discourse preferences of a particular culture as a whole, based on the necessarily insufficient empirical evidence permitted by finite resources. It is perfectly justified to work with a smaller corpus if the sample is representative of the total linguistic output. 20 There is no shortage of variables along which specialised corpora can be constructed: some possibilities are the intended audience, author gender or native language, language, date or place of publication for written texts, text type, discipline, topic etc. Numerous recently-compiled specialised corpora such as the Bergen Corpus of London Teenage Language (COLT), the International Corpus of Learner English (ICLE), the TEC corpus of English translational texts, the Wellington corpora of spoken and written New Zealand English (WSC/ WWC) and the Zürich corpus of early English newspaper language (ZEN) have already delivered exciting insights into language use in specific contexts. The same is true of specialised corpora of academic speech or writing, such as the Michigan Corpus of Academic Spoken English (MICASE), the British Academic Spoken English (BASE) corpus, the corpus of English as a Lingua Franca in Academic Settings (ELFA), the TOEFL 2000 Spoken and Written Academic Language Corpus (T2K-SWAL) and Hyland’s research article corpus. Specialised corpora such as these offer clear advantages over the sometimes haphazardly constructed text collections seen in many recent studies of academic discourse. 21 While specialised corpora may sometimes seem arcane in the specificity of their sampled populations, they promise reliable results from an 20 For example, although there was no doubt a countless number of reviews published in 2004, there was not a countless number of reviews published for one particular monograph, or in one particular journal, or by one particular author. 21 Specialised corpora also have clear advantages over the non-specialised mega-corpora, such as the Collins COBUILD or the World Wide Web, towards which some parts of the corpus linguistic community are currently moving. <?page no="54"?> 54 achievable study design. They complete a small part of the puzzle in an empirically-sound fashion, and this is ultimately superior to studies which aim far higher, but deliver potentially misleading results. The third issue that must be considered with regard to studies of academic discourse is the generalisability of the sample. In most recent contrastive studies, there is a marked discrepancy between the huge general trends or cultural differences posited and the nature of the supporting data. Peterlin (2005) for example aims to characterise English and Slovene academic writing based on 32 research articles, 16 in each language, taken from mathematics and archaeology. No justification is given for the choice of disciplines, and it is not explained why the two disciplines chosen are thought to be generalisable to academic discourse as a whole in the two languages. Likewise Buhl (1999), who tests hypotheses and makes claims about Englishand German-language academic discourse based on one British and one German author each. It is doubtful even whether these findings are generalisable to the two authors, since only one work (by the Briton) and two (by the German) were examined - regardless, they are definitely not generalisable to an entire academic style. A more positive example can be found in Adamzik and Pieth (1999), who examined prefaces to ten French and ten German histories of literature: this may not be a large sample numerically, but it is probably generalisable, since it represents a representative selection of the total number of this particular text type published in France and Germany in the chosen time period. Of course, the issue of generalisability is closely related to the first and second issues discussed above, and the solution is the same: research would be sounder if it were based on specialised corpora, which represent a reasonably-sized and representative sample of the total population. The fourth difficulty is that in many studies of academic discourse the factors are not operationalised in a quantifiable and falsifiable way. There is often a tremendous gap between the isolated phenomena analysed and what they are taken to represent. By way of example, Figure 2.1 lists some of the myriad categorisation schemes for Englishand German-language academic discourse; the same point could, however, have been made equally well by examining studies of another language pair. <?page no="55"?> 55 Figure 2.1. Various ways of categorising Englishand German-language (academic) discourse ______________________________________________________________________ Study English German ______________________________________________________________________ Galtung 1981 proposition-oriented theory-oriented Galtung 1981; Clyne 1987 linear digressive Edmondson et al. 1984; relationship-oriented content-oriented Kotthoff 1989 Clyne 1987, 1991 symmetrical asymmetrical Clyne 1987 deductive inductive Clyne 1987; Baumann 1998 writer-responsible reader-responsible Clyne 1991 reader-oriented author-oriented Clyne 1993 form-oriented content-oriented Gerzymisch-Arbogast 1997b; reader-oriented content-oriented Buhl 1999; Hutz 2001 Duszak 1997 expository contemplative personal impersonal dialogic / interactive monologic / less interactive House 1999 indirect direct addressee oriented ‘I’ oriented addressee orientation content orientation implicitness explicitness linguistic routines ad-hoc formulations Busch-Lauer 2001 highly impersonal personal turning toward the author turning away from the author ______________________________________________________________________ The problem is that such theories are usually built upon the analysis of very few phenomena, and these phenomena often do not directly support the posited theory. In the case of Galtung, of course, the theory is based on no analysis at all. Nonetheless, Duszak (1997) repeats Galtung’s claims verbatim, treating them as verifiable fact, without offering any additional supporting evidence, as does Clyne (1987; 1994). Hutz (2001) characterises German academic discourse as content-oriented, because the German authors he examined used fewer personal pronouns and marginally more intensifiers when offering criticism in book reviews. This may show that the German authors Hutz studied were perhaps less explicitly personal, or more directly confrontative, in their criticism, though of course their criticism is just as personal all the same. However, the link the researcher then draws to content orientation is not empirically justified. Baumann (1998) too maintains that German academic culture is contentoriented, in this case based on the observation that - in the three texts he analysed - the German authors he studied more often referred to <?page no="56"?> 56 “wissenschaftlich anerkannte Autoritäten” [academically-recognised authorities] than did the English-speaking authors (1998: 311-312). Yet Baumann does not explain how these “authorities” were defined, and the connection between reference to “authorities” and content-orientation is obscure. Gerzymisch-Arbogast (1997b), Buhl (1999) and Hutz (2001) characterise English, but not German, as “leserorientiert” [readeroriented], which seems problematic, since in a philosophical sense at least all communication is oriented towards the reader or listener (with the possible exception of those who really do talk to themselves). Buhl (1999) for example categorises texts beginning with known information, an example, or a humorous lead-in as reader-oriented, while a text begin containing new information is not reader-oriented (1999: 127). Personal comments, asides and examples are also taken to be reader-oriented (1999: 130), though one could equally well regard personal elements as author-oriented. Assuming that a reader dislikes jokes and personal asides, and wishes to read an article which starts with interesting new information rather than old material, an author who caters to such a reader’s wishes would therefore count as not reader-oriented. This makes no sense, and the fault lies in positing a static typology, which assigns one of two mutually-exclusive communicative purposes to each linguistic feature. Clearly, the factors have not been operationalised in a sound fashion: reader-orientation cannot be measured in this way. What these authors mean perhaps is that German academic discourse is oriented towards a different kind of reader, or oriented differently. Clyne’s (1987) characterisation of English academic discourse is also questionable. While a symmetrical, hierarchical textual macrostructure can be taken as indicative of a certain linearity in argument progression, it is not clear for example how increased use of advance organisers, or the definition of technical terms, by the English-speaking scholars in Clyne’s study contributes to ‘linearity’. Similar problems with operationalisation recur in much research on academic discourse. Research into academic discourse has long been broadly criticised for its lack of theory (Mauranen 2001: 43; see also Houghton & Hoey 1983; Graefen 1994: 138). In a sense, the discrepancy between theoretical characterisation and the phenomena analysed is evidence of an attempt to address this. However, empirical research that propounds theories just for the sake of having a theory, when this theory is not supported by the findings, cannot be the solution. The assignation of specific findings as supporting evidence for a particular theory sometimes seems almost random. For example, though Hutz (1997) and Busch-Lauer (2001) reach exactly contrary findings in their studies, nonetheless the two researchers present essentially the same characterisation of German academic discourse. It seems that the results are squeezed to fit a pre-formulated theory. While a <?page no="57"?> 57 neat set of one-word distinguishing epithets is understandably appealing when trying to characterise and contrast something as complicated as culture, characterisations formulated using such a method are deceptive. Reality is more complex, and not served by reductionistic dichotomous summaries. The fifth methodological issue in research on academic discourse concerns the treatment of variables. In a large number of studies, there is little awareness of the influence of variables other than the one being analysed - in this case usually culture - on linguistic production. For this reason, other potentially influential intervening variables are regularly left unaccounted for. Such laxity makes it impossible to discern whether results are truly due to the variable(s) ostensibly under consideration, or rather assignable to any number of other intervening factors. In contrast, even the earliest corpus linguists took care to balance their text collections according to variables such as text type, topic and composition date, and author native language. These are just some of the variables accounted for in firstgeneration computerised corpora such as the one million word Brown corpus of written US-American English texts, which was published in 1961. With such a data basis, it is possible to compile subsets of the data to test hypotheses about the influence of different variables on specific linguistic phenomena: for example, one can compare speeches given by men to those given by women (to test gender influence), and speeches given by women to articles written by women (to test text type influence). As we saw above, however, recent studies of academic discourse examine a very small number of texts spanning composition periods of up to 65 years (Buhl 1999), or do not control their data for author native language as well as language of publication (Oldenburg 1992; Hutz 2001; Fandrych & Graefen 2002; Martín 2003). These text collections are valid only if neither composition date nor author native language have any influence on textual production, which is not the case. Research into academic discourse would therefore benefit from adopting some of the more systematic sampling methods generally applied in corpus linguistics. In fairness, I will point out that the treatment of variables is lax even in some recent corpus linguistic studies (see footnote 7). Clearly, there is much to be improved on this front. The sixth and final methodological criticism of research on academic discourse is that the findings - even when supported by the data - are not subjected to appropriate statistical analysis. Despite the fact that many researchers try to compensate for inadequate data samples by concentrating on qualitative analysis, few can resist the temptation to present numbers nonetheless, and try to draw conclusions from them. Without even a basic knowledge of statistics, however, these conclusions are quite likely wrong, <?page no="58"?> 58 because the discerned ‘trends’ may well be nothing more than random variation due to chance. Some studies compare cumulative percentages across subcorpora, i.e. without case weighting. A cumulative percentage measure presupposes that there is an equal number of possible occurrences in each subcorpus. If, however, - and this is the case in most studies - the subcorpora differ in size, this condition is not met, and the results are therefore spurious (Sachtleber 1990; Dahl 2004). 22 There are also studies in which results are presented as absolute numbers, and contrasted across subcorpora of differing sizes (Swales, Ahmad, Chang, Chavez, Dressen & Seymour 1998; Graefen 2000b; Fandrych 2001; Fandrych & Graefen 2002; Samraj 2002; Martín 2003). Another point to consider is that research on academic discourse is essentially contrastive, comparing different cultures, disciplines and text types (Hoffmann 1992). Yet investigators do not employ appropriate statistical measures to test whether the tendencies they discuss are statistically significant, or are more likely to be due to chance. One admirable exception, however, is the study by Trumpp (1998), which skilfully applies a variety of statistical analyses. As the findings of this study will show, the human eye is a poor judge of significant results. Depending on the distribution of a particular sample (Gaussian or non-Gaussian, paired or unpaired and so on), there is a wide variety of different tests for significance (Mann Whitney, Kruskal Wallis, Chi Square etc). There is also no way of establishing or measuring correlation between different independent variables without using statistical measures such as Pearson Rho, the Phi Coefficient or Spearman Rho (some common measures of correlation for interval, dichotomous and ordinal scales). Researchers must know their data and choose the appropriate test for their study. Statistical analysis, properly applied, is indispensable. Crismore (2004) in fact does test her findings for significance, but then ignores the results of the tests, stating that “the findings, although not statistically significant, are nonetheless useful and important for pedagogical and practical reasons” (2004: 317). There follow 11 pages of discussion of the non-significant ‘differences’ between subcorpora. The method here is not sound. If the differences are not significant, it means precisely that: the results are of no relevance to anything, because they are most likely random patterns due to chance. It is therefore vital that researchers not just apply, but also understand basic statistical methods, and are able interpret test results. The absence and improper use of basic statistical methods is a crucial defect in previous research, since it fundamentally undermines the credibility of the findings, and the conclusions based upon them. 22 It is likely that many more studies make this mistake, but this can only be discerned where the researcher has revealed total word counts for individual subcorpora. <?page no="59"?> 59 Admittedly, lack of statistical knowledge can also be a problem in some corpus linguistic research. However, especially in newer corpus linguistic research, lack of statistical knowledge is neither as frequent nor as fundamental a problem as it is in academic discourse studies. Much recent corpus linguistic work in fact makes use of quite sophisticated statistical tests in order to demonstrate tendencies and relationships in large data samples (Biber, Conrad & Reppen 1998; Hundt 1998; Sigley 2003; Gries & Stefanowitsch 2004; Szmrecsanyi 2005; Hoffmann 2005). Again, research into academic discourse would do well to take a fresh leaf out of the corpus linguistic book. In conclusion, the research that has been carried out into academic discourse is marred by a number of serious methodological defects. These concern both the sampling methods (insufficient data size, nonrepresentative and non-generalisable samples, influential variables disregarded) and the general methodology (factors unsoundly operationalised, statistical analysis absent). The situation could be much improved if recent studies were not just to refer to their data as corpora, but strive for an understanding and application of corpus linguistic methodology. The fact is that even now, nearly forty years after the 1966 article by Kaplan which is often counted as its inception, research into academic discourse still lacks a cohesive sense of itself as a discipline, and a methodology of its own. The latter problem at least could be tempered by adopting some of the methods applied in corpus linguistics. 2.3 Combining corpus and discourse The combination of corpus linguistic with discourse analytic methodology is unusual. In the past, corpus linguists have not been particularly interested in discourse, preferring to concentrate on lexical and morphosyntactic analysis. Similarly, discourse analysts have seldom worked with corpora, preferring methods such as introspection, elicitation and the unsystematic collection of anecdotal evidence. The divide between corpus and discourse therefore runs both ways: corpus based studies are generally not concerned with discourse (McEnery & Wilson 1996, 2001) and discourse studies are “not typically corpus-based” (Biber, Conrad & Reppen 1998: 106). Aside from the traditional antipathy which has developed between these two approaches, there are also practical reasons for the ostensible incompatibility. The first corpora were compiled by researchers primarily interested in lexical and syntactic analysis: it therefore makes sense that neither the corpora nor the methodological tools they devised are <?page no="60"?> 60 particularly fitted to discourse analysis. This combination of lack of interest on the one hand, and unsuitable data format and instruments on the other, has tended to reinforce itself up to the present day, to the detriment of both fields of study. Given this background, it is not surprising that, in research into academic discourse, there has so far been both a perceived and a real conflict between the bottom-up methodologies dictated by corpus linguistic concordancing tools and the top-down approaches more characteristic of discourse and genre analysis (Swales 2002b; J. Flowerdew 2002b). There are several reasons why the largest corpora freely available at present, such as the BNC, ICE, FLOB and FROWN, are not ideal for discourse research. One problem is that, though they contain substantial samples of a large range of text types, these corpora are generally not capable of providing the tightly-controlled data set that would be required for empirically-sound studies of (academic) discourse. 23 In the BNC, for example, many extra-linguistic variables such as gender are not accounted for. For example, there is not a single female author represented in the entire natural sciences subcorpus of the BNC. In addition, corpora such as FLOB and FROWN do not respect or preserve textual integrity, but consist of 2,000 word samples of longer texts. From a discourse perspective, it is neither justifiable nor worthwhile to work with text fragments chosen randomly from some section of a research article or university lecture. In standard large corpora, there is also often little control over text type, intended audience or topic: texts are often taken from vaguely-defined, even ambiguous, genres that are not linguistically identifiable, and that are incompatible with the concepts of text type or genre as understood in discourse research. The Brown and LOB corpora, for example, contain the text categories “skills and hobbies”, “religious writing” and even “miscellaneous”. 24 In the BNC, one of the subcategories of the science genre is “gardening”, which is misleading, because gardening manuals do not constitute a scientific genre in the academic discourse sense. This situation creates immense problems for researchers interested in genre analysis or concerned to control their data for the influence of text type and/ or topic. These comments apply to most corpora freely available today, such as the BNC, the ICE project, FLOB, LOB, FROWN and BROWN, and the 23 Of course, as we saw above, many discourse studies have not yet been capable of providing this either, but that is another issue. 24 The FLOB categories, for example, are: Press: Reportage, Press: Editorial, Press: Review, Religion, Skills, Trades and Hobbies, Popular Lore, Belles Lettres, Biographies, Essays, Miscellaneous, Science, General Fiction, Mystery and Detective Fiction, Science Fiction, Adventure and Western, Romance and Love Story, Humour. No wonder then that Mair and Hundt call these “admittedly rough-and-ready textual categories“ (1999: 224). <?page no="61"?> 61 traditional disregard for discourse-oriented concerns when constructing corpora goes some way towards explaining why corpora are not generally used for discourse analysis. For discourse research purposes, many of the corpora currently in existence provide insufficient contextual information about the participants or circumstances surrounding the interaction recorded in the corpus. This frequent lack of contextual information allows Swales to emphasise, somewhat condescendingly, that “the genre analyst inhabits a more integrated discoursal world orthogonal to the enthusiastic but fragmented efforts of the corporist” (2002b: 152). This is, however, at most a criticism of particular corpora, not of the corpus linguistic method per se. A more serious criticism of corpus linguistic methods of analysis is that they “do not consider the socio-cultural context as they deal with decontextualised corpus data” (L. Flowerdew 2005: 321; see also Aston 1995; Widdowson 1998; Hunston 2002). This is true up to a point, in that corpora generally cannot retain the original visual and socio-cultural context of the texts sampled: the texts are no longer in situ, but become part of a corpus. Some level of decontextualisation is necessary if the texts are to be electronically collected and analysed in large numbers. However, it is possible to annotate corpora for extra-linguistic information, or select texts from a single well-defined context (specialised corpora). Researchers can also compensate for the loss of the original context with their own knowledge of the social, cultural, communicative and pragmatic context of the texts they study, and/ or by consulting specialist informants. In cases where researchers compile their own corpus, Flowerdew suggests that the “compiler-cum-analyst can therefore act as a kind of mediating ethnographic specialist informant to shed light on the corpus data” (L. Flowerdew 2005: 329). Though it is true that many currently-available corpora do not contain sufficient extra-linguistic information for discourse analysis, these difficulties can be and are addressed by the creation of the specialised corpora such as those mentioned in the previous section. Many of these corpora have been created specifically for discourse research, and therefore contain extensive contextual information. Another major reason why corpus and discourse are seldom combined is that corpus linguists generally rely on automated methods of analysis (concordancing software), which are largely unsuited to the kind of highlevel, context-dependent textual investigation most discourse phenomena require. In the case of evaluation, for example, Hunston (2004) notes that while manually examining a single text is rewarding, it risks being “hopelessly subjective” and does “not lead to a set of items which could be identified automatically in a large corpus” (2004: 185). Conversely, while the automatic identification and quantification of evaluative meaning in large <?page no="62"?> 62 corpora is possible, this method captures “only a limited set of realizations of evaluation” (2004: 186). This dilemma is common to analyses of most discourse phenomena: many cannot be identified solely on the basis of surface lexical or syntactic features (assuming one is lucky enough to work with a tagged corpus). The computerised analysis of discourse phenomena in (tagged) corpora is therefore at present mostly restricted to features which have lexical (or syntactic) indicators, such as back-channel devices, reaction signals, and discourse management devices such as tag questions and markers (Mandala 1999). Other major discourse phenomena, such as politeness or hedging, cannot be reduced to a closed set of automatically identifiable lexico-grammatical patterns. Conrad (2002: 77) suggests that researchers write their own specialised software for analysing discourse, but there cannot be many linguists with sufficient computer programming skills to achieve this. Even if linguists were to double as software engineers, the fundamental problem of identifying discourse features based upon surface characteristics remains. Discourse or pragmatic tagging may offer a solution for some phenomena, but it is immensely complex and there is as yet no widely-accepted or standardized form of annotation (Leech 1991: 25; Mandala 1999). Despite their own efforts in the field, Garside et al. describe discourse annotation as “at a fairly immature stage of development” (Garside, Fligelstone & Botley 1997: 83). A major problem with discourse tagging, as pointed out by McEnery and Wilson (1996: 52), is that the function of many rhetorical features depends on the context. This makes the automatic assignation of discourse tags extremely difficult. Nonetheless, there are some discourse-tagged corpora already in existence, such as the Santa Barbara Corpus of Spoken American English (SBCSAE), the Penn Discourse TreeBank (PDTB), the RST Discourse Treebank and the Cambridge and Nottingham Corpus of Discourse in English (CANCODE). However, the tagging systems, and tagged phenomena, in these and other discourse-tagged corpora are not comparable: while the SBCSAE is tagged for turns, intonation units and intonation features, the Penn Discourse TreeBank encodes coherence relations associated with discourse connectives, the RST Discourse Treebank is annotated according to Rhetorical Structure Theory, and CANCODE is tagged for the relationship between the speakers. Much work continues on discourse tagging, for example in projects such as the PDTB, the Discourse Resource Initiative, MATE and the Global Document Annotation initiative. 25 At present, however, it seems that some level of manual analysis will nearly always be necessary when studying discourse through corpora. 25 These projects are largely computational linguistic in origin and commercial in focus, and are concerned with possible applications in machine translation, information retrieval, information filtering, data mining, consultation and expert systems. <?page no="63"?> 63 The potential of corpus-based approaches to discourse is increasingly recognised, and two recent edited collections bring together a number of suggestions for bridging the gap between corpus and discourse (Connor & Upton 2004; Partington, Haarman & Morley 2004). Some of the papers in these two volumes expertly combine corpus and discourse methods, thereby demonstrating some of the insights a combined approach can offer (see for example Haarman, Hunston, Stenström and Biscetti in Partington et al. 2004, and Simpson, Hyland, Cheng, and Connor and Gladkov in Connor & Upton 2004). Many others, however, particularly in Partington et al. 2004, are methodologically unsophisticated, and corpus linguistic only in the most superficial sense, in that the examples discussed are naturally-occurring rather than made-up (see discussion in Sanderson 2006b). Familiar problems such as absence or inappropriate use of statistics, unrepresentative and non-generalisable samples and unsound operationalisation also recur. It is no coincidence that the better work here has generally come from corpus linguists with an interest in discourse, rather than from discourse analysts working with so-called ‘corpora’. 26 At present, corpus linguistic methods appear too little known among researchers working primarily in other fields. However, a symbiosis of corpus and discourse can be achieved only if researchers possess thorough knowledge of both methodologies. Recent studies, for example Fløttum, Dahl and Kinn (2006), Hyland and Bondi (2006), and Ädel and Reppen (forthc.), augurs well for continued progress in this area. If the combination of corpus and discourse is to be practicable and worthwhile, it is clear that neither the fully-automated analysis methods of the corpus linguist nor the exclusively manual methods of the discourse analyst are sufficient. At their worst, corpus linguistic approaches can be criticised for a microptic and blinkered pre-occupation with the lowest levels of linguistic performance. Similarly, at their worst discourse analytic approaches can be criticised for making fascinating yet all too sweeping generalisations, often based on insufficient empirical study, and/ or for positing intuitive and speculative tendencies or hypotheses, which have no clear textual manifestation, and are therefore not falsifiable. Both types of problems can be seen in contrastive studies of academic discourse, yet it is 26 There are a number of recent studies of academic discourse which describe their text collections as ‘corpora’ (Sachtleber 1990; Baumann 1998; Graefen 2000b; Fandrych 2001; Fandrych & Graefen 2002; Fandrych 2005). However, these ‘corpora’ are very different from that which most corpus linguists would accept, because they are unsystematic, not representative of the cultures and genres they attempt to characterise, and posit far-reaching conclusions which cannot be substantiated by the data. Partington comments diplomatically that “simply employing a corpus in one’s research does not necessarily make it a study in Corpus Linguistics” (Partington 2004: 12). <?page no="64"?> 64 possible to avoid them by moving away from an either/ or conception of appropriate methodologies and letting each approach moderate the other. Ideally, the combination of corpus and discourse would help to balance the weaknesses of both methods. Corpus linguistics could offer discourse analysis an empirically-sound methodology that is based on analysing objectively observable and quantifiable phenomena and posits theories which can be independently verified. Research into academic discourse has much to gain from such an approach. At the same time, the tendency of corpus linguistic research to become sometimes bogged down in bean-counting, ignoring communicative context and purpose, could receive an apt corrective from the inclusion of discourse perspectives. Perhaps the question is not so much how a corpus-discourse methodology can resolve the conflict between bottom-up and top-down approaches, but rather how such an approach can turn this tension into an advantage, by constantly negotiating the best methodology for examining each individual phenomenon in a particular study. This is similar to what Oldenburg called for in 1992, when he noted that since none of the textlinguistic or LSP approaches under discussion encompassed all levels of a text, in practice it would be necessary to select the approach best suited to the specific purpose of the study, or to develop one based on those already available (1992: 36-7). The corpus-based discourse researcher will therefore need to be flexible: each phenomenon will have to be analysed in a different manner, with a unique combination of quantitative and qualitative methodologies, in order to investigate optimally all aspects of its frequency and use. The potential benefits are enormous. 2.4 Aims of the present study The present study aims to combine discourse analytic with corpus linguistic methodology in order to reveal some salient features of humanities research articles, taken from five disciplines, written by native-speaker British, US-American and German academics. The underlying objective is to examine the possible influence of variables such as cultural background, discipline, gender and academic status on selected aspects of humanities research writing in a systematic fashion. The study aims to develop and implement a methodology that is empirically sound, since this is the major weakness of previous research. The research does not attempt to provide an exhaustive overview of all differences between academic writing styles in the five disciplines and three cultures examined. Nor does it claim to catalogue all disciplinary and cultural differences in the texts examined. The results are not intended to be representative of all academics, nor of all <?page no="65"?> 65 research genres, from these five disciplines and three cultures. It is hoped that the results will be of intrinsic interest and may also help inform specialist pedagogy, in particular for English NNS academic writers. The study concentrates on the ways in which humanities academics from Great Britain, the USA and Germany construct multifaceted and multilayered identities, and how they relate to their imagined readers, and to their own writing, in research articles. In a genre traditionally regarded as objective and therefore impersonal, social interaction is a particularly interesting object of study. The conception of writing as an inherently social activity, which underlies this study, informs a large and increasing number of studies of academic discourse (Hunston 1994, 2005; Duszak 1997; Ivani 1998; Myers 1999; Hyland 2000, 2005; Thompson 2001; Flowerdew 2002c; Mauranen 2003; Dahl 2004; Fortanet 2004; Ädel 2005; Harwood 2005a, 2005b; Tognini-Bonelli & Camiciotti 2005). The approach adopted here considers how academic texts reflect, encode and construct a complex network of social identities, relations and interactions. In order to analyse such phenomena, two features were selected: person reference and text comments. The analysis is based on a corpus designed and compiled specifically for this purpose by the author. First, then, an account of the corpus. 2.5 The SCEGAD corpus The Synchronic Corpus of English and German Academic Discourse (SCEGAD) was compiled by the author at the University of Freiburg in 2001 - 2003. The corpus was created to allow systematic contrastive investigation of specific features of native-speaker language use in English and German-language written academic texts. SCEGAD comprises 100 published humanities research articles (whole texts) taken from five humanities disciplines. It consists of three parallel parts: British English (25 texts), US-American English (25 texts) and Standard German (50 texts), making in total approximately one million words. The corpus therefore allows not only interlingual (German/ English), but also intralingual (British/ US-American English) and intercultural (German/ British/ US- American) comparisons to be made. It should be noted from the outset that, in this study, ‘British’ refers specifically to academics from Great Britain only (England, Wales and Scotland), not from Northern Ireland. I use the term ‘US-American’ to indicate that these are authors only from the USA, and not from Canada, Mexico or anywhere else in North or South America. Similarly, ‘German’ refers to authors from the Federal Republic of Germany, and not German- <?page no="66"?> 66 speaking academics from Austria, Luxemburg and Switzerland etc. 27 The reason for restricting the study to authors from Great Britain, the USA and Germany lies in the potential for cultural differences between academics within language groups. The possibility that cultural differences exist within a single language group is sadly ignored in most contrastive studies, which follow Galtung (1981) in talking for example of ‘Anglo-American’ discourse norms (Taylor & Chen 1991; Mauranen 1993b; Duszak 1997; Baumann 1998; Zhabotynska 2001; Fandrych & Graefen 2002; Giannoni 2002; Bassler 2003). Although intra-lingual differences are not the primary focus here, through systematic comparison of texts by British and US- American authors, the present study shows that the assumption of a common ‘Anglo-American’ norm for academic writing is a mistake. The very term is a misnomer. The same might well be true for different native varieties of German, as Clyne (1987) suggests, though this was not examined in the present study. Every single author in the corpus is a native-speaker of the variety in which the text is written. Thus instead of simply assuming that only nativespeakers publish in corresponding language journals (Giannoni 2002; Suau 2004), or guessing that surnames indicate native language (as for example in Taylor & Chen 1991 and apparently also in Oldenburg 1992; Fandrych & Graefen 2002 and Dahl 2004), the researchers’ native languages were verified through personal correspondence with the SCEGAD authors. 28 As we saw in sections 2.1 and 2.2, studies which do not secure such information risk comparing native-speakers of one variety with non-native speakers of another, or subsuming foreign-language writers in a group of ostensible native-speakers who are taken to represent one particular culture only. In addition, the SCEGAD authors are not only all native-speakers of the language they write in, but were all formally educated (school, university and graduate study) and have worked at universities solely in their home country. The aim here was to minimise the possibility of cultural 27 The inclusion of academics from countries where bilingualism or multilingualism is national policy, such as Canada or Switzerland, would have been particularly problematic: in a study of elite bilinguals such as academics in these countries, it would have been impossible to assume that these scholars are typical of any one (academic) culture. 28 Other studies simply give up: “the originally intended subdivision […] according to the cultural and linguistic backgrounds of the review authors could not be carried out, as in most cases it was impossible to determine whether the writers were native speakers of English and which regional variety of English they were using” (Römer 2005: 112). As the present study shows, such subdivision may be time-consuming, but it is not impossible, and is certainly worthwhile. See also footnote 7 above. <?page no="67"?> 67 ‘contamination’ through extensive contact with, or training in, another academic culture. 29 In addition to being balanced for the native language of the authors, the corpus is also controlled for gender, age and academic status. This information too was obtained through personal correspondence with the SCEGAD authors. Wherever possible, within each discipline, the corpus contains an equal number of articles written by male and female scholars, and a spread of ages and career status levels (both tenured and untenured). The academics represented in the SCEGAD corpus range in age from 28 to 75, and in academic status from junior lecturers/ pre-PhD scholars to emeritus professors. 30 In the UK and USA, the emeritus professor is the most venerable position in the academic hierarchy: it is a title awarded to some particularly distinguished professors upon their retirement from fulltime work. In Germany, the emeritus title is awarded to any and all Ordinarien [full professors with planned positions] appointed before 01.01.1980: upon retirement, these academics continue on full pay rather than a pension, and may continue to teach and/ or carry out research if they wish or if this is required by the university. SCEGAD therefore provides samples of writing by a broad cross-section of expert academic authors, spanning the entire professional academic career from junior lecturer to emeritus professor, meaning that its conclusions are more likely to be representative of this diverse discourse community as a whole. Based on these carefully-selected data, the study can examine the effect of many extra-linguistic variables - not just culture - on selected aspects of academic writing. Since the texts are coded for these extra-linguistic independent variables in addition to the dependent variable(s) under investigation, the corpus allows differences to be investigated, for example between authors of different academic status, even in those cases where it was not possible to obtain articles by exactly the same proportion of pre-PhD and post-PhD scholars, full professors and emeriti. While the ‘typical’ British, US-American or German humanities academic exists only as a theoretical 29 For a subsequent research project, it is intended to expand the corpus with the addition of pre-submission versions of an additional 50 research articles intended for journal publication, which have been written in the English language by German scholars. The systematic comparison of native speaker use, such as occurs in this study, of course constitutes a necessary first step for any such later comparison. In theory, for specialist DaF (German as a Foreign Language) teaching, it would also be interesting to expand the corpus to include German-language articles written by English-speaking academics, though it is somewhat doubtful whether sufficient texts could be gathered for such an analysis. 30 Academics of any lower status (i.e. undergraduates) were not found to publish in the leading journals selected for SCEGAD. <?page no="68"?> 68 construct, it is hoped nonetheless that these considerations will maximise the representativeness of the texts sampled. Figure 2.2. SCEGAD composition according to variables ______________________________________________________________________ Variable Number of corpus texts (total 100) ______________________________________________________________________ Author native-culture British 25 US-American 25 German 50 Discipline Philosophy 20 History 20 Folkore 20 English / German literary studies 20 English / German linguistics 20 Author gender Male (single authorship) 59 Female (single authorship) 39 Mixed (multiple-authorship) 2 Author age Under 30 2 30-39 47 40-49 24 50-59 15 60-69 4 70 and over 5 Multiple/ unknown 3 Author academic status Pre-PhD 2 Post-PhD 59 Full professor 26 Emeritus professor 10 Multiple/ unknown 3 ______________________________________________________________________ The corpus is synchronic: the vast majority of texts were published in the years 2001-2004, though there is also a small number of texts from 1996- 2000 in cases where technical difficulties (e-journal licensing agreements) precluded the inclusion of the most recent publications. As far as the teaching implications are concerned, I felt that a synchronic study would be more immediately useful, since LSP practitioners aim to teach the language as it is now, not as it was, and it is largely for this reason that <?page no="69"?> 69 SCEGAD contains only contemporary material. 31 No doubt a similar study with a diachronic focus would be fascinating, especially since significant developments in preferred styles of academic writing seem to have occurred over the last fifty years, particularly for German academic discourse (Wüest 1988; Gläser 1990; Olszewska 2003). However, this task must be left to another occasion and another study. The articles selected for SCEGAD were published in established scholarly journals. By ‘established’ I mean journals which have a publication history of at least fifty years, or in many cases over 100 years, and which are published by well-recognised academic publishing houses such as the CUP, OUP, de Gruyter, Blackwell, Routledge, Klostermann, R. Oldenbourg Verlag, Universitätsverlag Winter and Sage Publications, or by venerable learned societies such as the American Historical Association (founded in 1884) or the American Folklore Society (founded in 1888). The journals all claimed to be either a leading or even the leading publication in their field; this claim was tested where possible using the ISI Journal Citation Reports. In order to counteract the possibility that journal editors from a different cultural background might have substantially influenced the form and content of the texts, only articles written by British, US-American and German academics publishing in journals from their own home country were selected. Articles written in British or US- American journals by German academics, or articles written in German journals by English NS, were therefore excluded. Only monolingual journals were considered, thereby excluding some German journals which have lately begun to publish also in English, since these are perhaps less good representatives of culturally specific discourse traditions. 32 A list of the corpus journals is shown in Figure 2.3. 31 This is not to say that diachronic studies have no place in pedagogy: it can often be enlightening for students to know the background of what they learn, and it may even help them learn more effectively. However, diachronic findings do not provide the actual information which LSP learners wish to acquire. 32 Swales (2004b: 33-4) records that this trend towards English-language publication is not confined to German journals: French, Dutch, Swedish and Japanese journals too have changed their editorial policy to allow or even require English-medium publication. This trend is by no means new, as Lippert (1978) showed for German medical periodicals. <?page no="70"?> 70 Figure 2.3. Journals in the SCEGAD corpus ______________________________________________________________________ Discipline Journal ______________________________________________________________________ British US-American German ______________________________________________________________________ Philosophy Mind Journal of Philosophy Zeitschrift für philosophische Forschung History History The American Historical Historische Zeitschrift Review Folklore Folklore Journal of American Zeitschrift für Volkskunde Folklore Literary Essays in Journal of American Euphorion Studies Criticism Literature Linguistics Journal of Journal of English Zeitschrift für Linguistics Linguistics germanistische Linguistik ______________________________________________________________________ The corpus has a humanities focus. A major reason for the decision to focus on humanities texts rather than the natural sciences for example was that, as a large-scale survey by Stickel (2000: 131-2) and a smaller-scale study by Schiewe (1991: 23-4) revealed, the humanities is one of the few research areas where German-speaking academics still publish primarily in German rather than in English. Humanities texts also seemed more suited to an analysis of identity construction and social interaction in academic texts: in a mixed-discipline corpus, Hyland for example noted that the ‘soft sciences’ texts, taken from philosophy, marketing, sociology and linguistics, contained some 75% more interaction markers than did the engineering and science papers in his corpus (2005: 186-7). Over 80% of reader pronouns (‘you’, ‘your’ and ‘we’) for example were found in the soft sciences corpus texts (Hyland 2005: 188; see also similar findings in Lafuente 2005). It is also the case that, as Flowerdew notes, Arts subjects are not well represented in the EAP literature (L. Flowerdew 2002: 111); the current study is therefore a contribution towards redressing this imbalance. Finally, it is possible that cultural and disciplinary traditions and tendencies are more evident in humanities texts than they might be in natural sciences articles. Duszak suggests that textual conventions in the humanities are less formalised and internationalised than those in the natural sciences: “If experimental sciences are prone to show more similarities in textualization patterns, writings in the humanities and social sciences evidence more prominent variation” (1997: 11), and Ammon <?page no="71"?> 71 comments that specialist writing in the humanities is “weniger formalisiert” [less formalised] than in the natural sciences (1993: 21). Hyland too refers to the “highly standardised code” in which arguments are formulated in the hard sciences (2005: 189). It is worth mentioning, however, that a cross-cultural study by Taylor and Chen (1991) in fact noted considerable disciplinary variation across natural and physical sciences research articles, though there was apparently very little cultural variation. More work is needed in order to establish whether natural sciences texts really are so uniform across cultures and disciplines. However, for the purposes of this study, it is sufficient to note only that humanities texts evince great textual variation. It therefore seemed that humanities texts were perhaps better suited to an analysis of textual variation across cultures and disciplines than natural sciences texts, or a mixed corpus of texts, might have been. One British, one US-American and one German journal was therefore chosen from each of the following five humanities disciplines: history, philosophy, literature studies (English literature for the English-language texts, German literature for the German-language texts), folklore (in German Volkskunde) and Linguistics (English linguistics for the Englishlanguage texts, German linguistics for the German-language texts). Care was taken to ensure that the journals selected publish articles from all aspects of each discipline, and do not represent the production of only a particular sub-discipline or school. The journals are published in similar intervals, in most cases quarterly. In an attempt to represent different aspects of each discipline, articles were also selected from a wide variety of sub-disciplines. For example, in the history subcorpus, the sampled subdisciplines were ancient history, mediaeval and renaissance history, 19th and 20th century history, European, Asian and US-American history, African-American history, political history, intellectual history and women’s/ gender history. These categories are of course not mutually exclusive, and some texts spanned more than one subdiscipline. 33 In my own field, represented by the linguistics subcorpus, the subdisciplines covered were generative linguistics, applied linguistics, historical linguistics, sociolinguistics and clinical linguistics. By restricting the SCEGAD corpus to a relatively homogeneous selection of disciplines and controlling the corpus for sub-disciplines, the study aims to achieve a balance between size and representativeness. In a corpus of this size, there can be no question of an attempt to represent all disciplines, or even all humanities disciplines. However, it is both possible and highly important 33 An article on 19th century Japanese imperialist expansion in Taiwan, for example, could be counted as belonging to (East)-Asian history, 19th century history and political history. <?page no="72"?> 72 to sample a reasonable selection of disciplines within the humanities, and a reasonable selection of subfields within each discipline, in order to be more representative of the chosen academic subfield (humanities). The choice of disciplines was deliberate, and was motivated primarily by the necessity of finding high-quality research written by German academics in German, not in English. Thus folklore, history and philosophy are three disciplines with strong national traditions in Germany. In a survey of major German publishing houses, Ammon (1998: 176) named philosophy in particular as one of the so-called Nischenfächer [niche disciplines], that is, disciplines in which high-quality scientific work is still being conducted in the German language. According to the Peter Lang publishing house, philosophy is one of the very few disciplines in which non German-speaking countries still place substantial orders for German language monographs (Ammon 1998: 166-7); history too is a discipline in which high-quality German-language publication still takes place (Ammon 1998: 165-6). Skudlik classifies philosophy, history and in particular literary studies in Germany as “nicht anglophon-geprägte” [non anglophone-shaped] disciplines (1990: 213-6). This is not the case in the natural sciences, where, as Ammon has shown, German academics publish work they consider to be of high quality in English, not German (1998: 106). This is important, since it would not have been appropriate to compare the best-quality English-language research with second-rate Germanlanguage publications. Literary studies was selected since, as a pilot study indicated (Sanderson 2000), it is a discipline with much internal freedom, and therefore might usefully illustrate the extent of individual freedom within the confines of culture. Linguistics - according to Skudlik (1990: 213) an “anglophon-gesprägte” [anglophone-influenced] discipline in Germany - was chosen in order to test the hypothesis posited by House (2003a) that lingua franca English covertly influences the discourse norms of other languages. SCEGAD is a corpus tailor-made for discourse analysis. In contrast to most freely available corpora, which are intended primarily for morphosyntactic analysis, it therefore consists of whole texts rather than extracts, and contains extra-linguistic information on the study participants’ native language, cultural background, gender, age, academic status and educational background, as well as the date and place of publication, discipline and subdiscipline for each text. As noted above, this kind of additional information is absent, or only partially present, in most traditional corpora. For a satisfactory analysis of discourse phenomena, more than just a patchy idea of the communicative context is necessary. SCEGAD thus offers the usual advantages of a computer-based corpus, with its speedy and efficient access to and searching of large amounts of text, and in addition contains <?page no="73"?> 73 the kind of detailed information about the circumstances surrounding a communicative event which is usually the preserve of microlinguistic discourse studies. SCEGAD is also a specialised corpus: it is not intended to be representative of the totality of academic discourse in the English and German languages, but rather aims to sample a representative cross-section of high-prestige humanities writing in the major varieties of these two languages. The corpus is designed to provide a representative sample of this specialist population, and gather general insights on this subtype of academic discourse which may be applicable to the larger population of successful humanities researchers from Great Britain, the USA and Germany. In sum, the SCEGAD corpus differs from those employed in previous contrastive studies of academic discourse in four main respects. First, it consists of published texts written by expert native-speaker language-users for an expert audience, rather than of unpublished texts written by students and/ or non-native speakers, as has been the case in the vast majority of other studies (rare exceptions being Clyne 1987; Crismore, Markkanen & Steffensen 1993; Mauranen 1993b; Busch-Lauer 2001 and Fandrych & Graefen 2002). Second, the corpus has three parallel components containing texts written by German, British and US-American scholars, which permits both cross-linguistic (English/ German), cross-varietal (British/ US-American English) and cross-cultural (British/ US- American/ German) investigations to be carried out in a systematic fashion upon the same data. Third, the corpus is controlled and wherever possible also balanced for variables generally neglected in previous studies, such as text type, discipline and sub-discipline, as well as author native language, culture, gender, academic status and age, which means that the influence of these variables too can be investigated. Fourth, the humanities focus strikes a balance between size and representativity, thereby increasing the generalisability of the results, in contrast to other studies, which have often studied too small a sample of too large a population to enable any reliable or generalisable results. 2.6 Analysing social interaction and identity construction in academic texts This study investigates social interaction and identity construction in a corpus of academic texts drawn from five disciplines and written by scholars from Britain, the USA and Germany. The justification for focussing on these two related phenomena is that they appear to be key areas in <?page no="74"?> 74 which cultural and disciplinary tendencies manifest themselves, and therefore seemed potentially rewarding objects for a study interested in similarities and differences across cultures and disciplines. Duszak notes that “an academic text reflects the social self-image of the writer and his/ her perception of the readership” (1997: 13), and it emerges in this study that discipline and particularly culture are powerful influences on the way in which these images are co-constructed, which images are judged appropriate to which situation, as well as the perceived expectations of both authors and readers. This approach recognises a link between culture and social behaviour in written academic discourse. The idea is that culture influences, and is reflected in, the way social relations are realised in various contexts, among them academic writing. In research articles written by humanities academics, as in other forms of human interaction, the “reality constructed through language forms the basis of social organisation” (Crawford 1995: 18), and social organisation of course differs across cultures. This view of language, known as constructivism, holds that identity, knowledge and meaning - even reality - are human products, constructed socially and/ or culturally through dynamic processes of social interaction. Within this framework, writing is regarded as a “culturally-situated social activity” (Hyland 2000: 110). For present purposes, social constructivist theory is useful in that it encourages investigation of the co-operative function of language in creating meaning in texts, and the role of social interaction and identity construction in written academic discourse. The recognition that culture, language, meaning and even reality as perceived by individual academics are subjective and interpersonal can be enlightening. Constructivism can be traced to a variety of sources, among them the work of the social psychologist Mead (1913; 1934), who suggested that individual identity is constructed in relation to other people, and what we think other people think of us: “the whole self is reconstructed in its relation to the other selves whose relations are essential to its personality” (1913: 380). Mead saw the self as a “social object, i.e., an object answering to complex reactions of a number of individuals and also for the appearance of the self.” (1925: 268), and referred to his work as the “social psychology which derives the selves of individuals from the social process in which they are implicated” (1934: 222). Other important early influences on constructivist theory were the philosopher Dewey (1925), who insisted that human beings, and the human mind, could be best understood in relation to their social environment, and the philosopher and sociologist Schütz (1932), who argued that scientific facts are at least partly constructed by the observer. Influenced by Mead and Dewey, Habermas formulated the ‘Theorie des kommunikativen Handelns’ [theory of <?page no="75"?> 75 communicative action], which proposes that rationality resides in the structure of contextualised human communication and interaction through language, rather than in language-external entities such as the natural world or the knowing subject (1976a, 1976b, 1981). Mead’s pupil and successor Blumer (1969) described the school of thought concerned with the process of interaction in the formation of meanings for individuals as ‘symbolic interactionism’ (Blumer 1969). Symbolic interactionism was developed further by sociologists such as Berger and Luckmann (1966) into social-constructivist theory, which in its radical form suggests that reality is socially constructed in terms of embedded social relationships. Following Berger and Luckmann (1966), Kukla (2000) argues that there are no objective facts, only facts about human interaction. Accordingly, ontological reality does not exist other than as a social construct represented in social interaction. Radical social constructivism therefore opposes essentialism in all its forms: realism, rationalism and empiricism. This strong form of social constructivist theory, which Kukla (2000: 21) calls ‘constitutive constructivism’, is not subscribed to here. It can be distinguished from the weaker social constructivism of philosophers such as Searle (1995), who propose that external reality is only partly socially or culturally constructed. Alongside identity and reality, knowledge is also seen as a human product, and as such sociallyand culturally-constructed. In this way, individuals are held to create meaning through social interaction (Prawat & Floden 1994; Gredler 1997; Neuman 2000). The anthropologist Geertz (1973) examined cultural variation from a socialconstructivist perspective, defining culture too as a social construct: […] man is an animal suspended in webs of significance he himself has spun. I take culture to be those webs, and the analysis of it to be therefore not an experimental science in search of law but an interpretive one in search of meaning. (1973: 5) The relationship between language and social identity was further examined by researchers such as Gumperz (1982), who adopted an ethnographic approach to identity construction across cultures, and LePage and Tabouret-Keller (1985) who proposed that speech is an “act of identity”, and characterised self-construction through language as a dynamic process of alignment and distancing in relation to particular social groups. Hyland echoes LePage and Tabouret-Keller when he observes that “academic writing, like all forms of communication, is an act of identity” (Hyland 2002c: 1092; see also J. Flowerdew 2002a). According to a socialconstructionist viewpoint, language is not just a way of transmitting knowledge; it is the most important sign system for human society (Berger & Luckmann 1966: 36-7). Language is the medium through which social identity is created, social relations are maintained and social interaction is <?page no="76"?> 76 managed: human communication is thus a “form of joint social action” (Mühlhäusler & Harré 1990: 14). Social constructivism in its radical form has been criticised for the dubious logic of constructing a theory stating that theories are constructed: The central ambition of the cultural constructivist program - to explain the deepest and most enduring insights of science as a corollary of social assumptions and ideological agenda - is futile and perverse. The chances are excellent, however, that one can account for the intellectual phenomenon of cultural constructivism itself in precisely such terms. (Gross & Levitt 1998: 69) While the strong version of social constructivism is neither defensible nor helpful, for our purposes, the theory does usefully focus attention on academic texts as social phenomena, rooted in a particular socio-cultural context. This provides a valuable basis for studying language use in academic texts across different cultural contexts. As Mauranen notes: Texts as cultural products act out relevant social relationships within the culture, and in this way provide the keys to understanding themselves as well as other aspects of the culture. (Mauranen 2001: 53) Of course, culture is not the only contextual factor which influences social interaction and identity construction in academic texts. The choices writers make are shaped by interacting cultural, social, disciplinary, institutional and discoursal influences and norms. In recognition of this, the study also considers the disciplinary, institutional and wider social context in which each scholarly article is rooted, and examines the influence of factors such as discipline, gender and academic status on the ways in which academic authors construct identity and interact with their reader(s) in the corpus texts. There are many approaches to studying interaction and identity in academic texts, and scholars have examined different linguistic features under this same heading, or analysed similar features as part of studies of ‘stance’, ‘evaluation’ or ‘appraisal’ (see further discussion in chapter four). Extensive work on social interaction in academic prose has been carried out by Hyland, who has examined interaction in research articles, book reviews, abstracts, scientific letters and textbooks, and also carried out interviews with experienced researchers (Hyland 1999a, 1999b, 2000a, 2002b, 2002c, 2002d, 2003, 2005). These studies approach the connection between discourse and lexico-grammatical features from different directions. Some of Hyland’s work centres on a lexico-grammatical feature, and examines the discourse functions this feature fulfils: so for example personal pronouns are examined as part of authorial identity (2002c) or self- <?page no="77"?> 77 promotion (2003), and reporting verbs are interpreted as having evaluative functions (2002d). Other studies by the same author start with a discourse function, and then examine the linguistic features that can be employed for that purpose, for example Hyland (2000) on interpersonal and textual metadiscourse (104-131), or Hyland (2005) on stance and engagement, both of which examine the same linguistic features, but under different discourse headings. The general effect has been to show the great range of both linguistic features and discourse functions related to social interaction in academic writing. Researchers have advanced different opinions about whether discourse should be approached through surface features, or surface features examined for discourse functions. Myers (1999) regards a focus on explicit speech acts or pronouns as narrow, and cites Thompson and Thetela (1995), who consider modality, evaluation and the ways roles are enacted or projected in texts as facets of the writer and reader relationship. The present study approaches social interaction and identity construction from both directions, by investigating one lexico-grammatical phenomenon, and one discourse phenomenon. The categorisations applied by different researchers are still in the process of definition: Hyland (2005) for example treats evaluation as an aspect of social interaction, while Hunston and Thompson (2000) examine social interaction as one facet of evaluation (see chapter four for further discussion). Martin (1992, 1997, 2000), who applies a systemic function approach, treats interpersonal positioning and relationships as one aspect of appraisal. Different labels and initial directions aside, the general idea is nonetheless similar, and each of the studies has contributed to a re-shaping of the way academic discourse is viewed: Over the past decade or so, academic writing has gradually lost its traditional tag as an objective, faceless and impersonal form of discourse and come to be seen as a persuasive endeavour involving interaction between writers and readers. (Hyland 2005: 173) The social-constructivist view of academic writing certainly opens up fascinating new perspectives for research. However, it is not without problems. The difficulty with approaches that focus on discourse phenomena is that they are often ill-suited to the automatic corpus analysis of large samples of text, as was discussed in section 2.3. In the case of evaluation, for example, if one had a tagged corpus, theoretically all adjectives, all adverbs, all modal and speech-act verbs and much more besides would have to be examined and manually classified as to discourse purpose. This is not feasible. The other issue with such approaches is that the attempt to equate discourse functions such as evaluation with specific lexico- <?page no="78"?> 78 grammatical features is fraught with difficulty: most lexico-grammatical features can serve more than one discourse purpose, and most discourse purposes cannot be analysed by examining only a closed set of surface features. One should therefore be wary of studies which allot neat, mutually-exclusive sets of surface-level phenomena to each discourse purpose: the reality, as we shall see, is far more complex. Researchers must be willing to supplement automatic with manual analysis, and accept that no study can provide an all-encompassing picture of a discourse phenomenon. If this realisation is accepted, as it is for example by Hyland (2002d), research into social interaction in academic writing will continue to be rewarding. As part of the social interaction taking place in scholarly texts, scholars construct an identity as an ‘academic author’. This form of interaction is central to academic discourse, for, as Scollon (1994) notes, scholarly writing is as much the construction of an authorial self as the presentation of fact. Observations such as Scollon’s are based on a symbolic interactionist approach to academic writing, which goes against an essentialist view of identity, seeing identity instead as a social phenomenon, co-constructed through language by the participants in an interaction. In a seminal study of the discoursal construction of identity in academic writing, Ivani distinguishes between the “autobiographical self”, which “develops in the context of socially constrained access to possibilities to self-hood (1998: 28), and the “discoursal self”, which writers construct “not out of an infinite range of possibilities, but out of the possibilities for self-hood which are supported by the socio-cultural and institutional context in which they are writing” (Ivani 1998: 28). It follows, therefore, that the way in which the discoursal self is constructed will vary across cultures, disciplines and academic status levels, and this is shown in chapter three. Academic texts also reveal the complex and shifting network of public and private identities each scholarly author possesses. One of the main ways in which authors construct these identities within the text is through the use of personal pronouns, which are “the major indexical devices of language in use” (Mühlhäusler & Harré 1990: 46). As we shall see, identity construction is in fact just one of the many discourse functions personal pronouns fulfil in academic texts: it can also be used for example to claim authority as a researcher, to advertise one’s own work, and to build the relationship between writer and reader. Personal pronouns express social relations and attitudes, and encode personal and social deixis. They are “indicators of the complex relationships between selves and the societies these selves live in” (Mühlhäusler & Harré 1990: 47). For a study interested in the nature of relationships within an academic discourse community, and the way these relationships differ <?page no="79"?> 79 across cultures and disciplines, it is therefore of obvious interest to study the use of personal pronouns. In a recently-proposed model of social interaction in scientific discourse, Hyland (2005) treats first person singular pronouns, which he calls “self-mention”, under “stance”, and second person singular and first person plural pronouns, which he calls “reader pronouns”, under “engagement” (2005: 176, 181-2). Stance is defined as “textual voice” and engagement as how writers relate to readers (2005: 176). The distinction between stance and engagement is useful, and corresponds roughly to that between authorial identity and writer-reader interaction in the present study. In practice, however, these two aspects of writing are interwoven: identity is constructed in relation to the reader, and interaction with the reader is influenced by the author’s discoursal identity. The divination of mutually-exclusive communicative intentions and assignation of also mutually-exclusive linguistic features in Hyland’s (2005) model therefore seems contrived. Surely the way in which authors present themselves (stance) also plays a role in their interaction with the reader (engagement). Likewise, it is not clear how inclusive first person plural pronouns (engagement) are to be separated from authorial identity (stance). Clearly, many personal pronouns have to do with both identity construction and social interaction, and a discussion of pronouns should not attempt to separate these two functions along formal grammatical lines. It is worth noting that personal pronouns are not the only way in which person reference is encoded in academic texts. Authors occasionally also refer to themselves and their reader in the third person, and these instances too are considered in the present study. A too narrow focus on surface features such as pronouns in the construction of author and reader identity would neglect this aspect of person reference. Academic authors interact with different groups through their writing: most obviously and immediately, this is with the intended reader(s), but authors also interact with other researchers in their citation and discussion of colleagues’ work, with specific schools or established positions within their discipline, and possibly also with the editors and/ or reviewers of the journals in which their work is to appear. The corpus texts were written by specialists for a specialist audience, and show different types of interaction between members of the academic discourse community, across different disciplinary discourse communities, in the context of a single academic genre. Writer-reader dialogue in particular is increasingly recognised as one of the central features of academic writing (Hyland 2005). One of the main ways in which academic authors communicate directly with their readers is through metadiscourse, which can be defined most briefly as writing about writing, or those elements of the text that do not directly <?page no="80"?> 80 contribute to the informational content. As we shall see in chapter four, there is a very large and open group of linguistic devices available to the author for expressing this other type of content. However, for the purposes of this study, a specific subset of metadiscourse was selected: comments made by the author in the text about the text: (1) This raises…a point which will be considered later. (HIUK) (2) Ich beginne mit der Schilderung verschiedener… [I begin with the description of various…] (PHDE) Such items, which I shall call text comments, allow the author to perform a variety of tasks, such as organising discourse and drawing attention to this organisation, leading the reader through the text, commenting on or justifying the structure of the text and establishing a cooperative relationship with the reader in the shared knowledge-gathering process. The reader who is present in an academic text is not the actual person physically reading the text, but a textual construct created by the author: the author assigns expectations to the reader, and anticipates what the reader requires in order to understand the text. Written academic texts can therefore be seen as a record of a dialogue between writer and reader in which the “writer has to conduct the interaction by enacting the roles of both participants” (Widdowson 1984: 65; see also Purves 1984). In this way, scholarly writing can be seen as a “series of exchanges in an interaction conducted half by proxy” (Thomson 2001: 75). It follows then that the writer’s expectations of the readership, and anticipation of the reader’s expectations, are of great importance if the interaction is to be successful. Readers’ expectations of text structure, and of the relationship with the author, vary across cultures (Huckin & Olsen 1991: 406) and also across disciplines. Authors’ anticipation of these expectations, and the ways in which they seek to cater to them, also vary across cultures and disciplines, as we shall see for the British, US-American and German philosophers, historians, folklorists, linguists and literary scholars in this study. Text comments therefore reveal much about author and reader expectations regarding both discourse structure and the social relations encoded in academic texts. They reveal important aspects of the interaction between writers and readers, who are the two central interactants in a written text. This study therefore considers two main discourse topics - social interaction and identity construction - by analysing two main phenomena: person reference and text comments. Neither the discourse purposes, nor the linguistic features examined, are discrete phenomena, and they are not treated as such: the study recognises that identity construction is inseparable from social interaction, just as person reference is used in text <?page no="81"?> 81 comments. An exhaustive list of the linguistic features used to achieve any one discourse purpose is theoretically impossible to compile and practically impossible to analyse. For this reason, this study, like all other studies, focuses on selected features only. Person reference and text comments were chosen because they reveal important aspects of social interaction and identity construction, and seemed interesting from a crosscultural, cross-disciplinary perspective. 2.7 Method of analysis The methodology combines discourse analytic with corpus linguistic approaches. As suggested in section 2.2, corpus linguistics could offer a solution to many of the methodological problems which continue to plague academic discourse research. This in turn would increase the reliability of the results and their usefulness for pedagogy. In light of the challenges discussed in section 2.3, it is one of the aims of this study to demonstrate ways in which these two approaches can be combined more successfully. The first step in the analysis was to identify the two phenomena analysed in this study, person reference and text comments, in the corpus texts. This required two different approaches, one more typical of corpus linguistics, one more common in discourse analysis. Person reference could be initially identified automatically, using the standard corpus linguistic software package Wordsmith Tools. However, after the first automatic searches, some manual analysis was still necessary, since quotes were not counted, and neither were instances of German ‘Ich’ in the philosophical sense, which is often translated into English as ‘ego’. Pronouns used as examples in the discussion of (mainly linguistics) articles, examples from other languages and translations were also discounted. Thus, every single person reference token identified by the concordancing program had to be manually re-checked in context. Some examples of discarded hits are the following: (3) Mein Gedanke, Sie kennen ihn schon, führt den Wunsch, jetzt das Bedürfnis in sich: Sie Weimar, Sie mir selbst zu sichern. [My thought, you know it already, holds the wish, now the desire within it: to claim you for Weimar, and for myself.] (LTDE) [A quote from a letter by Carl Alexander] (4) ‘I dread the losing hold of forms’, Tennyson himself once declared. (LTUK) (5) Die Inauguraldisseration belegt nämlich auf eindrucksvolle Weise, daß Kant um 1770 noch nicht über eine Konzeption des denkenden Ich <?page no="82"?> 82 verfügt … [for the dissertation shows, in an impressive fashion, that in 1770 Kant did not yet possess a conception of the thinking ‘I’] (PHDE) (6) Beim Eintritt Eiseners in die Stube haben die beiden sich noch mit Sie angesprochen, dann jedoch fragt er … [When Eisner enters the room, the pair still addresses each other with ‘Sie’, but then he asks …] (LTDE) (7) (78) I would’ve asked you, only my mother told me not to. (LGUK) [a linguistic example] (8) The exceptions were that instances of the words I and my were each capped at 10 percent of an individual’s total tokens … (LGUS) (9) Similarly with Elizabeth (Elý` sheba ‘God is my oath’) and Rachael/ Rachel (Ra` he` l `ewe'). (LGUK) (10) The negative XP me niets moves overtly to Spec NegP from its underlying position … (LGUK) It was also necessary to filter out manually those cases of third person references to ‘reader’/ ‘readers’/ ‘Leser’ and ‘author’/ ‘Autor’ which did not constitute an indirect address to the reader or oblique reference to the author of the article in question, but referred instead to readers or the author of another work, or to authors in general: (11) Damit ist die Aufhellung der Umstände der geheimnisvollen Schwangerschaft nähergerückt, auf die die Leser seit langem warten. [Thus the revelation of the circumstances of the mysterious pregnancy, which the readers have long been waiting for, comes closer.] (LTDE) (12) The device continually reminds the silent, solitary reader of the distance between writing and hearing. (FLUS) (13) Der Autor des bellum Alexandrinum widmet allein den Vorbereitungen und dem Verlauf dieser Seeschlacht fünf Kapitel … [The author of the bellum Alexandrinum devotes five chapters only to the preparations and course of this sea battle …] (HIDE) (14) Although Harland’s novels were printed before the 1891 copyright act legitimized the author as the sole creator and owner of a work … (LTUS) Person reference was therefore identified largely using standard corpuslinguistic analysis methods, though these had to be supplemented by manual analysis. The text comments, on the other hand, could not be identified automatically at all, for while there are certain lexico-grammatical <?page no="83"?> 83 features commonly found in them, such as deictic expressions and first person pronouns, these features are not present in all text comments. An automatic analysis therefore would have uncovered only some of the examples, and since I considered this unsatisfactory, the text comments were identified completely manually by reading through each of the 100 articles three times. The first read-through was used to establish the taxonomy of text comments applied in the study, the second to identify and classify the text comments, and the third read-through to test that the classification was reliable. The manual analysis was extremely timeconsuming, and would probably not be practicable for corpora larger than SCEGAD. The second step was to conduct a quantitative analysis of the phenomena, comparing frequencies per 10,000 words across cultures, languages, disciplines, authorship (single or multiple), author gender, author age and academic status levels. Each text counted as a single observation, which was entered into a Microsoft Excel table, to facilitate later statistical analysis. The results tables therefore contained 100 observations arranged in 100 rows. To enable multivariate statistical analysis, the seven independent variables were coded as nominal-scale values, as can be seen in Figure 2.4. <?page no="84"?> 84 Figure 2.4. Coding of independent variables in frequency tables ______________________________________________________________________ Independent variable Numerical values assigned ______________________________________________________________________ Author native language English 1 German 2 Author native-culture British English 1 US-American English 2 German 3 Discipline Philosophy 1 History 2 Folklore 3 English / German literary studies 4 English / German linguistics 5 Authorship Single 1 Multiple 2 Author gender Male 1 Female 2 Mixed (multiple-authorship) 3 Author age Under 30 1 30-39 2 40-49 3 50-59 4 60-69 5 70 and over 6 Author academic status Pre-PhD 1 Post-PhD 2 Full professor 3 Emeritus professor 4 ______________________________________________________________________ Values which were unknown, such as author age and academic status in a few cases, were left blank. Initially, the results were recorded as raw (absolute) frequencies per text. In order to enable comparison across texts, the raw frequencies were then converted into frequencies per 10,000 words for each single text (not as a whole for each subcorpus), which ensured that the results were weighted for different text lengths. The f/ 10,000 measure was chosen because the average length of the corpus texts was 8,937 words: a measure per 1,000 words therefore would have risked artificially inflating the frequency of the phenomena, just as a measure of 100,000 words would have artificially deflated their frequency (Biber et al. 1998: <?page no="85"?> 85 262-4). The results were compared with those from other corpora, such as the BNC and FLOB and FROWN, to check the SCEGAD tendencies against a larger data sample. In most cases, however, the comparison with reference corpora was not particularly revealing, due to the abovementioned unsuitability of these corpora for discourse studies. Initially, several different subtypes of each phenomenon were distinguished. This set was then reduced by forming thematic groups of phenomena for the final analysis. The data table for person reference, for example, initially distinguished between 65 different subtypes of person reference, according to different grammatical case, formulaic lexical expressions, position in the text, pronoun referent and discourse purpose. Aggregate groups were then created according to common features: all first person reference, for example, or all pronouns referring to the academic community. A vital component of the quantitative analysis was the statistical analysis of the results, using the software program SPSS. 34 To give an idea of how the results were prepared for this analysis, Figure 2.5 shows the SPSS data input for a sample of 10 observations for person reference frequency (10 English-language folklore texts). The meaning of the nominal values assigned in the first seven columns (independent variables) can be seen in Figure 2.5. Figure 2.5. Example of SPSS data input for 10 English-language folklore articles (person reference frequency per 10,000 words) Lang. Cult. Disc. Auth. Gender Age Status 1p. sing. 2p. sing. 3p. sing. 1p. pl. all reference 1 1 3 1 1 2 2 3.51 0.00 0.00 1.76 5.27 1 1 3 1 2 2 2 0.00 0.00 0.00 1.94 1.94 1 1 3 1 1 4 2 33.20 0.00 0.00 7.38 40.57 1 1 3 1 2 1 1 5.26 0.00 0.00 12.28 17.54 1 1 3 1 1 2 2 20.95 0.00 0.00 19.02 36.97 1 2 3 1 2 4 3 22.63 0.00 0.00 4.16 30.49 1 2 3 1 1 3 2 14.33 0.00 0.00 45.85 60.18 1 2 3 1 1 5 3 8.68 0.00 0.00 43.49 52.07 1 2 3 1 1 4 19.14 2.94 1.47 8.83 32.39 1 2 3 1 2 3 2 73.48 0.00 0.52 49.51 123.51 The first step in the statistical analysis was to carry out (bivariate and multivariate) significance tests, in order to discern whether the differences between groups that could be perceived through human observation were indeed valid, and conversely to establish whether seemingly marginal differences in frequency were not in fact statistically significant. In general, I chose a standard statistical measure, regarding as significant all 34 SPSS (Statistical Package for the Social Sciences) is the most widely-used and comprehensive integrated system for statistical data analysis. <?page no="86"?> 86 results up to the “p is less than 0.05” degree (p<0.05). 35 In some specific cases, which are explained in the relevant sections, a lower significance level was chosen (chiefly with the multivariate significance tests). Since the data did not have a standard (Gaussian) distribution, for bivariate tests the Mann Whitney test was used. 36 For the multivariate significance tests (three or more independent variables), first the Kruskal Wallis test was carried out in order to discover whether there were any significant differences at all between medians in the data, and then the Mann Whitney test was performed on groups of variables in pairs to establish specific significance levels. It was thus possible to discover which independent variables had a discernable effect on the frequency of the selected phenomena, and which did not. The results of the significance tests were sometimes surprising, in that they were very different from what the naked eye discerned. The second type of statistical analysis carried out was a Spearman Rho correlation analysis. 37 Using correlation analysis in combination with backwards and forwards stepwise multiple regression, 38 it was then possible to establish which of the independent variables was best able to predict how often particular authors would employ person reference or text comments. The multiple regression analysis showed which independent variables had the greatest predictive power. 35 This means that all results where there was a more than 5% chance that they could have been due to chance were discounted as statistically insignificant. For a good general introduction to statistics see Bulmer (1979), Freedman, Pisani & Purves (1997), or, in German, Zöfel (2000). An introduction to statistics written especially for (corpus) linguists is Oakes (1998). 36 Mann Whitney is a non-parametric alternative to the two-sample t-test, and is used to compare the medians of two independent groups of sampled data. The test reveals whether two samples are from the same or different populations, expressed as a degree of probability. See the references in note 24 above for further literature on statistics. 37 Spearman Rho is a standard measure of correlation for non-Gaussian samples. Correlation analysis is used in order to discern the type (linear, non-linear), direction (positive or negative) and degree (between -1 and +1) of relationship that exists between the dependent and independent variables. The correlation coefficient reveals what percentage of variance in the dependent variable (a particular type of person reference or text comment, for example) can be attributed to variance in an independent variable (such as author culture or gender). See Backhaus, Erichson, Plinke & Weiber (2005), or the more general works listed in note 24 above. 38 Regression is used to predict the amount of variance in the dependent variable that can be accounted for given a set of predictors (independent variables). Both forwards and backwards step-wise multiple regression analyses were carried out. In forwards regression, one independent variable at a time is added to the regression model, with the predictor that accounts for most of the variance entered into the model first. With backwards regression, one independent variable at a time is removed from the regression model, with the predictor that accounts for most of the variance entered into the model removed last. Forwards regression is then carried out additionally to test the analysis. See Backhaus et al. (2005), or the more general works listed in note 24 above. <?page no="87"?> 87 Finally, a factor analysis was performed on the data. 39 The results produced a correlation matrix grouped for the variables which clustered together, thereby giving some indication of common ways in which sets of variables affected the observed frequencies. The quantitative results are presented in table form, showing relative frequencies and comparative differences in frequency, and also as bar graphs and prototypical profiles for different subgroups. Once the quantitative results had been gathered, calculated, statistically tested and discussed, the next step was to conduct a qualitative analysis. This part of the analysis was clearly more discourse analytic in method, and centred on questions of context of usage and discourse purpose. The analysis discerned a variety of different discourse strategies for each phenomenon, and examined cultural, disciplinary and gender differences in the use of these strategies. Close attention was paid to the connection between the actual phenomena analysed, and their discourse purposes at various points in the text, i.e. between person reference and text comments, and identity construction and social interaction. This was a vital step, since, as we saw in section 2.1 and 2.2, previous studies often lacked a clear connection between the micro-level linguistic features they had analysed, and the broader discourse implications of these features. In the case of person reference, discourse strategies, the relationship between grammatical form and discourse function, and the issue of multiple identification and allegiances in academic prose were also considered. In the case of text comments, the implications for author-reader interaction, and the relationship between textual metadiscourse and readability were discussed. The qualitative analyses therefore covered a range of discourse issues not otherwise considered in the quantitative analysis. In sum, then, the study combined a variety of automated corpus-linguistic methods, including appropriate statistical techniques, with manual analysis methods and discourse perspectives, in order to give a fuller picture of the phenomena under discussion. The general idea was to demonstrate that it is possible to analyse academic discourse contrastively in an empirically sound manner, using a combination of corpus and discourse techniques. In this way, the approach aims to provide results that are more reliable than are those of previous studies, which did not attempt to combine these two methodologies. The rationale behind this approach is by no means a desire to turn away from quantitative analyses, which are the basis of corpus-linguistic methodology. Rather, the aim is to integrate such methods into discourse analysis, by identifying, classifying and quantifying discourse phenomena 39 Factor analysis is used to detect structure in the relationships between independent variables. See Backhaus et al. (2005), or the more general works listed in note 24 above. <?page no="88"?> 88 in large passages of text. Corpus linguistics therefore becomes a tool for shedding light on discourse. There can thus be no question of surrendering corpus linguistic methodology for any perceived inadequacies. Instead, the study adapts and extends corpus linguistic methods to a new area of application. This point is worth stressing because it relates to the consistency of purpose in the present study. As we saw above, there have certainly been previous attempts to analyse discourse phenomena quantitatively. However, earlier work in this area was not informed by any consistent or specific methodology, and often lacked a sound empirical foundation upon which to base any conclusions. In a sense, these earlier studies (Clyne 1987; Buhl 1999; Baumann 1998; Fandrych & Graefen 2002; Suau 2004) were trying to offer numerical results with the adequacy of a corpus analysis. The problem, however, was that the results were based not on corpora, but on the kind of speculative, circumstantial evidence usually found in discourse analysis. Presenting the results as tables or graphs did not make the data any more reliable. Had these studies aimed only to provide a detailed discourse analysis of these data, then that would have been perfectly legitimate. The problem lay in the attempt made in these studies to extract results and posit conclusions not supported by the data. Now that a reasonable amount of data, in the form of the SCEGAD corpus, is available for addressing the questions of interest here, it would have been theoretically possible to have worked purely quantitatively. Indeed, much of the analysis offered here is concerned with quantitative data, and is quite clearly based on corpus linguistic methodology. However, rather than being the first and only step in the analysis, quantitative data are the basis, and the point of departure, for the detailed discussion of the discourse phenomena I have chosen to analyse. The central point is that, no matter how large or qualified the data base, for a study of social interaction, conclusions cannot really be reached solely on the basis of figures. Adamzik for example writes of the resistance on the part of competent language users to accept purely quantitative statements as an adequate description of their knowledge of a particular text type (1998: 114). It is far more important, Adamzik writes, to discuss questions such as who uses a particular text type, in which pragmatic context, and for which purpose (1998: 117). Conrad too argues against purely quantitative approaches in corpus studies, urging that “even the most sophisticated quantitative analyses must be tied to functional interpretations of the language patterns” (2002: 78). As a discourse analyst as well as a corpus linguist, I do not believe that quantitative analysis should be an end in itself, no matter what the phenomenon analysed. Figures are useful only insofar as they tell us something helpful, something new and revealing, about what is being examined. In this case, the object of study is <?page no="89"?> 89 text, or more specifically how processes of interaction and identity construction are reflected in academic texts. Interaction is a discoursal phenomenon, and as such is an object of linguistic study usually examined by discourse analysts, not corpus linguists. However, the linguistic elements used to construct, negotiate and manage interaction processes are no different from any other linguistic devices found in all written texts, and frequently examined in corpus linguistic studies. Consequently, I look here at some of the same linguistic phenomena as any other corpus linguistic study, yet importantly, I look at them in a different way, concentrating on issues such as socio-discoursal function and possible extra-linguistic motivations for specific uses, rather than focussing exclusively on frequency and distribution. Instead of examining grammatical features in isolation, treating them as a separate feature of language, I integrate them into other related features of the same text, and group them under the discourse strategies to which I judge they belong. This integrated approach takes lexico-grammatical frequency and explains it as part of discourse strategy, thereby helping to bridge the gap between traditional corpus linguistic studies and discourse analysis. It is hoped that the study will benefit from this combined approach, and, through it, be able to offer a more thorough analysis of the phenomena under discussion. <?page no="91"?> 91 Academic writing is not just about Conveying an ideational ‘content’, it is also about the representation of self. _______________________________ (H YLAND 2002c: 1091) ‘I’ is harder to write than to read. _______________________________ (B ARTHES 1982: 487) Chapter 3 Me, you and us: fashioning authorial, reader and social identity One of the most interesting features of academic discourse is the tension between the impersonality traditionally regarded as characteristic of the scholar, and the personal identity, transmitted through writing, which is inseparable from each academic author. In fact, scholars are no more impersonal than any other group of authors, yet, because of historical and theoretical considerations relating to conceptions of empirical research, the type of texts they produce are supposed to be somehow objective, factual and therefore impersonal. Scholarly language is intended to reflect, or at least not compromise, the logic of the scientific method, with its systematic processes of observation, deduction, hypothesis, prediction, experiment, and ultimately theory formulation. Though each of these steps, and the language used to report them, is inherently subjective, academic writing traditionally seeks to minimise the additional, overt subjectivity of personal content. Nonetheless, research articles are still personal texts, and at times the academic author clearly steps into the foreground. This chapter examines the circumstances under which authors of different cultural backgrounds do this, and the purposes for which explicit person reference is employed. By choosing between different ways of positioning and representing themselves and others via language, academic authors both construct and reveal <?page no="92"?> 92 multi-dimensional identities. Within the bounds of a specific genre, authors also exercise their freedom to create different kinds of communicative situations - monologic or dialogic, involved or distanced - and claim membership of different kinds of academic, cultural and social groups. By drawing upon the SCEGAD corpus, which is balanced not just for linguistic but also many extra-linguistic variables, it is possible to examine the influence of different factors on the language choices academics make in their writing. These factors are influential, but of course do not operate deterministically, and authors also exercise their individual freedom. First, I consider some ways of constructing and expressing identity in academic texts (section 3.1) and examine the role of hedging in this context (section 3.2). I then move on to discuss the taboo and reality of authorial presence in these texts (section 3.3). The following sections contain the quantitative analysis of person reference according to language, variety, culture, discipline, author gender and academic status (section 3.4), and a qualitative analysis of the discourse purposes for which person reference is used (section 3.5). In section 3.6, I discuss the role of multiple identities and multiple allegiances in academic prose, while the following section is concerned with the relationships between grammatical form and communicative function. 3.1 Constructing and expressing identity in academic texts The need for impersonality in academic discourse, or rather the seeming absence of any need for personal content, is explained by the main purpose of scholarly communication, which is to convey facts or factual information. Since, theoretically at least, facts are independent of both the form in which they are presented, and the person by whom they are presented, scholarly communication is supposed to have no need of the subjectivity associated with any one individual academic. In fact, there is any number of academic genres with a high level of subjective content, such as book reviews, letters to the editor, the literature review sections of PhD theses and so on. The other problem of course is that this information, even if it were to be solely factual, is being conveyed through the medium of language by human agents, and neither language nor humans are solely objective. In practice, these three factors - thematic content, linguistic medium and human agent - combine to shape the factual, social and individual aspects of a piece of academic writing. This means that, at the same time that academics are using language to convey information about their research (reporting findings, criticising previous work, recommending <?page no="93"?> 93 future directions and so on), they are also conveying information about their own identity or, more accurately, identities (personal, individual, gendered, social, discursive and cultural). Academic authors choose between different ways of positioning and representing themselves, and of course others too, via language. Along a sort of cline, some of these choices are more conscious (am I writing for an expert or novice audience? Am I writing a book review or a thesis? ), some less so (am I from a culture that frowns on person reference? Am I female and therefore less likely to say ‘I’? ). Through comparative study it is possible to discern the boundaries of individual freedom or variation within the confines of a particular culture; boundaries that are set at different places in different cultural environments. German, British and US- American academics working in universities in their home countries inhabit similar institutional settings, and though the possible manifestations of self-hood overlap at some points along the continuum of acceptable selfconstructions, the sets of possibilities are by no means the same: Possibilities for self-hood also socially construct the self as author: there are conventions for whether and how to establish authorial presence which differ from one type of writing to another. These conventions influence whether and how actual writers establish themselves as authors in their own writing. (Ivani 1998: 28) One of the most explicit ways in which academic authors express identity in their texts is by referring to themselves and the reader. 40 In so doing, they construct for themselves and the reader a social identity based on an individual conception of the interpersonal relations reflective of their culture, genre and discipline. The primary person-referential devices offered by British and US-American English and by German are personal pronouns. The “grammar of the forms of person reference and ascriptions to persons”, write Mühlhäusler and Harré (1990: 91), are “an essential medium for the acquisition by a person of the socially based concepts of self”, and are “indicators of the complex relationships between selves and the societies these selves live in” (1990: 47). Academic authors make particular use of the first person singular and plural pronouns and the second person singular pronoun to construct individual, group and reader identity. In traditional grammatical descriptions, the first person is said to refer to the speech act role of the speaker, the second person to that of the addressee (see for example Halliday and Hasan 1976; Lyons 1977; Comrie 1981; van Riemsdijk & Williams 1986). In practice, however, there is no 40 See chapters two and four for a discussion of other approaches to studying identity and interaction in academic prose. <?page no="94"?> 94 one-to-one correlation between the grammatical form of the pronoun and its referent in a particular text or moment of speech. The connection made between pronoun form and person, as Wales points out, is “but a traditional convenience”, since “personal pronouns are multi-functional in their roles in different contexts” (1996: 7). In addition, the social relations encoded by personal pronouns are not fixed: authors may be ‘I’/ ‘ich’ at one moment but ‘we’/ ‘wir’ at the next; an addressee ‘you’/ ‘Sie’ in one context, or part of ‘we’/ ‘wir’ in another. Pronouns are traditionally grouped together because of their morphosyntactic similarities - they can be inflected for person, number, and case, and in German, also for gender (see for example Leech & Svartvik 1994: 331-2 for one such definition). The literal sense of the word ‘pronoun’ (i.e. standing for, or in place of, a noun) is also used to explain its function: the Oxford English Dictionary for example has “a word used instead of a noun”. The German Duden Grammatik [Duden Grammar] 41 has the almost identical “Wort, das in einem Satz statt eines Substantivs stehen kann” [word, which can occupy the position of a noun in a sentence], 42 as does the Bertelsmann Grammatik der deutschen Sprache [German Grammar] (Götze & Hess-Lüttich 2002: 21, 262-3). Similar explanations focussing on replacement or substitution for a noun can also be found in many Englishand German-language grammars, linguistic dictionaries and textbooks (for example Bernard 1975; Heidolph, Flämig & Motsch 1981; Fromkin, Rodan, Collins & Blair 1996; Drosdowski & Eisenberg 1995; Crystal 2003a). Such definitions are unsatisfactory when it comes to personal pronouns, because these types of pronouns are in many cases more common in naturally-occurring language than the nouns they are said to replace (Sacks 1992: 353). This point is also made in the influential Comprehensive Grammar of the English Language (Quirk, Greenbaum, Leech & Svartvik 1985), which initially advises that “the term pronoun […] is at least partially appropriate in suggesting that a pronoun can serve as a substitute for a noun” (75), but then goes on to correct itself, explaining that “more generally, however, pronoun is a misnomer on two counts” (76), first because pronouns replace whole noun phrases, not just single nouns, and second because ‘replacement’ is too simple a term for the complex relationship between a pronoun and its antecedent. Mühlhäusler and Harré state simply that there is “surprisingly little support” for the view of pronouns as substitutes for nouns (1990: 49). Wales (1996) occupies an intermediate position between traditional and progressive views. She criticises pronoun definitions couched in terms of 41 The Duden publishing house is the most highly respected producer of dictionaries and lexica for German-speaking countries. 42 All translations from German are my own. <?page no="95"?> 95 substitution, noting that they are more fitted to the third than first and second person pronouns, and discusses the problem of deciding what pronouns are substitutes for (1996: 3, 21). However, Wales concludes that “it is clear that all the personal pronouns can be seen to be useful ‘shorthand’ referring expressions” (1996: 3, 4), which goes against the evidence of real language and the opinions of the researchers quoted above. There are various other definitions, which cite common functions of pronouns, such as anaphora/ exophora, deixis/ indexicality, reference or text coherence (Brown & Yule 1983; van Riemsdijk & Williams 1986; Kaplan 1989; Chomsky 1994; Zifonun, Hoffmann & Strecker 1997). The Cambridge Grammar of the English Language (Huddleston & Pullum 2002) for example cites anaphoric reference as the common feature of both pronouns and pro-forms, which latter term is defined as “an anaphor with little inherent semantic content of its own” (1461); the Zifoun et al. Grammatik der deutschen Sprache [German Grammar] offers the same functional definition (anaphora), but dispenses with the word ‘pronoun’ altogether and operates only with ‘Proterme’ [pro-terms] (1997: 37). Discourse analysts, on the other hand, emphasise the social dimension of pronouns and their role in the construction of ‘self’ and ‘other’ (Malone 1997; Sacks 1992; Schegloff 1996). Biber, Johansson, Leech, Conrad & Finegan (1999), in the corpus-based Longman Grammar of Spoken and Written English, approach such views in their definition of pronouns as “function words which make it possible to refer succinctly to the speaker/ writer, the addressee, and identifiable things or persons other than the speaker/ writer and the addressee” (328). The Langenscheidt Deutsche Grammatik: ein Handbuch für den Ausländerunterricht [Langenscheidt German Grammar: a Handbook for Teaching Foreigners] categorically disposes of traditional definitions, stating that “im Falle des Personalpronomens entspricht die Bezeichnung nicht der Funktion des Pronomens” [in the case of the personal pronouns, the description does not reflect the function], and explaining that personal pronouns do not in fact replace anything, but stand alone as the only adequate way of describing the speaker and addressee (Helbig & Buscha 2001: 208). The encoding of grammatical information or reference is not the only or even the main function pronouns may serve in real language; entirely missing from traditional definitions is the role pronouns, in particular personal pronouns, play in constructing authorial identity and structuring social interaction. Mühlhäusler and Harré are of the opinion that, due to this restrictive grammar-based view, “those aspects [of pronouns] that illuminate processes of human communication as a form of joint social action have remained grossly understudied” (1990: 14). Wales too laments the <?page no="96"?> 96 fact that descriptive grammars concentrate on form rather than function, and on sentence rather than discourse, thereby ignoring that pronouns have a “richness of rhetorical and social connotation beyond the strictly denotational” (1996: 50). The approach in this chapter goes beyond definitions of personal pronouns as indices of grammatical relations such as number or person. It considers the way pronouns are actually used in context, and the discursive functions they can fulfil in written academic texts. This kind of approach recognises that the “meaning of pronouns is textand contextderived” (Mühlhäusler & Harré 1990: 58). What the chapter does not offer is a discussion of hedging, even though the examination of person reference often falls under this heading. The reason for this is explained below. 3.2 Authorial Presence and hedging In the space of two decades, the term ‘hedge’ has developed from a term used to describe a way of modifying set-membership in formal logic (cf. Lakoff 1972) to a name for a complex, open-ended group of linguistic items and expressions related to speakers’ attitudes to what they are saying or writing. The path from one to the other is well-documented (see for example Markkanen & Schröder 1997; Clemen 1998, 2000; Graefen 2000a), and therefore need not be recounted here. Hedges were quickly taken up by researchers interested in discourse analysis and speech act theory and there is now a substantial quantity of literature concerned with them, much of it unrelated to the original meaning of the term (see bibliography in Markkanen & Schröder 1997: 249-271). Since there is no widely accepted definition of what a hedge or hedging is, an essentially absurd situation has arisen, in which, as Lewin (2003: 2) notes, the term is used to describe absolutely contradictory concepts. For example, the term hedge can cover the expression of both certainty and uncertainty, attribution through both personalisation and impersonalisation, and the use of both weakening and strengthening particles (Lewin 2003). Other authors distinguish between hedges, which reduce the force of statements, and boosters, which increase the force of statements (Holmes 1990; Hyland 2000), or between hedges and intensifiers (Grundy & Jiang 2001; Helms-Parka & Stapleton 2003) or hedges and emphatics (Grabe & Kaplan 1997; Hyland 1999a; Abdi 2002). In linguistics, hedges have been variously treated as a type of metadiscourse, metacommunication or metalinguistic device, as an aspect of modality or facework, as a lexical category or set of vocabulary items, as a type of modifier, politeness device or performance disclaimer, as one of many <?page no="97"?> 97 expressions of propositional attitude, and as a conversational feature related to distance and formality. Many researchers count all forms of person reference, in particular those instances where personal pronouns are used to express an opinion or qualify a statement, as belonging to hedging (Zuck & Zuck 1985; Skelton 1988; Darian 1995; Markkanan & Schröder 1997; Hyland 2000; Breitkopf 2005). However, as Varttala points out, the avoidance of personal pronouns has also been termed hedging (Varttala 2001: 38; see for example Luukka & Markkanen 1997). One study even treats both the use and the avoidance of personal pronouns, as well as the use of impersonal constructions, as hedging (Schröder 1998). This same study then laments that the term ‘hedge’ creates the impression of a “definatorischen Chaos” [chaos with regard to definition] (226). Not surprisingly, uncertainty about the definition of hedging carries over into empirical studies of the phenomenon. In the largest study of social interaction in (English-language) academic writing to date, Hyland stresses the “huge importance of hedging and boosting to the negotiation of the writer-reader relationship in [scientific] letters” (Hyland 2000: 90), and indeed, the items Hyland counts as hedging are certainly frequent in the texts he examined (17 to 20 occurrences per 1000 words). However, Hyland’s typology is thoroughly unclear (why is ‘could’ regarded a hedge, and ‘couldn’t’ a booster, while both ‘should’ and ‘shouldn’t’ are deemed hedges, for example? ). Any number of other items on the list seems debatable in its categorisation: ‘we think’, ‘we find’ and ‘perceive’ are regarded as expressions of certainty, while ‘predict’, ‘suggest’ and ‘I/ we claim’ express doubt. Some expressions, such as ‘surmise’ and ‘always’, are even present in both the ‘certainty’ and ‘doubt’ lists. Similarly, the word ‘necessarily’, as in ‘these results are not necessarily an indication of…’, is seen as expressing certainty, while ‘frequently’, as in ‘this feature occurs very frequently’ is said to express doubt, and so on. It is easy to think of many sentences in which the posited boosters express uncertainty, and the hedges certainty, or both express something else entirely. Such an approach totally disregards the meaning of these expressions in use, and the great semantic variability of a word depending on its context. The possible meanings of any number of (modal) verbs, adverbs and adjectives are suddenly reduced to two basic functions. Such work constitutes a gross over-simplification of semantics and contributes very little to our understanding of academic discourse. To be fair, however, this is less an individual failing on Hyland’s part than a basic problem with the concept of hedging and its accompanying terminology. <?page no="98"?> 98 Markkanen and Schröder (1997: 6) emphasise that “no linguistic items are inherently hedgy but can acquire this quality depending on the communicative context or the co-text”. Hedges can in fact be found in every part of speech, depending on the context, for which reason “no clear-cut lists of hedging expressions are possible”, although the authors nonetheless posit a list of approximately 50 hedging expressions. It is probably for this reason that Graefen (2000b: 118) calls hedging a “pseudo-Kategorie” [pseudocategory]. Hyland lists some 180 lexical expressions of hedging and boosting (Hyland 2000), and other, similarly eclectic, lists abound (for a summary, see the annotated bibliography in Markkanen & Schröder 1997). Ultimately, any list or categorisation of such disparate devices remains highly subjective and therefore does not lend itself to application in further studies. The linguistic category hedge is simply too vague to be consistently applied in different studies and, due to its lack of boundaries, is thus unworkable. Lakoff originally defined hedges as “words whose job is to make things fuzzier or less fuzzy” (1972: 195). As Clemen nicely puts it: “Der Begriff Hecke ist also in sich selbst fuzzy” [Thus the term hedge is in and of itself fuzzy] (Clemen 1998: 11). When it comes to the SCEGAD corpus, there are certainly examples where pronouns - the first person singular in particular - minimise the author’s commitment to the truth value of an utterance, that is, they have what is often regarded as a downtoning, qualifying, or hedging function, as can be seen in the following extracts: (15) As far as I know, this view of inquiry first appears in a fully developed form in the writings of C. S. Peirce. (PHUS) (16) Meiner Ansicht nach ist auch in diesem Falle wie oben zu argumentieren: …[In my opinion, in this case the argument above applies … ] (PHDE) This goes some way towards illustrating why some researchers count personal pronouns among hedging devices. However, in both languages, such phrases are part of a homogeneous group of highly formulaic expressions and constitute little more than a nod to the conventions of academic writing. They do not tell us much about the dynamics of the author-reader relationship, nor about the construction of authorial identity, and they reveal only a very small aspect of personal pronoun use in academic texts. In addition, the corpus shows that authors follow exactly opposite strategies with personal pronouns, using them not only to minimise (first extract), but also to strengthen their responsibility for the truth value of an utterance (second and third extract): <?page no="99"?> 99 (17) I have tried to show, though, that like the discredited plot summaries and bowdlerized translations of earlier generations of folklorists and anthropologists, the stories retain their power. (FLUS) (18) As I have shown, the relationship between road protesters and fairies is a complex one, but … (FLUK) (19) Not only does Della Rocca offer no argument for the falsity of (5), but, on the contrary, I have shown that an essentialist has an argument for the truth of (5). (PHUS) As we shall see in section 3.7 below, there are also cases in which the first person plural, used with a first person singular referent, fulfils a function which some researchers consider hedging. However, the SCEGAD authors use personal pronouns to perform numerous other unrelated functions, which cannot be classified as either minimising or strengthening commitment or responsibility. These also will be explored properly further below. Even these few examples from the corpus make clear why the definition of the term hedging has created such problems. They also demonstrate the major problems noted above: hedging as a term is either not definable, or covers absolutely contradictory rhetorical strategies. It is in any case far too narrow a way of looking at the construction of the author-reader relationship in academic texts. It is undoubtedly true that, as Hyland writes, “statements don’t just communicate ideas, they also indicate the writer’s attitude to them and to readers” (Hyland 2000: 88). The purely factual discourse does not exist, except perhaps in the symbolic representations of mathematics or theoretical physics. However, the question is whether hedging is the best way of describing and examining this aspect of written (and indeed spoken) language. In fact, hedging often appears little more than a collective term for anything and everything a particular researcher has left over, rather than a clearly delineated unit of analysis for scientific inquiry. A new unit of analysis must reveal aspects of language use that would otherwise remain unstudied or even impossible to study. This is not the case for hedging, since all of the phenomena often grouped under its heading, from passive to modality to opinion markers, can (more) fruitfully be studied without recourse to any other term. Therefore, I do not regard person reference as a subset of hedging, but treat it as a phenomenon worthy of separate examination. In addition, since - as the corpus examples showed - hedging covers exactly opposite rhetorical strategies, in the discussion of the results offered below it is not listed as a separate type of discourse purpose. In other words, I neither treat person reference as a subset of hedging, nor do I accord hedging a place as a subset of person reference. Instead, the <?page no="100"?> 100 corpus instances where person reference is used to modify (strengthen, weaken or otherwise) the author’s commitment to the truth value of an utterance are grouped with the cases of ‘personal opinion or comment’, this being a broader and also clearer definition of its pragmatic function. These decisions are intended to clarify my discussion of person reference, which is a linguistic feature sufficiently interesting to justify independent study, and sufficiently complex without additional classification. 3.3 The taboo and reality of authorial presence in English and German-language academic texts The explicit authorial presence indicated by person reference is a particularly interesting feature to examine in academic texts, because it has often been stated that these texts do not, or should not, contain such features. The recommendation to adopt an impersonal academic writing style follows from a positivistic view of science as “authoritarian, non-humanistic, objective, purely rational and empirical, universal, impersonal, socially sterile, and unencumbered by the vulgarity of human imagination, dogma, judgements, or cultural values” (Aikenhead 1996: 9; see also Harding 1991; Hyland 1999d). Ideally, a purely objective written style is therefore a reflection of the purely empirical and objective nature of the knowledgegathering process itself. In practice, something along the lines of Karl Popper’s searchlight theory operates, in that the process of accumulating scientific knowledge is subjectively selective even as it strives for objectivity (Popper 1979). The recognition that human perception is not objective or absolute, but personal and subjective, is by no means novel. Francis Bacon for example observed that: Falso enim asseritur, sensum humanum esse mensuram rerum; quin contra, omnes perceptiones, tam sensus quam mentis, sunt ex analogia hominis, non ex analogia universi. Estque intellectus humanus instar speculi inaequalis ad radios rerum, qui suam naturam naturae rerum immiscet, eamque distorquet et inficit. [it is a false assertion that the sense of man is the measure of things. On the contrary, all perceptions as well of the sense as of the mind are according to the measure of the individual and not according to the measure of the universe. And the human understanding is like a false mirror, which, receiving rays irregularly, distorts and discolours the nature of things by mingling its own nature with it] (Bacon 1620 IV: 54; translation Taggard & Thompson 1863) If human perception is subjective, yet scientific prose is to conjure objectivity, a certain tension ensues. It is in this sense that Geertz speaks of the “clash” between “author-evacuated” and “author-saturated” prose (1988: 9), and of the “oddity of constructing texts ostensibly scientific out of <?page no="101"?> 101 experiences broadly biographical” (10). Geertz is concerned with ethnographers engaged in field-work, and though most humanities scholars do not relate experiences that are personal or biographical in quite the same way as that of ethnographers, the same tension between the objective and subjective exists when they come to record their knowledge in written texts. The construction of an authorial presence within a text is therefore a central issue, especially in a genre which generally expects authors to ‘evacuate’, and person reference one of the clearest sites for examining this conflict in action. German scholars in particular have noted that person reference is unacceptable in academic writing in their language. Weinrich influentially stated that “Ein Wissenschaftler sagt nicht ‘ich’” [an academic does not say ‘I’] (Weinrich 1989: 132). This avoidance of first person pronouns has been christened the “Erstes Verbot” [first prohibition] in scholarly language (Weinrich 1989: 132), or the “Ich-Verbot” [I-taboo] (Kretzenbacher 1991: 120), and, along with the narrative and the metaphor prohibitions, is posited by Weinrich as typical of German scholarly prose. Kretzenbacher states that in German academic texts a parallel “unwritten rule”, of almost equal force, is that “Ein Wissenschaftler sagt nicht ‘du’” [an academic does not say ‘you’] (Kretzenbacher 1991: 120-1). Other researchers have made similar observations about person reference. Hutz notes that “so gilt an vielen Schulen im deutschsprachigen Raum oftmals immer noch die alte Maxime, daß die Verwendung der ersten Person in Aufsätzen unterlassen werden soll” [thus the old maxim, that one should desist from using the first person in articles, often still persists in many German-speaking scientific schools] (Hutz 1997: 232). Gläser too comments that the “streng sachbezogenen und unpersönliche (‘agensabgewandte’) Behandlung eines Sachproblems äußert sich in […] der Vermeidung der Ich-Form” [the strictly content-oriented and impersonal (‘agent-averted’) treatment of a problem is reflected in the avoidance of the I-form] (Gläser 1998: 485). Weinrich (1989: 133), Weingarten (1994: 129) and Gläser (1998: 485) maintain that the avoidance of ‘I’ derives from the need for academic objectivity - or rather, the need to give an impression of this. Hutz also associates the avoidance of ‘I’ with a wish for increased objectivity on the part of the author, as well as with a desire to protect oneself from criticism (2001: 125). Person reference therefore would hinder authors in striving to create the appearance of scientific objectivity in their written work. The tacit assumption here is that the avoidance of surface linguistic features such as personal pronouns is sufficient to ensure scientific objectivity. In a similar vein, Clyne quotes a German phrase: “Not he [the German professor] speaks, but scholarship through him” (1987: 238), thereby <?page no="102"?> 102 casting the individual author as (impersonal) instrument rather than (personal) agent when it comes to scholarly research. Person reference is thus seen as being in opposition to, or even mutually exclusive to, that set of desirable characteristics encapsulated by the essentially untranslatable German word ‘Wissenschaftlickheit’ [‘scientificness’], which is often cited in this context (Weinrich 1989; Kussmaul 1978; Hutz 1997). The conception of academic writing as lacking in personal content can also be found in reference to English, albeit in different sources. In English there is a whole genre of style and writing guides for both professional and student writers, which has no parallel in German. It is here, not in linguistics research articles, that one would expect to find similar comments on person reference for English-speaking academic writers. An initial examination of such style guides is, however, surprising: neither the MLA nor the Chicago Manual of Style, both regarded as definitive style guides for English-speaking academic authors, offer any advice at all when it comes to person reference. In a one-page section entitled ‘Language and Style’, the MLA Handbook for Writers of Research Papers blithely states that “good scholarship requires objectivity”, but correlates this requirement only with the avoidance of sexist pronoun use such as generic uses of he/ his/ his (Gibaldi: 2003, 60). The classic US-American style guide The Elements of Style recommends only “place yourself in the background. Write in a way that draws the reader’s attention to the sense and substance of the writing, rather than to the mood and temper of the author” (Strunk & White 2000: 70). This vague piece of advice, which was added by White only in later editions, could be construed as limiting person reference. The more modern Guide for Academic Authors (Luey 2002) again does not comment on person reference. At the same time, there are countless other style guides for non-expert authors, from primaryto tertiary-level students, which offer contradictory advice on person reference. These guides are generally concerned with a different academic text type from that in SCEGAD - essays rather than research articles - and their comments are therefore not directly comparable. However, the recommendations in such guides may still be relevant to the study of expert academic writing. The essay is the main expository text type for student authors, just as the research article is the main expository text type for professional academics. It is possible that the style advice given budding academics in their student days resonates far beyond their undergraduate careers, and may well influence their writing habits in later professional research activity. Yet the advice provided is contradictory. Three examples from student writing guides, ranging from prohibition through moderation to recommendation follow: <?page no="103"?> 103 Do not refer to yourself. Avoid ‘I,’ ‘we,’ ‘us,’ and so forth. Be the omniscient author. (Fry, Jennifer. 2003. Paper Requirements. History Department, King’s College, Pennsylvania. http: / / www.kings.edu/ jrfry/ guidelines.html) It is not forbidden to use ‘I’ in an academic essay, but neither is it good form to do so constantly. The ‘I’ should be used when you are stating your intention, giving an opinion, or when you want to pose a question (preferably not a rhetorical one). Whenever you are conveying knowledge (whether common or specialist) avoid ‘I’, and write objectively. (Lauret, Maria. undated. Notes on Essay Writing: Format, References and Bibliography. Department of American Studies, University of Sussex, http: / / www.sussex.ac.uk/ americanstudies/ documents/ notes_on_essay_writing.doc) Myths about academic writing: […] Never use ‘I’ in essays. […] In many cases, using the first person pronoun can improve your writing […] Humanities: It's usually all right to use ‘I’ in these fields, because the purpose is generally to offer your own analysis of history, language, ideas, or a work of art. (UNC-CH Writing Center. 2002. Writing Center, University of North Carolina at Chapel Hill, http: / / www.unc.edu/ depts/ wcweb/ handouts/ should_I_use_I. html) While we are primarily concerned with expert native speaker authors here, it is worth noting that this chaotic situation naturally leads to great confusion on the part of both native speaker students and non-native speakers writing academic texts in English. Hyland (2002c: 1095) notes that “selfmention can be a considerable problem for L2 undergraduate writers”, while Myers (2001: 65) observes that “issues of opinion in academic writing seem to pose problems for all sorts of students”. German students and non-native speakers writing German-language academic texts probably have an easier task, since, in the rare cases where advice for German academic authors does exist, it follows exactly the prohibitions posited by the German linguists quoted above: Allgemeine Hinweise: Neutraler Beobachter: Sprache / Verwendung von „ich“. Die Arbeit sollte aus der Warte eines neutralen Beobachters geschrieben werden. Ausdrücke wie „ich“, „mir“, „mein“ etc. sind deshalb nach Möglichkeit durch indirekte Umschreibungen (z.B. „Es ist festzustellen ... “) zu ersetzen. Selbst wenn der Verfasser seine persönliche Meinung zum Ausdruck bringen will, sollte er auf Personalpronomina verzichten und stattdessen auf Redewendungen wie „Der Verfasser vertritt die Auffassung ... “ o.ä. zurückgreifen. [General Advice: Neutral observer: Language / Use of ‘I’. The assignment should be written from the perspective of a neutral observer. Expressions such as ‘I’, ‘me’, ‘my’ etc. are therefore to be replaced with indirect paraphrases (e.g. ‘It can be observed … ’) if at all possible. Even when the author wishes to express a personal opinion, he should do so without personal pronouns and instead rely upon expressions such as ‘the author is of the opinion … ’ or similar.] (Schoder, Detlef. 2005. Advanced Course on Information Management, Department of Information Management, University of Cologne. http: / / www. <?page no="104"?> 104 wim.uni-koeln.de/ uploads/ media/ Hinweise_zum_Anfertigen_von_Arbeiten_ Mai05.pdf) There are three points to be made here. First, it is improbable that expert native speaker writers would consult undergraduate essay-writing guidelines for advice. Second, as we have seen, the style guides which professional writers do regularly use contain no recommendations on the use of person reference. Third, since these are only recommendations, they reveal nothing about how students, let alone expert authors, actually write. As with all recommendations, there is likely to be a discrepancy between what people are told to do, and what they in fact do. It is therefore not possible to extrapolate academic language use from style guides formulated for students, and this has not been the intention in quoting from such material. Nonetheless, the contrast between the wealth of recommendations for English-speaking student authors, and the dearth of advice for Englishspeaking academic authors is striking. It is also striking that German linguists can explain and discuss stylistic rules which are absolutely categorical and yet, at the same time, nearly totally unwritten. For English-speaking authors, it seems that, once they have reached a certain status, the rules disappear from view, and expert users adopt an explicitly personal style, which may well include elements previously forbidden in undergraduate style guides. The expert German authors in the SCEGAD corpus do not rigidly keep to the apparently binding prohibitions for their language either. As we shall see later, there are also important differences even among expert authors. Academic status emerges as a reliable determiner of person reference frequency, with professors in both language groups using person reference close to twice as often (p<0.05) as their more lowly colleagues (1.6 times as often for German academics, 1.98 times as often for the English-speaking academics). It seems, therefore, that the strength with which such prohibitions are uttered, and the stringency with which they are followed, declines with increasing academic status. For student authors, the situation seems to be different. A longitudinal study of how philosophy undergraduates express personal opinions revealed that over time the most successful students were those who learnt to express their opinions without the use of person reference, which was overused by the junior students (Binchy 2002). How this relates to the corpus findings just mentioned, that professors employ more person reference than their lower-status colleagues, is uncertain. It may be that more senior and more successful students employ less person reference because they have learnt that good grades depend on compliance with the composition rules they have been taught. While a more junior academic may be subject to ‘rules’ formulated by an academic supervisor, professors are at <?page no="105"?> 105 the top of the academic ladder and as such accountable only to themselves: they issue rules, but need not necessarily follow them. Of course, if a professor were to write in a grossly deviant fashion, the work might not be accepted in a journal, or might be unfavourably reviewed. In general, however, it seems that once composition rules are understood, they are more stringently applied and adhered to by the lower-status members of the community. The construction of academic discourse as purely objective and therefore impersonal may be an ideal - though it often seems only a topos - but it is certainly not the reality. Drescher (2004) argues that the impersonal nature of academic discourse is in fact nothing more than an ideological construct, while Elbow (1991, qtd. in Zamel 1997) states that a detached and impersonal stance is a pretence, because it is not possible to separate the “ideas and reasons and arguments from the person who holds them”. Hyland (2005: 173) notes simply that “a great deal of research has now established that written texts embody interactions between writers and readers”. It is important to note, however, that most studies of interaction in academic texts have not focused on person reference, but examined a variety of other phenomena or sets of phenomena, encompassing many different metadiscoursal, evaluative and appraising functions of academic discourse. In this sense, it has therefore been established that academic texts, like all other forms of language, contain (some type of) interaction, and this finding should not be at all surprising, because it is a truism, or rather a set of truisms. Communication is interactive, interaction is personal, and people are subjective beings. The point here is that academic style guides do not advise against having personal opinions. Quite the contrary in fact: as many student style guides state, it is precisely because everyone knows that academic texts are statements of personal opinion that explicit person reference is unnecessary. I will quote only one example here; more are in the notes: 43 Avoid constructions like in my personal opinion, I believe that , and it seems to me . Readers assume that you are the author of any ideas, opinions, or conclusions expressed in your paper not attributed to someone else. There is no need to draw attention to yourself needlessly. 43 A few more examples: “Write emphatically. Avoid the use of disclaimers such as, ‘in my opinion’, ‘feel’, ‘it seems to me’, and the like. If your name is on the paper, readers will know it is your opinion.” (Carney 2003); “Avoid redundant phrases such as ‘in my opinion’, ‘to me’, and ‘I think’ or ‘I believe’.” (Wooley 2003); and “first-person phrases […] add nothing that is not already implied. Readers will assume that the ‘beliefs’ expressed in your essay will be your beliefs, so why do you need to tell readers this information? ” (Rambo 2004). Because of limited space, I restrict myself to these three examples; there are many more. <?page no="106"?> 106 (Pocock, Emil. 2002. Writing for Clarity and Grace: And Practical Suggestions for Revising History Papers. Department of History and American Studies, Eastern Connecticut State University. http: / / www.easternct.edu/ personal/ faculty/ pocock/ writing.htm) What academic authors, at least the student authors for whom these guidelines were formulated, were told to avoid is the explicit marking of their opinions as such. They were either advised not to write in a personal way, or not to base ideas solely on personal opinions: this is not the same as being told to avoid opinions altogether, since this is an impossibility in a genre concerned with argument. Given that there is evaluation and appraisal in all texts, from adverts to articles, the interesting question becomes whether there is explicit person reference in academic texts. This is an entirely different question, and perhaps a more appropriate one for this genre too, because it is these explicit linguistic features with which both style guides and authors themselves seem preoccupied. The avoidance of surface linguistic features marking personal opinions is in this sense somewhat disingenuous. One of the top-selling US- American style guides, The Writer’s Repertoire, notes that “To the scientist it is unimportant who observed the chemical reaction: only the observation itself is vital” (Gong & Dragga 1995). This is nonsense, and suggests that academic authors should strive for a kind of objectivity which, due to the nature of human perception, does not and cannot exist. Further, there is a danger here, in that, through the avoidance of the surface features which explicitly encode subjectivity, academic authors suggest an objectivity they do not possess. Such sophistry may sometimes be a deliberate strategy on the part of many academic authors, and though it need not be consciously deceptive in purpose, it is often deceptive in effect. Ultimately, because scholars present their accumulated knowledge and opinions as universal rather than personal, readers may assume that scholars are objective, and therefore not question sufficiently the information or viewpoints being conveyed through the text. Invisible, neutral observers cannot be wrong. A British university style guide, which advises student authors that it is “important to avoid too many uses of ‘I’”, acknowledges the criticismneutralising effect of such avoidance: “One benefit of avoiding a subjective approach is that it is more difficult to criticise you personally for the ideas in your essay” (2003. University of Essex, Study Skills: Writing Skills Part 1 - Academic Style. http: / / www2.essex.ac.uk/ by/ studyskills/ writing-skills- 1.shtm). It is perhaps for this reason that Crismore and Farnsworth describe the idea that academic writing is simply the impersonal statement of facts as a “dangerous myth” (1990: 118). The other point worth considering is that, by not explicitly presenting or claiming their opinions and observations as such, academic authors in <?page no="107"?> 107 some ways evade responsibility for them. Unfortunately, academics have not always behaved so scrupulously as to make these points of no concern. Irresponsible, unchallenged science is not an appealing concept, and authors who try to hide themselves completely in their texts do little to counteract, if they do not even encourage this. Obviously, the nature of scientific responsibility and accountability is far more complex than just avoiding, or employing, person reference, as Ziman points out: In pursuit of complete ‘objectivity’ - admittedly a major virtue - the norm [of disinterestedness] rules that all research results should be conducted, presented, and discussed quite impersonally, as if produced by androids or angels. […] this ‘no ethics’ principle is not just an obsolete model that can be uninstalled by a keystroke. It is an integral part of a complex cultural form. (1998: 1813) A useful in-depth analysis of connected issues, with recommendations, can be found in Evers (2001), which is a report of the International Council for Science (Standing Committee on Responsibility and Ethics in Science). It is not the role of this study to judge the way scientists present their research, nor can any recommendations be made here. For the purposes of the present discussion, it is enough to note that, from a critical discourse analysis point of view, these issues make the investigation of person reference especially pressing and relevant. Ultimately, regardless of how academics are supposed to write, in all three cultures examined here there are nonetheless conditions under which the identity of the academic author steps into the foreground. However, the force of this pronoun prohibition, or the stringency with which it is observed by academics, varies across cultures, and also across disciplines and text type. As far as cultural differences go, numerous studies have noted that German academics write less personally than their Anglo-Saxon colleagues. Kussmaul (1978) noted that, in humanities articles, Germanspeaking academics referred to themselves markedly less often than their English-speaking colleagues did. Trumpp (1998) recorded the same trend in a corpus of Englishand German-language sports science texts. In the RA subcorpus, Trumpp found that the English-language texts contained significantly more (p<0.05) first person singular and plural pronouns than the German ones; the frequencies were roughly three times higher (1998: 100-1). In a study of Englishand German-language reviews, Hutz found that the English-speaking authors referred to themselves more than twice as often as their German colleagues did (2001: 124). Hutz also found that while 21 of the English-speaking academics whose reviews he analysed used ‘I’ to mark statements of personal opinion, only 10 of the 30 German authors used the first person pronoun at all (2001: 124). Hutz attributes these differences to the fact that “in German-speaking academic circles, the <?page no="108"?> 108 attitude still dominates that personal statements generally work against the intended effect of objectivity, neutrality and therefore scientificness.” (my trans.) (Hutz 2001: 123). Both Buhl (1999) and Busch-Lauer (2001) came to similar conclusions - German academics write in a less explicitly personal fashion - based on an analysis of extracts from German and English-language physics, medicine and linguistics articles. 44 Like Hutz, Buhl equates the lower level of personal pronoun use with a higher level of content-orientation in the German-language texts (Hutz 2001: 123; Buhl 1999: 138). In an overview of previous contrastive work in intercultural pragmatics, House (1999: 47-8) argues that English speakers prefer interpersonally-oriented discourse strategies, through which they can occupy a perspective directed towards the partner’s point of view (addressee orientation). German speakers, on the other hand, tend to prefer contentoriented interaction strategies orienting in the speaker’s point of view (self orientation). At the same time, House characterises German academic discourse as “I oriented” and English-language academic discourse as “addressee oriented”, though this does not result in German academics having a more explicit personal presence in their written texts (1999: 49). As regards differences across disciplines, numerous studies have noted considerable variation in personal pronoun frequency across disciplines (Busch-Lauer 2001, Hyland 2002a, Breivega, Dahl & Fløttum 2002, Harwood 2005a, 2005b). In emerges that ‘hard knowledge’ natural science authors generally employ fewer personal pronouns than their ‘softknowledge’ humanities and social sciences colleagues; these differences appear to hold across languages. As we shall see below, although SCEGAD contains only soft-knowledge research articles, there are significant differences across disciplines even within this subgroup. These results suggest that the hard/ soft dichotomy is too simplistic a categorisation, and does not capture the complexity of disciplinary and indeed sub-disciplinary variation in academic writing practices. This possibility is acknowledged by the concept of ‘discourse communities’ as applied to academic disciplines by Swales (1990) and Hyland (2000), which allows for a more differentiated approach to cross-disciplinary differences in academic discourse conventions. It may be that disciplinary variation is greater within the humanities than within the natural and social sciences. A study comparing two social sciences disciplines (phonetics and psychology) found no significant differences ( Mart í n 2003), though more research is needed here. 44 Interestingly, Busch-Lauer notes that the German academics’ preference for impersonal presentation extends to their summary of other researchers’ work. In contrast to the English-speaking authors, who couched literature reviews in phrases such as ‘X [name of author] suggests…’, the German authors in Busch-Lauer’s study preferred depersonalised phrases such as ‘The study suggests…’ (2001: 162). <?page no="109"?> 109 Several studies have also noted the influence of text type upon personal pronoun use in written academic discourse. As one would expect, personal pronouns appear to be more frequent in more involved/ informal text types such as book reviews (Gläser 1990; Hutz 2001). One suspects that the same may be true of academic textbooks, but since the vast majority of studies focus on research articles only, there is a need for more research here too. Before moving on to the results from the corpus, it is worth briefly mentioning some of the alternative strategies academic authors employ. Instead of using person reference, both German and English-speaking scholarly authors turn to the passive, and impersonal or attributive constructions. The use of impersonal constructions as a way of avoiding person reference is described by Weingarten (1994: 128) as the “Metapher des aktiven Textes” [metaphor of the active text]. German authors in particular make use of the impersonal pronoun ‘man’. The English equivalent, ‘one’ is used only extremely rarely by the SCEGAD authors. Studies of academic discourse in English and German (Graefen 2000b; Hutz 2001), English and Ukranian (Zhabotynska 2001) and English and Esperanto (Fiedler 1992) suggest that person reference and the passive occur in inverse proportions in academic writing. A study by Seoane-Posse (2004) on the diachronic development of passive frequency in British and US- American scientific texts showed that, although the US-American texts demonstrated the largest decrease in passive frequency (38% versus 27% in the British texts), the increase in first and second person pronouns was in fact smaller in the US-American texts (7.77% increase versus 9.76% increase in British texts). Seoane-Posse therefore concludes that the decrease in the use of the passive in scientific texts during the 20 th century was not caused by a change in style towards either a more subjective, more emotive or more author-centred discourse. 45 It is not possible to draw this conclusion based on these data (correlation is not the same as causation), and in the absence of a correlation analysis (that there are significant differences between the British and US-American texts does not tell us anything about the relationship between pronoun frequency and passive). There is also any number of other intervening variables which could be responsible for Seoane-Posse’s results, whose effect, in the absence of a regression analysis or similar statistical test, it is impossible to discern. Unfortunately, this hypothesis cannot be tested on the larger data sample 45 Seoane-Posse’s results were based on an examination of 80 2,000 word extracts from diverse scientific genres and disciplines contained in the FLOB/ LOB and Frown/ Brown corpora (total 160,000 words). <?page no="110"?> 110 contained in SCEGAD. 46 We now move on to the results for person reference. 3.4 Person reference according to language, variety, culture, discipline, gender and other variables In order to have a clearer picture of person reference in English and German humanities texts, personal pronouns in SCEGAD were analysed according to both form and discourse function. The pronouns analysed are shown in Figure 3.1, grouped according to formal grammatical categories. Figure 3.1. English and German-language pronoun forms analysed in the corpus ______________________________________________________________________ English German _________________ _________________ 1st person singular Nom. I ich Acc. me mich Gen. my/ mine mein/ e/ r/ n/ s/ es Dat. me mir Refl. myself mich 2nd person singular/ plural Nom. you Sie * Acc. you Sie Gen. your Ihr/ e/ r/ n/ s/ es Dat. your Ihnen Refl. yourself/ selves sich 1st person plural Nom. we wir Acc. us uns Gen. our/ s unser/ e/ r/ n/ s/ es Dat. us uns Refl. ourself/ selves uns ______________________________________________________________________ * This is the polite form of the German second person pronoun: the familiar form, ‘du’, would not be used in a formal context such as an academic text. The analysis concentrated on first and second person pronouns, because these have the clearest interactive and identity construction purposes in the discourse. Because of their “inherent instability of reference”, first and second person pronouns are highly context-dependent in meaning and use (Wales 1996: 51), and therefore well-suited to a combined corpus linguistic and discourse analytic analysis. Hyland (2005) terms second 46 Since the SCEGAD corpus is not grammatically tagged, in order to explore the relationship between person reference and the passive in the corpus texts, all passive constructions would have to be identified manually. In a corpus of one million words, this is unfortunately hardly feasible. <?page no="111"?> 111 person pronouns the “reader pronouns”, and notes that they are “the clearest way a writer can acknowledge the reader’s presence” (2005: 182); similarly, Wales (1996: 3) labels first and second person pronouns the “interpersonal pronouns”. Third person pronouns were not included in the analysis, because they generally do not perform an interpersonal function. However, third person singular addresses to the reader along the lines of ‘the reader may well wonder…’ were included in the analysis, as were oblique authorial self-references in the third person such as ‘the author wishes to thank x’ or ‘der Forscher wurde aufgefordert’ [the researcher was asked]. The English-language tokens searched for were ‘author’, ‘writer’, ‘reader’, ‘researcher’ and ‘scholar’. In German, the tokens were ‘Autor’ [author], ‘Verfasser’ [author], ‘Leser’ [reader], ‘Forscher’ [researcher] and ‘Wissenschaftler’ [scholar]. Only those tokens were counted which actually referred to the author or reader of the text, not to authors, scholars or readers in general, or some other researcher or reader. In addition, discipline-specific third person self-references to the author as a ‘philosopher’, ‘linguist’, ‘folklorist’, ‘ethnographer’, ‘historian’ etc were also counted in the analysis. Again, references to historians in general, or a specific historian other than the author, were discarded. All other instances of third person singular references or pronouns which did not construct authorial identity, or express the relationship between author and reader, were excluded from the analysis. 47 All figures below labeled third person singular thus refer to such reader addresses and self-references only, and not to third person singular personal pronoun forms. Other points to note are that for both languages, the second person singular and plural pronouns are grouped together, since they have the same form. In addition, although generic uses of ‘you’ have a less directly marked interactive purpose than the direct ‘you’ address to the reader, they were still included in the analysis, and in fact account for nearly two thirds of all second person singular pronouns in the English-language texts (62 of 97 occurrences). It is also worth noting that since German has many more reflexive verbs than English does, the frequency counts for the Germanlanguage texts are in a sense slightly inflated. Reflexive pronouns account for 4.63% of the German first person plural counts (21 of 454 occurrences) and 10.25% of the first person plural counts (76 of 779 occurrences). The results of the quantitative analysis are presented as mean occurrences per 47 Examples of third person self-reference which appeared to have been added by the editors were also excluded: in one German journal, for example, the author’s address was at the bottom of each article, captioned “address of the author”; this practice also appeared in several English-language articles. The same German-language journal also headed the acknowledgements section, where present, with “author’s note”. The uniformity strongly suggests that these examples are due to editorial policy, not authorial choice. <?page no="112"?> 112 10,000 words, since the average length of the SCEGAD texts is 8,937 words. 3.4.1 Person reference according to language Figure 3.2. Person reference: English and German-speaking authors compared (frequency per 10, 000 words) ______________________________________________________________________ f English f German Significance (Mann Whitney) ______________________________________________________________________ 1st p. sing. pronouns 28.41 (n=1322) 11.76 (n=454) 0.003 2nd p. sing. pronouns 3.71 (n=97) 0.00 (n=0) na * 3rd p. sing. reference 0.42 (n=26) 0.14 (n=5) na * 1st p. pl. pronouns 26.52 (n=1188) 20.91 (n=779) 0.042 TOTAL 59.06 (n=2633) 32.85 (n=1239) 0.019 ______________________________________________________________________ * Frequencies in the German subcorpus too small for the Mann Whitney test to be reliable As can be seen in Figure 3.2, the English-speaking authors represented in SCEGAD consistently use more person reference in their writing than do the German authors. The frequencies for the English-speaking authors are very similar to those found in Hyland (2003), which examined comparable phenomena in a 1.5 million word corpus of published research articles. 48 The difference between the two language groups is significant at the p<0.05 level: this means that the two language groups are different, and it is very unlikely (less than a five in one hundred chance) that this difference is due to chance. In other words, we can be 95% sure that the differences observed here between the German and English-speaking academics are not due to chance. For the first person singular pronouns, the difference is even more highly significant: here, there is a 3 in 1,000 chance that the difference between the means of the two groups is coincidental. The Mann Whitney test was not applied for the second person pronouns and third person reference, since it is not suitable for frequencies lower than five. However, a significance test was deemed unnecessary given the fact that the difference between the German and English-speaking authors was so great in these two cases. The difference across the two language groups is striking in the frequency of second person pronouns. While, in the English-language 48 Hyland (2003) examined the tokens “I, me, my, we, us our”, authorial self-citation and third person self-reference: the average occurrences were 47.9/ 10,000 words across texts from the “soft sciences” subcorpus, which included marketing, philosophy, applied linguistics and sociology (2003: 253). <?page no="113"?> 113 subcorpus, the authors directly addressed the reader as ‘you’, or used the generic ‘you’, on average 3.71 times per 10,000 words, in the entire German subcorpus there was not a single second person singular pronoun. The German authors did not directly address the reader with ‘you’ at all, and preferred to perform the generic ‘you’ function with the indefinite pronoun ‘man’, as can be seen by comparing the following typical examples: (20) In 1946, just after the ending of the Second World War, you can still find the poet Auden … (LTUK) (21) Man findet diese Variante, soweit ich sehe, nicht in der Novellistik, sondern im Roman. [One finds this variant, as far as I can see, not in the novella but in the novel] (LTDE) (22) So, you cannot make the conjunction of LOT and CT compatible with psychology. (PHUS) (23) Man kann den Widerspruch durch die Annahme vermeiden , daß L- Andrea und R-Andrea identisch sind. [One can avoid the contradiction by assuming that L-Andrea and R-Andrea are identical] (PHDE) The relatively frequent use of ‘man’ in German-language academic texts was documented by Trumpp, who found that the indefinite pronoun was used over five times more frequently (p<0.001) in the German-language compared to the English-language subsection of her corpus (1998: 191). As two ways of fulfilling a very similar discourse purpose, it makes perfect sense that the indefinite and the second person singular pronoun should occur in roughly inverse proportions. In addition to a higher frequency of direct reader address, the Englishspeaking authors in the corpus were almost four times as likely as their German colleagues were to address either themselves or the reader in the third person (‘my reader’, ‘the reader’, ‘the author’, ‘der Verfasser’ [the author]). Unfortunately, the frequencies were too low for this difference to be tested for significance. Third person references to the reader served a similar discourse purpose across all three cultures, enabling the author to establish a dialogue with the reader without the need for direct address (‘you’/ ‘Sie’): (24) The reader may well wonder how considerations concerning informational transactions can, as here, be supposed to be at all relevant to the present issue (PHUK) (25) Der folgende Abschnitt wurde in der Absicht geschrieben, den Leser zu einer rekonstruktiven Überprüfung … zu bewegen. [The following <?page no="114"?> 114 section was written in order to move the reader to a reconstructive examination of …] (PHDE) It is worth remembering that direct reader address did not occur in the German subcorpus at all: though the German authors too entertain a dialogue with their readers, direct address is probably perceived as being even more, or too much, involved. Third person references to the author served a variety of different discourse purposes, though no one purpose appeared exclusive to or more typical of a particular linguistic or cultural group (this type of self-reference is discussed further in section 3.7). As mentioned above, of the person reference types tested for statistical significance, the most highly significant difference between the two groups lies, however, in their use of the first person singular pronouns: the English-speaking authors are almost 2.5 times more likely to employ this type of person reference than are their German colleagues (p<0.005). Finally, when it comes to first person plural pronouns, the two language groups are also significantly different at the p<0.05 level, with the Englishspeaking academics demonstrating higher frequencies also of this type of personal pronoun. In sum, this means that English and German-speaking academics used significantly different amounts of all forms of person reference analysed. There is a consistent trend for the English-speakers in SCEGAD to employ more person reference in their articles than the Germans do. The prototypical profile for the Englishand German-language texts, which is given in Figure 3.3, shows that while both language groups follow a similar pattern, with the first person pronouns more frequent than other types, there are no points at which the two groups completely converge (the graph point for third person reference is deceptive: see the numbers in Figure 3.4). Again, when interpreting the graphs it must be born in mind that the figures for the third person refer not to personal pronouns, but only to indirect addresses to the author or reader in the third person. The figures must not be interpreted as a commentary on all types of personal pronoun in the corpus texts, because they are only concerned with those uses of personal pronouns, and of third-person reference, which construct the authorial identity or the author-reader relationship. The results for all pronoun tokens (including ‘it’, ‘he’, ‘she’, ‘they’ etc) would be much different. Figure 3.3 shows that the comparative degree of difference between the two language groups is much greater for the first person singular than for the first person plural pronouns. In addition, while the English-speaking academics write ‘I’ and ‘we’ with almost equal frequency, for the Germanspeaking academics the situation is very different. The German authors employ first person plural pronouns far less frequently than do the English- <?page no="115"?> 115 speakers, and this in turn is nearly twice as frequently as they use first person singular pronouns. It seems that, for the German academics, the ‘Itaboo’ is alive and well. No such taboo seems to operate for their Englishspeaking colleagues: they are even slightly more likely to refer to themselves explicitly as individuals than as members of a group. Figure 3.3. Person reference: prototypical profiles for English and German-speaking authors compared (frequency per 10,000 words) 0.00 5.00 10.00 15.00 20.00 25.00 30.00 1st p.sing. pronouns 2nd p.sing. pronouns 3rd p. sing. reference 1st p. pl. pronouns English German Figure 3.4. Person reference: comparative difference in frequency for Englishand German-speaking authors (frequency per 10,000 words) ______________________________________________________________________ f English f German Difference in f ______________________________________________________________________ 1st p. sing. pronouns 28.41 11.76 16.65 2nd p. sing. pronouns 3.71 0.00 3.71 3rd p. sing. reference 0.42 0.14 0.28 1st p. pl. pronouns 26.52 20.91 5.61 TOTAL 59.06 32.85 26.21 ______________________________________________________________________ There is a close relationship between different types of pronouns, with the correlation coefficient (Spearman Rho) for all types significant at the p<0.01 level. This means that authors who use more of one type of pronoun will also use more of all the others, and suggests that authors do not distinguish between types. The German authors not only have an ‘I taboo’, as researchers have stated, but a general ‘person reference taboo’. The English-speakers are not (only) self-advertisers, but partners in a dialogue: if they explicitly refer to themselves more often, they also address the reader more often as well. These are the differences between <?page no="116"?> 116 the two language groups, but what of differences between the two varieties of English contained in the corpus? 3.4.2 Person reference according to variety As can be seen in Figure 3.5, in general, the US-American authors represented in SCEGAD are somewhat more likely than their British colleagues are to use person reference, with the British authors providing 43% of all instances found in the English-language corpus, and the US- Americans 57%. Figure 3.5. Person reference: British and US-American authors compared (frequency per 10, 000 words) ______________________________________________________________________ f British f US-American Significance English English (Mann Whitney) ______________________________________________________________________ 1st p. sing. pronouns 27.95 (n=609) 28.89 (n=713) 0.327 2nd p. sing. pronouns 0.71 (n=18) 6.71 (n=79) 0.359 3rd p. sing. reference 0.16 (n=5) 0.71 (n=21) 0.042 1st p. pl. pronouns 21.13 (n=496) 31.91 (n=692) 0.148 TOTAL 49.95 (n=1128) 68.21 (n=1505) 0.233 ______________________________________________________________________ There does appear to be some consciousness of the possible intervarietal difference here. For example, a British university postgraduate style guide comments that British academic English probably uses more passive/ impersonal forms than American English but this is in factual/ objective description sections personal pronouns may be acceptable in conclusions/ comment sections. (Du Boulay: 2001) However, despite appearances, the difference between the British and US- American authors is statistically significant only when it comes to third person references to the author and reader. This may serve as a reminder for anyone inclined to neglect statistics in favour of human observation. The US-American authors were significantly more likely to address the reader, and refer to themselves, in the third person than were their British colleagues. However, the US-American scholars were not significantly more likely to address the reader directly with ‘you’. In order to clarify this situation, the same pronoun types were compared for frequency in the Jsection (Learned and Scientific Writings) of the FLOB and FROWN corpora. The J-section subcorpora contain 80 approximately 2,000 word extracts of scientific publications. Each J-section subcorpus in FLOB and <?page no="117"?> 117 FROWN therefore has a total of roughly 160,000 words. The same method was followed as for SCEGAD: references to authors and readers who were not the actual author and reader of the text in question were discarded, as were examples occurring as part of quoted material. However, it must be noted that the FLOB and FROWN results are not directly comparable to those from SCEGAD. While SCEGAD contains whole texts from the humanities only, and of a single academic text type, the FLOB and FROWN J-sections contain only extracts, rather than whole texts, taken from a large variety of both soft and hard-sciences disciplines, and not controlled for text type, gender, author age or status. In such a small set of data, all of these unaccounted-for variables skew the results. For example, one of the FROWN scientific texts (J69) is a computer programming manual, which demonstrates roughly 20 times the average frequency of generic ‘you’ forms present in the other text types. The source of each of the FLOB and FROWN text samples is given, and it would therefore be possible in theory to code the texts for all of these variables, but the corpora are so one-sided (many text types are sampled only once) and contain so many variables other than author culture, that the results would be most likely meaningless. For example, there are only 10 women authors sampled in the FROWN J-section, alongside many texts by groups of authors or corporate authors: this makes it near impossible to investigate the effect of gender on language use based on these data. Similarly, the 80 texts in the FROWN J-section cover at least nine text types and 24 disciplines: there are simply too few examples of each genre and discipline for any conclusions to be drawn about text type of disciplinespecific variation. It is also not the case that the two J-sections sample the same number of texts from any of these variables (genre, discipline, author gender etc). The FLOB and FROWN corpora therefore give a very broad picture of academic discourse in British and US-American English. It is, however, impossible to discern whether any differences between the two subcorpora are due to author culture or any of the other independent variables. Nonetheless, the data are given in Figure 3.6. <?page no="118"?> 118 Figure 3.6. Person reference: FLOB and FROWN corpora compared (frequency per 10, 000 words) ______________________________________________________________________ FLOB FROWN Significance (Mann Whitney) ______________________________________________________________________ 1st p. sing. pronouns 7.56 (n=120) 5.75 (n=92) 0.687 2nd p. sing. Pronouns 1.44 (n=23) 1.56 (n=25) 0.262 3rd p. sing. reference 0.00 (n=0) 0.00 (n=0) 0.00 1st p. pl. pronouns 35.06 (n=561) 42.08 (n=673) 0.365 TOTAL 44.06 (n=704) 49.38 (n=790) 0.328 ______________________________________________________________________ Unfortunately, the FLOB/ FROWN comparison is also inconclusive. Neither the FLOB nor the FROWN corpus J-sections contain a single instance of third person reference to the writer or reader of the text, thereby preventing those SCEGAD results from being tested, for which there was a statistically significant intervarietal difference for this type of person reference. As in SCEGAD, it again appears as though there is a slight tendency for the US-American authors to employ more person reference, but the trend is even less pronounced than in SCEGAD and not statistically significant for any of the pronoun types. Although other large corpora of academic prose exist, for example Hyland’s 1.5 million word corpus of published research articles (Hyland 2003), unfortunately, they are not controlled for different varieties of English, and so cannot be used to test hypotheses about possible British/ US-American differences. When frequency alone is considered, there are therefore highly significant differences across languages, but only marginal statistically significant intervarietal differences in the use of person reference. We now move on to an analysis of person reference across cultures. 3.4.3 Person reference according to culture The intercultural comparison between British, US-American and Germanauthored texts, seen in Figure 3.7, reveals an unexpected pattern. While the profile for the US-American and German authors is very different, the British authors hover somewhere in-between the two groups, at some points indistinguishable from the US-Americans, and at others, almost identical to the German authors. The English-speakers are nearly identical in their use of first person singular pronouns: the ‘I-taboo’ really is peculiar to the German academics. However, when it comes to using first person plural pronouns to claim group identity or express solidarity with a group, the British authors behave like their German colleagues, and less like the <?page no="119"?> 119 US-Americans. It is also worth noting that, when the relative frequency of first person singular and plural pronoun use is compared, it is the US- Americans who are the piggy-in-the-middle. Here the US-Americans more closely resemble the German authors than their fellow English-speakers, in that they slightly more frequently claim group identity than personal identity, though the difference is nowhere near as great as it is for the German academics. The gross difference between the US-Americans and the other two cultural groups when it comes to the second person pronouns is largely due to one US-American philosophy text, which, though it is only 3, 510 words long, contains nearly 50 direct addresses to the reader. A typical passage: (26) But, according to CT, you cannot have a lcatl [sic] until you have the ability to detect cats. According to psychology, you cannot have the ability to detect cats until you have a theory of cats. According to LOT, you cannot have a theory of cats until you have lcatls. So, you cannot make the conjunction of LOT and CT compatible with psychology. (PHUS5) In contrast, there is not a single example of such an address in the entire German sub-corpus. Were this one article to be disregarded, the British and US-American authors would in fact show extremely similar results for second person pronoun frequency. Figure 3.7. Person reference: prototypical profiles for British, US-American and German authors compared (frequency per 10, 000 words) 0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 1st p.sing. pronouns 2nd p.sing. pronouns 3rd p. sing. reference 1st p. pl. pronouns British English US-American English German <?page no="120"?> 120 Figure 3.8. Person reference: British, US-American and German-speaking authors compared (frequency per 10, 000 words) ______________________________________________________________________ f British f US-American f German ______________________________________________________________________ 1st p. sing. pronouns 27.95* 28.89 11.76 2nd p. sing. pronouns 0.71 6.71 0.00 3rd p. sing. reference 0.16 0.71 0.14 1st p. pl. pronouns 21.13 31.91 20.91 TOTAL 49.95 68.21 32.85 ______________________________________________________________________ * The two most similar values are highlighted The picture here is more complex than for the comparison between all English-speaking with the German-speaking academics. Certainly, the general trend observed above holds: the English-speaking authors, both British and US-American, are consistently more likely to employ person reference than their German colleagues. However, the US-American authors employ person reference not only more often than do their German colleagues, but also somewhat more often than do their British colleagues, though, once again, it must be borne in mind that the differences in frequency between the British and US-American scholars for the second person pronouns and third person reference are not significant. The similarity between the British and US-American scholars is greatest in the case of the first person singular pronouns: the low frequencies for the German scholars may well be evidence of the ‘I-taboo’ in action. For all of the other types of pronouns examined, the US-American academics have little in common with either of the other cultural groups: they write in a unique fashion. Apart from the first person singular pronouns, the frequencies for the British academics more closely resemble those of the German authors rather than the US-Americans. It appears that the typological similarity of British and US-American English does not translate into a similarity in discourse norms in this respect. The pedagogical implications of this finding are potentially far-reaching. There are two major tendencies which have emerged through both the inter-lingual and inter-cultural comparisons. First: both British and US- American scholars employ person reference more frequently than German academics do. Second, German authors are relatively unlikely to employ explicit person reference in the form of first person singular pronouns: this appears to be the ‘I-taboo’ in action. In sum, then, it seems that, just as authorship in English academic writing “carries a culturally constructed individualistic ideology” (Hyland 2002c: 1110), so in German it seems to carry a - similarly constructed - less personal and more collectivistic ideology. The cross-linguistic and cross-cultural comparisons are, however, <?page no="121"?> 121 not the only ones possible with the SCEGAD data: in the following, some of the other factors which influence this aspect of language use will be examined. 3.4.4 Person reference according to discipline Since the SCEGAD corpus is balanced not only for language and culture, but also for discipline, it is also possible to compare the mean frequency (per 10,000 words) of person reference across the five disciplines in the corpus: philosophy, history, folklore, literary studies and linguistics. The results for the British and US-American scholars are represented in a single column, since the differences between the two English-speaking groups were marginal (see section 3.4.2 above). Figure 3.9. Person reference according to discipline: bar graph (frequency per 10,000 words) 0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00 160.00 PHIL LING FOLK LIT HIST English Texts German texts Figure 3.10. Person reference by discipline: comparative difference in frequency (frequency per 10, 000 words) ______________________________________________________________________ f English f German Difference in f ______________________________________________________________________ Philosophy / Philosophie 148.76 77.79 70.97 Linguistics / Sprachwissenschaften 68.94 20.25 48.69 Folkore / Volkskunde 40.48 34.37 6.11 Literary Studies / Literaturwissenschaften 25.95 7.39 18.56 History / Geschichte 11.16 24.44 13.28 AVERAGE 59.06 32.85 26.21 ______________________________________________________________________ <?page no="122"?> 122 The results across disciplines show that, with the exception of history, frequencies are consistently higher in the English-language texts across all disciplines, which is what we would expect, based on the overall results examined above. However, within this general trend, there is great variation across disciplines, with some interesting individual results: the crossdisciplinary differences between language groups for total person reference frequency are highly significant in the linguistics subcorpus (p<0.01) and significant in both the philosophy and literary studies subcorpora (p<0.05). The high degree of disciplinary variation is not unexpected, for there is a growing number of studies which confirm the crucial influence of discipline on person reference, as well as on engagement, involvement and hedging (Busch-Lauer 2001; Breivega et al. 2002; Harwood & Hadley 2004; Hyland & Tse 2004; Harwood 2005b). A word of warning, therefore: any study of academic discourse which does not control for the variable of discipline will paint a distorted and incomplete picture of the phenomena being analysed. The smallest differences in frequency between the two languages are in the history and folklore texts, with the German historians in fact using person reference even more frequently than do their English-speaking colleagues, though this difference is not statistically significant (p<0.05). These unusual results are explained by the effect of two influential intervening variables - status and gender - which, as we shall see further below, are powerful predictors of pronoun frequency (higher status and male authors use more pronouns). In the case of the history subcorpus, the high results in the German sample were mainly due to unusually high frequencies in two texts by Emeritus professors of history, who also happened to be male. Likewise in the folklore subcorpus, three German texts showed unusually high levels of person reference: these were the texts written by the three highest-status authors, all of whom were also male. The results for history were particularly influenced by these two variables because the discrepancy between the English and German subcorpora for this discipline was greater than between the two folklore subcorpora. While the English-language subcorpus sampled only one historian with the academic status level of ‘professor or above’, the German sample contained four such senior researchers, and it was precisely the two researchers highest in status who used the most pronouns. Similarly, while the English-language history subcorpus contained equal numbers of men and women, the German subcorpus contained only three female authors. This is three times the proportion of female authors who are published in the actual journal (Die Historische Zeitschrift): of the 36 articles published in 2003 and 2004, only four were by female historians (11%). While it is not possible to create a corpus that is balanced for every single variable in every <?page no="123"?> 123 subcorpus (in this case because the required texts simply do not exist), the fact that each SCEGAD text is marked for extra-linguistic variables means that it is possible to track the influence each variable has on written production. What looks like an exception to the general tendency - that English-speakers demonstrate higher frequencies of person reference - emerges not to be one at all, and instead can be seen to be due to the influence of intervening variables. The most significant difference between languages was found in the linguistics texts (p<0.01). This is interesting, because the only reason linguistics texts were included in the SCEGAD corpus was because I wanted to investigate whether German linguists were behaving in the same way as all their other humanities colleagues, or if reading and publishing so much in English had somehow ‘rubbed off’ on them. The hypothesis that lingua franca English dominance covertly influences the discourse norms of other languages has been posed in a research project led by Juliane House: the research project found no evidence of such influence on German-language texts (Baumgarten, House & Probst 2001; House 2003a). The SCEGAD data support this finding. It appears that nothing has rubbed off: the German linguists do not even approximate English-speaker academic behaviour when it comes to this feature of scholarly writing, but retain their own, culturally specific writing patterns. The difference in person reference frequency between the German and English-speaking authors is also significant in literary studies (p<0.05). Literary studies is the discipline in which the German authors employ the least, and the English-speakers the second least person reference, which is surprising, since it is certainly not the least subjective of the disciplines examined here. The literary studies scholars sampled in SCEGAD of course express opinions in these texts, but they comparatively seldom mark these opinions with personal pronouns. Contrast the following pairs of examples, the first in each pair taken from a literary studies text, the second from another discipline: (27) But the two positions… may have more in common than one might initially wish to allow. (LTUS) (28) In our final analysis, we might think of Ryder's diary as thirty-five years of psychiatrist's notes… (HIUS) (29) Betrachtet man die Lyrik des Barock unter diesem Aspekt, dann fällt auch dort die Tendenz auf… [If one examines Baroque poetry in this regard, then the tendency makes itself noticed that…] (LTDE) (30) Sehen wir von dem genannten „Mißverständnis“ ab und betrachten wir die grundsätzlichen Deutungsmuster der Theologen, so… [If we ignore <?page no="124"?> 124 the misunderstanding mentioned above and if we examine the basic interpretative patterns of the theologists, then…] (FLDE) By far the most striking result, however, is the high frequency of person reference in the philosophy subcorpus, which was noted also by Hyland in his study of interpersonal metadiscourse (2000: 114). When all disciplines are compared to each other, the only statistically significant difference which emerges is that between the philosophy subcorpus and all other subcorpora. This might prompt the question: what on earth is going on in philosophy? The frequency of person reference in philosophy texts in both languages is more than double the average across all disciplines. In the English-language subcorpus, this difference in total pronoun frequency between philosophy and all other disciplines is significant when compared to all other disciplines except for linguistics (p<0.005). In the Germanlanguage subcorpus, the difference in total pronoun frequency between philosophy and all other disciplines is significant when compared to all other disciplines except folklore (p<0.005). The singularity of philosophical writing is underscored by the results of a study of English-language academic writing by Hyland (2000). Here, the philosophy texts contained more than double the cross-disciplinary average frequency of interpersonal metadiscourse, principally because of heavy use of personal pronouns (2000: 114). This would all seem to suggest that a cross-linguistic feature of philosophy texts might be responsible for the observed pattern, though this feature is not influential enough to iron out the large differences between languages. When the results are broken down into frequencies for each type of person reference, it emerges that the high level of person reference in the philosophy articles is related to the use of different types of person reference in the two languages. The English-speaking philosophers employ first person singular and plural forms with very similar frequency (67.14 1st p. sing. / 64.42 1st p. pl.), and use both types far more often than their colleagues in all other disciplines except for linguistics (p<0.005). The German philosophers, on the other hand, employ significantly more first person plural pronouns than first person singular pronouns (p<0.05), and do not use first person singular forms significantly more often than their compatriots do in any of the other represented disciplines. One could therefore say that, even for the German philosophers, the ‘I-taboo’ still operates. The German philosophers write ‘wir’ (‘we), ‘uns’ (‘us’), ‘unser/ e/ r/ n/ s/ es’ (‘our’) and ‘uns’ (‘ourselves’) far more frequently than do the German authors in the other four disciplines, and exactly as often as the Englishspeaking philosophers. The cross-cultural similarity therefore lies in the level of first person plural pronouns only. <?page no="125"?> 125 The main cross-linguistic reason for the comparatively high occurrence of first person plural pronouns in this discipline is the tendency to generalise. Both the Englishand German-language philosophy texts in the corpus were often concerned with general statements about the nature of human perception and thought, and the authors therefore refer a great many times to ‘we’, meaning ‘we humans’. Two extracts from corpus texts illustrate this typical feature: (31) For this attack, whether it be in Berkeley or the sense-datum theorists, precisely focuses on the phenomenal character of the conscious experiences that we enjoy when we perceive the world. There is supposed to be something about the phenomenology of sense-experience (crucially, of sight, which almost all take to be central) which entails that we are never directly aware of the physical world. (PHUK) (32) Denn wäre nicht auch das reale Außer-uns der Objekte Inhalt unserer Wahrnehmung, so wäre es für uns nicht nur nichts, sondern wir würden die Objekte auch nicht an einem anderen Ort, als wo wir uns befinden, sehen. [For if the real outside-ourselves of the objects were not also the content of our perception, then it would not only be nothing to us, but we would also not see the objects in a different place from where we are.] (PHDE) However, while such examples constitute the vast majority (77.30%) of all instances of first person plural pronouns in the German texts, they do not constitute the majority in the English-language texts. Here, by far the most common type of first person plural referent is not the ‘human we’, but the group of academics, or the ‘we’ of the author and the reader. These referents account for 65.05% of all cases of first person plural pronouns in the English-language texts, whereas ‘human we’ accounts for only 12.97% of the occurrences. Two illustrative examples from the English-language subcorpus show how the English-speaking authors use first person plural pronouns not to make general statements, but to express solidarity with the reader: (33) Thus the sentence ‘- (A - -A)’ leads to the contradiction ‘-A . -- A’, whence we may infer ‘-- (A - -A)’. If we could delete double negations, we should obtain the law of excluded middle. So, since double negation elimination fails intuitionistically, we can rule it out that any sentence is neither true nor false, but we cannot affirm that every sentence is either true or false. (PHUK) (34) We can, if we like, conjoin or disjoin statements about leptons with statements about trees. If we bring mathematics in, we can count both leptons and trees. (PHUS) <?page no="126"?> 126 The Englishand German-speaking philosophers represented in the corpus are therefore not as similar as they may initially have seemed. It is true that both groups employ more person reference than their colleagues in the other disciplines do. However, for the German authors, this feature is due only to the high level of ‘human we’ first person plural pronouns, and can be overwhelmingly attributed to the nature of philosophical argument as it is represented in the corpus texts. The person reference in the German philosophers’ articles only very rarely fulfils an interactive function. For the English-speaking philosophers the situation is very different. They are much more likely than are their colleagues in other disciplines to use not just first person plural, but also first person singular pronouns. They do use first person plural pronouns with nearly identical frequency to the German philosophers, but these pronouns serve an interactive function in their texts: their use is not due to any characteristic of philosophical argument. This fits well with Hyland’s finding that, in his English-language philosophy texts, there was a relatively high frequency of inclusive ‘we’, which is used by the authors in his corpus to “invite readers into the discussion by drawing on shared general knowledge or the previous discourse to lead the reader through an argument” (2000: 127). This kind of generalised first person plural form was not employed with comparable frequency in any of the other disciplines Hyland studied. The same pronoun serves very different functions across cultures, and numbers, therefore, are not everything. While the purely quantitative results might have tempted us to declare a real cross-linguistic, discipline-specific feature (philosophers generalise with ‘human we’), it emerges that the influence of discipline on this feature in the corpus texts is marginal. The question which remains is why English-speaking philosophers write in such an explicitly personal and dialogic fashion. The unusually high frequency of both first person singular and plural pronouns in these texts is a feature uniformly represented across all articles in the corpus. At times, the argument in these texts more closely resembles a sort of stream of consciousness internal dialogue, as is illustrated by the following extract: (35) Next, and more important for what follows: Could that very vapour have been a solid or a liquid? I do not mean to ask whether the stuff that makes up the vapour comes in solid or liquid form. I know that it does. I mean to ask whether, say, T. S. Eliot’s yellow fog which “rubbed its back against the window pane” could have been or could come to be identical with a block of ice or a puddle of water. I do not mean to ask whether the fog could solidify into a block of ice or liquefy into a puddle of water. I know that it could. I mean to ask whether the fog could be <?page no="127"?> 127 either of these things. It seems to me that the fog could not be either of these things. (PHUS) When not carrying on a kind of personal conversation as in this extract, the authors ally themselves with the reader or with the community of their fellow philosophers, as in extracts (19) and (20), or use person reference to lead the reader through the text, as in Hyland’s study. The fact that the German philosophers do not behave in this fashion makes it unlikely that this feature is due to any innate characteristic or requirement of philosophical discourse: one can write a philosophy article perfectly well without recourse to any higher frequency of person reference than any other type of humanities academic. It appears that the English-speaking philosophers simply continue the culturally-specific trend evident in the work of their English-speaking colleagues in the other corpus disciplines. The English-speaking philosophers may be more extreme than their fellows are, but they use the same linguistic devices, in the same proportions, for the same purposes. In conclusion, the corpus analysis shows that disciplinary affiliation is not as decisive as cultural affiliation. Similarities within a discipline which, based on a purely quantitative analysis, at first appeared to be crosslinguistic in nature, emerge in a qualitative analysis to be due to different factors. The English-speaking authors in SCEGAD write in a more explicitly interactive fashion, and the philosophers demonstrate this tendency most strongly. The German philosophers do not write any more personally or interactively than do any of their humanities colleagues, and use first person plural pronouns only to express the kind of generalised statements central to the argument in their texts. The cultural differences we have already observed between English and German-speaking scholars remain constant. 3.4.5 Person reference according to gender Another variable for which the SCEGAD corpus is balanced is gender. Figure 3.11 shows a consistent trend across all sub-corpora, and across the corpus as a whole: the male authors employ person reference more frequently than do their female counterparts. This trend holds across all types of person reference investigated, though it is statistically significant only for ‘total pronouns’ (all pronoun types counted together). The two-tailed asymptotic significance value across all pronoun types is 0.044, using the Mann Whitney test (significant to the value of p<0.05). The gender difference for first person plural pronouns was very nearly significant (two-tailed significance value 0.058). <?page no="128"?> 128 Figure 3.11. Person reference by gender: bar graph (frequency per 10,000 words) 0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 German authors British authors US-american authors Whole corpus M ale Female Figure 3.12. Person reference by gender: comparative difference in frequency (frequency per 10, 000 words) ______________________________________________________________________ f men f women Significance (Mann Whitney) ______________________________________________________________________ German authors 35.01 30.95 0.826 British authors 74.03 16.15 0.011 US-American authors 77.42 57.29 0.285 Whole corpus 54.33 33.12 0.044 ______________________________________________________________________ The results in the three subcorpora (German, British and US-American authors) are striking: while the German academics appear to behave in a nearly unisex fashion, there is a huge intervarietal difference between the British and US-American scholars. The use of person reference is a stylistic difference in writing that is apparently far more highly marked for gender in British than in US-American English. Figure 3.13 shows the prototypical profiles for both genders, and all types of person reference, in the three cultures under examination. This graph further emphasises the singularity of the British results. The British authors show the most pronounced gender differences of all groups: the British men employ both singular and plural first person pronouns significantly more often than do the British women (p<0.05). By contrast, in the US-American subcorpus, it is the women who employ more first person self reference. However, none of the gender differences among the US- American scholars is statistically significant. The German academics too <?page no="129"?> 129 show no significant gender difference for person reference use. The crossgender differences, which are significant when measured across the corpus as a whole, are therefore in large part due to the pronounced discrepancy between the British men and women. Figure 3.13. Person reference by gender, cultural profiles (frequency per 10,000 words) 0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 1st p.s. pn 1st p.s. pn 2nd p.s. pn 2nd p.s. pn 3rd p.s. ref 3rd p.s. ref 1st p.pl. pn 1st p.pl. pn M F M F M F M F German Authors British Authors US- American Authors It seems as though, for female British academics, there is not only a sort of ‘I-taboo’, which operates even more strongly than it does for German authors, but in fact a sort of general ‘person reference taboo’, since the British women are less likely to write both ‘I, me, my’ and ‘we, us, our’ etc. than are the German academics. The only two authors in the entire corpus who employed no person reference at all were two British female historians. No such taboo operates for the US-American women scholars. The results in the British subcorpus are also not due to one or two outliers, but evidence of a consistent stylistic feature of the British women’s writing in the corpus. It is important to note that these results are not due to the intervening influence of the variable discipline: although the British female subcorpus contains no philosophy texts (the journal concerned, Mind, published no articles by female British academics in the years 2001- 2005), 49 even if all philosophy texts are excluded from the analysis, the 49 The journal published only two articles by female philosophers in these five years: one US-American and one German scholar. There were also three articles with multiple authors of mixed gender, which obviously cannot be used for gender comparison. It appears that the natural sciences are not the only male-dominated scientific domain. <?page no="130"?> 130 same trend still emerges. It is not possible to speculate here why there is such a huge difference across genders specifically for the British scholars. 50 However, it may be worthwhile considering some possible explanations for the general trend across all three cultural groups. There is little literature on gender-marking in written discourse, and even less specifically on academic discourse. 51 However, one possible reason for this gender-difference may be that the use of person reference is not innate to the author’s physical gender, but in some way connected to power. 52 This may be supported by the fact that, when a factor analysis was conducted on the person reference data, gender, age and academic status formed one factor. As we shall see in the next section, higher-status academics (professors) use significantly more person reference than lowerstatus scholars (non-professors) do. The female scholars in SCEGAD, particularly the British women, therefore demonstrate a feature (lower relative person reference frequency) associated with lower academic status. This is not reflective of any actual inferiority in the status of these women: of the British scholars sampled, 30.77% of the men and 30.00% of the women were professors. Despite factually equal academic status, the female British scholars in particular still adhere to an ‘I-taboo’ not felt by their male colleagues. However, since person reference can serve both authority- 50 Though see Morley’s paper on the situation of female academics in British universities, in which she notes that discrimination against women in the academy is still rife, and that British women scholars remain consigned largely to the lowest status and pay levels (1995). 51 By this I mean that there is little empirical, gender-contrastive literature (though see Hiatt 1977 for a notable exception). In the United States particularly, within the field of composition studies, there is a tradition of research into so-called ‘women’s ways of writing’. However, such research starts from the ideologically-motivated premise that there are “female types of thinking and writing” (Allison 1993: 108): this hypothesis is uncritically treated as a fact and never supported by any empirical evidence. There are a number of ethnographic studies of women’s writing along these lines (see for example Allison 1993; Cole 1997). The stereotypes reproduced therein, for example of women’s writing as circular, sensual and subjective (Allison 1993: 105), are blatant, and possess no sound factual basis. 52 This suggestion harks back to a central debate in research on gender and language, and to O’Barr and Atkins’ influential 1980 study challenging Lakoff’s (1975) suggested features of female speech. The authors found that, in their study of courtroom discourse, linguistic differences between male and female speakers were due not to gender, but to situation-specific authority and power. What Lakoff (1975) called ‘women’s language’ they suggest should in fact be renamed ‘powerless language’, a register used by both men and women depending on their status and on the communicative context. In SCEGAD, the communicative context was the same for male and female academics in the corpus, and there were no differences in actual status. It is therefore impossible to discern whether these results indicate a felt, if not a real, inferiority in status on the part of the women academics (power explanation), or an innate difference in male and female academic writing (gender explanation). <?page no="131"?> 131 enhancing (explicit claim to knowledge) and authority-reducing (statements given as opinions rather than facts) functions, it is difficult to gain any clear picture of possible reasons for gender-specific behaviour here. One strongly authority-enhancing use of person reference, in order to advertise the author’s own work, shows no significant difference across the genders. Research on automatic text categorisation according to gender has sought to prove that women’s written language is more involved and personal than that of men (Argamon, Koppel, Fine & Shimoni 2003). Other researchers regard this dichotomous approach as “a gross oversimplification of the complexity of language and gender” (Mullany 2002: 4; see also Graddol & Swan 1989). Crawford quotes a “methodologically sophisticated study” which argues that the stereotype of women being more people-oriented than men is wrong, since it found that the women studied did not employ as much person reference as men (1995: 30). In their study, Agarmon et al. (2003) examined various linguistic features, among them pronouns, in the BNC, and found that female authors employed significantly more person reference than did male authors. They conclude that […] even in formal writing, female writing exhibits greater usage of features identified by previous researchers as ‘involved’ while male writing exhibits greater usage of features which have been identified as ‘informational’. (2003: 321) Argamon et al. relate their findings to Holmes’ (1993) suggestion that there are ‘universals’ of female language. One such universal is the greater attention women are said to pay to the relationship-building function of conversation, for which person reference is also used. It is, however, extremely doubtful whether the BNC results can be used to support such a hypothesis, or whether these findings are applicable to academic discourse. A possible ‘universal’ would have to operate in all linguistic contexts: the SCEGAD results prove that this is not the case. As regards academic discourse, it is impossible to test whether the SCEGAD results are mirrored in the appropriate subsections of the BNC, since the entire BNC contains (an extract from) only one single humanities periodical (The Women’s Art Magazine). If one widens the criteria to include two further BNC text-medium categories, books and published texts, the results mirror those for the whole corpus: female authors use more person reference than males. However, whether this trend really applies to academic discourse is questionable, since the 15 texts included in the corpus include a user’s guide to art criticism (more popular science than academic prose proper) and a book entitled The Perfect English Country House, among other unsuitable samples. In addition, many of the hits are in fact quotes from <?page no="132"?> 132 various novels under discussion in the corpus texts. Any conclusions based on such data are extremely doubtful. When the criteria are changed to examine not humanities texts, but texts classified as natural sciences, no gender comparison is possible, since the BNC contains no female-authored texts in this category. The BNC is thus not capable of providing any data on gender differences in texts even remotely comparable to the SCEGAD corpus. Though Argamon et al.’s results are interesting, a study based on the BNC is not qualified to make statements regarding either language universals, or on academic discourse. Likewise, bold and unsupported assertions that women employ the inclusive ‘we’ more often than men do, because women are more ‘person-oriented’, and supposedly inspired by a cooperative rather than a competitive spirit, could not be confirmed in this study (Lakoff 1975; Maltz & Borker 1982; Bailey 1992). On the contrary, the female authors in SCEGAD used fewer first person plural pronouns than do the males. In general, extreme caution should be exercised in positing ostensibly universal gender-specific tendencies. Hiatt’s (1977) detailed empirical investigation showed a large range of pronouncements and assumptions regarding women’s writing to be false (though many unfortunately endure). Although Hiatt did find discernable differences between maleand female-authored texts, and thus a sort of ‘feminine style’, this style had nothing whatsoever in common with popular stereotypes of women’s language (1977: 121). 53 If the data on which theories of gender and language rest are not solid, researchers risk repeating - or worse even enforcing - unhelpful and potentially harmful stereotypes. It is worth mentioning an example of a spurious tendency due to inadequate data here. Starting with Lakoff’s concept of ‘women’s language’ (1975), hedges, as expressions of insecurity and diffidence, have often been associated with women’s speech (see also Holmes 1995). In the MICASE corpus of spoken academic discourse, it is indeed the case that the female speakers employ significantly more hedges than the males do. However, as Poos (1999, cited in Swales 2002: 164) and Poos and Simpson (2002b) showed, these differences are attributable not to gender, but to another influential intervening variable. Due to the skewed nature of the MICASE corpus, perhaps only reflecting the skewed nature of research-life, there were far fewer women represented in the science and engineering sections than in the humanities and social sciences. Since terminology in the humanities and social sciences is often less clear-cut in meaning than in the hard sciences, more hedges are necessary in order to talk about central terms and con- 53 Hiatt (1977) unfortunately did not examine firstor second-person pronouns. The range of other phenomena she analysed, based on a computerised corpus of 200,000 words of prose from 100 books (50 by men, 50 by women), is however impressive. <?page no="133"?> 133 cepts. This interpretation is supported by Hyland and Tse’s finding that hedges were well over twice as common in the soft-sciences section of their academic writing corpus (2004: 173). Thus, a posited gender difference is not due to gender at all but to discipline. This indicates that we should take care to consider possible intervening variables before proclaiming gender differences (or indeed any other sorts of differences). With regard to the current study, more research is needed in order to clarify whether the SCEGAD results are indicative of wider tendencies in scholarly writing, or if they are an anomaly, and if so, to what the anomaly is due. 54 It would certainly be interesting if academic discourse were to constitute an exception to, or even the disproof of, the widespread stereotype that women’s language is more personal than that of men. The SCEGAD results indicate that this may indeed be the case. 3.4.6 Person reference according to other variables Of the other extra-linguistic features for which the SCEGAD corpus is balanced, it turns out that while age has no significant effect on person reference frequency, the author’s academic status does. SCEGAD contains articles by academics of all status levels, from pre-PhD to emeritus professor. In the corpus texts, there were significant differences between professors (current and emeritus) and non-professors for all types of person reference except third person reference (Mann Whitney test p<0.05). There was no significant difference between current and emeritus professors, or between pre and post-PhD scholars without a professorship. It seems that the appointment to a professorship is such a milestone for humanities academics that it significantly changes the way they present themselves, and interact with their readers, when they produce articles for publication. The importance of academic status in adopting a more or less explicitly personal authorial voice is recorded by Holstein (Holstein & Bleich 2001). In the introduction to a volume on the social character of humanities writing, Holstein comments that it fills her with unease to write explicitly personal academic texts. She explains that her male co-editor, professor David Bleich, however, may write in any way he chooses, because “he has earned the right to be personal and given his authority within the profession, he is undiminished by its use [original emphasis]” (2001: 8). The strange thing about this comment is that Holdstein too is a highly respected 54 Interestingly, a pilot study by Römer (2005), which examined hedging in a specialised corpus of academic book reviews, found that the female academics in the corpus in fact did not employ more hedges than did their male colleagues, thereby disputing another posited universal of female language. However, since the study design is unsound, these results are not necessarily reliable (see discussion in chapter two). <?page no="134"?> 134 professor: is it possible that gender is responsible for her reluctance to write in the same personal fashion as her nominally equal-status colleague? How else is the “right” to be personal earnt, if not through promotion to the highest academic level, and peer recognition as a fine scholar? Holdstein’s is a revealing comment on phenomena that seem to be mirrored in the SCEGAD texts: higher-status academics and male scholars adopt a more explicitly personal authorial voice than do more junior and/ or female colleagues. Even where male and female authors possess the same high academic status, gender differences persist. Again, this serves as a reminder that linguistic behaviour has multiple conditioning factors. Therefore, studies which neglect extra-linguistic variables, such as communicative situation, gender, age and relevant social factors, or control their data only for one or two variables, will not provide reliable results. Language, as a social phenomenon, cannot be satisfactorily investigated or explained using a monocausal model. Bivariate and multivariate significance tests have shown that a number of variables, both linguistic (language, culture, discipline) and extralinguistic (gender, academic status) have a significant effect on the use of person reference. With stepwise multiple regression, it is possible to establish which of all of these variables is the best able to predict how often an author will employ person reference. Figure 3.14. Which variables best predict person reference frequency: stepwise multiple regression model summary a. Predictors: (Constant), Text from philosophy b. Predictors: (Constant), Text from philosophy, German author c. Dependent variable: Person reference frequency (all forms) In Figure 3.14, the value in the column “adjusted R squared” is a measure of the relative prediction power of the model (here we see the values for the two most powerful models) and reveals what proportion of the total variance in the dependent variable (person reference frequency) can be determined using only the values of the independent variable (in model one: discipline, in model two: discipline and cultural background). Underneath the table, the explanation for predictors a and b reveals that the most Model R R squared Adjusted R squared Std. Error of the Estimate Change Statistics R squared change F Change df1 df 2 Sig. F change 1 .602(a) .362 .356 45.23606 .362 55.628 1 98 .000 2 .645(b) .416 .404 43.49356 .054 9.010 1 97 .003 <?page no="135"?> 135 important predictors of person reference frequency in academic texts are culture (is the text by a German author? ) and discipline (is the text a philosophy article? ). If both these pieces of information are provided, it is possible to predict over 40% of the total variance in person reference frequency, which is considered a very high percentage in statistical terms. This means that the other factors examined above, although many had a significant effect on the frequency of person reference, are not as effective predictors as the variables discipline and cultural background. When all texts in the corpus are integrated into the model, it emerges that by far the most influential variable for determining the frequency of person reference is discipline: in particular, the knowledge whether any one text is from philosophy or not. This underlines the singularity of the SCEGAD philosophy texts with regard to the use of person reference, as was noted above. These results at first seem to contradict the conclusion reached above that cultural background is more influential than discipline. The multiple regression analysis shows that the best mathematical predictor of frequency is discipline: this is due to the unusually high frequency of person reference across both languages. The qualitative analysis above showed, however, that there was no common reason across the English-language and German-language philosophy texts for this high frequency, and that the commonalities (relatively high frequency in philosophy texts) were not due to the influence of discipline. Both analyses are right in their own way: if one is interested solely in predicting frequency, discipline emerges as the most influential factor; if one is interested in consistent reasons for linguistic behaviour, then cultural background emerges as the most influential. 3.5 Person reference and discourse strategies So far, the results have been grouped according to formal considerations of pronoun form, specifically person and number. However, it is also possible to group the results according to discourse purpose. Therefore, an examination of the purpose person reference serves in academic texts now follows. Personal pronouns, and third-person author-reader references, serve myriad functions in the SCEGAD texts: through them, authors claim membership of an array of different groups, or present themselves as individuals. They offer opinions, qualify statements, address the reader, comment on their methodology and, through text comments (see chapter four), lead the reader through the text. For the analysis, the discourse purposes served by each instance of person reference in the corpus texts were divided into eleven mutually exclusive <?page no="136"?> 136 categories, as illustrated in the following chart (arranged in order of frequency in the English language texts). These are: 1. ‘Personal opinion’: in order to express a personal opinion or make a personal comment 2. ‘TC’: in text comments 3. ‘Group: academics’: for expressing group identity as an academic (historian, philosopher, etc.) 4. ‘Group: author & reader’: for expressing group identity and solidarity with the reader 5. ‘Group: humans’: for expressing group identity as a human being 6. ‘Group: other’: for expressing various other types of group identity, for example as a member of a particular research group, or group of students in personal reminiscences, or for expressing solidarity with the people one is studying 7. ‘Methodology’: for explaining the methodology employed in the article 8. ‘Address reader’: for directly addressing the reader (not talking about readers in general: such examples were excluded from the analysis) 9. ‘Self-advertisement’: for referring to, and thus advertising, the author’s own previous work 10. ‘Acknowledgement’: for expressing thanks to various groups (colleagues, institutions, family members) for providing data, funding or personal support (among other things). 11. ‘Reference to secondary sources’: for referring the reader to work carried out by other researchers <?page no="137"?> 137 Figure 3.15. Person reference by discourse purpose: English and German-language texts (percentage of total occurrences) 0% 5% 10% 15% 20% 25% 30% 35% 40% Personal opinion TC Group: academics Group: author+reader Group: humans Methodology Address reader Self-adv. Acknowled. Group: other Ref sec.source English German The English-speaking authors most commonly used person reference to express a personal opinion ‘I’, whereas the German authors, in particular the philosophers, employed person reference as a way of indicating membership of the group of human beings, through using the inclusive ‘we’. The German philosophers’ use of this strategy overshadows all other uses of person reference in the German subcorpus, and constitutes an important exception to the otherwise less frequent use of person reference in the German-language texts. The second most frequent strategy for which both English and German-speaking scholars used person reference was to lead the reader through their articles with text comments. Strategies for which person reference was far more frequently employed by English-speaking authors include addressing the reader, and self-advertisement (reference to the author’s own previous work). This is not at all surprising, and fits very well with the general finding that English-speaking authors emphasise interpersonal aspects of their writing more than German scholars do, since these two strategies, reader address and self-promotion, are highly (inter)personal in nature. For both groups, text comments accounted for roughly 20% of all cases of person reference use. In some cases, these devices were incredibly specific in their localisation and organisation of material: <?page no="138"?> 138 (36) As I shall show at the end of my final section, physicalism can be exhibited as being explanatory in just the way we expect scientific hypotheses to be. (PHUS) (37) Finally, let me give the briefest outline - no more than a hint - of the evidence for physicalism that I mentioned a few paragraphs back. (PHUS) In the relatively few cases in which German authors expressed personal opinions with ‘I’, it is often possible to detect a measure of discomfort on the part of the author. This particularly awkward sentence, from an article by a German historian, shows the author hedging his bets between the passive and the personalised active in referring to the processes of knowledge gathering in his work: (38) Diese Eigenschaften wurden zudem auch Gott selber zugeschrieben. Daraus läßt sich schließen - ich habe es getan - , daß die Herrscher des 10. Jahrhunderts „höchstwahrschein-lich“ , bei Handlungen der Huld, Milde und der siegbringenden Stärke ähnliche „Erfahrungen“ hatten wie Gregor VII. [These characteristics were also ascribed to god. From this it may be concluded - I have done so - that the rulers of the 10th century ‘most probably’ had similar experiences to Gregor VII concerning grace, clemency, and victory-bringing strength.] (HIDE) Once again, we see the influence of the ‘I-taboo’ inaction. The use of person reference in explaining methodology was particularly common in the folklore articles (52.21% of all cases in the Englishlanguage texts, 37.50% of all cases in the German-language articles). Particularly in some ethnographically-oriented folklore articles, the authors come very much to the fore in reporting on research. The process of gathering information, because it is inseparable from the process of the authors getting to know the subjects interviewed and studied, becomes personalised, with reminiscences of when the authors first met informant x or y, or how the authors experienced the foreign culture under examination. One representative example from the English-language subcorpus: (39) In 1992 when I was translating Sitti Djaoerah into English in Bungabondar (Soetan Hasoendoetan 1997), my translation colleague Baginda Hasudungan Siregar, a ceremonial orator, school principal, and folktale writer and recorder, encouraged me to discuss Angkola heritage with his lineagemate, Sutan Habiaran, down the road. So I would drop by Sutan Habiaran’s house casually several times a week. He was still vigorous in his eighties and was vastly enjoying his retirement from his lumber business in Medan. He had become an antiquarian writer of no small repute, having published over ten small volumes on Sipirok adat (heritage) lore and a tome on the Siregar clan founder “back in Toba” <?page no="139"?> 139 (1974). Sutan Habiaran and I had met on my initial Saturday in Sipirok in 1974 when I was invited out to a village to attend a bone reburial ceremony. (FLUS) There is an almost autobiographical aspect to this usage: though ostensibly reading an article about the life and habits of a foreign culture, one also learns much about the researcher. It is possible that the emphasis on the extensive fieldwork carried out by the author also has a self-promotional aspect. The author presents herself as an experienced fieldworker, and an accepted member of the foreign community she studies. The intensely and unavoidably personal nature of much folklore research is also evident in the German texts. Note, however, that the last example of self-reference in this extract is in the more indirect third person: (40) Als ich wohlvorbereitet und wie immer adäquat gekleidet eine Händlerin ansprach, die gerade das Börsengebäude zum Dienstschluß verließ, fertigte mich die Dame kurzerhand mit den Worten ab: „Interessiert mich überhaupt nicht“. Sprach’s und ließ den verdutzten Ethnographen stehen. [When I - well-prepared and as ever adequately dressed - addressed a broker just as she was leaving the stock exchange building after work, the woman quickly disposed of me with the words ‘doesn’t interest me at all’. Spoke and left the dumbfounded ethnographer standing there.] (FLDE) The same author continues to refer to himself in the third person for the rest of the explanation of his knowledge-gathering process: (41) Der Forscher wurde sogar im Verlauf der Feldarbeit mehrmals ausdrücklich aufgefordert, … zu erläutern. [In the course of the fieldwork, the researcher was even explicitly asked to explain … a number of times] (FLDE) (42) Demgemäß kleidete sich der Ethnograph stilistisch wie Börsianer; Krawatte und ärmellose Weste inbegriffen. [Thus the ethnographer dressed in stockbroker-style; tie and sleeveless waistcoat included] (FLDE) The examples above can be taken as evidence of some discomfort felt by this German author when explaining his methods in the field. The effect is somewhat odd, particularly in example (28), where the author talks about himself in the third person even when describing something so very personal as the clothes he wore. However, the author apparently prefers this relatively awkward type of oblique self-reference to the use of first person singular pronouns. The English-speaking folklorists apparently feel no <?page no="140"?> 140 such discomfort when explaining their methodology, since they consistently prefer first person self-reference. The prominence of first person singular pronouns for explaining methodology in the folklore texts has something to do with the dilemma faced by ethnographers in studying and interpreting foreign cultures, of which Geertz writes in Works and Lives: The Anthropologist as Author. According to Geertz, the problem lies in “how to get an I-witnessing author into a they-picturing story” (1988: 84), and he recommends that “the most direct way to bring field work as personal encounter and ethnography as reliable account together is to make the diary form […] something for the world to read” (84). In some of the SCEGAD folklore texts, in particular those with an ethnographic, field-work focus, person reference seems to be one indication of the authors’ attempts to commit themselves to what Geertz calls “an essentially biographical conception of Being There” (84). Since this is a dilemma unique to the social sciences, it therefore makes sense that it is the folklore texts in SCEGAD which should show signs of it. Theoretically, the same sort of situation could arise in linguistics, which is often considered a social sciences discipline. However, because of the particular journals selected for the linguistics subcorpus, and the fact that each subcorpus was balanced for subdiscipline, none of the linguistics articles in the corpus is comparably ethnographic in focus. However, something similar can be observed in a history article, which is ethnographic in method and focus. The article is a case study of the effects of racial categorisation on one US-American family, and is based largely on the author’s ‘fieldwork’ interviews with the protagonists. This is frequently acknowledged in the footnotes: (43) 5 Genealogy in family Bible in possession of William Warren Conolly, and author’s conversation with … (HIUS) (44) 18 … author’s conversation with Theoline Conolly McCoy (b. 1919) … (HIUS) (45) 20 … author’s conversation with Lou Connolly Coleman (b. 1928) … (HIUS) (46) 31 Author’s telephone conversation with James T. Conolly, Brooklyn, N.Y., November 11, 1996 (untaped) … (HIUS) (47) 34 … author’s conversation with Dorothy McLean Welcome (b. 1917, a Conolly on her mother’s side), East End, Grand Cayman, July 15, 1999; Theoline Conolly McCoy conversation, July 13, 1999; author’s conversation with Lou Connolly Coleman, East End, Grand Cayman, August 27, 2000; Aurellia Conolly interview, p. 4. (HIUS) <?page no="141"?> 141 The difference between this history article and the folklore texts in SCEGAD is that here the personal nature of the knowledge-gathering process is acknowledged indirectly, by using the third rather than the first person, and appears only in the footnotes, rather than being thematised as narrative content and foregrounded in the body of the article. The point here is that this way of using language is not confined to a particular discipline, nor is it a feature of any discipline per se, but is rather evidence of a particular way of gathering knowledge. As a methodology, field work is by its very nature personal, and this is reflected - to varying extents - in the language used to report on it. This leads on to some more general observations about person reference strategies, which hold across cultural boundaries. In all three cultural groups represented in the corpus, one of the most unusual ways of constructing authorial identity within the corpus texts was in direct relationship to the potential reader, either by direct address with ‘you’/ ‘Sie’ or by the periphrastic ‘the reader’/ ‘der Leser’ or ‘my reader’. 55 In the German subcorpus, there were no examples of direct second person reader address, and only one example of oblique third person reader address, which is given below (example 37). Both types of address seem an unexpected communicative style in a research article, since this text type otherwise backgrounds direct personal interaction and involvement. This is not to say that written texts per se are in any way less personal than spoken texts: a letter to a lover or a diary entry is certainly more personal than a Finance Minister’s speech on the current account deficit. As mentioned above, there are also written academic genres with a high degree of personal involvement and therefore subjective content, such as book reviews and academic references. For our purposes, it is thus more useful to distinguish between involved and distanced than spoken and written communication. It is also not necessarily the case that a text is less personal, simply because it contains less person reference, since there are also other ways in which authors are manifested in the text, such as through evaluation and modality. However, it is true that research articles are generally not regarded as personal texts, and are generally not explicitly constructed as dialogues. This may be because it seems non-intuitive to address nameless people one shall never meet. However, even though the author and reader of an academic publication are separated in both time and place, there are nonetheless instances where the corpus authors directly address their putative readers: 55 Direct reader address with second person singular pronouns appears to be uncommon across disciplines: Trumpp (1998) found only a two examples of it in the Englishand German-language components of her sports science corpus, both in an English-language article (1998: 105). <?page no="142"?> 142 (48) If you look back to (97), you will notice that the two States involve stage level predication. In fact, the more ‘stage-like’ the predicate is, the better for the Telic Clause. (LGUK) (49) It is easy to get confused about this, because one imagines someone who has not yet learned to recognize cats reliably looking at a cat and having the same percept as you and I would have. (PHUS) (50) Not that the readers of this publication need to be educated about the politics of Standard English (I suspect I am preaching to the choir). (LGUS) (51) Der folgende Abschnitt wurde in der Absicht geschrieben, den Leser zu einer rekonstruktiven Überprüfung seiner die Extension des Handlungsbegriffs betreffenden Überzeugungen zu bewegen. [The following section was written with the intention of moving the reader to a reconstructive examination of his views regarding the extension of the concept of action.] (PHDE) The discourse strategies behind such addresses are various, as can be seen above. In the first case, the pronouns are used in a directive: the author asserts his right to direct the reader’s attention where he sees fit. In the second case, an attempt is made to bring the imagined reader on side, with the author assuming or anticipating that the reader will share a particular perception with the writer. Similarly, in example (36), specific attitudes and knowledge are attributed to the readers: the author shows her knowledge of the target audience, and at the same time, pays this audience an implicit compliment. In the last example, the address to the reader emphasises the author’s attempt at persuasion. By using this persuasive effort as the justification for a section of the article, the author draws the reader’s attention to his explicitly-stated aim. Assuming the reader wants to be persuaded, this very direct metadiscoursal strategy may be successful; other readers may prefer to be prompted less baldly. Like the direct address with the second person, generic uses of ‘you’ also signal author-reader dialogue and involvement. However, generic ‘you’ fulfils different discourse purposes. It is not used for issuing directives or assigning opinions to the reader, but rather to add immediacy and personal content to the writing: (52) It is part of our cultural knowledge that when you open a new bank account, an ATM card is posted to you and that when you buy a winning lottery ticket, you are entitled to your cash prize. (LGUK1) <?page no="143"?> 143 In the example above, ‘you’ could just have well been replaced by ‘one’ - indeed, according to traditional grammatical prescriptions, it ought to have been. However, through the second person pronoun, the reader’s direct involvement is summoned, and this would not have occurred had the indefinite pronoun been used. The use of the second person appears to be a conscious choice on the part of the author: it is true that ‘one’ can sound stilted, and might have been avoided for that reason, but a passive construction could equally well have been used. However, the author chose to express a generic meaning with the pronoun form typically reserved for reader address, and I suspect that the author did so deliberately, in order to achieve a specific stylistic effect. Though the generic ‘you’ is usually regarded as informal, since it is common in casual speech, it was relatively common in the Englishlanguage corpus texts (see section 3.4.1), and was nearly twice as common as direct second person address. The generic ‘you’ appeared in all disciplines except history; it was particularly common in the philosophy articles. Though the use is more frequent than in other disciplines, the discourse purpose - to involve the reader - is essentially the same, as can be seen in the following example: (53) But, according to CT, you cannot have a lcatl [sic] until you have the ability to detect cats. According to psychology, you cannot have the ability to detect cats until you have a theory of cats. According to LOT, you cannot have a theory of cats until you have Icatls. So, you cannot make the conjunction of LOT and CT compatible with psychology. (PHUS) Even though the second person is not used here to address the reader directly, the effect of this passage is still essentially dialogic. A group of philosophical theories is couched in the second person as a set of directives about what is and is not possible, and this creates a markedly informal lively tone, which is likely to capture the reader’s interest. A particularly striking example of the generic ‘you’ being used to engage the reader’s attention is found in the initial sentences of a philosophy text: (54) You finally meet the love of your life. After an exciting series of dates, he decides to propose. There you are together at the restaurant, Lespinasse, enjoying a simply ravishing meal. Just before the dessert is served he produces a small box, the sort of box that could very well contain an engagement ring. You open it with anticipation, only to find inside a nugget of gold and some chimney soot. Your heart sinks. You knew he was a philosopher, but you realize with growing alarm exactly what this means as he offers the following curious speech … (PHUS) <?page no="144"?> 144 Depending on the reader’s expectations, this sort of language use could be either engaging or off-putting. The writer’s intention, clearly, is to capture the reader’s interest at the very beginning of the text with a humorous made-up story. The same type of strategy is pursued in journalistic texts: the first sentence, known as the ‘hook’, must immediately intrigue the reader, and ensure that he or she continues reading. In a specialist academic journal, however, where no advertising revenues depend on entertaining the audience, the strategy seems strangely out of place. There are many other ways of engaging the reader. Some authors construe their readers as possessing certain ideas or preconceptions, which they then proceed to argue against, thereby constructing a sort of straw man with whom they can hold a dialogue and whom they can eventually convince: (55) Though many readers think of journalism as the realm of fact and the newspaper article as a transparent presentation of fact, journalists have always thought of what they do as telling stories. (FLUS) (56) Some may already think that (1) is a misleading way of expressing an identification of water and H 2 O. If you think that, feel free to take a peek at section VI where an alternative way of parsing this identification is found wanting. (PHUS) This increases the interactive, explicitly argumentative character of the text, and is also used as a way of anticipating reader criticism, as in examples (41) and (42) above. Although both the English and Germanlanguage texts contain both anticipated reader-criticism and ‘straw-man’ readers, only the British and US-American authors achieve this through direct address to the reader with second person singular pronouns or oblique third person reference. The German authors prefer indefinite pronouns such as man (one), which was not counted with the other pronouns, since it is not a form of person reference, but rather a way of avoiding person reference: (57) Man würde es sich zu leicht machen, dies allein mit dem zaudernden Charakter des Pompeius oder der Unfähigkeit bzw. Untätigkeit seiner Unterfeldherrn zu erklären. [One would make it too easy for oneself by explaining this solely through Pompey’s procrastinatory character or the incapability or inactivity of his generals] (HIDE) Despite the fact that written scientific texts are not generally thought of as being explicitly dialogic, as Kretzenbacher points out, all academic texts are argumentative and thus persuasive, and every sort of argumentation is essentially interactive and dialogic (1998: 137). As can be seen for <?page no="145"?> 145 example in the extract (43), argumentation need not take the form of direct address to the reader. Therefore, it would be wrong to conclude that the texts which do not include such features are any less dialogic than those that do. The difference lies in how this dialogue is constructed, and with whom it is held. Kretzenbacher emphasises implicitly dialogic devices, such as the use of citations, and the dialogue with other researchers (1998: 137): I would add the use of indefinite pronouns, directives and rhetorical questions. Across languages, explicit dialogue with the reader is not a regular feature of the academic texts in SCEGAD: it is an unusual strategy in the English-language corpus texts, and save one example (37) completely absent from the German ones. 3.6 Multiple identification and multiple allegiances in academic prose Personal pronouns are more than just a means to achieve specific discourse purposes: they are social words that serve social functions. They are the devices through which academic authors construct identity, or do ‘identity work’ (Malone 1997; Sacks 1992: 1 & 2; Schegloff 1996; Bramley 2001). Personal pronouns are also the devices through which authors build alliances with the reader. From this social constructivist perspective, writing becomes an ‘act of identity’ (LePage & Tabouret-Keller 1985), in which people “align themselves with socio-culturally shaped possibilities for self-hood, playing their part in reproducing or challenging dominant practices and discourses, and the values, beliefs and interests which they embody.” (Ivani 1998: 32; see also Watson 1987; Sacks 1992; Malone 1997). The situation is particularly complex because academic authors do not just construct identity, they construct identities. Each author in fact possesses multiple and multi-layered identities - personal, individual, group, gender, social, discoursal, cultural and so on - all of which may be summoned at any point within the text, sometimes concurrently. Similarly, there are many different shared allegiances which a writer may highlight in order to strengthen the relationship with the reader. Writer and reader can be bound together by shared disciplinary affiliation, by shared linguistic or cultural heritage, or by shared profession. These identities and allegiances are signalled most often through the first person plural pronoun: Hyland (2005: 182) for example found inclusive ‘we’ to be “the most frequent engagement device in academic writing”. But who exactly does ‘we’ include? One of the most prominent features of the first person plural pronoun as it is used in the corpus texts is the great diversity of groups that ‘we’ can refer to: as Wales points out, the “actual discourse referents for <?page no="146"?> 146 ‘we’ are seemingly limitless” (1996: 63). The multiplicity of possible referents allows the writer great flexibility in their own identity construction and their interaction with the reader. Up to now, the emphasis has been on cultural comparison and empirical data, i.e. on contrast. However, this focus can be limiting: authors from different cultures sometimes follow very similar strategies, and there are aspects of language use which are better illuminated with a more qualitative analysis. In this section, I aim to illustrate some of the ways in which British, US-American and German academic authors foreground multilayered identities and allegiances, and some examples follow. First, authors may explicitly claim membership of a specific academic community and express their identity as part of this community: (58) As folklorists, we need not relinquish analysis of complex and contradictory subjectivities to scholars of literature and media. (FLUS) (59) Therefore, by applying the transitivity of identity, we philosophers have discovered that snow is numerically identical with water vapor. (PHUS) (60) Man berief sich auf die uns bekannten und als Vorläufer des Faches betitelten Gelehrten Jeremias Gotthelf… [One referred to teachers wellknown to us and titled pioneers of the discipline, such as Jeremias Gotthelf...] (FLDE) (61) Gerade diese fuzzy category des Alltags erwies sich als enorm produktiv in der Geschichte unserer Disziplin , die es als ihre Verpflichtung ansieht… [Precisely this everyday fuzzy category has emerged as enormously productive in the history of our discipline, which regards it as its duty…] (FLDE) Through this type of group membership claim, authors draw attention to and strengthen their identity as researchers, positioning themselves as part of a traditionally recognised discipline, and allying themselves with a respected social role as scholars. Foregrounding their belonging to the group of researchers can be a powerful means of legitimising opinions and lending hypotheses credibility. This is therefore at once an authority-invoking and an authority-building strategy. Using the inclusive ‘we’ to mean ‘we researchers’ may also strengthen the relationship with the reader, by emphasising shared community membership. Hyland (2005) noted an “enormous emphasis on binding writer and reader together through inclusive ‘we’” (2005: 182) in the texts in his corpus, and the same is true in the SCEGAD corpus. Summoning this kind of allegiance creates a cooperative spirit: through the academic ‘we’, the reader is invited to “participate as something more than an audience” (Mühlhäusler & Harré 1990: 129). This strategy encourages the reader to view the text not so much as <?page no="147"?> 147 the personal presentation or creation of the writer, but as a contribution to the shared enterprise of knowledge-building within a specific academic community. Authors may also claim membership of a particular ethno-linguistic group, this type of group identity expression being particularly common among the linguists: (62) Als Resultat haben wir im Deutschen die häufig benutzten Hilfs- und die starken Verben, die ja alle sehr unregelmäßig sind. [As a result, in German we have the frequently-used auxiliary and strong verbs, which are all very irregular] (LGDE) (63) Zum einen ist Weltwissen zentral: In unseren Kulturkreisen geht man normalerweise, wenn es regnet, nicht unbedingt los, sondern wartet eher ab, bis der Regen nachlässt. [For one thing, societal knowledge is central: in our culture, one does not usually go outside when it is raining, but waits until the rain stops.] (LGDE) This strategy is not related to authority, but to solidarity with the reader through the assumption of a common linguistic or cultural origin. It is particularly useful when the authors are comparing cultures or languages, and is a sort of call to the reader to remember on which side he or she is on. The unspoken alliance between author and reader is explicitly summoned, and this strategy therefore serves to emphasise and consolidate the author-reader relationship. The underlying assumption on the part of the author is that the reader has the same first language and culture. If this assumption proves to be false, i.e. in the case of a foreign reader, the first person plural pronouns would have to be interpreted not as inclusive, but as exclusive. In this case, rather than drawing the reader closer to the author, the strategy will more likely have a distancing effect, since it highlights the difference between the author and reader. It is interesting that this type of group membership claim occurred only in the Germanlanguage texts: it is possible that only the German authors in SCEGAD take for granted that their readership will consist solely of other Germans, whereas the English-speaking academics assume an international readership, with whom they cannot claim solidarity solely on the basis of shared linguistic or cultural background. Other types of group identity regularly constructed by the SCEGAD authors include institutional identity, as a member of a particular research group, as in the following example: (64) Die finanziellen Mittel zur Durchführung des Projekts verdanken wir der DFG, die es unter der Nummer SE 699/ 3-3 fördert. [We owe the <?page no="148"?> 148 financial means for this project to the DFG, which supports it under grant number SE 699/ 3-3.] (LGDE) Group identity as part of a research group is commonly expressed in the acknowledgments section of an article, in order to thank a funding body for financial support. This type of claim occurs only in the German texts, and reflects the structure of humanities research funding in Germany, where there are many more group project grants than individual grants. The institution referred to in the first extract above is the Deutsche Forschungsgemeinschaft (DFG) [German Research Foundation], which is Germany’s centralised research funding organisation for all branches of the sciences and humanities. Although English-speaking authors also thank institutions for funding their research, they do so without claiming group membership, since they have received personal funding through individual research grants. Although the primary function of a sentence such as this is certainly to thank funding bodies, the group membership claim may also serve a number of other functions. In Germany, the membership of a research group, and the external funding this guarantees, is prestigious. This type of claim therefore may also be a way of implicitly building academic authority as a member of an elite group. At the same time, summoning the group may serve a protective hedging function: the author is not alone in his/ her opinions. In this way, the argument and claims in the article are lent an extra weight: the author’s statements represent the opinions not of one researcher, but of many, and are thus perhaps more difficult to rebuke. Other authors juggle multiple professional identities, in this case as not just a linguist but also a teacher: (65) In short, if we as teachers do not confront prior understandings and grapple with the ways in which they may conflict with what we hope to teach as ‘fact’, we seem to do little to change people’s knowledge in any lasting or meaningful way. (LGUS) This type of dual identity expression occurred in all corpus texts with a pedagogical focus, regardless of discipline or the author’s cultural origin. It highlights the author’s involvement in the activity (teaching) they are writing about, and avoids the impression arising on the part of the reader that some ivory-tower academic with no teaching experience is trying to make suggestions to teachers. The atmosphere therefore becomes more collegial, with the author acknowledging that they have the same concerns as the reader, and that they too stand to benefit from their own suggestions. Writing becomes a co-operative enterprise, and writer and reader meet at eye level. <?page no="149"?> 149 A third type of group membership claim, which was mentioned above in regard to the philosophy texts, was social identity as a member of the group ‘human beings’: (66) […] der von Kant in Sectio II, § 10 vertretenen Auffassung übt, daß ein intuitus intellectualium für uns Menschen unmöglich sei. [the view represented by Kant in section II, § 10 opines that a intuitus intellectualium is impossible for us humans.] (PHDE) This type of claim arises from the requirements of the argument pursued in the text. If, as in many of the German-language philosophy texts in particular, the argument centres on a general exposition of the nature of human perception or thought, the author cannot avoid referring to humans as a group. The phrase ‘us humans’ makes explicitly clear that the ‘we’ is not just a broadly inclusive ‘we’ encompassing author and reader, but refers to all human beings. This avoids confusion and highlights the wider implications of the author’s reasoning. Because academics often switch between these and many other different types of identities, it is therefore up to the reader to interpret from the context which particular identity authors are highlighting at any particular time. However, much ambiguity ensues, and it is at times nearly impossible to discern the referent of particular pronouns. Jespersen observed that “the word ‘we’ is essentially vague and gives no indication whom the speaker wants to include besides himself” (1924: 192). The Longman Grammar of Spoken and Written English also notes this ambiguity in scholarly writing specifically, and comments that “in some cases, academic authors seem to become confused themselves, switching indiscriminately among the different uses of we” (Biber et al. 1999: 320). I would question whether the switching is really indiscriminate, though it may initially seem so because of its complexity. Rather, the alternation of different ‘we’/ ‘wir’ memberships in quick succession shows academic authors distinguishing between the many different identities they concurrently occupy. In fact, if we consider the extracts above, the opposite appears to be the case: each type of membership claim serves a specific purpose within the text, and authors appear absolutely aware of this. This kind of behaviour seems to be by no means restricted to Englishand German-speaking academics, nor to academic discourse: a recent study of conversational Spanish, for example, focused on the “indeterminacy or ambivalence inherent in speaker-hearer reference and its utility as a strategic resource to be exploited by speakers and hearers in the pursuance of their own goals.” (Stewart 2001: 166). As the extracts above have indicated, the SCEGAD authors too do not switch indiscriminately, but alternate in their presentation of group alliances so that they can follow the strategy which best suits them at a particular point <?page no="150"?> 150 in the text, be this authority-enhancement, solidarity with the reader, or the exact specification of a ‘we’ group referent. It is in this sense that, in their work on organisational discourse, Taylor and Cooren talk of the “interactive construction of a web of interrelated identities” (1997: 436): the parallel with academic discourse seems apt. The alternation is by no means inevitable, but deliberate and strategic: pronoun users are “social beings engaged in purposeful social behaviour in social space” (Mühlhäusler & Harré 1990: 60). Any ambiguity which ensues is on the part of the reader, not the author. The alacrity and dexterity with which academic authors summon different alliances is impressive. The rapid alternation of pronoun referents was also noted in a study of meeting talk, where Fasulo and Zucchermaglio observed “high rates of pronoun switch within the same turn and even change of the sense conveyed by the pronoun across different turn units”, with ‘we’ being the most variable pronoun of all (Fasulo & Zucchermaglio 2002: 1125). Similarly, in the SCEGAD texts, the authors sometimes switch allegiances and identities even within a single sentence: (67) Folklorists may claim faithfulness to the complete story, but texts like Charles Briggs and Julian Vigil’s The Lost Gold Mine of Juan Mondragon (1990) show us what a rigorously verbatim transcript looks like. (FLUS) In the extract above, the ‘us’ is in opposition to the academic community of fellow folklorists, to which the author certainly belongs. The ‘us’ might refer to the author and readers, or even to the wider community of academics, though this must remain speculation. The same type of flexibility is also inherent in the German pronominal system, and the German authors exploit this characteristic in the same way, and with the same routine dexterity, as the English-speaking academics: (68) Die Beobachtung, daß wir zwischen verschiedenen Formen wählen können, finden wir einerseits in der Umgangssprache, vgl. z. B. die Aussprachevarianten … [We find the observation, that we can choose between different forms, on the one hand in colloquial speech, compare the variant pronunciations...] (LGDE) In this extract, the first ‘wir’ is an inclusive ‘we’, meaning ‘we (German speaking) people’, while the second ‘wir’ refers to the academic community, which is able to analyse the linguistic behaviour of the former group. In some cases, the concurrent group-identification is even threefold. In the following extract from a multiple-authored text, the ‘we’ could be either the two authors, or the authors and their reader(s), or also the academic <?page no="151"?> 151 community of historians, of which the authors feel part (or any combination thereof): (69) The point of a spiritual journal was to record the feelings: the penitence for sins about which we can only speculate, the despair induced by an unnamed employee, the gratitude after returning from travels to places never described more specifically than ‘abroad’. (HIUS) This juggling of multiple, changing identities sometimes borders on linguistic play, as can be seen in the past sentences of a US-American philosophy article, where the first ‘we’ most likely expresses the professional identity (‘we philosophers’), while the second ‘we’ could also be taken as a so-called pluralis majestatis ‘we’, expressing a personal challenge to other philosophers: (70) We are not, that is, allowed to employ the naturalistic program, with its commitment to the Peircean notion of honest inquiry, across the board to include not only theory of knowledge but theory of meaning as well. We want to know why not. (PHUS) This kind of usage suggests that the juggling of allegiances is a conscious manipulation of language on the part of expert language-users. This last extract also raises an important point. In examples (53) - (55) above, the first person plural pronouns in particular were ambiguous because they could have referred to a number of different groups. However, this ambiguity is in fact even more complex, as the last example (56) begins to indicate. 3.7 Relationships between grammatical form and communicative function The traditional grammatical definition of pronouns relies upon a static relationship between the form of the pronoun and its referent: ‘I’ refers to the speaker, ‘you’ to the listener and so on. There is, however, no such constancy to be found in academic texts. In fact, the discrepancy between the form and referent of the personal pronoun is one of the most striking aspects of person reference in academic texts. This discrepancy is by no means unique to this text type (Wales 1996: 3, 51). Be it academic writing in general, or written texts as a whole, the dynamic relationship between grammatical form and referent is one of the most interesting features of the English and German pronominal systems. A few examples of mismatches <?page no="152"?> 152 between pronoun form and pronoun referent include the following (all are possible in both languages): • first person plural for first person singular referent (the pluralis majestatis, and also the ‘editorial we’) • second person singular for first person singular referent (as in ‘you could tell she was annoyed’) • third person singular for first person singular referent (‘The author would like to thank…’) or second person singular referent (‘The reader may wonder … ’) • first person plural for second person reference (as used by nurses - ‘How are we feeling today? ’ and teachers ‘Last lesson, we learnt that … ’) 56 • first person plural for third person plural reference (common when referring to the sports results for teams of which one is not a part: ‘We won the cricket! ’) The grammatical distinctions of number and person simply do not hold in spontaneous speech or writing: Mühlhäusler and Harré state simply that “any pronoun can be used for any person” (1009: 133). This type of pronominal flexibility can be very useful for academic authors, and it is actively exploited in the corpus texts in order to achieve specific discourse goals. In academic texts, the most common of these goals is to reduce the personal content of an utterance by using a first person plural for first person singular referent, though a third person singular form is also used occasionally. Originally, the editorial ‘we’ was used by the editors of publications to reflect the fact that their opinions represented those of the publishers. This usage therefore had a factual basis and the discourse strategy was transparent and based on honesty: editors drew the reader’s attention to the fact that in writing they were expressing not their own personal opinions, but those of a group of others. However, in the case of academic authors, there is no group of others and therefore no concrete motivation for such usage. The royal ‘we’, on the other hand, never had a concrete motivation, but was informed by perceptions that plural forms were somehow more polite or deferential than singular forms, along the lines of “more means great” (Wales 1996: 64). This type of usage - now uncommon even among the exalted few - was therefore a signal of authority, and the use of plural forms to refer to oneself reserved for Popes, rulers and other powerful 56 Jespersen (1924) terms this the “paternal we”, and notes that it is common in Germany (1924: 217-8). <?page no="153"?> 153 people. 57 This also seems ever so slightly inappropriate for academics. Why, then, does this feature occur in academic texts? The reasons why academic authors would wish to reduce or avoid the subjectivity signaled by first person singular forms are varied. On the one hand, deciding not to foreground oneself may be regarded as a form of modesty through a desire to shrink into the background, emphasising the factual content of the article rather than its originator. On the other hand, consistently couching personal opinions and actions in plural pronoun forms could also be a strategy for increasing the gravity of these opinions, and underlining one’s own importance: it is not for nothing that the second person singular polite forms of address in many languages are grammatically plural forms (Albanian ‘ju’, Czech ‘vykání’, Estonian ‘teie’, French ‘vous’, Finnish ‘te’ , German ‘Sie’, Greek ‘eseis’, Scottish Gaelic ‘sibh’ , Ukranian ‘vy’ and so on). Jespersen (1924) explained how this came to be the case: When a person speaks of himself as ‘we’ instead of ‘I’ it may in some cases be due to a modest reluctance to obtrude his own person on his hearers or readers; he hides his own opinion or action behind that of others. But the practice may even more frequently be due to a sense of superiority, as in the ‘plural of majesty.’ This was particularly influential in the case of the Roman emperors who spoke of themselves as nos and required to be addressed as vos. This in course [sic] of time led to the French way of addressing all superiors (and later through courtesy all equals, especially strangers) with the plural pronoun vous. In the Middle Ages this fashion spread to many countries. (1924: 193) This feature - using first person plural forms to indicate a first person singular referent - is therefore mistakenly regarded by some researchers a part of the discourse strategy of hedging, when in reality things are far more complex. The confusion arises because the expression of personal opinions is a point at which the two categories, person reference and (some definitions of) hedging, intersect. As discussed in section 3.2, there is little justification for regarding all forms of person reference as a type of hedging, and little use in labelling only some forms of person reference hedging. Person reference fulfils a great variety of functions in academic texts, of which the qualification of statements or expression of personal opinions is only one. In fact, in the corpus data, this personal function was 57 Wales is of the opinion that the royal ‘we’ is not dying out at all, but “can be relabelled the ‘presidential or premier ‘we’”, as in Margaret Thatcher’s famous “we are a grandmother”. Wales argues that while the ‘we’ of some politicians is ostensibly exclusive, it “often hides an egocentric agency” (1996: 64). I would suggest, however, that this is a very different phenomenon from the royal ‘we’ - which was used for address and self-reference, and used consistently - and therefore ought to be kept separate from it. <?page no="154"?> 154 not the dominant function of person reference at all, as can be seen in table 3.15. In any case, even for this restricted aspect of person reference - the use of first person plural forms for singular referents - the functions served are more diverse and more complex than simply hedging. For this and other reasons also explained above, the present study therefore does not treat hedging as a separate category of analysis. The three examples below illustrate a strategy which is particularly common in the German subcorpus, whereby the authors express absolutely personal speech acts such as speaking or recognising with plural pronoun forms: (71) Mit Norbert Elias können wir hier von einer “Figuration” sprechen, d. h. von einer Gestalt, die sich aus Mustern der Interdependenz fügt. [With Norbert Elias we can here speak of a ‘figuration’, i.e. of a form, which consists of patterns of interdependency]. (FLDE) (72) In diesem Sinne, der diese Gruppen ohne Unterscheidung umfaßt, sprechen wir im folgenden von Friedrich Barbarossa. [It is in this sense, which encompasses these groups without differentiation that we speak of Friedrich Barbarossa in the following.] (HIDE) (73) Hier erkennen wir auch die Nähe zur konditionalen Verwendung von wo. [Here we recognise the proximity to the conditional use of where.] (LGDE) The same type of strategy occurs in text comments, where single authors explain their personal intent in, or comments on the progression of, the article in the plural (exclusive ‘we’): (74) An dieser Stelle ist zu betonen, daß wir natürlich nicht das Denken des Menschen Friedrich von Staufen erfassen können und wollen. [At this point, it must be emphasised that we of course neither can nor wish to cover the thinking of the man Friedrich von Staufen.] (HIDE) (75) Wir wollen die berechtigte Frage nach der Begründung des Postulats der genetischen und logischen Priorität der sozialen vor der objektbezogenen Form des Handelns zurückstellen und festhalten, dass… [We wish to defer the legitimate question about the justification for the postulate of the genetic and logical priority of the socially before the object-related form of action, and note that…] (PHDE) (76) In this study we seek to provide an appropriate syntactic analysis for residual OV order in non-literary LME, on the basis of these findings. We also relate OV remnants to another syntactic construction […]. (LGUK) <?page no="155"?> 155 (77) Taking the phrase ‘remarkable outcome’ in the original formulation of the conditions on reference of the Telic predicate, we can now offer a more finegrained definition. (LGUK) The Longman Grammar of Spoken and Written English comments on the discourse purpose behind such usage: “by choosing the plural pronoun ‘we’ rather than ‘I’, a single author avoids drawing attention to himself/ herself, and the writing becomes somewhat more impersonal” (Biber et al. 1999: 330). In fact, it is questionable whether academic writing really becomes more impersonal through such devices. In real terms, the writing is exactly as personal when the authors refer to themselves as ‘we’ as when they do so with ‘I’: the opinions, observations and perceptions are still those of the same person, the same person has still decided how to organise the text etc. In terms of reader perception or general tone, it is also doubtful whether readers are hoodwinked by such devices. Examples (60) - (63) above are clearly the exclusive ‘we’, and as such simply provide camouflage for the individual author. It would be most surprising if readers were to mistake the examples above for the inclusive ‘we’, or failed to notice that it is a personal, individual author who is wishing, seeking and offering in the extracts above. It is also possible that the strategy backfires: single authors who consistently use plural forms risk drawing attention to themselves, since this strategy may be perceived as unnatural or affected. An alternate interpretation is offered by Mühlhäusler and Harré, who venture that the “narrative convention [of using ‘we’ for ‘I’] has the effect of a rhetorical distancing of the speaker from an overt self-reference to make the personal source of advice or knowledge or whatever it may be more palatable” (1990: 129). The distancing effect is certainly present, but it depends on reader expectations whether the pronoun swap is perceived as more or less “palatable” than explicit self-reference, always assuming that this is actually what the writer intended to achieve. Other cases of pluralis majestatis occur when (single) authors engage in self-advertisement: (78) In Ingham (1998) we analysed a large database of English familial correspondence of the 15th century, and found many instances of OV in auxiliated clauses where the object NP is negated, as in (2) above. (LGUK) This is a very strange example, since the publication to which the author refers was written by himself only, and was not part of a group research project. There is thus no factual or grammatical reason for employing a plural form here. The discourse strategy behind this is perhaps the attempt <?page no="156"?> 156 to increase the author’s importance: work carried out by groups of people is possibly perceived as being of greater worth. On the other hand, the plural form might be motivated by a sense of embarrassment at advertising one’s own work, and an attempt at modesty. The same author also uses first person plural forms to express personal opinions: (79) It seems to us that to take verse or even literary prose source material as evidence for the state of core syntactic rules in English of the 15th century requires some justification. (LGUK) Ultimately, the use of first person plural forms for authorial self-reference seems to form part of an author’s personal style, and style, of course, is a matter of opinion: A single author should also use ‘we’ in the common construction that politely includes the reader: ‘We have already seen ...’ But never use ‘we’ as a mere substitute for ‘I,’ as in, for example, ‘In our opinion ... ,’ which attempts modesty and achieves the reverse; either write ‘my’ or resort to a genuinely impersonal construction. (American Institute of Physics Style Manual 1997: 14) The use of the third person for authorial self-mention is rare in the corpus (0.22 occurrences per 10,000 words), but serves a variety of functions. At one extreme, third person self-reference could be a sign of modesty, as it apparently is in languages such as Japanese. This may be the part of the motivation for this German example: (80) Es handelt sich um Daten aus einer Tagebuchstudie von A., der einsprachig deutsch aufwachsenden Tochter der Autorin. [The data are taken from a study of the diary of A., the monolingually-raised daughter of the author.] (LGDE) Here, a very personal state of affairs - the author has studied her own daughter - is couched in an impersonal way: oblique self-reference helps the author reduce the personal content in explaining the intensely private origin of the article’s source material. This draws attention away from the author’s identity as a mother rather than as an author, and suggests a measure of professional detachment hardly likely given the nature of the data. At the other extreme, third person self-reference can also suggest excessive pride. One is reminded of many famous people who have referred to themselves in the third person, from historical figures such as Julius Caesar and Jesus of Nazareth, to Bob Dole in the USA 1996 election campaign. This precedent may colour reader attitudes towards - generally not so famous - academics adopting a similar strategy. Historical figures used third person self-reference for a variety of reasons. In the case of <?page no="157"?> 157 Caesar’s De Bello Gallico, the “choice was that of a master manipulator of language and people, meant to suggest the absolute trustworthiness and objectivity of his account” (Lakoff 2001: 33; also noted by Jespersen 1924: 217). The reader is thus lulled into accepting a necessarily coloured account of events as a statement of historical truth. Jesus of Nazareth regularly referred to himself in the third person, either as ‘he’, or with a variety of epithets, such as ‘Son of Adam’, ‘Son of Man’ or ‘Son of God’. The epithetical self-references emphasise Jesus’ human origin and his claim to be the Messiah prophesied in the Hebrew Bible (in which incidentally both God and King David also refer to themselves in the third person). The third person references in SCEGAD most likely have a similar rhetorical motivation as in Caesar’s case: increased objectivity, trustworthiness and the predictive power of utterances. At a psychological level, third person self-reference is indicative of a certain perceived discrepancy or dissociation between the public persona as an academic, and the private ‘I’ persona: juggling multiple identities is not always easy. Sometimes there are more than two personae waiting in the wings. Third person self-reference can be useful when an author wishes to emphasise a particular aspect of their identity in a way not possible with a first person singular pronoun: (81) Wie alt ist die Globalisierung? Das Problem kann hier nicht beiläufig auch noch geklärt werden, doch sei dem Historiker wenigstens eine launige Bemerkung erlaubt. [How old is globalisation? This problem too cannot be solved here in passing, but the historian might be allowed at least one witty comment] (HIDE) (82) Es sei einer ostdeutschen Sprachwissenschaftlerin gestattet, im Folgenden das Phänomen des für ostdeutsche Sprecherinnen und Sprecher so typischen Gebrauchs maskuliner Bezeichnungen für weibliche Personen in den Blick zu nehmen. [In the following, an East- German linguist might be permitted to consider the phenomenon - so typical of East-German speakers - of using masculine categories for female people] (LGDE) In both of these cases, the explicit foregrounding of one aspect of the author’s identity is used to justify the argument and lend weight to opinions. The first author emphasises his professional identity as a historian, and cites this role as justification for his commenting on globalisation. The second author foregrounds her professional identity as a linguist, and her biographical identity as an East-German: both, she argues, qualify her to write about her chosen topic. The assumption implicit in such comments, which are both about asking for permission to express opinions, is that those not properly qualified to talk about a particular topic have no <?page no="158"?> 158 right to make comments. The British and US-American authors do not use oblique person reference for this purpose. 58 The use of third person self-reference to highlight specific aspects of the speaker’s identity is not confined to written language, or to academics. Lakoff (2001) mentions the case of Lt. Col. Oliver North’s congressional testimony during the Iran-Contra Affair, in which North switched between the first person and several different forms of third person self reference, chosen from “several different evocative options, depending on which image of himself he wished to project at any given point” (2001: 33). In a psychologically dubious yet rhetorically effective fashion, North thereby created a “stageful of characters, all parts of ‘Oliver North’, yet each distinct; none identical with the person (‘I’) who was testifying” (Lakoff 2001: 33). At times, academic authors are similarly adept at manipulating language and identity in order to persuade their audience. Thankfully, their intentions are not similarly nefarious. There are places in the text, and discourse functions, where third person self-reference appears to be formulaic, such as in footnotes, particularly when commenting on the source of information. The corpus text with the highest frequency of third person self-reference (4.46 times/ 10,000 words) uses this strategy solely in the footnotes, and always in a formulaic context: (83) 5 Massachusetts Vital Records, Dracut, 1869, vol. 218, p. 166, Massachusetts State Archives, Boston, Mass. (hereafter, MSA). Genealogy in family Bible in possession of William Warren Conolly, and author's conversation with William Warren Conolly (b. 1920), East End, Grand Cayman, July 22, 1998. Tapes and transcripts of all conversations cited are in the possession of the author, unless otherwise noted. (HIUS) The use of third person self-reference also appears to be formulaic when giving thanks (see also Sanderson 2005: 76): (84) The authors would like to thank … (HIUS) (85) The author wishes to acknowledge the British Academy who provided the small grant that made this research possible. (FLUK) 58 It is tempting to draw cultural parallels with the German emphasis on expert-knowledge and specialisation, as evidenced by the high productivity of ‘fach’ [‘specialist’ or ‘specialised’] as the first element in nominal compounds of the type: Fachfremde [layman], fachgerecht [correct, professional], Fachkraft [specialist, skilled employee], Fachmann/ Fachfrau [male/ female expert], Fachhandel [specialist shop], Fachgespräch [specialist discussion], fachkundig [knowledgeable, expert, skilful], Fachpresse [specialised press], Fachwelt [experts] etc. <?page no="159"?> 159 Oddly enough, in the same article, the acknowledgements section continues with the next sentence, in which the self-reference is now in the first person: (86) I also wish to thank friends and colleagues who provided assistance, advice and additional references, in particular … (FLUK) This is not the only case where self-reference strategies differ across the same text: in one corpus text, though the author refers to himself in the first person in the body of the article, and does so with near exact average frequency for his culture (20.02/ 10,000 words versus US subcorpus average 28.89), in the abstract, he consistently employs the third person: (87) While past researchers have argued that … , the author shows that … (LGUS) (88) The author demonstrates that there is considerable evidence for … (LGUS) It is possible that some authors wish to avoid explicit person reference in exposed positions, such at the beginning of an article, as in an abstract, or at the very beginning of a section, such as the acknowledgements. 59 This interpretation is supported by the fact that, in these articles, third person self-reference occurs only in a textually prominent position; later on, in the body of the text or body of the text-section, direct self-reference is preferred. This fits in well with Hyland’s (2005) finding that self-mention was far more frequent in the body text of research articles than in abstracts (34.2 occurrences per 1,000 words in articles versus only 0.8/ 10,000 words in abstracts). In the end, as for plural ‘we’ self-reference, third person selfreference may simply be a feature of personal style: as for the other examples of grammatical incongruence between pronoun form and referent, we see how academic authors purposefully and creatively exploit the inherent flexibility of the pronominal system in English and German in order to achieve a great variety of discourse purposes. 59 Another conceivable explanation for the anomalous absence of personal pronouns in the abstract of this US-American linguistics article may be that the abstract was not written by the author, but provided by the publisher or editor. However, since none of the other articles taken from this journal demonstrates a similar tendency, this possibility appears unlikely. <?page no="160"?> 160 3.8 Conclusion The conception of academic discourse as purely objective and therefore impersonal is a theoretical construct, and does not reflect the reality of language use in written academic texts. Academic discourse is evaluative and argumentative, and these communicative functions are intrinsically personal. Given that academic texts can in fact be described as personal, the interesting question becomes whether there is explicit person reference in academic texts. At the student level, academic authors are often advised to disguise this fact by avoiding person reference. At the professional level, German academics in particular are characterised by avoidance of person reference. No measure of avoidance renders these texts any less personal or their authors any more objective: as Rahman puts it, “an author is always present in his work, in every word of it, by virtue of his being the writer” (2004: 35; see also Davidson 1996: 1072). However, avoidance of explicit person reference may indeed create in the reader an impression of impersonality, which is at times actively exploited by scholarly authors in a somewhat duplicitous fashion. Personal pronouns, in particular first person singular and plural pronouns, and second person singular pronouns, are an important means of constructing social interaction in written academic texts. While third person reference to the author and reader is rare, it too serves a variety of interesting functions in managing author-reader interaction. This interaction is extremely complex, since scholarly authors - generally adept writers - employ person reference for a wide variety of interpersonal functions. The corpus analysis shows that authors pursue exactly contrary strategies with person reference, using it both to minimise and to strengthen their responsibility for the truth value of an utterance. Similarly, academic authors employ first person plural pronouns for self-reference in ways that demonstrate both modesty and at times self-aggrandisement. An author’s language background dictates the linguistic resources available for the construction of self and other identity, and factors such as cultural background, discipline, gender and professional status influence how academic authors use these resources. Person reference use in academic texts is therefore motivated by many interacting factors. In the case of English and German, the linguistic resources offered for this purpose are relatively similar. However, it transpires that the way SCEGAD authors from these two language groups structure social interaction in their texts is very different. The English-speaking scholars represented in the corpus are significantly more likely to refer to themselves and to address the reader in their texts. They are much more ‘visible’ as authors, and often construct their <?page no="161"?> 161 texts as an explicit dialogue with the reader. English-speaking authors in SCEGAD use person reference to establish a rapport with the reader and align themselves with a multitude of professional and social groups. The German-speaking authors in the corpus, on the other hand, largely adhere to the ‘I-taboo’ that has been posited for their language. Unless the content of their writing absolutely requires it, the German-speaking authors in SCEGAD refrain from person reference in general, and first person singular pronouns in particular. The German scholars studied are more likely to couch their personal opinions in first person plural pronoun forms, and prefer not to address the reader directly. While the English-speaking scholars are more likely to refer to themselves as individuals than as members of a group, the German authors in the corpus are twice as likely to claim group membership as they are to explicitly refer to themselves in the text. The English-speaking authors therefore use person reference to construct their texts as at once more individual and more personal than those of their German colleagues. Although the primary focus of this study is to contrast language use among English and German-speaking researchers, the data permit a variety of other comparisons. The intervarietal contrast British/ US-American English shows that the US-American authors represented in SCEGAD were somewhat more likely than were their British colleagues to use person reference. However, this trend is not statistically significant and more research would be required before any firm pronouncements about could be made about differences between British/ US-American personal pronoun usage in humanities research articles. The two groups of Englishspeaking academics are nearly identical in their use of first person singular pronouns: the ‘I-taboo’ therefore appears to be peculiar to the German academics. There is a high degree of disciplinary variation across the corpus texts. However, with the exception of the history articles, frequencies are consistently higher in the English-language texts across all disciplines. The anomalous results for the history subcorpus are due to the effect of two influential intervening variables - status and gender. If this effect is taken into account, the trend for English-speaking authors to employ person reference more frequently still holds. The largest difference between languages was found in the literary studies texts, and this was also the discipline in which all authors used person reference least frequently. Linguistics articles had been included in the corpus in order to test the hypothesis, posited by House, that lingua franca English dominance covertly influences the discourse norms of other languages. The corpus analysis found no evidence for such a claim. The frequency of person reference in philosophy texts in both languages is more than double the average across <?page no="162"?> 162 all disciplines. However, the high level of person reference in the philosophy articles is related to the use of different types of personal pronouns in the two languages, and to different communicative purposes. Because many of their articles centre on general philosophical topics, the German philosophers refer far more frequently than the other German academics to ‘wir’ [‘we’] meaning ‘we humans’. They do not, however, use first person singular pronouns any more frequently, and the comparatively frequent person reference in their texts rarely serves an interpersonal function. The English-speaking philosophers, on the other hand, use both first person plural and singular pronouns far more frequently than do their colleagues in the other corpus disciplines (p<0.01). 60 In contrast to the German academics, they use first person plural pronouns for interactive purposes, not because the argument would seem to necessitate it. The English and German-speaking philosophers are not as similar as they may initially have seemed, since even the same pronoun can serve very different functions. Cultural background therefore emerges as a more consistently influential variable than discipline. The comparison across genders showed that, across all cultural groups, the male authors in SCEGAD employ significantly more person reference than do the females. While German academics seem to be behaving in a nearly unisex fashion, the difference between genders was highly marked for the British academics, with the female British academics adhering to a kind of ‘I-taboo’ even more stringently than the German scholars do. It is possible that academic discourse forms an exception to the general stereotype that women’s language is more personal than that of men; however, further research is necessary to confirm these results. Ethnographic study of university departments in different countries as communities of practice would also be needed in order to clarify possible reasons for the observed trend. The SCEGAD authors were at all stages of the academic career, from pre-PhD to emeritus professor. It emerges that there is a significant difference between professors (current and emeritus) and non-professors when it comes to this linguistic feature. The more senior researchers used more personal pronouns of all types. Interestingly, they were not more likely to indirectly refer to themselves or their reader in the third person (indirect self-reference and indirect reader-address). As far as discourse strategies go, the study showed that the Englishspeaking authors in the corpus most commonly use person reference to 60 This value represents the minimum level of statistical significance as measured in bivariate tests across the five disciplines, with one exception: there was no significant difference in the frequency of first person plural pronouns between the English-language philosophy and linguistics texts. <?page no="163"?> 163 express a personal opinion ‘I’, whereas the German authors, in particular the philosophers, employ person reference as a way of indicating membership of the group of human beings, through using the inclusive ‘we’. The second most frequent strategy for which person reference was used, by both groups of authors, was for text comments. Strategies for which the English-speaking authors employed person reference far more frequently include addressing the reader and self-advertisement, both of which are highly interactive in character. The grammatical distinctions of number and person do not hold in research articles, and the relationship between grammatical form and referent is one of the most interesting features of the English and German pronominal systems as reflected in the corpus texts. The academic authors in the corpus actively exploit the pronominal flexibility of their native language in order to achieve specific discourse goals. The analysis highlights the fact that academic authors possess multiple and multi-layered identities - personal, individual, group, gender, social, discoursal, cultural and so on - all of which may be summoned at any point within the text, sometimes concurrently. Academics often switch between these and many other types of different identities, and it is therefore up to the reader to interpret from the context which particular identity authors are highlighting at any particular time. The SCEGAD academic authors do not switch between pronouns indiscriminately, as some researchers have suggested, but alternate in their presentation of group alliances so that they can follow the strategy which best suits them at a particular point in the text, be this authority-enhancement, solidarity with the reader, or clarification of a ‘we’ group referent. The alternation is by no means inevitable, but deliberate and strategic, and if ambiguity ensues, it is on the part of the reader, not the author. In conclusion, although the German academics represented in SCEGAD are less likely to employ person reference, it would be wrong to surmise that their texts are less personal or less dialogic than are those of the English-speaking scholars. The interpersonal content of the German scholars’ writing, and their interaction with the reader, is not, however, usually expressed through person reference. In the context of the traditional view of academic writing as impersonal, the posited ‘I-taboo’ in German academic writing, and contradictory style guide advice for English-language academic writing, an examination of how academic authors actually behave in their published work appears especially interesting. It emerges that the ‘I-taboo’ in the German texts is not absolute: there are cases in which the German academics explicitly refer to themselves. However, when compared to their English-speaking colleagues, the German authors do so markedly less often. In this sense, the <?page no="164"?> 164 ‘I-taboo’ appears to hold for German academic writing in the humanities as it is represented in this corpus. There is also a discourse critical aspect to such examination, in that, through trying to hide themselves and their opinions from their readers, academic authors are attempting to claim an objectivity they cannot own. Readers may be insufficiently critical of personal opinions and ideas which are deliberately presented as though they were facts. Since academic authors observe and perceive in exactly the same, subjective, fashion as all other authors, everything they write is innately personal in nature. What is unique to academic authors is the attempt to obscure the personal origins of their perception by disclaiming their opinions and constructing their texts as a monologic listing of facts and passively observed events rather than a dialogue. The analysis in this chapter has shown that some groups of academics try to hide themselves more than others do. <?page no="165"?> 165 Metadiscourse - those aspects of texts that affect the relations of authors to readers. _______________________________ (C RISMORE 1989: 237) Every successful academic text displays the writer’s awareness of both its readers and its consequences. _______________________________ (H YLAND 2005: 173-4) Chapter 4 Writing about writing: author-reader interaction through textual metadiscourse Metadiscourse is the rhetorical strategy authors use when they talk about their own text. It is a way of organising discourse and explaining this organisation to readers, and helps structure and guide author-reader interaction within the text. Metadiscourse therefore marks the intersection of author, reader and text. It is evidence of authors’ awareness of their own unfolding texts as discourse, and of their anticipated audience. There are many strategies used for cultivating and managing author-reader interaction, and many types of metadiscourse, but one of the main ways in which academic authors use metadiscourse to communicate directly with the reader, and thereby reveal their conception of this reader, is by commenting on their text. Authors employ a specific type of metadiscourse, which I shall call text comments, to explain the text to the reader, justifying its structure, stating their intentions and pointing out connections between different parts of the argument. Because text comments serve to facilitate the discourse, they are central devices in the co-construction of knowledge. For Crismore, they are even “as important to successful communication as <?page no="166"?> 166 are semantics and syntax” (1989: 5). For a study interested in social interaction in academic prose, text comments are therefore an attractive object of analysis. An examination of this type of metadiscourse in the research articles of the SCEGAD corpus reveals that the ways in which academic authors structure their interaction with the imagined reader, and the resources they use for doing so, differ across cultures and disciplines. I begin by considering the interaction of author, reader and text in written academic discourse (section one). In section two, I discuss various definitions of metadiscourse, before moving on to examine the relationship between metadiscourse and closely-connected phenomena such as evaluation and appraisal (section three). The following sections concern the connection between textual and interpersonal metadiscourse (section four), and the connection between textual metadiscourse and text comments (section five). In section six, I introduce the taxonomy of text comments developed for this study. The next section places text comments in the context of previous research on devices used for discourse organisation and management (section seven), while section eight discusses the findings of previous contrastive studies of text comments. In section nine, I provide a mainly quantitative analysis of text comments based on the corpus data. Section ten presents the more qualitative analyses, focussing on discourse strategies and possible implications for the author-reader relationship in academic prose, while the final section briefly considers the relationship between metadiscourse and readability. 4.1 Author, reader and text The text is the place where author and reader meet; it is what author and reader have in common. Though research articles are factually a monologue, in that the reader of an academic text cannot reply, and the author cannot listen, at least not directly, there are nonetheless two partners in the scholarly communicative process, and a form of interaction certainly takes place between them. This interaction is special, however, in that one of the participants within the exchange exists only as imagined by the other. Readers are present in the text not as a real people, though these real-world people of course exist, but as the ‘reader in the text’ (Thompson & Thetela 1995), a ‘model’ or ‘implied’ reader imagined by the author (Dillon 1981: xvi). 61 There are therefore two readers, the actual and the imagined reader, 61 In a strict sense, of course, the author as a real person is not present within the text either, but rather projects a persona, which need not be identical with his or her biographical self. This is related to Ivani ’s concept of the autobiographical and discoursal self (1998: 28), as discussed in section 6 of chapter 2. <?page no="167"?> 167 which latter figure is a projection of the author. The kind of reader an author projects, and the way in which the author communicates with this putative reader, reveal much about the author and also about the scholarly communication process. Authors can also write with multiple audiences in mind: a general imagined reader, academics following a specific school of thought, even one particular colleague whose work is discussed, not to mention the editor(s) of the journals they wish to publish in. The relationship between the author and imagined readership is not pre-ordained, but is constantly shifting and being negotiated throughout the text (Crismore 1989: 126). The form this negotiated relationship takes, and the aspects of interaction that are open to negotiation, are influenced by cultural and disciplinary norms as well as individual authorial preference. Each author has a slightly different conception of and relationship with his or her readership; this is part of what constitutes personal style. One of the most obvious and important places where negotiation takes place within research articles is where the author steps outside of the text and comments on it for the reader(s). Academic authors comment upon what they write, and how, why and when they write it, often citing reasons for their behaviour. They anticipate criticism that a section is too long or in the wrong place, or that important issues are left out altogether. Such comments reassure readers that their needs are being considered. They are the means by which authors shape and manage the ongoing relationship with their audience. By involving the reader(s) in the ongoing chain of argument, informing them and explaining to them how the communication proceeds, text comments also help authors retain audience attention. In the first chapter, I noted that academics publishing in learned journals write for a very specific, specialised audience. If they are to persuade this audience, scholarly authors must try to anticipate its needs, background, level of general and specific knowledge and preferred style: the learned article […] assumes vast amounts of shared information as well as common interests and outlook. Many have observed that the less you know about whom you are communicating with, the harder it is to communicate with them. (Dillon 1981: 24) Numerous researchers have observed that an awareness, or rather an accurate imagining, of the potential audience, as well as an “ability to reflect and exploit that awareness in the way a text is written”, is vital to successful scholarly communication (Thompson 2001: 58; see also Grabe & Kaplan 1996; Johns 1997; Hyland 2000, 2005). The reason why this awareness is so vital to successful academic writing is that in a written text the interaction is non-reciprocal: “since there is no possibility of immediate reaction he [the writer] has to anticipate what is likely to be and provide for <?page no="168"?> 168 any possible misunderstandings and unclarity arising from a lack of shared knowledge” (Widdowson 1984: 65; see also Ehlich 1982: 119; Graefen 1994: 136-7). If an academic author misjudges or mismanages an aspect of the communication, no feedback can occur, since the actual interaction partner is not present. If information is to be transferred successfully, it is therefore vital that the author’s picture of the potential reader is as close as possible to the actual reader. Authors who misjudge their audience hamper the reception of their argument, and risk engendering confusion or even annoyance rather than agreement. The way in which a text is organised and the way this organisation is signalled is directly related to a writer’s awareness of the potential audience (Thompson 2001: 58; see also Hyland & Tse 2004: 164). Some readers might wish to be led step-by-step through an explicitly ordered text, while others might be willing to embark on an unpredictable textual journey with no signs and destination unknown. Those readers who prefer to be guided through the text may well differ in the kind of organisational signals they expect and where these should occur. Audiences probably also vary with regard to the information they expect, at which point in the text they expect it, and in what form. These expectations are part of what defines a specific text type and discourse community. In order to meet audience expectations, academic authors rely upon their intuition, which is informed by their experience as members of a specific discourse community and culture, and based not a little on their own reading preferences projected onto the imagined audience. In the case of research articles, academic authors could also research their audience, reading articles from their chosen journal in order to gain a picture of appropriate style. Since academic authors are also academic readers, and part of the same specialised discourse community as their audience, they are well placed to imagine accurately the needs and wishes of (this part of) their imagined readership. On the basis of their knowledge, experience and intuition, scholarly authors then attempt to project an imagined reader that will meet with the actual readers’ approval. It is in this sense that Dillon asserts that the “process of imagining a reader is […] a projecting of self that readers will try on and find agreeable” (Dillon 1981: 163-4). Ideally, the imagined reader and actual reader will be very similar, meaning that the text is appropriately written for a particular readership. For Thompson, this “convergence of the reader with the reader-in-the-text is a crucial step in most types of argumentative, persuasive text, including academic papers” (2001: 62). Text comments play a vital role in ensuring that this process, and thus the text as a whole, succeeds. There is another side to this whole process, for while authors imagine readers, readers imagine texts: <?page no="169"?> 169 Based on their knowledge of social and language conventions, as well as their beliefs and perceptions of the author, readers also make use of social imagination in constructing and composing rhetorical texts. Readers project roles and footings for authors and for themselves and then act or react according to the projected or assumed roles and footing as they in engage in, enjoy, understand and learn from […] texts. (Crismore 1989: 126) Readers approach texts with expectations about what they will find: in the case of readers of scholarly periodicals, these expectations have been shaped by years, often decades, of immersion in the language and text types of their part of the academic community. This web of mutual expectations, assumptions and perceptions can be studied - at least in part - by examining how authors comment within the text on the structure of their writing. 4.2 Defining metadiscourse In chapter two, alluding to two common definitions, I explained metadiscourse briefly as ‘writing about writing’, or those elements of the text that do not directly contribute to the informational content. The first of these definitions is attractively simple. It is sometimes re-phrased as ‘discourse about discourse’, ‘communication about communication’ or ‘talk about talk’, and appears in many studies on metadiscourse (Rossiter 1974: 36; Schiffrin 1980: 199; Vande Kopple 1985: 83; Williams 1981: 195, 1989: 27-8; Crismore 1989: 7; Geisler, 1994: 11; Rahman 2004: 29). The second definition of metadiscourse which I cited in chapter two is also standard fare, with a large number of researchers distinguishing between primary/ informative discourse, which informs the reader about primary topics, and secondary or metadiscourse, which is writing that guides the reader (Williams 1989; Vande Kopple 1985; Dillon 1986; Crismore 1989; Crismore & Farnsworth 1990; Markkanen, Steffensen & Crismore 1993; Hyland 2000; Lee 2000). 62 The distinction between primary and secondary discourse parallels Searle and Vanderveken’s (1985) distinction between illocutionary acts and propositional content, and has led to speech-act theories of metadiscourse, as developed for example by Beauvais (1989). Both types of definition are ultimately problematic: they are vague, rest on artificial dichotomies and can lead to taxonomies so broad as to be 62 Hyland and Tse dislike the terms ‘primary’ and ‘secondary’ discourse, since metadiscourse is “not somehow ‘secondary’ to the text […] metadiscourse is not simply the ‘glue’ that holds the more important parts of the text together, but is itself a crucial element of its meaning” (2004: 161). <?page no="170"?> 170 unworkable. Neither gives any indication of the features which may be regarded as metadiscourse, and it is perhaps for this reason that metadiscourse has become an amorphous term covering a vast and disparate array of phenomena from gestures to tone of voice, hedging and text comments. 63 Though the distinction between metadiscourse and (primary) discourse is theoretically useful, it is difficult to uphold in practice, because the two levels or planes of discourse are inextricably linked. Crismore accepts that naturally-occurring language is a mixture of text and metadiscourse (1989: 52) and admits that “what is metadiscourse in one situation may be discourse in another” (1989: 49). Hyland and Tse too find that the two levels of text are sometimes difficult to distinguish (2004: 160). The problem here is that, as Scollon and Scollon note, essentially all communication serves both a communicative (informational) and a metacommunicative (relationship) function (2001: 151). Definitions of metadiscourse that rely upon a clear-cut distinction between discourse (content) and metadiscourse (non-content) are therefore shaky. Especially when analysing naturally-occurring speech, it cannot be assumed that all forms of communication about communication can be adequately separated from communication itself. The binary model is ultimately problematic, since the two proposed levels of discourse are neither practically separable nor mutually exclusive. It is possible to avoid this dualistic view and its concomitant disadvantages. Instead of defining metadiscourse as a level or plane of language, or a distinct unit separate from the primary discourse, metadiscourse can be conceptualised as a rhetorical strategy used by speakers and authors to talk about their own talk (Crismore 1989: 86). This is essentially a functional/ discourse-oriented as opposed to a formally-oriented view. If we reject the notion that metadiscourse is a separate and linguisticallyidentifiable discourse unit within the primary discourse, it becomes instead a rhetorical phenomenon. Crismore recommends just such an approach when she suggests that it is more interesting to examine the function and 63 It should be noted from the outset that metadiscourse is a phenomenon with many names. Three of the more mundane alternative terms are ‘metacommunication’ (Rossiter 1974; Gläser 1990; Busch-Lauer 1992a), ‘metatext’ (Crismore & Farnsworth 1990; Mauranen 1993a; Markkanen, Steffensen & Crismore 1993; Valero-Garces 1996; Bunton 1999; Rahman 2004) and ‘meta-talk’ (Schiffrin 1980). Crismore (1989) uses all four terms as interchangeably. Swales (2004b: 121) says that he now prefers to use the term ‘metatext’ for textual metadiscourse categories, reserving ‘metadiscourse’ for interpersonal categories, yet this distinction is confusing and in any case unnecessary in light of suggestions to be made below. Other slightly more exotic alternative epithets include ‘peri-text’ or ‘epi-text’ (Kinneavy 1971, cited in Crismore 1989), ‘text/ discourse reflexivity’ (Mauranen 1993b, 2003) and ‘discourse deixis’ (Schiffrin 1980; Lenz 1997). For the sake of clarity, in the following I shall work only with the term ‘metadiscourse’. <?page no="171"?> 171 communicative purpose of metadiscourse within a specific textual context, rather than try to establish its linguistic peculiarities when contrasted to primary discourse (1989: 86). Metadiscourse therefore becomes a “social, rhetorical instrument which can be used pragmatically to get things done” (Crismore 1989: 4). For a cross-cultural and cross-disciplinary study, a functional/ rhetorical perspective can be particularly informative, since it focuses attention on the ways in which authors from different cultural and disciplinary backgrounds employ metadiscourse for potentially divergent rhetorical purposes. For a study of social interaction in academic writing, definitions of metadiscourse that emphasise its role in structuring and managing authorreader interaction are obviously far more useful than those which define metadiscourse in terms of a discourse/ metadiscourse distinction. Dahl for example defines metadiscourse as material which “overtly express[es] the writer’s acknowledgement of the reader” (Dahl 2004: 1811), and both Williams and Crismore underline the importance of metadiscourse to effective and active reader participation in the knowledge making process (Williams 1981: 195; Crismore 1989: 7). Other researchers who have studied metadiscourse foreground its role in aiding reader comprehension, in that it helps readers to organise, classify, interpret, evaluate, and react to content (Vande Kopple 1985: 83), its role as an interactive tool (Rossiter 1974), and as a way of interactionally generating, sustaining and eventually ending arguments (Schiffrin 1980: 231). The conceptualisation of metadiscourse as a rhetorical strategy rather than a plane of language offers a practical way of approaching and studying metadiscourse that is well-fitted to the current study. However, it is not just the conceptualisation of metadiscourse that is problematic; its operationalisation is notoriously difficult as well. The precise specification of the features assigned to metadiscourse, is chaotic, involving numerous conflicting definitions, terminologies and taxonomies (Crismore 1989: 239; Schiffrin 1980: 201). In recognition of this fraught state of affairs, scholars have referred to metadiscourse as “vague” (Schiffrin 1980: 201; Hyland & Tse 2004: 156), “fuzzy” and “extremely dishevelled” (Crismore 1989: 60; Hyland 2000: 108), “under-theorized” (Hyland & Tse 2004: 156), “admittedly messy” and “a somewhat gnarled area of study” (Crismore, Markkanen & Steffensen 1993: 54). Hyland and Tse lament that this “lack of theoretical vigour and empirical confusion” has prevented metadiscourse from realising its explanatory potential (2004: 156). At a practical level, the main problem lies in restricting the scope of the phenomenon (Crismore 1989: 94; Dahl 2004: 1811; Ädel 2005: 155). Because standard definitions of metadiscourse have often suggested that language can be separated into discourse, which is informative, and <?page no="172"?> 172 metadiscourse, which is apparently not, metadiscourse frequently ends up as a catch-all term for everything apart from propositional content (Ädel 2005: 156). Like Ädel, I would suggest that this is far too much ground for one term to cover, and I offer an alternative suggestion in section 4.4. The delimitation problem has been present from the very beginning of metadiscourse research. Rossiter for example defines metadiscourse as “all messages about communication […] whether they are verbal or non-verbal, about communication in general (e.g. books and articles about communication) or about some specific communication interchange” (1974: 36). A definition this broad is completely unworkable. According to such definitions, metadiscourse becomes almost a synonym for language itself, since “almost all properties of spoken and written discourse may implicitly or explicitly signal various dimensions of the communicative situation, such as the speaker/ hearer, author-reader, and the goals of the speaker/ author” (Crismore 1989: 49). It might be easier to define what is not metadiscourse. Mauranen comments wryly that metadiscourse “covers such a wide area of language use that […] saying that some item in a text is metadiscourse does not say much” (1993b: 143). Slightly less extreme than Crismore (1989), but similarly unworkable, is Crystal’s definition of metadiscourse as: those features in the organization or presentation of a text which help the reader to interpret or evaluate its content. They include features of textual organisation (e.g. headings, spacing and connectives such as first and next) as well as such interpersonal elements as hedges (perhaps), attitude markers (frankly) and dialogue features (for example, see Figure 1). (2003b: 289) The same dilemma resurfaces when Crismore defines metadiscourse as “an interpersonal device that fulfils the interpersonal function of language” (Crismore 2004: 311). Such a definition alludes to the systemic-functional view of language prominently represented by Halliday (1973, 1985), who proposed that the semantic systems of language can be divided into the ideational, interpersonal and textual functions. Within this framework, metadiscourse falls within both the interpersonal and textual categories, since it is used - sometimes simultaneously - to establish and maintain relationships, and to structure text. Aside from the fact that utterances cannot be neatly divided into three functions, it is again neither realistic nor advantageous to expect metadiscourse to cover all interpersonal and textual functions of language. Doing so sanctions the relegation of metadiscourse to a nebulous term that is all things to all people, but nothing specific to anyone. Difficulties in operationalising metadiscourse persist in recent research. Hyland defines metadiscourse very broadly, as “the devices <?page no="173"?> 173 writers use to explicitly organise their texts, engage readers and signal their attitudes to both their material and their audience” (2000: 104). Geisler adopts a similar approach, defining metadiscourse as “the discourse that calls attention either to the relationship between the author and the claims in the text or to the relationship between the author or the text’s readers (1994: 11). Hyland and Tse define metadiscourse at length as the linguistic resources used to organise a discourse or the writer’s stance towards either its content or the reader […] an umbrella term to include a heterogeneous array of cohesive and interpersonal features which help relate a text to its context by assisting readers to connect, organise and interpret material in a way preferred by the writer and with regard to the understandings and values of a particular discourse community. (Hyland & Tse 2004: 157) The danger with such expansive definitions is that there is a vast group of phenomena which can serve organisational, reader engagement and attitude signalling functions. All-encompassing definitions such as the above therefore lead to operationalisations of metadiscourse that are unworkable, in that they attempt to cover an open list of all sorts of grammatical and discoursal features, from verbs, conjunctions, adverbs and pronouns to discourse markers, hedges and boosters (see for example the list in Hyland 2000: 190-3). As we shall see in the following section, many of these phenomena have been treated under terms other than metadiscourse, and their relation to metadiscourse has never been established. If metadiscourse is defined so broadly, it loses all explanatory power, and becomes only a synonym for a range of phenomena that already possess numerous labels of their own. 4.3 Metadiscourse, evaluation, appraisal and co. Research on metadiscourse intersects with many other areas of research, overlapping with the study of phenomena such as coherence (Van Dijk 1977; Bublitz, Lenk & Ventola 1999; Lee 2000; Bublitz 2001), discourse markers (Schriffrin 1980, 1987), evaluation (Hunston 1994; Hunston & Thompson 2000; Bondi & Mauranen 2003), hedging (Markkanen & Schröder 1997; Clemen 1998; Hyland 1998), discourse deixis (Webber 1988; Lenz 1997; Levinson 2004) and argumentative zoning (Teufel 1999). It is closely related to the research on stance, engagement and involvement that has emerged from systemic-functional appraisal theory (Martin 2000; White 2001, 2003; Miller 2002; Hood 2004; Martin & White 2005). Each of these phenomena encompasses a shifting catalogue of linguistic features as individually defined by each researcher. Despite <?page no="174"?> 174 the many different terms employed, what all areas of research have in common is an interest in the interpersonal features of texts or discourse, and the devices used to express so-called secondary content. An interest in the interpersonal dimension of writing has always figured prominently in both systemic functional and social constructionist theory (Hyland 2005: 174). However, coherence and discourse marker research is often more concerned with the structural, organisational functions of language. Evaluation, hedging, and appraisal research, on the other hand, generally place more emphasis on expressive, intersubjective properties. The relation between different phenomena is unclear. Evaluation and/ or hedging are treated by some scholars as subsets of metadiscourse (Williams 1981; Kinneavy 1971; Vande Kopple 1985, 1988; Crismore 1989; Markkanen, Steffensen & Crismore 1993; Hyland 2000; Harwood & Hadley 2004; Hyland & Tse 2004). Confusingly, at other points some of the same researchers consider evaluation and hedging as independent areas of research (Hyland 1995, 1996, 1998; Markkanan & Schröder 1997), or as subsets of another phenomenon altogether, such as Hyland’s inclusion of evaluation and hedging in his model of stance (or writer positioning) and engagement (or reader positioning) (Hyland 2005; see also Hyland 1999d). As one would expect, the same linguistic features reappear under a multitude of umbrellas: personal pronouns, for example, have been variously treated as metadiscourse (Vande Kopple 1985; Crismore & Farnsworth 1990; Crismore, Markkanen & Steffensen 1993; Mauranen 1993b; Hyland 1998; Crismore 2004; Hyland & Tse 2004), (discourse) deixis (Fillmore 1975; Lyons 1979; Webber 1988), stance and/ or engagement (Hyland 1999d, 2005), and hedging (Zuck & Zuck 1985; Skelton 1988; Darian 1995; Markkanan & Schröder 1997; Hyland 2000; Breitkopf 2004). At times, the terms chosen, and the particular features they are taken to encompass, seem arbitrary. While terminological confusion is a constant problem in all areas of research, the confusion here seems extreme, in that most of these phenomena possess no precise definition at all, or none that is not applied to many other phenomena also. There is often little difference between definitions of metadiscourse, evaluation, hedging and co. For each topic, there are researchers who propound tremendously wide definitions, thereby claiming as much ground as possible for their chosen object of study (see chapter three for a discussion of this issue with regard to hedging). The features which each term is taken to encompass are often very similar: essentially, all language other than the purely propositional or factual (though sometimes this too). Scholars recycle the same definition for more than one phenomenon: for example, Hyland’s definition of stance and engagement - the resources writers use to “represent themselves, their positions and their readers” (2005: 190) - is <?page no="175"?> 175 almost identical to his definition of metadiscourse. Both are little more than paraphrases for ‘everything but propositional content’. Hunston and Thompson define evaluation as “the broad cover term for the expression of the speaker or writer’s attitude or stance towards, viewpoint on, or feelings about the entities or propositions that he or she is talking about”, and observe that this includes areas sometimes referred to as ‘stance’, ‘modality’, ‘affect’, or ‘appraisal’ (Hunston & Thompson 2000: 5). Stance as defined by Hyland is therefore not the same as stance according to Hunston and Thompson, though both overlap, to different degrees, with many other terms, all of which are also defined differently by different researchers. Elsewhere, Hyland states that interpersonal metadiscourse “concerns evidentiality, relation and affect” (2000: 112), thereby suggesting that is it is essentially synonymous with evaluation, though evaluation for Hunston and Thompson is a good deal narrower than it is for Hyland. Again, though this is a common situation not only in metadiscourse research, it is nonetheless a great hindrance. The phenomena mentioned above are intimately related to metadiscourse and to each other. There will never be a perfect model which separates them completely. It is therefore not evidence of any fault with previous studies that they have been unable to provide such a model. However, it is vital that researchers are absolutely clear about their approach and the phenomena they study, lest all terms merge into a meaningless conglomeration. In the midst of this exegetical bedlam, it was doubly important for this study to find a consistent way to conceptualise metadiscourse, and thereby differentiate it from other phenomena. 4.4 Textual and interpersonal metadiscourse As we have seen, broad definitions of metadiscourse cover both those elements which refer explicitly to discourse organisation, and those which elucidate the writer’s stance towards the readership and/ or the text content. These two aspects are often labelled ‘textual’ and ‘interpersonal’ metadiscourse (Vande Kopple 1985, 1988; Crismore 1989; Crismore & Farnsworth 1990; Crismore, Markkanen & Steffensen 1993; Hyland 2000; Hyland & Tse 2004). The terminology is misleading, since textual metadiscourse is just as interpersonal as so-called ‘interpersonal’ metadiscourse. Indeed, metadiscourse is interpersonal per se. Nonetheless, the terms appear to be firmly established. 64 A similar typology is offered by 64 As Hyland and Tse discovered in their study, the multifunctionality of language means that “in practice there are serious difficulties with this attempt to identify two single, discrete functions of metadiscourse” (2004: 162). Textual metadiscourse is not just <?page no="176"?> 176 Schiffrin (1980), who distinguishes between ‘organisational meta-talk’, which serves referential and organisational functions, and ‘evaluative meta-talk’, which is used for expressive and evaluative purposes. Crismore too differentiates ‘informational’ (goals, previews, reviews and topicalisers) and ‘attitudinal’ (saliency, emphatics, hedges and evaluatives) metadiscourse (1989: 158-9; see also Crismore & Farnsworth 1990: 121-2). The distinction in such definitions between textual/ discourse organisation on the one hand, and writer stance and evaluative elements on the other hand, is important, in that it shows that two different phenomena are quite clearly being lumped together under one heading. For while textual metadiscourse is concerned with comments on the form and/ or organisation of a piece of writing, so-called interpersonal metadiscourse focuses on comments on the content of the writing and/ or its expression (Rahman 2004: 34, 38). Certainly, both textual and interpersonal metadiscourse play a role in writer/ reader interaction. Comments on form are a kind of content, and comments on content can also be a form of organisation. However, intuitively we know that there is a distinction between commenting on the organisation of our writing, and commenting on what we say. The former is concerned primarily with structural features, and the latter primarily with matters of opinion. This distinction, though not perfect, is fundamental and central. The writer stance (‘interpersonal’) aspects of metadiscourse are in fact more commonly and appropriately known as evaluation. When Mauranen (1993b: 142-3) distinguishes between broad (‘integrative’) and narrow (‘non-integrative’) approaches to defining metadiscourse the difference, as Ädel points out, lies mainly in whether the approaches include or exclude evaluation (2005: 155). Rather than subsuming evaluation in metadiscourse, Ädel (2005) argues that the two terms describe distinct phenomena and should be systematically differentiated. Ädel (2003, 2005) demonstrates how this may be achieved using Jacobsen’s framework of metalinguistic, directive, expressive and referential functions of language (Jakobsen 1995), arguing that metadiscourse is characterised by the metalinguistic at least, and often the directive and expressive functions of language as well (2005: 157-8). Put simply, while metadiscourse is reflexive language, evaluation is not (Ädel 2005: 158). Rahman agrees, emphasising that metadiscoursal expressions are “text-internal; that is, they refer to the text itself” (2004: 47). The text-referring characteristic of metadiscourse is also central to Mauranen’s (1993) discussion of ‘text ‘textual’ in function at all. The authors therefore propose the terms ‘interactive’ and ‘interactional’ metadiscourse, following Thompson (2001). A better and simpler solution to the terminological problem is that suggested by Ädel (2005), as discussed in the following. <?page no="177"?> 177 reflexivity’ (metadiscourse) and Hyland and Tse’s description of metadiscourse as “self-reflective material” (2004: 156). There is thus both a theoretical justification and a practical set of criteria for distinguishing (textual) metadiscourse from evaluation. In consequence, narrower definitions of metadiscourse exclude the writer’s attitudes to what is said (evaluation), and focus instead on the role of metadiscourse in text organisation and reader understanding of text organisation (Busch-Lauer 1992a; Mauranen 1993a, 1993b; Valero-Garces 1996; Bäcklund 1998: Trumpp 1998; Bunton 1999; Ädel 2003, 2005; Rahman 2004). According to non-integrative definitions, metadiscourse consists of the “linguistic elements that show how the text is organised or that make references to the text itself” (Ädel 2005: 156). Metadiscourse therefore involves commenting on and demonstrating relationships between concepts and propositions as they unfold, in which case it could be more aptly defined as ‘discourse about discourse management’, rather than the original catch-all definition of ‘discourse about discourse’ (Rahman 2004: 37-40). A phenomenon such as this is theoretically and practically different from evaluation, which is “the expression of the writer’s or speaker’s opinion” (Thompson & Hunston 2000: 2). A distinction between metadiscourse and evaluation fits in very well with those broader definitions of metadiscourse which distinguish between textual and interpersonal types. The textual metadiscourse of integrative definitions is therefore metadiscourse proper, while so-called interpersonal metadiscourse is in fact evaluation. In this way, research on evaluation which has nothing to do with the organisational features of discourse (such as Channell 2000; Conrad & Biber 2000; Hunston & Sinclair 2000; Römer 2005; Bednarek forthc.) is freed from being forcibly categorised as metadiscourse. A narrower, non-integrative definition of metadiscourse therefore has a number of advantages. For our purposes, textual or organisational metadiscourse - metadiscourse proper, according to narrower definitions - is of most interest, since it is the type consistently identified as playing a major role in author-reader interaction. For Hyland, textual metadiscourse is used to organise content in ways that the imagined audience is expected to find “coherent and convincing” (2000: 112). Its use therefore depends on “the writer’s assessment of a reader’s possible processing difficulties, intertextual requirements and need for interpretative guidance” (Hyland 2000: 111). For Crismore, “textual metadiscourse makes the primary discourse and [author] seem coherent and logical to readers” (1989: 43). Scholars who adhere to a narrower definition of metadiscourse emphasise the role it plays in facilitating communication between author and reader by ensuring the adequacy of the author’s communicative actions, managing and <?page no="178"?> 178 guiding the reception of content, and ensuring reader comprehension (Busch-Lauer 1992a: 49). Trumpp, who likewise adopts a non-integrative position, defines metadiscourse as “verstehensfördernde Elemente” [elements which aid comprehension]: such elements structure the text, provide orientation, direct the reception process and guarantee the exchange of ideas (1998: 60). Since notions of logical structure are cultureand perhaps discipline-specific, as indeed are structural patterns in written academic discourse (see chapter one), it is germane to consider how authors from different cultural and disciplinary backgrounds use metadiscourse to render their texts logical and coherent to readers. In so doing, we can examine the possibility that British, US-American and German scholars, writing in five humanities disciplines, may have different expectations of their audience, and different ways of catering to readers’ anticipated needs. 4.5 From textual metadiscourse to text comments Textual metadiscourse, or metadiscourse proper, has been taken to encompass a variety of different features. Vande Kopple’s classification, which has been adopted with some variations by many later researchers, lists ‘text connectives’ (‘first’, ‘next’, ‘however’), ‘illocution markers’ (‘to sum up’, ‘to give an example’), ‘code glosses’ (‘in other words’, ‘namely’) and ‘narrators’, which explain the source of an utterance or piece of information (1980, quoted in Crismore 1989: 93). In a later categorisation, Vande Kopple (1985) distinguishes between text connectives, code glosses and action markers (the same as illocution markers in his previous categorisation), while in a yet later classification, Vande Kopple (1988) reduces textual metadiscourse to only two categories: connectives and code glosses; illocution markers were re-classified as belonging to interpersonal metadiscourse. Crismore (1989) largely adopts Vande Kopple’s (1985) categorisation, as does Hyland (2000), though Hyland adds some subcategories of his own, listing textual metadiscourse features as logical connectives, frame markers, code glosses, endophoric markers and evidentials (Hyland 2000: 111). Mauranen (1993b) distinguishes between ‘reflexivity’ of high and low explicitness: in the former, writers explicitly refer to the communication process (organisation and language of the text, the writing process, the reader), whereas in the latter, they use devices such as enumerations and conjunctions to structure the text without explicitly referring to it (1993b: 180-1). Mauranen (1993a) discerns three subtypes of ‘commentary’: reviews, previews and action markers. Crismore, Markkanen and Steffensen (1993) divide textual metadiscourse into <?page no="179"?> 179 ‘textual markers’ (logical connectives, sequencers, reminders and topicalizers) and ‘interpretive markers’ (code glosses, illocution markers and announcements). It is unclear what is ‘interpretive’ rather than ‘textual’ about giving examples (‘illocution markers’) or announcing coming material. In a more recent study of textual metadiscourse, Crismore examines logical connectors and action markers only, as well as what are termed “miscellaneous textual metadiscourse items”, such as ‘in fact’, ‘in order to’ and ‘in short’ (2004: 320). Like all categorisations, these suggested frameworks are subjective and idiosyncratic; many are also inconsistent. 65 None is to be understood as comprehensive. The unifying principle behind previous operationalisations of textual metadiscourse is an interest in the role of diverse features in organising or facilitating the reception of discourse. Presumably, the main idea behind previous categorisations is that readers who know what is going on at a particular point, or who are alerted to the relation between different parts of the text, will find it easier to process the material being conveyed and make connections between pieces of information. References to information provided at other points of the text draw attention to the text structure, and may help to emphasise the fact that a text is well-written or well-organised (at least in the author’s opinion), and a text which is easier to follow and appears better organised may well be more persuasive. Of course, such references not only aid the reader, but help the author to organise discourse as well. The role that metadiscourse plays in organisation and managing discourse is also central to the conceptualisation of metadiscourse adopted here. However, the specific focus of this study requires a different operationalisation of metadiscourse from that implemented in previous work. Though previous classification schemes served the purposes of past studies, for which they were conceived, they are not well-suited to the present investigation, which uniquely combines social interaction with cross-cultural and cross-disciplinary analysis. For a study interested in social interaction, it seems of marginal interest to count the number of 65 Rahman (2004) laments the inconsistency of metadiscourse classifications, especially those in Crismore and Farnsworth (1990), Crismore, Markkanen and Steffensen (1993) and Mauranen (1993a). Rahman suggests that the classifications proposed by Crismore and colleagues are the source of the problem, and points out that Crismore et al “have shown inconsistency even within the pages of a single study […] they violate their own criteria” (Rahman 2004: 34). Commentary, for example, is at one point part of interpersonal metadiscourse, and at another in the same study both interpersonal and textual (Crismore & Farnsworth 1990). Part of the problem lies in the terminology: as I pointed out above, the separation of so-called ‘interpersonal’ from ‘textual’ metadiscourse sets up a trap, since all metadiscourse is multifunctional. This is a separate matter from the distinction between metadiscourse and evaluation, both of which can also fulfil textual and interpersonal functions, and many other functions besides. <?page no="180"?> 180 times an author introduces an example into the text (connectives/ code glosses), or to tally the conjunctions and enumeration signals a research article contains (also connectives). 66 The reason for this is that text connectives such as ‘but’, ‘therefore’, ‘first’ and ‘second’, and enumeration signals, are more the devices used to structure text than a comment on the text. 67 In saying ‘but’, or structuring points in a numbered list, the author does not step outside of the text to comment on it for the reader. In contrast to numerous previous studies, connectives were therefore not examined here (Vande Kopple 1980, 1985; Crismore 1989; Crismore, Markkanen & Steffensen 1993; Mauranen 1993a; Hyland 2000; Rahman 2004). Likewise, features included in Mauranen’s (1993b) ‘reflexivity of low explicitness’, expressions which “organise the text and indicate functions of its parts without explicitly referring to the communication process which is taking place via the text” (1993b: 80), were not counted here. For a study of social interaction, such items are obviously of lesser interest than are those places where the writer explicitly refers to the communication process (‘highly explicit reflexivity’), because social interaction occurs where authors talk to readers about the textual organisation, not where authors simply employ organisational devices such as numbers. The role played by citations such as ‘according to X / (1990)’ or ‘Z states’, which Hyland christens ‘evidentials’ (2000: 111), in writer/ reader interaction is also minimal. ‘Visual metadiscourse’ (Crismore, Markkanen & Steffensen 1993), which includes devices such as exclamation marks and underlining (Hyland 2000: 100) or font size and white space (Crismore 2004: 312), is also not examined here, because it too seems at best a marginal aspect of writer/ reader interaction. As regards writer/ reader interaction, the use of 12 point rather than 10 point type face might perhaps reveal, at best, that the author expects a readership with poor eyesight. References to tables and examples did not constitute a separate area of investigation, as they did in Hutz (1997) and Fandrych and Graefen (2002). However, unlike Dahl (2004), I decided not exclude all references to tables and examples from the analysis. Simple references to tables such as ‘This table shows…’ were excluded from the analysis, since they do not comment on the text. Explicit references to tables or examples provided at an earlier or later point in the 66 Crismore, Markkanen and Steffensen (1993) variously refer to enumerations such as ‘first’, ‘second’ and ‘third’ as both ‘sequencers’ and ‘topicalisers’ (1993: 42). The distinction between sequencers and topicalisers is obscure. 67 Hyland states that he studied only those connectives which fulfilled a metadiscoursal function, “helping the reader to interpret links between ideas, rather than simply contributing to syntactic coordination” (Hyland 2000: 112). Unfortunately, since no corpus examples of this distinction are provided, one is left wondering how items such as ‘and’, ‘but, ‘also’ and ‘so’ help the reader, and how this function was distinguished from their syntactic function. <?page no="181"?> 181 text were counted however. Finally, Schiffrin (1980) devotes much attention to demonstratives, which were too low level and too formal a phenomenon to be included separately in the present discourse-oriented categorisation. Many of these same features are also unsuited to a cross-cultural, cross-disciplinary study. It is doubtful whether important or even interesting insights about cultural or disciplinary differences in academic discourse can be gained from comparing the use of conjunctions and enumeration signals: one might discover, for example, that authors from culture A employ the number 3 more often than authors from culture B, or that philosophers write ‘but’ more often than historians, all of which seems rather slight and dull. Granted, citation practice across cultures and disciplines probably is an interesting area of analysis, but since there is no way of automatically identifying intertextuality, and the topic in any case seems worthy of an investigation of its own, I leave this object of study to another occasion. Rahman’s (2004) distinction between references to discourse acts (‘As will be noted below…’), discourse entities (‘In the following section’) and discourse labels (‘This section…’) is also interesting, but not adopted here, since it is of less interest to a cross-cultural and crossdisciplinary study. The ‘miscellaneous’ items examined by Crismore (2004) seem of little interest from either an interactional, cross-disciplinary or cross-cultural perspective. In light of this, the categories established in previous studies are not adopted here. Instead, the analysis focuses on a type of metadiscourse that is particularly relevant to the study of social interaction in academic texts, namely text comments. Text comments are a specific type of metadiscourse which fulfils a markedly interpersonal function and also appears to differ in use across cultures and perhaps disciplines. Text comments are those places within texts where authors explicitly comment on the structure, organisation or progression of their writing; they are not remarks on the content of the text. Text comments are part of the way in which academic authors shape the text interactively to fit their audience (Thompson 2001: 61). Through such devices, authors attempt to make their writing more easily comprehensible, more acceptable, and therefore more persuasive to potential readers. Text comments play an important role in author-reader interaction, and exemplify a number of interlinked features which scholars have perceived as central to metadiscourse, such as consciousness of text producing processes (Rossiter 1974; Rahman 2004), social interaction (Enkvist 1975; Williams 1981; Vande Kopple 1985; Crismore 1989; Crismore & Farnsworth 1990; Mauranen 1993a, 1993b), reader guidance (Crismore 1989; Rahman 2004; Hyland 2005) and textual focus (Ädel 2005). There is <?page no="182"?> 182 therefore much to be said for analysing this specific segment of metadiscourse in more detail. 4.6 Types of text comments For the purposes of the investigation, text comments were divided into seven types, as shown in Figure 4.1. A detailed discussion of the discourse purposes for which authors use each type is provided in section 4.10. The taxonomy was derived using bottom-up methods, that is, by first reading through the texts in the corpus in their entirety and then considering a framework for distinguishing different types. The manual analysis was rechecked twice in order to ensure consistent assignation to the various TC categories. 68 It is hoped that the bottom-up approach frees the taxonomy of cultural bias, since the categories are based not on preconceived notions of types and functions, but derived from an open-ended examination of the texts themselves. 69 The initial impression was that, though text comments were used for a great variety of discourse purposes, nonetheless they could be divided into several main groups, which occurred across all three cultures and five disciplines represented in the corpus. There were comments on the overall aims and structure of the text, which generally occurred near the beginning of the articles, and explained what would be done. In both discourse purpose and text position, these references constituted a distinct type of text comment. In analogy to this type, there were references to what would or could not be done; this was another type, with a different discourse purpose from positive statements of intention. Upon reading the corpus texts, I noted that the German articles in particular included a large number of comments on the necessary or obligatory nature of particular textual actions or pieces of information. The frequency of this type, or rather its unusual prominence in the German texts, suggested that it should be placed in a separate category. Endophoric reference was clearly a category of its own, or rather two categories, anaphora and cataphora, following Halliday & Hasan (1976). There were also comments, usually at the end of the article, which referred to the aims of the article as a whole, and what the author hoped to have achieved or shown. These too 68 I am very grateful to Daniela Kolbe for helping me establish intercoder reliability of the text comment types. 69 This question is particularly important in a cross-cultural study. The challenge is to establish a relevant, culturally-sensitive taxonomy which does not inadvertently establish one culture as the norm and then examine the other in terms of its statistical and societal deviance from this norm. This is the danger of working with pre-established dichotomies and frameworks. The solution is to start from the texts, formulating a taxonomy only after extensive open-ended observation. <?page no="183"?> 183 seemed to form a distinct category. Having established these categories, I then began the close analysis, re-reading all of the corpus texts. Figure 4.1. Taxonomy of text comments developed for the present study ______________________________________________________________________ 1. Overarching text comments: what the article is about My aim is to illustrate how…(HIUS) Gegenstand dieser Untersuchung sind…[The subject of this study is…] (LGDE) 2. Exclusive text comments: what the article is not about One might also challenge…But I will leave that debate for another day. (FLUS) Die damit berührte Frage der…kann hier nicht weiter verfolgt werden. [The related question of…cannot be pursued further here.] (LTDE) 3. Introductory text comments: what is being done at present / will be done immediately following In this section, we argue that the facts about…pose severe difficulties for…(PHUK) Zunächst wende ich mich dem Realitätsproblem zu. [First I turn to the reality problem.] (FLDE9) 4. Explanatory text comments: why things are being/ must be done In order to reach an understanding of why…background needs to be brought into focus…(FLUK) Für die Detailanalyse reicht diese Idealisierung nicht aus. Dazu müssen weitere Eigenschaften von Grammatikalisierungsprozessen eruiert werden. [This idealisation is insufficient for a detailed analysis. For this purpose, further characteristics of the grammaticalisation process need to be elicited.] (LGDE) 5. Prospective text comments: references to what will be done later The reason for…will emerge later. (PHUK) Wie wir noch genauer sehen werden,…[As we shall see more clearly…] (PHDE) 6. Retrospective text comments: references to what was done before … as was shown in an earlier section. (LGUS) Wir sahen bereits: …[We have already seen…] (HIDE) 7. Concluding text comments: what the article showed In conclusion, we see no obvious way of adding an axiom that will distinguish between S and S*. (PHUK) Die Untersuchung ergibt, dass…[The study shows that…] (HIDE) ______________________________________________________________________ Overarching text comments have as their topic the objectives and organisation of the entire article. Some overarching text comments are very <?page no="184"?> 184 long, consisting of several sentence-length declarations of intent, immediately following one another: (89) The analysis in this article is essentially threefold. The first section discusses how….It considers…The second part examines the…Finally, a brief consideration is given to…(HIUK) (90) Die folgenden Ausführungen gliedern sich in drei Abschnitte. Zunächst soll… begründet werden…Der anschließende Abschnitt ist einer…gewidmet…Der dritte Teil fasst…zusammen und skizziert abschließend…. (91) [The following discussion is divided into three parts. First…is accounted for…The following section is concerned with…The third section summarises…and finally outlines…] (PHDE) Following Hyland and Tse (2004: 157), multiple-sentence overarching TCs were regarded as one discourse unit, and therefore counted only once. By this I mean that, as in Schiffrin (1980), the unit of analysis was not formal, that is syntactic (the sentence), but discoursal, as seems fitting for the investigation of a discourse phenomenon. Where the sentences did not immediately follow one another, but were dispersed over a whole text section, they were counted individually. Exclusive text comments too are concerned with the objectives and organisation of the entire article. In analogy with overarching TCs, it seemed interesting to consider the places in the text where authors explain what they will not be doing: what the aims of the article are not, and what topics will not be discussed. Often, what authors chose not to do is just as revealing as what they do chose to do. This type of text comment has no parallel in previous investigations. (92) For our immediate purposes, we may sidestep this theoretical disagreement, since the two approaches converge on the main point being made here: …(LGUK) (93) There are interesting philosophical questions about….; but it would take us too far afield to consider them here. (PHUS) (94) Auf die Gründe und Umstände dieses Phänomens braucht hier im Einzelnen nicht eingegangen zu werden; wichtig ist, daß… [There is no need to go into the reasons and circumstances of this phenomenon in any detail here; the important thing is that…] (HIDE) Introductory text comments refer to actions being carried out at that moment, or announce those which are to follow immediately. Included in the introductory TC type are exhortations, often in the form of collective <?page no="185"?> 185 first-person imperatives or, in German, also expressed with an impersonalpronoun: (95) Let us now turn to the issue of how… (LGUK) (96) We can turn next to the signature feature of…and ask how…(PHUS) (97) Beginnt man mit der…[If one begins with…] (LTDE) Explanatory text comments refer to why the article is structured in a particular fashion, or why a particular piece of information is or must be provided at a certain point. They are therefore explanatory in function. This type gives an indication of the constraints under which academic authors operate, as perceived by the authors. The justification implicit in such explanations also gives an indication of imagined reader expectations, as perceived by the author, and helps to involve the reader in the text. Other text comments in this category are less explanatory and more obligatory in nature, noting what must be done, or should be done, without necessarily explaining why this is so. Prospective text comments refer to material to be provided later in the text, or in a later text section, while retrospective text comments refer to material provided at a previous point in the text. Finally, concluding text comments offer a summary of what has been shown in the article: a concluding comment on the text. 70 It must be accepted that no taxonomy is perfect. Since language is multifunctional, there will always be some degree of overlap between categories, as virtually all scholars recognise (Rossiter 1974: 37; Schiffrin 1980: 201; Vande Kopple 1980: 6, quoted in Crismore 1989: 72; Crismore 1989: 49, 60, 94; Hyland & Tse 2004: 175; Ädel 2005: 158-9; Hyland 2005: 176). In the current investigation, this meant that there were occasionally examples that fell into more than one category; these were coded once only, in the category which seemed best to represent their function. The following example was coded as an explanatory TC, though it could also have been a prospective TC, because the main discourse purpose here is to justify the provision of the examples, not explain that they are coming: (98) Die folgenden Beispiele mögen zur Illustration der vielfältigen Bezeichnungsvarianten dienen: …[The following examples may serve to illustrate the diverse descriptive variants…] (LGDE) 70 This is not to imply that texts without concluding TCs have no conclusion. It simply means that the conclusions were not explicitly marked as such with a text comment. <?page no="186"?> 186 Example (12) was coded as a retrospective TC, though it could also have been a prospective TC, because the main discourse purpose here is to refer to previous material, not prepare the reader for what is coming: (99) Die genannten Sätze will ich unter dem Terminus weiterführender w- Relativsatz, kurz w-wRS, fassen. [I wish to capture the sentences mentioned with the term continuative w-relative clauses, in short wwRS.] (LGDE) Example (13) was coded as an explanatory TC, though it could also have been an introductory TC, because the main discourse purpose here is to justify the fact that something must be repeated, not simply comment on the current speech action: (100) Allerdings muss hier wiederholt werden, dass die Abgrenzung zwischen Objekten und Adverbialen keineswegs trennscharf ist. [However, it must be repeated here that the distinctions between objects and adverbials is by no means selective.] (LGDE) In the present study, there were only very few cases in which a text comment spanned two categories, and thus no need for a ‘mixed’ or ‘miscellaneous’ category. In most cases, what seemed like pragmatic overlap, or ambiguous discourse purpose, on the surface, in fact was not ambiguous when examined in the wider context of the paragraph or text section. Despite the fact that all taxonomies are imperfect, including that applied in this study, an imperfect taxonomy is still preferable to none at all. The reason for this is that different types of text comments serve different discourse purposes. Without a taxonomy of some type, it is not possible to distinguish between these different purposes; this would prevent detailed analysis of the sort necessary for a study of social interaction in academic writing. Before we proceed to the results from the corpus, I wish to place text comments in the context of previous research. An awareness of the broader body of research into text comments and related phenomena can enrich our perspective. It also helps to clarify the relationship between text comments and other conceptualisations of the devices that help to organise and manage discourse. <?page no="187"?> 187 4.7 Text comments, discourse deixis, advance organisers and co. Under a variety of aliases, text comments and closely-related phenomena have been the subject of much previous research. As befits the focus of this study, the following survey covers only research on Englishand/ or German-language academic discourse. Some previous studies are listed in Figure 4.2. Meyer (1975, 1985; Meyer & Poon 2001) focuses on ‘signalling’, which is defined as textual elements that emphasise or point out certain aspects of the propositional content without being content themselves. Signalling types include summaries, pointer words, devices for specifying relations between content, and references to material that occurs later in the text. Tyler (1994) defines signalling as metalinguistic devices that explicitly organise discourse by guiding reader and listener interpretation of coming information. Like Meyer (1975), Tyler (1994) lists previews, summarisers and pointer words or emphasis markers as signalling devices, as well as logical connectives, which are familiar from Vande Kopple’s work. Keller (1979) refers to the devices used by speakers to structure content and conversational procedure as ‘gambits’. Keller maintains that gambits are primarily used to indicate what the speaker is about to say next (semantic information), and prepare the listener for this, though he in fact distinguishes seven major categories of semantic frame gambits that fulfil many more functions. Dillon writes of ‘traffic signals’ (Dillon 1981: 60), such as ‘turning now to the second point’, which Vande Kopple would classify as action markers. <?page no="188"?> 188 Figure 4.2. Various terms for text comments and closely-related phenomena in previous research ______________________________________________________________________ Study Terms for text comments and closely-related phenomena ______________________________________________________________________ Fillmore 1975; Lyons 1979; discourse deixis Levinson 1983, 2004; Lenz 1997 Meyer 1975, 1984; Meyer & Poon 2001 signalling Keller 1979 gambits Schiffrin 1980 discourse brackets Vande Kopple 1980 illocution markers Dillon 1981 traffic signals Vande Kopple 1985; Crismore 1989; action markers Hyland 2000 Clyne 1987; Schröder 1987; Trumpp 1998 advance organisers Schiffrin 1987; Pulcini & Furiassi 2004 discourse markers Crismore 1989 signposts, sequencers, logical connectors Kretzenbacher 1990 Organisationshilfen [aids to organisation] Mauranen 1993a commentary (reviews, previews and action markers) Mauranen 1993b reflexivity of high explicitness Tyler 1994 signalling devices Hutz 1997 metakommunikative Äußerungen [metacommunicative utterances] Baumann 1998 Gliederungssignale [structuring signals] Graefen 2000b; Fandrych 2001; Textkommentierungen, Textkommentaren Fandrych & Graefen 2002 [text commentaries], text commenting devices Hyland 2000, 2005 endophoric markers, frame markers Crismore, Markkanen & Steffensen 1993 textual markers (logical connectives, sequencers, reminders and topicalizers) and interpretive markers (code glosses, illocution markers and announcements) Crismore 2004 logical connectors, action markers Swales 2004b prospective roadmapping ______________________________________________________________________ <?page no="189"?> 189 Schiffrin (1980) focuses on ‘discourse brackets’, which are a form of meta-talk with an “organisational function” (1980: 216). This is similar to Schiffrin’s later work on ‘discourse markers’, defined as “sequentially dependent elements which bracket units of talk” (1987: 31). Like text comments, discourse markers establish coherence in discourse by signalling relations between units of talk. Indeed, a recent definition of discourse markers, as “items which are used by speakers to structure and control conversation, and organise discourse in terms of cohesion and coherence between its parts”, could serve equally well for textual metadiscourse (Pulcini & Furiassi 2004: 108). There is even the same distinction as in metadiscourse between interpersonal/ interactional and textual/ metatextual functions (Pulcini & Furiassi 2004: 109). The main difference between discourse markers and textual metadiscourse is that discourse marker research has always focused primarily on spoken interaction. For this reason, some of the linguistic devices listed in discourse marker taxonomies do not feature in classifications of textual metadiscourse. 71 This notwithstanding, the two phenomena are nonetheless very similar in function. Clyne (1987), Schröder (1987) and Trumpp (1998) work with the concept of ‘advance organisers’, that is, devices which “explain the path and organisation of a paper” (Clyne 1987: 229). Trumpp slightly alters the concept, defining advance organisers as “alle metakommunikative Äußerungen, welche die folgende Textstruktur ankündigen” [all metacommunicative elements which announce the structure of following text] (1998: 60), thereby excluding those comments which refer to past material. For Schröder, advance organisers are defined functionally, and serve the “expliziten Steuerung und Organisation des Diskurses” [explicit direction and organisation of discourse] (1987: 159). Again, there is a great deal of overlap here with text comments. Another closely-related phenomenon is ‘discourse deixis’, which Levinson defines as references to “earlier or forthcoming segments of the discourse”, such as ‘as mentioned before’, ‘in the next chapter’, or ‘I bet you haven’t heard this joke’ (2004: 48; see also Fillmore 1975; Lyons 71 Pulcini and Furiassi, for example, list discourse marking items ranging from non-verbal signals (gestures and gazes), hesitation sounds (er, erm, mm), interjections (oh, ah, uhu), coordinators/ conjunctions (and, but), connectors/ adverbs (well, now, then, actually) to clauses (y’know, I mean, I think) (2004: 108). Obviously, many of these features do not occur in written academic discourse. The same is of course true for all taxonomies based on informal spoken language rather than formal written texts: tag questions, turn-taking signals and ‘communication control signals’ (Keller 1979; Bublitz 2001) such as ‘is that clear? ’ or ‘I don’t get you’ also do not occur in academic writing, though they might occur in informal written genres (such as internet chats) or formal spoken genres (such as academic lectures). <?page no="190"?> 190 1979; Levinson 1983; Lenz 1997). Since one must know where the current coding/ reading/ recording point is in order to interpret such references, Levinson regards such devices as “quintessentially deictic in character” (2004: 48). Discourse markers, as in Schiffrin (1987), are viewed as one subset of discourse deixis (Levinson 2004: 50). The functional definition of discourse deixis as “to do with keeping track of reference in the unfolding discourse” (Meliani 2002: 7) is very close to that of text comments. There are various other idiosyncratic conceptualisations. Kretzenbacher coins the term ‘Organisationshilfen’ [aids to organisation], and distinguishes between ‘previously offered’ types, which refer to previous material, and ‘preparatory’ types, which prepare the reader for coming material (Kretzenbacher 1990: 14). Hutz (1997) analyses ‘metakommunikative Äußerungen’ [metacommunicative utterances], which are closely related to text comments in function and scope, as we shall see below. Baumann (1998) prefers to talk of ‘Gliederungssignale’ [structuring signals], which are devices used to facilitate ‘cognitive Strukturierung’” [cognitive structuring] for the reader (1998: 312-3). Fandrych and Graefen (Graefen 2000b; Fandrych 2001; Fandrych & Graefen 2002) formulate the concept of ‘Textkommentierungen’ and ‘Textkommentaren’ [text commentaries], which they translate in their English-language publications as ‘text commenting devices’. These are basically the same as text comments, though the subtypes are somewhat different, as we shall see in the next section. Crismore works with a variety of phenomena, all subtypes of metadiscourse, which are essentially synonymous with text comments, such as ‘signposts’, ‘sequencers’ and ‘logical connectors’, all three of which are used by authors to assist readers’ interpretation of content (Crismore 1989: 51). Finally, Swales (2004b: 191) considers prospective metadiscourse, or ‘roadmapping’, which can be seen as a subtype of text comments. While the terminology is not shared, the conceptualisation of many of these phenomena, and the elements in these taxonomies, have much in common with text comments. Signalling, gambits, traffic signals, discourse markers, advance organisers, discourse deixis, roadmapping, call it what you will: text comments, or phenomena very closely related to them, have been the subject of much research precisely because of the central role they play in organising discourse and making it coherent to readers. The breadth of previous work on phenomena closely-related to text comments is in part a reflection of a shift of emphasis in text-linguistic research, as discussed by Ädel, who perceives that up to the 1990s the focus was largely on propositional meaning in texts (2005: 156). From this perspective, early work on signalling (Meyer 1975, 1984), which is defined as non-propositional material, can be seen as a reaction to the almost exclusive focus at <?page no="191"?> 191 the time on the subject matter of texts rather than on functional or rhetorical aspects. Though the early conceptualisation of metadiscourse as all non-propositional material was relatively unsophisticated, it opened up new perspectives for research, as reflected in Keller’s (1979) work on gambits, and Dillon’s (1981) work on traffic signals, which are aspects of non-propositional content defined by their communicative function. Schiffrin’s work on discourse brackets and discourse markers continued the functional perspective (1980, 1987), as did later work on advance organisers (Clyne 1987; Schröder 1987; Trumpp 1998). Text comments too are defined primarily by the discoursal function they serve. Like gambits, traffic signals, discourse markers and advance organisers, text comments organise and manage (segments of) discourse, aiming thereby to ease comprehension. This is exactly the same function as that at the centre of concepts such as Kretzenbacher’s ‘Organisationshilfen’ [aids to organisation] (1990), Hutz’s ‘metakommunikative Äußerungen’ [metacommunicative utterances] (1997), Baumann’s ‘Gliederungssignale’ [structuring signals] (1998) and Fandrych and Graefen’s text commenting devices. These phenomena are very closely related indeed. Though research on discourse deixis emerged from a different line of thought, and is part of a much larger range of phenomena that are unrelated to text comments, nonetheless discourse deixis too is very similar in function to text comments. Both concepts focus on how discourse management is achieved through text reference. In sum, though the terminology, conceptualisation and operationalisation of these many phenomena of course differ across studies, the guiding principle and central interest behind much of this research is the same as that behind text comments. We see, therefore, that text comments not only exemplify a number of functions which previous researchers have identified as central to metadiscourse; they are also very closely related to a number of other phenomena that scholars have perceived as vital to the organisation and management of discourse, and to the interaction between author/ speaker and reader. In addition, text comments have attracted much interest from a cross-cultural perspective. 4.8 Text comments from a cross-cultural perspective From a cross-cultural perspective, text comments are of particular interest since a number of studies have indicated the existence of culturally specific text commenting patterns among British and / or US-American and German academics. Indeed, if as Crismore proposes, the “choices and inferences that authors and readers make are a result of their attitudes and beliefs <?page no="192"?> 192 about the world and the other” (1989: 124), one would expect authors from different cultural backgrounds to communicate differently with their potential readers. However, there is as yet no consensus among researchers about how or even if cultural preferences differ in this respect. In fact, as we shall see, the results of previous research on text comments in Englishand German-language academic prose directly contradict each other. It is therefore all the more important to revisit this type of metadiscourse here. Focussing on text comments permits the results from the SCEGAD corpus to be contrasted with those of previous studies; claims about cultural tendencies made in other studies can also be examined based on the larger quantity of data which SCEGAD provides. In addition, the closelycontrolled data set offered by the SCEGAD corpus permits the influence of many other variables apart from culture to be investigated as well. It is hoped that a clearer picture of this aspect of language use may thus emerge. First, however, let us examine the results of previous contrastive research into text comments in Englishand German-language scholarly writing. In chapter two, I noted that cross-cultural research into academic discourse relies upon dozens of different methodologies, and I discussed a variety of approaches, and a still greater variety of phenomena that have already been examined. Accordingly, it is rare to encounter a phenomenon that has been the subject of more than one or two separate investigations. Text comments constitute a valuable exception to this general rule; valuable in the sense that being able to draw on a number of studies ideally permits a fuller picture to be gained of a phenomenon that many scholars have perceived as central to social interaction in academic writing. I shall discuss five previous contributions to cross-cultural research on text comments: Clyne (1987), Hutz (1997), Baumann (1998), Trumpp (1998) and a group of studies by Fandrych and Graefen (Graefen 2000b; Fandrych 2001, 2003; Fandrych & Graefen 2002; Fandrych 2005). 72 The first researcher to examine text comments in Englishand German-language academic writing was Clyne (1987). Although no figures for relative frequencies are provided, Clyne states that the US- American, British and Australian scholars in his corpus of research articles were “more likely” than their German colleagues to use advance organizers (1987: 229). No subcategories are distinguished. As the name suggests, it appears that Clyne’s term refers to comments on coming material only. 72 For obvious reasons, this overview concentrates on contrastive research into Englishand German-language academic texts. Some other language combinations in studies of metadiscourse are English, Norwegian and French (Dahl 2004), English and Slovenian (Peterlin 2005), French and German (Sachtleber 1990), English and Finnish (Mauranen 1993b) and English and Swedish (Melander, Swales & Fredrickson 1997). <?page no="193"?> 193 The study found that the English-speaking scholars were also more likely to place advance organisers at the start of their articles (1987: 229). For Clyne, the use of advance organisers is connected to linearity. This is hard to understand, since theoretically at least it would be absolutely possible for an author to “explain the path and organization of a paper” that is not structured in a linear fashion (1987: 229). Conversely, a text structured in a linear fashion need not necessarily contain any advance organisers. Of course, in Clyne’s data it might be the case that linearity and use of advance organisers correlate, but this is not demonstrated. In Clyne’s opinion, the greater use of advance organisers makes the papers by the English-speaking scholars easier to understand, because “people know what to expect” (1987: 229) and the texts are “more predictable” (1994: 163). The connection between advance organisers and predictability seems to be a non-sequitur: surely, if text structure were entirely predictable, no advance organisers would be necessary. The very notion of predictability is in any case only meaningful in relation to culturally specific expectations of discourse patterns: no doubt German scholars find texts written by Germans just as predictable as British academics do British-authored texts. 73 Expectations, that is, are culturally relative. Clyne reports that, in contrast to their English-speaking colleagues, the German authors in his corpus give advance organisers, if at all, in an “obscure location”, by which Clyne presumably means in a location different from that expected by himself and preferred by the English-speaking scholars in the corpus. In the German texts Clyne examined, advance organisers more often occurred not at the beginning, but in a later location (not further specified by Clyne), “almost as if to express embarrassment about this formal adherence to the conventions of an international journal.” (1987: 229). The phrasing of this last observation explains in part the offence which Clyne’s study caused, and the adverse reactions of some German researchers to this section of it. For Clyne, the use of advance organisers is connected to Hind’s notion of writer versus reader-responsibility (1980, 1987): since English-speaking scholars come from a writer-responsible culture, it is they who must make the effort to explain their text to readers, rather than the readers being responsible for understanding the text. German scholars, originating from an ostensibly reader-responsible culture, apparently feel no such need to 73 Clyne’s own work in fact confirms this: when German and Australian academics were asked for their response to doctored (very ‘Anglo-Saxon’and very ‘German’structured) texts, it emerged that the German scholars were more easily able to understand texts that were less linear in structure than were the Australian scholars. The German academics were also more tolerant of non-linear text-structure. It is therefore all the more surprising that Clyne himself treats linearity and comprehensibility as one and the same, and judges non-linear structure as deficient (see Clyne, Hoeks & Kreutz 1988; Clyne 1991). <?page no="194"?> 194 guide or aid their audience. Thus, “the English compliment that one’s academic text is ‘easy to follow’ could be interpreted as an insult in German” (Clyne 1994: 187). These comments are not supported by any empirical evidence; they certainly do not follow from Clyne’s (undisclosed) results. 74 We have once again descended into cultural stereotypes. No wonder then that a German linguist responded thus: Seit Michael Clyne lastet sozusagen der Verdacht auf deutschen Wissenschaftlern, dass sie stark inhaltsbezogen (quasi selbstverliebt) und wenig leserfreundlich schreiben, wenn nicht gar rücksichtslos gegen den Leser sind. [Since Michael Clyne’s work, the accusation has weighed on German scholars that they write in a strongly content-orientated (basically self-infatuated) and insufficiently readerfriendly manner, even that they are ruthlessly inconsiderate towards the reader] (Graefen 2001: 123) 75 Graefen’s reaction highlights a fundamental problem in contrastive research, which will be discussed in more detail in the following chapter. The constant temptation, and the constant danger, in cross-cultural studies is to connect observable linguistic features with personal observations about a foreign culture. This practice is usually inaccurate, since it is based on culturally-biased perceptions of the other group, as judged from outside. It also risks creating a climate in which cultural stereotyping, presented as research, is given far more credence and legitimacy than it deserves. It requires the correction of a cultural insider such as Graefen, a fellow linguist, to uncover the no-doubt unintentional cultural bias innate in Clyne’s dichotomy. The second of the five contrastive studies of text comments (Hutz 1997) examined ‘metakommunikative Äußerungen’ [metacommunicative utterances] in social-psychology research articles. The utterances were defined functionally as serving reader reception through text-organisational 74 Similar comments about German academic culture have been made even by German scholars: Clyne (1994) quotes Ströbe (1976), who opens a book review with the words “German students are brought up in the belief that anything which is easy enough for them to understand is dubious and probably unscientific” (1976: 509; see also Spillner 1982: 48). However, what Clyne (1994) does not mention is that in this review Ströbe severely criticises the work of a German colleague precisely because of its lack of reader-friendly presentation. At the very least, the admiration of impenetrable style is therefore not a universal good in German culture; I suspect also that attitudes towards appropriate scientific style may have changed considerably in Germany over the past few decades (see chapter 2, section 5). Ultimately, the onus is upon researchers themselves to offer empirical evidence for or against such hypotheses, rather than endlessly repeating them, adding at most an anecdote or two. 75 The German rücksichtslos in this quote is somewhere between English ‘inconsiderate’, ‘ruthless’ and ‘oblivious’: it is certainly a most negative epithet with which to describe a culture. <?page no="195"?> 195 signals, helping to lead the reader through the text, ease comprehension and structure the text (1997: 154, 166). Several types were distinguished, as can be seen in Figure 4.3: Figure 4.3. Taxonomy of metacommunicative utterances according to Hutz (1997) ______________________________________________________________________ Metacommunicative utterances relating to discourse goals and the topic of the whole text: 1. Metacommunicative utterances relating to text organisation and reception management 1. References to coming material and announcements 2. References to past material 3. Text structuring and references to the textual macrostructure (‘First…second…’, ‘The discussion of the results will concentrate on…Before this,…In a final point,…’) 3. Metacommunicative utterances to ensure comprehension 1. Definitions 2. Examples 3. Paraphrases 4. Metaphors, comparisons, analogies 5. Conclusions (‘In general…’) 4. Metacommunicative utterances relating to the visualisation of text information (references to tables, pictures and diagrams) 5. Metacommunicative utterances addressed to the reader (directives; indirect directives with German ‘man’ [one]) ______________________________________________________________________ Though Hutz (1997) expected to find a higher frequency of metacommunicative utterances (MU) in the US journal he examined, the results showed virtually no difference in frequency between the US and German texts (average 16.6 MU/ 100 sentences in the US-American journal versus 16.5 MU/ 100 sentences in the German journal). 76 Hutz maintains that the results for the German journal were skewed by a greater proportion of references to tables (category four), which comprised 27.8% of all MU in the German journal, but only 9.9% of all MU in the US journal (1997: 166). This begs the question of why references to tables were counted in the first place. 77 76 Hutz records the average frequency for the German journal as 16.4, not 16.5, but this appears to be a calculation error. Since the number of total sentences in the two subcorpora, and the raw numbers of MU are provided (1997: 82, 165), I have been able to double check the figures. Note also that the unit of comparison here, f/ 100 sentences, is problematic (see chapter 2, section 1) 77 What this really shows is that both the taxonomy and the method in this study are inadequate. Ostensibly, any differences in average frequency should actually be taken as <?page no="196"?> 196 Personally, I should think that the number of times an author refers to a table is of little to no interest from either a cross-cultural, cross-disciplinary or a social-interaction perspective. 78 At any rate, if the results for category four are discounted, the author maintains that the US-American journal contains a “deutliches Übergewicht” [clear prevalence] of MU. This is not the case: the results without category four would show a slight prevalence (average 15.0 MU/ 100 sentences for the US journal versus 12.0 for the German journal), which is most likely not statistically significant. Unfortunately, it is not possible to test this without access to Hutz’s data; the results are therefore inconclusive. For most MU types, there were only marginal differences in relative frequency across the two journals examined. It does appear that conclusions were more frequent in the German journal (0.18 versus 0.62/ 100 sentences), as they were in the German subcorpus of SCEGAD, but in the absence of significance tests one should not make too much of these results. The third study to analyse text comments in Englishand Germanlanguage academic discourse is Baumann (1998). Text comments, or ‘Gliederungssignale’ [structuring signals], are assigned a variety of functions. They cognitively organise and structure text content, emphasise and summarise central propositions, highlight factual differences, similarities and agreements, mark the referential unity of factors, and signal functional order in text progression (1998: 312). No examples are given for any of these functions. Further, Baumann opines that structuring signals “weisen darauf hin, dass der Autor den inhaltlichen Überblick über seine Darstellung besitzt. Sie sind ein Hinweis auf die Klarheit des Denkens” [show that the author possesses an overview as regards the content of his presentation. They are evidence of clarity of thought] (1998: 312). The parallel between structuring signals and clear thinking is completely unfounded. The study distinguishes no subtypes of structuring signals, though the two examples given indicate that both enumerations evidence of legitimate cultural differences, but apparently not even the author believes this in the case of references to tables. Having realised perhaps that his taxonomy is dysfunctional, the researcher then manipulates the results, in hindsight, to fit foregone conclusions: in this case, that the US journal contains more metacommunicative utterances. Findings that contradict the foregone conclusion are then declared to be illegitimate. 78 I am open to persuasion, of course. However, in this case, interesting results are not even theoretically possible, because the method is faulty. The inclusion of references to tables is problematic, because there is no way of distinguishing between articles that simply contain no tables or figures, and those which contain them, but do not refer to them. One therefore has neither a summary of how many texts contain figures and tables, nor a summary of how many texts, out of the total number which contain such elements, refer to these figures and tables. The same criticism applies to Fandrych and Graefen’s work, which is discussed below. <?page no="197"?> 197 (‘…(a)…(b)…’) and conclusions (‘In summary…’) were counted. The results show a roughly five-times higher frequency of structuring signals in the three English-language texts examined when compared to the three German-language texts. However, it is not clear whether this difference is statistically significant. In any case, these findings are neither representative nor generalisable (see chapter two, section one). Like Clyne (1987), Baumann interprets the higher frequency in the English-language texts as evidence of English-speaking authors’ emphasis on writer responsibility and corresponding desire to make their content easier for readers to process (1998: 313). Since ease of processing depends on what a reader expects and is used to, this parallel is unwarranted. Both the results and their interpretation are thus highly questionable. A fourth contrastive study (Trumpp 1998) examines advance organisers in a corpus of Englishand German-language sports science writing. Trumpp paraphrases advance organisers as “metakommunikative bzw. metadiskursive Äußerungen” [metacommunicative, that is metadiscursive, utterances]. These are defined as “situations-, kommunikations- und textbezogene Sprechhandlungen in Form von Kommentaren zum laufenden Kommunikationsereignis” [situation-, communicationand text-specific speech acts in the form of comments on the ongoing communicative event] (1998: 60). Trumpp uses the term in Clyne’s sense, to indicate coming material only, and does not distinguish any subtypes. The analysis revealed no significant differences in either the frequency or length of advance organisers across languages (1998: 94, 157), 79 which categorically refutes Clyne’s (1987) findings (1998: 54). This refutation is important, since Trumpp’s study is methodologically sophisticated, and is also the only one of the contributions discussed here to apply appropriate statistical analysis to the results. The final contrastive study, or rather group of studies, focuses on ‘text commentaries’, or ‘text commenting speech actions’, in a heterogeneous collection of 17 English-language and 19 German-language research articles (Graefen 2000b; Fandrych 2001, 2003; Fandrych & Graefen 2002; Fandrych 2005). Text commentaries are defined as “speech actions by which writers comment explicitly on the textual arrangement they have adopted” (Fandrych & Graefen 2002: 17). The function of these devices is explained as follows: Text commenting supports the reader in his/ her understanding and orientation with regard to certain aspects of the text. The author steps back for a moment in a way 79 Trumpp did find differences across text types, however: advance organisers were found to be more frequent in research articles than in non-specialist scientific articles and book reviews (1998: 178, 190). <?page no="198"?> 198 and adopts the role of a commentator of his/ her own text, explicitly directing the reader’s attention towards the structure of text or its internal coherence. (Fandrych & Graefen 2002: 20) The researchers identify six types of ‘text-commenting speech action’, as shown in Figure 4.4. For Fandrych and Graefen, ‘advance organisers’ are a sub-type of text comments, rather than an independent object of study (as in Clyne 1987; Schröder 1987 and Trumpp 1998). Like Hutz (1997), the authors include references to tables and figures in their tallies (see note 18), though such references are not discussed or analysed in the study because, as the authors explain, they are too frequent and are formulaic in realisation (Fandrych & Graefen 2002: 26). This again begs the question of why such references were counted in the first place. Figure 4.4. Taxonomy of text-commenting devices according to Fandrych and Graefen (2002) ______________________________________________________________________ 1. Declaration of main objective(s), topics and approaches of an article 2. Introductory qualification of speech actions (introduction or announcement of a thematically or illocutionarily new text section, which follows immediately) 3. Advance organisers (announcement of speech actions, which follow in a later text section) 4. Reactivation of information (reference to previous speech acts) 5. Self-assessment of the author (comment on findings of whole text) 6. Data integration (integration of figures and tables into the text) ______________________________________________________________________ Fandrych and Graefen’s general finding is that text commenting is generally not more common in the English-language texts they examined, though no overall trend across all 19 disciplines could be established (Graefen 2000b: 114; Fandrych 2001: 375; Fandrych & Graefen 2002: 21, 26). It is worth mentioning that the point of departure for the research conducted by Fandrych and Graefen was their dissatisfaction with the way in which Clyne (1987) had characterised German academic discourse (Fandrych & Graefen 2002: 35). The authors set out to prove that German scholars are not inconsiderate towards their readers, since they do not use fewer text-commenting devices (a non sequitur: see the comments on Clyne above). This issue aside, the other difficulty with this group of studies is their methodological unreliability (see also chapter two, section one). The results are presented as raw numbers, compared across non- <?page no="199"?> 199 parallel subcorpora of different sizes, and are based on a small text corpus controlled for neither discipline, composition date nor author native language. It is therefore not surprising that there was a very high level of internal variation within the subcorpora (Fandrych & Graefen 2002: 21, 26). As regards discipline-specific trends, the authors find that text commenting devices are most frequent in economics and psychology and least frequent in biology and literary studies research articles. However, since these results are based on the analysis of only one or two texts per discipline in each culture, they cannot be taken as representative. The cross-linguistic comparison of sub-type frequency is meaningless and incomprehensible, since only raw-number totals are given, rather than case-weighted average frequencies, with no indication of the word count of the two subcorpora or the length of each text (see figures in Fandrych & Graefen 2002: 28). In sum, therefore, we are left with one study that discloses no figures yet proclaims a clear trend (Clyne 1997) and two studies that proclaim a trend that is probably statistically insignificant (Hutz 1997; Baumann 1998). The fourth study (Trumpp 1998), which is the only one that is methodologically reliable, contradicts the findings of the first three studies, as does the final group of studies (Fandrych & Graefen 2002 etc.), which is, however, methodologically unsound. Additional complicating factors are the different definitions adopted for common terms (‘advance organisers’), the influence of uncontrolled variables, and the nonrepresentative non-generalisable small data samples. Despite the comparatively large number of previous studies concerned with (approximately) the same phenomenon, the situation is therefore as clear as mud. In view of these findings, text comments seemed a particularly suitable object of investigation for the present study. Perhaps a more empirically sound analysis of this phenomenon, based on the SCEGAD corpus, can shed some light on the situation. This brings us, therefore, to the results of the corpus analysis. 4.9 Text comments according to language, variety, culture, discipline, text position and other variables In contrast to the features examined in the previous chapter, which could be identified automatically, text comments had to be identified and coded manually. It is worth explaining briefly why automated analysis was not feasible. In an automated analysis, researchers must start with a list of linguistic items they wish to search for. In the case of text comments, however, it is not clear which items should be chosen, since text comments <?page no="200"?> 200 appear in all sorts of linguistic realisations, with and without person reference, adverbs of time and place and lexical items such as ‘study’/ ‘Studie’, ‘article’/ ‘Artikel’ and ‘section’/ ‘Abschnitt’. It is also of little use to search automatically for discourse deictic terms such as ‘here’/ ‘hier’ and ‘following’/ ‘folgend(e/ r/ s/ n/ m/ er/ en)’, since there is a great number of cases in which these items do not refer to the text at all. The following extracts, for example, contain ‘here’ tokens that do not refer to the text at all (the examples were all selected, for the sake of illustration, from the same randomly-selected subcorpus): (101) A party of about 30 men and an officer was cut off in a trench, when the officer said to his men, “Look here, we must either stay here and be caught like rats in a trap, or make a sortie against the enemy. We haven’t much of a chance, but personally I don’t want to be caught here.” (FLUK), quoting a secondary source. (102) The need for rewriting here cannot be disputed, if the song was to be published at all. (FLUK), referring to the text of a folksong. (103) Here the heavy machinery used in, say, the construction of a road, is damaged so as to prevent it being used again and to escalate the costs. (FLUK), referring to ‘acts of eco-sabotage’. Though they contain a lexical item often found within text comments, none of these extracts actually contains a text comment. The formal diversity of text comments, and the multifunctionality of the linguistic items contained within them, therefore makes automatic identification extremely difficult. 80 It must be noted, however, that the manual analysis was also difficult, since it was extremely time-consuming. This tension between accuracy and representativity on the one hand, and practicability on the other, is one of the fundamental dilemmas in studying discourse phenomena using corpora. One either works manually, in which case the phenomenon will be captured in all its diversity, including those linguistic realisations the researcher might not have expected, or one works automatically, which is quicker, but returns only those forms the researcher already knows exist and therefore actively looks for. 81 The other problem with automated identification is that, even if researchers were to have an exhaustive list of features to search for, they 80 These difficulties of course apply to the automatic identification of metadiscourse as a whole, not just to the study of text comments. 81 The tension between these two extremes is clear in recent work on metadiscourse. Hyland, for example, waxes lyrical about the context-dependent nature of metadiscourse, observing that it cannot be identified on the basis of linguistic form alone, but then studies the phenomenon only as reflected in 300 automatically-identified “commonly used items” (2000: 110-1). <?page no="201"?> 201 would still have to cope with the fact that these items do not always fulfil a metadiscoursal function (see the ‘here’ examples above, extracts 14 - 16). Many of the returned hits will therefore be of no relevance to the phenomenon under investigation, and must be excluded. Here at the very latest manual analysis becomes indispensable, since no corpus concordancing software can identify discourse functions. A recent study of metadiscourse, which applies automated analysis methods, approaches this problem by sampling 600 returns for each item and then calculating a percentage of metadiscoursal uses (Hyland & Tse 2004). Here, the problem mentioned above returns, in that sampling is less work than manual reanalysis of all hits, but correspondingly less accurate. How much less accurate this method is will depend on what percentage of returned hits are re-analysed. It should be born in mind, however, that this method effectively entails the sampling of what is already only a sample (the corpus): the representativity and generalisability of the findings are therefore compromised. In the present study, a more accurate and fuller picture of text comments was achieved through manual analysis than could ever have been gained with automated methods, precisely because, when it comes to text comments, the context is extremely important. An examination of decontextualised lines in a concordancer would not have permitted accurate identification (let alone categorisation) of text comments in the SCEGAD corpus. Reading each corpus text three times, always one text at a time, allowed every single text comment to be considered holistically, with a view to its rhetorical function at a particular point in a particular text. 4.9.1 Quantitative analyses of text comments according to language Only two corpus texts, both by German authors, contained no text comments whatsoever. All other texts contained between 1.03 (HIUS5) and 57.45 (PHDE6) text comments per 10,000 words. As noted in other studies, there was a considerable degree of individual variation in the frequency of text comments (Dillon 1981: 60; Williams 1981: 197; Hyland 2005: 191). However, some cross-linguistic tendencies could be discerned nonetheless. <?page no="202"?> 202 Figure 4.5. Text comments: English and German-speaking authors compared (frequency per 10,000 words) ______________________________________________________________________ f English f German Significance (Mann Whitney) ______________________________________________________________________ Overarching TC 3.09 (n=142) 2.35 (n=90) 0.504 Exclusive TC 0.94 (n=50) 1.88 (n=77) 0.085 Introductory TC 2.66 (n=117) 1.74 (n=66) 0.290 Explanatory TC 1.55 (n=77) 4.06 (n=152) 0.0001 Prospective TC 2.49 (n=133) 2.94 (n=135) 0.582 Retrospective TC 4.06 (n=193) 6.20 (n=24) 0.093 Concluding TC 1.55 (n=70) 2.12 (n=85) 0.206 TOTAL 16.16 (n=782) 21.30 (n=849) 0.025 ______________________________________________________________________ The categories of text-comments cited are those explained in Figure 4.1. As can be seen in Figure 4.5, in SCEGAD it is not the case that the German-speaking authors consistently employ fewer text comments than do their English-speaking colleagues, as was suggested in previous studies (Clyne 1987; Hutz 1997; Baumann 1998). However, it is also not the case that there are no differences between the two groups of authors, as was suggested in yet other studies (Trumpp 1998; Fandrych & Graefen 2002). In fact, it is the German academics in SCEGAD who employ more text comments. The difference between the two language groups is significant at the p<0.05 level only for the total frequency, and for the TC type ‘explanatory text comments’ (results highlighted). Two other subtypes, ‘exclusive text comments’ and ‘retrospective text comments’, approached but did not reach the p<0.05 level. As such, the results for the total counts are largely due to the use of the explanatory TC type. For this type, the results are striking. In general, a value of p<0.001 is considered highly significant: the results for explanatory text comments are ten times less likely than this to be due to chance (i.e. there is a one in 10,000 chance that the results are due to random variation, rather than being evidence of actual differences between the two groups). This surprising trend begs further investigation in another sample of academic discourse. The trend for explanatory text comments is so conspicuous that it aroused my suspicion: I therefore re-examined all 229 corpus examples of explanatory TCs several times, to ensure that the difference in frequency was not due to differences in the language system erroneously coded as discourse preference differences. One hypothesis tested was that, since ‘sollen’ [should] can also fulfil a future-time reference function in German that it does not in English, perhaps some instances of ‘sollen’ had been erroneously coded as explanatory TCs (obligatory meaning) when they <?page no="203"?> 203 were really overarching TCs or even prospective TCs (future-time reference). This was, however, not the case: ‘sollen’ [should] constructions formed only a small percentage of explanatory TC examples (5.19% English language and 12.5% German), and there were many cases of such constructions also in the overarching TC and prospective TC categories. It became clear that the analysis had indeed been motivated consistently by functional-pragmatic considerations, not lexical ones. The difference between the two language groups is therefore discoursal in nature: the German scholars in the corpus explain and justify their speech actions, or express them as obligatory duties, far more often than their Englishspeaking colleagues. The implications of this finding for the author-reader relationship are discussed in section 4.10 below. The prototypical profiles in Figure 4.6, and the calculations provided in Figure 4.7, show that the comparative degree of difference between the two language groups is much greater for the explanatory and retrospective TC types than for the other types. Overarching and concluding text comments, which are both general comments on the structure of the text as a whole, are used by the two groups with very similar frequency. Both groups use retrospective text comments the most frequently of all types: in relatively long and complex texts such as research articles, reminding the reader of what has already been said or done appears to be a very useful way of leading the reader through the text and strengthening knowledge claims. Of all text comment types, the English-speaking authors in SCEGAD are least likely to explain what their text will not do (exclusive TC), and this strategy was also relatively uncommon in the German RAs. The German academics, on the other hand, are least likely to comment on the speech action they are currently carrying out, or about to undertake (introductory TC) - this is, however, the third most popular type among the English-speaking scholars. <?page no="204"?> 204 Figure 4.6. Text comments: prototypical profiles for English and German-speaking authors compared (frequency per 10,000 words) 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 OT XT IT ET PT RT CT English German Figure 4.7. Text comments: comparative difference in frequency for English and German-speaking authors (frequency per 10,000 words) ______________________________________________________________________ f English f German Difference in f ______________________________________________________________________ Overarching TC 3.09 2.35 0.74 Exclusive TC 0.94 1.88 0.94 Introductory TC 2.66 1.74 0.92 Explanatory TC 1.55 4.06 2.51 Prospective TC 2.49 2.94 0.45 Retrospective TC 4.06 6.20 2.14 Concluding TC 1.55 2.12 0.57 TOTAL 16.16 21.30 5.14 ______________________________________________________________________ The very different results across types suggest that it is important for crosscultural studies to distinguish between different discourse purposes in text comments, rather than lumping all sorts of metacommunicative utterances together in one category (as in Clyne 1987; Hutz 1997; Trumpp 1998). For Rossiter, metadiscourse requires the “willingness and ability to comment about […] interaction that is in process” (1974: 42). While the ability is certainly present in both languages, it appears that the willingness varies greatly across types. It was also investigated whether this willingness differed across the two varieties of English represented in the SCEGAD corpus. <?page no="205"?> 205 4.9.2 Quantitative analyses of text comments according to variety The intervarietal comparison between British and US-American authors, which can be seen in Figure 4.8, reveals no significant differences for any type of text comment. It therefore appears that the use of text comments is not marked for the two major national varieties of English as represented in the SCEGAD corpus. The only type of text comment for which there is anything even near a significant difference is the overarching type: here, there is an approximately nine per cent chance that the difference between the British and US-American authors is due to random variation. This chance is too high for the null hypothesis to be discounted; there are therefore no significant intervarietal differences in text comment frequency as reflected in the corpus texts. Figure 4.8. Text comments: British and US-American authors compared (frequency per 10, 000 words) ______________________________________________________________________ f British academics f US-American academics Significance (Mann Whitney) ______________________________________________________________________ Overarching TC 3.78 (n=88) 2.39 (n=54) 0.092 Exclusive TC 1.03 (n=30) 0.85 (n=20) 0.732 Introductory TC 3.36 (n=73) 1.96 (n=44) 0.314 Explanatory TC 1.33 (n=34) 1.58 (n=43) 0.992 Prospective TC 2.99 (n=91) 2.00 (n=42) 0.552 Retrospective TC 4.13 (n=101) 3.98 (n=92) 0.471 Concluding TC 1.62 (n=39) 1.48 (n=31) 0.879 TOTAL 18.08 (n=456) 14.24 (n=326) 0.764 ______________________________________________________________________ 4.9.3 Quantitative analyses of text comments according to culture The intercultural comparison between British, US-American and Germanauthored texts, seen in Figure 4.9, presents a similar picture to the crosslinguistic analyses offered in section 4.9.1 above. Since there were no significant inter-varietal differences among the Anglophone scholars, this similarity is of course to be expected. For all three cultural groups, retrospective TCs are the most common text comment type; the relative dominance of explanatory TCs in the German subcorpus is also apparent: while the difference between the two English-speaking cultural groups was not significant, the differences between the German scholars and both the <?page no="206"?> 206 British and US-American scholars respectively are highly significant for this type. As can be seen in the histogram in Figure 4.9, the trend for the German-speaking academics to employ exclusive text comments relatively frequently holds across cultures: the tendency to state not only what one shall do, but what one shall not do, appears to distinguish the German scholars in SCEGAD from the other two groups. Interestingly, the graph also shows that the US-American authors share their German colleagues’ relative dislike of overarching text comments: only the British authors are particularly keen on frequent statements of their goals and structure over the whole text. The British academics are also unusual in their relative liking for introductory text comments, which function to announce actions as they occur, or announce them shortly beforehand. It seems, therefore, that the British authors place a comparatively strong emphasis on retaining constant contact with the reader. In terms of frequency, the only significant difference between the three groups was for the explanatory TC type, and here only between the German authors and both other groups (p<0.001 in both cases, DE/ UK and DE/ USA). Figure 4.9. Text comments: British, US-American and German authors compared: bar graph (frequency per 10, 000 words) 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 OT XT IT ET PT RT CT British US-American German <?page no="207"?> 207 Figure 4.10. Text comments: British, US-American and German-speaking authors compared (frequency per 10, 000 words) ______________________________________________________________________ f British academics f US-American academics f German academics ______________________________________________________________________ Overarching TC 3.78 2.39* 2.35 Exclusive TC 1.03 0.85 1.88 Introductory TC 3.36 1.96 1.74 Explanatory TC 1.33 1.58 4.06 Prospective TC 2.99 2.00 2.94 Retrospective TC 4.13 3.98 6.20 Concluding TC 1.62 1.48 2.12 TOTAL 18.08 14.24 21.30 ______________________________________________________________________ * The two most similar values are highlighted We see therefore that, according to the SCEGAD results, the differences between linguistic and cultural groupings are nowhere near as great as proposed by previous studies. Though there are differences in the relative frequency of different types, the only consistently significant difference between all linguistic and cultural groups lies in the higher frequency of explanatory text comments in the German subcorpus. The discourse implications of this finding are discussed in section 4.10. 4.9.4 Quantitative analyses of text comments according to discipline We move on now to the cross-disciplinary comparison. Several studies have reported differences across disciplines in the frequency of metadiscourse (Dahl 2004; Hyland & Tse 2004; Peterlin 2005). However, none has previously examined the possibility of cross-disciplinary variation within the humanities. Figure 4.11 provides a comparison of text comment frequency across the philosophy, history, folklore, literary studies and linguistics RAs represented in the corpus. The results for the British and US-American scholars are represented in a single column, since there are no statistically significant differences between the two English-speaking groups (see section 4.9.2 above). The disciplinary ranking is almost identical to that given in the previous chapter for person reference (section 3.4.4). In other words, both phenomena examined in this study, as devices for constructing social interaction in academic prose, reflect the same discipline-specific trends. The differences between disciplines are not consistent for any one TC type. For example, while there are no statistically significant differences in the frequency of any TC type at all between folklore and literary studies texts, the differences between history and <?page no="208"?> 208 linguistics RAs are significant for all TC types except explanatory text comments. Interestingly, the one TC type where the intercultural differences were significant, explanatory TCs, is also the only TC type where there are no significant interdisciplinary differences: here at least it can be said that the influence of culture clearly overrides that of discipline. The overall TC frequencies were significantly different only between the linguistics and folklore/ literary studies/ history, and philosophy and history texts. Figure 4.11. Text comments by discipline: bar graph (frequency per 10, 000 words) 0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 LING PHIL FOLK LIT HIST English Texts German texts Figure 4.12. Text comments by discipline: comparative difference in frequency (frequency per 10, 000 words) ______________________________________________________________________ f Englishf German- Difference speakers speakers in ______________________________________________________________________ Linguistics / Sprachwissenschaften 31.77 27.29 4.48 Philosophy / Philosophie 25.36 20.39 4.97 Folklore / Volkskunde 11.73 17.34 5.61 Literary Studies / Literaturwissenschaften 7.15 25.82 18.67 History / Geschichte 4.80 15.66 10.86 AVERAGE 16.16 21.30 5.14 ______________________________________________________________________ As noted above, the average difference in frequency between the two language groups represented in the corpus is slight overall. This holds true for the linguistics, philosophy and folklore articles; in history and particu- <?page no="209"?> 209 larly in literary studies, however, the comparative difference in frequency across languages is relatively large. In literary studies, for example, the English-speaking scholars employ nearly 2.5 times fewer (p<0.0001) text comments than their German-speaking colleagues. While the Englishspeaking literary studies scholars and historians employed all TC types less frequently than their German colleagues, the differences in frequency are particularly marked for concluding and exclusive text comments. In literary studies, the English scholars employed over 30 times fewer (p<0.005) concluding text comments, and in the history texts nearly 20 times fewer (p<0.0001). It is not that the frequencies of this TC type were particularly high in the German texts (relative to the German subcorpus average), but that they were particularly low in the English-language ones. It is important to note that these results are not due to the influence of an intervening variable such as text length. This possibility was investigated since it seemed plausible that an author would feel less of a need to summarise his or her findings in a relatively short text as opposed to a relatively long one. However, the English-language history and literary studies texts are in fact, on average, longer than those in the Germanlanguage subcorpus. In any case, text-length did not emerge as an influential factor on concluding TC use, nor indeed on text comment use as a whole (see the statistical analysis in section 4.9.7 below). In addition, the English-speaking scholars also employed over ten times fewer exclusive text comments than the German academics in both disciplines (13 times less in literary studies; 11 times less in history). As with the concluding TCs, it is not that the frequencies for the exclusive TC type were particularly high in the German texts (measured by distance from the German average), but that they were particularly low in the English-language ones. Though the trend is less extreme, the same is true of all other TC types in the English-language literary studies and history texts: the frequencies in these two disciplines are well below that in the German-language texts in the same disciplines. It seems, therefore, that the British and US-American historians and literary scholars feel less need to comment on their texts than do their German colleagues in the same disciplines. In particular, the Englishspeaking historians and literary scholars are far less inclined to give a resumé of their findings (concluding TC), or explain what they will not be doing in their texts (exclusive TC). When compared to their Englishspeaking compatriots in the other three disciplines, it emerges that the English historians and literary scholars also employ far fewer overall, introductory and prospective text comments. The English-speaking academics in these two disciplines are therefore generally restrained in their use of text comments. It is not possible here to discern the reasons <?page no="210"?> 210 behind these differences, and any suggestions can be only at the level of conjecture. Nonetheless, it is possible that the British and US-American literary scholars and historians in SCEGAD use fewer TCs because they expect a more active reader than both their German colleagues in the same disciplines, and than their English-speaking colleagues in the other three disciplines. Perhaps British and US-American historians and literary scholars imagine a reader who does not particularly need to have the continual contact represented by introductory TCs, and who is more willing or able to draw conclusions without having them explicitly commented upon by the author. In some ways, perhaps, there may also be the expectation of a less critical reader, hence the lower frequency of exclusive TCs, which explain what the author cannot or will not aim to achieve in the article. However, there could be any number of other motivations behind these disciplinary differences. The other difficulty with conjecture such as this is that it assumes that the same textual action (or lack of it) will yield a universal, predictable response in the reader, and that there is a universal, predictable motivation on the part of the writer. However, this cannot be assumed, particularly in a cross-cultural study. Ultimately, all that may be recorded reliably is the observed results: the way in which writers project readers into their texts differs across disciplines. It is perhaps worth remembering that, as we saw in the previous chapter, precisely these two disciplines (across both languages) were also relatively conservative in their use of person reference. The corpus results suggest that the historians and literary scholars, in particular the English-speakers, favour a less involved, or rather differently-involved, style of presentation, where they interact less overtly and directly with the reader: this entails fewer comments on the text and also fewer explicit references to themselves and their audience. 82 The lower than average frequency of text comments in the history texts appears to constitute a relatively stable cross-linguistic feature of this discipline. Measured across the entire corpus, the total TC frequency is a quarter (English) or half (German) the average frequency in the respective language group. Across the whole corpus, however, the total TC frequency is significantly lower in the history subcorpus only when compared to the linguistics (p<0.0001) and philosophy (p<0.05) texts, but not when compared to the folklore and literary studies RAs. For the English-language subcorpus, TC frequency was significantly lower in history than in all disciplines studied apart from literary studies (p<0.005 for history when compared to folklore, and p<0.0001 for history when compared to the 82 It is true that the German literary scholars constitute an exception to this posited tendency, in that they employ more TCs than the average for their language group. However, the trend does hold for person reference. <?page no="211"?> 211 philosophy and linguistics texts). In the German subcorpus, however, the frequency of TCs in history was significantly lower only when compared to the folklore texts. The difference between the historians and their colleagues in other disciplines is therefore not as consistently marked in the German-language texts as it is in the English-language ones. In general, this trend fits well with the results for person reference, where frequencies in the history texts were also well below average. The English-language texts are particularly marked for this discipline-specific feature: the British and US-American historians used the fewest text comments and least person reference of all disciplines sampled in the SCEGAD corpus. The reasons for this trend are, I would suggest, a sort of general discoursal reluctance among the historians in SCEGAD, particularly the Englishspeaking historians, to interact directly and explicitly with the reader. Possibly the historians represented in SCEGAD imagine a reader who prefers a more disconnected, less personally invasive, style of presentation, but this of course is one possible interpretation only. And as historians well know, no one - not even historians themselves - is capable of describing (let alone analysing and interpreting) historical events neutrally, independent of whether they explicitly comment on this description or not. In this sense, even if the historians in the corpus were trying to present an objective, neutral persona, the nature of knowledge-making in their discipline precludes actual neutrality and objectivity, since it is based on personal interpretation and perception. The historians’ behaviour is in marked contrast to that of the philosophers in both language groups, who consistently evince very high frequencies of both these interactive features, text comments and person reference. One possible effect of such usage in the philosophy texts, at least as judged from my own admittedly biased (British/ Australian) perspective, is that of an author and reader engaged in a cooperative enterprise. The SCEGAD philosophers provide their readers with more information about their personality - their aims, wishes and opinions - and encourage the imagined audience to participate, through text comments, in the co-construction of text: OT (104) My aim is to shore up the…argument…by…exposing flaws in some cases…I shall start with two arguments that aim to show that…Next, I shall turn to the…Then, after discussing the idea that…I shall rebut an…argument that…At the end, I shall turn back to the question of…(PHUS) (105) Es ist das Ziel meiner Untersuchung, die Heterogenität des…zu mildern…[The aim of my study is to reduce the heterogeneity of…] (PHDE) <?page no="212"?> 212 (106) I feel, however, that a detailed consideration of the issues involved may be of interest for at least three reasons. First,…Secondly,…And thirdly…(PHUK) (107) Ich möchte mich im folgenden mit einem Problem aus dem Contrat social auseinandersetzen,…[In the following I wish to examine a problem in the contrat social…] (PHDE) In exclusive TCs, the philosophy authors sampled in the corpus explicitly voice their assessment of their and their reader’s shared needs; they assure the reader that they will not lead them astray or take them on any argumentative detours (author and reader presumably want to reach a common goal on a tight timetable), or detain them with needless technicalities (apparently author and reader are both in a hurry): XT (108) There are interesting philosophical questions about…; but it would take us too far afield to consider them here. (PHUS) (109) The issues here are far too technical for the scope of the present work. (PHUK) The British, US-American and German philosophers represented in the corpus also use exclusive TCs to solicit the audience’s understanding of limitations that force them to exclude particular topics from the discussion: (110) Eine experimentelle Überprüfung der…Hypothese würde freilich den Rahmen eines Aufsatzes übersteigen. [An experimental examination of the…hypothesis would of course go beyond the scope of this article.] (PHDE) (111) Wie das…Argument…ausgearbeitet werden kann,…kann hier nicht dargestellt, aber doch knapp angedeutet werden: …[The elaboration of the…argument,…cannot be described here but only briefly touched upon…] (PHDE) In introductory TCs, on the other hand, the SCEGAD philosophers sometimes ask for reader permission to include specific topics in the discussion: IT (112) Finally, let me give the briefest outline - no more than a hint - of the evidence for…(PHUS) <?page no="213"?> 213 In the corpus texts, academic writers in this discipline also use introductory text comments to conjure a picture of author and reader together turning their attention from one topic to the text, thereby treating the reader as a sort of textual travel companion: IT (113) Let us now turn to the issue of how…(LGUK) (114) Wenden wir uns nun den Darlegungen der…zu. [Let us now turn to the analysis of…] (PHDE) The philosophers represented in SCEGAD use explanatory text comments to cite their very personal motivation for particular textual decisions: ET (115) I can best give my reasons for this belief if I first dispel the paradox which X claims to find in my criterion for applying…(PHUK) (116) Der folgende Abschnitt wurde in der Absicht geschrieben, den Leser zu einer…zu bewegen. [The following section was written with the intention of moving the reader to…] (PHDE) Even something as routine as anaphoric and cataphoric reference receives a distinctly personal, interactive touch in the philosophy corpus texts: PT (117) Later we shall argue that the possibility of…ought to be taken seriously as…(PHUK) (118) As I shall show at the end of my final section…(PHUS) (119) Auf die Frage, um welche Art…es sich dabei handelt, werde ich später noch eingehen. [I will go into the question of what kind of…it is concerned with later on…] (PHDE) RT (120) What I trust we have seen so far is that…(PHUK) (121) In light of our writing of…and the argument of the last section,…(PHUK) (122) Ich habe schon auf die Stelle hingewiesen, an der…[I have already indicated the part in which…] (PHDE) <?page no="214"?> 214 Concluding text comments too are used to emphasise the co-operative and/ or personalised nature of knowledge construction: CT (123) I think we must conclude, therefore, that…is false. (PHUK) (124) In any case, however these special examples turn out, I should sum up like this: …(PHUS) (125) In der Frage, ob…können wir also ein skeptisches Resümee ziehen. [Regarding the question of whether…we can therefore give a sceptical summary.] (PHDE) The combination of text comments with person reference, which can be seen in examples (36) - (38) above, is typical of the philosophy texts in SCEGAD, and strengthens the personal, interactive mood of the texts. As far as possible rhetorical effects go, I would venture to say that the two strategies reinforce each other, creating an author-reader relationship marked by collegial dialogue, communal discovery and the co-construction of meaning. On the other hand, of course, the very same two strategies could also be perceived as off-putting or even condescending. Such is the nature of relationship creation, and of positing possible effects of linguistic behaviours. In any case, the writer/ reader relationship created through extensive use of text comments sets philosophy apart from the other disciplines represented in the corpus. The smallest comparative difference across languages is to be found in linguistics, philosophy and folklore, with the linguistics scholars the most uniform across language groups in their use of text comments. While the linguists and philosophers employ text comments similarly frequently, the folklorists employ these devices similarly infrequently across language groups. The primary reason for the similarity across languages in the linguistics texts lies in the high frequency of references to coming or past material (prospective and retrospective TC) in this discipline. Very often, the text comments refer to other parts of the argument and/ or sections of text, but also to case studies, examples and other data. It is possible that the empirical nature of most of these RAs encourages a relatively elaborate textual macro-structure, with many levels of numbered subsections, which the authors frequently announce and refer back to. However, the highly subdivided textual structure might also be a reflection of editorial preference. In any case, in the SCEGAD linguistics texts, text comments provide a way of explicitly indicating the order of relatively large numbers of (sub)sections and items of supporting evidence. Some illustrative examples (four prospective and four retrospective): <?page no="215"?> 215 PT (126) The semantic underpinnings of this contextual effect are explored in section 4.2. (LGUK) (127) Die semantischen Bezugsgrößen werden dann im Abschnitt 4 genauer untersucht. [The semantic bases will then be described in more detail in section 4] (LGDE) (128) In particular, each poet has one especially striking intonation pattern that I will discuss separately in more detail. (LGUS) (129) Beide Varianten der Kontur werden im Folgenden genauer beschrieben und mit Beispielen belegt. [In the following, both versions of the contour will be described in more detail and supported by examples] (LGDE) RT (130) It is clear from the above data that while…(LGUS) (131) Und tatsächlich haben - wie die Daten verdeutlichen - Interagierende (in der Regel) keine Probleme…[And in point of fact - as the data show - interactants usually have no problem…] (LGDE) (132) (It is also what occurs in those varieties of primary English which otherwise show the broken pattern described above.) (LGUK) (133) … im Vergleich mit zuvor beschriebenen ähnlichen Konturen des Berlinischen…[compared to the similar patterns in Berlin dialect described above…] (LGDE) The comparatively high frequency of such endophoric markers in linguistics texts was also noted by Hyland, who recorded average frequencies of 37/ 10,000 words in the applied linguistics subsection of his corpus, but on average only 8/ 10,000 words in the sociology and 3/ 10,000 words in philosophy subcorpora (2000: 114). We are therefore dealing, it seems, with a cross-linguistic characteristic of linguistic RAs: the authors frequently connect new information with old, and preview information to be provided later in the text. This strategy of constantly explaining textual organisation to the reader, both prospectively and retrospectively, is most likely intended as an aid to reader comprehension, and as a clarification of the methodology and textual structure employed by the author (see discussion in section 4.10 below). Two disciplines, linguistics and philosophy, oppose the general trend for the German academics to employ more text comments than their English-speaking colleagues. In both language groups, the linguists and philosophers employed more introductory, concluding, prospective and <?page no="216"?> 216 retrospective text comments than average for their linguistic group. They were therefore more likely to comment on the overall structure and aims of their article, more likely to announce an immediately-following speech action, and they more frequently indicated the relationship between different parts of their text. In addition, the English-speaking linguists also employed overarching and exclusive TCs more frequently than the English-language subcorpus average. The general picture is that the linguistics and philosophy RAs are more explicitly interactive, across both language groups, than those in the other three disciplines: the trend is simply quantitatively more pronounced in the English-language articles (see also chapter three, section 4.4). Thus, the general trend for the German academics to employ more text comments is reversed in these two disciplines. Some examples from the linguistics subcorpus will serve to demonstrate the markedly interactive nature of text comments in this discipline (for philosophy examples see extracts 17 - 38 above): OT (134) Working from this assumption that part of any effective solution is educating future educators, I have chosen to focus this article on the question of… (LGUS) (135) Der Aufsatz ist wie folgt gegliedert: Nach einer kurzen Begründung dafür, warum…werde ich die…diskutieren. Ich werde insbesondere nachweisen, daß…Ich wirde…genauer beschreiben und zeigen, daß...Abschließend werde ich…untersuchen und argumentieren, daß…[The article is structured as follows: after a short explanation of why…I shall discuss…In particular, I shall show that…I shall describe…in more detail and show that…Finally I shall examine…and argue that …] (LGDE) XT (136) Although many of the observations which I make about nevertheless also apply to utterance initial still, I shall confine the discussion to…nevertheless. (LGUK) (137) Wir können also…aber auch das sog …. außen vor lassen und brauchen die…nur kurz zu erwähnen. [We can therefore ignore…and also the socalled…and need to mention…only briefly.] (HIDE) IT (138) Before I begin the analysis of the variables themselves, I describe the sociohistorical and linguistic situation of Hyde County. (LGUS) <?page no="217"?> 217 (139) Bevor ich näher auf den Nürnberger Schreibusus eingehe, möchte ich zunächst an einem Text zeigen, was diesen Usus gerade nicht kennzeichnet. [Before I address the Nürnberg writing customs in more detail, I wish to use one text to show first of all precisely what does not characterise these customs.] (LGDE) CT (140) In this paper I have offered an account of the… (LGUK) (141) Fassen wir abschließend die wichtigsten Gedanken zusammen, die teils in…bereits formuliert wurden, teils sich aus der Vorstellung von …ergeben. [In conclusion, let us summarise the most important thoughts, which have already been formulated partly in…and partly arose from the presentation of…] (LGDE) As in the philosophy examples given earlier (examples 17 - 38), I would suggest that in the linguistics subcorpus too the SCEGAD authors use text comments to emphasise both their personal role and encourage the reader’s role in the knowledge construction process. According to this interpretation, text comments therefore figure in two related rhetorical strategies: the creation of reader involvement and indication of author involvement. The linguistics texts in the corpus could therefore be described as at once more cooperative in spirit, and also more personal. What these strategies have in common is their focus on meaning construction as a dialogic enterprise: text comments are a way for academic authors, particularly in linguistics and philosophy, to structure their texts as a series of explicitly interactive stages. In sum, much the same comments apply to all of the observed trends: there are clear discipline-specific tendencies - more TCs in philosophy and linguistics, fewer in history, folklore and literary studies - which are consistently more marked in the English-language than in the Germanlanguage texts. When it comes to text comments, the German scholars behave in a more uniform manner across disciplines than do their Englishspeaking colleagues. It is possible that the German scholars represented in SCEGAD assume a reader with preferences that are less strongly demarcated along disciplinary lines, or perhaps the English-speaking authors in the corpus feel more free to cater to presumed discipline-specific audience expectations. Again, it is not possible to decide which of these two possible explanations is more accurate, or if there is another explanation entirely: this is the difference between interpretation and corpus data. For the English-speaking scholars, discipline seems to be a greater influence on text comment frequency than culture: frequencies vary more widely across disciplines, reflecting discipline-specific trends more strongly than in the German texts. For the German scholars, cultural tendencies appear to be <?page no="218"?> 218 more decisive than disciplinary affiliation: the frequencies across disciplines are more uniform and these scholars adjust this aspect of their writing less strongly to audiences from different disciplines. This does not necessarily mean that the German scholars in SCEGAD make fewer allowances for their readers: it is entirely possible that their target audiences simply do not differ as much along disciplinary lines. Overall, it is not necessarily the case that authors in one academic field make any less effort than do those in another to communicate with and help their readers through the use of text comments. Rather, I would suggest that authors from different disciplines perhaps have differing types of imagined readers in mind, and strive to respond to their anticipated needs appropriately. The philosophers and linguists in SCEGAD employ text comments in order to emphasise the dialogic nature of knowledge construction in their discipline; the linguists also use text comments to help the reader understand a methodology and textual structure that is often more complex than that in other disciplines. The historians and folklorists in the corpus, on the other hand, seem to prefer a less explicitly interactive style, since they draw attention to the textual organisation, and their role in it, relatively rarely in both language groups. These differences are indicative of strategic purposes that vary across disciplines; across languages, the SCEGAD scholars pursue rhetorical purposes specific to their academic field. This is of interest in light of Hyland’s comment that “writers and readers make connections, through texts, to their disciplinary cultures” (Hyland 2005: 190). In addition, there are certain tendencies present particularly in one language group: the English-speaking literary scholars are the least inclined to use text comments, whereas the German literary scholars employ even slightly more text comments than average for their language group. Text comments therefore play an important role in each of the SCEGAD subcorpora; like Hyland, however, I conclude that this role is realised to different degrees and in different ways across disciplines (2000: 115). 4.9.6 Quantitative analyses of text comments according to text position As noted above, Clyne found that the English-speaking scholars he studied were more likely than their German counterparts to place advance organisers at or near the beginning of their papers (1987: 229). The German academics were found to give advance organisers more often later in the text, though Clyne does not specify any precise location. It seemed worthwhile to test this finding on the data in the SCEGAD corpus. Clyne’s advance organisers correspond most closely with overarching text <?page no="219"?> 219 comments in the present study: his findings will therefore be contrasted with those for this type of text comment. In the SCEGAD texts, overarching text comments in fact appeared in all manner of locations, from the abstract to footnotes. Abstracts are a matter of editorial policy and therefore cannot be interpreted as indicating cultural or disciplinary preferences - only 7 of the 20 journals sampled in SCEGAD contained any. All other text locations can, however, be examined with a view to testing the tendencies discerned by Clyne in the placement of this TC type. The analysis reveals that there are indeed statistically significant differences across linguistic and cultural groups in overarching TC position. The analysis in SCEGAD was more fine-grained than in Clyne’s study. In all, five text positions were distinguished: (1) in the abstract, (2) at the very beginning of the body of the text. (3) in the middle or at the end of the first section, (4) in the second section and (5) at an even later location or in a footnote. The results show that the German scholars are more likely to place overarching text comments at the end or middle rather than the very beginning of the first section. However, so are the British scholars. Only in the US- American texts is it the case that overarching TCs are indeed more closely tied to the very beginning of the first text segment. Here is evidence yet again that the concept of ‘Anglo-American’ discourse norms is a fallacy (see discussion in section five of chapter two). Figure 4.13. Overarching text comment placement in SCEGAD: cultural comparison (frequency per 10, 000 words) ______________________________________________________________________ British academics US-American academics German academics ______________________________________________________________________ Abstract 1.45 (n=37) 0.28 (n=7) 0.16 (n=7) Very beginning 0.85 (n=20) 1.39 (n=30) 0.86 (n=31) End / mid. sec. 1 1.18 (n=23) 0.21 (n=6) 1.10 (n=42) Section 2 0.22 (n=6) 0.50 (n=10) 0.07 (n=3) Later / footnote 0.08 (n=2) 0.02 (n=1) 0.17 (n=6) ______________________________________________________________________ * Positions where significant differences between linguistic and/ or cultural groups exist highlighted When tested for significance across languages (Mann Whitney bivariate test), the difference between the Englishand German-speaking academics is significant for three text positions, one of which is placement in the abstract (p<0.0001). This is a prime example of an extremely significant tendency that tells us nothing, other than the fact, as noted above, that many of the German journals contained no abstracts. What is more interesting, however, is that the differences between the Englishand Germanspeaking scholars for the text positions ‘very beginning’ and ‘end / middle <?page no="220"?> 220 section one’ were also significant to the p<0.05 level (asymptotic twotailed significance values 0.038 and 0.009 respectively). Here we do have evidence of salient contrasts: while the Anglophone scholars are more likely to place overarching TCs at the very beginning of their texts, the German academics are more likely to place them in the middle or at the end of the first section. When compared across cultures, the differences between the two groups of Anglophone scholars are not significant for any text position other than in the abstract: despite appearances, there are therefore no salient intervarietal differences. As can be seen in the table, the differences between language groups are due solely to the US-American authors. The British academics are in fact not more likely than their German colleagues to place overarching TCs at the very beginning of their texts, as Clyne’s study suggested, nor are they less likely to place them in the middle or at the end of the first section. The only salient contrasts, therefore, are between the German and US-American scholars. When these two groups are compared, only the difference in placement in the middle or at the end of the first section is significant, though this is highly significant. We can conclude, therefore, that the German scholars are far more likely than are the US-Americans to place overarching TCs in the middle or at the end of the first section (asymptotic two-tailed significance value 0.002). Clyne studied US-American, British and Australian scholars as one group, and then contrasted this mixed group with German scholars. This approach is problematic, because it obscures possible cultural differences within the Anglophone scholars. In SCEGAD, cultural differences can be confirmed only between the US-American and German academics. For the sake of comparison, in Figures 4.14 and 4.15, the SCEGAD results are expressed in exactly the same way as Clyne’s. Such a comparison entails subsuming three of the text positions distinguished above into the category ‘later in text’, as in Clyne’s study. It is unclear whether Clyne counted those instances that occurred in abstracts. However, for the SCEGAD corpus, such examples have been excluded from the tallies in Figure 4.14., since not all corpus texts contained abstracts. It emerges that while the results for the German-speaking scholars are very similar, the Anglophone scholars in Clyne’s study demonstrate a tendency for early placement not evident in the SCEGAD data. However, if the table listed US-American scholars only, the results would be nearly identical to Clyne’s (64% at very beginning, 36% later). The cultural trends in the SCEGAD corpus are therefore not consistent across language groups; they are evidence of culturally specific discourse tendencies, not of differences between languages. It would have been interesting to see the results in Clyne’s corpus for the different groups of <?page no="221"?> 221 English-speaking scholars: perhaps the trends were not consistent across languages here either. Figure 4.14. Advance organiser placement according to Clyne (1987: 229): percentage of all instances (n=unknown) ______________________________________________________________________ Advance organiser placement English-speaking German-speaking academics academics ______________________________________________________________________ At or near text begin 59% 42% Later in text 41% 58% ______________________________________________________________________ Figure 4.15. Overarching text comment placement in SCEGAD: percentage of all instances (n=181) ______________________________________________________________________ OT placement English-speaking German-speaking academics academics ______________________________________________________________________ Very beginning 51.02% (n=50) 37.35% (n=31) Later in text 48.98% (n=48) 62.65% (n=52) ______________________________________________________________________ * SCEGAD results excluding those overarching TC which occurred in abstracts If these tendencies are what Clyne meant by “obscure locations”, then his findings are partially confirmed by those of the present investigation (1987: 229): here, however, the obscure locations are favoured not only by the German but also by the British academics. There is no trend for all English-speaking scholars to place overarching TCs in any particular location. Even if there were, I would not wish to replicate the value judgement implied in characterising one location as normal (that is, unmarked) and the other as obscure. Are these results evidence of writer-responsibility in English-language (here: US-American) writing, and the greater importance assigned to form rather than content in English-speaking countries (here only the USA), as Clyne concludes? I think not. It is a brave researcher who posits universal tendencies based on so little evidence. Perhaps the results show a difference in reader expectations: it may be that the position of overarching text comments is more codified in US- American humanities RAs, whereas the British and German authors enjoy a greater degree of freedom in regard to overarching TC placement. The position of overarching TCs is an intercultural difference that is particularly marked for the US-American scholars represented in SCEGAD. The significantly greater tendency among the US-American authors to place overarching TCs at the beginning of the text might therefore be evidence of <?page no="222"?> 222 US-American readers’ expectation that writers state their intentions at the very beginning of the text. 83 Since overarching text comments are given in different locations across cultures, it is possible that the placement of concluding text comments might also be culturally specific. The concluding text comments in SCEGAD were therefore classified according to text position: either in the final section (often marked ‘conclusion’) or at an earlier location. There was indeed a tendency for the German authors to place concluding TCs outside of the final text segment; however, the results were not significant across any of the linguistic or cultural groups represented in the corpus (Mann Whitney and Kruskal Wallis tests). Cultural background therefore has a significant influence on the placement of one type of TC examined in the study. The position of the remaining six types appears to be determined by other factors; foremost among these are most likely the communicative needs and preferences of the author and imagined reader at any one point in a specific text. 4.9.7 Quantitative analyses of text comments according to other variables As regards the other extra-linguistic features for which the SCEGAD corpus is balanced, it emerged that, with one minor exception, neither the gender, age nor academic status of the author had any significant effect on the frequency with which any type of text comment was employed. The exception is that there was a statistically significant tendency for the female authors to employ more introductory text comments than the males (asymptotic two-tailed significance value 0.028, i.e. significant at the p<0.05 level). The overall frequency for all types of text comments, however, was not significantly different across genders (asymptotic twotailed significance value 0.606). It therefore appears that the women sampled in SCEGAD may place more emphasis on retaining a constant connection with the reader, since they comment significantly more often on their immediate intentions and current actions throughout the course of the text. This relatively minor difference aside, however, it appears that extra-linguistic variables have little influence on the frequency of text comments. It was also not decisive whether a text was written by a single or multiple authors. 83 This is not to deny that reader expectations are at least partly the result of writers’ habits. See chapter 5, section 2, for a discussion of the chicken-and-egg relationship between texts, cultures and expectations. <?page no="223"?> 223 Another extra-linguistic variable whose effect on text comment frequency was tested was text length. Several studies have suggested that the length of a text influences the number of text comments employed. Trumpp found that advance organisers were less more frequent in the (shorter) review subsection of her corpus than in the research article subsection. Only 10% of the reviews examined contained any advance organisers at all; the proportion for RAs is unfortunately not specified (1998: 127, 190). By way of explanation, Trumpp observes that, the longer the text, the more appropriate metadiscourse seems, since it elucidates textual progression and eases text comprehension (1998: 128). Hyland and Tse too believe that the variations in metadiscourse frequency in their study are due in part to text length: the texts in the PhD subcorpus are roughly twice as long as those in the Masters subcorpus, which according to the researchers necessitates a higher frequency of interactive devices to structure what they term “more discursively elaborated” arguments (2004: 171). While these findings sound logical, they could not be confirmed in the SCEGAD corpus. The results of the (Spearman Rho) correlation analysis are shown in Figure 4.16. They show that there was no significant correlation between text-length and text comment frequency for any of the types distinguished in the analysis. 84 In general, the correlation coefficient value is regarded as significant if it explains upwards of 40% of variance: as can be seen in Figure 4.16, in the SCEGAD texts, the length of the article explains at best 19% of the variance (for prospective text comments). This does not necessarily mean that text length has no influence on text comment frequency in general: it is possible that, since the SCEGAD texts were relatively similar in length, the influence of this variable was simply too slight. Figure 4.16. Text comments and text length: Spearman Rho correlation analysis ______________________________________________________________________ OT XT IT ET PT RT CT Total ______________________________________________________________________ Correlation coefficient 0.059 0.095 -0.082 -0.139 0.192 -0.05 -0.053 -0.068 (Spearman Rho) ______________________________________________________________________ 84 A perfect relationship would mean that 100% of the variance in the one variable is explained by variance in the other, i.e. text comment frequency can be entirely explained by text-length. A correlation of +1 would mean that there were a perfect positive linear relationship between the two variables (as one goes up, the other does too), while a correlation of -1 would mean that there were a perfect negative linear relationship between the two variables (as one goes up, the other sinks); a correlation of zero would mean that there were no relationship at all between them. <?page no="224"?> 224 In a further correlation analysis, the relationship between TC types was examined. It emerged that there is a statistically highly significant relationship between the frequency of the text comment types overarching, exclusive, introductory and concluding TC (p<0.01). This means that authors who state what they wish to do and/ or what they are doing at present, are more likely to also state what they do not wish to do, and what they have done. This makes a kind of sense, but should not be interpreted too deeply. There is on the other hand no correlation between the explanatory, prospective and retrospective TC types. In other words, there is no connection between the frequency with which authors state that they must do something, or refer to what they will do and/ or have done; the frequency of explanatory, prospective and retrospective TCs does not help to explain author behaviour when it comes to any other TC type. The fact that the correlation is not significant across all TC types suggests that authors distinguish between different kinds of text comments, using them for specific rhetorical purposes, rather than employing all types with impunity. The final statistical analysis I shall offer in this section is stepwise multiple regression, shown in Figure 4.17. In contrast to the significance tests offered above, which establish what effect (if any) particular independent variables have on TC frequency, stepwise multiple regression examines the effect of all independent variables together, removing them from the model one at a time (backwards regression) until the single variable remains which explains the greatest percentage of variance in the dependent variable (in this case text comment frequency). 85 Stepwise multiple regression uncovers the model able to predict the highest proportion of variance in TC frequency using the smallest number of independent variables. The three models below show the variable, or combination of variables, best able to predict how often an author will employ text comments. 86 Stepwise multiple regression is therefore an assessment of the relative predictive power of the independent variables included in the study. 85 The model is then tested, as was done here, by performing a forwards regression analysis with the variables contained in the backwards regression model. If the regression analysis is valid, the variables in each model should be the same. In this case, they are: the regression analysis is therefore valid. Figure 4.17 shows the results of the second (tested) model, using forwards regression. 86 The figure for R squared is the explained variance. Model three, for example, explains roughly 25% of variance in TC frequency. <?page no="225"?> 225 Figure 4.17. Which variables best predict text comment frequency: stepwise multiple regression model summary a. Predictors: (Constant), Text from linguistics? b. Predictors: (Constant), Text from linguistics? , text from philosophy? c. Predictors: (Constant), Text from linguistics? , text from philosophy? , US-American author? Dependent variable: Text comment frequency (all forms) The results of the analysis are interesting, but not dramatic. They are not dramatic in the sense that the total variability that can be predicted is only 15% (model one) to about 25% (model three). The predictive power of these models is thus not particularly high (as explained above, approximately 40% explained variance and upwards would be considered high); this is not through any fault in the method, but rather an indication of the fact that the extra-linguistic variables controlled for in the SCEGAD corpus are not the primary motivation behind text comment frequency. As mentioned above, I suspect that the primary motivation behind text comment use is rather the rhetorical effect the author wishes to achieve, and this cannot be controlled for in any corpus. The results of the regression analysis are nonetheless interesting, in that they reveal the relative predictive power of the variables. The most influential variables concern the discipline to which the text belongs; culture appears only as the third variable, and here it is decisive whether the author is a US- American or not. This tells us that the US-American authors are the most easily identifiable on the basis of (relatively low) text comment frequency. For the British and German scholars, there are more important influences than their cultural background when it comes to this aspect of language use. These results fit well with the observations made above: relatively few significant differences across linguistic groups and a high degree of crossdisciplinary variation. In sum, therefore, while an author’s cultural background is an important influence on text comment frequency, it is not the most important influence of all. Of the variables for which the SCEGAD corpus is controlled, discipline is the most influential. I close this section with a quote from Crismore: Model R R squared Adjusted R squared Std. error of the estimate Change Statistics R squared change F change df1 df2 Sig. F change 1 .387(a) .150 .141 13.00228 .150 17.254 1 98 .000 2 .462(b) .214 .198 12.56613 .064 7.921 1 97 .006 3 .498(c) .248 .225 12.35163 .034 4.398 1 96 .039 <?page no="226"?> 226 Quantitative information is necessary for identifying the existence of and relative emphasis given to different metadiscourse types in samples. Purely quantitative analysis cannot, however, convey the flavour of the text materials (1989: 163) In this spirit, therefore, we move on to the more qualitative analyses. 4.10 Relating to the reader in academic prose One of the main reasons for examining text comments in this study was the crucial role such devices play in structuring and constructing the relationship between author and imagined reader. So far, however, such issues have been somewhat in the background while the influence of variables such as language, culture and discipline on frequency was considered. In what follows therefore the major emphasis will rest with qualitative analysis of the discourse function and strategy behind text comment use; nonetheless, quantitative analyses will be offered where appropriate, in particular where they reveal salient cultural differences. The discussion will proceed type by type, since, as we saw above, different types of text comments are associated with different rhetorical strategies. I will illustrate some of the different discourse purposes academic authors pursue with text comments, and, based on these observations, offer some suggestions on what different discourse strategies reveal about the relationship between author and imagined reader. The general idea behind these suggestions is that text comments are evidence of the sensitivities which texts stimulate and tap into, and of the resources which mediate readers’ expectations. Schiffrin observes that when writers and speakers use metadiscourse, they project a different part of the self, assuming a persona as an animator, which is the part of the authorial self active in producing utterances. For Schiffrin, metadiscourse is therefore about speakers exercising control (1980: 231). Projections aside, it is absolutely clear that, in a communicative situation such as this, it is the writer who controls the interaction. The corpus analysis reveals that the extent to which authors emphasise or downplay their active role in structuring and managing this unequal process of interaction differs across cultures. Across disciplines, text comments reveal aspects of the preferred level and type of interaction between author and reader, and of the ways in which authors manage social interaction in their texts. From a more general perspective, text comments reveal aspects of the author’s anticipation of the reader, and of the reader’s expectation of the text, including the author’s appropriate role within this text. As argued above, the interplay between these anticipations and expectations is of <?page no="227"?> 227 central importance in the communicative process. Mead (1934) proposed that society consists of structures created through interaction; by analogy, texts too consist of structures that are the agreed result of interaction. We can study the text not as a product but as a process, by examining the social acts, involving individuals, that make up the whole. It is here that text comments fit, since they are evidence of the social acts authors and (indirectly) readers engage in when constructing texts. Text comments are evidence of writing as a textual process that has the co-construction of meaning as its aim. Text comments are therefore of interest from a crosscultural, cross-disciplinary and a more general academic discourse perspective, and will be approached from all of these directions in the following analysis. It must be stressed that the suggestions offered here remain speculative: without asking the SCEGAD authors about their intentions and expectations of their intended readership, one cannot know exactly what is going on. Indeed, even if one were to engage in post-hoc interviews with the interactants, not all authors and readers are self-reflective enough to provide cogent responses about hundreds of relatively minor textual decisions, assuming of course that these decisions are even conscious choices. As with all informed speculation, there is more than one possible interpretation, and I hope to present these openly. 4.10.1 Interaction and overarching text comments Overarching text comments are general comments on the subject matter of the article. They consist of two main moves: a statement of intentions and an explanation of the structure. Not all overarching TCs contain both moves, however, and the order of the moves is variable, though the aims usually precede the explanation of the structure. The two sections answer the anticipated reader questions ‘what…? ’ (theme/ purpose) and ‘how…? ’ (structure). As was mentioned earlier, the structure move is sometimes very long, consisting of multiple sentences detailing the organisation of each segment, section by section. This type of exhaustive shopping-list overarching TC is probably intended for a reader who wants to know exactly what to expect and when. It highlights the methodical progression of the argument in a way likely to appeal to a reader who is similarly methodical. Other readers, however, might regard it as unnecessary or over-long. In longer overarching TCs, some authors explain text structure as a chronological progression of interconnected arguments: <?page no="228"?> 228 (142) My aim is to shore up the…argument…by…exposing flaws in some cases…I shall start with two arguments that aim to show that…Next, I shall turn to the…Then, after discussing the idea that…I shall rebut an…argument that…At the end, I shall turn back to the question of…(PHUS) (143) Nach der…Darstellung von Daten zu…werden kurz die Gemeinsamkeiten und Unterschiede der…besprochen. Anschließend erfolgt ein Erklärungsversuch anhand eines…Ansatzes. [Following the presentation of data on…the similarities and differences of…will be discussed. Afterwards a tentative explanation of…using a…method will be carried out.] (LGDE) However, other authors conceive of their texts as a series of numbered sections or parts: (144) Ich werde im folgenden zunächst die einschlägigen Fakten vorstellen und interpretieren (II), sie dann in den…Rahmen der…einordnen (III); und abschließend die Folgen für…(IV) aufzeigen. [In the following I shall first present the relevant facts and interpret them (II), then place them in the framework of…(III); and finally highlight the implications for…(IV).] (HIDE) (145) Im ersten Abschnitt wird eine knappe Darstellung der…gegeben….Im zweiten Abschnitt, geht es darum, die…zu rekonstruieren und…herauszuarbeiten. Im dritten Abschnitt wird dargestellt, wie…Schließlich wird der…problematisiert, um vor allem am Beispiel der…zu dokumentieren, wie…[In the first section a brief account of…is offered…The second section is concerned with reconstructing and elaborating…In the third section…is accounted for. Finally…is problematised using mainly the example of…in order to document …] (HIDE) The first method emphasises the logical progression of the argument, and is more narrative in structure, while the second draws attention to the logical order of the constituent sections, and is more standardised and repetitive in structure. Depending on what the reader expects, both methods could be equally effective in informing the audience of what awaits them. The way in which authors present their texts varies. Some authors couch their topics and intentions as blank statements of fact: (146) This article is about…and the way...(FLUK) (147) Gegenstand dieser Untersuchung sind…[The subject of this study is…] (LGDE) <?page no="229"?> 229 (148) I show, first, that…and, second, that these differences (LGUK) (149) Ich werde diese Supervenienzthese präzise fassen, sie auf zwei Weisen motivieren und hierbei zeigen, dass…[I shall conceptualise the supervenience thesis precisely, motivate it in two ways, and thereby show that …] (PHDE) Other authors use hedges to characterise their texts as attempts, and talk of trying to achieve particular aims or show particular things: (150) I hope instead to provide some ideas about how…(LGUS) (151) This is what with some trepidation the rest of this article will try to do. (HIUK) (152) Ich werde im folgenden zu zeigen versuchen, daß…[I will try to show in the following that…] (PHDE) (153) In diesem Zusammenhang soll deshalb im Weiteren auf einige wenige Aspekte der Sprachentwicklung in der DDR eingegangen werden. [In this context therefore, in the following a few aspects of language development in the GDR shall be discussed.] (LGDE) The first strategy could be interpreted as more businesslike and perhaps may even appear more objective. The second, because it implicitly raises the possibility of failure, appears more personalised, casting the text as an individual human endeavour whose success will be judged by the reader. Readers who prefer a more visible and human (even modest) author could prefer the second strategy; others might not be interested in ‘attempts’, preferring instead an impersonal statement of aims. Alongside the seemingly matter-of-fact or even modest authors, there are also some (English-speaking) authors who try to ‘sell’ their text, presumably in the expectation of a harried reader looking for an excuse to read something else instead: (154) This article…uses a fresh, more complete and more rigorous analysis of...to show that…(FLUK) (155) I want to take the opportunity for detailed reexamination of Bauman’s foundational text that this anniversary affords to attempt to explain, at long last, how and why performance theory and feminist epistemology are incongruent. (FLUS) The way in which authors refer to their articles probably reveals something of how they see their work, and perhaps a little of their personality, or at least their persona within the text. As these examples demonstrate, it is not <?page no="230"?> 230 easy for academic authors to strike the right tone, especially when writing for an audience whose preferences they can only intuit. 4.10.2 Interaction and exclusive text comments Exclusive text comments are statements of what the article is not about. In analogy to the first TC type, they explain what the article does not or cannot aim to achieve. The motivation behind exclusive TCs is diverse. At the most basic level, exclusive TCs are evidence of the formal constraints academic authors labour under, which sometimes force them to exclude material they would have liked to cover: (156) For reasons of space I will give here only the most striking and significant example, which is a combination of a stative main verb and a stative infinitival verb, as in (97). (LGUK) (157) There are more other variations than space allows here. (LGUS) (158) Aus Platzgründen kann auf zwei weitere Funktionen nur kurz eingegangen werden: …[For reasons of space, two further functions can be addressed only briefly: …] (LGDE) By explicitly referring to the restrictions imposed upon them from outside, authors absolve themselves of responsibility for the exclusion of possibly vital material. Such exclusive TCs are not indicative of cultural or disciplinary differences, but rather allude to a general feature of the RA genre, in which word limits, and editorial interventions, are a matter of course. Where there are repeated exclusive TCs within a single text, the background is often that the article is based on a larger research project such as a PhD, in which case the author has typically had trouble compressing material into the RA form: (159) Owing to limitations of space this debate cannot be examined in any detail here...(FLUK; research based on a PhD thesis) (160) In diesem Aufsatz können aus Platzgründen nur einige ausgewählte Belege vorgestellt werden. Eine umfangreichere Belegsammlung zu diesem Phänomen ist im Internet unter der Adresse…zu finden. [For reasons of space, in this article only a few selected examples can be presented. A more detailed collection of examples of this phenomenon can be found on the internet under the address: …] (LGDE; findings from a much larger research-group project) <?page no="231"?> 231 This kind of exclusive TC is therefore an indication of textual history, namely how the author has cut material in order to fit the required length and format; as such it too is discourse-motivated, and due to the circumstances surrounding the writing, rather than being conditioned by culture or discipline. The declaration that a particular matter cannot be treated at present can lead to, or could perhaps even be the excuse for, self-advertisement: (161) Some of these are also problems for meter and other aspects of poetics, but they are compounded when one looks at intonation. Some are beyond the scope of this article (see AUTHOR NAME 1998)…(LGUS) 87 (162) For present purposes, however, such developments may be sidelined ( AUTHOR NAME , forthcoming). (PHUK) (163) Ich kann hier nur die Grundidee einer solchen Deutung skizzieren und verweise für eine formal ausgearbeitete Analyse auf AUTHOR NAME (2001). [I can only sketch the basic idea of such an interpretation here and refer to AUTHOR NAME (2001) for a formally elaborated analysis.] (LGDE) (164) Für eine ausführliche Fassung dieser Arbeit mit Beispielen zu allen Konturen,…siehe AUTHOR NAME (2002). Hier nur eine kurze Zusammenfassung der wichtigsten Unterschiede zwischen dem Dresdenerischen und dem Berlinischen. [For a more detailed version of this study with examples of all contours,…see AUTHOR NAME (2002). Here is only a short summary of the most important differences between the Dresden and Berlin dialects.] (LGDE) (165) In der vorliegenden Arbeit werden…dargestellt, meine Untersuchung zu…werde ich an anderer Stelle darlegen ( AUTHOR NAME 2001b). [In the present study…will be presented, I will present my study of…at another point ( AUTHOR NAME 2001b)] (LGDE) 88 Self-advertisements can be rather subtle: most of the examples above contain no personal pronouns and only an oblique, even bracketed, reference to the author’s own work. If readers have briefly forgotten whose work they are currently reading, which can of course happen after some pages, they might not even recognise self-promotion as such. The SCEGAD authors cite themselves in exactly the same way as they cite everyone else: this is either modesty or somewhat disingenuous. In any 87 Author names have been removed for privacy reasons. 88 In this case, the self-promotion seems extreme: seven of the 37 works in the references list of this article are by the author. In 13,435 words, there are 17 self-citations. This is roughly five times the average self-citation frequency per 10,000 words Hyland recorded for the applied linguistics RAs in his study (Hyland 2003). <?page no="232"?> 232 case, I hasten to add that such self-citation was not confined to any particular culture or discipline (though see section five of chapter three for a discussion of self-advertising and person reference). Similarly, exclusive TCs can be used to stake a claim for later research, indicating the areas that an author wishes to examine, and publish on, at more length in the future: (166) Defending such an approach is, however, a task for another occasion. (PHUK) (167) Although I must save that enquiry for another occasion, let me say in advance why…is so important. (PHUS) (168) One might also challenge…, since arguably…But I will leave that debate for another day. (FLUS) Such claims can be vital: in a highly competitive environment, academic authors need to find and publicise a niche as quickly and widely as possible in order to ensure continued research success. The effect on the reader is perhaps a little like a film preview: one is persuaded that something is fascinating, though one cannot have it now. Some authors appear to play with this suspense, almost teasing the reader: (169) This is a crucial move on Quine’s part and deserves close examination - but I shall not do that here. (PHUS) (170) Ich lasse dabei…außer Acht - obwohl es auch eine lohnende Aufgabe wäre, zu fragen… [I therefore pay no attention to…though it would be a worthwhile task to ask…] (FLDE) Exclusive text comments also introduce references to the work of other researchers, thereby delineating the present study from those conducted by others: (171) (Whether the accessing asymmetry is modelled in terms of…or in terms of …is not germane to the point in hand; see Leonard 1992, Macken 1992 and Menn & Mattei 1992 for reviews of the competing positions.) (LGUK) (172) Auch kurze Wörter können schwer sein wegen komplexer Silbenränder. Darauf gehen wir hier nicht ein (vgl. aber Werner 1978). [Short words too can be difficult due to complex syllable division. We do not discuss this here (but see Wetner 1978)] (LGDE) There is evidence here of authors’ desire not to replicate previous research, thereby wasting both their time and that of the reader: <?page no="233"?> 233 (173) This practice has been widely criticized (e.g., Cameron and Coates 1984; Nichols 1983, 64-65) and will not be discussed in this paper. (LGUS) (174) Da mittlerweile ein breiter Konsens über…besteht, soll an dieser Stelle darauf verzichtet werden, den…nachzuzeichnen. [Since there is now a broad consensus on…tracing the…shall be abstained from here.] (HIDE) The institutional context of these texts explains this usage: academics in all three countries, and no doubt in many others besides, do not have the luxury of endless time. Such comments also show the authors’ awareness of previous research in their area, and consciousness of where their contribution fits. In this way, they help scholarly authors construct an expert identity as a competent specialist. The cross-cultural analyses above showed that the German authors in SCEGAD employed exclusive TCs more often than both their US- American and British colleagues. These differences were not statistically significant, but it is worth noting that there are other types of TCs which also have an exclusive TC character in German: ET (175) Drei Hinweise müssen hier genügen, um anzudeuten, welche besonderen Kompetenzen und Erfahrungen des Faches dabei nutzbar gemacht werden können: … [Three hints must suffice here to indicate which specialised skills and experience in the discipline could be usefully turned to this purpose: …] (FLDE) OT (176) Ich beschränke mich im folgenden auf die kritische Untersuchung… [In the following I shall restrict myself to a critical examination of…] (PHDE) The choice of words here is most likely not deliberate: whether one expresses intentions in the form of what one shall cover or what one shall restrict oneself to is probably not a matter of conscious deliberation. However, the exclusive TC tone of such statements, and the relative frequency of exclusive TCs in the German texts, does seem to suggest that the German authors in the corpus may anticipate a more critical reader. Exclusive text comments suggest a defensive relationship between author and reader, in which the author attempts to combat anticipated criticisms in advance. Since, as was noted above, many of the metadiscoursal features typical of spoken interaction, which help to ensure comprehension, do not occur in written academic texts, it is all the more important for <?page no="234"?> 234 scholarly authors to try and anticipate these expectations. 89 Indeed, Hyland and Tse perceive such anticipatory assuagement as vital to rhetorical success in academic writing (2004: 170). In addition to those already cited, one particularly clear example of a pre-emptive defence in the SCEGAD corpus is the following: (177) Die…Aspekte des…werden dabei zunächst bewußt außer acht gelassen. [In doing so, aspects of…will at first be deliberately disregarded.] (LTDE) Here one can distinctly hear the anticipated voice of the reader: ‘What? You’ve left out X? ’ and the writer’s reply ‘Yes, but don’t think it wasn’t deliberate! ’ Thompson (2001) points out that negation - and exclusive TCs are about negation - is one of the sites where the reader-in-the-text emerges. Quoting Jordan (1988), Thompson observes that negatives refute a positive proposition that is somehow under consideration, in many cases because the author assumes that the reader believes it (2001: 76). In example (90) above, the positive proposition would be that the author has disregarded aspects of a particular phenomenon by mistake, out of ignorance or carelessness. The author anticipates this proposition, and hastens to correct it. Applied to exclusive TCs in general, the idea that refutations are a reaction to and reflection of implied statements or assumptions means that in stating what they shall not do, the SCEGAD authors reveal their implicit assumptions about what readers expect they should or could do. This offers us a new perspective on exclusive TCs. Let us therefore examine some of the assumptions that academic authors make, as reflected in exclusive TCs. A particularly common feature of exclusive TCs in the German-language texts is a reference to completeness or exhaustiveness of presentation: (178) Es kann hier nicht um eine vollständige Auswertung aller Artikel und Beiträge der Sprachpflege zum Thema gehen. Erfasst wurden lediglich einige Texte, die für…wesentlich erschienen. [The point here cannot be a complete analysis of all articles and contributions on the topic of language cultivation. Only a few texts were covered which seemed relevant to…] (LGDE) 89 I am thinking here of devices such as tag questions, turn-taking signals and communication control signals such as ‘is that clear? ’ or ‘I don’t get you’ (Keller 1979; Bublitz 2001). None of these devices occurs in a formal written genre such as a research article, because they only make sense in synchronous interaction. They are of course possible in formal spoken genres (such as meetings) as well as in informal written language (such as chat-rooms), since these types of interaction are synchronous. It is the synchronicity, that is, not the written/ spoken or formal/ informal nature of interaction, that enables communication control signals. <?page no="235"?> 235 (179) Die…erschöpfend zu behandeln, würden den Rahmen dieses Aufsatzes sprengen. [It would go beyond the scope of this article to treat…exhaustively.] (PHDE) (180) Die Vielfalt der…kann hier natürlich nicht im einzelnen analysiert werden. [The diversity of…of course cannot be analysed here in detail.] (HIDE) (181) Eine minutiöse inhaltliche Auswertung aus…Perspektive kann daher auch nicht das Ziel dieser Betrachtungen sein; … [A meticulous textual analysis from a…perspective therefore cannot be the goal of these observations.] (LTDE) (182) Alle möglichen Grammatikalisierungsaspekte von Präpositionalobjekten zu behandeln, würde den Rahmen dieses Beitrages sprengen. [It would go beyond the scope of this article to discuss all possible grammaticalisation aspects of prepositional objects.] (LTDE) (183) Dabei geht es nicht um enzyklopädische Vollständigkeit, sondern darum,… [The point here is not encyclopaedic exhaustiveness, but…] (LTDE) (184) Natürlich müssen in einer umfassenden Deutung auch andere Fragestellungen und Aspekte berücksichtigt werden, und bei…ist ohnehin ein anderer Zugang erforderlich. [In a comprehensive interpretation other questions and aspects would also have to be considered of course, and for…a different approach is in any case necessary.] (LTDE) ‘Exhaustive’, ‘detailed’, ‘meticulous’, ‘encyclopaedic’ and ‘comprehensive’: this is the ideal these German academics feel they currently cannot attain. Such comments can only be made if there is already the assumption - in the author and/ or imagined reader’s mind - that a text such as this should or could offer an exhaustive picture of the phenomenon at hand. As Dillon points out: All writing is interpreted against a background of beliefs, knowledge, and concerns that the writer believes are accessible to the reader; to imagine an audience involves deciding what you can take for granted. (Dillon 1981: 59) The German scholars in SCEGAD imagine an audience that takes comprehensiveness for granted. It seems therefore that, for German humanities articles, completeness appears to be a major criterion against which work is judged. Measured against this ideal, the authors of these particular texts feel that their work will fall short, and therefore feel obliged to justify their behaviour. <?page no="236"?> 236 In contrast, the English-language texts contain only one comparable instance of such a defence, in which the author cites his inability to provide a complete picture of the phenomenon at hand: (185) A full elaboration of these claims is not possible here, but I will examine the…(LTUS) Therefore, it must be assumed that these criteria are not as important to English-speaking readers as they are to their German colleagues. As noted above, reader expectations, and writing traditions, seem to differ across cultures. A related assumption, also particular to the German-language articles, is that a discussion should be appropriate to its subject matter, that is, lengthy and detailed enough. If complex objects or topics cannot be analysed in a satisfactory manner, many authors apparently feel that there is no point discussing them at all: (186) Der hier gemeinte Begriff der…kann hier nicht zufrieden stellend geklärt werden. [The intended notion of…cannot be explained satisfactorily here.] (PHDE) (187) Das Problem eines etwaigen…soll hier nicht ausdiskutiert werden (und kann beiläufig gar nicht angemessen diskutiert werden)…[The problem of the possible…is not to be discussed here (and incidentally cannot be discussed adequately here anyway)…] (HIDE) One wonders who decides what an adequate or satisfactory presentation is: do these statements represent the author’s opinion, or his/ her assumptions of the audience’s (colleagues, editors, readers of journal X) judgement? Possibly the two cannot be separated. Not only the notions of comprehensiveness and exhaustiveness appear crucial to the German academics in SCEGAD; the concept of a correct, appropriate discussion of specialised topics appears to be central too. What lies in wait for an author who discusses matters without the due length and detail? One corpus text explicitly sets out the dangers of treating an issue superficially: (188) Aber sie so substantiell aufzuzeigen und für die Interpretation von…derart fruchtbar zu machen, wie das ohne die Gefahr plakativer Vereinfachungen unerlässlich ist, erforderte eine eigene Untersuchung. [But to present them so substantially and render them as productive for the interpretation of…as is absolutely necessary to avoid the danger of gross simplification, would require a separate study.] (LTDE) Research is obviously a very complicated endeavour. Critical readers expect articles to treat challenging, demanding topics and concepts in an <?page no="237"?> 237 appropriately complex fashion, that is, in all due length and detail. As a preparatory defence against such readers, German authors go to great lengths to comment in advance on their inability to provide what is apparently expected of them. This necessitates an almost breathless emphasis on the insufficiency of the text: (189) Der komplexe Inhalt kann hier nur stichwortartig in Bezug genommen werden. [The complex content can only be related in note form here.] (HIDE) (190) In der Untersuchung eines…kann nicht der Frage nachgegangen werden, wie…Lediglich eine Überlegung sei hier angeführt. [In the study of…the question of how…cannot be addressed. Let merely one consideration be cited here.] (LTDE) (191) Gewiß kann dieser Aufsatz ein Desiderat vom umrissenen Ausmaß nur sichtbar machen, nicht beheben. Wenn der eine oder andere seiner provisorischen Zugriffe zur Diskussion eines vernachlässigten Themas anregte, wäre vielleicht schon etwas gewonnen. [Certainly this article can only reveal, not solve, a desideratum of this size. If one or other of its tentative suggestions were to stimulate discussion of a neglected topic, then perhaps something could be gained.] (HIDE) Whether these authors really perceive their texts as inadequate, or are simply casting them as such in order to avert criticism, as a sort of facesaving technique, is impossible to discern. If the German academics in the corpus seem more concerned about completeness and detail, Englishspeaking scholars seem more focused on including only strictly relevant material, as a number of the authors stress: (192) This is not a point that we need take up here, however, since it has little bearing on the… (LGUK) (193) …and these other cases are not central here. (PHUK) (194) Whether this position forms the coda portion of a closed rhyme or an onset is a question we can set on one side here. Of more immediate relevance is the fact that… (LGUK) The assumption behind these comments is that the writer should only include, and the reader only wishes to receive, that information which is directly and immediately relevant to the point at hand: (195) For our immediate purposes, we may sidestep this theoretical disagreement, since …(LGUK) <?page no="238"?> 238 (196) For present purposes, however, such developments may be sidelined ( AUTHOR NAME , forthcoming). (PHUK) (197) For the purposes of the present argument, however, it is not necessary to determine whether the negative images… (FLUS) Accordingly, too much information is not a blessing, but a hindrance, since it weighs down the argument with unnecessary baggage: (198) The technicalities need not detain us; … (PHUK) The Anglophone authors in the corpus appear to abide by a rule that additional information, even if important and interesting, must be reserved for another occasion: 90 (199) Important though this debate undoubtedly is within psychology, its resolution, as such, is wholly immaterial to the issue with which we are concerned. (PHUK) (200) There are other important similarities between…and…, but here I am only concerned with…(PHUS) (201) The teaching of standard written conventions is another important topic, beyond the scope of this article. (LGUS) For the Anglophone academics in SCEGAD, the criterion of immediate relevance is related to the concept of ‘scope’. The following are a few examples of the many in the corpus: (202) It is beyond the scope of this article to suggest definite answers to the problem, but given…(LGUS) (203) Some are beyond the scope of this article (see Cooper 1998), but…(LGUS) (204) The issues here are far too technical for the scope of the present work. (PHUK) (205) It is beyond the scope of this article to map the… (FLUS) 90 It would be interesting to examine the relation between the rule of direct relevance implicit in the English-language corpus texts, and the explicit style advice given in British and US-American academic writing guides and style sheets. It may well be that the notion that all irrelevant material must be excluded is not so much an unwritten rule as a stringent norm enforced through prescriptive pedagogy. <?page no="239"?> 239 (206) It is beyond the scope of this article to offer a critique of this position, or of the …However, it must be noted that…(FLUK) For the British and US-American scholars, ‘scope’ seems to be a seminal concept in writing a research article, and presumably a primary criterion according to which readers will judge communicative success. An article which includes irrelevant material will be condemned for waffling. There is no parallel here with the German scholars’ concern that their work will be condemned for excluding additional material. The general picture here is almost exactly opposite to that in the German-language texts. While the English-speaking authors in the corpus comment on the fact that they have excluded ‘extraneous’ material relating to the central topic, even if it is important and interesting, the German academics in SCEGAD remark on the fact that they cannot include all ‘necessary’ material, and therefore must exclude whole topics. In Gricean terms, the Anglophone scholars appear to place more explicit emphasis on the maxim of relation, assuring readers that everything irrelevant has been excluded, and that, conversely, everything excluded was irrelevant. 91 In contrast, their German colleagues seem to value more highly the maxim of quantity, conscientiously informing readers when they cannot make their contribution as informative as might be required (Grice 1975). 92 The English-speaking scholars studied seem to anticipate a reader who dislikes the inclusion of ‘irrelevant’ information, and desires constant reassurance that only truly central matters are being discussed. The German scholars analysed, on the other hand, anticipate a reader who expects all ‘relevant’ information to be included, and desires an explanation if it is not. Whether any of this material is really relevant or not is of course an entirely subjective judgement. The final question that is of interest from a cross-cultural perspective is whether authors give a reason for their behaviour. In the Germanlanguage corpus texts, it is often the case that the authors state simply that something will, can or need not be done, without explaining to the reader why this is so, as in these examples: 91 It is worth mentioning, however, that the English-speaking scholars’ emphasis that the material they leave out is irrelevant might also be an offensive strategy to combat reader criticism by devaluing uncovered ground: this would mean that, while the German scholars in SCEGAD combat criticism defensively, by emphasising the ideals they cannot realise, the British and US-American academics meet criticism more offensively. There would therefore be culturally distinct ways of dealing with uncovered ground. 92 The maxim of quantity relies upon the concept of what is “required” (Grice 1975: 45). Obviously the nature of what is required is defined by the (socio-cultural, discoursal, disciplinary and textual) context: the German scholars’ emphasis on the requirement for information to be complete, detailed and exhaustive appears to be a culturally specific, and possibly genre-specific, interpretation of this concept. <?page no="240"?> 240 (207) …aus verschiedenen Gründen, die uns hier nicht zu interessieren brauchen…[…for various reasons, which need not interest us here…] (PHDE) (208) Mit dieser Möglichkeit der Rechtfertigung werde ich mich in der Folge nicht beschäftigen…[In the following I shall not be concerned with the possibility of this justification.] (PHDE) (209) Jenseits der…sah das bis 1200 durchaus anders aus; das muß hier jedoch nicht interessieren. [For the…on the other hand the situation up to 1200 was completely different; this need not be of interest here however.] (LTDE) (210) An dieser Stelle ist zu betonen, daß wir natürlich nicht das…erfassen können und wollen. [At this point it must be emphasised that we of course neither can nor want to cover…] (HIDE) The absence of a reason for excluding material or topics is a defining feature of the German-language corpus texts: the difference between the German scholars and both the British and US-American scholars is statistically significant at the p<0.05 level (UK/ DE p=0.030, USA/ DE p=0.044; UK+USA/ DE p=0.009). There is no significant difference between the two Anglophone groups for this aspect of exclusive TC use. The assumption behind such usage in the German corpus texts seems to be that the author and reader have common interests: if a topic does not interest the author, it naturally need not interest the reader. In some cases, the author’s projection of his or her own needs onto the reader may assume a commonality that does not exist, in which case this type of expression may appear arrogant. On the other hand, perhaps the German scholars in SCEGAD simply assume that their readers are not interested in the reasons for particular authorial choices. It is possible that German readers might regard personal explanations of the motivation behind textual decisions as onerous or irrelevant, and are more interested in simply knowing what will be treated and what disregarded. Again, this question cannot be resolved here. In all three cultures examined here, exclusive TCs appear markedly interactive in character, since they are a reply to anticipated reader expectations. However, the needs and wishes authors assume their readers to possess, and the texts readers appear to expect, seem to differ across cultures. 4.10.3 Interaction and introductory text comments Introductory TCs comment on what is being done at present, or will be done immediately following. Introductory text comments fulfil three main <?page no="241"?> 241 discourse strategies: announcements, current comments and directives. The first type constitutes an explicit announcement of the next rhetorical move in an author’s argument (Crismore 1989: 4), and introduces the speech action, topic or text section that immediately follows: (211) As a step in exploring the uneasy relation between performance theory and feminist theory, however, I am going to advance an outrageous claim; namely, that… (FLUS) (212) I therefore now turn to the conditions on reference for telos in more detail. (LGUK) (213) Folgende These soll in diesem Abschnitt geprüft werden: … [The following hypothesis is to be examined in this section: …] (LGDE) Though this first type of text comment is similar in form to many examples of the prospective TC type, the discourse context is different; here the reference is to a future action that follows immediately, rather than at some later point. Introductory text comments with this discourse purpose typically appear either at the very end of a section, referring to the very next, or at the very beginning of a new section. They prepare the reader for what is about to come, and are a way of retaining constant contact throughout the text. The second type of introductory text comment, which I shall call current comments, refers to the speech action or argumentative move being carried out at that very moment. Here are three examples from the literary studies texts: (214) I am suggesting, then, that… (LTUK) (215) Both the poets we are presently considering apparently…(LGUS) (216) …in der hier in Rede stehenden Zeit…[…in the period being discussed here…] (LTDE) The discourse purpose behind these comments is slightly different: here the motivation is not so much to prepare the reader, but to clarify the author’s current intentions. Current comments are therefore an answer to the implicit reader question ‘What is the author trying to do right now? What is s/ he getting at here? ’. This type of introductory TC commonly occurs in the middle of a text passage, and helps to orientate the reader in the chain of argument. Such comments can also serve as a reminder of the central topic or concern of the article: <?page no="242"?> 242 (217) … the two approaches converge on the main point being made here: the… (LGUK) (218) The claim we are making in this study predicts the… (LGUK) (219) In den hier im Mittelpunkt stehenden Periodika entwirft Reichard…[In the periodicals that are central here Reichard…] (LTDE) These introductory TCs provide a delicate jog of the reader’s memory, should he or she momentarily forget, or fail to understand, the end point of an author’s argument. As such, this type of introductory TC aids reader comprehension, and helps ensure that author and reader remain at each other’s side while on the path to a specified textual goal. Their use reflects the author’s assessment of which are the particularly difficult stages in the argument, or concern that the central point has not been made clearly enough for the reader to discern it. The author therefore explicitly emphasises the current aim with the help of a text comment. The third type of introductory TC serves not to remind readers of textual or communicative aims, but to direct them to where authors wish them to go. While it is clear that authors, and not readers, control the progression of an academic article, a text without a more-or-less voluntary reader following in the author’s wake is of little use. Since authors often wish to achieve a great number of complex communicative goals within any one text, readers may need to be told what is required of them at a particular point. Such signals ensure that readers do not get lost somewhere along the communicative journey. Academic authors do not issue orders as such, of course, but encourage, suggest and invite their fellow colleagues to carry out the rhetorical actions necessary at specific junctures. Some invitations are rather direct: (220) Let us go through some of these arguments and relate them to…(PHUK) (221) Setting aside more general differences between the two speakers’ uses of language, let us now focus more specifically on…(LGUS) (222) Vergegenwärtigen wir uns kurz die…[Let us briefly bring…to mind] (HIDE) In English, these more direct invitations take the ‘let us…’ form; in German, they are formed with an inverted subject present tense verb and inclusive ‘we’ pronoun. Wales calls ‘let us’ a sort of “chummy critical idiom”, and suggests that is it used to disguise an I-instigation: ‘Disguises’ is the crucial word: despite the surface meaning of modesty and joint activity which ‘we’ cohesively reinforces, the real agentives, the ‘I’ and the ‘you’, <?page no="243"?> 243 are often apparent in interpretation, and the authoritative, persuasive voice of the speaker cannot really be avoided. (1996: 66) Academic readers are no doubt perfectly aware of the real agentives: if they were not, the communicative strategy would not function. I doubt therefore whether disguise and avoidance are the central issues here. There is no lack of British and US-American authors in the corpus who write ‘I shall examine’ rather than ‘Let us examine’. However, the underlying discourse strategy behind this usage is to encourage the reader to co-operate in the construction of the intended meaning. The SCEGAD authors are not primarily trying to disguise their agency, but rather to create co-agency, a communal agency that includes the reader. The other strategy certainly exists, even in text comments (see the discussion in chapter three, section seven), but one should not discern it in each case where there is a seeming mismatch between grammatical form and discourse referent. There is in fact no mismatch here: certainly the action is instigated by the author, but if the reader were not to co-operate, that is also go through the arguments, consider X or focus on Y, the communication would fail. We are therefore dealing here with the textual configuration of co-operative action, and not primarily with the avoidance of authorial presence. The third type of introductory TC is therefore a way for authors to exert influence over the meanings which readers construct from their words, and ultimately helps ensure that readers draw the conclusions authors wish them to draw. Hyland regards ‘let’s’ as pragmatically equivalent to ‘we will’; both are “collective first person imperatives”, through which authors “attempt to lead readers by inviting them to participate, rather than by making demands on them” (2002b: 227). This distinction is vague: authors are clearly making demands here, whether couched as an invitation or in any other fashion; the difference lies in how these demands are expressed. Relative to other ways of expressing demands attested in SCEGAD, ‘let us’ is in fact relatively direct. Other directives are even more oblique: (223) We can see this point from another angle. (PHUS) (224) We might now enquire how the… (LGUK) (225) Erwägt man schließlich den…[Finally if one considers the…] (LTDE) In English, the more oblique directives are formed with ‘we’ + modal; in German, they are expressed as above with an inverted subject present tense verb, but here always with an impersonal subject ‘man’ [one]. This indirectness is essentially a politeness strategy: I very much suspect that <?page no="244"?> 244 readers know full well that ‘we might’ and ‘if one’ are nothing more than paraphrases for ‘I want you to’. 93 A bloody-minded reader might be inclined to reply that ‘we can’, but ‘I don’t want to’. As with all conversational implicature, authors assume here that their readers apply the cooperative principle (Grice 1975), otherwise they could not express wishes in such an indirect fashion. 94 Directive introductory TCs in particular are evidence of author-reader interaction. Thompson regards such in-text commands issued by the writer as one of the main devices for performing overt dialogic reader interaction (2001: 66), and Hyland too found that authors use the combination of imperatives and inclusive pronouns to create a more personal relationship with the audience by directly involving readers as participants in the textual process (2002b: 228). This combination occurs in both the Englishand German-language introductory TCs across all five disciplines in the corpus. None of these discourse strategies is particular to a specific cultural group, nor, as we saw above, are there any significant differences in frequency between cultural groups for this TC type. The general impression is that academic authors, whether British, US-American or German, are at pains to ensure that their readers understand and can follow them at all stages of the text, and interpret no more and no less than is intended. 4.10.4 Interaction and explanatory text comments Explanatory text comments provide the reason behind specific textual decisions, and answer the implicit reader questions ‘why this? ’ and/ or ‘why this now? ’ Through explanatory TCs, authors explain why they have done particular things, why they have done or must do them at this point rather than another, and why they have done so in a particular way. Explanatory text comments are particularly interesting features to examine in academic writing, since they offer an insight into individual authors’ conception of their obligations, and their readers’ obligations, within the written medium and as part of a specific discourse community. As shown in the quantitative analysis above, explanatory text comments are the only TC type for which there are significant differences in 93 This is not to say that the rhetorical effect of impersonal versus first person plural pronouns is necessarily the same across cultures. For example, a US-American colleague to whom I showed these examples described the first person plural pronoun invitations common in the English-language texts not as polite, but as ‘horribly presumptuous’. Politeness, of course, is a culture-specific phenomenon. 94 The cooperative principle runs as follows: “make your conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged” (Grice 1975: 41). <?page no="245"?> 245 frequency across cultures. It emerges that German scholars in the corpus are far more likely to employ this type of text comment than either of the Anglophone groups (p<0.001 in both cases). The highly significant difference between the German authors and both other cultural groups indicates a difference in rhetorical strategy. The German scholars represented in SCEGAD consistently represent their textual decisions as the result of external obligations. In this way, they deny or disguise their own communicative agency and intentions, suggesting that they are not the driving force behind the text, but merely the instrument of external factors and obligations. In some cases, this may reflect actual restrictions imposed from outside, such as editorial decisions or word limits. In most cases, however, it appears to be more a rhetorical strategy. In contrast, the British and US-American scholars studied are more likely to assume explicit personal responsibility for their textual choices. The representation of textual decisions as external obligations in the German texts makes it harder for the reader to criticise, and indeed this may be part of the motivation for the German scholars’ frequent use of such strategies in the SCEGAD corpus. Explanatory text comments are employed for a variety of discourse purposes. First, they explain the purpose of particular pieces of information: (226) A brief survey of the…will show how… (HIUK) (227) A closer examination of the…reveals how the…(HIUS) (228) Eine Betrachtung dieser…Optionen verdeutlicht, welche Bedeutung…[An examination of these…options clarifies what importance…] (HIDE) In addition, authors from all three cultural groups use purpose clauses to explain textual decisions: (229) To understand the unacceptability of (48), the interaction of the Telic Clause and the only preface needs to be considered in more detail. (LGUK) (230) To begin to answer these questions, I think it is important to step back for a moment to look at…(LGUS) (231) Ich stelle dieses Beispiel an den Anfang, weil es deutlich macht, dass… [I place this example at the beginning, because it demonstrates that…] (FLDE) <?page no="246"?> 246 In this way, authors justify their approach and the information they chose to provide. Like introductory TCs, explanatory text comments can, however, also be used to issue reader commands. A particularly forceful way of couching these is with ‘must’ / ‘muß’: (232) If… is to be assessed, the….must be taken into account, as must… (FLUK) (233) The elucidation and defence of…must begin with an explanation of what I mean by…(PHUS) (234) Diese…Verwobenheit…muß berücksichtigt werden, wenn man Rolle und Funktion…bestimmen will. [This interweaving…must be taken into account, if one wishes to assign a role and function….] (LTDE) A possible discourse purpose behind such statements is the projection of an authoritative writer stance; here, unlike elsewhere in the RAs, the emphasis is not at all on the co-operative construction of meaning, but on the writer’s control of the text, and the influence of outside demands on the writer. It is not clear whether the writer or some external authority has decided what ‘must’ be done here; it is absolutely clear, however, that it is not the reader who has the power to decide. There are more subtle ways of issuing commands: (235) Perhaps, as a fourth attempt, we should consider…and consider the formulation…(PHUK) (236) Whether Waldorf is historically Southern or South Midland, then, two things should be particularly noted to give context. (LGUS) (237) Vielleicht sollte der Ursprung, die Idee zur Love Parade, nicht vergessen werden: … [Perhaps the origin, the idea of the Love Parade, should not be forgotten: …] (FLDE) For this reason, Thompson regards Swales et al.’s (1998) study of imperatives in academic writing as too narrow in focus: a fuller picture would have also included “modulated statements functioning as writer commands”, such as ‘It should be borne in mind that…’ (Thompson 2001: 76). Another way of issuing oblique commands is for writers to comment on what is necessary, of course solely from their own point of view: (238) Since this is arguably the dominant approach to…among psychologists of the present day, we need to give it serious consideration. (PHUK) <?page no="247"?> 247 (239) To connect future work solidly with this base…we need both to probe previously undetected gaps and simply to remind ourselves of the purposes for which these theories were created… (FLUS) (240) Dazu ist es erforderlich, auf ein Problem einzugehen, das…[To this end it is necessary to consider a problem, which…] (PHDE) In the German-language corpus texts, there is a far greater variety of other devices for expressing reader obligation, many of which translate only approximately into English. Some examples: (241) Vorschläge sind gemacht; es gilt, sie zu überprüfen. [Suggestions have been made; it is necessary to examine them.] (LTDE) (242) Dies ließe sich am Beispiel der…erläutern, lässt sich aber vielleicht auch auf den Punkt bringen mit…[This could be explained using the example of…but can perhaps also be summed up by …] (FLDE) (243) Bevor ich auf…eingehe, ist jedoch an das…zu erinnern. [Before I address…a reminder of…is however to be given.] (PHDE) (244) Einige dieser Beobachtungen seien genannt. [Let some of these observations be cited here.] (HIDE) (245) Die zu diskutierende Frage lautet somit: …[The question to be discussed thus runs: …] (PHDE) (246) Verflechtungen und komplexe Beziehungen wären hier nachzuzeichnen. [Connections and complex relationships ought to be traced here.] (LTDE) (247) …, als Namen mögen Victor Turner oder Peter Burke genügen. [As far as names go, may Victor Turner or Peter Burke suffice.] (FLDE) The greater variety of linguistic realisations of this rhetorical strategy is most likely a reflection of the greater frequency of explanatory TCs in the German texts, rather than evidence of a greater inventory of expressions for indicating obligation in the German language. For Hyland and Tse, obligation modals, which encourage a particular thought or action on the part of the reader, are important devices for integrating readers into the text as “participants in an unfolding dialogue” (2004: 172). The researchers therefore perceive such devices as engagement features, and indeed they do bring the reader’s anticipated reactions into the text: (248) Es ist die zunächst vielleicht befremdliche Frage zu stellen, ob… [The perhaps initially strange question of whether…must be posed.] (FLDE) <?page no="248"?> 248 (249) Es müssen daher zunächst vermeintliche Nebenschauplätze…betreten werden, die jedoch für das Verständnis der…unerlässlich sind. [Thus issues of ostensibly secondary importance, which are however vital to the understanding of…. , must be…considered.] (HIDE) However, this engagement is of an imbalanced kind, because readers are clearly the inferior partners in this dialogue, with no option but to obey what are essentially commands, whether they find them ‘strange’ or not. It does not matter that the issue may appear unimportant to the reader - the author considers it important, and so it will be discussed nonetheless. Likewise, the list may suffice or it may not: either way, the author has decided that no more examples will be provided, and the reader must accept this. Such is the nature of RA interaction. In order to ensure readers’ continued acceptance of their textual subjugation, it appears to help academic authors to couch their commands as objectively necessary tasks: not ‘You do this’, but ‘it should/ must be done’. At times, it seems that the academic writers in SCEGAD are conscious of the high demands they make of their readers, mixing some coaxing and pleading in with the coercion: (250) Some of the highlights of this academic work deserve at least a brief overview, since… (FLUS) (251) …the question of…nevertheless deserves a few words if only to highlight more clearly how… (HIUS) (252) Wenigstens auf eine literarisch realisierte sei hier noch hingewiesen. [Let reference be made at least to one literarily realised one here.] (LTDE) Just one, just at the margins, at least a brief overview, in a few words: such phrasing suggests an anticipated reader who has little patience left for being told what do to, and little time in which to construct meaning actively. What is evident in explanatory TCs, at least, is the author’s concern for how the imagined reader will react to such bidding. However, this type of engagement is still very different from the co-operative speech actions encoded for example in introductory text comments. In introductory TCs, writers issue what are essentially commands as invitations or exhortations; in explanatory TCs, they couch them as obligatory actions. Both types of device involve the reader, and in both cases readers ultimately have no choice but to obey if they wish to continue reading, but the two types involve readers in a different role. A mismatch between the communicative situation and the interactive role reduces communicative success: few people would reply to a party obligation, and an invitation to <?page no="249"?> 249 stop at a traffic light would hardly be effective. Academic authors use both strategies, at different moments, to ensure continued reader involvement and co-operation in the textual process. Readers may have no choice but to obey, but authors are still concerned to project some semblance at least of a reciprocal, voluntary, egalitarian interaction. The level of concern expressed varies across TC types, and according to cultural background. Explanatory text comments have to do with obligations. From a discourse point of view, it makes a difference whether authors couch these obligations in terms of their own or the reader’s needs (personal, internal motivation and justification) or in terms of requirements imposed upon them from outside (impersonal, external motivation and justification). The two types of modality have been assigned various labels. The first type, expressing personal, internal obligation, is known as dynamic, circumstantial, or volitive modality, while the second type, expressing impersonal, external obligation, is known as deontic or necessity modality (Lyons 1977; Kärkkäinen 1987; Jacobsson 1994; Lew 1997; Palmer 2001). I shall use the terms dynamic and deontic modality, following Palmer (2001). The corpus texts reveal significant cultural differences in the preference for either dynamic or deontic modality in explaining textual decisions. When compared across cultural groups, it emerges that the German authors are far more likely than both the British and US-American scholars to express explanatory TCs using deontic rather than dynamic modality: this difference is extremely significant in statistical terms (p<0.0001 for both comparisons). There are no significant differences between the British and US-American scholars for this feature (p=0.867). Authors who express requirements in terms of their own, and their interaction partner’s needs, draw attention to the fact that the structure of their text is orientated to their own wishes and the anticipated wishes and needs of their audience. This is the discourse strategy pursued by the British and US-American scholars in SCEGAD. It seems to me that by personalising the reason for specific textual decisions, the Anglophone authors create a persona which assumes responsibility for the shape of the text. The author constructs a persona which has decided what is appropriate and necessary and when, while bearing in mind the needs of the imagined audience. The English-speaking scholars in the corpus are therefore visible as the driving force behind the text and emphasise their own role in individually shaping its structure. The German authors in SCEGAD, on the other hand, explain textual decisions in terms of general obligations or external requirements. One possible effect of this strategy on the reader is that the structure of the text appears to be orientated not towards the authors’ or their reader’s wishes, but dictated by external laws and rules about what is appropriate or necessary and when. This may not be the <?page no="250"?> 250 authorial intent behind this strategy, and it is of course possible that the strategy could create a different impression among German readers. Other possible intents and effects are politeness (‘I don’t want to order you around, but X simply must be done’) or perhaps an appeal to accepted routines (‘X must be done, as we both well know’). From my own perspective, however, it seems that by de-personalising the reason for specific textual decisions, the German scholars surrender or attempt to obscure their responsibility for the shape of the text. The German academics in the corpus are less visible as the driving force behind the text, creating a writer persona that is the instrument of forces and/ or circumstances beyond its control. The difference in presentation and tone, even for the same communicative intention (in this case ‘posing a rhetorical question’), is marked: (253) The question we had to ask on first reading…, is whether… (HIUS) (254) Und deshalb bleibt nichts anderes übrig, als sich der Frage zu stellen, wie…[And this there is no choice but to ask oneself how…] (PHDE) (255) Es wird danach zu fragen sein, ob hier zwei Parameter bzw. Komponenten von Konturen kombiniert werden: …[It will have to be asked, whether two parameters or respectively components are being combined here: ….] (LGDE) The distinction between dynamic and deontic modality is one way of examining explanatory text comments. Another way of examining this type of text comment, which is also interesting from a discourse point of view, is to distinguish between those explanatory TCs where a reason or explanation is given for a textual decision, and those where there is no reason given, but rather an appeal to mute necessity, or a bald statement of obligation. This is related to, but not the same as the question of dynamic or deontic modality: it is possible to give a reason that is not personal, just as it is possible to explain the reason why external circumstances dictate a particular textual decision. Rahman notes that explanatory TCs can aid reader comprehension, since explanations of why certain choices were made can prevent misreading (Rahman 2004: 40). However, where no reasons are given, aiding reader understanding cannot be the discourse motivation behind such comments. There are significant differences across cultural groups in the frequency with which explanatory TCs cite reasons when explaining textual decisions. When compared across language groups, it emerges that the German authors give reasons for their behaviour far less frequently than their Anglophone colleagues (p<0.0001). This difference is extremely significant in statistical terms. When compared across cultural groups, it <?page no="251"?> 251 emerges that the difference between the German and British scholars is highly significant (p<0.005); that between the Germans and US-Americans is extremely significant (p<0.0001). There is no significant difference, however, between the two English-speaking groups when it comes to this aspect of linguistic behaviour. From an interactional perspective, it makes a difference whether authors explain the reason behind their actions or not. Authors who state that particular things must be done, or done at a certain point, without explaining why this is so, create an author persona that can seem authoritarian and somewhat paternalistic: (256) Die Aufmerksamkeit muß…gelten, desgleichen und ganz besonders aber auch…[Attention must be directed towards…but likewise and especially also towards…] (LTDE) (257) Es darf aber nicht übersehen werden, daß…[But it may not be overlooked that…] (LTDE) (258) Dabei sollte jedoch nicht vergessen werden, dass…[In doing so, it should however not be forgotten that…] (FLDE) The suggestion here is something along the lines of ‘I know what is required, and how, and feel no need to explain why’. The rhetorical effect of this strategy is to create a relationship with the reader that is markedly asymmetrical in its power structure. On the other hand, as for exclusive TCs, another possible interpretation is that German authors give no reasons simply because they anticipate that their intended readers are not interested in hearing them. The lack of cited reasons could also indicate the trust that exists between author and reader, along the lines of ‘I know what is required, and how, and know that you trust me enough not to need an explanation of why’. However, even in this case, the relationship between author and reader is still somewhat authoritarian. It might also be the case that the German authors in the corpus simply assume that expert readers already know why something is necessary, either because it is obvious from the context: (259) A, B und C treffen hier in so grundsätzlicher Weise aufeinander, daß wir nicht umhin können,…unsere Aufmerksamkeit zuzuwenden. [A, B and C collide here in such a fundamental fashion that we simply must to pay attention to….] (HIDE) or because specialist knowledge of the disciplinary context or the literature must (the author assumes) immediately suggest to them the reason for decisions: <?page no="252"?> 252 (260) Neben der Obligatorik ist bei der paradigmatischen Variabilität auch die Frage nach den Realisierungsalternativen zu stellen. [With regard to paradigmatic variablity, alongside obligation, the question of alternative realisations must also be posed.] (LGDE) (261) An Monographien sind zu nennen: … [Monographs that must be mentioned are: …] (LGDE) (262) Aus einer ethnologischen Perspektive muss gefragt werden, was… [From an ethnological perspective, it must be asked what…] (FLDE) Ultimately, we can only speculate on the effect these explanatory TCs would actually have on the target audience. In contrast, authors who give reasons for their behaviour might be regarded as creating a more egalitarian relationship with the reader. This is the strategy pursued far more frequently by the English-speaking scholars in SCEGAD. One possible rhetorical effect of this strategy is the creation of a more personal, but also more individual, and perhaps even somewhat idiosyncratic tone, especially when combined with dynamic modality: (263) I can best give my reasons for this belief if I first dispel the paradox which Dummett claims to find in my criterion for applying…(PHUK) (264) To illustrate what I have in mind, we can consider a simpler case. (PHUS) (265) We consider it worthwhile to detail the reasons for reaching this conclusion - the purpose of this section. (LGUK) At worst, usage such as this could create the impression in the audience that they are at the mercy of the author’s whims. This would be a very different, but also possible, rhetorical effect from the egalitarian relationship cited above (examples 176 - 178). Alternatively, it is also possible that explanatory TCs that cite reasons could be a sign of insecurity, and act as a defensive measure against the reader. English-speaking writers in particular apparently feel obligated to explain constantly not just that things must be done, but why they must be done: (266) The reason my argument has revolved around…is that…(LTUS) (267) To explain why it is that …we will need to show how the interpretation… (LGUK) (268) I organize things in this way in part because it makes the argument easier to understand, and in part because the stage-two thesis - that CT entails <?page no="253"?> 253 LOT - has some independent interest and is therefore worth separating from the rest of the argument. (PHUS) In these cases, perhaps there is simply not enough trust present for the simple appeal to obligatory requirements to satisfy the reader that authorial decisions are legitimate and cogent. Ultimately, it cannot be decided here what effect these devices would have on the target discourse community: readers must and do judge for themselves based on their own reaction to the texts, and these reactions are coloured by culturally specific expectations. No doubt the rhetorical effects could be quite different upon readers who are used to another strategy. 4.10.5 Interaction and prospective text comments Prospective text comments are references to what will be done later in the text: they are therefore a form of cataphoric reference. Prospective TCs highlight the relationship between what has been done already, and what will be done at a later stage. Such references can help readers prepare for the tasks, arguments and concepts that will confront them further on, thereby helping them to make connections between different text sections. The primary discourse function therefore is to aid reader comprehension. Sometimes authors feel the need to prepare readers for coming material by announcing their intentions regarding the following section, explaining what the following examples are to show, or indicating the topic of the following text extract: (269) In the following section we will ask whether the use… (LGUK) (270) Weniger stark und abrupt ausgeprägt, aber in der Form dennoch zu erkennen, ist der finale Anstieg in den nächsten Beispielen. [The final rise is less strongly and abruptly pronounced, though still recognisable in form, in the following examples.] (LGDE) (271) Im folgenden Ausschnitt reflektiert Enno darüber, mit welchem Verkehrsmittel er in die Stadt Xlingen fahren soll: … [In the following extract Enno reflects upon which mode of transport he should go to Xlingen by…] (LGDE) Such references are particularly common in the linguistics texts, as discussed above (section 9.4). They help readers to make sense of the text, interpret examples, and grasp the theme of specific pieces of information. Without this additional information, readers would not be able to follow the text as easily or in the way the author intended. <?page no="254"?> 254 Sometimes prospective TCs provide information that is not just additional, but indispensable: (272) Die sprachlichen Merkmale, um die es mir geht, sind in der folgenden Textwiedergabe durch Fettdruck markiert. [The linguistic features with which I am concerned are marked in bold in the following text extracts.] (LGDE) (273) …diese bilanzierende Sentenz Aleida Assmanns gilt es für die folgenden Überlegungen im Auge zu behalten… […it is essential to bear this evaluation by Aleida Assmann in mind during the following discussion…] (FLDE) If readers cannot recognise the phenomena being discussed, or if they lose sight of a quote that sums up the whole point of the article, they will experience great difficulty in trying to construct any knowledge from the text. Prospective text comments can refer to all sorts of entities, from text sections, analyses and data to future actions and even states of mind. References to future text sections and data have the most clearly organisational, formally-oriented function: (274) The data in the Appendix shows that all sixteen possibilities are available. (LGUK) (275) … the next section deals with the methodology of data extraction and the delimitation of the sample used in this study. (LGUS) (276) Für eine weitere Frage der FrDD10 mit Treppenkontur und weiterer Steigung am Ende siehe dd10-0823. [For a further question from FrDD10 with an upward staircase contour and further rise at the end see dd10-0823.] (LGDE) References to coming actions have a more narrative function, in that they situate upcoming speech actions in the series of steps that make up the argument: (277) I shall briefly indicate later how…(PHUK) (278) My argument in the rest of this section is that this…(LGUK) (279) Im folgenden werde ich zeigen, daß dies nicht richtig ist, daß vielmehr...[In the following I shall show that this is incorrect and that in fact …] (LGDE) <?page no="255"?> 255 References to future states of mind are partly persuasive in function, guiding the reader in their thoughts even before any supporting evidence for these upcoming conclusions has been presented: (280) As we shall see at the end of this paper, problems do lurk in such a use of the term… (PHUK) (281) Rather, the…set in motion circumstances that, as shall become apparent, at first denied Eunice the privileges of white womanhood... (HIUS) (282) Es wird jedoch klar werden, daß gerade diese Beispiele theoretisch von besonderem Interesse sind. [However, it will become clear that these examples in particular are of especial theoretical interest.] (PHDE) The frame of reference for prospective TCs sometimes spans only a few sentences, as in examples (190) and (191). At other times the references relate to material to be dealt with far later, or even at the very end of the paper: (283) The semantic underpinnings of this contextual effect are explored in section 4.2. (LGUK: in section one) (284) As I shall show at the end of my final section…(PHUS: in section one) (285) Auf unterschiedliche Forschungspositionen gehe ich in Abschnitt V ausführlich ein. [I respond to the various views found in the research in more detail in section V.] (LTDE: in section one) While some references are incredibly specific: (286) … a point we return to in section 5.2. (LGUK) (287) … noch eine zweite folgt (siehe Abschnitt 3.2. dieser Arbeit). […a second follows (see section 3.2 of this article).] (LGDE) Other prospective text comments are extremely vague in reference: (288) This raises the significance of…, a point that will be considered later. (HIUK) (289) It will be suggested below that….(PHUK) (290) Diese letztere Kontur, die an anderer Stelle beschrieben werden soll,…[This last contour, which will be described at a different point,…] (LGDE) <?page no="256"?> 256 The discourse strategy is the same here, whether authors refer to a specific text section or only to an imprecise later location. The only difference perhaps is that authors who cite a specific segment enable readers to check more easily whether they really have kept their textual promises. More specific prospective TCs also enable the reader to scan the text for particular pieces of coming information. If an author refers only to ‘later’ or ‘below’, readers must go through the entire text if they wish to find the argument announced in the text comment. As can be seen in examples (201) - (203) above, authors conceive of their text as both a spatial and a temporal unit: spatial conceptions place more emphasis on the text as a physical object (product), whereas temporal conceptions figure the text as unfolding discourse (process). Textual organisation is not always linear: in fact, the corpus authors relatively often raise a topic only to return to it later. In some cases, scholars announce the fact that this will occur: (291) I shall return to this point later. (PHUS) (292) …work (I will return to this potentially hypocritical stance below). (LGUS) (293) [11]…(more on this in fn. 23). (LGUK) (294) Ich werde nachher noch auf diese sehr außergewöhnlichen Ereignisse zurückkommen. [I will return to these very unusual events later.] (FLDE) Comments such as these aid reader understanding of a text structure that is unpredictable and potentially difficult to follow. When authors choose, for whatever reasons, not to group thematically-related material together, it can ease comprehension for their audience if they provide such TCs. In other cases, instead of announcing that they will return to an issue later, authors state that they will address it later or in more detail: (295) …(this will be discussed in detail in section 4 - 2). (LGUK) (296) In particular, each poet has one especially striking intonation pattern that I will discuss separately in more detail. (LGUS) (297) Ich werde auf diese Problematik später noch eingehen. [I shall address this problem later on.] (FLDE) Such comments serve as a kind of ‘that’s not all folks! ’ discourse flag. The likely rhetorical strategy behind them is the anticipation of reader criticism. By referring to their future intentions, authors attempt to defend them- <?page no="257"?> 257 selves against the charge that they have not addressed a relevant topic, or not addressed it in enough detail. At times, this defence becomes overt: (298) … (explanation will be provided shortly)… (PHUK) (299) I am well aware that my claims require substantiation, and I will attempt to provide it, not point by point, but rather from several different angles. (FLUS) (300) Die Relativierung des Begriffs der…wird noch erläutert (vgl. Abschnitt 3). [The relativisation of the concept of…will be explained later (see section 3).] (PHDE) Prospective text comments can therefore be a very useful way for authors to assuage anticipated reader criticism. By promising to attend to probable concerns later, authors hope to delay expected negative judgements or objections until such time as they can address them. In some cases, such flags appear to be used to create something approaching suspense: (301) For reasons soon to be explained,… (PHUK) (302) Much more will be said about this question below; … (PHUK) (303) The transnational family story to be told here centers on… (HIUS) With luck, readers’ curiosity will be aroused, and they will want to continue reading in order to discover the ‘much more’ promised and the ‘soon’ to be explained. Though this strategy is a marginal one in the SCEGAD articles, it is an interesting one, and gives an indication of how authors attempt to capture and maintain reader interest over the course of the text. The challenge, of course, is that it requires quite some effort and time to read a research article, and there are a great many good research articles about. One of the hallmarks of written texts, as Dillon points out, is the convention that authors structure their written work more than they would for example an impromptu speech (1981: 30). Research articles, for example, are commonly divided into sections and often also subsections. Within the sections, the argument is structured so that one point leads to the next, with arguments building on what has come before, and connecting with what is yet to come. This is evidence of the amount of attention academic authors pay to textual organisation; nothing similar occurs in spontaneous informal interaction either written or spoken. 95 Genres which obligate authors or 95 Something similar does however occur in prepared formal speech, such as in academic lectures for example. <?page no="258"?> 258 speakers to structure their work closely also obligate readers or listeners to pay more attention to how the text is organised, and to the relationship between the sections. This of course assumes that readers can competently discern formal patterns in stretches of discourse (Dillon 1981: 31). The ability to recognise text structure, and make connections between parts, is both a sort of genre knowledge and a measure of specialist reading competency. It is crucial to understanding a text properly. If we do not know that recipes specify the required quantities of each ingredient in a separate textual location from the chronologically organised instructions, or that obituaries place the deceased person’s full name at the end of the text, the cake will not rise and we will not know who has died. Similarly, knowledge of the RA genre probably includes knowledge of the component parts (often: abstract, introduction, body of the text, conclusion, references) and the order in which they typically appear. This by itself, however, does not give readers the ability to discern automatically the connections and relationships between these parts. It is here that authors must sometimes help, by commenting on the more intricate sorts of textual structuring they have carried out. As we have seen, one way of doing this is with prospective text comments. Another method is through retrospective text comments. 4.10.6 Interaction and retrospective text comments Retrospective text comments are references to what has already been done in the text: they are therefore a form of anaphoric reference. In all three cultural groups examined here, retrospective TCs were the most frequently used type of text comment. The high frequency reflects the cardinal role retrospective TCs play in ensuring the effective transfer of knowledge from author to reader. In the same way as in prospective TCs, retrospective text comments too can refer to all sorts of entities (evidence, speech actions, arguments, analyses, data), and can be more or less specific in indicating where exactly the material referred to was situated: (304) …the evidence for…I mentioned a few paragraphs back. (PHUS) (305) Ich hatte oben erwähnt, dass… [I had mentioned above that…] (FLDE) (306) I used a form of argument, at the end of section IV, which…(PHUS) (307) Im Dresdenerischen werden ähnliche Konturen wie die oben in Abschnitt 3.1 beschriebenen…verwendet… [In the Dresden dialect similar contours to those described above in section 3.1…are used…] (LGDE) <?page no="259"?> 259 Retrospective text comments can take a variety of forms, from the highly involved, personal and colloquial: (308) If you look back to (97), you will notice that… (LGUK) (309) All that I have said up to now might, I think, be acceptable to Gibbard…(PHUK) (310) The volume, as we’ve begun to see,… (LTUS) (311) Ich habe schon auf die Stelle hingewiesen, an der…[I have already referred to the place where…] (PHDE) (312) Hier dürfte, wie gesagt, das Absinken der Tonhöhe nach der ersten Version der… [The fall in pitch here after the first version of the…should, as was said…] (LGDE) to the detached, routinely impersonal and formal: (313) Locke’s point…as will be recalled…(PHUK) (314) As has already been observed,… (FLUK) (315) In the analysis of copula absence above, it was shown that… (LGUS) (316) Die vorgeführte Dialoganalyse berechtigt also…[The dialogue analysis performed therefore justifies…] (PHDE) (317) Die vorgestellte Lösung sollte…[The solution presented should…] (LTDE) The SCEGAD authors often include references to what they have previously argued and/ or shown. This is commonly done when the authors wish to advance a knowledge claim: (318) But since, as we have written it and we have argued, GMR is not conceptually modal, the problem does not materialize. (PHUK) (319) This is not surprising since, as we have seen, it is difficult…(LGUK) (320) Die angeführten Textstellen dürften deutlich machen, dass… [The text extracts cited ought to have demonstrated that…] (LGDE) (321) Die Betrachtungen der vorangegangenen Abschnitte führen zu dem Ergebnis, dass…[The considerations of the previous sections lead to the result that…] (PHDE) <?page no="260"?> 260 (322) Nun, angesichts des eben erläuterten Fortgangs kann…offenkundig nicht so aufgefaßt werden, als sollte...[Now, in light of the progress just explained,…obviously cannot be understood as though …] (PHDE) This strategy of course makes perfect sense: authors provide material in support of a particular point or certain position, and it is therefore vital that they indicate the connection between the evidence they have collected and the conclusions they wish to draw. The discourse purpose here is to persuade the reader by referring to (one hopes) solid and substantial material in support of the writer’s case. Retrospective text comments explicitly indicate the relationship between new and old material. This strategy helps readers to process content by making connections with what has been presented before. The high frequency of retrospective TCs across all three cultural groups is especially interesting since research in cognitive science, educational psychology and instructional design suggests that the ability and opportunity to relate new information to existing knowledge structures is a nearly universal feature of effective learners (and also teachers) (Ausubel 1968; Adams & Bruce 1980; Anderson, Spiro & Montague 1984; Pearson & Dole 1987; Voss 1987; West, Farmer & Wolff 1991; Merrill 2002). Either consciously or subconsciously, academic authors take account of this in their writing practice. The anaphoric reference contained in retrospective text comments probably plays a vital role in the acquisition of new information and generation of new meaning. 4.10.7 Interaction and concluding text comments Concluding text comments explicitly draw the audience’s attention to what an article has shown. They are concluding in both senses of the word: they form the end of the article, and they express its final judgements: (323) What all this adds up to is that… (FLUS) (324) On the basis of these observations, I conclude that… (LGUK) (325) Nach all dem, was wir bisher über frühneuzeitliche Nürnberger Texte wissen, ergibt sich zusammenfassend folgendes Bild: … [After everything we now know about Early New period Nürnberg texts, in conclusion the following picture emerges: …] (LGDE) In concluding TCs, authors summarise main points, remind the reader of the main thesis formulated at the beginning, and direct readers to a specific <?page no="261"?> 261 point of view. Concluding TCs therefore fulfil both descriptive and persuasive functions. The descriptive function is most fully served by reviewing the article step by step: (326) In order to elucidate this cultural connection, the foregoing discussion has emphasized several ways that…First, the sources exaggerated…Second, they used analogies to….Third, they framed…Finally, they explained…(HIUS) (327) In diesem Aufsatz wurden die…untersucht. Dabei wurde erstens nachgewiesen,dass...Hinsichtlich der…wurde zweitens argumentiert, daß…Drittens wurde gezeigt, daß der…[In this article the…was examined. In doing so it was demonstrated first that…With respect to…it was argued second that…Third it was shown that the …] (LGDE) Detailed summaries such as these remind readers of the main sections and points in the article, and aim to increase their retention of the text. In the SCEGAD corpus, they are a relatively unusual strategy in both languages, however. This is possibly because they require much space without contributing any new information. If an article is convincing, readers will have been persuaded of the author’s position by the end of the text. The success of this persuasive venture can be increased by explicitly stating what the author believes the analysis to have shown. Scholars differ, however, in the degree of certainty and the degree of agency with which they present their final assessments. Some of the US-American and German scholars are comparatively tentative at this point. However, while the US-Americans in the corpus emphasise their own agency even here (I try, I hope), the German academics emphasise what should have occurred, rather like an explanatory TC in the conclusion: (328) In conclusion, I hope to have shown that…(PHUS) (329) I have tried to show that arguments…(PHUS) (330) I have tried to show, though, that…(FLUS) (331) Der Versuch…verlangt zweifellos nach weiterer Vertiefung. Bei aller Vorläufigkeit…dürfte jedoch klar geworden sein, wie…[Without doubt the attempt…requires further examination. Despite all precursiveness…it ought however to have become clear how…] (HIDE) (332) Zumindest so viel sollte aus der Behandlung der vorgestellten Zeugnisse sichtbar geworden sein…[This much at least should have become apparent from the discussion of the preceding evidence…] (HIDE) <?page no="262"?> 262 (333) Damit ist, denke ich, gezeigt, daß…[It has therefore been shown, I think, that…] (LGDE) Other colleagues speak of their findings not as attempts and precursiveness, but as established facts: (334) Besides offering insights into the syllabic dimensions of vowel disorder, the case study provides external confirmation of the independence of the prosodic and melodic facets of phonological representation. (LGUK) (335) The study of copula absence in…has shown that not only may speakers vary widely in the relative frequency of use of…, but they may also vary according to...(LGUS) (336) Die These, dass Präpositionalobjekte Eigenschaften aufweisen, die für Grammatikalisierungsprozesse typisch sind, hat sich damit bestätigt. [The hypothesis that prepositional objects exhibit characteristics which are typical of the grammaticalisation process has therefore been confirmed.] (LGDE) The manner here has more in common with a mathematical proof than a personal claim. The rhetorical effect is to create an air of absolute certainty that at first sight seems unlikely to exist in humanities disciplines. However, it is interesting that these examples occurred primarily in the linguistics texts: in this discipline, the results might in fact be falsifiable, for which such assurance may well be legitimate. It seems that we have evidence here of a way of knowledge making specific to this discipline. 96 The degree of personal agency which authors accord themselves in their conclusions also varies widely across the corpus texts. Indirect assessments of what the article has shown cast the author as simply the presenter of objective facts which speak for themselves: (337) The Swansea case thus illustrates the potential for… (HIUK) (338) The data presented in this article show…(LGUS) (339) The comparison between nevertheless and but has demonstrated that…(LGUK) (340) Die Analysen der wo-Konstruktionen in der (gesprochenen) Umgangssprache haben verdeutlicht, dass…[The analyses of ‘where’constructions in informal (spoken) language have shown that…] (LGDE) 96 By which I mean that it occurs here only among the five disciplines represented in the SCEGAD corpus. This is not to suggest that similar patterns do not occur in other humanities disciplines not examined here. <?page no="263"?> 263 (341) Die Untersuchung ergibt, dass…[The study shows that…] (HIDE) (342) Der vorliegende Beitrag postuliert keine…Er versucht jedoch…und gelangt zu dem Ergebnis, dass…[The present contribution does not postulate any…It tries however to…and comes to the conclusion that…] (HIDE) Here, cases, data, comparisons, analyses and studies become agents of their own. There is no mention of the fact that it is really the author who has done the illustrating, showing and demonstrating: ‘it tries’, not me. In contrast, direct, explicit assessments of what the article has shown cast the author as the active creator of knowledge: (343) In characterizing the construction I have argued in support of the view that… (LGUK) (344) Thus, I believe that… (PHUS) (345) I have argued that…; and I have argued that… (PHUS) (346) …wie ich gezeigt habe,… […as I have shown,…] (PHDE) (347) Mein Resümee…lautet… [My conclusion…is…] (HIDE) (348) Ich habe dafür argumentiert, dass…[I have argued in favour of…] (PHDE) Such assessments acknowledge the role that the human author plays in the construction of knowledge, reflecting the (sometimes-disguised) truth that scholarly writing is more than the passive transmission of facts. This strategy can be contrasted with the co-operative construction of knowledge figured in other conclusions: (349) In the end, we face the question of whether…(LGUS) (350) I think we must conclude, therefore, that…is false. (PHUK) (351) We may more modestly conclude that…(PHUK) (352) Unser Ergebnis ist somit: …[Our result is therefore: …] (HIDE) (353) Zusammenfassend können wir festhalten: …[In conclusion we can record: …] (PHDE) (354) Fassen wir abschließend die wichtigsten Gedanken zusammen, die teils in den …bereits formuliert wurden, teils sich aus der Vorstellung <?page no="264"?> 264 von…ergeben. [In conclusion let us summarise the most important points, which were in part already formulated in the…and in part emerged from the presentation of…] (LGDE) In all three cultures examined here, the final stage in the communal textual journey can be figured as an impersonal statement of findings, a personalised statement of individual achievements, or as a co-operative enterprise. Concluding text comments provide readers with clear signals of where they should be, and, based on the new information and arguments presented in the article, the opinions they should now hold. The vital argumentative, interpretative step in the process of academic reasoning lies in the connection of evidence to a thesis. This connection enables knowledge construction. It is therefore important for scholarly authors to indicate explicitly the link between what they have presented, and what they believe their presentation to have shown. Through concluding TCs, authors articulate this link. 4.11 Metadiscourse and readability Finally, a brief word on the implications of metadiscoursal devices such as text comments on readability. Numerous studies have postulated a link between the use of metadiscourse, of which text comments are a part, and comprehensibility (Conley 1979; Vande Kopple 1985; Clyne 1987, 1994; Crismore 1989; Crismore, Markkanen & Steffensen 1993; Cheng & Steffensen 1996; Hutz 1997; Hyland 2000; Lee 2000; Camiciottoli 2004). As we have seen, text comments often do appear to aid reader understanding. However, this does not mean that more text comments equals better comprehension: I could not understand a quantum physics or molecular biology RA no matter how many text comments they contained. Conversely, I have no more trouble understanding and following the two SCEGAD texts that contain no text comments at all (PHDE7 and PHDE10) than I do understanding any of the other German philosophical articles. 97 Even within the same specialist target group, it is not clear that a higher frequency of text comments will ensure better comprehension. Crismore for example initially suggests that using a great deal of textual metadiscourse is the sign of a “considerate author” and warns that attitudes to texts and even learning can be affected by the use of metadiscourse (1989: 14, 129). Tellingly, Crismore’s study of reader attitudes found that there was no preference among the respondents for texts manipulated to 97 Indeed, if I had not been analysing the frequency of text comments in these texts, I doubt whether I would have even noticed their complete absence from these two articles. <?page no="265"?> 265 include more metadiscourse, nor were these texts rated as easier to read. Alhough this may be a fault in the study design, as Crismore asserts (1989), it may also be an indication that the connection between metadiscourse frequency and readability is not as clear-cut as the scholars cited above suggest. In fact, it seems to me that concentrating on frequency misses the point altogether. The first point to note is that there are many ways of creating a readerfriendly text. From a cross-cultural point of view, it would be a mistake to assume that the construction of ‘readability’ relies upon the same devices in each culture. The second point to note is that, as the analysis above showed, metadiscourse fulfils many other functions apart from aiding comprehension, such as creating solidarity with the reader, anticipating criticism, and self-advertisement. To reduce metadiscourse (even a subset of metadiscourse such as text comments) to one function is a misrepresentation of the myriad rhetorical strategies academic writers employ. The third and most fundamental point is that the very concept of readability is meaningful only within a specific communicative context. The reason for this is that readability is a value judgement, not an intrinsic property of text: Impressions that sentences flow together well or jump around, that emphasis falls where it should or it doesn’t, are readily voiced by composition students. Describing these characteristics as properties of the text, or of sentences, probably reflects the common illusion that takes aspects of the reader’s expectations to be aspects of the text. (Dillon 1981: 101) Lack of comprehension, therefore, is not a function of the frequency of metadiscoursal devices such as text comments. Rather, difficulty in following a text arises when there is a mismatch between reader expectations and authorial strategies. 98 It is just as likely that readers will feel alienated by too many text comments as too few, as Williams (1981) and Rahman (2004) for example recognise. 99 What one reader finds helpful, another may find unnecessary, obstructive or even bothersome: it is all a question of expectations. A more fundamental point concerns attitudes to readability in academic texts. Essentially, readability is the reader’s subjective assessment of the effort required to understand the text. We therefore end up with roughly the following argument: 98 I am not suggesting, of course, this is the only reason for textual incomprehensibility. 99 Williams points out that excessive use of metadiscourse can become “mechanical and obtrusive” (1981: 197), while Rahman observes that it may interfere with the reading process and may look “outright imposing and condescending” (2004: 47). <?page no="266"?> 266 1. Good writing easy to understand minimal effort 2. Metadiscourse reduced effort therefore 3. High frequency of metadiscourse low level of effort easy to understand good writing The first step here is possibly culturally specific, the second is as we have seen a non-sequitur, and the third combines both these faults. Anglophone researchers and educators assume that readability is a positive and desirable trait: the US-American researcher Dillon for example states that “good writing is writing that allows intake (extraction) of propositional content with the least effort on the reader’s part” (1981: 17-18, see also Clyne 1987, 1994; Crismore 1989; Hyland 2000; Lee 2000). However, this judgement is not necessarily universally shared. First, effort, as an individual and subjective measure, will depend on the reader’s expectations. If a text conforms to conventions familiar to the reader, the effort will be less than if it deviates from them. Second, the value judgement here may be culturally specific: is minimum effort really the ultimate virtue in academic discourse? Should scholars aim to produce pre-digested material, calculated to cause minimum disturbance or exertion on the part of the reader? Dillon’s presentation of effort minimisation as a universally-accepted and universally-desirable goal is debatable and appears biased. On the contrary, some research aims to disquiet the reader, because it challenges conventional wisdom, and this requires reader effort. Many scholarly topics and concepts are complex, and readers may have to expend considerable energy if they are to understand them. Surely the quality of academic writing cannot be assessed according to the effort a reader must make to understand it. It seems to me that reader effort is in fact an extremely poor judge of ‘good writing’. German academic readers at least seem to agree: leading humanities scholars such as for example Niklas Luhmann and Theodor W. Adorno are recognised and appreciated despite their incomprehensible style. 100 This does not mean that authors should not strive to accommodate their readers. The point is simply that ease of comprehension - in any case an individual, subjective impression - is not necessarily a mark of ‘good writing’. Since there is also no direct connection between metadiscourse frequency and readability, it appears that the argument 100 It is worth noting that this difficulty arises not because the academics in question cannot express themselves in any other way: Luhmann for example is perfectly comprehensible when giving lectures. <?page no="267"?> 267 represented above is entirely faulty. Metadiscourse is a reflection of the shared assumptions and routines that, for Hyland, are the substance of a discourse community (2005: 191). The analyses above show that these assumptions and routines differ across disciplines and cultures. 4.12 Conclusion Text comments are one of the most important ways in which academic authors construct their texts as a dialogue with the imagined reader. The dialogue in academic texts is, however, not the direct, turn-taking interaction of conversation. In fact, the communicative situation in research articles does not allow actual readers any direct role in the interaction at all: rather, they figure as personae, as textual constructions conceived of by the author. Readers cannot give direct feedback if they do not understand, nor can they actively influence the course of the text. However, in order for the communication to be successful - claim established, knowledge transferred, audience persuaded - authors must ensure that readers do understand and can follow the text. Without knowing or hearing directly what effect their textual decisions have on the prospective audience, academic writers have no choice but to try to anticipate possible reader expectations and assumptions. The closer these anticipated characteristics match those of the actual reader, the more efficiently authors can communicate with and ultimately persuade the reader. The anticipated reader figures particularly clearly in text comments, for it is here that the writer turns to the audience, to comment on and explain the path of the text. The kinds of comments authors make reveal their conception of the imagined readership. Defensive text comments reveal the expectations authors imagine readers to hold, negative text comments the information authors imagine readers expect to encounter. For this study, a taxonomy of seven text comment types was developed. Overarching TCs explain the aims and structure of the entire text, exclusive TCs explain aims not set and topics not covered, and introductory TCs comment on the actions currently being carried out. Explanatory TCs explain textual decisions in terms of necessities or obligations, prospective TCs refer to coming material and retrospective TCs to past material. The last type, concluding TC, summarises main points and draws attention to what the article has shown. Each of these seven types is employed for a variety of discourse purposes; each plays an important role in constructing and managing the interaction between author and reader. The very different results across TC types underscore the importance of distinguishing between different kinds of metacommunicative utterances. <?page no="268"?> 268 As regards text comment frequency across cultures, it emerges that there are statistically significant differences between the cultural groups in SCEGAD only for the explanatory text comment type, which was far more common in the German than the British and US-American texts. The findings of previous, smaller studies, which indicated a much higher level of text commenting in British and/ or US-American-authored than in German-authored texts, therefore appear to be more a reflection of poor study design and author bias than of salient cultural differences. There were no significant inter-varietal differences among the Anglophone scholars in the corpus. The differences between cultures lie not so much in frequency, but in the way in which text comments of different types are used to construct the author-reader relationship, and in the way these comments are realised. In addition, the analysis showed a high degree of disciplinary variation across the corpus texts. The differences between disciplines are not consistently significant for any one TC type. For explanatory TCs, there are no significant interdisciplinary differences at all: here, cultural background seems to override disciplinary membership as a determiner of TC frequency. Philosophy texts contained the highest frequency of text comments and history articles the lowest in all three cultural groups examined in the study. Particularly in the philosophy texts, text comments are used to create overt author-reader interaction. There is a sense of meaning construction as a co-operative dialogue and communal journey through the text. In contrast, the SCEGAD historians prefer less explicit references to the text, and less overt interaction with the reader. These disciplinary tendencies fit well with the results in chapter three concerning person reference. Other variables examined were text position, text length, author gender, age and status and number of authors. As suggested in previous studies, the positions of TCs within the text differed significantly across linguistic and cultural groups. The Anglophone scholars in SCEGAD are more likely to place overarching TCs at the very beginning of the text, while their German-speaking colleagues prefer to place them at the end/ middle of the first section (p<0.05). These trends do not hold across cultures however. Here, it emerges that the sole significant difference between groups is that the German scholars are more likely to place overarching TCs at the end/ middle of the first section than are their US- American colleagues (p<0.05). This is a further indication of the inapplicability of the term ‘Anglo-American’ to discourse norms. Cultural background had no significant influence on the placement of concluding text comments. <?page no="269"?> 269 The age and academic status of the SCEGAD authors, and the number of authors per text, also had no significant effect on text comment frequency. The gender comparison revealed that the female scholars employ more introductory TCs than the males (p<0.05), but not more TCs overall. This may perhaps indicate a concern among the female academics sampled in SCEGAD for maintaining a more constant connection with the reader. Correlation analysis revealed no significant relationship between the length of a text and the frequency of (any type of) the text comments. The stepwise multiple regression analysis showed that the three best predictors of TC frequency are whether the text was from linguistics, from philosophy or by a US-American author. However, the predictive power of the models is not particularly high, suggesting that TC use is motivated less by disciplinary or cultural membership than by discourse purpose. The rhetorical strategies behind text comments were examined by type. Overarching text comments are used to indicate aims and also explain structure. Overarching TCs vary widely in length, depending on the amount of detail authors provide regarding the structure of individual sections. Some of the corpus authors present their texts as a narrative progression of arguments, others as a chronological progression of text segments. Authors also seem to differ in the confidence with which they state their aims. These differences reveal something about the way in which writers view their own work, and their anticipation of the level of initial orientation readers desire. Exclusive text comments are a defensive reaction to anticipated reader expectations. Since the statement that something cannot be done presupposes that readers wish or expect it to be done, the exclusive TCs offer valuable insights into reader expectations and assumptions. While the German authors in the corpus appeared to expect RAs to be exhaustive, detailed and complete in their presentation of material, the Englishspeaking authors examined seemed to place more emphasis on relevance, desiring the exclusion of all material not immediately relevant to the topic at hand. The exclusion of material can also be motivated simply by lack of space. The German scholars in SCEGAD, however, very often give no reason at all: they are significantly more likely than are both their British and US-American colleagues to cite no reason (p<0.05). It is possible that the German authors in the corpus assume their readers to be uninterested in knowing why material was excluded. Alternatively, the German authors might assume that if a topic is of no interest to them, it automatically will not interest the reader. If readers desire no explanation, this might indicate a more trusting author-reader relationship; if authors are unwilling to give explanations, this suggests a more authoritarian author-reader relationship. Which of the two, or what other possibility, applies in this case cannot be <?page no="270"?> 270 decided here. Exclusive TCs are also used for self-promotion, staking claims for later research and for delineating one’s contribution from that of other researchers. Introductory text comments are used to announce material that will follow immediately, comment on current actions or arguments, and issue directives to the reader. While announcements help to prepare the reader for what follows, current comments help to clarify authorial intentions, explaining the aims of specific sections or arguments. Directives steer readers to the actions and thoughts desired by the author, and are often couched as invitations to communal action. The discourse purpose here is to retain contact with readers throughout the text, ensuring that they can follow each step of the way, and understand what the author intends at a specific moment. Introductory TCs are therefore a strategy for increasing reader involvement in the text. They were the least common TC type in the German corpus texts, and the third most common in the British and US- American RAs. Explanatory text comments too create reader engagement, but of a particular type. The expression of authorial textual decisions and requests as obligatory actions creates an author-reader relationship that is more authoritative and less co-operative and egalitarian. Readers are assigned the role of accepting obligations, not invitations (as in introductory text comments). There are extremely significant differences between cultures in the level of responsibility authors accept for textual decisions. The German academics sampled in the corpus were more likely to employ deontic than dynamic modality, presenting themselves as the mere instruments of outside forces, rather than active creators of text. This may indicate an attempt to deflect reader criticism in advance by hiding behind external restrictions. In contrast, the British and US-American scholars analysed were more likely to explain textual decisions in terms of their own and the reader’s needs (UK/ DE and USA/ DE differences significant at the p<0.0001 level). The Anglophone scholars in SCEGAD therefore seemed to create an author-reader relationship that was both more personal and more egalitarian, though at times also more idiosyncratic. These findings are underlined by the SCEGAD German academics’ tendency not to explain why authors or readers must perform certain actions; here too the differences between the German and British/ US- American scholars are extremely significant (p=0.0001). Couching textual progression as a series of unexplained externally-imposed requirements makes for a more impersonal and authoritative tone in the German corpus articles, and may indicate a different conception of the textual process, and the role of both participants within it. The German authors in the corpus enact both themselves and the reader as the instruments of outside require- <?page no="271"?> 271 ments: just as authors note that they must carry out particular actions, so too do they expect the reader to perform the actions required of them. Meaning is therefore constructed within the text as a process of mutual obligation, in which parties fulfil their roles usually without needing to explain (author), or requiring an explanation (reader). This TC type shows the importance not just of noting that authors involve readers, but also of examining how authors involve the reader, and what role they offer the reader in the textual interaction. Ultimately, readers of course have to accept whatever role they are offered, if they wish to read the text, and no feedback will ever reach the author. However, the roles readers are offered differ greatly across the three cultures examined in this study. Prospective text comments are used primarily to aid reader comprehension, helping to prepare the reader for the tasks, arguments and concepts yet to come. Such comments occur where authors judge their text progression to deviate from reader expectations, for example, when the same point is considered in more than one section, or where explanations or detailed analyses occur long after a concept or topic has been introduced. In these cases, references to later material help to deflect reader criticism. Through prospective TCs, authors explain the relationship between what has been done already, and what will be done later. This assists readers in grasping the text as a coherent unit, and making connections between its parts. Prospective TCs can also serve persuasive functions, suggesting to readers what they will conclude upon seeing data not yet presented. Another way in which authors help readers grasp the intricate structure of text is by referring to information already provided. Retrospective text comments are the most frequent TC type across all three cultures examined here, and play possibly the most important role in ensuring reader comprehension. The ability to connect new information with what is already known appears to play a cardinal role in effective comprehension and learning; retrospective TCs indicate academic authors’ most likely subconscious attempt to engage this ability in the creation of meaning. Concluding text comments usually occur at the end of the text. They describe the achievements of the text, as judged by the author, and explicitly remind readers of the opinions authors expect them to hold in light of the information and arguments presented. In this way, concluding TCs strengthen the knowledge claims made in the article and help to convince readers of an author’s point of view. The SCEGAD authors vary widely in the degree of agency they accord themselves in summarising textual achievements (‘I have shown’ / ‘the article has shown’ / ‘it has been shown’), and in the degree of confidence with which they advance <?page no="272"?> 272 claims. Concluding TCs constitute the vital argumentative link between what has been presented, and what an author considers the presentation to have shown. Each of the text comment types discerned in the analysis is used for a different combination of discourse purposes. Some demonstrate clear cultural and disciplinary preferences in frequency and use, while others are used in essentially the same fashion and with essentially the same frequency across all three cultures and five disciplines examined here. The fact that there was no consistent correlation between the use of all TC types underlines the fact that authors distinguish between types, using specific sorts of text comments for specific rhetorical purposes. This suggests that future studies would do well to distinguish clearly between different types of metadiscourse and consider them from a qualitative as well as quantitative perspective, as was done here. There is no single discourse strategy underlying all forms of metadiscourse, as some studies have simplistically suggested. Text comments are not only about ensuring reader comprehension, but about something wider by far: the interaction between author and reader. <?page no="273"?> 273 …numbers alone give little insight about language. _______________________________ (C ONRAD 2002: 78) …the acquisition of unfamiliar cultural and linguistic rules allows no instant solutions. _______________________________ (H YLAND 1995: 39) Chapter 5 Implications and applications of the study This study set out to examine how scholars from different cultures, writing in different disciplines, use their native language in research articles. It has shown that there are empirically verifiable, statistically significant differences in the ways British, US-American and German scholars construct and express various identities within a defined social group, how they interact with their imagined readers and the wider academic community, and how they explicitly refer to the structure of their texts. The high degree of internal variation within cultural groups supports the argument that culture is neither monolithic nor deterministic (Scollon & Scollon 2001: 167-176). There is more than one British, one US-American and one German way of achieving the same discourse purposes, and many academics enjoy a considerable amount of individual freedom in their selfpresentation and their interaction with the imagined reader. Since the SCEGAD corpus is balanced for many extra-linguistic variables, the analysis has also shown that discipline, author age, status and gender exercise a significant influence on specific aspects of language use. Scholarly writing is not the disconnected, faceless transmission of universally-valid facts, but rather the personal, dialogic construction of knowledge through socially-situated author and imagined reader interaction. Therefore, <?page no="274"?> 274 academic discourse can best be analysed with respect to its social, cultural, institutional and disciplinary context. This final chapter begins with a critical summary of the findings (section one). Then follows an examination of the place of culture in contrastive studies, and the difficulty of analysing cultures through cultural products (section one). In the third section, I consider the implications of research such as mine for EAP teaching and learning. This leads on to section four, in which I briefly discuss approaches to EAP pedagogy in the context of English-language dominance. In section five, I make some recommendations for reducing the disadvantages currently faced by NNS scholars. Section six places the study within the context of wider implications for the situation of scholarly languages other than English. The final segment (section seven) suggests some directions for future research. 5.1 Summary of findings The findings of this study are two-fold. For person reference, substantial differences were found in both the frequency and discourse purposes across cultures. Text comments, on the other hand, were found to demonstrate fewer cultural differences than previously thought (Clyne 1987; Hutz 1997; Baumann 1998). These joint results affirm that empirical studies of academic discourse can have surprising findings, both confirming and undermining conceptions of cultural tendencies. As regards disciplinary differences, there was evidence both of considerable variation within disciplines and of significant discipline-specific trends which held across cultural groups. The results thus confirm the findings of various studies that discipline, as well as culture, exercises a substantial influence on scholarly writing practices (see for example Russell 1991; Taylor & Chen 2001; Oldenburg 1992; MacDonald 1994; Hutz 1997; Busch-Lauer 2000; Hyland 2000; Fandych & Graefen 2002 and Dahl 2004). Furthermore, this investigation demonstrated that it is possible to plot a middle path between lexico-grammatical and discourse features, between quantitative and qualitative analyses, and between formalist and functional approaches. The study was concerned with two features, one more traditional and formally-defined, and one more discoursal and contextdefined; both features, person reference and text comments, were analysed using a combination of corpus-linguistic and discourse-analytic methods. This investigation showed the benefits of analysing a formal feature (also) in a discourse fashion, and a discourse feature (also) in a frequencyorientated fashion. In the case of person reference, traditional grammatical descriptions focus almost exclusively on formal and paradigmatic aspects, <?page no="275"?> 275 and thereby ignore functional, pragmatic and sociolinguistic characteristics. As Wales notes, this means that many of the most interesting uses and features of person reference are either disregarded completely or confined to footnotes (1996: 8). I hope to have shown that some of these pragmatic ‘footnotes’ are worthy of the more detailed examination granted them here. Textual metadiscourse is, as the name suggests, a discourse phenomenon, and traditional discourse analysis has seldom been interested in calculating frequencies, or analysing larger collections of text. Previous studies of textcomments across cultures offered some quantitative results, yet these often lacked empirical support, and were derived from unsound study designs. In consistently applying corpus-linguistic methodology to a discourse phenomenon, I have shown that stepwise multiple regression and pragmatics can mix. In addition, by analysing each phenomenon (also) with methods usually reserved for the other object of study, I have provided evidence that such methodological ‘cross-fertilisation’ can provide a fuller picture of language use in specific contexts. In other words, corpus and discourse approaches can learn from and with each other. The analysis in chapter three showed that person reference is an important device for the construction of authorial, imagined reader and group identity, as well as for the management of social interaction in academic prose. From self-advertisement to reader address and group membership claims, person reference was used by the expert writers sampled in the corpus for a large variety of discourse purposes. These purposes varied widely across cultures, languages, disciplines, genders and academic status levels, though not across age groups. The British and US- American scholars were more likely than were German scholars to refer to themselves and address the imagined reader. The German scholars in SCEGAD preferred to express personal opinions with first person plural forms, and largely refrained from both authorial self-reference and imagined reader address. While the English-speaking academics in the corpus more frequently presented themselves as individuals rather than members of a group, the German scholars were far more likely to express group rather than personal identity. The disciplinary trends were largely consistent across cultural groups, with person reference the most frequent in philosophy texts and the least frequent in history research articles. Bearing in mind that the SCEGAD texts were chosen for their assumed disciplinary proximity, the high level of interdisciplinary variation in the corpus was a surprising finding. While the British and US-American philosophers used person reference mainly to express personal opinions (first person singular pronouns) and actively engage with the imagined reader (inclusive ‘we’), the German philosophers in the corpus employed person reference largely to express general statements about the nature of human <?page no="276"?> 276 perception and thought (human ‘we’). The cultural trends therefore remain. Across all three cultural groups, the male academics represented in the corpus employed more person reference than did their female colleagues. The most pronounced gender differences were found among the British academics. This trend is in exact contrast to often-aired conceptions that women’s language is more involved and ‘personal’ than men’s (Lakoff 1975; Maltz & Borker 1982; Holmes 1993; Argamon et al. 2003), and therefore calls for further research into both posited gender universals and gender marking in academic discourse. Finally, authors of higher academic status employed all types of personal pronouns, though not third-person reference, more frequently than did lower status scholars. The general picture is that there is still a large discrepancy between the innately personal nature of academic activity, and the outward reflection of subjectivity through language. At present, it seems that British and US-American humanities academics in SCEGAD are more readily prepared to emerge from the shadow of conventionalised impersonal language than their German colleagues are. At the heart of these findings lies the problematic role of the academic author as a gatherer and conveyor of knowledge. The recognition that scholarly observation and writing are subjective and personal entails a profound conflict for those researchers and academic readers who regard scholarly activity as the search for and presentation of an objective truth outside of ourselves. As a US-American humanities academic comments, “for some of us, I think, using the personal in academic scholarship provokes a profound sense of unease” (Holdstein in Holdstein & Bleich 2001: 7). For many researchers, the avoidance of person reference, in particular of the scholarly ‘I’, is a way of expunging themselves from the process of knowledge construction, in the hope of indicating or perhaps ensuring a sort of objectivity no human can ever possess. There is something of a paradox in the attempt to create what cannot be created by avoiding what cannot be avoided. The grammatical elimination of the author from the text through the avoidance of authorial self-reference, and the minimisation of other explicitly interactive forms of person reference, is sometimes a conscious, almost duplicitous strategy, and at others a subconscious routine. Some academics are tired of what they consider to be ‘beating around the bush’. In their 2001 volume Personal Effects: The Social Character of Scholarly Writing, Holdstein and Bleich expound the view that academics’ personal and social circumstances are not only intimately connected to, but in fact part of their subject matter (2001: 4-5). Although they had previously attempted to eliminate themselves from their scholarly writing through the absence of the “personal ‘I-voice’” (2001: 7), Holdstein and Bleich now explain that they are no longer prepared to adhere to this <?page no="277"?> 277 traditional writing convention. They describe the conventional ways of reducing the human presence in the humanities, through avoidance of selfreference and self-reflection, as “coercive” and “arrogant”, and urge humanities academics to recognise that their own lived experiences are intimately entwined with their subject matter (2001: 2, 12). The authors therefore plead for a radically different view of scholarship, academic research and writing in the humanities. It is questionable whether such proposals will catch on; there may be too much historical, socio-cultural and institutional baggage to cast away. Nevertheless, they do provide a fascinating insight into the tension some academics perceive between their professional and personal identity, and the difficulty scholars can feel in allowing for and reconciling both identities in their research writing. The analysis in chapter four revealed that text comments are central devices for the construction and management of the author/ imagined reader relationship in humanities research writing. Text comments are used not only or even primarily to ensure reader comprehension, as previous studies have suggested, but also for a variety of rhetorical purposes, from imagined reader engagement to the anticipation of imagined reader criticism. The fundamental characteristic of text comments, however, is their role in structuring the relationship between the author (persona), the imagined reader, the actual reader and the text. Despite the suggestions of previous studies, text comment frequency was not found to be higher in either the British or US-American corpus texts. On the contrary, the German academics represented in SCEGAD in fact employed text comments significantly more often than both groups of English-speaking scholars did. These results are primarily due to the influence of one TC type in particular, explanatory text comments; here, the difference between cultures is extremely significant (p<0.0001), with the German corpus scholars employing this type of TC far more often than their British and US-American colleagues. Explanatory TCs are used to explain and justify textual decisions in the face of anticipated reader expectations. In some cases, the only reason given is that a particular action is necessary, in which case the character of these devices is less explanatory and more obligatory. The German scholars in SCEGAD were significantly more likely than were either the British or US-American scholars to employ deontic rather than dynamic modality, and to cite no reasons for textual decisions in their explanatory text comments. In this sense, one could say that explanatory TCs were explanatory in the English-language corpus texts, and obligatory in the German ones. These surprising findings call for more research. As for person reference, there was a high degree of crossdisciplinary variation in text comment frequency, with the philosophers the most likely to comment on their text, and the historians the least likely. <?page no="278"?> 278 The cross-disciplinary trends were largely consistent across cultures, languages and examined phenomena. Again as for person reference, author age did not emerge as a significant influence on frequency. The general picture is that readers from different cultural groups, linguistic backgrounds, disciplines, genders and status levels employ (different types of) text comments in order to meet the anticipated expectations of the imagined reader, and in order explicitly to enact and manage social interaction in their texts. The notion that readability is directly proportionate to the number of text comments was contested by arguing that readability is in fact a function of the congruence, or lack of congruence, between reader expectations of texts and actual textual patterns. Obviously, there is more to social interaction than person reference and text comments. However, the present investigation never aimed to encompass all aspects of interaction in academic discourse; it would have been folly to attempt and vain to suggest this. The aim was to investigate some phenomena which encode social interaction in academic texts, primarily from a cross-cultural perspective, and to do so in an empiricallysound fashion. The design of the corpus permitted a number of observations to be made about the influence of other variables in addition to culture, such as discipline, gender and academic status. The premise of the study is that words are more than words, or, as Bazerman puts it, “writing matters”, since “the word draws on and ties together writers, readers, prior texts, and experienced reality to constitute the domain symbolic knowledge” (1988: 18). Bazerman of course is a social-constructivist, and, as explained in the second chapter, it is this theory which informs the present study. The same can be said of Hyland, whose work stems from the perception of writing as a “culturally-situated social activity” (2000: 110). According to such a framework, not only are words not just words, but knowledge is not just knowledge; it is “derived by someone and told to someone. It is knowledge that this person found in this community or society and is sharing with this other group of people [original emphasis]” (Holdstein & Bleich 2001: 2). A belief in the personal nature of knowledge, and its socially-mediated and socially-constructed characteristics, has provided the foundation for this study. As I argued in the second and third chapters, the general idea is that language as used by academics is not atypical, devoid of personal content, purely rational, completely objective and so on. Academic communication in fact abides by much the same mechanisms as everyday communication does (Schröder 1998: 183). Like diaries, recipe books and wills, academic discourse is intimately personal by virtue of its human origin. When Schröder states that he assumes that science is not just content (1998: 184), this realisation <?page no="279"?> 279 must be the starting point for all studies of academic discourse, and all teaching of the same. This has been the ethos behind this study. 5.2 Anglo-Saxon empiricism and Teutonic theorisation: beyond the stereotypes In chapters two and four, I quoted some reactions from German scholars to Clyne’s (1987) seminal study of differences between Englishand German-language academic texts. Part of the problem, certainly, lay in Clyne’s tendency to express his findings in terms of value judgements, according to which German academic style was inferior or somehow deficient. However, another part of the problem lay in Clyne’s establishment of a dichotomy between content/ author-orientation and reader-orientation (Clyne 1991, 1993). As discussed in chapter two, a recurring problem in studies of academic discourse is the large gap between the isolated phenomena analysed and the cultural tendencies they are taken to represent. Clyne’s terms offered an elegant solution to this problem. However, the dichotomy he suggested is mistaken, and has had unfortunate and misleading consequences, as numerous researchers have attempted subsequently to squeeze collections of phenomena into this artificial overall framework (see section 2.2). The whole idea of reader-orientation as opposed to some other type of orientation is dubious, since it suggests that for German academics, for example, communication is not participantoriented. The fact is that authors from different cultural backgrounds assume audiences with different preferences. Indeed, it appears that readers from different cultural backgrounds do have different preferences when it comes to discourse structure, as Clyne’s own work showed (Clyne, Hoeks & Kreutz 1988; Clyne 1991). Most authors, regardless of cultural background, try to cater to the assumed discourse expectations of their readership. The difference lies in the kind of imagined readers towards which academic authors orient their writing. The suggestion that this orientation towards the imagined reader is characteristic of only one culture, and that it is mutually exclusive with a concentration on content or authorial identity, is absolutely untrue and even prerogative. It is this treatment of two cultures as if they were polar opposites that is the foundation of ideological statements, or stereotyping; human behaviour is assumed to be fully determined by cultural membership, and a limited view of one or two dimensions of behaviour is regarded as the whole picture (Scollon & Scollon 2001: 168-170). However, content versus reader orientation is not the whole picture of language use in the academic or in any other domain. <?page no="280"?> 280 Much the same comments apply to the other one-word descriptions of Englishand German-language academic discourse listed in chapter two: researchers have suggested that German scholars are not interested in writing well-ordered texts (Galtung 1981; Clyne 1987), or that Englishspeaking academics are more concerned with relationship-building than expressing content (Edmondson, House, Kaspar & Stemmer 1984; Kotthoff 1989). 101 Another much-quoted binary contrast, that between writer/ speakerand reader/ listener-responsible cultures, also appears not to hold. The origin of this notion was Hinds’ (1987) influential paper on discourse structure in English and Japanese, in which he stated that Japanese texts require greater processing effort on the part of the reader. The categorisation was later adopted by studies of totally unrelated cultures, as if it were a statement of universal cultural truths. 102 Thus, German is apparently reader-responsible (Clyne 1991; Baumann 1998), as are Czech (Cmerjková 1996), Fârsi (Moradian 1999), Finnish (Mauranen 1993a), French (Dahl 2004), Polish (Duszak 1994) and Spanish (Álvarez 2005). Norwegian on the other hand is writer-responsible (Dahl 2004). In later work, Hinds included Chinese, Korean and Thai writers alongside Japanese writers in the reader-responsible category (1990). None of this work is convincing, because the categorisation is a gross oversimplification and subjective distortion of cultural differences in writing styles (as Kirkpatrick 2004a shows for Chinese). It is not clear why work on Japanese texts should be automatically valid and applicable to work on texts from all other cultures. Part of the problem here lies in approaching texts with certain preconceived notions about how cultures differ. Hinds’ work on Japanese texts may have led him to formulate a theory, but it would be unwise to use this theory as the starting point for other studies of completely different cultures. As I noted above, the findings of this study both confirm and undermine notions of cultural differences. The hypothesis that authors from different cultural backgrounds use language differently was therefore accepted in some cases and rejected in others. Such is the nature of scientific enquiry. It is therefore concerning that much previous research starts from the assumption or even the conviction that there are readilyidentifiable cultural types, and begins the analysis with a fixed theoretical framework into which all results are forced. Research should begin with a falsifiable hypothesis - of course informed by previous research and intuition - but not with a foregone conclusion that (a) cultures differ 101 The last set of epithets reminds me of those applied to women’s language. Here, too, they do not appear to be accurate (see chapter three, section 4.5) 102 This is, no doubt, much as Hinds intended it: the 1987 article is subtitled “a new typology [of language]”. <?page no="281"?> 281 dramatically and (b) they differ according to the attribution of responsibility for encoding utterances. Aside from being an oversimplification not necessarily relevant to the study of other cultures, the theory of writerand reader-responsible cultures is in fact not valid, even for Japanese, the culture upon which it was originally based. McCagg (1996) reanalyses the data upon which Hinds based the theory, and shows that the assumptions Hinds makes are false. According to McCagg, greater processing effort is required only for readers who are not familiar with Japanese discourse conventions: as long as writer and reader share the same set of cultural beliefs, life experiences, as well as similar conceptual and linguistic abilities, comprehension of Japanese messages in general does not require any greater cognitive effort on the part of the reader than understanding of English messages does. (McCagg 1996: 239) Essentially, what Hinds (1987) established was that he personally found the single Japanese text he studied harder to understand than its Englishlanguage counterpart. He had by no means discovered a new typology of language. The truth of the matter is that Hind’s categorisation was a culturallybiased representation of differences in discourse preferences. So too are the studies which adopted his framework. As Kubota states, in Hinds’ study legitimate differences were “oversimplified and overgeneralised, resulting in a static, exotic, and monolithic image of Japanese written language divorced completely from English” (1998: 2). For Chinese, Kirkpatrick (2004a) has noted that the distinctions proposed in Hinds (1990) do not hold either: an examination of Chinese-language rhetoric handbooks showed that Chinese writers are just as concerned as English-speakers are to write in a clear fashion comprehensible to readers (see also criticism in Liu 1996). The findings in this study likewise do not support the idea that German (academic) writing is reader-responsible. The whole idea should be surrendered, and replaced with a focus on how authors from different cultures orient themselves towards their target audience, and how NNS of whatever language may be helped to do this more effectively in the target culture. When it comes to responsibility, Matalene comments, as researchers “our responsibility is surely to try to understand and appreciate, to admit the relativity of our own logic, and to realise that logics different from our own are not necessarily illogical” (1985: 807). Researchers must take time and care to devise methodologicallysound study designs before beginning their investigation, and must resist manipulating the results by forcing them to conform to pre-existing taxonomies, especially those fraught with cultural bias. Tendencies <?page no="282"?> 282 certainly exist, and empirical studies should attempt to conceptualise, operationalise and quantify them: in chapter three, for example, we saw that the German scholars employed less person reference than did their British and US-American colleagues, making their texts less explicitly personal. However, as I was careful to note, this does not mean that German academic style as a whole is less personal, as the frameworks of Duszak (1997) and Busch-Lauer (2001) propose, or even that the German RAs in SCEGAD are impersonal. Identifying tendencies is useful; enforcing rigid dichotomies is not. The same is true of those studies, such as Kaplan (1966) and Galtung (1981), which represent cultural styles as line drawings. There is more to scholarly cultures than can be expressed in this fashion, as Kaplan himself later admitted (2000: 83). The central point is that many of these judgements are relative: if the preferences of one culture are judged from the perspective of another (usually the researcher’s native culture), the ‘other’ culture is all too easily represented as a deviation from the ‘norm’, because it does not fulfil the researcher’s own (culturally-biased) expectations. 103 In fact, the other culture is not a deviation from the norm at all; it simply follows a different norm. For Dillon, linguistic expectations are codified as a set of patterns which we bring to texts and apply to their interpretation. Such patterns, Dillon notes, “form the basis of communication insofar as writers (and speakers) assume that they can and will be applied by readers (and hearers)” (1981: 52). In communication between cultures, and in analysing a foreign culture, these expectations are quite often not met. Unfortunately, when confronted by patterns unknown to us, our first reaction is nearly always judgemental and often negative; it is certainly culturally-biased. When judged from within the culture of origin, however, the question of what constitutes a truly difficult text for example is probably quite different. One might just as well suggest that Japanese texts are exactly right, and English-speaking writers talk down to their readers. This might lead to a theory about how Japanese writers are democratic, treating their readers as equals, while English-language discourse conventions dictate the hierarchical presentation of knowledge, based on condescending assumptions of writer superiority. Normality and difficulty are relative concepts. The same can be said of Clyne’s ideas of linearity and digressiveness, on which Graefen remarks that the concept of a ‘digression’ is not a characteristic of a particular text at all, but rather a description of the inability of the researcher to understand its structure (1994: 148). A German 103 Of course, what all of this also shows is the considerable influence of discourse preferences: even linguists are quick to judge those texts negatively which do not meet their expectations. <?page no="283"?> 283 reader would most likely not judge as digressive the same texts Clyne describes as badly-planned, disorganised, difficult to read and awkward (1987). Again, these judgments are culturally relative: correct interpretations of discourse structures can only be made by researchers who are capable of understanding texts in their original context, or at least make some reference to how such features might be regarded in the target discourse community. Some rhetoricians know this: a classic US-American college style guide for example notes that “a work is considered coherent when the sequence of its parts […] is controlled by some principle which is meaningful to the reader” (Hughes & Duhamel 1962: 20). The crucial phrase here is “to the reader”: we can only judge rhetorical meaning and effect in context. If linguistic analysis is divorced from socio-cultural context, the results may not be valid. In addition, as Kubota (1998) notes, the mere fact that a particular discourse pattern occurs more often in culture x does not necessarily mean that it is preferred: to assume this confuses frequency with import. A tendency in previous research to focus upon quantitative analyses, rather than examine discourse purpose and meaning in context, has perhaps not been the most helpful approach. Bicultural researchers, or mixed-culture research teams, would be necessary here, but even a monocultural researcher working alone could consult foreign informants about the effect specific patterns might have in their culture. This would lead to more accurate and appropriate descriptions of discourse conventions and preferences across cultures. Cross-cultural analyses which have adopted such methods have concluded that ideas of good writing are in fact very similar across cultures (see Kubota 1998 on English and Japanese, and Scollon & Scollon 1997, Zhu 1997 and Kirkpatrick 2004a on English and Chinese) The preoccupation with categorical differences and clear categorisations is therefore more a comment on researchers themselves than a reflection of actual language use across cultures. If the frameworks already proposed are unsatisfactory, we must ask what could or should take their place. Despite the enduring appeal of simple answers, this study shows that, in studying something as complex as culture, simple answers seldom exist. Researchers fixated on dichotomies miss the point: taking the position that members of a cultural group write in a prescribed, monolithic manner tends to discount or overlook complex factors like the heterogeneity of writers, multiplicity of writing experiences within a cultural group, and their notions of audience. (Zainuddin & Moore 2003: 1) Any serious empirical attempt at investigating the influence of culture on language use will produce tendencies and strategies, not black and white <?page no="284"?> 284 answers. Fixed classifications are almost dangerously reductionistic; dangerous in that they risk perpetuating prejudices. When scholars attempt to reduce their findings to a neat set of adjectives, or a striking diagram, the appropriate reaction is caution and scepticism. As Scollon notes, alluding to Kaplan’s (1966) ‘doodles’: A very broad range of studies have shown that no language or culture can be reduced to one or two diagrammatic structures that might be applied across the board from internal cognitive schema to paragraph structure, whether these might fly under the flags of circular, direct, indirect, zigzag, inductive, or deductive. (1997: 353) Another instance of the attempt to simplify what is in fact far more complex is the reference in many contrastive studies to ‘Anglo-Saxon’ or ‘Anglo-American’ discourse norms. As noted in chapter two, this term was introduced by Galtung (1981) and continues to be used widely (Mauranen 1993b; Baumann 1998; Zhabotynska 2001; Fandrych & Graefen 2002; Giannoni 2002; Bassler 2003). The differences between British and US- American scholars discerned in this study show that this term is a fiction. Had the SCEGAD corpus sampled work by other groups of L1 Englishspeaking scholars also, such as Scots, Canadians and Australians (not to mention Bermudans, Indians and Namibians), I anticipate that this point could have been demonstrated even more strongly. Future studies should therefore consider the possibility that cultural differences exist within language groups, not just across them. In this way, we might hope to capture some of the cross-cultural diversity of language use within a specific communicative context; diversity that almost certainly exists in academic contexts just as it does in other domains (see for example collected papers in Hellinger & Ammon 1996; Spencer-Oatley 2000 and Luzio, Gunther & Orletti 2001). If there are differences across cultures in the packaging of information, the question arises of where these differences come from. Clyne (1987) suggested that cultural preferences are largely transmitted through the education system, and observed that there is more emphasis on teaching essay-writing form in the USA, England and Australia than there is in Germany. This led Clyne to believe that in English-speaking countries, the form of a text is regarded as more important than its content: students are required to demonstrate their mastery of formal rules, rather than their knowledge of content (1987: 92). Without comparative studies of NS writing instruction in different countries, it is difficult to assess the accuracy of these statements. My own non-representative survey of German colleagues in fact suggests that great emphasis is placed on learning the formal rules of argumentative text types, known as <?page no="285"?> 285 Schulaufsätze [school essays], such as the Erörterung [argumentative text]. The Erörterung has two subtypes, the textgebundene Erörterung [textbound essay] und the freie Erörterung [free essay] or Gesinnungsaufsatz [opinion essay], and these types are practised extensively in German schools. For my own part, I remember finding the formal rules of the Erörterung structure, which I was taught as part of my university degree in German, far more constraining and rigid than anything I was ever taught in an Australian school about essay writing. There is certainly a need for empirical research here. A more basic question concerns the relationship between NS learner text types, such as the student essay and the Erörterung, and professional scholarly writing. Clyne (1987) seems to assume that the conventions of the school text types are immediately transferable to the research article and Forschungsartikel [RA]. Though it is entirely possible that adult scholars are influenced by what they learnt at school, the text types are still generally quite different, and should be treated as such. 104 Graefen (1994) argues that since - according to Clyne - cultural differences in RA language use are really due to differences in pedagogical institutions, these differences in fact have nothing to do with either culture or academic writing at all. English-language RAs are in fact student essays (Graefen 1994: 151). Apart from the fact that it is wrong to assume that all differences in language use are attributable to differences in the education system, the argument here is exactly circular. If texts differ across cultures only because school systems are different, and school systems are different because cultures are different, then different cultures lead to differences in texts. Schools therefore condition text-production; text reception conditions cultures; cultures condition both text production and school systems. Put in another way, the argument is a sort of chicken-and-egg situation: do school systems influence culture, or do cultural systems influence schooling? I would suggest that the one is not distinct from the other. Texts are cultural products which can be understood in terms of their socio-cultural context. It must be remembered, however, that the norms which are taught at school form only a small part of this socio-cultural context; a much larger part concerns the role of academics within a society, 104 Another point worth considering is that, even if some differences in language use across cultures are attributable to different education systems, these systems differ widely between countries that share a national language (compare the USA with the UK, Singapore or Uganda, or Germany with Switzerland). Assuming that academic writing norms are at least partly influenced by explicitly taught writing skills and guidelines, it would therefore be a mistake to treat authors from these countries as a single group. In addition, in a federalist country such as Germany, where educational policy is the responsibility of the states, there are major differences between education systems even within a single country. <?page no="286"?> 286 as both producers and receivers of scholarly texts, and the role and appropriate use of language in conveying scholarly findings. Both synchronic and diachronic analyses of these contexts are important. Differences in academic discourse across cultures, that is, cannot be reduced to the formal, secondary socialisation that takes place within the educational system. Some other aspects of culture which may be significant in a crosscultural context are the ideology (history and worldview), primary (informal) socialisation, and the systems of social organisation within a group (Scollon & Scollon 2001: 140-1). Though it is true that membership of the academic discourse community is gained in large part through formal instruction, institutionalised learning and socialisation (Scollon & Scollon 2001: 149), nonetheless the social context of scholarly texts can and should not be reduced to the education system. Duszak believes that cross-cultural differences in academic communication styles are at least in part attributable to “differences in underlying social and cultural values” (1997: 17). The problem here, of course, is that it is incredibly difficult to quantify such differences without descending into reductionistic and possibly offensive stereotypes. In reality culture can be studied only through its products - texts, language use, customs, food and so on - and it is up to a researcher to interpret what if anything these products reveal about a culture. Ultimately, there are three main problems with previous attempts to uncover the reasons for cultural differences. First, in their search for a finite set of characteristics or features that encapsulate a particular culture, such discussions risk essentialising culture. Second, just as this study showed that academics possess a wide range of different identities, so too are they members of a variety of cultures and subcultures (cultural, national, linguistic, disciplinary, even subdiscplinary). Academics, as Myers puts it, are “cultural hybrids” (2004). The attempt to reduce all differences between texts to the influence of the writers’ native cultures is therefore reductionistic. Third, in such conceptions, influence is often presented as deterministic and unidirectional - culture influences textual patterns - which is an oversimplification, and denies the human and cultural agency of the individual. As Kubota puts it, talking about novice writers, “students are both cultural products and cultural agents” (2000; see also Leki 1997). This point is very clear when it comes to academic discourse, since scholars are both writers and readers, producers and receivers, of culturally specific norms. Though their cultural background clearly influences their textual production, as this study has shown, it is likely that their textual choices also influence the culture of which they are members. Communities shape discourse, and discourse, communities. The relationship between the two can be studied and interpreted by examining cultural products, but we will never be able to classify definitively or <?page no="287"?> 287 quantify it for all time. Perhaps this explains some of the fascination that cross-cultural studies hold. 5.3 Implications for EAP teaching and learning The results of this study have implications for a variety of different domains, from specialist translation to the teaching of academic writing to NS and NNS from the undergraduate to professional level. 105 Since this study has been concerned with professional language users rather than students, I wish to focus here on advanced-level academic writing tuition for professional researchers, who are NNS of English. There is a great deal of excellent research on EAP teaching for both NS and non-professional NNS, to which I refer the interested reader (Cowie & Heaton 1977; Robinson 1988; Swales 1990; Huckin & Olsen 1991; MacDonald 1994; Swales & Feak 1994, 2000; Johns 1997; Jordan 1997; Leki 1998; Thurstun & Candlin 1998; Thompson 1998, 2000b; Hyland 2000; Schleppegrell & Colombi 2002; Upton 2004). 106 One premise of this study has been the dominant position of English as the language of international scientific communication. The study was concerned with disciplines in which there is still a tradition of high-quality research production in the German-language and not just in English. The fact is, however, that even for these few disciplines, the proportion of work being undertaken in German is still relatively insignificant on an international scale. Based on the largest-available discipline-specific databases, Ammon conducted research on the language of scientific publications. 107 In philosophy, one of the posited German ‘niche disciplines’, the percentage of German-language publications in 1995 was 3.2%. The percentage of English-language publications in the same year was 85.5% (1998: 166). In history, another native-language holdout discipline, the percentages in 1995 were 5.3% for German-language and 78.0% for English-language 105 On the implications of cross-cultural and corpus-based approaches for specialist translation, see Enkvist (1991), Baker (1993, 1995), Gerzymisch-Arbogast (1993, 1997a, 1997b), Granger, Lerot and Petch-Tyson (1993), Wright and Wright (1993), House (1997, 1999), Stolze and Deppert (1998), Olohan (2004) and Schreiber (2004). 106 Certainly, this investigation also has implications for specialist Deutsch als Fremdsprache [German as a foreign language] tuition for academics using German as a scholarly language. However, since the potential audience for such tuition is extremely limited, I shall not discuss these implications separately here. Though I know of no precise figures, the number of non German-speaking academics who publish in German can be nothing like the number of NNS, Germans and others, who currently publish in English. 107 The figures quoted here are based on The Philosopher’s Index and Historical Abstracts on Disc. See Ammon (1998: 165-6). <?page no="288"?> 288 publications (1998: 165). And these are the disciplines in which there is still comparatively much publication in German. Even in the humanities, there is no escape from English for scholars who wish to reach a wide and international audience. Academics from EFL countries realise that they must publish (also) in English in order to advance their career: Ammon writes that German scholars have “letztlich kaum eine andere Wahl, wenn sie international zur Kenntniss genommen werden wollen” [ultimately no other choice, if they want to be noticed internationally] (1993: 21). Some scholars become aware of this reality even as undergraduates, particularly in those disciplines where the majority of the reading material is already in English. Others, working in disciplines where there is still some level of specialist publication in their native language, may notice the situation only at the post-doctoral level, when they begin to plan an academic career. At present, there is little preparation for this state of affairs. In light of the fact that, as the analyses cited above show, the dominance of English is a relatively recent phenomenon, and is still increasing (see chapter one), it cannot be expected that academics effect such a dramatic change without any help. The kind of assistance NNS academics require depends very much on their individual needs and of course their pre-existing knowledge of general and academic English, not to mention their general aptitude for written expression, which is an individual matter in both the native and foreign tongue(s). Some scholars, in particular those from disciplines where English is especially dominant, such as linguistics and sociology (not to mention of course the physical and natural sciences), may need relatively little help in order to publish successfully in English. Years of reading English-language publications will have furnished them with the specialist vocabulary necessary for their field, and have given them some familiarity with the accepted format of publications in their discipline. If these academics have also enjoyed extensive English-language tuition, at either school or university, probably only copy-editing will be required. 108 Other academics will need more help: lack of familiarity with the language of their discipline, and/ or a lack of sufficient tuition in English, might make it more practical for their work to be written in the mother tongue 108 In Germany, the level of general English proficiency is also partly a function of age, since younger academics have generally received more tuition, from an earlier age, than their older colleagues have. Younger academics have also not had to acclimatise laboriously to the changed scholarly status of their native tongue, since they have probably never known any situation other than English academic language predominance. <?page no="289"?> 289 and then translated. 109 However, whatever their proficiency level, NNS scholars who publish in English could and should also be assisted by (different kinds of) EAP writing courses. In order for such courses to be effective, however, scholars already require a high level of general English-language proficiency. In cases where this does not already exist, non-specialist English courses will have to be provided first. There is sometimes a distinction drawn between English for general and English for specific academic purposes (EGAP and ESAP; see Blue 1988; Jordan 1997). EGAP is essentially study skills, while ESAP is subject-specific academic writing knowledge. In all likelihood, NNS academics already possess both study skills and specialist writing skills in their own culture, language and discipline. However, these skills are not necessarily transferable to English. The problem with the EGAP/ ESAP distinction, from a cross-cultural point of view, is that the idea of EGAP as a ‘common-core’ (Coffey 1984) of academic writing knowledge suggests that this core is constant across and within cultures. However, many of the features Jordan (1997: 4-8) for example lists as ‘study skills’, such as giving oral presentations, summarising, and writing in an academic style organised ‘appropriately’, have been shown to differ across cultures (Clyne 1987; Taylor & Chen 1991; Sachtleber 1993; Mauranen 1993b; Seidlhofer 1995; Busch-Lauer 2001; Zhabotynska 2001; Ventola, Shalom & Thompson 2002; Bassler 2003; Peterlin 2005). Since the idea of a ‘common core’ is a fallacy, it therefore cannot be assumed that EAP teaching for any target group can begin at the ESAP level. Conversely, many NNS academics may not need extensive ESAP teaching anyway - they are already part of a specialised, and probably international, discourse community. Extensive reception of English-language specialist texts within their discipline will most likely make them more of an expert on economics or folklore discourse than any generalist EAP teacher could be. The distinction between EGAP and ESAP therefore seems rather questionable. Far more appropriate is a needs analysis, based on what exactly the students know, and how far this knowledge is applicable to the target discourse community. There will be no standard answers to such a survey. EAP teaching therefore needs to be specific, not only in that it conveys language as used in specific contexts, but specific also in that it is tailored to the current needs of a particular group of learners. Not only English for specific purposes (ESP), therefore, but also English for specific needs. The rough size and nature of the distance between native and target 109 Mühleisen argues that translation is essential if NNS are to have equal access to academic domains, and calls for professional translators and editors at universities and international publishing houses to ensure that non-native speakers are not at a disadvantage at the expense of their scientific work (2003: 117). <?page no="290"?> 290 language discourse norms depends on which are the native and target cultures, languages and disciplines. The differences will, however, pervade all levels of language use, from the banal and formal, such as formatting citations and footnotes, to the situated and subjective, such as establishing an appropriate relationship with the imagined reader. EAP teachers will need to account for this, and adjust their materials, methods and style accordingly. With regard to the specific situation of NNS professional users of academic English, I wish to make three main points. The first is that NNS, in particular practising academics, have different needs, and therefore require a different pedagogical approach, from both NS learners and also from NNS novices. As we shall see, much previous work in EAP, and many previous teaching recommendations, do not apply to this group of learners. While it is true that there are areas all groups find difficult - I mentioned in chapter three that issues of opinion is one of these areas - even the same content will have to be taught in a different way, from a different perspective, when dealing with NNS academics as opposed to NS or NNS novices. The important point is that NNS academics, as students of EAP, are both learners and experts. While their knowledge of EAP writing may not be sufficient to publish without help, they no doubt already have extensive experience reading English-language texts in their area of expertise, and are already members of a specific disciplinary community. The central point is that NNS scholars are already full members of disciplinespecific communities of practice in a way that NS and NNS students are not. Most research on EAP is aimed at addressing the needs either of foreign students living in English-speaking countries or of NS undergraduates. While this situation is understandable, it means that many of the issues raised in such research are not relevant to NNS scholars wanting to improve their professional EAP skills. Weingarten, for example, emphasises the importance of “scientific socialisation” in learning an appropriate academic style (1994: 124), and L. Flowerdew too argues that writers need to know the social context surrounding texts if they are to write appropriately (2005: 323; see also Belcher 1994; Connor & Mayberry 1995; Prior 1995). NNS academics have already been socialised into the academic community in their native culture, and most likely into a disciplinary community that is international anyway. They are therefore well aware of the social context surrounding the specific text types they read and produce. Other researchers talk of “enculturation” or “acculturation” (Furneaux, Robinson & Tonkyn 1988; Gee 1989; Haas 1994; Meyer 1995; Belcher 2004). This term might well be applicable to foreign students trying to find their way around a British or US-American university system <?page no="291"?> 291 and adjust to living in a foreign culture, but has a totally different meaning for NNS academics resident in their home country, who simply wish to publish in international English-language journals, or give papers at English-language conferences. On the one hand, NNS academics already have an in-depth knowledge of the academic and disciplinary cultures they wish to participate in, and on the other, they have no need of a full-scale cultural adjustment to a British or US-American tertiary education environment. In contrast to NS and NNS novices, NNS scholars wishing to publish in English are already successful academics; they are already relative insiders of a particular discipline. Enculturation is not what they need. NNS scholars need a pedagogy that is less focused on academic socialisation, initiation and enculturation. Instead, they need to know how to adapt and exploit the expert knowledge they undoubtedly possess in order to achieve success in the target culture as well. NNS scholars require detailed instruction about the specific adjustments they will need to make in their native writing (speaking, reading etc.) practices. This can only come from a comparison of academic interaction norms in their nativeculture with the practices of the target (English-language) culture, as relevant to their specific disciplinary community. This recognition leads to the second point I wish to make, namely that EAP tuition for NNS professionals must be contrastive. In the previous chapter, I referred to research in cognitive science, educational psychology and instructional design, which has suggested that people learn best when they can connect new knowledge with old. Since NNS academics possess considerable knowledge of academic interaction in their own culture, there is much that can, and should, be built upon in the advanced EAP classroom. With reference to NS novices, Dillon observed that “what learners need to know is how far their commonsense intuitions about interactions and communication can be followed and trusted” (1981: xi). If one substitutes “commonsense” for “native-culture”, one can see the challenge for EAP pedagogy of NNS academics. Instructors should therefore ideally be acquainted with academic discourse norms in both the source and target culture. Another advantage of explicitly contrastive teaching is that it enables teacher and students to regard patterns that do not conform to the target cultural norms not simply as errors or inadequacies, but as evidence of what scholars do know about a different rhetorical tradition (Leki 1991: 138). Reid’s suggestions for the multicultural classroom are applicable here: open discussion of differences between the native and target culture can be a good way of emphasising that the question is not so much what is right and what is wrong, but what is different (1992: 212). I suspect that this type of approach might be particularly advantageous in dealing with adult learners, intellectuals at <?page no="292"?> 292 that, who are likely to be overly self-critical and expect far too much of themselves. It might cost NNS academics quite some effort to admit that they are not able to publish successfully in English without assistance, and the situation will become far worse if they are made to feel inadequate. Contrastive curriculum is particularly important when it comes to creating an appropriate relationship with the intended reader. Metadiscourse is a case in point. Crismore, Markkanen and Steffensen observe that “proficiency in this area [metadiscourse] is notoriously difficult to attain in a foreign language” (1993: 41). The reason for this, I suggest, is that metadiscourse is connected to textual expectations, and to the anticipation of imagined reader expectations. NNS academics will have a good idea of what they expect of academic texts, as readers, and of how to create an author-reader relationship appropriately in their native culture and discipline. However, since textual expectations differ across cultures, they may feel unsure about what is expected or appropriate in the target culture. Linguistic insecurity may also hinder NNS scholars even from achieving the same rhetorical effects in a foreign language. Reid emphasises that the problem NNS have with academic composition in English is their “limited or skewed perception of what is expected” (1992: 221): this applies not only to what teachers expect of students, but also to what readers expect of academic writers. For Intraprawat and Steffensen, “skilled writers” are aware of their readers’ needs and command strategies for rendering their writing more “considerate” and “accessible” to the reader. Authors who are unable to create considerate texts are classed as “poor writers” (1995: 253). This seems harsh: it is not necessarily true that the NNS Intraprawat and Steffensen studied, students who had performed poorly in ESL essays, are inconsiderate at all; more likely, they either lack the foreign language skills to achieve the effects they desire, or they are simply considering a different kind of imagined reader. Contrastive examination of reader expectations, or rather of the different kinds of imagined readers, common in the native and target culture, would be of great help here. The third point to be made is that EAP teaching for NNS scholars must be discipline-specific. Non-specific EAP courses will be useful for both NS and NNS with little experience of academic writing, and little specialist academic training. Ideally, general EAP courses would be provided across all subjects at undergraduate level. However, generally speaking, a German philosopher or French chemist need courses that are discipline-specific; professional academics will have little time for and even less need of a general introduction to academic writing. As many studies including the current one have shown, ways of writing differ considerably across disciplines, even with the humanities and social sciences, and most certainly across the humanities and natural sciences (Russell <?page no="293"?> 293 1991; Taylor & Chen 2001; Oldenburg 1992; MacDonald 1994; Hutz 1997; Busch-Lauer 2000; Hyland 2000; Fandych & Graefen 2002; Dahl 2004). In light of such research, it is questionable whether one can even still talk of general ‘academic writing’ norms. Perhaps the term itself is little more than a convenient omnium-gatherum construct, and it would be more appropriate to talk of specific norms in specific disciplines. Certainly, when the results of the present investigation are considered, it appears that there are such great differences between even closely-related disciplines, that a general course for all NNS scholars would be of little use. Again, this of course increases the requirements for instructors: they must not only be able to teach contrastively, but also with specialist knowledge of the relevant disciplines. MacDonald (1994) suggests that, since EAP practitioners naturally cannot be expert in all disciplines, close cooperation with specialist faculty members will be necessary (see also Flowerdew & Miller 1992). Ultimately, it is doubtful whether general writing pedagogy will lead to success: a German historian who has been taught to write like a philosopher or physicist in her second language is unlikely to convince readers, and unlikely to place her work in leading journals. Writers must adhere to discipline-specific textual norms, and these can best be taught in discipline-specific courses. In the present study, there is evidence both of disciplines with similar profiles across cultures, and of others with highly divergent patterns across cultures. The situation in a particular cultural constellation can be determined only by examining texts from the target discipline in both the native and foreign culture, and from here trying to establish which commonalities and differences exist. Either through active research, or through their own experience as disciplinary insiders, specialist EAP teachers will need to do this. Ultimately, as Swales notes, “high frequency does not entail high pedagogical priority” (2002b: 163): rather, in EAP teaching, high priority must be assigned to those textual features in a specific discipline which are not common to students’ native and target cultures. While EAP courses for NNS academics are useful and necessary, they do not eliminate the need for other types of assistance for English-language publication. Ideally, each department, or at least each faculty, would have a permanent staff of specialised NS instructors on hand to help NNS academics as required. Failing this, perhaps each university could have an English-language editing section, again comprised of NS EAP teachers from different academic specialisations. These staff should also include specialist translators, since there are some cases where it will be easier and better to translate material into English, rather than edit it extensively. This of course requires translators who are well acquainted not only with the academic culture of the NNS scholars, but also with the disciplinary <?page no="294"?> 294 culture relevant to each source text. Since academics speak as well as write, there is also a need for training in English-language presentation skills, and in spoken academic text types such as lectures and conference presentations. Ideally, EAP tuition, editing and translation services would be offered free of charge to all academics who wish to publish in English. There is a huge need for specialised, contrastive, EAP tuition for NNS academics from EFL countries. 110 The recommendations made here concern the ideal type of tuition for this specialised target group, and I recognise that it may well prove impossible to implement all of these suggestions. The challenge for this area of EAP pedagogy may lie in finding ways to approach the ideal; much would be won if the teaching were contrastive at least, or the courses could be discipline-specific. Having an ideal in mind might stimulate work on improving the current state of EAP services provided to NNS academics publishing in English. If all of this seems utopian, it could be worth reminding universities that they stand to benefit from increased English-language research activity by their academic staff. Since leading journals publish in English, and English-language publications are cited more often than those published in other languages (see chapter one), universities will profit just as much as individual academics from the academic exposure that may be gained. 111 From this profit springs the obligation to offer assistance to NNS academics publishing in English, rather than leaving scholars to bear the costs, while universities (also) reap the benefits. At present, NNS scholars pay handsomely, often from their own purse, for English-language editing and translation services, and the provision of English-language services for faculty in tertiary institutions is erratic at best. This situation needs to change, in order to decrease the disadvantage suffered by NNS scholars when publishing in English. 112 With the assistance of editing and 110 It seems to me that, though this need is often unrecognised or unsatisfied by academic institutions, it is acutely felt by many NNS academics. Anne Pallant of Reading University reports the success of a distance course offered to French-NS academics publishing in English (personal communication), as does Robin Bellers, who teaches specialised ESP courses for faculty at the Central European University in Budapest (personal communication; see also Sengupta, Forey & Hamp-Lyons 1999). 111 I am thinking here not just of general academic prestige. With the advent of evaluation and performance management in the tertiary education sector, the number and quality of publications each university produces can directly influence funding outcomes. Publication activity is also taken into account in university rankings, with highly ranked institutions attracting better students and staff, which in turn benefits the university. 112 The needs of NNS academics will change if EFL countries continue to replace national languages with English for tertiary education. The increased introduction of Englishmedium instruction from the undergraduate level would ensure a more uniform level of English-language competency among young academics, and would offer the possibility of integrating basic EAP courses into the curriculum at an earlier stage. At present, such <?page no="295"?> 295 translation services, and specialist EAP courses which are disciplinespecific and taught contrastively, much could be done to help NNS scholars cope better with the current dominance of English. Having discussed the pedagogical implications of this study, in particular for NNS scholars publishing in English, let us now move on to the language political implications. 5.4 Approaches to EAP in a lingua franca world Early work on cultural differences in academic discourse was overtly pedagogical and prescriptive in focus. It operated from the assumption that there is a right and wrong way of writing, and that NNS learners, like NS learners, should be taught the right way. The role of contrastive research was to determine what exactly the differences between cultures were, and thereby aid teachers in discerning the reasons for their NNS students’ errors. It was automatically assumed, therefore, that NNS should and would want to adopt target-language discourse norms: […] each language and each culture has a paragraph order unique to itself, and that part of the learning of a particular language is the mastering of its logical system. (Kaplan 1966: 14) The explicitly pedagogical focus of EAP research, including genre analysis, has continued to the present day (Swales 1990, 2004b; Flowerdew & Peacock 2001; Hyland & Hamp-Lyons 2002). Flowerdew for example writes of the role EAP research can play in uncovering preferred expectations in different cultures and contexts, noting that “these preferential expectations can be examined and the resulting descriptions can form the basis of pedagogical materials” (J. Flowerdew 2002c: 3). However, in view of the role of English as a global lingua franca, some recent work has begun to question the relevance of NS judgements and to contest the place of NS norms in language teaching (Pennycook 1995; Zamel & Spack 1998; Kubota 1999; Ramanathan & Atkinson 1999; Benesch 2001; Phillipson 2003; Kirkpatrick 2004b; Kubota & Lehner 2004). Connor, who has been involved in much contrastive-rhetoric research, writes of the ideological problem presented by deciding which norms and standards to teach, and argues that teaching NS norms risks “perpetuating established power hierarchies” (2002: 9). Benesch is even more explicit, stating that EAP has economic roots, and is the result of “efforts of courses are often available only to those students majoring in English, even though students from all disciplines could benefit from them. <?page no="296"?> 296 governments and private companies to promote English worldwide for political and commercial purposes” (2001: 25). This parallels Phillipson’s conception of the conspiracy of major commercial interests behind the “English language industry” (2003: 16), and illustrates that much of the rhetoric behind critical EAP is simply borrowed steam from the wider debate on international English. Parallel to the broader debate over who ‘owns’ English as a global language, there is controversial discussion regarding the ownership of international scientific English. Wood for example opines that: international users of scientific English should have the right to determine the type of English deemed acceptable or standard in that community. Scientists should determine that regardless of their native language. […] The owners of international scientific English should be international scientists not Englishmen or Americans. (Wood 1997) which is fine in theory, so long as one does not try to have an article published in a journal edited by native-speakers, since they appear to be unaware of any such right. The problem with such arguments, as Thompson points out, is that it is not helpful for teachers to refrain from demanding adherence to target-culture discourse norms if work is going to be judged later according to precisely these norms (2001: 77). This is the situation in academic writing, as both personal accounts (Ammon 1989) and numerous studies show (Gosden 1995, 2003; Canagarajah 1996; Prince, Blicblau & Soesatyo 2000; J. Flowerdew 2001; Burrough-Boenisch 2002). It seems, therefore, that critical considerations of the place of NS norms in EAP are missing the point. At present, the debate should centre not on the responsibility of EAP teachers and students to examine critically and to challenge norms, but rather on the need for more egalitarian power relations in the international scholarly community. This could begin with increased tolerance among the gatekeepers of academic discourse, in particular journal editors. Tolerance is likely to come from an increased awareness of cultural variations in academic discourse styles (Mauranen 2001: 53-4) and it is here that contrastive research could be useful. Tolerance could also emerge from the economic recognition that NNS are an increasingly important readership block, and are unlikely to subscribe to journals in which their work and that of their colleagues seldom appears. The middle line between the prescriptivism of early work in contrastive rhetoric and the rejectionist discourses of later research is to raise students’ awareness of differences between discourse norms in their native and the target culture (Kaplan 1988; Reid 1989; Leki 1991; Connor 1996; <?page no="297"?> 297 Grabe & Kaplan 1996; Johns 1997). 113 However, such awareness-raising must also include knowledge of the consequences of non-adherence. This does not preclude critical reflection upon and questioning of discourse norms in both cultures, but students must be equipped and permitted to judge for themselves whether they wish to adhere to target-culture discourse norms or not, rather than being forced to become critical discourse analysts and fighters against English-language ‘hegemony’. Students can only make an informed choice between adherence, challenge or perhaps another strategy entirely, of course, if EAP instructors do not stop at awareness, but explicitly communicate the differences between native and the target-culture discourse norms. The sad fact is that no NNS scholar ever had an article accepted for publication simply because it was ‘aware’. Some researchers appear to view teaching and learning, as opposed to awareness-raising, as something deleterious. Johns for example argues that: full involvement or affiliation in academic discourse communities requires major cultural and linguistic trade-offs from many students. Faculty expect them to accept the texts, roles, and contexts of the discipline, but acceptance requires much more sacrifice and change than the faculty may imagine (1997: 66) This seems a strangely negative way of describing the learning process. Why is change here automatically equated with sacrifice? Cannot change be enrichment too? Indeed, as anyone who has learnt a foreign language, or acclimatised to a foreign culture, will attest, one does not surrender or sacrifice anything, but gains. When I learnt German, I did not sacrifice my identity as an Australian and Briton, even if I often choose to behave according to German cultural norms in most situations when I am in Germany. Assuming that a student is voluntarily studying at a foreign university, it would seem to be a great achievement and source of pleasure, even pride, to become a member of an additional discourse community. There is no trade-off. In similar fashion to Johns, Allison speaks of “painful” learning experiences and “cultural clashes”, and argues that the “dilemma is how to help non-native speakers retain their identity in 113 Flowerdew places such approaches in the context of the New Rhetoric school (Freedman & Medway 1994), which argues for a “pedagogy developed along the lines of general consciousness raising rather than overt didactics” (J. Flowerdew 2002c: 3). A teacher who offers no rules at all, not even some orientation, leaves all the work to the students, who must extrapolate rules from their own observation. This is likely to be much more time-consuming and probably inaccurate than teacher-guided experimentation. Pedagogy without didactics therefore seems to me irresponsible; ideally, however, the two could complement each other. <?page no="298"?> 298 composition classrooms” (1993: 122). Again, this could be rephrased equally well as ‘the challenge is how to help non-native speakers gain a new identity in composition classrooms’. The configuration of cultural learning as sacrifice, conflict and dilemma is puzzling and disturbing. Clyne too views the idea of cultural learning as discomforting: Such pragmatic features are so closely connected with the own culture and individual personality that it would not be appropriate to expect people to adjust to foreign norms. (Clyne 1993: 4) It is entirely appropriate to expect people to engage actively with other cultures. It would be better if this exchange were not completely one-sided; assuming NS journal editors, for example, are interested, they could certainly learn much about other cultures by asking foreign contributors about academic discourse norms in their native culture. However, given the power relations of this situation - an academic gatekeeper who is paid to assess critically the work of a NNS colleague - one must be realistic about the direction in which most of the learning will take place. Teachers who only raise awareness but do not teach are therefore failing their students: in all the very modern ideas of teaching as a cooperative process of discovery, of students as researchers etc. (for a typical example of this approach, see Johns 1997), it is easy to lose sight of the fact that teachers exist to convey information to their students. Instructors who do not prepare their students for the fact that their work will be judged according to its conformity to specific NS textual patterns, and do not teach students these patterns, are leaving their students vulnerable and unprepared for the often harsh reality of international academic publishing in English. If a NNS scholar, based on this knowledge, decides not to conform to target-language discourse norms, that is at least an informed decision. Ultimately, contrastive research into preferred styles of academic presentation has two main areas of application. First, the findings of such research can inform pedagogy, and thereby improve the chances of NNS academics to participate more equally in English-medium international scholarly exchange. Second, contrastive research can help to sensitise writers of all cultural backgrounds to previously unsuspected areas of difference, and thereby encourage a more tolerant environment for scientific exchange at the international, that is cross-cultural, level. Both of these areas are vital to improving the situation of NNS academics reliant on English as a lingua franca. <?page no="299"?> 299 5.5 Improving the situation of non-native speaker researchers A situation in which one language predominates in a specific functional domain will never be egalitarian, especially if there are large numbers of NS of this language enjoying a competitive advantage through no effort or merit of their own (see discussion in Phillipson 2003). However, a number of improvements could lessen the inequality somewhat. First, it is important that, for NNS, L1 publications also count in performance reviews. Vandenbroucke states that, in the Netherlands, publications in the local language are not counted by inside or outside reviewers of academic performance when scientific pecking orders are calculated to rank persons or departments. Dutch publications are judged somehow as local folklore. (1989: 1461) If this situation does not change, there is little incentive for scholars to continue academic publication in their mother tongue; the mother tongue, the local community, and possibly the conduct of science as a whole, will suffer as a result. In the absence of research on the subject, it is unclear whether the Dutch policy is representative of European or even wider international practices as a whole. Performance reviews in Germany do not appear to be language-specific. However, in its performance review guidelines, the Humbolt Universität in Berlin requires the “Nennung von drei bis fünf (bei großen Abteilungen maximal 10) ‘key publications’” [List of three to five (for large sections maximum 10) key publications], first and foremost RAs, which are to be submitted “mit ‘peer-review’” [with peer review] (2003: 4). It is patently clear here what language these key publications are assumed to be in. The Wissenschaftliche Kommission [Scientific Commission] of the north-German state of Niedersachsen notes in a performance review that the number of publications alone cannot be a sufficient assessment criterion. Rather, a qualitative ranking is important “wobei der impact dieser Publikationen, wie er etwa in Rezensionen, Zitaten und Bezugnahmen, Neuauflagen oder Übersetzungen manifest wird, ein wichtiges Indiz ist.” [for which the impact of these publications, as manifested for example in reviews, citations and references, new editions or translations, is an important indicator] (2005: 9). The injustice of this, for academics who publish in German, is that their work of course reaches a smaller audience, and therefore has a smaller possible chance of being cited and reviewed, than if they were to publish in English. It seems, therefore, that even in an EFL country where performance reviews do not explicitly weight mother-tongue publications lower than English-language <?page no="300"?> 300 ones, there is still evidence of subtle pressures on NNS academics to transfer all scholarly activity to English. This only enforces the negative consequences of English-language dominance for other speech communities. In theory, at least, it would be possible to assign extra credit to nativelanguage scholarly activity, as part of a positively discriminatory proactive language protection policy. This would, however, require government recognition of, and concern for, loss of the national language’s scientific functionality. The second suggestion for lessening NNS disadvantage is that universities in English-speaking countries should not remove the language requirement from their higher degrees. Swales notes that, in the USA, the requirement for PhD candidates to possess at least reading competency in one or more foreign languages has been largely dropped (2004b: 39). If NS future academics do not learn foreign languages, they will be less able to empathise with NNS colleagues, and risk appearing arrogant in their assurance of their native language’s dominant status. Since the predominance of English is not set in stone, and there is still research being published in other languages, foreign language skills could also be advantageous to the NS researcher: monolingual NS academics will be unable to read research published in any other language, and will be at a tremendous disadvantage to their multilingual colleagues if the status of English should change. Fishman warned in 1977 of the parochialism that can result when NS of a dominant speech community cease learning foreign languages (1977b: 334-5). He argued that the linguistic and cultural insularity of English NS stifles scientific and technological innovation, since there is no exchange, and that this situation may ultimately lead to the erosion of English language dominance. There are therefore enough arguments for encouraging even purely self-interested NS to learn about other languages and cultures. Though NS will still unfairly profit from English-language dominance, multilingual NS could assuage, at least symbolically, some of the injustice perceived by NNS scholars. Third, international journals need to become actually more international. Editorial boards should include a variety of NNS speakers, and indeed also NS from underrepresented and/ or newly-emergent varieties, who can work to prevent bias based on cultural differences, and journals should also appoint accomplished NNS referees, who can select and judge contributions with reference to the writing norms in the culture of origin (De Swaan 2001: 79). Anglophone journals must open themselves to international contributions, including those by NNS of English. Apart from inviting recognised NNS scholars to join editorial boards, and increasing the use of international peer reviewers, journals could also profile international research topics, present more content online, increase the presence of <?page no="301"?> 301 US journal (commissioning) staff at international research meetings, and offer writing workshops for NNS at conferences (Iverson 2002; Swales 2004b). Ammon pleads for NNS’ “right to linguistic peculiarities” (2000: 114), as does Kirkpatrick, who argues that since monolingual academic publication encourages a “one-way flow of ideas and this is dangerous”, in order to stimulate the exchange of ideas across cultural barriers, journal editors must accept contributions which do not conform to target-culture rhetorical principles (2004a: 2). Given that English is renowned for having adopted foreign words, one wonders whether it could not also adopt some foreign discourse traditions, or at least develop a measure of tolerance towards them. Duszak argues for constant negotiation of preferred styles of academic discourse or at the very least for negotiation of “acceptability thresholds” for NNS academic texts, though she gives no indication of how or where such negotiation is to take place (1997: 19, 23). Mauranen calls not just for tolerance, but also for a greater understanding of cultural variation in academic texts among editors of academic journals (2001: 53-4). Such understanding, Mauranen argues, will not only increase fairness in international scholarly publishing, but also make use of the “hybrid vigour of text and thinking arising from academic traditions in contact”; both sides could therefore benefit (2001: 54; see also Allison 1993: 104). Ultimately, if a journal presents itself as international, its target discourse norms should be defined not on cultural, linguistic or national grounds, but should reflect those of the international academic community interested in the particular subject or discipline of the publication. The international academic community, of course, cannot be expected to abide by solely US- American or British discourse norms. Assuming that journal editors can alter their expectations, the result could therefore be a profitable exchange for both parties. The situation in medicine, where many of these suggestions are already being implemented by journals and have met with considerable success, is encouraging (Swales 2004b: 42). 5.6 The situation for scholarly languages other than English As noted in chapter one, at the beginning of the 20th century, and in fact even well into it, French and German were international languages of scholarly publication alongside English. As we saw, this is no longer the case. For former scientific linguae francae, the loss of this particular function is accompanied by a loss of prestige, both on the international <?page no="302"?> 302 stage, and also in the minds of these languages’ native speakers. It appears that a downhill slide in the scientific functionality of these languages sets in, with each decision against the mother tongue and for English in academic publishing only enforcing the trend. Academics from former lingua franca speech communities may feel torn between responsibility to their native language and concern for their own professional advancement. It is hardly realistic, however, to expect all French and German scholars to publish in their mother tongue so that its scientific function may be somehow rescued or preserved, especially if this decision is to the detriment of their careers. It may be enough to encourage scientists to reflect upon the possible consequences of their language use by fostering an increased language awareness in this context. This may lead individual academics to reconsider their publishing behaviour, perhaps to the benefit of retained functionality for the mother tongue. Swales notes that many NNS academics derive a considerable amount of satisfaction from continuing to publish in their native language alongside English (2004b: 39). If most scholarly writing now takes place in a foreign language, a greater effort will need to be made in order to communicate the worth and import of scholarly findings to the wider community. The displacement of the vernacular from the academic domain risks research becoming a secret, suspicious activity, inscrutable in its motives and incomprehensible in its achievements. This should worry researchers reliant on external sources of funding, since few companies will fund projects they do not understand and whose advantages cannot be explained to the populace. To a lesser extent, governments may be faced by the same problems in trying to explain research funding policy to voters (Beesley 2003). Since academics are generally employed by taxpayer-funded institutions, the public responsibility to popularise and disseminate findings in the mother tongue should not be taken lightly. Exclusivity will ultimately be to the detriment of both parties. 114 For smaller speech communities, whose languages never functioned as international scholarly tongues, the advent of English as the scientific lingua franca has essentially meant the exchange of one foreign language for another for publication purposes. Such is the case in countries such as Sweden and the Netherlands, where scholars relied upon Latin up to the mid 18th century, and in the first half of the 20th century often used 114 Exclusivity is of course a relative measure: if academics from EFL countries publish in English, they exclude the non English-speaking parts of the national population from their work. On the other hand, if academics from EFL countries publish in their native tongue, they then exclude all other scholars who cannot speak that particular national language. In this sense, NNS scholars cannot win; someone will always be excluded. This one of the reasons why the translation and popularisation of scientific writings is so important. <?page no="303"?> 303 German (Vandenbroucke 1989; Winsa 2000; Gunnarsson 2001). The switch to English is therefore just one more swap to a foreign language of science. For multilingual speech communities, it seems that the dominance of English marginalises the second and third languages routinely used for scholarly writing. In Finland, for example, the rise of English has coincided with, and most likely accelerated, the demise of scientific publication in Swedish and German, which were widely used as scholarly languages up to the second half of the 20th century (Swales 2004b: 36; Wilson 2002: 10). Similarly, in multilingual Switzerland, there are concerns that English is replacing German and French for internal lingua franca communication, and marginalising all other languages for scholarly purposes (Trabant 2000). On the other hand, it appears that while the second (and possibly third) languages in such situations decline, the proportion of academic publications in the major national language remains relatively stable (Swales 2004b: 36). Swales therefore perceives a trend towards both the global and the local (2004b: 36). There is a functional niche for research published in a local language, but it seems to be large enough to sustain only one language, not several. Essentially, something like scientific pragmatism should operate: English for studies or findings that are of international import, and the local language(s) for issues of national or regional concern. The result would be a kind of academic diglossia or perhaps triglossia motivated not just functionally (all science is in English) but also by the target audience (local/ national/ international). This sort of motivated language use, as opposed to a blind and complete switch to English, is probably the best outcome. For disciplines that are of primarily domestic import, this ought to mean that the majority of research is written in the domestic language(s). Such a practice might ensure the continued existence of languages other than English for scientific publication. Unfortunately, at present there is much evidence of unfounded and essentially unjustifiable replacement of national languages with English. Disciplines which one would assume to be of mainly local concern are represented by Englishlanguage publications. Taking German as an example, there appear to be very few leading literary studies journals that still publish solely in German: the only one known to me, Euphorion, was the source of the texts in the corpus. Other leading German studies periodicals, such as the Deutsche Vierteljahrsschrift für Literaturwissenschaft und Geistesgeschichte, the Beiträge zur Geschichte der deutschen Sprache und Literatur, the Wei marer Beiträge and the Zeitschrift für deutsche Philologie, have now switched to bilingual publication. One wonders how long it will be before English completely displaces German even here. <?page no="304"?> 304 These decisions are difficult to understand: is it really the case that literature scholars the world over, who specialise in German literature, cannot understand or publish in German? Editorial decisions such as this smack more of what, in German, would be called ‘vorauseilender Gehorsam’ - a sort of anticipatory obedience to a state of affairs not yet decided in one’s disfavour. In so doing, editors create the situation they have already surrendered to, instead of realising, and possibly making use of, their power to influence the linguistic status quo. Defeatist thinking such as this can perhaps be reduced through awareness-raising research and targeted lobbying for the place of national languages. All of this of course assumes that there is something at stake here; that something will be lost if English eventually reigns supreme and singular as the sole international scholarly language. Yet it is difficult to quantify or even describe exactly what could be lost. An essential recognition is that the switch from the mother tongue to English for academic activity is more than just changing labels, or decanting the same wine from one bottle to another (Schiewe 1996). The linguistic switch entails a fundamental and far-reaching change in styles of thinking, since science conceived of and explained in English is not the same as science conceived of and explained in German or French. If this switch leads to multilingualism and multiculturalism, the potential benefits are huge. If it leads to English monolingualism and monoculturalism, however, the disadvantages may be considerable. Without wanting to expound a strongly Whorfian position, what I am suggesting is that academic languages are repositories of the scholarly history and thought models of a particular cultural and discourse community. 115 However, in the absence of draconian government intervention on the side of the mother tongue, language choice is a personal matter, and few academics are romantics. If English only is the current fashion in academic publication, many appear happy to follow the trend. It would not be the first language shift. 115 Everything from the terminology to the preferred format for research articles is reflective of a specific, perhaps unique, way of seeing the world and of expressing this world view through language. By this I mean that language does not dictate thought, but it does influence it. The human diversity of these world views may not be lost, but it is certainly hindered by expression in a foreign language. Languages have concepts, or words for concepts, that are absent in English, taxonomies more refined or simply different, theories or scholars so grounded in their native culture that they do not readily translate into English. In this sense, professional translators and advanced bilingual/ multilingual academics have a vital role to play in ensuring that this diversity is transferred to and retained in English. Aside from romantic arguments about the value of each language per se, it is possible that reduced diversity in this respect might also be detrimental to scientific research itself, which generally profits from a variety of ideas and points of view. A switch to English only may therefore have unexpected - though by no means unavoidable - negative consequences. <?page no="305"?> 305 There is of course the possibility that English will be superseded by another (scholarly) lingua franca. Entering the realm of divination, it is worth noting that the Hispanic subpopulation in the USA has now overtaken the African-American community to become the largest ethnic minority, with around 41.9 million members, or 14.5% of the US population (2005 American Community Survey figures). 116 The future influence of Spanish as an international (academic) language will depend on the economic situation and research output of Spanish-speaking countries and of the USA, on the socio-economic status of the Spanishspeaking population in the USA, and on whether later generations of US- American Latinos still speak Spanish. If one considers the forecast continued population growth for the Hispanic minority in the USA, and the fact that Spanish, like English, has considerable geographical diffusion due to its colonial history, Spanish appears to be a possible candidate for replacing global (scholarly) English. Another prominent candidate for future lingua franca status is Mandarin Chinese. If population alone were decisive, Mandarin would already be the global language. However, far more important than the number of people alone is the general economic situation, in particular the investment in research and development. It is here that China has enormous potential. At present however, despite rapid economic growth, and fast-expanding research activity and international output, China comes nowhere near the scholarly might of the USA. In fact, Chinese research productivity is comparatively low: China is not even in the top ten of research paper countries of origin, while the USA is at number one (Gibbs 1995). 117 Discussions of alternative languages of science remain speculation and not fact, and though I have mentioned two possible alternative scholarly languages, some researchers do not think that English will be replaced in this function, or its wider lingua franca function, any time soon (De Swaan 1993; Ammon 2001; Mair 2003). Personally, I am not at all convinced that English will remain the global lingua franca; at present, 116 According to the latest United States census data (www.census.gov), for the period July 1st 2003 to July 1st 2004, the Hispanic minority had the highest population growth rate of any ethnic group in the USA, and more than triple the average population growth rate for the whole country. It is projected that the Hispanic minority will reach around 102 million people, or 24% of the total population. by 2050 (see also Grieco & Cassidy 2001). 117 As measured by the national research article output as a percentage of the articles cited in the Science Citation Index (SCI). The SCI covers 3,700 premier scientific journals in over 100 disciplines, and though it may be somewhat biased towards US-American publications, since it is a US-American initiative, it is doubtful whether bias alone could account for the huge discrepancy between the USA and China noted here. <?page no="306"?> 306 however, English is the lingua franca, and since genethlialogy is not a linguist’s forte, I shall leave more elaborate predictions to others. Ultimately, while the conduct of scientific research in a common language, no matter whether English, Spanish or Mandarin Chinese, does bring certain practical benefits, for other languages the development is not particularly advantageous. 5.7 Directions for future research There is a need not for just more studies, but for a greater number of empirically sound studies, of academic discourse across cultures. Some of the ways in which this could be ensured were discussed in the second chapter. With a well-constructed corpus and some knowledge of statistics, there are many interesting areas for further research. For a start, it would be worth extending the study of social interaction in academic prose to other disciplines and cultures. Regrettably, at present there is comparatively little known about academic discourse norms in non-English-speaking cultures: while the lopsided interest is understandable, and from a pedagogical point of view necessary, there are still whole worlds of discourse waiting to be investigated. There is also any number of academic disciplines whose discourse patterns we know nothing about. Work based on the MICASE corpus has shown that there is much to be learnt from examining spoken academic discourse, not just written texts. Studies of metadiscourse (Mauranen 2001, 2003; Swales 2001; Swales & Malczewski 2001), evaluation (Mauranen 2002; Mauranen & Bondi 2003; Swales & Burke 2003; Fortanet 2004a; Swales 2004a) and hedging (Poos 1999; Lindemann and Mauranen 2001; Mauranen 2004) demonstrate that spoken interaction in academic contexts is a fruitful area for further research. More research is also needed concerning the influence of gender on written (academic) discourse. 118 One of the most interesting and unexpected findings of the study concerned the influence of an extra-linguistic variable other than culture on the features studied. It would have been of great help had there been more research on gender-marking in academic writing with which to compare and test the SCEGAD results. There is a range of phenomena not covered in this study that are of great potential interest to further studies of social interaction in academic discourse. The use of questions, raised by Thompson (2001) in a noncontrastive study, could be of interest also from a cross-cultural perspective. The use of topic sentences - another metadiscoursal device that can 118 Gjesdal’s (2005) study of the use of feminine adjectival endings in French-language academic prose is a recent contribution in this direction. <?page no="307"?> 307 aid reader comprehension - also appears to be worth further investigation across cultures (Trumpp 1998: 61; see also Popken 1987: 211). Another potentially rewarding area for further study is citation, or intertextuality. Non-contrastive studies of citation (Cronin 1984; Thompson 2000a; Thompson & Tribble 2001; Hyland 1999b, 2003; Jakobs 1999, 2003; Cronin & Shaw 2002), as well as a new contrastive study of intertextuality in Englishand German-language linguistics RAs by Griffig (2005), suggest that there is much yet to be discovered here. I suspect also that the use of redundancies, which are an important interactive figure, would be worth considering. This list is long and open-ended, and offers exciting perspectives for future research. From a methodological perspective, there are a number of points future studies might like to consider. First, it would be a tremendous boon if more contrastive studies were carried out by bicultural researchers or mixed-culture research teams. Since, as Swales notes, it is “intuition that underlies insight into and exploitation of corpus material” (2002b: 151), such constellations would bring unique insights to cross-cultural studies. Another methodological possibility which deserves reconsideration is Clyne’s (1987) work with doctored texts, as a way of ascertaining (possibly culturally specific) reader reactions to particular discourse patterns. Reader response studies with authentic texts that happen to display all of the desired discourse features would of course be even better. It would also be valuable to examine a number of texts written by the same researchers. One could study paired samples of texts, in an ideal case on the same topic, written in the native and non-native language(s). This might reveal the adjustments multilingual scholarly authors make when writing in a foreign language, and deliver insights useful for pedagogy. It would also be interesting to examine texts written by the same author but for different audiences: studies including such pairs (Crismore and Farnsworth 1990; Buhl 1999) could be extended by a cross-cultural perspective in order to discover how scientific popularisation is carried out across cultures. An additional methodological suggestion which might be worth considering, if resources permit, is to carry out studies with more than two cultures. Some noteworthy recent examples are the work of Trumpp (1998) on French, German and English-language texts, the ongoing KIAP project on French, Norwegian and English-language academic discourse (Breivega, Fløttum & Dahl 2002; Dahl 2004; Fløttum, Dahl & Kinn 2006; Fløttum, Kinn & Dahl 2006) and Bassler and Breitkopf’s work on German, Russian and English-language sociology texts (Bassler 2003; Breitkopf 2005). The advantage of such studies is that there is a wider focus on languages other than English, which is welcome in view of the relatively <?page no="308"?> 308 one-sided focus of previous research (see section two). In addition, for corpus-linguistic studies, once a data collection method and text selection criteria have been formulated, it is often relatively easy to extend the corpus to include more languages. The International Corpus of Learner English (ICLE) project, for example, has grown since 1990 from a collection of English-language essays by French NS students to include over 2 million words of argumentative essays by advanced learners of English from 19 different language backgrounds. This is one of the advantages of principled collections of text such as the SCEGAD corpus. A major area for future research lies in the combination of discourse analysis with corpus linguistic methodology to investigate language use in specific contexts. This study was designed to show that such a combination is not only practicable, but also opens up new avenues for future research. Indeed, there is at present much interest in combined corpus and discourse approaches, in particular when applied to the study of academic discourse (see for example Partington, Haarman & Morley 2004; Connor & Upton 2004; Camiciotti & Tognini-Bonelli 2004; Tognini-Bonelli & Camiciotti 2005; Hyland & Bondi 2006; Fløttum, Dahl & Kinn 2006) 119 Even in critical discourse analysis, an area of discourse research that has previously been markedly unempirical in focus, there is a growing recognition that, as Hardt-Mautner puts it, “even the crudest techniques of corpus linguistics can make useful contributions to the study of discourse from a critical perspective.” (1995: 5; see also Garzone & Santulli 2004). Flowerdew calls for a “cross-fertilisation” not only between corpus linguistics and other linguistic fields, but also between descriptive and applied EAP research, in order to maximise the benefits of corpus-based studies for the teaching of academic discourse (J. Flowerdew 2002b: 114). In practice, I suspect that it will be difficult for any one study to achieve both these aims. However, there is certainly much to be gained by corpus-based research increasingly informing pedagogy, as a great many researchers point out (Aston 1998; Williams 1998; Coxhead 2000; Briggs & Lee 2002; Bowker & Pearson 2002; L. Flowerdew 2004, 2005; Gavioli 2005; Scott & Tribble 2006). 120 If discourse analysts are willing to engage with corpus linguistic methods, there is no end of topics that might be fruitfully investigated. As Kennedy observes, “the research topics in a machine-readable corpus are potentially as various and wide-ranging as are the facts about a language and the use of that language” (1998: 274). The present investigation aimed to integrate corpus with discourse approaches, from stepwise multiple regression and 119 Another volume on the topic of discourse analysis and corpus linguistics, to be edited by Ädel and Reppen, is in preparation. 120 Though see Swales (2002b) for a critical assessment of the place of corpora in ESP pedagogy. <?page no="309"?> 309 correlation analyses to the consideration of discourse strategies and rhetorical effects. These are not methods or perspectives that sit easily with each other, and it is precisely for this reason that their combination has so much to offer. This is not to deny that the combination of corpus and discourse has limitations. It is difficult to work with a plain-text, unannotated corpus. Hardt-Mautner argues persuasively that the use of concordancing software does “not necessarily mean restricting oneself to below-sentence-level linguistic description but can in fact open a window into larger-scale discoursal processes.” However, as the author notes, none of this is possible without a “word-based ‘peg’ to hang the analysis on”, or in other words, a “discrete, searchable item around which the higher-level phenomenon that we are after is expected to cluster” (1995: 19). The problem is that this peg is often not present in discourse phenomena, or rather, it is present, but shared with dozens of other phenomena. In discourse analysis, there is no unique, stable correspondence between linguistic form and discourse function. Even if one finds a peg, it is still necessary to weed out the instances of all other discourse phenomena which display the same surface features. In some cases - this was the case for text comments in the present study, for example - the manual reanalysis would have been far more labour-intensive than simply carrying out a completely manual analysis from the outset. In the absence of any satisfactory system for discourse tagging (see discussion in chapter two), researchers cannot avoid often extensive manual analysis. As a result, the phenomena that can be studied, and the amount of text that can be examined, are severely limited (Aston & Burnard 1998: 27). Römer notes that when studying evaluation she could not examine all the phenomena identified manually, because they were too hard to trace and handle (2005: 100; see also Hunston 2004: 185-6). This dilemma, common to all corpus-based studies of discourse phenomena, will persist for a long time. However, size is not everything. As the present study shows, there are worthwhile insights to be gained with a relatively small corpus if it is sufficiently specialised and controlled for influential variables. I should like to conclude with a few comments on the pitfalls of culturally contrastive analysis. It was noted above that many of the dichotomies posited in previous investigations are biased and inaccurate. Judged from the perspective of a cultural outsider, textual patterns are described as deviant, and assigned communicative intentions and meanings they do not possess in the culture of origin. The underlying problem is that, rather than simply recording what they see, researchers extrapolate reasons for observed linguistic features. When analysing and contrasting the products of different cultures, it is particularly tempting to attribute divergent <?page no="310"?> 310 motives to divergent linguistic behaviours. Behind this is the human need to interpret and to create patterns in what we observe. Linguists are no more immune than anyone else is to the seductive power of cultural categories. However, it is important that linguists in particular are scrupulous in their avoidance of extrapolation. Positing fixed cultural categories and dichotomies risks encouraging stereotypes, and precludes the open-ended research that might tell us what is really going on in a specific communicative context. Since most language users themselves are not even conscious of the purpose(s) their speech acts serve, it is unlikely that monocultural linguists alone will be able to judge these motivations definitively by examining only linguistic products. When researcher and object of study are from different cultures, there is even less likelihood of accurate interpretation. Even those linguists who do not set out to present culturally-biased conjecture often end up doing so. In practice however, as I know from experience, it is hard to avoid. This is one area, perhaps, where we would do well to be less personal, and more detached. <?page no="311"?> 311 References Abdi, Reza. 2002. “Interpersonal Metadiscourse: An Indicator of Interaction and Identity”. Discourse Studies 4 (2): 139-145. Adams, Marilyn and Bruce, Bertram. 1980. Background Knowledge and Reading Comprehension. Urbana: University of Illinois Center for the Study of Reading. Adamzik, Kirsten. 1998. “Methodische Probleme kontrastiver Textsortenanalysen.”. In Darstellungsformen der Wissenschaft im Kontrast. Aspekte der Methodik, Theorie und Empirie, Lutz Danneberg and Jürg Niederhauser (eds), 103-130. Tübingen: Gunter Narr. Adamzik, Kirsten and Pieth, Christa. 1999. “Rhetorik der wissenschaftlichen Kommunikation im Sprachvergleich (Deutsch - Französisch). Am Beispiel von Vorwörtern zu Literaturgeschichten”. In Rhetoric and Argumentation. Proceedings of the International Conference, Lugano, April 22-23, 1997, Eddo Rigotti and Sara Cigada (eds), 73-85. Tübingen: Niemeyer. Ädel, Annelie. 2003. The Use of Metadiscourse in Argumentative Texts by Advanced Learners and Native Speakers of English. Diss. Göteborg University, 2003. Ädel, Annelie. 2005. “On the Boundaries Between Evaluation and Metadiscourse”. In Strategies in Academic Discourse, Elena Tognini-Bonelli and Del Lungo Camiciotti, Gabriella (eds), 153-162. Amsterdam: John Benjamins. Ädel, Annelie and Reppen, Randi. Forthc. Exploring Discourse through Corpora (working title). Amsterdam: John Benjamins. Aikenhead, Glen S. 1996. “Science Education: Border Crossing into the Subculture of Science”. Studies in Science Education 27: 1-52. Ainsworth, Claire. 2001. “Special Report Science in Germany: Germany Bites the Biobullet”. New Scientist. 29 September 2001. 20 August 2005 <http: / / www.gpneuro.uni-goettingen.de/ download/ press/ newScientist_29_09_01.pdf> Al-Ali, Mohammad and Holme, Randal. 1999. “The Flight from a Perfect World: Rethinking the Notion of Genre in Language Teaching”. In Issues in EAP Writing Rresearch and Instruction, Paul Thompson (ed.), 2-14. Reading: CALS. Alexander, Richard J. 2003. “G.lobal L.anguages O.ppress B.ut A.re L.iberating, too: The Dialectics of English”. In The Politics of English as a World Language: New Horizons in Postcolonial Cultural Studies, Christian Mair (ed.), 87-95. Amsterdam, New York: Rodopi. Al-Khatib, Mahmoud. 2005. “English in the Work Place. An Analysis of the Communication Needs of Tourism and Banking Personnel”. Asian EFL Journal 7 (2): 175-195. Allison, Elizabeth. 1993. Sociocomposition: Academic Discourse and the Culture of Exclusion. Diss. University of South Florida, 1993. <?page no="312"?> 312 Álvarez, Susana Álvarez. 2005. “Retórica contrastiva y calidad de la traducción: los patrones organizativos de las lenguas como motor principal del proceso traductor”. In Quality in Translation - Academic and Professional Perspectives, María Elena Garcia Garcia, Antonio Gonzalez Rodrigeuz, Claudia Kunschak and Patricia Scarampi (eds). Madrid: Universidad Europea de Madrid Ediciones. 12 January 2006 <http: / / www.uem.es/ web/ fil/ invest/ publicaciones/ web/ EN/ autores/ alvarez_art.htm> American Institute of Physics. American Institute of Physics Style Manual. 4th ed. New York: American Institute of Physics, 2005. Ammon, Ulrich. 1989. “Die Schwierigkeiten der deutschen Sprachgemeinschaft aufgrund der Dominanz der englischen Sprache”. Zeitschrift für Sprachwissenschaft 8 (2): 257-272. Ammon, Ulrich. 1993. “Wissenschaftssprache: Kein Grund zur Resignation... aber die goldenen Zeiten der Wissenschafts-Sprache Deutsch sind vorbei”. Deutsche Universitätszeitung 18: 20-22. Ammon, Ulrich. 1998. Ist Deutsch noch internationale Wissenschaftssprache? Englisch auch für die Hochschullehre in den deutschsprachigen Ländern. Berlin, New York: de Gruyter. Ammon, Ulrich. 2000. “Towards More Fairness in International English: Linguistic Rights of Non-native Speakers? ”. In Rights to Language: Equity, Power and Education, Robert Philippson (ed.), 111-116. Mahwah: Lawrence Erlbaum Associates. Ammon, Ulrich (ed.). 2001. The Dominance of English as a Language of Science. Effects on Other Languages and Language Communities. Berlin: Mouton de Gruyter. Anderson, Greg. 2002. “How to Write a Paper in Scientific Journal Style and Format”. 25 August 2005 <http: / / abacus.bates.edu/ ~ganderso/ biology/ resources/ writing/ HTWtoc.html> Anderson, Richard C., Spiro, Rand J. and Montague, William E. (eds). 1984. Schooling and the Acquisition of Knowledge. Hillsdale: Lawrence Erlbaum. Applequist, Jon. 1981. “The Use of Modern Languages in Chemical and Biomedical Sciences”. ADFL Bulletin 12 (3): 20-22. Argamon, Schlomo, Koppel, Moshe, Fine, Jonathan and Shimoni, Anat Rachel. 2003. “Gender, Genre, and Writing Style in Formal Written Texts”. Text 23 (3): 321-346. Arunachalam, Subbiah. 1998. “Information Haves and Have-nots: Technology Enhances Inequalities in Scientific Research, Too”. Educom Review 33 (6): 40- 45. Aston, Guy. 1995. “Corpora in Language Pedagogy: Matching Theory and Practice”. In Theory and Practice in Applied Linguistics, Guy Cook and Barbara Seidlhofer (eds), 257-270. Oxford: Oxford University Press. Aston, Guy. 1998. “What Corpora for ESP? ”. In L'apprendimento linguistico all'università: le lingue special, Maria Pavesi and Giuliano Bernini (eds), 205- 226. Rome: Bulzoni Editore. Aston, Guy and Burnard, Lou. 1998. The BNC Handbook: Exploring the British National Corpus with SARA. Edinburgh: Edinburgh University Press. <?page no="313"?> 313 Ausubel, David P. 1968. Educational Psychology: A Cognitive View. New York: Holt, Rinehart and Winston. Bach, Christopher L. 2006. U.S. International Transactions in 2005. U.S. Department of Commerce, Bureau of Economic Analysis. Washington D.C.: Bureau of Economic Analysis, 2006.20 September 2006 <http: / / www.bea.gov/ bea/ ARTICLES/ 2006/ 04April/ 0406_ITA.pdf > Backhaus, Klaus, Erichson, Bernd, Plinke, Wulff and Weiber, Rolf. 2005. Multivariate Analysemethoden. Berlin: Springer. Bacon, Francis. 1863. Novum Organum Scientiarum. The Works of Francis Bacon. Vol. 8. Trans. James Spedding, Robert Leslie Ellis and Douglas Denon Heath. Boston: Taggard and Thompson. Translation of Novum Organum Scientiarum. Londinii: J. Billium, 1620. 25 November 2005 <http: / / www.constitution.org/ bacon/ nov_org.htm> Bäcklund, Ingegerd, 1998. Metatext in Professional Writing: A Contrastive Study of English, German and Swedish. Upsala: Upsala University. Bailey, Richard W. 1992. Images of English: A Cultural History of the Language. Cambridge: Cambridge University Press. Bailey, Richard W. and Görlach, Manfred (eds). 1984. English as a World Language. Cambridge: Cambridge University Press. Baker, Mona. 1993. “Corpus Linguistics and Translation Studies: Implications and Applications”. In Text and Technology: In Honour of John Sinclair, Mona Baker, G. Francis & Elena Tognini-Bonelli (eds), 233-250. Amsterdam: John Benjamins. Baker, Mona. 1995. “Corpora in Translation Studies: An Overview and Some Suggestions for Future Research”. Target 7: 223-243. Banks, David. 1998. “Vague Quantification in the Scientific Journal Article”. ASp, la Revue du GERAS 19 (22): 17-27. Barthes, Roland. 1982. “Deliberation”. In A Barthes Reader, Susan Sontag (ed.), 479-95. New York: Farrar, Straus & Giroux. Bassler, Harald. 2002. “Identitätskonstruktion russischer SoziologInnen: Wissenschaftstexte in Deutschland und Russland: Subjektivität und Verortung des Autors im Wissenschaftsdiskurs”. Paper. Research Symposium of the Volkswagen Stiftung “Konstruktionen des ‘Fremden’ und des ‘Eigenen’: Prozesse interkultureller Abgrenzung, Vermittlung und Identitätsbildung”. Albert-Ludwigs-Universität, Freiburg. 24-26 April 2002. Bassler, Harald. 2003. “Russische, deutsche und angloamerikanische Zeitschriftenabstracts der Soziologie: Worin unterscheiden sie sich? ”. In Sprache und politischer Wandel, Helmut Gruber, Florian Menz and Oswald Panagl (eds), 189-212. Frankfurt am Main: Peter Lang. Baumann, Klaus-Dieter. 1998. “Die sprachliche Realisierung von Wissensstrukturen in Fachtexten des Englischen und Deutschen”. In Darstellungsformen der Wissenschaften im Kontrast: Aspekte der Methodik, Theorie und Empirie, Lutz Danneberg and Jürg Niederhauser (eds), 299-317. Tübingen: Gunter Narr. Baumann, Klaus Dieter and Kalverkämper, Hartwig (eds). 1992. Kontrastive Fachsprachenforschung. Tübingen: Gunter Narr. <?page no="314"?> 314 Baumgarten, Nicole, House, Juliane and Probst, Julia. 2001. Untersuchungen zum Englischen als lingua franca in verdeckter Übersetzung: Theoretischer Hintergrund, Weiterentwicklung des Analyseverfahrens und erste Ergebnisse. Arbeiten zur Mehrsprachigkeit 20. Hamburg: Universität Hamburg. Bazerman, Charles. 1988. Shaping Written Knowledge: The Genre and Activity of the Experimental Article in Science. Madison: University of Wisconsin Press. Beauvais, Paul J. 1989. “A Speech Act Theory of Metadiscourse”. Written Communication 6 (1): 11-30. Beck, Stefan. 2004. “Alltage, Modernitäten, Solidaritäten. Soziale Formen und kulturelle Aneignung der Biowissenschaften - Plädoyer für eine vergleichende Perspektive”. Zeitschrift für Volkskunde 1: 1-30. Bednarek, Monika. forthc. Evaluation in Media Discourse. Analysis of a Newspaper Corpus. London: Continuum. Beesley, Lisa. 2003. “Science Policy in Changing Times: Are Governments Poised to Take Full Advantage of an Institution in Transition? ”. Research Policy 32 (8): 1519-31. Belcher, Diane D. 1994. “The Apprenticeship Approach to Advanced Academic Literacy: Graduate Students and their Mentors”. English for Specific Purposes 13 (1): 23-34. Belcher, Diane D. 2004. “Research in Teaching Writing”. Annual Review of Applied Linguistics 24: 70-106. Benesch, Sarah. 2001. Critical English For Academic Purposes. Mahwah: Lawrence Erlbaum Associates. Berger, Peter and Luckmann, Thomas. 1966. The Social Construction of Reality. New York: Doubleday. Bernard, J.R. 1975. A Short Guide to Traditional Grammar. Sydney University Press, Sydney. Bhatia, Vijay. 1993. Analyzing Genre: Language Use in Professional Settings. London: Longman. Bhatia Vijay. 2002. “A Generic View of Academic Discourse”. In Academic Discourse, John Flowerdew (ed.), 21-39. Harlow: Longman. Biber, Douglas, Conrad, Susan and Reppen, Randi. 1998. Corpus Linguistics. Cambridge: Cambridge University Press. Biber, Douglas, Conrad, Susan, Reppen, Randi, Byrd, Pat and Helt, Marie. 2002. “Speaking and Writing in the University: A Multi-dimensional Comparison. TESOL Quarterly 36: 9-48. Biber, Douglas, Johansson, Stig, Leech, Geoffrey, Conrad, Susan and Finegan, Edward. 1999. Longman Grammar of Spoken and Written English. Harlow: Longman. Binchy, James. 2002. “Will I, won't I? Personal Pronouns, Grades and Changes over Semesters in Student Academic Writing”. Abstract. “Evaluation in Academic Discourse” Conference. Certosa di Pontignano. 16 June 2003. 15 August 2005 <http: / / www.unisi.it/ eventi/ ead/ binchy. pdf> Biscetti, Stefania. 2004. “Using Corpus Techniques to Study Pragmatic Meaning: The Case of ‘bloody’”. In Corpora and Discourse, Alan Partington, John Morley and Louann Haarman (eds), 285-302. Bern: Peter Lang. <?page no="315"?> 315 Blue, George. 1988. “Individualising Academic Writing Tuition”. In Academic Writing: Process and Product, Pauline Robinson (ed.), 95-99. London: Modern English Publications/ British Council. Blumer, Herbert. 1969. Symbolic Interactionism: Perspective and Method. Englewood Cliffs, N.J: Prentice-Hall. Bondi, Marina and Mauranen, Anna (eds). 2003. Evaluation in Academic Discourse. Special Issue of Journal of English for Academic Purposes 2 (4): 269-374. Bowker, Lynne and Pearson, Jennifer. 2002. Working with Specialized Language: A Practical Guide to Using Corpora. London: Routledge. Bramley, Nicolette. 2001. Pronouns of Politics: The Use of Pronouns in the Construction of ‘Self’ and ‘Other’ in Political Interviews. Diss. Australian National University, 2001. 20 December 2004 <http: / / thesis.anu.edu.au/ uploads/ approved/ adtANU20040817.141350/ public/ 02whole. pdf> Breitkopf, Anna. 2005. “Hedging in deutschen und russischen wissenschaftlichen Aufsätzen: Sprachliche und funktionale Unterschiede”. In Sprache lehren - Sprache lernen: Beiträge der 32. Jahrestagung DaF 2004, Armin Wolff, Claudia Riemer and Fritz Neubauer (eds). Regensburg: Iudicium. Breivega, Kjersti Rongen, Dahl, Trine and Fløttum, Kjersti. 2002. “Traces of Self and Others in Research Articles: A Comparative Pilot Study of English, French and Norwegian Research Articles in Medicine, Economics and Linguistics”. International Journal of Applied Linguistics 12 (2): 218-239. Briggs, Sarah and David Lee. 2002. “Developing a Lexical Database of Academic Spoken English (LDASE) for Language Testing: Problems & Prospects”. Poster presentation. 4th North American Symposium on Corpus Linguistics and Language Teaching, Indianapolis, Indiana, November 2002. 20 October 2005 <http: / / www.lsa.umich.edu/ eli/ micase/ LDASE_PosterAAACLabstract.htm> Brown, Gillian and Yule, George. 1983. Discourse Analysis. New York: Cambridge University Press. Brown, Gordon. 2005. Speech. Academy of Social Science, Beijing. 21 February 2005. 14 June 2005 <http: / / www.hm-treasury.gov.uk/ newsroom_and _speeches/ press/ 2005/ press_20_ 05. cfm> Bublitz, Wolfram. 2001. “Formen der Verständnissicherung in Gesprächen”. In Text- und Gesprächslinguistik. Ein internationales Handbuch zeitgenössischer Forschung, Klaus Brinker, Gerd Anton, Wolfgang Heinemann and Sven F. Sager (eds), 1330-1340. Berlin: de Gruyter. Bublitz, Wolfram, Lenk, Ute and Ventola, Eija (eds). 1999. Coherence in Spoken and Written Discourse: How to Create It and How to Describe It. Amsterdam: John Benjamins. Buck, Vincent. 2002. “One World, One Language? ”. Communicate! - Le cybermagazine des interprètes de conférence et du secteur des congrès: April- May 2002. 18 August 2005 <http: / / www.aiic.net/ ViewPage.cfm/ article520.htm> Buhl, Silke. 1999. “Gestaltungsprinzipien wissenschaftlicher Texte im Sprachenpaarvergleich Deutsch - Englisch am Beispiel von Texten Einsteins und Russells zur Relativitätstheorie”. In Wege der Übersetzungs- und <?page no="316"?> 316 Dolmetschforschung, Heidrun Gerzymisch-Arbogast (ed.), 117-141. Tübingen: Gunter Narr. Bulmer, Michael. 1979. Principles of Statistics. New York: Dover Publications. Bunton, David R. 1999. “The Use of Higher Level Metatext in Ph.D Theses”. English for Specific Purposes 18: 41-56. Burrough-Boenisch, Joy. 2002. Culture and Conventions: Writing and Reading Dutch Scientific English. Diss. Utrecht Institute of Linguistics, 2002. 3 February 2006 <www.lotpublications.nl/ publish/ issues/ Burrough/ index.html> Busch-Lauer, Ines-A. 1992a. “Zur Rolle der Metakommunikation in englischen Fachtexten”. In Aktuelle Probleme der anglistischen Fachtextanalyse, Rosemarie Gläser (ed.), 45-58. Frankfurt am Main: Peter Lang. Busch-Lauer, Ines-A. 1992b. “Zur Untersuchung der funktionalen Makrostruktur in englischen Fachtexten der Medizin und Psychologie”. In Beiträge zur Fachsprachenforschung: Sprache in Wissenschaft und Technik, Wirtschaft und Rechtswesen, Theo Bungarten (ed.), 14-28. Tostedt: Attikon Verlag. Bush Lauer, Ines-A. 2000. “Titles of English and German Research Papers in Medicine and Linguistics Theses and Research Articles”. In Analysing Professional Genres, Anna Trosborg (ed.), 77-94. Amsterdam, Philadelphia: John Benjamins. Busch-Lauer, Ines-A. 2001. “Kulturspezifik in englischen und deutschen Originalarbeiten - Medizin und Linguistik im Vergleich”. In Zur Kulturspezifik von Textsorten, Ulla Fix, Stephan Habscheid and Josef Klein (eds), 51-67. Tübingen: Stauffenburg. Camiciottoli, Belinda C. 2004. “Walking on Unfamiliar Ground: Interactive Discourse Markers in Guest Lectures”. In Corpora and Discourse, Alan Partington, John Morley and Louann Haarman (eds), 91-106. Bern: Peter Lang. Camiciotti, Gabriella Del Lungo and Tognini-Bonelli, Elena (eds). 2004. Academic Discourse: New Insights into Evaluation. Bern: Peter Lang. Canagarajah, Suresh A. 1996. “‘Nondiscursive’ Requirements in Academic Publishing, Material Resources of Periphery Scholars, and the Politics of Knowledge Production”. Written Communication 13 (4): 435-483. Carney, Todd F. 2003. “Quick Tips for Better Writing”. Homepage. Southern Oregon University Department of History. 23 August 2003 <http: / / www.sou.edu/ history/ carney/ quick.htm> Census.gov. 2006. United States Census Bureau. 3 February 2006 <http: / / www.census.gov> Channell Joanna. 2000. “Corpus-Based Analysis of Evaluative Lexis”. In Evaluation in Text: Authorial Stance and the Construction of Discourse, Susan Hunston and Geoff Thompson (eds), 38-55. Oxford: Oxford University Press. Cheng, Winnie. 2004. “TOOK / / did you / / from the miniBAR / / : What is the practical relevance of a corpus-driven language study to practitioners in Hong Kong's hotel industry? ”. In Discourse in the Professions, Ulla Connor and Thomas A. Upton (eds.), 141-166. Amsterdam: John Benjamins. Cheng, Xiaoguang and Steffensen, Margaret. 1996. “Metadiscourse: A Technique for Improving Student Writing”. Research in the Teaching of English 30 (2): 149-81. Chicago Manual of Style. 2003. 15th ed. Chicago: University of Chicago Press. <?page no="317"?> 317 Chinatoday.com. 2004. “Education: Facts, Figures & Links”. 15 May 2005 <www.chinatoday.com/ edu/ a.htm> Chomsky, Noam. 1994. Lectures on Government and Binding: The Pisa Lectures. Berlin: Walter De Gruyter. Clemen, Gudrun. 1998. Hecken in deutschen und englischen Texten der Wirtschaftskommunukation: eine kontrastive Analyse. Diss. Universität Gesamthochschule Siegen, 1998. 2 June 2004 <http: / / www.ub.unisiegen.de/ pub/ diss/ fb3/ 1999/ clemen/ clemen.pdf> Clemen, Gudrun. 2000. “‘Hedging’ als neue Kategorie? Ein Beitrag zur Diskussion von Gabriele Graefen, München: Eine Replik von Gudrun Clemen”. 2 June 2004 <http: / / www. sw2.euv-frankfurt-o.de/ Forschung/ Hedging/ replik. clemen.html> Clyne, Michael. 1980. “Writing, Testing and Culture”. The Secondary Teacher 11: 13-16. Clyne, Michael. 1981. “Culture and Discourse Structure”. Journal of Pragmatics 5: 61-66. Clyne, Michael. 1984. “Wissenschaftliche Texte Englisch- und Deutschsprachiger: Textstrukturelle Vergleiche”. Studium Linguistik 15: 92-97. Clyne, Michael. 1987. “Cultural Differences in the Organization of Academic Texts - English and German”. Journal of Pragmatics 11: 211-247. Clyne, Michael. 1991. “Zu kulturellen Unterschieden in der Produktion und Wahrnehmung englischer und deutscher wissenschaftlicher Texte”. Info DaF - Informationen Deutsch als Fremdsprache 4 (18): 376-383. Clyne, Michael. 1993. “Pragmatik, Textstruktur und kulturelle Werte: eine interkulturelle Perspektive”. In Fachtextpragmatik, Hartmut Schröder (ed.), 3- 18. Tübingen: Gunter Narr. Clyne, Michael. 1994. Inter-cultural Communication at Work: Cultural Values in Discourse. Cambridge: Cambridge University Press. Clyne, Michael, Hoeks, Jimmy and Kreutz, Hans-Josef. 1988. “Cross-cultural Responses to Academic Discourse Patterns”. Folia Linguistica 22: 457-474. Cmerjková, Svetla. 1996. “Academic Writing in Czech and English”. In Academic Writing: Intercultural and Textual Issues, Eija Ventola and Anna Mauranen (eds), 137-152.. Amsterdam and Philadelphia: John Benjamins. Coffey, Bernard. 1984. “ESP-English for Specific Purposes”. Language Teaching 17 (1): 2-16. Cole, Caroline. 1997. Woman, Feminist, Lesbian: Negotiated Identities in Working with Man-made Discourse Conventions. Diss. University of Illinois at Urbana- Champaign, 1997. Commonwealth of Australia. Department of Foreign Affairs and Trade. Trade 2006. Canberra: Department of Foreign Affairs and Trade, 2005. 20 September 2006 < http: / / www.dfat.gov.au/ trade/ trade2006/ index.html > Comrie, Bernard. 1981. Language Universals and Linguistic Typology: Syntax and Morphology. Chicago: University of Chicago Press. Conley, Thomas. 1979. “Ancient Rhetoric and Modern Genre Criticism”. Communication Quarterly 27: 47-53. Connor, Ulla. 1996. Contrastive Rhetoric: Cross-cultural Aspects of Secondlanguage Writing. New York, NY: Cambridge University Press. <?page no="318"?> 318 Connor, Ulla. 2002. “New Directions in Contrastive Rhetoric”. TESOL Quarterly 36 (4): 493-510. Connor, Ulla. 2004. “Intercultural Rhetoric Research: Beyond Texts”. Journal of English for Academic Purposes 3 (4): 291-304. Connor, Ulla and Gladkov, Kostya. 2004. “Rhetorical Appeals in Fundraising Direct Mail Letters”. In Discourse in the Professions, Ulla Connor and Thomas A. Upton (eds.), 257-286. Amsterdam: John Benjamins. Connor, Ulla and Mayberry, Susan. 1995. “Learning Discipline-specific Academic Writing: A Case Study of a Finnish Graduate Student in the United States”. In Academic Writing: Intercultural and Textual Issues, Eija Ventola and Anna Mauranen (eds), 231-253. Amsterdam: John Benjamins. Connor, Ulla and Upton, Thomas A. 2004. Discourse in the Professions: Perspectives from Corpus Linguistics. Amsterdam: John Benjamins. Conrad, Susan. 2002. “Corpus Linguistic Approaches for Discourse Analysis”. Annual Review of Applied Linguistics 22: 75-95. Conrad, Susan and Biber, Douglas. 2000. “Adverbial Marking of Stance in Speech and Writing”. In Evaluation in Text: Authorial Stance and the Construction of Discourse, Susan Hunston and Geoff Thompson (eds), 56-73. Oxford: Oxford University Press. Cowie, Anthony and Heaton, Brian (eds). 1977. English for Academic Purposes. London: British Association for Applied Linguistics/ Special English Language Materials for Overseas University Students. Coxhead, Averil. 2000. “A New Academic Word List”. TESOL Quarterly 34 (2): 213-238. Crawford, Mary. 1995. Talking Difference: On Gender and Language. London: Sage. Crismore, Avon. 1989. Talking with Readers: Metadiscourse as Rhetorical Act. New York: Peter Lang. Crismore, Avon. 2004. “Pronouns and Metadiscourse as Interpersonal Rhetorical Devices in Fundraising Letters: A Corpus Linguistic Analysis”. In Discourse in the Professions, Ulla Connor and Thomas A. Upton (eds.), 307-330. Amsterdam: John Benjamins. Crismore, Avon and Farnsworth, Rodney. 1990. “Metadiscourse in Popular and Professional Discourse”. In The Writing Scholar: Studies in Academic Discourse, Walter Nash (ed.), 118-36. Newbury Park: Sage Publications. Crismore, Avon, Markkanen, Raija and Steffensen, Margaret. 1993. “Metadiscourse in Persuasive Writing: A Study of Texts Written by American and Finnish University Students”. Written Communication 10 (1): 39-71. Cronin, Blaise. 1984. The Citation Process: The Role and Significance of Citations in Scientific Communication. London: Taylor Graham. Cronin, Blaise. 1995. The Scholar’s Courtesy: The Role of Acknowledgements in the Primary Communication Process. London: Taylor Graham. Cronin, Blaise and Shaw, Deborah. 2002. “Identity-creators and Image-makers: Using Citation Analysis and Thick Description to Put Authors in their Place”. Scientometrics 54 (1): 31-49. Crystal, David. 1997. English as a Global Language. Cambridge: Cambridge University Press. <?page no="319"?> 319 Crystal, David. 2003. The Cambridge Encyclopaedia of the English Language. Cambridge: Cambridge University Press. Dahl, Trine. 2004. “Textual Metadiscourse in Research Articles: A Marker of National Culture or of Academic Discipline? ”. Journal of Pragmatics 36 (10): 1807-1825. Darian, Steven. 1995. “Hypotheses in Introductory Science Texts”. International Review of Applied Linguistics in Language Teaching 33 (2): 83-108. Davidson, Cathy. 1996. “Critical Fictions”. PMLA 111 (5): 1069-1072. Day, Robert. 1979. How to Write and Publish a Scientific Paper. Philadelphia: ISI Press. De Swaan, Abram. 2001. “English in the Social Sciences”. In The Dominance of English as a Language of Science: Effects on Other Languages and Language Communities, Ulrich Ammon (ed.), 71-84. Berlin: Mouton De Gruyter. De Swaan, Abram. 1993. “The Evolving European Language System: A Theory of Communication Potential and Language Competition”. Special Issue of International Political Science Review 14: 241-55. Dewey, John. 1925. Experience and Nature. New York: Dover. Dietz, Gunther. 2001. “The Pragmatics of Scientific Titles Formulated as Questions”. In Approaches to the Pragmatics of Scientific Discourse, András Kertész (ed.), 19-33. Frankfurt am Main: Peter Lang. Dillon, George. 1981. Constructing Texts. Bloomington: Indiana University Press. Dillon, George. 1986. Rhetoric as Social Imagination: Explorations in the Interpersonal Function Of Language. Bloomington: Indiana University Press. Drescher, Martina. 2004. “Sprache der Wissenschaft, Sprache der Vernunft? Zum affektleeren Stil in der Wissenschaft”. Paper. Colloquium “Wissenschaftsstile im Kontrast”. Albert-Ludwigs-Universität, Freiburg. 29 April 2004. Drosdowski, Günther and Eisenberg, Peter. 1995. Duden Grammatik der deutschen Gegenwartssprache. 5th ed. Mannheim: Duden Verlag. Du Boulay, Doreen. 2001. “Writing for Postgraduates”. Department homepage. University of Sussex, Brighton. Sussex Language Institute. 29 January 2006 <http: / / www.sussex.ac.uk/ languages/ 1-6-8-4-3.html> Dudley-Evans, Tony and Henderson, Willie. 1990. “The Organisation of Article Introductions: Evidence of Change in Economics Writing”. In The Language of Economics: The Analysis of Economics Discourse, Tony Dudley-Evans and Willie Henderson (eds), 67-78. London: Modern English Publications. Duszak, Anna. 1994. “Academic Discourse and Intellectual Styles”. Journal of Pragmatics 21 (3): 291-313. Duszak, Anna. 1997. Culture and Styles of Academic Discourse. Berlin, New York: Mouton de Gruyter. Edmondson, Willis, House, Juliana, Kaspar, Gabriele and Stemmer, Brigitte. 1984. “Learning the Pragmatics of Discourse: A Project Report”. Applied Linguistics 5: 113-127. Eggington, William. 1987. “Written Academic Discourse in Korean: Implications for Effective Communication”. In Writing across Languages: Analysis of L2 Text, Ulla Connor and Robert Kaplan (eds), 153-168. Reading: Addison- Wesley. <?page no="320"?> 320 Ehlich, Konrad. 1982. “Deiktische und phorische Prozeduren beim literarischen Erzählen”. In Erzählforschung. Ein Symposium, Eberhard Lämmert (ed.), 112.129. Stuttgart: Metzler. Ehlich, Konrad. 1992. “Scientific Texts and Deictic Structures”. In Cooperating with Written Texts, Dieter Stein (ed.), 201-229. Berlin: de Gruyter. Ehlich, Konrad (ed.). 2003. Mehrsprachige Wissenschaft europäische Perspektiven. Eine Konferenz zum Europäischen Jahr der Sprachen 2001. München, Institut für Deutsch als Fremdsprache / Transnationale Germanistik. Enkvist, Nils Erik. 1975. Basic Concepts in Text Linguistics. Helsinki: Gaudeamus. Enkvist, Nils Erik. 1991. “Discourse Type, Text Type, and Cross-Cultural Rhetoric”. In Empirical Research in Translation and Intercultural Studies, Sonja Tirkkonen-Condit (ed.), 5-16. Language in Performance 5. Tübingen: Gunter Narr. Evers, Kathinka. 2001. Standards for Ethics and Responsibility in Science: An Analysis and Evaluation of Their Content, Background and Function. International Council for Science, Standing Committee on Responsibility and Ethics in Science (SCRES). September 2001. 3 February 2005 <http: / / www.icsu.org/ Gestion/ img/ ICSU_DOC_DOWNLOAD/ 218_DD_ FILE_Background_1.pdf> Fandrych, Christian. 2001. “‘Dazu soll später noch mehr gesagt werden’: Lexikalische Aspekte von Textkommentaren in englischen und deutschen wissenschaftlichen Artikeln”. In Proper Words in Proper Places : Studies in Lexicology and Lexicography in Honour of William Jervis Jones, Marie C. Davies, John L. Flood and David N. Yeandle (eds), 375-398. Stuttgart: Heinz. Fandrych, Christian. 2003. “Herausarbeiten vs. illustrate: Kontraste bei der Versprachlichung von Sprechhandlungen in der englischen und deutschen Wissenschaftssprache”. In Mehrsprachige Wissenschaft europäische Perspektiven, Konrad Ehlich (ed.). München: Institut für Deutsch als Fremdsprache / Transnationale Germanistik. 1 January 2006 <http: / / www.eurosprachenjahr.de/ onlinepub.htm> Fandrych, Christian. 2005. “‘Räume’ und ‘Wege’ der Wissenschaft: Einige zentrale Konzeptualisierungen von wissenschaftlichem Schreiben im Deutschen und Englischen”. In Zwischen Lexikon und Text: lexikalische, stilistische und textlinguistische Aspekte, Ulla Fix, Gotthard Lerchner, Marianne Schröder and Hans Wellmann (eds), 20-33. Leipzig: Verlag der Sächsischen Akademie der Wissenschaften. Fandrych, Christian and Graefen, Gabriele. 2002. “Text-commenting Devices in German and English Academic Articles”. Multilingua 21: 17-43. Fasulo, Alessandra and Zucchermaglio, Cristina. 2002. “My Selves and I: Identity Markers in Work Meeting Talk”. Journal of Pragmatics 34 (9): 1119-1144. Fiedler, Sabine. 1992. “Die pädagogische Rezension im Englischen und Esperanto”. In Kontrastive Fachsprachenforschung, Klaus-Dieter Baumann and Hartwig Kalverkämper (eds), 147-161. Tübingen: Narr. Fillmore, Charles. 1975. The Santa Cruz Lectures on Deixis. Bloomington: Indiana University Linguistics Club. <?page no="321"?> 321 Fishman, Joshua. 1992. “Sociology of English as an Additional Language”. In The Other Tongue: English Across Cultures, Braj Kachru (ed.), 19-26. Urbana: Univ. of Illinois Press. Fishman, Joshua. 1977a. “Knowing, Using and Liking English as an Additional Language”. In The Spread of English, Joshua Fishman, Robert L. Cooper and Andrew Conrad (eds), 309-326. Rowley: Newbury House. Fishman, Joshua. 1977b. “English in the Context of International Societal Bilingualism”. In The Spread of English, Joshua Fishman, Robert L. Cooper and Andrew Conrad (eds), 329-336. Rowley: Newbury House. Fishman, Joshua, Cooper, Robert L. and Conrad, Andrew W (eds).1977. The Spread of English. Rowley: Newbury House. Fix, Ulla, Habscheid, Stephan and Klein, Josef (eds). 2001. Zur Kulturspezifik von Textsorten. Tübingen: Stauffenburg. Fløttum, Kjersti, Dahl, Trine and Kinn, Torodd. 2006. Academic Voices: Across Languages and Disciplines. Amsterdam: John Benjamins. Fløttum, Kjersti, Kinn, Torodd and Dahl, Trine. 2006. “We Now Report On …” Versus “Let Us Now See How…”: Author Roles and Interaction with Readers in Research Articles”. In Academic Discourse Across Disciplines, Ken Hyland and Marina Bondi (eds), 203-224. Bern: Peter Lang. Flowerdew, John. 2001. “Attitudes of Journal Editors to Nonnative Speaker Contributions”. TESOL Quarterly 35 (1): 121-150. Flowerdew John. 2002a. “Ethnographically Inspired Approaches to Academic Discourse”. In Academic Discourse, John Flowerdew (ed.), 235-252. London: Longman. Flowerdew John. 2002b. “Corpus-based Analyses in EAP”. In Academic Discourse John Flowerdew (ed.), 95-114. London: Longman. Flowerdew, John (ed.). 2002c. Academic Discourse. London: Longman. Flowerdew, John and Dudley-Evans, Tony. 2002. “Genre Analysis of Editorial Letters to International Journal Contributors”. Applied Linguistics 23 (4): 463- 489. Flowerdew, John and Miller, Lindsay. 1992. “On the Notion of Culture in Second Language Lectures”. TESOL Quarterly 29 (2): 345-73. Flowerdew, John and Peacock, Matthew. 2001. Research Perspectives on English for Academic Purposes. Cambridge: Cambridge University Press. Flowerdew, Lynne. 2000. “Using a Genre-based Framework to Teach Organizational Structure in Academic Writing”. ELT Journal 54 (4): 368-378. Flowerdew, Lynne. 2004. “The argument for using English specialized corpora to understand academic and professional language”. In Discourse in the Professions, Ulla Connor and Thomas A. Upton (eds.), 1-33. Amsterdam: John Benjamins. Flowerdew, Lynne. 2005. “An Integration of Corpus-based and Genre-based Approaches to Text Analysis in EAP/ ESP: Countering Criticisms Against Corpus-based Methodologies”. English for Specific Purposes 24 (3): 321-332. Fortanet, Inmaculada. 2004a. “Verbal Stance in Spoken Academic Discourse”. In Academic Discourse: New Insights into Evaluation, Gabriella Del Lungo Camiciotti and Elena Tognini-Bonelli (eds), 99-120. Bern: Peter Lang. <?page no="322"?> 322 Fortanet, Inmaculada. 2004b. “The Use of ‘we’ in University Lectures: Reference and Function”. English for Specific Purposes 23: 45-66. Frath, Pierre. 2001. “Hégémonie de l'anglais: fantasmes et dangers”. Les Langues modernes 3 : 31-38. Freedman, David, Pisani, Robert and Purves, Roger. 1997. Statistics. 3rd ed. New York: W.W. Norton, Inc. Fromkin, Victoria, Rodman, Robert, Collins, Peter and Blair, David. 1996. An Introduction to Language. 3rd ed. Sydney: Harcourt Brace. Fry, Jennifer. 2003. “Paper Requirements”. Homepage. King’s College, Pennsylvania, Department of History. 22 August 2003 <http: / / www.kings.edu/ jrfry/ guidelines.html> Furneaux, Clare, Robinson, Pauline and Tonkyn, Alan. 1988. “Making Friends and Influencing Tutors: Strategies for Promoting Acculturation in the EAP Classroom”. In Individualization and Autonomy in Language Learning, Arthur Brookes and Peter Grundy (eds), 88-101. London: The British Council. Galtung, Johann. 1981. “Structure, Culture and Intellectual Style. An Essay Comparing Saxonic, Teutonic, Gallic and Nipponic Approaches”. Social Science Information 20 (6): 817-856. Garfield, Eugene. 1990. “The Languages of Science Revisited: English (Only) Spoken Here? ”. Current Comments 32: 280-294. Garside, Roger, Fligelstone, Steve and Botley, Simon. 1997. “Discourse Annotation: Anaphoric Relations in Corpora”. In Corpus Annotation: Linguistic Information from Computer Text Corpora, Roger Garside, Geoffrey Leech and Tony McEnery (eds), 66-84. New York : Addison Wesley Longman. Garzone, Giuliana and Santulli, Francesca. 2004. “What Can Corpus Linguistics do for Critical Discourse Analysis? ”. In Corpora and Discourse, Alan Partington, John Morley and Louann Haarman (eds), 351-368. Bern: Peter Lang. Gavioli, Laura. 2005. Exploring Corpora for ESP Learning. Amsterdam: John Benjamins. Gee, James P. 1989. “What is Literacy? ”. Journal of Education 171 (1): 18.25. Geertz, Clifford. 1973. The Interpretation of Cultures: Selected Essays. New York: Basic Books. Geertz, Clifford. 1988. Works and Lives: The Anthropologist as Author. Stanford: Stanford University Press. Geisler, Cheryl. 1994. Academic Literacy and the Nature of Expertise: Reading, Writing and Knowing in Academic Philosophy. Mahwah: Lawrence Erlbaum Associates. Gerzymisch-Arbogast, Heidrun. 1993. “Contrastive Scientific and Technical Register as a Translation Problem”. In Scientific and Technical Translation, Leland D. Wright and Sue Ellen Wright (eds), 21-51. American Translators Association Scholarly Monograph Series. Vol. 6. New York: SUNY. Gerzymisch-Arbogast, Heidrun. 1997a. “Translating Cultural Specifics: Macroand Microstructural Decisions”. In Machine Translation and Translation Theory, Christa Hauenschild and Susanne Heizmann (eds), 51-67. Berlin: Mouton de Gruyter. <?page no="323"?> 323 Gerzymisch-Arbogast, Heidrun. 1997b. “Der Leserbezug in Sigmund Freuds psychoanalytischen Schriften im Spiegel der englischen Übersetzungen”. In Modelle der Translation - Models of Translation, Gerd Wotjak and Heide Schmidt (eds), 231-233. Frankfurt am Main: Vervuert. Giannoni, Davide Simone. 2002. “Worlds of Gratitude: A Contrastive Analysis of Acknowledgement Texts in English and Italian Academic Articles”. Applied Linguistics 23 (1), 1-31. Gibaldi, Joseph. 2003. MLA Handbook for Writers of Research Papers. 6th ed. New York: Modern Language Association of America. Gibbs, Wayt. 1995. “Lost Science in the Third World”. Scientific American 273: 92-9. Gjesdal, Anje Müller. 2005. “Scientific Writing and Gender: The Case of Personal Pronouns”. Akademisk Prosa 3: 101-112. Gläser, Rosemarie. 1990. Fachtextsorten im Englischen. Tübingen: Gunter Narr. Gläser, Rosemarie. 1998. “47. Fachtextsorten der Wissenschaftssprachen I: Der wissenschaftliche Zeitschriftenaufsatz”. In Fachsprachen. Ein internationales Handbuch zur Fachsprachenforschung und Terminologiewissenschaft, Lothar Hoffmann, Hartwig Kalverkämper, Herbert Ernst Wiegand, Christian Galinski and Werner Hüllen (eds), 482-488. Berlin: Walter de Gruyter. Gnutzmann, Claus and Oldenburg, Hermann. 1991. “Contrastive Text Linguistics in LSP Research: Theoretical Considerations and Some Preliminary Findings”. In Subjectoriented Texts: Languages for Special Purposes and Text Theory, Hartmut Schröder (ed.),103-136. Berlin, New York: Walter de Gruyter. Godin, Benoit. 1995. Endangered Species? Une nouvelle estimation de la part du français dans les communications scientifiques. Quèbec: Conseil de la langue français. Gong, Gwendolyn and Dragga, Sam. 1995. A Writer’s Repertoire. New York: Harper Collins. Gordon, Raymond G., Jr. (ed.). 2005. Ethnologue: Languages of the World. 15th ed. Dallas, Texas: SIL International. 14 August 2005 <http: / / www.ethnologue.com/ > Gosden, Hugh. 1995. “Success in Research Article Writing and Revision: A Social Constructionist Perspective”. English for Specific Purposes 14 (1): 35-57. Gosden, Hugh. 2003. “Why Not Give us the Full Story? : Functions of Referees’ Comments in Peer Reviews of Scientific Research Papers”. Journal of English for Academic Purposes 2: 87-101. Götze, Lutz and Hess-Lüttich, Ernest. 2002. Bertelsmann Grammatik der deutschen Sprache: Sprachsystem und Sprachgebrauch. Gütersloh: Bertelsmann Lexikon Verlag. Grabe, William and Kaplan, Robert. 1996. Theory and Practice of Writing: An Applied Linguistic Perspective. London: Longman. Grabe, William and Kaplan, Robert. 1997. “On the Writing of Science and the Science of Writing: Hedging in Science Text and Elsewhere”. In Hedging and Discourse: Approaches to the Analysis of a Pragmatic Phenomenon in Academic Texts, Raija Markkanen and Hartmut Schröder (eds), 151-167. Berlin: de Gruyter. Graddol, David. 1997. The Future of English? . London: British Council. <?page no="324"?> 324 Graddol, David. 2004. “The Future of Language”. Science 303 (5662): 1329-1331. Graddol, David and Swan, Joan. 1989. Gender Voices. Cambridge: Blackwell. Graefen, Gabriele. 1994. “Wissenschaftstexte im Vergleich: Deutsche Autoren auf Abwegen? ”. In Texte und Diskurse. Methoden und Forschungsergebnisse der funktionalen Pragmatik, Gisela Brünner and Gabriele Graefen (eds), 135-157. Opladen: Westdeutscher Verlag. Graefen, Gabriele. 1997. Der Wissenschaftliche Artikel - Textart und Textorganisation. Frankfurt am Main: Peter Lang. Graefen, Gabriele. 2000a. “‘Hedging’ als neue Kategorie? Ein Beitrag zur Diskussion”. 2 June 2004 http: / / www.sw2.euv-frankfurt-o.de/ Forschung/ Hedging/ graefen.html Graefen, Gabriele. 2000b. “Textkommentierung in deutschen und englischen wissenschaftlichen Artikeln”. In Sprache und Kultur, Horst Dieter Schlosser (ed.), 113-124. Frankfurt am Main: Peter Lang. Granger, Sylviane, Lerot, Jacques and Petch-Tyson, Stephanie (eds). 1993. Corpus-based Approaches to Contrastive Linguistics and Translation Studies. Amsterdam: Rodopi. Gredler, Margaret. E. 1997. Learning and Instruction: Theory into Practice. 3rd ed. Upper Saddle River, NJ: Prentice-Hall. Grice, Paul H. 1975. “Logic and Conversation”. In Syntax and Semantics Volume 3: Speech Acts, Peter Cole and Jerry L. Morgan (eds), 41-58. New York: Academic Press. Grieco, Elizabeth M. and Cassidy, Rachel C. 2001. Census 2000 Brief: Overview of Race and Hispanic Origin. U.S. Department of Commerce, Economics and Statistics Administration, U.S. Census Bureau. Washington D.C.: U.S. Census Bureau 2001. 27 January 2006 <http: / / www.census.gov/ prod/ 2001pubs/ c2kbr01-1.pdf> Gries, Stefan T. and Anatol Stefanowitsch. 2004. “Extending Collostructional Analysis: A Corpus-based Perspective on ‘Alternations’”. International Journal of Corpus Linguistics 9 (1): 97-129. Griffig, Thomas. 2005. Intertextualität in linguistischen Fachaufsätzen des Englischen und Deutschen. Diss. Rheinisch-Westfälische Technische Hochschule Aachen, 2005. 29 January 2006 <http: / / deposit.ddb.de/ cgibin/ dokserv? idn=977880591&dok_var=d1&dok_ext=pdf&filename=97788059 1.pdf> Gross, Paul R. and Levitt, Norman. 1998. Higher Superstition: The Academic Left and its Quarrels with Science. Baltimore: John Hopkins. Grundy, Peter and Jiang, Yan. 2001. “Ideological Ground and Relevant Interpretation in a Cognitive Semantics”. In Language and Ideology, René Dirven, Bruce Hawkins and Esra Sandikcioglu (eds), 107-140. Amsterdam: John Benjamins. Gumperz, John (ed.). 1982. Language and Social Identity. Cambridge: Cambridge University Press. Gunnarsson, Britt-Louise. 2001. “Swedish, English, French or German: The Language Situation at Swedish Universities”. In The Dominance of English as a Language of Science: Effects on Other Languages and Language <?page no="325"?> 325 Communities, Ulrich Ammon (ed.), 287-316. Berlin, New York: Mouton de Gruyter. Haarman, Louann. 2004. “‘John, What’s Going On? ’”. In Corpora and Discourse, Alan Partington, John Morley and Louann Haarman (eds), 71-88. Bern: Peter Lang. Haas, Christina. 1994. “Learning to Read Biology: One Student’s Rhetorical Development in College”. Written Communication 11 (1): 43-84. Habermas, Jürgen. 1976a. “Was heißt Universalpragmatik? ”. In Sprachpragmatik und Philosophie, Karl-Otto Apel (ed.), 174-272. Frankfurt am Main: Suhrkamp. Habermas, Jürgen. 1976b. Zur Rekonstruktion des Historischen Materialismus. Frankfurt am Main: Suhrkamp. Habermas, Jürgen. 1981. Theorie des kommunikativen Handelns (Vol.1: Handlungsrationalität und gesellschaftliche Rationalisierung, Vol. 2: Zur Kritik der funktionalistischen Vernunft). Frankfurt am Main: Suhrkamp. Haggan, Madeline. 2004. “Research Paper Titles in Literature, Linguistics and Science: Dimensions of Attraction”. Journal of Pragmatics 36: 293-317. Halliday, Michael. 1973. Explorations in the Functions of Language, London: Edward Arnold. Halliday, Michael. 1985. An Introduction to Functional Grammar. London: Arnold. Halliday, Michael and Hasan, Ruqaiya. 1976. Cohesion in English. London: Longman. Harding, Sandra. 1991. Whose Science? Whose Knowledge? Thinking from Women’s Lives. Ithaca, NY: Cornell University Press. Hardt-Mautner, Gerlinde. 1995. “Only Connect: Critical Discourse Analysis and Corpus Linguistics”. UCREL Technical Papers 6. 16 January 2006 < http: / / www.comp.lancs.ac.uk/ ucrel/ tech_papers.html> Harwood, Nigel. 2005a. “‘I Hoped to Counteract the Memory Problem, but I Made no Impact Whatsoever’: Discussing Methods in Computing Science Using I”. English for Specific Purposes 24 (3): 243-267. Harwood, Nigel. 2005b. ““Nowhere has Anyone Attempted…In This Article I Aim to Do Just That’. A Corpus-based Study of Self-promotional I and WE in Academic Writing Across Four Disciplines”. Journal of Pragmatics 37 (8): 1207-1231. Harwood, Nigel. 2005c. “‘We Do Not Seem to Have a Theory...The Theory I Present Here Attempts to Fill This Gap’: Inclusive and Exclusive Pronouns Academic Writing”. Applied Linguistics 26 (3): 343-375. Harwood, Nigel and Hadley, Gregory. 2004. “Demystifying Institutional Practices: Critical Pragmatism and the Teaching of Academic Writing”. English for Specific Purposes 23 (4): 355-377. Heidolph Karl Erich, Flämig, Walter and Motsch, Wolfgang. 1981. Grundzüge einer deutschen Grammatik. Berlin: Akademie-Verlag. Helbig, Gerhard and Buscha, Joachim. 2001. Langenscheidt Deutsche Grammatik: ein Handbuch für den Ausländerunterricht. München: Langenscheidt. Hellinger, Marlis and Ammon, Ulrich (eds). 1996. Contrastive Sociolinguistics. Berlin: Gruyter. <?page no="326"?> 326 Helms-Park, Rena and Stapleton, Paul. 2003. “Questioning the Importance of Individualized Voice in Undergraduate L2 Argumentative Writing: An Empirical Study with Pedagogical Implications”. Journal of Second Language Writing 12: 245-265. Hiatt, Mary. 1977. The Way Women Write. New York: Teachers’ College Press. Hinds, John. 1980. “Japanese Expository Prose”. Papers in Linguistics 13 (1): 117- 158. Hinds, John. 1987. “Reader Versus Writer Responsibility: A New Typology”. In Writing Across Languages: Analysis of L2 Text, Ulla Connor and Robert Kaplan (eds.), 141-152. Reading: Addison-Wesley. Hinds, John. 1990. “Inductive, Deductive, Quasi-inductive: Expository Writing in Japanese, Korean, Chinese and Thai”. In Coherence in Writing: Research and Pedagogical Perspectives, Ulla Connor and Ann Johns (eds), 87-110. Alexandria: TESOL. Hoffmann, Lothar. 1992. “Vergleiche in der Fachsprachenforschung”. In Kontrastive Fachsprachenforschung, Klaus-Dieter Baumann and Hartwig Kalverkämper (eds), 95-107. Tübingen: Gunter Narr. Hoffmann, Sebastian. 2005. Grammaticalization and English Complex Prepositions. A Corpus-Based Study. London: Routledge. Holdstein, Deborah H. and Bleich, David (eds). 2001. Personal Effects: The Social Character of Scholarly Writing. Logan: Utah State University Press. Holmes, Janet. 1990. “Hedges and Boosters in Women’s and Men’s Speech”. Language and Communication 10 (3): 185-205. Holmes, Janet. 1993. “Women’s Talk: The Question of Sociolinguistic Universals”. Australian Journal of Communication 20 (3): 125-149. Holmes, Janet. 1995. Women, Men and Politeness. London: Longman. Hood, Sue. 2004. Appraising Research: Taking a Stance in Academic Writing. Diss. University of Technology Sydney, 2004. Houghton, Diane and Hoey, Michael. 1983. “Linguistics and Written Discourse: Contrastive Rhetorics”. In Annual Review of Applied Linguistics III, Robert Kaplan, Randall L. Jones & G. Richard Tucker (eds), 2-22. Rowley: Newbury House. House, Juliane. 1997. “Interkulturelle Pragmatik und Übersetzen”. In Modelle der Translation - Models of Translation, Gerd Wotjak and Heide Schmidt (eds), 21-40. Frankfurt am Main: Vervuert. House, Juliane. 1999. “Zur Relevanz kontrastiv-pragmatischer und interkultureller Diskursanalysen für das Fachübersetzen”. In Wege der Übersetzungs- und Dolmetschforschung, Heidrun Gerzymisch-Arbogast, Daniel Gile, Juliane House and Annely Rothkegel (eds), 43-55. Jahrbuch Übersetzen und Dolmetschen. Tübingen: Narr. House, Juliane. 2003a. “English as Lingua Franca and its Influence on Discourse Norms in Other Languages”. In Translation Today, Gunilla Anderman and Margaret Rogers (eds), 168-180. Clevedon: Multilingual Matters. House, Juliane. 2003b. “English as Lingua Franca: A Threat to Multilingualism? ”. Journal of Sociolinguistics 7 (4): 556-579. <?page no="327"?> 327 Huckin, Thomas. 2001. “Abstracting from Abstracts”. In Academic Writing in Context Martin Hewings (ed.), 93-103. Birmingham: University of Birmingham Press. Huckin, Thomas and Olsen, Leslie. 1991. Technical Writing and Professional Communication for Non-Native Speakers. New York: McGraw-Hill. Huddleston, Rodney and Pullum, Geoffrey. 2002. Cambridge Grammar of the English Language. Cambridge: Cambridge University Press. Hughes, Richard E. and Duhamel, P. Albert. 1962. Rhetoric: Principles and Usage. Englewood Cliffs: Prentice-Hall. Humboldt-Universität zu Berlin. 2003. Forschungsevaluation: Leitfaden. Berlin: Humboldt-Universität zu Berlin, 2003. 3 February 2006 <http: / / www.huberlin.de/ step/ leitfaden_fo.pdf> Hundt, Marianne. 1998. New Zealand English Grammar - Fact or Fiction? A Corpus-Based Study in Morphosyntactic Variation. Amsterdam, Philadelphia: Benjamins. Hunston, Susan. 1994. “Evaluation and Organization in a Sample of Written Academic Discourse”. In Advances in Written Text Analysis, Malcolm Coulthard (ed), 191-218. London: Routledge. Hunston, Susan. 2002. Corpora in Applied Linguistics. Cambridge: Cambridge University Press. Hunston, Susan. 2004. “Counting the Uncountable: Problems of Identifying Evaluation in a Text and in a Corpus”. In Corpora and Discourse, Alan Partington, John Morley and Louann Haarman (eds), 157-188. Bern: Peter Lang. Hunston, Susan and Sinclair, John. 2000. “A Local Grammar of Evaluation”. In Evaluation in Text: Authorial Stance and the Construction of Discourse, Susan Hunston and Geoff Thompson (eds), 74-100. Oxford: Oxford University Press. Hunston, Susan and Thompson, Geoff (eds). 2000. Evaluation in Text: Authorial Stance and the Construction of Discourse. Oxford: Oxford University Press. Hutz, Matthias. 1997. Kontrastive Fachtextlinguistik für den fachbezogenen Fremdsprachenunterricht. Fachzeitschriften der Psychologie im interlingualen Vergleich. Trier: Wissenschaftlicher Verlag. Hutz, Matthias. 2001. “‘Insgesamt muss ich leider zu einem ungünstigen Urteil kommen’ - Zur Kulturspezifik wissenschaftlicher Rezensionen im Deutschen und Englischen”. In Zur Kulturspezifik von Textsorten, Ulla Fix, Stephan Habscheid and Josef Klein (eds), 109-130. Tübingen: Stauffenburg. Huwyler, Friedrich. 2002. Nachqualifikation der Lehrpersonen der Primarschule im Fach Englisch; Verpflichtungskredit: Bericht und Vorlage an den Kantonsrat. Beschluss Nr. 1040/ 2002. Regierungsrat des Kantons Schwyz. 14 June 2005 <http: / / www.sz.ch/ information/ Mm_2002/ Englisch_Bericht.pdf> Hyland, Ken. 1995. “The Author in the Text: Hedging Scientific Writing”. Hong Kong Papers in Linguistics and Language Teaching 18: 33-42. 5 January 2006 <http: / / sunzi1.lib.hku.hk/ hkjo/ article.jsp? book=4&issue=40007> Hyland, Ken. 1996. “Writing Without Conviction? Hedging in Science Research Articles”. Applied Linguistics 17 (4): 433-454. Hyland, Ken. 1998. Hedging in Scientific Research Articles. Amsterdam: John Benjamins. <?page no="328"?> 328 Hyland, Ken. 1999a. “Disciplinary Discourse: Writer Stance in Research Articles”. In Writing: Texts, Processes and Practices, Christopher Candlin and Ken Hyland (eds), 99-121. Essex : Longman. Hyland, Ken. 1999b. “Academic Attribution: Citation and the Construction of Disciplinary Knowledge”. Applied Linguistics 20 (5): 541-567. Hyland, Ken. 1999c. “Talking to Students: Metadiscourse in Introductory Textbooks”. English for Specific Purposes 18 (1): 3-26. Hyland, Ken. 1999d. “Persuasion in Academic Articles”. Perspectives 11: 73-103. 19 December 2005 <http: / / sunzi1.lib.hku.hk/ hkjo/ article.jsp? book= 10&issue=100020> Hyland, Ken. 2000. Disciplinary Discourses: Social Interactions in Academic Writing. Harlow: Longman. Hyland Ken. 2002a. Teaching and Researching Writing. London, Longman. Hyland, Ken. 2002b. “Directives: Argument and Engagement in Academic Writing”. Applied Linguistics 23 (2): 215-239. Hyland, Ken. 2002c. “Authority and Invisibility: Authorial Identity in Academic Writing”. Journal of Pragmatics 34 (8): 1091-1112. Hyland Ken. 2002d. “Activity and Evaluation: Reporting Practices in Academic Writing”. In Academic Discourse John Flowerdew (ed.), 115-130. London, Longman. Hyland, Ken. 2003. “Self-citation and Self-reference: Credibility and Promotion in Academic Publication”. Journal of the American Society for Information Science and Technology 54 (3): 251-259. Hyland, Ken. 2004. “A Convincing Argument: Corpus Analysis and Academic Persuasion”. In Discourse in the Professions, Ulla Connor and Thomas A. Upton (eds.), 87-114. Amsterdam: John Benjamins. Hyland, Ken. 2005. “Stance and Engagement: A Model of Interaction in Academic Discourse”. Discourse Studies 7 (2): 173-192. Hyland, Ken and Bondi, Marina. 2006. Academic Discourse Across Disciplines. Bern: Peter Lang. Hyland Ken and Hamp-Lyons Liz. 2002. “EAP: Issues and Directions”. Journal of English for Academic Purposes 1 (1): 1-12. Hyland, Ken and Tse, Polly. 2004. “Metadiscourse in Academic Writing: A Reappraisal”. Applied Linguistics 25 (2): 156-177. Intaraprawat, Puangpen and Steffensen, Margaret S. 1995. “The Use of Metadiscourse in Good and Poor ESL Essays”. Journal of Second Language Writing 4 (3): 253-273. Ivani , Roz. 1998. Writing and Identity: The Discoursal Construction of Identity in Academic Writing. Amsterdam: John Benjamins. Iverson, Cheryl. 2002. “US Medical Journal Editors’ Attitudes toward Submissions from Other Countries”. Science Editor 25: 75-8. Jacobsson, Bengt. 1994. “Recessive and Emergent Uses of Modal Auxiliaries in English”. English Studies 72 (2): 166-182. Jaffe, Sam. 2003. “No Pardon for Poor English in Science”. The Scientist 17 (5): 44-5. <?page no="329"?> 329 Jakobs, Eva-Maria. 1999. Textvernetzung in den Wissenschaften. Zitat und Verweis als Ergebnis rezeptiven, reproduktiven und produktiven Handelns. Tübingen: Niemeyer. Jakobs, Eva-Maria. 2003. “Reproductive Writing: Writing from Sources”. Journal of Pragmatics 35: 893-906. Jakobsen, Roman. 1995. On Language. Ed. Linda R. Waugh and Monique Monville-Burston. Cambridge: Harvard University Press. Jamsen, Kirsten, Knutson, Susan, Scepanski, Rachel and Strain, Kimberly. 2004. “Undercurrents: EFL Student Voices in American ESL Academic Writing”. In Proceedings of the 2004 Midwest Writing Centers’ Association Regional Conference. Minnesota: Midwest Writing Centers Association 2004. 16 November 2004 <http: / / www.ku.edu/ ~mwca/ conference/ proceedings/ Undercurrents.pdf> Jespersen, Otto. 1924. The Philosophy of Grammar. London: Allen & Unwin. Johns, Ann M. 1997. Text, Role and Context: Developing Academic Literacies. Cambridge: Cambridge University Press. Jordan, Michael P. 1988. “The Power of Negation in English: Text, Context and Relevance”. Journal of Pragmatics 29: 705-52. Jordan, Robert R. 1997. English for Academic Purpose: A Guide and Resource Book for Teachers. Cambridge: Cambridge University Press. Kachru, Braj (ed.). 1992. The Other Tongue: English Across Cultures. 2nd ed. Urbana: University of Illinois Press. Kaplan, David. 1989. “Demonstratives: An Essay on the Semantics, Logic, Metaphysics and Epistemology of Demonstratives and Other Indexicals”. In Themes from Kaplan, Joseph Almog, John Perry and Howard Wettstein (eds), 481-563. New York and Oxford: Oxford University Press. Kaplan, Robert. 1966. “Cultural Thought Patterns in Intercultural Education”. Language Learning 16: 1-20. Kaplan, Robert. 1987. “Cultural Thought Patterns Revisited”. In Writing Across Languages: Analysis of L2 Text, Ulla Connor and Robert Kaplan (eds), 9-21. Reading, MA: Addison-Wesley. Kaplan, Robert. 1988. “Contrastive Rhetoric and Second Language Learning: Notes Toward a Theory of Contrastive Rhetoric”. In Writing Across Languages and Cultures. Issues in Contrastive Rhetoric, Alan Purves (ed.), 275-304. Newbury Park, CA: Sage. Kaplan, Robert. 2000. “Contrastive Rhetoric and Discourse Analysis: Who Writes What to Whom? When? In What Circumstances? ”. In Discourse and Social Life, Srikand Sarangi and Malcolm Coulthard (eds), 82-101. London: Longman. Kärkkäinen, Elise. 1987. “Towards a Pragmatic Description of Modality”. In Proceedings from the Third Nordic Conference for English Studies, Ishrat Lindblad and Magnus Ljung (eds), 149-161. Stockholm: Almqvist & Wicksell. Kearsey, John and Turner, Sheila. 1999. “Evaluating Textbooks: The Role of the Genre Analysis”. Research in Science and Technological Education 17 (1): 35- 44. Keller, Eric. 1979. “Gambits: Conversational Strategy Signals”. Journal of Pragmatics 3: 219-38. <?page no="330"?> 330 Kemp, Paul. 2002. “Why the Journal’s Structure is Different”. Limnology and Oceanography: Methods. 26 August 2005 <http: / / aslo.org/ lomethods/ editor/ structure.html> Kennedy, Graeme. 1998. An Introduction to Corpus Linguistics. London: Longman. King, David A. 2004. “The Scientific Impact of Nations: What Different Countries Get for their Research Spending”. Nature 430: 311-316. Kinneavy, James L. 1971. A Theory of Discourse: The Aims of Discourse. Englewood Cliffs: Prentice-Hall. Kirk, Elizabeth and Speier,Wolfgang. 1998. “U.S. and German Approaches to International Scientific Cooperation: perspectives for the 21st Century”. Report. Colloquium of the American Association for the Advancement of Science, the German-American Academic Council and the German Ministry for Education, Science, Research and Technology. Washington D.C. 15-16 October 1998. 20 August 2005 <http: / / www.aaas.org/ international/ eca/ GAAC/ wrkfinalprtintr.html> Kirkpatrick, Andy. 2004a. “Writing styles in Chinese and English compared. Implications for China English and ELT in China”. Paper. 4th International Symposium on ELT in China. Beijing. 21-15 May 2004. 3rd February 2006 <http: / / www.humanities.curtin.edu.au/ pdf/ sol/ Writing_styles.pdf> Kirkpatrick, Andy. 2004b. “Teaching Academic Writing: Whose Norms? ”. Paper. Australian Council of TESOL Associations Conference 2004. Perth. 1-3 October 2004. 4 February 2006 <http: / / www.humanities.curtin.edu.au/ pdf/ edu/ Teaching_AcademicWritingAC.pdf> Kotthoff, Helga 1989. Pro und Kontra in der Fremdsprache. Frankfurt am Main: Peter Lang. Kourilová, Magda.1996. “Interactive Function of Language in Peer Reviews of Medical Papers Written by NN Users of English”. UNESCO-ALSED LSP Newsletter 19 (1): 4-21. Kretzenbacher, Heinz L. 1990. Rekapitulation: Textstrategien zur Zusammenfassung von wissenschaftlichen Fachtexten. Tübingen: Gunter Narr. Kretzenbacher, Heinz L. 1991. “Syntax des wissenschaftlichen Fachtextes”. Fachsprache 13: 118-138. Kretzenbacher, Heinz L. 1995. “‘...sicherlich von Interesse, wenngleich...’. Das Peer Review als bewertende Textsorte der Wissenschaftssprache”. In Linguistik der Wissenschaftssprache, Heinz L. Kretzenbacher and Harald Weinrich (eds), 175-215. Berlin/ New York: de Gruyter. Kretzenbacher, Heinz L. 1998. “Fachsprache als Wissenschaftssprache”. In Fachsprachen. Ein internationales Handbuch zur Fachsprachenforschung und Terminologiewissenschaft, Lothar Hoffmann, Hartwig Kalverkämper, Herbert Ernst Wiegand, Christian Galinski and Werner Hüllen (eds), 133-142. Berlin: Walter de Gruyter. Kreutz, Heinz and Harres, Annette. 1997. “Some Observations on the Distribution and Function of Hedging in German and English Academic Writing”. In Culture and Styles of Academic Discourse, Anna Duszak (ed.), 181-201. Berlin, New York: Mouton de Gruyter. <?page no="331"?> 331 Kubota, Ryuko. 1998. “An Investigation of Japanese and English L1 Essay Organization: Differences and Similarities”. Canadian Modern Language Review 54 (4). 3 February 2006 <http: / / www.utpjournals.com/ product/ cmlr/ 544/ 544-Kubota.html> Kubota, Ryuko. 1999. “Japanese Culture Constructed by Discourses: Implications for Applied Linguistics Research and English Language Teaching”. TESOL Quarterly 33 (1): 9-35. Kubota, Ryuko. 2000. “Contexts of L2 Writing”. Paper. Symposium on Second Language Writing. Purdue University Indiana, USA. September 16, 2000. 28 January 2006 <http: / / logos.unh.edu/ sslw/ 2000/ keynote.html > Kubota, Ryuko and Lehner, Akita 2004. “Toward Critical Contrastive Rhetoric”. Journal of Second Language Writing 13 (1): 7-28. Kukla, André. 2000. Social Constructivism and the Philosophy of Science. New York: Routledge. Kussmaul, Paul. 1978. “Kommunikationskonventionen in Textsorten am Beispiel deutscher und englischer geisteswissenschaftlicher Abhandlungen”. Lebende Sprachen 23 (2): 54-57. Lafuente, Enrique. 2005. “Self-mention in Academic Writing: The Presence of the Writer in Research Articles from Different Academic Disciplines”. Paper. Conference “Cross-cultural and Cross-linguistic Perspectives on Academic Discourse”. University of Turku. 21 May 2005. Lakoff, George. 1972. “Hedges: A Study in Meaning Criteria and the Logic of Fuzzy Concepts”. In Papers from the Eighth Regional Meeting, Chicago Linguistic Society, Paul M. Peranteau, Judith N. Levi and Gloria C. Phares (eds), 183-228. Chicago: Chicago Linguistic Society. Lakoff, Robin. 1975. Language and Woman’s Place. New York: Harper and Row. Lakoff, Robin. 2001. The Language War. Berkeley: University of California Press. Lauret, Maria. undated. “Notes on Essay Writing: Format, References and Bibliography”. Department homepage. University of Sussex, Brighton, Department of American Studies. 22 August 2003 <http: / / www.sussex.ac.uk/ americanstudies/ documents/ notes_on_essay_writing.doc> Lee, Icy. 2000. “Exploring Reading-Writing Connections Through a Pedagogical Focus on ‘Coherence’”. Canadian Modern Language Review 57 (2). 19 December 2005 <http: / / www.utpjournals.com/ jour.ihtml? lp=product/ cmlr/ 572/ 572_TCL_Lee.html> Leech, Geoffrey. 1991. “The State of the Art in Corpus Linguistics”. In English Corpus Linguistics. Studies in Honour of Jan Svartvik, Karin Aijmer and Bengt Altenberg (eds), 8-29. London: Longman. Leech, Geoffrey and Svartvik, Jan. 1994. A communicative grammar of English. London: Longman. Leki, Ilona. 1991. “Twenty-Five Years of Contrastive Rhetoric: Text Analysis and Writing Pedagogies”. TESOL Quarterly 25 (1): 123-39. Leki, Ilona. 1997. “Cross-talk: ESL Issues and Contrastive Rhetoric”. In Writing in Multicultural Settings, Carol Severino, Juan C. Guerra and Johnnella E. Butler (eds.), 234-44. New York: Modern Language Association. Leki, Ilona. 1998. Academic Writing. 2nd ed. London: Cambridge University Press. <?page no="332"?> 332 Lenz, Friedrich. 1997. Diskursdeixis im Englischen: Sprachtheoretische Überlegungen und lexiko-grammatische Analysen. Tübingen: Niemeyer. Le Page, Robert and Tabouret-Keller, Andrée. 1985. Acts of Identity. Creole-based Approaches to Language and Ethnicity. Cambridge: Cambridge University Press. Levinson, Stephen C. 1983. Pragmatics. Cambridge: Cambridge University Press. Levinson, Stephen C. 2004. “Deixis and Pragmatics”. In The Handbook of Pragmatics, Laurence R. Horn and Gregory G. Ward (eds), 97-121. Oxford: Blackwell. Lew, Robert. 1997. “Towards a Taxonomy of Linguistic Jokes”. Stidia Anglica Posnaniensia 31: 132-152. Lewin, Beverly. 2003. “Hedging: A Question with Probably No Answer”. 28 July 2003 <http: / / www.tau.ac.il/ ~lewinb/ uteli.html> Lindemann, Stephanie and Mauranen, Anna. 2001. “It’s Just Real Messy: The Occurrence and Function of ‘Just’ in a Corpus of Academic Speech”. Supplement 1 English for Specific Purposes 20: 459-475. Lippert, Hans. 1978. “Rückzug der deutschen Sprache aus der Medizin”. Medizinische Klinik 73 (14): 487-496. Liu, Yameng. 1996. “To Capture the Essence of Chinese Rhetoric: An Anatomy of a Paradigm in Comparative Rhetoric”. Rhetoric Review 14: 318-335. Luey, Beth. 2002. Handbook for Academic Authors. Cambridge: Cambridge University Press. Luukka, Minna-Riitta and Markkanen, Raija. 1997. “Impersonalization as a Form of Hedging”. In Hedging and Discourse: Approaches to the Analysis of a Pragmatic Phenomenon in Academic Texts, Raija Markkanen and Hartmut Schröder (eds), 168-187. Berlin: de Gruyter. Luzio, Aldo Di, Gunther, Susanne and Orletti, Franca (eds). 2001. Culture in Communication. Amsterdam: John Benjamins. Lyons, John. 1977. Semantics. Cambridge: Cambridge University Press. Lyons, John. 1979. “Deixis and Anaphora”. In The Development of Conversation and Discourse Terry Myers (ed.), 88-103. Edinburgh: Edinburgh University Press. MacDonald, Susan Peck 1994. Professional Academic Writing in the Humanities and Social Sciences. Carbondale: Southern Illinois University Press. Mair, Christian. 2003. “‘Everybody espick a leetal English’: Die Weltsprache als Motor und Symptom der Globalisierung”. Paper. Tagung des SFB 541. Albert- Ludwigs-Universität, Freiburg. 10 October 2003. Mair, Christian and Hundt, Marianne. 1999. “‘Agile’ and ‘Uptight’ Genres: The Corpus-based Approach to Language Change in Progress”. International Journal of Corpus Linguistics 4 (2): 221-242. Malone, Michael J. 1997. Worlds of Talk: The Presentation of Self in Everyday Conversation. Oxford: Polity Press. Maltz, Daniel N., and Borker, Ruth A. 1982. “A Cultural Approach to Male - Female Miscommunication”. In Language and Social Identity, John Gumperz (ed.), 196-206. Cambridge: Cambridge University Press. Mandala, Susan. 1999. “Corpora and Speech Acts”. Paper. BAAL/ CUP Seminar “Investigating Discourse Practices Through Corpus Research: Methods, <?page no="333"?> 333 Findings and Applications”. University of Reading, Reading. 18-19 October 1999. 5 December 2005 <http: / / www.rdg.ac.uk/ app_ling/ baalcup_ abstracts.htm#mandala> Markkanen, Raija and Schröder, Hartmut (eds). 1997. Hedging and Discourse: Approaches to the Analysis of a Pragmatic Phenomenon in Academic Texts. Berlin: de Gruyter. Markkanen, Raija, Steffensen, Margaret S. and Crismore, Avon. 1993. “Quantitative Contrastive Study of Metadiscourse: Problems in Design and Analysis of Data”. Papers and Studies in Contrastive Linguistics 28: 137-151. Markl, Hubert. 1986. “Die Spitzenforschung spricht Englisch”. In Deutsch als Wissenschaftssprache, Hartwig Kalverkämper and Harald Weinreich (eds), 20- 25. Tübingen: Gunter Narr. Martin, James R. 1992. English Text: System and Structure. Amsterdam: John Benjamins. Martin, James R. 1997. “Analysing Genre: Functional Parameters”. In Genres and Institutions: Social Processes in the Workplace and School, Frances Christie and James R. Martin (eds.), 3-39. London: Cassell. Martin, James R. 2000. “Beyond Exchange: Appraisal Systems in English”. In Evaluation in Text: Authorial Stance and the Construction of Discourse, Susan Hunston and Guy Thompson (eds), 142-175. Oxford: Oxford University Press. Martin, James R. and White, Peter P. R. 2005. The Language of Evaluation: Appraisal in English. London: Palgrave Macmillan. Martín, Pedro Martín. 2003. “Personal Attribution in English and Spanish Scientific Texts”. Bells 12. 9 December 2005 <http: / / www.publicacions.ub.es/ revistes/ bells12/ PDF/ art09.pdf> Martín, Pedro Martín. 2005. The Rhetoric of the Abstract in English And Spanish Scientific Discourse: A Cross-cultural Genre-analytic Approach. Bern: Peter Lang. Matalene, Carolyn. 1985. “Contrastive Rhetoric: An American Writing Teacher in China”. College English 47: 789-807. Mauranen, Anna. 1993a. “Contrastive ESP Rhetoric: Metatext in Finnish-English Economic Texts”. English for Specific Purposes 12: 3-22. Mauranen, Anna. 1993b. Cultural Differences in Academic Rhetoric: A Textlinguistic Study. Frankfurt am Main: Peter Lang. Mauranen, Anna. 1996. “Discourse Competence - Evidence from Thematic Development in Native and Non-Native Texts”. In Academic Writing: Intercultural and Textual Issues, Eija Ventola and Anna Mauranen (eds), 195- 230. Amsterdam: John Benjamins. Mauranen, Anna. 2001. “Descriptions or Explanations? Some Methodological Issues in Contrastive Rhetoric”. In Academic Writing in Context, Martin Hewings (ed.), 43-54. Birmingham: University of Birmingham Press. Mauranen, Anna. 2002 “‘A good question’: Expressing Evaluation in Academic Speech”. In Domain-specific English: Textual Practices Across Communities and Classrooms, Giuseppina Cortese and Philip Riley (eds), 115-40. Frankfurt am Main: Peter Lang. <?page no="334"?> 334 Mauranen, Anna. 2003. “’But There’s a Flawed Argument’: Socialization Into and Through Metadiscourse”. In Corpus Analysis: Language Structure and Use, Pepi Leistyna and Charles F. Meyer (eds), 19-34. Amsterdam: Rodopi. Mauranen, Anna. 2004. “They’re a Little Bit Different… Observations on Hedges in Academic Talk”. In Discourse Patterns in Spoken and Written Corpora, Karin Aijmer and Anna-Brita Stenström (eds), 173-197. Amsterdam: John Benjamins. Mauranen, Anna and Bondi, Marina. 2003. “Evaluative Language Use in Academic Discourse”. Journal of English for Academic Purposes 2 (4): 269- 271. McCagg, Peter. 1996. “If You Can Lead a Horse to Water, You Don’t Have to Make it Drink: Some Comments on Reader and Writer Responsibilities”. Multilingua 15 (3), 239-256. McConnell, Grant. 2001. “The Expansion of English as a Language of Science and Communication: East and Southeast Asia”. In The Dominance of English as a Language of Science: Effects on Other Languages and Language Communities, Ulrich Ammon (ed.), 113-136. Berlin: Mouton de Gruyter. McEnery, Tony and Wilson, Andrew. 1996. Corpus Linguistics. 1st ed. Edinburgh: Edinburgh University Press. McEnery, Tony and Wilson, Andrew. 2001. Corpus Linguistics. 2nd ed. Edinburgh: Edinburgh University Press. Mead, George Herbert. 1913. “The Social Self”. Journal of Philosophy, Psychology and Scientific Methods 10: 374- 380. Mead, George Herbert. 1925. “The Genesis of the Self and Social Control”. International Journal of Ethics 35: 251-277. Mead, George Herbert. 1934. Mind, Self and Society. Chicago: University of Chicago Press. Meier, Alan. 1992. “How to Review a Technical Paper”. Energy and Buildings 19: 75-78. 26 August 2005 <http: / / eetd.lbl.gov/ EA/ Buildings/ ALAN/ PUBLICATIONS/ how.to.review.html> Meierkord, Christiane. 1996. Englisch als Medium der interkulturellen Kommunikation: Untersuchungen zum non-native-/ nonnative-speaker-Diskurs. Frankfurt am Main: Peter Lang. Melander, Björn, Swales, John M. and Fredrickson, Kirstin M. 1997. “Journal Abstracts from three Academic Fields in the United States and Sweden: National or Disciplinary Proclivities? ”. In Culture and Styles of Academic Discourse, Anna Duszak (ed.), 251-272. Berlin, New York: Mouton de Gruyter. Meliani, Augustina. 2002. The Study of Deixis in Ten Stories Taken from the Book of A 2nd Helping of Chicken Soup for the Soul. Diss. Petra Christian University, 2002. 3 January 2006 <http: / / digilib.petra.ac.id/ adscgi/ viewer.pl/ jiunkpe/ s1/ SING/ 2002/ jiunkpe-ns-s1-2002-11497034-619-deixischapter1.pdf> Merrill, M. David. 2002. “First Principles of Instruction”. Educational Technology Research and Development 50 (3): 43-59. Meyer, Bonnie J. F. 1975. The Organization of Prose and its Effects on Memory. New York: Elsevier. <?page no="335"?> 335 Meyer, Bonnie J. F. 1985. “Signaling the Structure of Text”. In The Technology of Text, David H. Jonassen (ed.), 64-89. Englewood Cliffs: Educational Technology. Meyer, Bonnie. J. F. and Poon, Leonard. W. 2001. “Effects of Structure Strategy Training and Signaling on Recall of Text”. Journal of Educational Psychology 93 (1): 141-59. Meyer, Lisa. 1995. “Academic Acculturation for Foreign Graduate Students: Meeting New Concepts of Research Writing”. College ESL 5 (2): 83-91. Miller, Donna R. 2002. “Multiple Judicial Opinions as Specialized Sites of Engagement: Conflicting Paradigms of Valuation and Legitimation in Bush v. Gore 2000”. In Conflict and Negotiation in Specialized Texts, Maurizio Gotti, Dorothee Heller and Marina Dossena (eds), 119-141. Bern: Peter Lang. Minogue, Angela and Weber, Siegfried. 1992. “Der Textvergleich als Untersuchungsmethode in der Fachsprachenforschung”. In Kontrastive Fachsprachforschung, Klaus-Dieter Baumann and Hartwig Kalverkämper (eds), 49-60. Tübingen: Gunter Narr. Modern Language Association of America. MLA Handbook for Writers of Research Papers. By Joseph Gibaldi. 6th ed. New York: Modern Language Association of America, 2003. Montgomery, Scott. 2004. “Of Towers, Walls, and Fields: Perspectives on Language in Science”. Science 303 (5662): 1333-1335. Montgomery, Scott L. 1996. The Scientific Voice. New York: Guildford Press. Moradian, Mahmood Reza. 1999. A Contrastive Rhetoric Study: The Use of Deductive, Inductive, and Quasi-inductive Writing Styles in Persian and English Newspaper Editorials. MA thesis. University of Teheran, 1999. 6 February 2006 < http: / / ffl.ut.ac.ir/ departments/ showthese.asp? t=1&ID=95> Morley, Louise. 1995. “An Agenda for Gender: Women in the University”. In European Journal of Women’s Studies 2, 271-275. Mühleisen, Susanne. 2003. “Towards Global Diglossia? English in the Sciences and the Humanities”. In The Politics of English as a World Language: New Horizons in Postcolonial Cultural Studies, Christian Mair (ed.), 107-118. Amsterdam, New York: Rodopi. Mühlhäusler, Peter and Harré, Ron. 1990. Pronouns and People: The Linguistic Construction of Social and Personal Identity. Oxford: Basil Blackwell. Mullany, Louise. 2002. “’I Don’t Think You Want Me to Get a Word in Edgeways Do You John ? ’: Re-assessing (Im)politeness, Language and Gender in Political Broadcast Interviews”. Working Papers on the Web 3 (Linguistic Politeness and Context). 7 December 2005 <http: / / www.shu.ac.uk/ wpw/ politeness/ mullany.htm> Myers, Greg. 1992. “Textbooks and the Sociology of Scientific Knowledge”. English for Specific Purposes 11 (1): 3-17. Myers, Greg. 1995. “Words and Pictures in a Biology Textbook”. In Approaches to Written Text: Classroom Applications, Thomas Miller (ed.), 93-104. Paris: USIA. Myers, Greg. 1996. “Strategic Vagueness in Academic Writing”. In Academic Writing: Intercultural and Textual Issues, Eija Ventola and Anna Mauranen (eds), 3-17. Amsterdam: John Benjamins. <?page no="336"?> 336 Myers, Greg. 1999. “Interaction in Writing: Principles and Problems”. In Writing: Text, Processes, and Practices, Christopher N. Candlin and Ken Hyland (eds), 40-61. Harlow: Addison Wesley Longman. Myers, Greg. 2001. “‘In my Opinion’: The Place of Personal Views in Undergraduate Essays”. In Academic Writing in Context, Martin Hewings (ed.), 63-78. Birmingham: University of Birmingham Press. Myers, Greg. 2004. “Contrastive Rhetoric and Academic Writing”. New Directions in Language Analysis. Course homepage. Department of Linguistics and Modern English Language, University of Lancaster. 31 January 2006 <http: / / bowland-files.lancs.ac.uk/ staff/ greg/ genre/ ContrastiveRhetoric.htm> Nationmaster.com. 2005. “English Status”. 26 August 2005 <http: / / www.nationmaster.com/ graph-T/ lan_eng_sta> Neff, Joanne, Dafouz, Emma, Diez, Mercedes, Prieto, Rosa and Chaudron, Craig. 2004. “Constrastive Discourse Analysis: Argumentative Text in English and Spanish”. In Discourse Across Languages and Cultures, Carol Lynn Moder and Aida Martinovic-Zic (eds), 267-83. Philadelphia: John Benjamins. Neuman W. Laurence. 2000. Social Research Methods: Qualitative and Quantitative Approaches. 4th ed. Toronto: Allyn and Bacon. Niiya, Minako. 1998. Differences between student and teacher expectations in an English language classroom in Japan. B.A. thesis. Sophia University, Japan, 1998. 10 June 2005 <http: / / pweb.sophia.ac.jp/ ~yosida-k/ expectations.pdf> Oakes, Michael P. 1998. Statistics for Corpus Linguistics. Edinburgh: Edinburgh University Press. O’Barr, William and Atkins, Bowman. 1980. “Women’s Language or ‘Powerless Language’? ”. In Women and Language in Literature and Society, Sally McConnell-Ginet, Ruth A. Borker and Nelly Furman (eds), 93-110. New York: Praeger. Oldenburg, Antje. 1995. “Methodologische Grundlagen der kontrastiven Fachtextlinguistik”. Fachsprache 3 (4): 107-115. Oldenburg, Hermann. 1992. Angewandte Fachtextlinguistik - ‘Conclusions’ und ‘Zusammenfassungen’. Tübingen: Gunter Narr. Olohan, Maeve. 2004. Introducing Corpora in Translation Studies. Bern: Peter Lang. Olszewska, Danuta. 2003. “Metatextuelle Formulierungsstrategien am Beispiel der Zielexplizierungen in wissenschaftlichen Texten”. 23 August 2003 <http: / / www.germanistik.uni-halle.de/ t_reader7.pdf> Ostler, Shirley E. 1987. “English in Parallels: A Comparison of English and Arabic Prose”. In Writing Across Languages: Analysis of L2 Text, Ulla Connor and Robert Kaplan (eds), 177-205. Reading: Addison-Wesley. Palmer, Frank R. 2001. Mood and Modality. 2nd ed. New York: Cambridge University Press. Partington, Alan. 2004. “Corpora and Discourse: A Most Congruous Beast”. In Corpora and Discourse, Alan Partington, John Morley and Louann Haarman (eds), 11-20. Bern: Peter Lang. Partington, Alan, Haarman, Louan and Morley, John. 2004. Corpora and Discourse. Bern: Peter Lang. <?page no="337"?> 337 Pearson, P. David and Dole, Janice A. 1987. “Explicit Comprehension Instruction: A Review of Research and a New Conceptualization of Instruction”. Elementary School Journal 88: 151-165. Pennycook, Alastair. 1994. The Cultural Politics of English as an International Language. Oxford: Oxford University Press. Pennycook, Alastair 1995. “English in the world/ The world in English”. In Power and Inequality in Language Education, James W. Tollefson (ed.), 34-58. Cambridge: Cambridge University Press. Peterlin, Agnes Pisanski. 2005. “Text-organising Metatext in Research Articles: An English-Slovene Contrastive Analysis”. English for Specific Purposes 24: 307-319. Phillipson, Robert. 1992. Linguistic Imperialism. Oxford: Oxford University Press. Phillipson, Robert. 2003. English-only Europe? Challenging Language Policy. London: Routledge. Phillipson, Robert and Skutnabb-Kangas, Tove. 1995. “Linguistic Rights and Wrongs”. Applied Linguistics 16 (4): 483-504. Piitulainen, Marja-Leena. 2001. “Zur Selbstbezeichnung in deutschen und finnischen Textsorten”. In Zur Kulturspezifik von Textsorten, Ulla Fix, Stephan Habscheid and Josef Klein, 159-173. Tübingen: Stauffenburg. Pocock, Emil. 2002. “Writing for Clarity and Grace: And Practical Suggestions for Revising History Papers”. Homepage. Eastern Connecticut State University, Willimantic, Connecticut, Department of History and American Studies. 22 August 2003 <http: / / www.easternct.edu/ personal/ faculty/ pocock/ writing.htm> Pöckl, Wolfgang. 1995. “Nationalstile in Fachtexten? Vom Tabuzum Modethema”. Fachsprache 17: 98-106. Poos, Deanna. 1999. “A Question of Gender? Hedging in Academic Spoken Discourse”. Paper. Michigan Linguistic Society. Michigan State University, East Lansing, Michigan. October 1999. Poos, Deanna and Simpson, Rita C. 2002. “Cross-disciplinary Comparisons of Hedging: Some Findings from the Michigan Corpus of Academic Spoken English”. In Using Corpora to Explore Linguistic Variation, Randi Reppen, Susan M. Fitzmaurice, and Douglas Biber, 3-23. Amsterdam: John Benjamins. Popken, Randall L. 1987. “A Study of Topic Sentence Use in Academic Writing”. Written Communication 4 (2): 209-228. Popper, Karl R. 1979. Objective Knowledge: An Evolutionary Approach. Oxford: Oxford University Press. Prawat, Richard S. and Floden, Robert E. 1994. “Philosophical Perspectives on Constructivist Views of Learning”. Educational Psychology 29 (1): 37-48. Prince, Anne, Blicblau, Aaron and Soesatyo, Bambang. 2000. “Implicit Actions and Explicit Outcomes: Cultural-academic Interactions in Writing Journal Research Articles”. In Papers from the Australian Association for Research in Education Conference 1999, Bill Russell (ed.). 2 February 2006 < http: / / www.aare.edu.au/ 99pap/ bli99489.htm> Prior, Paul. 1995. “Redefining the Task: An Ethnographic Examination of Writing and Response in six Graduate Seminars”. In Academic Writing in a Second Language: Essays on Research and Pedagogy, Diane D. Belcher and George Braine (eds), 47-82. Norwood: Ablex. <?page no="338"?> 338 “Pronomen”. Duden Grammatik der deutschen Gegenwartssprache. 6th ed. 1998. “Pronoun”. The Shorter Oxford English Dictionary. 5th ed. 2002. Pulcini, Virginia and Furiassi, Cristiano. 2004. “Spoken Interaction and Discourse Markers in a Corpus of Learner English”. In Corpora and Discourse, Alan Partington, John Morley and Louann Haarman (eds), 107-124. Bern: Peter Lang. Purves, Alan. 1984. “The Teacher as Reader: An Anatomy”. College English 46 (3): 259-265. Quirk, Randolph, Greenbaum, Sidney, Leech, Geoffrey and Svartvik, Jan. 1985. Comprehensive Grammar of the English Language. London: Longman. Rahman, Mujib. 2004. “Aiding the Reader: The Use of Metalinguistic Devices in Scientific Discourse”. Nottingham Linguistic Circular 18: 29-48. 4 January 2006 <http: / / www.nottingham.ac.uk/ english/ nlc/ rahman.pdf> Rahman, Sayeedur. 2005. “Orientations and Motivation in English Language Learning: a Study of Bangladeshi Students at Undergraduate Level”. Asian EFL Journal 7 (1): article 3. 10 June 2005 <http: / / www.asian-efl-journal.com/ march_05_sr.php> Ramanathan, Vai and Atkinson, Dwight. 1999. “Individualism, Academic Writing, and ESL Writers”. Journal of Second Language Writing 8 (1): 45-75. Rambo, Randy. 2004. “English Composition 1”. Homepage. Illinois Valley Community College. Oglesby, Illinois, Department of English. 22 August 2003 <http: / / www.ivcc.edu/ rambo/ eng1001/ eng1001_words.htm> Read, Julia and Ambrose, Marguerite. 1998. “Towards a Multilingual Dictionary of Academic Words”. Paper. Third International Conference of the African Association for Lexicography. University of Pretoria, Pretoria. 29-30 June 1998. 26 August 2005 <http: / / www.sun.ac.za/ wat/ translex/ A&R03.html> Reid, Joy M. 1989. “English as Second Language Composition in Higher Education: The Expectations of the Academic Audience”. In Richness in Writing: Empowering ESL Students, Donna M. Johnson and Duane H. Roen (eds), 220-234. New York: Longman. Reid, Joy M. 1992. “Helping Students Write for an Academic Audience”. In The Multicultural Classroom: Readings for Content-Area Teachers, Patricia A. Richard-Amato and Marguerite Ann Snow (eds), 210-21. White Plains: Longman. Rentel, Nadine. 2005. “Interlingual Varieties in Written Academic Discourse: Evaluation in Italian and French Research Articles in Linguistics”. Paper. “International Conference on Evaluation and Text Types”. University of Augsburg. 22 July 2005. Robinson, Pauline (ed.). 1988. Academic Writing: Process and Product. London: Modern English Publications/ British Council. Römer, Ute. 2005. “‘This seems somewhat counterintuitive, though…’: Negative Evaluation in Linguistic Book Reviews by Male and Female Authors”. In Strategies in Academic Discourse, Elena Tognini-Bonelli and Gabriella Del Lungo Camiciotti (eds), 97-115. Amsterdam: John Benjamins. Rossiter, Charles M. Jr. 1974. “Instruction in Metacommunication”. Central States Speech Journal 25: 36-42. <?page no="339"?> 339 Rowley-Jolivet, Elizabeth and Carter-Thomas, Shirley. 2005. “The Rhetoric of Conference Presentation Introductions: Context, Argument and Interaction”. International Journal of Applied Linguistics 15 (1): 45-70. Rummel, Kärt. 2005. How to Write Reader-friendly Texts: Common Problems in the English Academic Writing of Estonian Writers. Masters thesis. University of Tartu, 2005. 2 February 2006 <http: / / www.utlib.ee/ ekollekt/ diss/ mag/ 2005/ b17555073/ rummel.pdf> Russell, David. 1991. Writing in the Academic Disciplines, 1870-1990: A Curricular History. Carbondale: Southern Illinois University Press. Sachtleber, Susanne. 1990. “Linearität vs. Digressivität: Wissenschaftliche Texte im zweisprachigen Vergleich”. Folia Linguistica 24 (1-2): 105-122. Sachtleber, Susanne. 1993. Die Organisation wissenschaftlicher Texte: eine kontrastive Analyse. Frankurt am Main: Peter Lang. Sacks, Harvey. 1992. Lectures on Conversation. Volume 1 & 2. Oxford: Basil Blackwell. Salon, Albert. 2005. “Lettre aux sociétés savantes”. Letter from the President of the Society Avenir de la langue française to the Société française d'Astronomie et d'Astrophysique, 27 April 2005. 22 June 2005 <http: / / www.imperatiffrancais.org/ dossiers/ dossiers.php? id_dossier=3669> Samraj, Betty. 2002. “Disciplinary Variation in Abstracts: The Case of Wildlife Behaviours and Conservation Biology”. In Academic Discourse, John Flowerdew and Christopher Candlin (eds), 40-46. London: Longman. Samraj, Betty and Swales, John. 2000. “Writing in Conservation Biology: Searching for an Interdisciplinary Rhetoric? ”. Language and Learning Across the Disciplines 3 (3): 36-56. Sanderson, Tamsin. 2000. Gestaltungsprinzipien literaturwissenschaftlicher Texte im Sprachvergleich Deutsch - Englisch. Unpublished term paper. Albert- Ludwigs-Universität, Freiburg, 2000. Sanderson, Tamsin. 2005. “Kontrastive Fachtextpragmatik deutsch- und englischsprachiger wissenschaftlicher Texte: Danksagungen im interlingualen Vergleich”. In Text-Arbeiten: Textsorten im fremd- und muttersprachlichen Unterricht an Schule und Hochschule, Kirsten Adamzik and Wolf-Dieter Krause (eds), 59-83. Tübingen: Gunter Narr. Sanderson, Tamsin. 2006a. “Bridging the Gap Between English-language Competency and Academic Success in English for the Non-native Speaker”. In Linguistics, Language Learning and Language Teaching, David J. Allerton, Cornelia Tschichold and Judith Wieser (eds), 41-56. Basel: Schwabe. Sanderson, Tamsin. 2006b. Rev. of Corpora and Discourse by Alan Partington, John Morley und Louann Haarman (eds). Arbeiten aus Anglistik und Amerikanistik 31 (1). Sano, Hikomaro. 2002. “The World’s Lingua Franca of Science”. English Today 72 (18): 45-49. Sapir, Edward. 1925. “Memorandum on the Problem of an International Auxiliary Language”. Romanic Review 16: 244-256. 9 December 2005 <http: / / www.geocities.com/ Athens/ Forum/ 5037/ sapir.html> <?page no="340"?> 340 Schegloff, Emanuel. 1996. “Some Practices for Referring to Persons in Talk-in- Interaction: A Partial Sketch of a Systematics”. In Studies in Anaphora, Barbara A. Fox (ed.), 437-485. Amsterdam: John Benjamins. Schiewe, Jürgen. 1991. “Wissenschaftssprachen an der Universität Freiburg”. Freiburger Universitätsblätter 113: 17-51. Schiewe, Jürgen. 1996. Sprachenwechsel. Funktionswandel. Austausch der Denkstile: die Universität Freiburg zwischen Latein und Deutsch. Tübingen: Niemeyer. Schiffrin, Deborah. 1980. “Meta-Talk: Organisational and Evaluative Brackets in Discourse”. Sociological Inquiry 50: 199-236. Schiffrin, Deborah. 1987. Discourse Markers. Cambridge: Cambridge University Press. Schleppegrell, Mary J. and Colombi, M. Cecilia (eds). 2002. Developing Advanced Literacy in First and Second Languages. Mahwah: Lawrence Erlbaum Associates. Schoder, Detlef. 2005. “Hinweise zum Anfertigen von Arbeiten”. Homepage. University of Cologne, Department of Information Management. 27 October 2005 <http: / / www.wim.uni-koeln.de/ uploads/ media/ Hinweise_zum_ Anfertigen_von_Arbeiten_Mai05.pdf> Schreiber, Michael. 2004. “Kontrastive Linguistik und sprachenpaarbezogene Translationswissenschaft”. In Claims, Changes and Challenges in Translation Studies, Gyde Hansen, Kirsten Malmkjær and Daniel Gile (eds), 83-98. Amsterdam: John Benjamins. Schröder, Hartmut. 1987. “Kontrastive Textanalysen: Ein Projekt zur Erforschung des Zusammenhangs von Diskurs, Kultur, Paradigma und Sprache in argumentativen Fachtexten der Gesellschaftswissenschaften”. Finlance 6: 145- 173. Schröder, Hartmut. 1998. “‘Ich sage das einmal ganz ungeschützt’: Hedging und wissenschaftlicher Diskurs”. In Darstellungsformen der Wissenschaften im Kontrast, Lutz Danneber and Jürg Niederhauser (eds), 183-197. Tübingen: Gunter Narr. Schütz, Alfred. 1932. Der sinnhafte Aufbau der sozialen Welt. Eine Einleitung in die verstehende Soziologie. Wien: Springer. Scollon, Ronald. 1994. “As a Matter of Fact: The Changing Ideology of Authorship and Responsibility in Discourse”. World Englishes 13: 33-46. Scollon, Ronald. 1997. “Contrastive Rhetoric, Contrastive Poetics, or Perhaps Something Else? ”. TESOL Quarterly 31 (2): 352-363. Scollon, Ron and Scollon, Susanne Wong. 1997. “Point of View and Citation: Fourteen Chinese and English Versions of the ‘Same’ News Story”. Text 17 (1): 83-125. Scollon, Ron and Scollon, Suzanne Wong. 2001. Intercultural Communication. 2nd ed. Malden: Blackwell. Scott, Mike and Tribble, Christopher. 2006. Textual Patterns: Key Words and Corpus Analysis in Language Education. Amsterdam: John Benjamins. Searle, John R. 1995. The Construction of Social Reality. New York: Free Press. Searle, John, and Vanderveken, Daniel. 1985. Foundations of Illocutionary Logic. Cambridge: Cambridge University Press. <?page no="341"?> 341 Seidlhofer, Barbara. 1995. Approaches to Summarization: Discourse Analysis and Language Education. Tübingen: Gunter Narr. Sengupta, Sima, Forey, Gail and Hamp-Lyons, Liz. 1999. “Supporting Effective English Communication Within the Context of Teaching and Research in a Tertiary Institute: Developing a Genre Model for Consciousness-raising”. Supplementary Issue. English for Specific Purposes 18: 7-23. Seoane-Posse, Elena. 2004. “Changing Styles: On the Recent Evolution of Scientific British and American English”. Paper. “13th International Conference on English Historical Linguistics”. University of Vienna. 24 August 2004. Sigley, Robert. 2003. “The Importance of Interaction Effects”. Language Variation and Change 15: 227-253. Simpson, JoEllen. 2000. “Topical Structure Analysis of Academic Paragraphs in English and Spanish”. Journal of Second Language Writing 9 (3): 293-309. Simpson, Rita. 2004. “Stylistic Features of Spoken Academic Discourse: The Role of Formulaic Expressions”. In Discourse in the Professions: Perspectives from Corpus Linguistics, Ulla Connor and Thomas A. Upton (eds), 37-64.. Amsterdam: John Benjamins. Simpson, Rita. 2004. “Stylistic Features of Academic Speech: The Role of Formulaic Expressions”. In Discourse in the Professions, Ulla Connor and Thomas A. Upton (eds.), 37-64. Amsterdam: John Benjamins. Skelton, John. 1988. “The Care and Maintenance of Hedges”. ELT Journal 42 (1), 37-43. Skudlik, Sabine. 1990. Sprachen in den Wissenschaften - Deutsch und Englisch in der internationalen Kommunikation. Tübingen: Gunter Narr. Skutnabb-Kangas, Tove. 2004. “The Right to Mother Tongue Medium Education: The Hot Potato in Human Rights Instruments”. Plenary Paper. “2nd Mercator International Symposium”. Tarragona. 27 February 2004. 1 December 2005 <http: / / www.ciemen.org/ mercator/ pdf/ simp-skuttnab.pdf> Spencer-Oatley, Helen (ed.). 2000. Culturally Speaking: Managing Relations in Talk Across Cultures. London: Cassell. Spillner, Bernd. 1982. “Formen und Funktionen wissenschaftlichen Sprechens und Schreibens“. In Wissenschaft - Sprache - Gesellschaft. Über Kommunikationsprobleme zwischen Wissenschaft und Öffentlichkeit und Wege zu deren Überwindung, Karl Ermert (ed.), 33-57. Rehburg-Loccum: Evangelische Akademie. Stenström, Anna-Brita. 2004. “What is Going on Between Speakers? ”. In Corpora and Discourse, Alan Partington, John Morley and Louann Haarman (eds), 259- 284. Bern: Peter Lang. Stewart, Miranda. 2001. “Pronouns of Power and Solidarity: The Case of Spanish First Person Plural nosotros”. Multilingua 20 (2), 155-169. Stickel, Gerhard. 2000. “Deutsch als Wissenschaftssprache an außeruniversitären Forschungsinstituten”. In Deutsch als Wissenschaftssprache im 20. Jahrhundert. Vorträge des Internationalen Symposions vom 18./ 19. Januar 2000, Friedhlem Debus, Franz Krollmann and Uwe Pörksen (eds), 125-142. Mainz: Akadamie der Wissenschaften und der Literatur. <?page no="342"?> 342 Stolze, Radegundis and Deppert, Alex. 1998. “Übersetzung und Verständlichkeit deutscher und englischer Wissenschaftstexte”. Fachsprache 20 (2): 116-130. Stroebe, Wolfgang. 1976. “Is Social Psychology Really All That Complicated? ”. European Journal of Social Psychology 6 (4): 509-511. Strunk, William Jr. and White, Elwyn Brooks. 2000. The Elements of Style. 4th ed. New York: Longman. Suau, Francisca. 2004. “The Difficult Balance Between Author’s Power and Scientific Community’s Power in Research Articles: A Comparison of English and Spanish Discourse in Linguistics as a Discipline”. Paper. “1st International Conference on Critical Discourse Analysis”. Valencia. 6 May 2004. Swales, John. 1990. Genre Analysis. Cambridge: Cambridge University Press. Swales, John 2001. “Metatalk in American Academic Talk: The Cases of ‘Point’ and ‘Thing’”. Journal of English Linguistics 29 (1): 34-54. Swales, John. 2002a. “On Models in Applied Discourse Analysis”. In Research and Practice in Professional Discourse, Christopher Candlin (ed.), 61-77. Hong Kong: City University of Hong Kong Press. Swales, John. 2002b. “Integrated and Fragmented Worlds: EAP Materials and Corpus Linguistics”. In Academic Discourse, John Flowerdew (ed.), 150-164. Harlow: Longman. Swales, John. 2004a. “Evaluation in Academic Speech: First Forays”. In Academic Discourse: New Insights into Evaluation, Gabriella Del Lungo Camiciotti and Elena Tognini-Bonelli (eds), 31-53. Bern: Peter Lang. Swales, John. 2004b. Research Genres: Explorations and Applications. Cambridge: Cambridge University Press. Swales, John, Ahmad, Ummul K., Chang, Yu-Ying, Chavez, Daniel, Dressen, Dacia F. and Seymour, Ruth. 1998. “Consider This: The Role of Imperatives in Scholarly Writing”. Applied Linguistics 19 (1): 97-121. Swales, John and Burke, Amy. 2003. “‘It’s Really Fascinating Work’: Differences in Evaluative Adjectives Across Academic Registers”. In Corpus Analysis: Language Structure and Use, Pepi Leistyna and Charles F. Meyer (eds),1-18. Amsterdam: Rodopi. Swales, John and Feak, Christine. 1994. Academic Writing for Graduate Students. Ann Arbor: University of Michigan Press. Swales, John and Feak, Christine. 2000. English in Today’s Research World: A Writing Guide. Ann Arbor: University of Michigan Press. Swales, John and Malczewski, Bonnie. 2001 “Discourse Management and New Episode Flags in MICASE”. In Corpus Linguistics in North America: Selections from the 1999 Symposium, Rita Simpson and John Swales (eds), 145-164. Ann Arbor: University of Michigan Press. Szmrecsanyi, Benedikt. 2005. “Creatures of Habit: A Corpus-linguistic Analysis of Persistence in Spoken English”. Corpus Linguistics and Linguistic Theory 1 (1): 113-149. Taylor, Gordon and Chen, Tingguang. 1991. “Linguistic, Cultural, and Subcultural Issues in Contrastive Discourse Analysis: Anglo-American and Chinese Scientific Texts”. Applied Linguistics 12: 319-336. <?page no="343"?> 343 Taylor, James R. and Cooren, François. 1997. “What Makes Communication ‘Organizational’? How the Many Voices of a Collectivity Become the One Voice of an Organization”. Journal of Pragmatics 27: 409-438. Templer, Bill. 2002. “The Political Economy of TEFL: A Levy on English as a Native Language Privilege”. TESOL Matters 12 (3). 2 February 2006 <http: / / www.tesol.org/ pubs/ articles/ 2002/ tm12-3-09.html> Teufel, Simone. 1999. Argumentative Zoning: Information Extraction from Scientific Articles. Diss. University of Edinburgh, 1999. 23 December 2005 <http: / / www.cl.cam.ac.uk/ users/ sht25/ az.html> Thielmann, Winfried. 1999a. “‘Justification’: The Importance of Linguistic Action Patterns for the Success of Intercultural Communication”. In Striving for the Third Place. Intercultural Competence through Language Education, Joseph Lo Bianco, Chantal Crozet and Tony Liddicoat (eds), 167-179. Melbourne: Language Australia. Thielmann, Winfried. 1999b. “Begründungen versus advance organizers: Zur Problematik des Englischen als lingua franca der Wissenschaft”. Deutsche Sprache 4 (99): 370-378. Thompson, Geoff. 2001. “Interaction in Academic Writing: Learning to Argue with the Reader”. Applied Linguistics 22 (1): 58-78. Thompson, Geoff and Hunston, Susan. 2000. “Evaluation: An Introduction”. In Evaluation in Text: Authorial Stance and the Construction of Discourse, Susan Hunston and Geoff Thompson (eds), 1-27. Oxford: Oxford University Press. Thompson, Geoff and Thetela, Puleng. 1995. “The Sound of One Hand Clapping: The Management of Interaction in Written Discourse”. Text 15: 103-127. Thompson, Paul (ed.). 1998. Issues in EAP Writing Research and Instruction. Reading: Centre for Applied Language Studies, University of Reading. Thompson, Paul. 2000a. “Citation Practices in PhD Theses”. In Rethinking Language Pedagogy from a Corpus Perspective, Lou Burnard and Tony McEnery (eds), 99-101. Frankfurt am Main: Peter Lang. Thompson, Paul (ed.). 2000b. Patterns and Perspectives: Insights into EAP Writing Practice. Reading: Centre for Applied Language Studies, University of Reading. Thompson, Paul and Tribble, Chris. 2001. “Looking at Citations: Using Corpora in English for Academic Purposes”. Language Learning and Technology 5 (3), 91-105. 3 February 2006 <http: / / llt.msu.edu/ vol5num3/ thompson/ default.html> Thurstun, Jennifer and Candlin, Christopher. 1998. Exploring Academic English: A Workbook for Student Essay Writing. Sydney: National Centre for English Language Teaching & Research. Tognini-Bonelli, Elena and Camiciotti, Gabriella Del Lungo. 2005. Strategies in Academic Discourse. Amsterdam: John Benjamins. Trabant, Jürgen. 2000. “Umzug ins Englische: über die Globalisierung des Englischen in den Wissenschaften”. Philologie im Netz 13: 3 February 2006 <108-126. http: / / web.fu-berlin.de/ phin/ phin13/ p13t5.htm> Truchot, Claude. 1997. “The Spread of English: From France to a More General Perspective”. World Englishes 16 (1): 65-76. Trumpp, Eva. 1998. Fachtextsorten kontrastiv: Englisch-Deutsch-Französisch. Tübingen: Gunter Narr. <?page no="344"?> 344 Tsunoda, Minoru. 1983. Les langues internationales dans les publications scientifiques et techniques. Sophia linguistica: 144-155. Tyler, Andrea. 1994. “The Role of Syntactic Structure in Discourse Structure: Signaling Logical and Prominence Relations”. Applied Linguistics 15: 243- 262. University of Essex English Language Teaching Centre. 2003. “Study Skills: Writing Skills Part 1: Academic Style”. 23 August 2003 <http: / / www2.essex.ac.uk/ by/ studyskills/ writing-skills-1.shtm> University of North Carolina at Chapel Hill Writing Center. 2002. “‘I’: When to Use It”. Homepage. 23 August 2003 <http: / / www.unc.edu/ depts/ wcweb/ handouts/ should_I_use_I.html> Upton, Thomas A. 2004. Reading Skills for Success: A Guide to Academic Texts. Ann Arbor: University of Michigan Press. Valero-Garces, Carmen. 1996. “Contrastive ESP Rhetoric: Metatext in Spanish- English Economics Texts”. English for Specific Purposes 15 (4): 279-94. Vande Kopple, William J. 1980. Experimental Evidence for Functional Sentence Perspective. Diss. University of Chicago, 1980. Vande Kopple, William J. 1985. “Some Exploratory Discourse on Metadiscourse”. College Composition and Communication 36: 82-93. Vande Kopple, William J. 1988. “Metadiscourse and the Recall of Modality Marker”. Visible Language 22 (2): 233-72. Vandenbroucke, Jan P. 1989. “On Not Being Born a Native Speaker of English”. British Medical Journal 289: 1461-2. Van Dijk, Teun A. 1977. Text and Context: Explorations in the Semantics and Pragmatics of Discourse. London: Longman. Van Dijk, Teun A. 1980. Macrostructures: An Interdisciplinary Study of Global Structures in Discourse, Interaction and Cognition. Hillsdale: Lawrence Erlbaum Associates. Vanhala-Aniszewski, Marjatta. 2004. “The Role of Metadiscourse in Russian and Finnish Academic Texts”. Paper. Symposium “Wissenschafliches Schreiben”. University of Klagenfurt. 25 October 2004. Van Riemsdijk, Henk and Williams, Edwin. 1986. An Introduction to the Theory of Grammar. Cambridge: M.I.T. Press. Varttala, Teppo. 2001. Hedging in Scientifically Oriented Discourse: Exploring Variation According to Discipline and Intended Audience. Diss. University of Tampere, 2001. 2 January 2006 <http: / / acta.uta.fi/ pdf/ 951-44-5195-3.pdf> Ventola, Eija. 1996. “Packing and Unpacking of Information in Academic Texts”. In Academic Writing: Intercultural and Textual Issues, Eija Ventola and Anna Mauranen (eds), 153-194. Amsterdam, John Benjamins. Ventola, Eija. 2003. “Akademisches Englisch der Nicht-Muttersprachler - Chance oder Hindernis? ”. In Mehrsprachige Wissenschaft europäische Perspektiven, Konrad Ehlich (ed.). München: Institut für Deutsch als Fremdsprache / Transnationale Germanistik. 1 January 2006 <http: / / www.eurosprachenjahr.de/ onlinepub.htm> Ventola, Eija, Shalom, Celia and Thompson, Susan. 2002. The Language of Conferencing. Frankfurt am Main: Peter Lang. <?page no="345"?> 345 Vogel, Beate and Buck, Margrit. 2002. Vorlage Nr. L 130/ G 172 für die Sitzung der Deputation für Bildung am 18. April 2002 Fremdsprachlicher Unterricht in der Grundschule. 16 June 2005 <http: / / www.schule.bremen.de/ sprachen/ fuig.pdf> Voss, James F. 1987. “Learning and Transfer in Subject-matter Learning: A Problem Solving Model”. International Journal of Educational Research 11: 607-622. Wales, Katie. 1996. Personal Pronouns in Present-day English. Cambridge: Cambridge University Press. Watson, D. Rod. 1987. “Interdisciplinary Considerations in the Analysis of Proterms”. In Talk and Social Organisation, Graham Lee Button and John R.E. Lee (eds), 261-289. Clevedon: Multilingual Matters. Webber, Bonnie Lynn. 1988. “ Discourse Deixis: Reference to Discourse Segments”. In Proceedings of the 26th annual meeting on Association for Computational Linguistics, 113-122. New York: Association for Computational Linguistics. 3 January 2006 <http: / / acl.ldc.upenn.edu/ P/ P88/ P88-1014.pdf> Weingarten, Rüdiger. 1994. “Zur Stilistik der Wissenschaftssprache: Objektivitäts- und Handlungsstil”. In Texte und Diskurse: Methoden und Forschungsergebnisse der Funktionalen Pragmatik, Gisela Brünner and Gabriele Graefen (eds), 115-136. Opladen: Westdeutscher Verlag. Weinrich, Harald. 1989. “Formen der Wissenschaftssprache”. Jahrbuch 1988 der Akademie der Wissenschaften zu Berlin, 119-158. West, Charles K., Farmer, James A. and Wolff, Phillip M. 1991. Instructional Design: Implications from Cognitive Science. Englewood Cliffs: Prentice-Hall. White, Peter R. R. 2001. Appraisal Website. <www.grammatics.com/ appraisal>. White, Peter R. R. 2003. “Beyond Modality and Hedging: A Dialogic View of the Language of Intersubjective Stance”. Text 23 (2): 2594-8. Widdowson, Henry G. 1979. Explorations in Applied Linguistics. Oxford: Oxford University Press. Widdowson, Henry G. 1984. Explorations in Applied Linguistics 2. Oxford: Oxford University Press. Widdowson, Henry G. 1998. “Context, Community and Authentic Language”. TESOL Quarterly 32 (4): 705-716. Willemyns, Roland. 2001. “English in Linguistic Research in Belgium”. In The Dominance of English as a Language of Science: Effects on Other Languages and Language Communities, Ulrich Ammon (ed.), 329-42. Berlin: Mouton de Gruyter. Williams, Geoffrey. 1998. “Collocational Networks: Interlocking Patterns of Lexis in a Corpus of Plant Biology Research Articles”. International Journal of Corpus Linguistics 3 (1): 151-171. Williams, Joseph M. 1981. “Literary Style: The Personal Voice”. In Style and Variables in English, Timothy Shopen and Joseph M. Williams (eds), 117-216. Cambridge, M.A.: Winthrop. Williams, Joseph M. 1989. Style: Ten Lessons in Clarity and Grace. 3rd ed. Glenview: Scott, Foresman & Co. <?page no="346"?> 346 Wilson, David. 2002. The Englishisation of Academe: A Finnish Perspective. Jyväskylä: University of Jyväskylä Language Centre. Winsa, Birger. 2000. “Language Planning in Sweden”. In Language Planning in Nepal, Taiwan and Sweden, Richard B. Baldauf and Robert Kaplan (eds), 376- 473. Clevedon: Multilingual Matters. 3 February 2006 <http: / / www.multilingual-matters.net/ jmmd/ 020/ 0376/ jmmd0200376.pdf> Wissenschaftliche Kommission Niedersachsen. 2005. Forschungsevaluation an niedersächsischen Hochschulen und Forschungseinrichtungen: Slavistik und Finnougristik Ergebnisse und Empfehlungen. Hannover: Wissenschaftliche Kommission Niedersachsen, 2005. 3 February 2006 <http: / / www.wk. niedersachsen.de/ Materialien/ FE_Slavistik_Finnougristik.pdf> Wood, Alastair. 1997. “International Scientific English: Some Thoughts on Science, Language and Ownership”. Science Tribune. 15 October 2004 <http: / / www.tribunes.com/ tribune/ art97/ wooda.htm> Wooley, Debra. 2003. “Some Rules of Formal Writing, Presentation of Assignments And Class Conduct”. Homepage. Florida International University Department of English. 23 August 2003 <http: / / comptalk.fiu.edu/ wooley.htm> Wright, Sue Ellen and Wright, Leland D (eds). 1993. Scientific and Technical Translation. Amsterdam: John Benjamins. Wüest, Jakob. 1988. “Textsorten kontrastiv betrachtet: Die Präsenz des Autors in linguistischen Publikationen”. In Linguistische Studien: Studien zur Sprachkonfrontation, Gerd Wotjak (ed.), 125-136. East Berlin: Akademie der Wissenschaften der DDR. Young, Robert E. 1996. Intercultural Communication: Pragmatics, Genealogy, Deconstruction. Clevedon: Multilingual Matters. Zamel, Vivian. 1997. “Toward a Model of Transculturation”. TESOL Quarterly 31: 341-352. Zamel, Vivian and Spack, Ruth (eds). 1998. Negotiating Academic Literacies: Teaching and Learning Across Languages and Cultures. Mahwah: Lawrence Erlbaum Associates. Zainuddin, Hanizah and Moore, Rashid A. 2003. “Audience Awareness in L1 and L2 Composing of Bilingual Writers”. Teaching English as a Second or Foreign Language 7 (1). 12 January 2006 <http: / / www-writing.berkeley.edu/ TESL- EJ/ ej25/ a2.html> Zhabotynska, Svitlana. 2001. “Author-profile in Scholarly Papers: Anglo- American vs. Ukranian/ Russian”. In Approaches to the Pragmatics of Scientific Discourse, Anrás Kertesz (ed.), 73-89. Frankfurt am Main: Peter Lang. Zhu, Yunxia. 1997. “An Analysis of Structural Moves in Chinese Sales Letters”. Text 17 (4): 543-566. Zifonun, Gisela, Hoffmann, Ludger and Strecker, Bruno. 1997. Grammatik der deutschen Sprache. Berlin: Walter de Gruyter. Ziman, John. 1998. “Why ´Must Scientists Become More Ethically Sensitive Than They Used to be? ”. Science 282 (5395): 1813-4. Zöfel, Peter. 2000. Statistik verstehen. München: Addison Wesley. Zuck, Joyce Gilmour and Zuck, Louis V. 1985. “Hedging in Newswriting”. In Beads or Bracelet? How do we Approach LSP? : Selected Papers from the 5th <?page no="347"?> 347 European Symposium on LSP, Anne-Marie Cornu, Johan Vanparjis, Mieke Delahaye and Lut Baten (eds), 172-180. Oxford: Oxford University Press. Zurghoul, Muhammad Raji. 2003. The Language of Power and the Power of Language in Higher Education in the Arab World: Conflict, Dominance and Shift. 2 January 2006 <http: / / members.fortunecity.com/ mrzughoul/ Language %20of%20Powr.htm> <?page no="349"?> 349 Appendix: corpus texts Philosophy (PH) Mind (2003), 112(447): 399-432. Mind (2002), 111(442): 305-322. Mind (2002), 111(442): 289-296. Mind (2002), 111(441): 15-36. Mind (2000), 109(435): 481-518. Journal of Philosophy (1997), 94(12): 622-637. Journal of Philosophy (1997), 94(12): 599-621. Journal of Philosophy (1997), 94(11): 564-583 Journal of Philosophy (1997), 94(11): 543-563. Journal of Philosophy (1997), 94(10): 535-542 Zeitschrift für philosophische Forschung (1999), 53(4): 507-529. Zeitschrift für philosophische Forschung (1999), 53(4): 487-506. Zeitschrift für philosophische Forschung (1999), 53(4): 383-406. Zeitschrift für philosophische Forschung (1999), 53(4): 355-382. Zeitschrift für philosophische Forschung (1999), 53(2): 236-258. Zeitschrift für philosophische Forschung (1999), 53(2): 188-213. Zeitschrift für philosophische Forschung (1999), 53(2): 53-74. Zeitschrift für philosophische Forschung (1998), 52(4): 497-520. Zeitschrift für philosophische Forschung (1998), 52(2), 170-194. Zeitschrift für philosophische Forschung (1996), 50(1): 309-322. History (HI) History (2003), 88(289), 32-52. History (2003), 88(289), 17-31. History (2002), 87(287), 353-372. History (2002), 87(287), 336-352. History (2002), 87 (287), 322-335. The American Historical Review (2003), 108(1): 84-119. The American Historical Review (2003), 108(1): 50-83. The American Historical Review (2003), 108(1): 20-49. The American Historical Review (2002), 107(2): 388-418. The American Historical Review (2002), 107(1): 55-83. Historische Zeitschrift (2003), 276(3): 613-646. <?page no="350"?> 350 Historische Zeitschrift (2003), 276(2): 323-368 Historische Zeitschrift (2003), 276(1): 37-73. Historische Zeitschrift (2003), 276(1): 1-36. Historische Zeitschrift (2002) 275(1): 1-55. Historische Zeitschrift (2002): 274(3): 633-664. Historische Zeitschrift (2002), 274(1): 61-85. Historische Zeitschrift (2001), 273(1): 1-29. Historische Zeitschrift (2000), 271(2): 281-309. Historische Zeitschrift (2000), 270(2): 345-388. Folklore (FL) Folklore (2002), 113(2): 215-236. Folklore (2002), 113(2): 151-173. Folklore (2002), 113(1): 11-34. Folklore (2001), 112(2): 147-161. Folklore (2000), 111(2): 282-305. Journal of American Folklore (2003), 116(460): 159-175. Journal of American Folklore (2003), 116(460): 129-158. Journal of American Folklore (2003), 116(459): 19-35. Journal of American Folklore (2003), 116(459): 9-18. Journal of American Folklore (2002), 115(455): 28-61. Zeitschrift für Volkskunde (2004), 100(1): 1-30. Zeitschrift für Volkskunde (2003), 99(2): 237-247. Zeitschrift für Volkskunde (2003), 99(2): 177-188. Zeitschrift für Volkskunde (2003), 99(1): 65-78. Zeitschrift für Volkskunde (2002), 98(2): 257-270. Zeitschrift für Volkskunde (2002), 98(1): 34-58. Zeitschrift für Volkskunde (2002), 98(1): 1-15. Zeitschrift für Volkskunde (2001), 97(1): 49-67. Zeitschrift für Volkskunde (2001), 97(1): 15-28. Zeitschrift für Volkskunde (2001), 97(1): 1-14. Literary Studies (LT) Essays in Criticism(2002), 52(4): 314-332. Essays in Criticism (2002), 52(3): 235-244. Essays in Criticism (2002), 52(1): 56-75. Essays in Criticism (2001), 51(1): 51-67. Essays in Criticism (2000), 50(4): 329-353. Journal of American Literature (2003), 75(1): 119-140. Journal of American Literature (2003), 75(1): 91-118. Journal of American Literature (2003), 75(1): 61-90. <?page no="351"?> 351 Journal of American Literature (2003), 75(1): 31-60. Journal of American Literature (2003), 75(1): 1-30. Euphorion (2003), 97(3): 405-417. Euphorion (2003), 97(3): 327-348. Euphorion (2003), 97(2): 163-191. Euphorion (2003), 97(1): 51-71. Euphorion (2003), 97(1): 19-50. Euphorion (2002), 96(4): 468-493. Euphorion (2002), 96(4): 449-467. Euphorion (2002), 96(3): 321-348. Euphorion (2002), 96(3): 303-320. Euphorion (2002), 96(1): 1-26. Linguistics (LG) Journal of Linguistics (2001), 37(2): 313-327. Journal of Linguistics (2000), 36(3): 463-486. Journal of Linguistics (2000), 36(1): 13-38. Journal of Linguistics (2000), 36(1): 1-12. Journal of Linguistics (1999), 35(2): 489-525. Journal of English Linguistics (2003), 31(1): 3-33. Journal of English Linguistics (2002), 30(4): 339-352. Journal of English Linguistics (2002), 30(2): 122-137. Journal of English Linguistics (2002), 30(1): 6-27. Journal of English Linguistics (2001), 29(4): 329-345. Zeitschrift für germanistische Linguistik (2003), 31(1): 78-98. Zeitschrift für germanistische Linguistik (2003), 31(1): 1-43. Zeitschrift für germanistische Linguistik (2002), 30(3): 310-341. Zeitschrift für germanistische Linguistik (2002), 30(2): 147-168. Zeitschrift für germanistische Linguistik (2002), 30(1): 56-72. Zeitschrift für germanistische Linguistik (2001), 29(1): 40-69. Zeitschrift für germanistische Linguistik (2001), 29(1): 23-39. Zeitschrift für germanistische Linguistik (2001), 29(1): 1-22. Zeitschrift für germanistische Linguistik (2001), 29(2): 192-218. Zeitschrift für germanistische Linguistik (2001), 29(2): 167-191.