Text linguistics for the contrastive study of online customer comments
Text-linguistic patterns in German, Dutch, Spanish and French hotel comments and reviews
0116
2017
978-3-8233-9015-2
978-3-8233-8015-3
Gunter Narr Verlag
Raul Sánchez Prieto
The goal of this corpus based research monograph is twofold. On the one hand, the volume delivers a practice relevant theoretical framework which will enable linguists and internationally operating businesses to evaluate product and service comments of foreign customers written in their mother tongue and intended for their fellow nationals. The theoretical framework should be applicable to any language combination. On the other hand, a large-scale empiric study has been carried out building on our theoretical premises with the aim of determining the text linguistic structure and language use in Spanish, Dutch, French and German customer comments on hotel accommodation in two relevant applications of the Social Web (social network services like Facebook and comments on the webs of hotel reservations agencies). Our findings will enable Spanish and German companies to interpret the linguistic behavior of their German/Spanish clients sharing their accommodation experiences on the Social Web. The goal of this corpus based research monograph is twofold. On the one hand, the volume delivers a practice relevant theoretical framework which will enable linguists and internationally operating businesses to evaluate product and service comments of foreign customers written in their mother tongue and intended for their fellow nationals. The theoretical framework should be applicable to any language combination. On the other hand, a large-scale empiric study has been carried out building on our theoretical premises with the aim of determining the text linguistic structure and language use in Spanish, Dutch, French and German customer comments on hotel accommodation in two relevant applications of the Social Web (social network services like Facebook and comments on the webs of hotel reservations agencies). Our findings will enable Spanish and German companies to interpret the linguistic behavior of their German/Spanish clients sharing their accommodation experiences on the Social Web.
<?page no="0"?> The goal of this corpus based research monograph is twofold. On the one hand, the volume delivers a practice relevant theoretical framework which will enable linguists and internationally operating businesses to evaluate product and service comments of foreign customers written in their mother tongue and intended for their fellow nationals. The theoretical framework should be applicable to any language combination. On the other hand, a large-scale empiric study has been carried out building on our theoretical premises with the aim of determining the text linguistic structure and language use in Spanish, Dutch, French and German customer comments on hotel accommodation in two relevant applications of the Social Web (social network services like Facebook and comments on the webs of hotel reservations agencies). Our ndings will enable Spanish and German companies to interpret the linguistic behavior of their German/ Spanish clients sharing their accommodation experiences on the Social Web. Text linguistics for the contrastive study Sánchez Prieto Raúl Sánchez Prieto Text linguistics for the contrastive study of online customer comments Text-linguistic patterns in German, Dutch, Spanish and French hotel comments and reviews ISBN 978-3-8233-8015-3 Studien zur kontrastiven deutschiberoromanischen Sprachwissenschaft SkodiS 2 <?page no="1"?> Studien zur kontrastiven deutsch-iberoromanischen Sprachwissenschaft (SkodiS) Band 2 <?page no="2"?> Studien zur kontrastiven deutsch-iberoromanischen Sprachwissenschaft (SkodiS) Herausgegeben von Meike Meliss, Bernhard Pöll und Raúl Sánchez Prieto Wissenschaftlicher Beirat Joachim Born (Universität Gießen) José Antonio Calañas Continente (Universitat de València) Mireia Calvet Creizet (Universitat de Barcelona) Juan Cuartero Otal (Universidad Pablo de Olavide, Sevilla) Paul Danler (Universität Innsbruck) María José Domínguez Vázquez (Universidade de Santiago de Compostela) Brigitte Eggelte (Universidad de Salamanca) Christian Fandrych (Universität Leipzig, Herder-Institut) Marta Fernández Villanueva (Universitat de Barcelona) María Jesús Gil Valdés (Universidad Complutense de Madrid) Sybille Große (Universität Heidelberg) José Luis Herrero Ingelmo (Universidad de Salamanca) Thomas Hüsgen (Universidade do Porto) Rafael López Bodineau (Universidad de Sevilla) Macià Riutort Riutort (Universitat Rovira i Virgili, Tarragona) Paloma Sánchez Hernández (Universidad Complutense de Madrid) Bernd Sieberg (Universidade de Lisboa) María Teresa Zurdo Ruiz Ayúcar (Universidad Complutense de Madrid) <?page no="3"?> Raúl Sánchez Prieto Text linguistics for the contrastive study of online customer comments Text-linguistic patterns in German, Dutch, Spanish and French hotel comments and reviews <?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.dnb.de abrufbar. © 2017 · Narr Francke Attempto Verlag GmbH + Co. KG Dischingerweg 5 · D-72070 Tübingen Das Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt. Jede Verwertung außer halb 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 Printed in Germany ISSN 2365-3337 ISBN 978-3-8233-8015-3 This book has been printed with financial support from the State Secretariat for Research, Development and Innovation of the Spanish Ministry of Economy and Competitiveness (“Comments 2.0”, FFI2011-23428). Raúl Sánchez Prieto is professor of German language at the University of Salamanca. <?page no="5"?> Table of Contents 1 Introduction: A text-linguistic study of consumer comments on the social web � 7 2 Text genre „comments“ on the Social Web and corpus description � � � � � � � � � � � � 11 2�1 Comments on online travel and hotel booking websites� � � � � � � � � � � � 11 2�2 Comments on social networking services � � � � � � � � � � � � � � � � � � � � � � � � 26 2�3 Comments on video-sharing platforms and Wiki discussion pages � � 33 2�4 Corpus description � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 35 3 Analysing social web conversations and discourses from a contrastive point of view � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 39 3�1 Communicative macrostructure: text functions and text actions in consumer comments � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 39 3�2 Text-grammatical structures in consumer comments � � � � � � � � � � � � � � 90 4 Conclusions � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 155 5 References � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 159 <?page no="7"?> 1 Introduction: A text-linguistic study of consumer comments on the social web Text linguistics is as a well-established scientific discipline. In fact, describing text structures has been a popular topic in linguistics since the so-called “pragmatic turn”. Over the past 30 years, a significant and growing canon of mostly theoretical works has been published on the study of texts from a linguistic perspective: Halliday/ Hasan (1976 and 1989) paved the way for a better understanding of textual cohesion, Van Dijk (1977, 1980) laid down the essentials of the semantics and pragmatics of text and discourse, De Beaugrande and Dressler (1980, 1981) were among the first scholars to establish a holistic and pivotal text theory, and Coseriu’s hermeneutic approach (1980) defined text linguistics as an individual language activity. The most influential introductions to text and discourse linguistics may be those written by Brinker (1985), Renkema (1993 and 2004), and Adamzik (2004). These and many other works seek to define the scope of text linguistics, and provide generic tools for the study of texts. This monograph builds on three decades of text linguistic research and practice, and focuses on the text linguistic tools that are relevant for the study of customer comments and reviews on the so-called social web. Therefore, this work has to be understood as an attempt to provide a useful, practical and simple method for the contrastive analysis of the linguistic patterns of customer comments. Although especially designed for hotel comments on online travel and booking websites (e.g., Booking.com and Expedia) and on social networking services (e.g., Facebook), it may hopefully be readily adapted to the contrastive studies of other customer comments� Since the monograph intertwines two fields of research that do not often come into contact, namely, linguistics and marketing, it is also intended to be a contribution to studies in marketing communication, albeit from a purely text-linguistic approach. In this respect, the empirical analysis of customer comments as customer-to-customer interactions may prove to be a valuable tool for understanding how consumers motivate their decisions, discuss the quality of a given product or service, recommend its acquisition, praise the producer/ company or its employees or, conversely, advise potential customers not to purchase the goods or services in question. The main purpose of this book is thus to provide text linguists and internationally operating European companies (particularly those active in the tourist industry) with a working tool for evaluating the product and service comments <?page no="8"?> 8 1 Introduction: A text-linguistic study of consumer comments on the social web made by their international customers by determining the communicative macrostructure and the text-grammatical structures that give texture to those comments and reviews� The practicality of the proposed analytical method and its theoretical premises will be tested in a subsequent corpus-based contrastive study of German, Dutch, Spanish and French customer comments on hotel accommodation� This could be considered the secondary aim of this work� The achievement of both goals will be the focus in chapter three, which is the central one here. In the preceding chapter (chapter two), we will address the difficult task of characterising the text genre (or subgenre) “comments” on the social web� The contrastive approach established in the third chapter rests on two pillars, which may be regarded as part of the intended comparative description of customer comments, and will be applied to the corpus to be explored. The first pillar involves the text functions and actions present in customer comments, shaping what we call the communicative macrostructure. The contrastive description of the text functions and text actions displayed in customer comments will focus on revealing the purpose and text features of the informational, appellative and argumentative text functions established by Brinker (1983: 139, 1985: 59). Special attention will be paid to the language used for persuasive purposes, because this is precisely where the main differences between languages (and cultures) may lie at this level of analysis� The second pillar will concern the text-grammatical structures that are relevant for a bior multilingual contrastive analysis. Among them we will study the linguistic elements that create grammatical (and lexical) cohesion and coherence in online customer comments� Concerning the way in which the textual components of a comment and the actual comments that are part of a discussion thread are mutually connected, we will explore the pertinence of the inclusion of cohesive devices of recurrence, deixis and textual connection (De Beaugrande/ Dressler 1981: 3) in the contrastive approach, as has already been suggested for the linguistic analysis of Web 1�0 (Schütte 2004: 103)� The achievement of this monograph’s second aim requires building a multilingual corpus of consumer reactions to hotel stays. As a methodological procedure, the so-called parallel text analysis will be applied (Spillner 1981, van der Eijk 1994, and for its linguistic uses, see Adamzik 2001: 14 and Lüger/ Lenk 2008: 19), giving us four parallel textual corpora (German, Dutch, Spanish and French) with a similar number of threads and comments discussing identical topics in the same social web applications: the comments pages of online travel and hotel booking agencies (Booking.com, TripAdvisor and Expedia), social networking services (Facebook), video-sharing platforms (YouTube), and Wikis (Wikipedia). The corpus data will be evaluated according to the contrastive approach briefly <?page no="9"?> 1 Introduction: A text-linguistic study of consumer comments on the social web 9 outlined above, which will hopefully allow us to draw significant contrastive conclusions from the corpus, and shed interesting light on the commenting behaviour of German, Dutch, Spanish and French hotel clients. Although our contrastive framework will be applied solely to customer comments written in four Western European languages, the text-linguistic approach proposed here should be applicable to any language combination� We hope that researchers will thus be empowered to detect interlingual differences regarding the communicative and text-grammatical structures of customer comments on the social web� <?page no="11"?> 2 Text genre “comments” on the Social Web and corpus description 2.1 Comments on online travel and hotel booking websites Among the wide range of text types used for describing a company’s products or services, the most relevant for the tourism industry may well be the comments posted by former, current or prospective customers on the websites of travel agencies and hotel booking agencies� Unlike comments on social networking services, video-sharing platforms, or blogs, which first need to be discovered by the interested customer, comments on booking agencies such as booking.com, hrs.com, expedia.com or tripadvisor.com, are often just a mouse click away for online customers. Although it is nearly impossible to precisely assess the importance and impact customer reviews have in terms of the booking decisions of hotel guests, having easy access to independent comments on the services offered by a particular hotel is greatly appreciated by those internet users thinking about booking accommodation� The reviews hotel customers share with each other on the net may have a huge impact on booking decisions, so customer comments should be seen as an integral part of customer relationship management (also called customer care management, Ebel 2007: 161). From an economic point of view, these type of comments are, in fact, a written form of word-of-mouth communication� The text type “comments pages of online travel and hotel booking agencies” could be succinctly described as an asynchronous CMC text type that belongs to the so-called social web, and whose main goal it is to discuss the quality of hotel accommodation and recommend or discourage booking a room in a given hotel� Since there is no universally accepted text classification, the focus of the linguistic characterisation of this text type will be based on text patterns. As Gansel/ Jürgens (2002: 53) affirm, “aufgrund der Multidimensionalität der Kategorie Text ist es praktisch nicht möglich, alle potentiellen Texte entsprechend einer einzigen verbindlichen Klassifikation einzuordnen”. Thus, at a descriptive level, this chapter will outline the main textual features which can be found in the text type we will be considering: text function, situationality, text structure, and wording patterns. At the end of the chapter, an attempt will be made to identify the possible differences in text patterns between languages. <?page no="12"?> 12 2 Text genre “comments” on the Social Web and corpus description 2.1.1 Text function and text actions When studying text functions, it should be borne in mind that a language utterance does not “fulfil only one function, to the total exclusion of others” (Brown/ Yule 1983: 1). A hotel customer posting a comment is usually pursuing different aims or intentions expressed by one or more text functions, as can be clearly observed in the following example 1 : (1) Das Hotel hat eine hervorragende Lage. In 5 Minuten ist man mitten in der Fußgängerzone. Die Mitarbeiter waren sehr freundlich und hilfsbereit. Das Frühstücksbuffet war sensationell. Die Sauberkeit der Zimmer und Bäder waren vorbildlich. Wir würden jederzeit wieder dort einchecken. Das Hotel kann man mit ruhigem Gewissen weiterempfehlen (EC D Booking: Jens, Gruppe, Alsfeld, Deutschland, 6. April 2012). In this text, one assertion is undoubtedly informative (“in 5 Minuten ist man mitten in der Fußgängerzone”), while other sentences convey information and persuasion (“das Hotel hat eine hervorragende Lage”, “die Mitarbeiter waren sehr freundlich und hilfsbereit”, “das Frühstücksbuffet war sensationell”...), and the last two assertions are straightforward recommendations (“wir würden jederzeit wieder dort einchecken”, “das Hotel kann man mit ruhigem Gewissen weiterempfehlen”). Information and persuasion go hand-in-hand here, although as is usually the case in texts (Brinker 2005: 89), this comment reflects a dominant function of a persuasive-appellative nature that outweighs the informative one� Although we have pointed out above that a text producer may have different intentions when posting a comment, intention, described by De Beaugrande as a “user-centered notion” and as one of the seven standards of textuality (De Beaugrande 1980: 21), is a “concept defined from the viewpoint of the sender” (Nord 2005: 53), and the only way we can ascertain the intentions of a commenting client is through the analysis of text functions� Customer comments posted on the comments pages of online travel and hotel booking agencies seem to have dominant appellative and informative text functions, according to the definitions of text functions provided by Brinker (1983: 139, 1985: 59). Text functions can be examined in very different ways depending on the study’s proposed aims and on the theoretical framework used by the author. If the approach is an eminently practical one aimed at describing the text structures and functions displayed in multilingual customer comments from a contrastive point of view, an in-depth analysis of the text actions present in informative and appellative settings may be useful. Text actions were initially described by Polenz (1980: 142) in the fol- 1 For practical reasons, no translation of the examples quoted from our corpora will be provided� <?page no="13"?> 2.1 Comments on online travel and hotel booking websites 13 lowing terms: “Jeder Satzinhalt enthält mindestens eine Sprecherhandlung; es können mehrere Satzinhalte die gleiche Sprechhandlung enthalten; jede Sprechhandlung ist in eine übergeordnete Kontexthandlung eingebettet, deren oberste eine wesentliche Texthandlung ist”. Recent text-linguistic studies, such as those by Bendel (1998), Schröder (2003: 7) and Janich (2005), tend to use the term to describe the main linguistic actions underpinning text functions� We will use the term in a somewhat (but not exactly) similar way. For us, linguistic actions denote the use of concrete language patterns pursuing certain textual intentions in a given medial context. In the German example mentioned above, the sentence “die Sauberkeit der Zimmer und Bäder waren vorbildlich” would be classified as “assessing the cleanliness”, which would in return be part of an appellative (rather than informative) text function. As can be inferred, since the context is clear and the text functions present in customer comments are usually reduced to two, no differentiation will be made between main and secondary text actions (“Texthandlungen” and “Teilhandlungen” according to Janich 2005: 122)� After a thorough analysis of both our exemplary and general corpus 2 , the text-type “comments pages of online travel and hotel booking agencies” usually includes at least some of the following text actions underlying the appellative and informative text functions: - Describing the room and the hotel premises (usually appellative, but in some contexts also informative) - Assessing the cleanliness (usually appellative) - Stating or commenting on the hotel location (usually informative, seldom appellative) - Assessing the price/ performance ratio (usually appellative, but in some contexts also informative) - Commenting on the performance of hotel staff (usually appellative) - Describing breakfast choices (informative and/ or appellative) - Indicating parking availability or commenting on parking-related problems (informative and/ or appellative) - Commenting on quietness and privacy (usually appellative) - Recommending or discouraging a stay at a given hotel (usually appellative) 2 For this study, two different corpora have been built up. The “Exemplary Corpus” (or EC) will be mostly used for illustrative purposes in the description of the text genres that are being analysed� The “General Corpus” (or GC) will provide the empirical data used to draw conclusions about the communicative macrostructure and text-grammatical structures of hotel comments and reviews� <?page no="14"?> 14 2 Text genre “comments” on the Social Web and corpus description It may be posited that this text type needs to be thematically organised around one or more of the text actions illustrated above. In a contrastive study analysing text functions and text actions in comments written in different languages, special attention should be devoted to the language used for persuasive (and hence appellative) purposes, as this is precisely where the main differences may lie between languages (and cultures) on this level of analysis� 2.1.2 Situationality Following De Beaugrande/ Dressler (1981: 9), the term situationality has been used for referring to “the factors which make a text relevant to a situation of occurrence”; that is, to the communicative situation in which the text producer and reader are involved� Text linguistics used to disregard the medium and concentrate on the “distribution and meaning of medium-independent word-forms” (Esser 2009: 95). This traditional approach is obviously no longer possible in the case of transmedia text types such as websites, where text and visual (and even audiovisual) components are usually interwoven, leading to a semantic complexity (Lüger/ Lenk 2008: 25) that must be considered when analysing the text and its structure� This is also true for user-generated content on the internet: most social web applications allow the user to post not only text but also upload and display images, videos, etc. As for the text type “comments pages of online travel and hotel booking agencies”, there are no common standards on the textual environment that characterizes the situationality of the textual occurrences, so the researcher is compelled to describe the specific textual settings of every comments page. Adapting the term used by De Beaugrande/ Dressler to our descriptive needs, we assume that there are at least three relevant factors that need to be taken into account when rendering the situationality of this text type: the channel and communicative form, the superficial text structure, and the visual text structure. Out of these three, only the channel and communicative form may be similar across the comments pages of online travel and hotel booking agencies� We will outline the situationality of the customer comments of three large, internationally active online travel and hotel booking agencies: Booking.com, Expedia and TripAdvisor. All three websites allow hotel customers to post their experiences during a given hotel stay on the hotel’s comments page. On one level, the hotel reviews are similar to private online texts like those posted on chats or mailing lists: they are intended only for a (relatively) small group of people interested in the hotel. On another level, however, the comments are as universally accessible as any other website: everyone can read these comments, even those people the posts were not intended for� This text type might best be <?page no="15"?> 2.1 Comments on online travel and hotel booking websites 15 described, according to Sánchez Prieto (2011: 20, tab. 1), as a half-private communicative form. Like blogs or fora, comments are intended for a particular audience, but can be read by all internet users; some of them are typical reviews and hence isolated, while others refer to other comments, and could be seen as part of a developing conversation among commenting hotel customers� Private texts Emails, chats, mailing lists, newsgroups, text on social media platforms Half-private texts Blog posts, forum posts, customer comments Universally accessible texts Webs Table 1. Online communicative forms in computer-mediated communication (Sánchez Prieto 2011: 20) Although the communicative form and the channel are very similar for the comments posted on the web of the three travel agencies, there are at least two significant differences between TripAdvisor and the other two: - Whereas customer comments on Booking.com and Expedia are verified, or, as Expedia puts it, “They [the commenting customers] paid and stayed. We double-checked”, anyone can review a hotel on TripAdvisor. - On TripAdvisor, the hotel management is given the opportunity to respond to the reviews, so customer comments can turn into conversations. The superficial text structure is somewhat different on Booking.com, Expedia and TripAdvisor. Besides the formal restrictions imposed by the different applications used on the websites (e.g. for Booking.com there is a limit on the number of characters, whereas TripAdvisor imposes a 200 character minimum for reviewers), the use of perigraphemic elements differs with regard to the font design, font size and font colour(s). Since perigraphemic characters organise the superficial text structure optically (Schütte 2004: 94), it would be worth noting the differences between the comments on the three webs: - Booking.com uses a rather small classical web font design in dark blue against a white background (Ill. 1). - On Expedia.com, comments are also shown in a classical web font design in dark grey against a white background, but in comparison to Booking.com there is a large heading (Ill. 2). - On Tripadvisor.com, only the first lines of customer reviews are listed on the comments page� The whole review can be read on a separate node� The typeface is bigger and black against a white background, and there is also a blue colour heading� <?page no="16"?> 16 2 Text genre “comments” on the Social Web and corpus description Ill. 1 Typical comment on Booking.com Ill. 2 Typical comment on Expedia Ill. 3 Typical comment on TripAdvisor As we may infer from the three screenshots, there are also certain differences in the visual text structure: <?page no="17"?> 2.1 Comments on online travel and hotel booking websites 17 - The most striking one may be the possibility of uploading an avatar image on TripAdvisor. A reviewer’s avatar alongside the hotel comment should help to develop mutual trust between reviewers and potential customers. From a textual point of view, it would be interesting to explore the relationship between an avatar image and the comment. Avatar images cannot be posted on Booking.com and Expedia. - Another difference to be exploited in a linguistic analysis is the presence of flags denoting the origin of the commenting client (or their country of residence) on Booking.com. The flag displayed is usually linked to the language in which the comment has been drafted� - A recommendation ribbon may be found on reviews featuring hotels recommended by customers on Expedia. - A fourth visual difference between comments posted on the webs of these three big booking agencies concerns the score awarded to the hotel by clients. It appears in a blue speech bubble on Booking.com, as a large yellow figure before a bullet graph on Expedia, and as a green-coloured bullet graph made up of circles, but lacking a scoring cipher, on TripAdvisor. Additionally, TripAdvisor also provides supplementary scoring graphs for at least six service items directly connected with text actions (value, location, sleep quality, rooms, cleanliness, and service). 2.1.3 External structure Customer comments have a certain external appearance depending on the host website of the online travel and hotel booking agency� The term “external structure” is used here according to Gaddy et al. (2001: 102), and refers primarily to “text-structure cues, such as titles and headings, [that] operate by means of structuring the text as a whole”. Thus, we understand external structure to be the constituent parts of the text type� This should not be confused with the term “text structure”, the commonly accepted denomination for the internal text structure, and described by Vater (1994: 63) and Brinker (2005: 20) as “Gefüge von Relationen, die zwischen den Sätzen bzw. den Propositionen als den unmittelbaren Strukturelementen des Textes bestehen und die den inneren Zusammenhang, die Kohärenz des Textes bewirken”. The constituent parts of the text type “comments pages of online travel and hotel booking agencies” are the headline, the score, the main body of the comment or review, the information concerning the guest reviewer and the hotel stay, as well as the feedback request. A hotel reviewer can usually draft the headline and the main body freely, although the review applications based on social web technology do not allow customising the typography. It should be <?page no="18"?> 18 2 Text genre “comments” on the Social Web and corpus description noted that comments posted on Booking.com do not have a heading. The information about the reviewer is also given by the commenting hotel guest in a standardised way. By contrast, the information on the hotel stay is provided by the booking agency (with some exceptions on TripAdvisor). Feedback requests are automatically posted at the end of the hotel reviews by the social web application, and are therefore neither written nor influenced by the commenting guest� There follows a brief description of the constituent parts of this text type� For illustrative purposes, we will divide the three comments shown above in their original form into their constituent parts (tab� 2 to 4)� Information about reviewer/ hotel stay Ralf Älteres Paar, Bad König, Deutschland 26. Dezember 2012 Headline (-) Main body + Die Datails im Haus sind perfekt ausgewählt - alles stimmig - man fühlt sich sofort wohl. Der Blick auf die Stadt - gerade bei Nacht - genial. - Parkmöglichkeiten sind sehr gering. Booking.com hat kein Frühstück im Angebot beinhaltet bzw. ausgewiesen, obwohl das Frühstück beinhaltet war! Somit haben wir es dann leider aus Nichtkenntnis verpasst. Score 8�8 Feedback Fanden Sie diese Bewertung hilfreich? Ja Nein Table 2. Parts of a comment posted on Booking.com (EC D Booking) <?page no="19"?> 2.1 Comments on online travel and hotel booking websites 19 Information about reviewer/ hotel stay juankirm las palmas 2 opiniones Opiniones en 2 ciudades Se alojó el Diciembre de 2006, viajó en pareja Heading “el peor hotel en el k he estado” Main body Me aloje en este hotel en diciembre de 2006, tres noches. Las dos primeras nos kdamos en una habitacion doble y la ultima en una triple� La habitacion doble era enana. El baño era tan grande como un armario y el equipamiento era viejo. La limpieza era pesima, el suelo estaba pegajoso y el “balcon” estaba cerrado con verjas y la calle era muy ruidosa� Solo desayunamos el ultimo dia y ojala no lo hubieramos hecho� el desayuno es un pan k parecia del dia anterior con mantequilla y un cafe con leche. Me gustó — absolutamente nada No me gustó — limpieza y el tamaño de la habitacion Score 1/ 5 Feedback ¿Ha sido útil esta opinión? Sí ¿Algún problema con esta opinión? Pregunta a juankirm sobre Cantabrico Hotel Esta opinión es la opinión subjetiva de un miembro de TripAdvisor, no de TripAdvisor LLC. Table 3. Parts of a comment posted on TripAdvisor (EC E TripAdvisor) <?page no="20"?> 20 2 Text genre “comments” on the Social Web and corpus description Information about reviewer/ hotel stay für alle von Ein verifizierter Reisender Empfohlen Gesendet am 22. Februar 2013 von einem verifizierten Hotels.com Kunden Heading zentrumsnah und guter Anschluss Nahverkehr Main body Sehr guter Service beim Empfang, nett , freundlich und hilfsbereit. Lage für Stadtbesuch Heidelberg super. Sehr guter Anschluss Nahverkehr. Frühstücksangebot entspricht den Erwartungen, frisch, reichlich und ausreichend. Zimmer waren topp zum wohlfühlen. Hotel kann man stets weiterempfehlen. Score 4 Feedback Hilfreiche Bewertung? Danken Sie dem Bewerter 0 Table 4. Parts of a comment posted on Expedia (EC D Expedia) Headings are titles preceding the main text body and indicate “what the central theme of the text is about” (Gunter 1987: 260). Following Kozminsky’s terminology and its reception in media studies, titles would be “advance organizers” (Kozminsky 1977: 482). Since there are hardly any studies on the headings of hotel reviews, we will make do with the research on news headlines by wellrenowned text linguists such as Van Dijk, whose definition of news headlines (1985: 74) can be adapted and re-defined to fit the particularities of comment headings as follows: titles posted by commenting guests are brief sentences that summarise the gist of the hotel review in a personal, catchy style. As is the case with newspaper headlines, the drafting of comment headings appears to differ in different languages and cultures. For example, Spanish headings tend to have an eminently verbal nature, whereas German ones are often defined by their nominal form 3 � The comments body of text is the main component of the hotel review� Unlike other text types or text genres, no textual formalities are expected of the comment’s author. There is no need to structure the text in a particular way, or even sequence the information, so neither a greeting nor a leave-taking expression is expected. The reader’s expectations are reduced to the positive or negative experiences the commenting guest has undergone during the hotel stay� The commenting hotel guest usually launches straight into the review. From the point of view of text linguistics, the review is a succession of text actions that 3 This assertion is based on the results of a preliminary study on online newspaper headlines and comment headings in German, Dutch and Spanish (Sánchez Prieto 2016). <?page no="21"?> 2.1 Comments on online travel and hotel booking websites 21 can be seen as typical for this text type, and which have already been described above� The text actions do not appear to be strung together in a particular sequence, but most reviewers share some common textual assumptions about the order in which they are presented: among the first text actions the reader comes across are “stating or commenting on the hotel location” and “describing the room and the hotel premises”, whereas “recommending or discouraging a stay at a given hotel “ is usually placed at the end of the review� The information concerning the guest reviewer and the hotel stay, as well as the feedback request, are technically speaking not part of the comment, but they help us to understand the context in which the comment is made� 2.1.4 Wording patterns of the text type Like all text types and genres, customer comments are also defined by certain characteristic text grammar, syntactic, morphological and lexical features. These features concern the manner in which comments are linguistically composed from an internal point of view, and will be referred to here as “wording patterns”� One of the most important wording patterns for defining a text type is text grammar� Text grammar in general and cohesion in particular are responsible for creating the “texture” of a given text (Halliday/ Hasan 1976: 2)� There follows an outline of how the texture of customer comments is achieved based on the first data obtained from our multilingual corpus. As defended in Sánchez Prieto (2011: 85) and other authors, such as Schnotz (1994), Von Stutterheim (1997) and Storrer (2004b: 3), we will now ignore the distinction between cohesion and coherence for practical purposes 4 � The most striking characteristic of customer comments regarding their text grammar may be the almost complete absence of cataphoric references, and the limited number of anaphoric ones. The only anaphoric references that often feature in hotel reviews are pronominal forms and some repeated lexical items (like “hotel” or “room”): sentences are linked simply through the systematic repetition of personal and possessive pronouns (usually in the first person) or certain key words, as can be seen in the three following comments (the anaphoric pronominal references are in italics): (2) Somos um casal jovem na casa dos 30, adoramos a nossa estadia! Cama de casal muito confortável... A disponibilidade dos funcionários foi ex- 4 As Rickheit/ Schade (2000: 275) point out, the use of both terms is not clear enough to be applied without further problems. Among text linguists, it appears to be a “nicht unumstrittene Differenzierung” (Storrer 2004a: 16). Nevertheless, we will differentiate both terms later on in the chapter dealing with text-grammar structures� <?page no="22"?> 22 2 Text genre “comments” on the Social Web and corpus description celente, super simpáticos , deixaramnos muito á vontade, arranjaramnos um mapa da cidade e através deles conseguimos saber os melhores locais para visitar, comer, sair á noite etc...Já conhecíamos o conceito de Hostel na net, mas foi a nossa primeira experiência e adora-mos��� se voltar-mos a Braga voltamos ao Braga Pop Hostel ! (EC P Booking: Casal jovem, Almada, Portugal, 9 de Novembro de 2011) (3) Tres surpris par la facade de cet établissement “ pour vivre heureux, vivons caché” nous sommes allé de surprise en surprise. La maitresse de maison est décoratrice et les deux personnes qui nous ont reçu ont été trés agréable� Nous y reviendrons. (EC F Booking: Michel couple d’âge mûr, Frouzins, France, 21 mai 2012) (4) Das Hotel hat sehr interessante und schöne Zimmer, das Personal war sehr freundlich ( wir durften einen Blick in alle leeren Zimmer werfen) und das Frühstück war auch gut. Das beste ist die absolut zentrale Lage! Wir waren im Dublinzimmer, das hat uns gut gefallen. Der Preis des Hotels ist allerdings ein bisschen zu hoch unserer Meinung nach. (EC D TripAdvisor: Voyageuse Pittore, Itzehoe, Deutschland). The lack of more cohesion and coherence devices may be due to the concise nature of the text type itself, whose authors concentrate on text actions, and do not usually elaborate on the opinions expressed� Only when the commenting guest is describing some experience at length is it common to introduce other co-references within the review text. This is, for example, the case in (5), where two lexical replacements can be found (“fin de semana” - “de viernes a domingo”, “excelente - bueno - bien”): (5) Estuvimos pasando un fin de semana ( de viernes a domingo ) en este hotel, y decir que es excelente . Quisimos recordar la primera vez que fuimos a Salamanca y nos hospedamos en este hotel (hace unos 10 años) y decir, que sigue siendo igual de bueno , lo unico que ha cambiado es parte del nombre, pero por lo demas sigue igual de bien (...). (EC E Expedia: Roberto Movi de Madrid, España). Deixis is another means used in customer comments to establish cohesion and coherence, as well as to create texture. The authors of this text type show a clear preference for the first person singular (or plural) in the case of the personal deixis. As for place deixis, which “concerns the specification of locations relative to anchorage points in the speech event” (Levinson 2003: 79), absolute spatial deictic structures are relatively common� These place relators are usually lexical items performing the syntactic function of a prepositional object (as <?page no="23"?> 2.1 Comments on online travel and hotel booking websites 23 in the following Dutch speaking comment), and not grammatical items such as demonstrative adverbs: (6) Compleet ingericht, luxe, comfortabel appartement midden in de Utrechtse binnenstad . Super sfeervol ingericht (heleboel kaarsen), direkt aan de gracht (EC NL Booking: Anoniem, Jong stel, Nederland, 11 juli 2012). Time deixis is also common in hotel reviews, but unlike place deixis, it is more often than not related to the point of utterance, as in (7): (7) Vor einem Monat war ich mit meinem Freund in Heidelberg und wir haben im Holländer Hof übernachtet. Das Hotel befindet sich in mitten der Altstadt (also eine super Lage) und ist sehr gemütlich eingerichtet (...) (EC D TripAdvisor: Urlaubsfee2010, Borchen). From a syntactic point of view, the nominal style prevails in many of the comments that are part of the exemplary corpus, not only in those written in Germanic languages (see 8). When customers elaborate on their experiences, they usually choose the verbal style, as in (9). In those cases, sentence complexity (Stahlheber 1992: 173) is rather limited: simple sentences are more frequent than complex ones. The preferred coordinating conjunction is, in all languages, the copulative coordinator “and”� Juxtaposition is also fairly commonplace� (8) Buena ubicación entre la ciudad y el aeropuerto, servicio de autobús del hotel de enlace con el aeropuerto desde las 5: 00 a�m� Totalmente satisfecho (EC E Expedia: Un viajero verificado Recomendado) (9) Sincèrement, je voyage souvent pour le travail et c’est la seule fois ou tant de prestations étaient proposé ! Qui plus est à ce prix ! ! ! L’accueil y est remarquable, les propriétaires, qui il faut le reconnaitre, sont de vrais passionnés, sont à votre disposition avec un sourire sincère ! Mais réunir, et j’insiste, à ce prix (...) (EC F Booking: Stéphane famille avec enfants Montmorency, France). Those comments written in the verbal style have certain common morphological features� The use of tenses depends largely on the text action the commenting hotel guest is dealing with: stating or commenting on the hotel location is usually expressed in the present tense (example 4), recommending or discouraging a stay at a given hotel normally implies the use of a conditional or a future tense (examples 1 and 3); and guests tend to use past tenses when describing the room, the hotel premises, and breakfast choices (example 1), as well as when commenting on the performance of hotel staff (example 2). Since <?page no="24"?> 24 2 Text genre “comments” on the Social Web and corpus description most comments are written in a personal and colloquial style, the active voice prevails by and large� An initial lexical analysis of the exemplary corpus reveals a very disparate semantic characterisation of customer comments depending on the positive or negative experiences expressed in the hotel review� Text actions that shed a positive light on a hotel are larded with superlatives and adjectives belonging to positive lexical fields: (10) Accueil très aimable ! Un oasis au milieu des immeubles, une belle surprise� Petite piscine sympa pour les enfants, le jacuzzi agrèable pour les parents� Tout était propre . La climatisation fonctionnait bien dans l’appartement du 1. étage. Décor originel au jardin, un petit jungle où se détendre. On a pu garer la voiture dans la rue à 30 m. L’accés au centre est facile et rapide avec le tram, qu’on joint à 300 m. (EC F Booking: Mikko, famille avec adolescents, Céret, France, 5 juillet 2012). Some expressions could even be interpreted as Hochwertwörter or Schlüsselwörter in the sense used by Römer (1968: 99, 132), Janich (2005: 120) and Golonka (2009: 207)� Detrimental comments also use superlatives and comparatives, albeit in a negative context. Alongside adjectives with negative polarity, adverbs entailing a negative assertion about the implied text action are also common, such as “alleen” or “te veel” in (11): (11) De kamer was steenkoud ; verwarming leek het in eerste instantie ook niet te doen� Trappen heel erg steil en smal ( lastig met bagage). Bij het ontbijt alleen witbrood van drie dagen oud . Verlichting op de kamer minimaal (een centrale lamp)� Gehorig . Ontbijtruimte was minder schoon dan de kamer. Ik had alleen een bed en een douche nodig; die kreeg ik ook. Maar € 79.50 voor een overnachting inclusief ontbijt vind ik voor het gebodene toch te veel . (EC NL Booking: Karin, Individuele reiziger, Berkhout, Nederland, 1 februari 2012) 2.1.5 Identifying possible differences between languages So far, text type has been defined and described without referring to any possible differences between languages. This approach is justified by the fact that text type is very common in nearly all Europe’s main languages, and it largely shares common textual features -at least apparently. However, since text types are “cultural entities” (Fix 2006: 260), it would not be surprising at all if an indepth analysis of a larger corpus of customer comments revealed differences in <?page no="25"?> 2.1 Comments on online travel and hotel booking websites 25 the way European hotel customers write hotel reviews. Obviously, such findings can only be made through a detailed study and analysis of a larger corpus� Nevertheless, some possible differences between cultures and languages could be provisionally singled out simply by reflecting upon the language use of international customer comments in our exemplary corpus� One of the most striking differences found in the corpus is the thematic organisation of text actions� Whereas the description of the room and the hotel premises appears to be customary in all languages, Spanish, Portuguese and Italian hotel customers tend to concentrate on assessing the cleanliness of the room and commenting on staff performance, rather than on assessing the price/ performance ratio or describing breakfast choices. Dutch, German and French customers seem to pay more attention to the hotel location, breakfast, and quietness and privacy than their Southern European counterparts. Yet this North- South divide breaks down when text functions (and not actions) are analysed: Spanish, German and Dutch comments tend to be rather informative, while Italian and French ones are more appellative. When present, comment headings (and to a lesser extent the body of the comment) differ considerably: German and Dutch headings are usually written in the nominal style characteristic of some Germanic languages, whereas Romance headings tend to be verbal, if anything. As for wording patterns, the main differences concern the textual grammar and the use of tenses and lexical items. In order to create texture, German, French and Dutch comments have a higher percentage of pronominal forms than Spanish, Portuguese or Italian reviews. In this respect, Iberian customers make greater use of repeated lexical items and lexical replacements. Overall, and without discriminating between text actions, the Dutch, Germans and Spanish tend to use the present tense more often than clients from other countries� Portuguese customers, by contrast, choose the simple past more regularly than other nationalities� The lexical choice also varies greatly from language to language, especially regarding negative comments: German customers are more inclined to use superlatives or comparatives in a negative context than in a positive one, whereas French, Spanish and Portuguese clients employ superlatives above all in positive reviews. Negative comments preferably involve adjectives from negative semantic fields. <?page no="26"?> 26 2 Text genre “comments” on the Social Web and corpus description 2.2 Comments on social networking services Since social networking services are extremely diverse and have quite different goals and website layouts, no linguistic or cultural characterisation of a common text of the type “comments on social networking services” is possible� The social networking environments of a photo-sharing site such as Instagram or of a professional networking site such as LinkedIn are necessarily different from each other, as well as from Facebook. Most of these social networking services are, however, irrelevant to hotels and their clients. Generally speaking, neither hotels nor their potential clients consider networking sites to be a useful platform for advertising, exchanging opinions and interacting with each other. The most useful platform may well be Facebook: it allows hotels to be present and produce networking content in a virtual reality made up of millions of users interacting with each other. This means that Facebook users can share information or ask for it by commenting or posting on a Facebook profile. Consequently, the text type (or, more accurately, subtype) “Facebook comment” could be described as an asynchronous CMS text type that allows multi-party talk and adding media content within or outside an online community� When it comes to hotels, these online communities are usually found around the socalled “Facebook pages” of hotels, that is, business Facebook profiles. In the following, the main textual features of the text type “Facebook comment” will be described� 2.2.1 Text function and text actions As with the comments pages of online travel and hotel booking agencies, Facebook comments from both present and prospective hotel clients and other Facebook users and the hotel management do not usually have just one text function. Even if a comment or a posting is clearly informative, an appellative element may also be present. However, one of the competing textual functions is generally dominant. In the following example extracted from the Portuguese exemplary corpus, potential clients are informed by the hotel management about an event at the hotel restaurant� The appellative function clearly outweighs the informative one, which can be morphologically detected, among other things, by the use of imperative forms: <?page no="27"?> 2.2 Comments on social networking services 27 (12) Sheraton Lisboa Hotel & Spa Queremos que o DIA DA MULHER seja simplesmente delicioso...Junte as suas amigas para almoçar ou jantar e venha celebrar no Lobby Bistro com 50% de desconto� (EC P Facebook: Sheraton Lisboa Hotel & Spa, Lisboa) Assertions posted by users on Facebook hotel pages are usually appellative or informative. The contact function can also be found in some comments. An initial analysis of our multilingual exemplary corpus has shown that Facebook comments on hotels are usually marked by at least two appellative or informative text actions: - giving some piece of information about the hotel, its premises and its services - commenting on events that will take place or have taken place at the hotel It is especially surprising that text actions like “asking for information about the hotel” and “assessing the hotel stay” do not seem to be at a first glance among the most usual ones. It seems that Facebook is not used for sharing information about hotels, but for exchanging opinions about it. As for the assessment of hotel stays, although we come across such a text action in a special review section on the hotel profile (which will not be analysed here), most customers and readers do seem to prefer reviewing portals that allow them to make bookings, like Booking.com or Expedia. Thus, from a macro-communicative point of view, it is constitutive for this text type to be thematically organised around the text actions “giving information about the hotel” and “commenting on events taking place at the hotel”� A thorough empirical study is needed to verify these first impressions and hypothesis� There may also be some divergence concerning both the text actions and the text functions among the languages considered here� 2.2.2 Situationality Since situationality, as described as one of the seven standards of textuality (De Beaugrande/ Dressler 1981: 11), is essential “to consider the situation in which the discourse has been produced and dealt with” (Renkema 2004: 50), it needs to be included in a general description of the text type “Facebook comments”. As with the comments on online travel and hotel booking websites, three main aspects dealing with textuality need to be considered when describing the situationality of this text type: the channel and the communicative form, the superficial text structure, and the visual text structure. <?page no="28"?> 28 2 Text genre “comments” on the Social Web and corpus description On Facebook, hotel comments are transmitted over a computer-mediated multi-channel medium that is usually universally accessible due to the privacy settings chosen by the hotel managers: as with other businesses, hotels prefer open profiles that allow absolute visibility. However, since an internet user must access Facebook in some way to read the postings and the comments and join the social networking site in order to post a comment, this text type can be best described as a half-private communicative form� The superficial text structure on Facebook is very clear and straightforward: a Facebook profile or page consists of a variable number of sections. The central sections are the timeline page, where all conversations between the hotel community manager and the Facebook users can be read, the contact section called “about”, and visitor posts. Whereas the comments posted by possible customers or hotel fans on the timeline page are usually reactions to a posting written by the hotel management, the visitor posts are comments made to the hotel that are not related to any wall post. There are, of course, many other sections that can be displayed on the Facebook page, such as events or a links sections (“upcoming events” and “liked by this page”), and even a section where users can review the hotel (“reviews”). However, for most internet users, only the main wall and the comments posted by people other than the hotel management are relevant� This means that three different, albeit closely related, types of comments can be distinguished: management posts, reactions to these posts, and visitor posts (and the reactions to them). All of them are organised in text boxes, which can also include different media content besides text. Notwithstanding the origin of the posts, visitor posts and posts from the community manager are situated below the profile name and the post’s date of publication. The text appears in black against a white background (ill. 4 and 5). Reactions to the posts by both the community manager and a visitor in the form of comments are also in black, but this time against a pale blue background (ill� 5 and 6)� <?page no="29"?> 2.2 Comments on social networking services 29 Ill. 4 Typical post from the community manager on Facebook Ill. 5 Typical visitor post on Facebook <?page no="30"?> 30 2 Text genre “comments” on the Social Web and corpus description Ill. 6 Typical reactions to a management post The visual structure of Facebook comments is fairly similar for comments posted by the community manager or visitors and the reactions to them� The comment itself is always preceded by the name of the Facebook user (or the profile or page name), the exact date of publication, and the avatar image that identifies the Facebook user. After the text itself, one or more photos, as well as videos, may be added to illustrate the message that the user is trying to convey. Most of the posts the hotel management is responsible for include some sort of graphic material, which can be original or shared. Some of the comments posted by others in the visitors section are also accompanied by visual or audiovisual content. Apart from commonly used emoticons, reactions to these comments are generally only textual� 2.2.3 External structure As with the customer comments posted on the websites of online travel and hotel booking agencies, the external structure of Facebook comments is pre-set by the social networking service. In this respect, “the layout, structure and communicative properties of the software service [i.e., Facebook] itself show characteristics of closed hypertexts” (Eisenlauer 2013: 102). Therefore, only some of the following constituent parts of Facebook comments can be written and edited by the user. A typical Facebook post includes a profile or avatar image, which is displayed over the cover photo and, in our case, is usually the hotel logo. The profile’s name is placed to the left of the profile image, and identifies the author of the post. If the author has posted shared content or has uploaded visual material it is usually stated after his or her name� The posting date and other edition settings, such as information about whether the post has been edited or it is public are stated directly below the profile name. The main body of <?page no="31"?> 2.2 Comments on social networking services 31 text, which is the first part of the post that can be fully composed by the post’s author, follows the posting date. One or more photos, or some other uploaded graphic content, are displayed below the main body. If the community manager or the Facebook user is sharing content from other Facebook pages, other users or other websites, the shared material is embedded in a box that is commonly placed under the photos� Some further information about other edition settings can be displayed afterwards. The final parts of a post provide an opportunity for interaction and communication between users, which have interactive buttons at their disposal (the so-called “Like Comment Share bar”). Users can check how many “likes” the post receives (likes counting bar), and may themselves comment on the topic (comments tab)� These comments can be composed by the users. They are structured in exactly the same way as the main posts. In table 5, a post written by a French community manager has been broken down for illustrative purposes� Profile image Author of the post Hidden Hotel **** Paris shared UNE beauty’s event. Posting date and other edition settings 26 December 2012 Main body Le 27 juin, profitez d’un brunch bio 100% végétarien au Hidden Hotel! Au programme: un délicieux buffet de produits frais, les nouveautés de chez UNE beauty à découvrir, un Nail Colour Bar pour chouchouter vos mains, un atelier DIY pour personnaliser le cabas littéraire Balzac Paris ainsi qu’un bar à thés Løv Organic! Sur réservation uniquement. Photo(s) Embedded shared content - Further edition settings shared text from another user Interactive buttons Like Comment Share Likes counting bar 12 people like this� Comments tab Table 5. Parts of a typical Facebook comment/ post (EC F Facebook) <?page no="32"?> 32 2 Text genre “comments” on the Social Web and corpus description Since the main body is practically the only part of a Facebook comment or post that is composed by a user, no other component of the comment can be considered in a contrastive study� 2.2.4 Wording patterns of the text type Some text-grammatical, syntactic, morphological and lexical features could be considered typical for this text type. Concerning text grammar, the most relevant wording patterns of Facebook comments or posts on hotels are the very common use of pronouns as a cohesive reference device, as well as the widespread use of all types of personal pronouns as a deictic texturing device. Lexical recurrence and other reiterative textual mechanisms that achieve texture also seem to be relatively common among comments posted by both community managers and other Facebook users. In contrast, morphosyntactic cohesion does not seem to appear very frequently, with the possible exception of syntactic parallelisms. An initial, and therefore superficial, analysis of the exemplary corpus has also shown that adjuncts are far less common in this text type than in hotel reviews posted on online travel and hotel booking agencies� Regarding syntax, Facebook contributions are mostly composed in the verbal style, although some comments to these contributions are written in a nounor adjective-based style. In any case, sentence and phrase complexity is limited. As for the lexicon, posts published by community managers tend to unlock positive emotions in the reader by using positively loaded lexical items, such as, inter alia, “Hochwertwörter” or “lexikalische Mittel der semantischen Aufwertung” (Sánchez Prieto 2011: 111, see also Janich 2005: 120). In this respect, comments posted by other Facebook users are more restrained, although they also seem to prefer vocabulary that belongs to positive lexical fields. 2.2.5 Identifying possible differences between languages Although some textual features are expected to be similar in Western European Facebook comments, there should be some differences across these languages in terms of the macro-communicative and text-grammatical structure� At a first glance, the thematic organisation of the text actions in the exemplary corpus does not seem to differ greatly from one language to another. The only exception to this early observation may be the higher use of Facebook postings for asking and giving information about the hotel in Spanish and French (as well as in Italian and Portuguese) comments. As for the wording patterns, the main differences may be found in the use of more texturing devices in languages with a higher level of morphologically <?page no="33"?> 2.3 Comments on video-sharing platforms and Wiki discussion pages 33 marked politeness, such as French and German, and in the lexical choice. In this regard, German and Dutch Facebook users resort to “Hochwertwörter” far more than other Western European ones. 2.3 Comments on video-sharing platforms and Wiki discussion pages Video-sharing platforms (such as YouTube), Wiki discussion pages and blog comments do not seem to be an appropriate channel for customers or online users in general to provide feedback on hotels. Although digital marketing still sees these social web applications as a tool for the management of a hotel’s reputation (see, for instance, Douyère/ Sosthé 2014: 107), and some advertising and marketing companies do offer online marketing for YouTube or Wikipedia, hotels, clients and potential hotel clients do not use online commenting tools other than those present on online travel and hotel booking agencies (particularly on Booking.com, Expedia and TripAdvisor) and on Facebook. This may be due to the booking agencies’ very straightforward business model, where comments on hotels allow potential customers to make a booking decision very quickly and then book immediately on the same website. The multimodal nature of Facebook, which supports audio, video and photo material, as well as the increased connectivity that it offers as a global online social network with millions of real people interacting may also help to explain the fact that no other social networking sites have been able to gain ground as sites for commenting on and reviewing hotels� However, for the sake of completeness, and since there are a few comments on hotels on video-sharing platforms such as YouTube, and also (albeit even fewer) on Wikipedia, a very brief description of the main textual features of those text genres will be provided in the following lines� Only very few text actions can be detected in hotel comments posted on YouTube and Wikipedia. On YouTube, most users either assess the hotel stay in a succinct way, or they asses the quality of the video posted. In either case, the appellative text function prevails. On Wikipedia, potential clients mostly come across just one text action in comments on the Talk page related to articles on hotels: a suggestion made by Wikipedians that the article itself is not relevant� They usually request its deletion (or the deletion of a text section). As for situationality, YouTube and Wikipedia comments could not be more different. Although both text genres are computer-driven and belong to the social web, YouTube comments are embedded in multimodal nodes, and refer mostly to audiovisual content organised in threads that can easily develop <?page no="34"?> 34 2 Text genre “comments” on the Social Web and corpus description into conversations. In this respect, one of the most accurate terms to describe this type of comment may be “schriftliche Dialogizität” (Sánchez Prieto 2015: 292). As with some Facebook comments, this computer-mediated discourse stages orality. No such thing can be asserted from the very scarce Wikipedia comments: they are written utterances embedded in an encyclopaedic context whose only goal is to “provide space to editors to discuss changes to the associated article” (Zidrasco 2010: 578). However, discussions on hotel articles are rare on Wikipedia: with the exception of some editors, whose task is to question the encyclopaedic suitability of articles, most Wikipedians do not seem to be interested in hotels, so no real conversations have arisen. The external structure is slightly more similar on both social websites. In current YouTube comments sections (as in table 6), a profile image appears in first place on the left, if one is available. The author of the comment and the posting date are stated afterwards� The main body is the only subsection of the comment that is not automatically added by YouTube, and can hence be freely composed by the comment’s author. The main body is followed by an interactive menu including a reply button, a likes counter, and a like and dislike icon bar. Profile image (image if available) Author of the comment and posting date AbejaCojonera hace 1 año Main body El acabado sobrio me gusta : ) y tiene una pinta de caro que es para alucinar. Es bonito y sencillo el mueblebar. El detallito del gato de goma de Mariscal mola. Interactive menu: - Reply button - Likes counter - Like and dislike icon bar Discussion tab Table 6. Parts of a typical YouTube comment (EC F YouTube) The external structure of Wikipedia comments is simpler. Apart from the discussion tab, which can be accessed through the edit button in brackets, it includes only three textual parts: a section headline, in which the subject of a new discussion topic about the entry is provided, the main body of the comment, and the user signature with the time and date (see table 7)� This last one is the only part of the comment that cannot be added or edited by a Wikipedian� <?page no="35"?> 2.4 Corpus description 35 Section headline and edit button Lemma [Bearbeiten] Main body Offiziell heißt es Schlosshotel im Grunewald - sollte verschoben werden...? User signature with time and date PerKu 23: 42, 5. Jun 2006 (CEST) Discussion tab Table 7. Parts of a typical Wikipedia comment (EC G Wikipedia) Due to the scarcity of comments on hotels on both YouTube and Wikipedia, it may be an impossible task to identify even basic wording patterns in these text types. Along very general lines, it could be stated that cohesion and coherence in YouTube comments are certainly influenced by the very colloquial nature of YouTube conversations (Wikipedia comments tend to be slightly more formal). This means that cohesion is not expected to be very frequent on YouTube, and that only additive relations should be relevant for the coherence device� The verbal style seems to dominate in nearly all the comments� Contributions on the Wiki Talk page can be expected to provide a wider lexical choice than those on YouTube, as well as a much more elaborate syntactic structure. In any case, the few comments on hotels that can be detected on both social website applications are certainly not enough to identify possible differences between languages. 2.4 Corpus description As has already been affirmed in the introductory chapter, this monograph has a twofold purpose: on the one hand, to provide a feasible working method for the contrastive study of customer comments on the social web based on text linguistics, and on the other, to examine the text linguistic patterns that are present in German, Dutch, Spanish and French hotel comments and reviews. Since our first goal is not circumscribed to the four languages we will be dealing with later on in this book, two partly different sets of data have been collected: - The first set of data, which will be called “Exemplary Corpus” or EC, has been used for illustrative purposes in the description of the computer-mediated text genres that are being analysed (second chapter), as well as in those subchapters of chapter three where the empirical contrastive method is presented. This first corpus is slightly smaller than the second one, but it includes hotel customer comments in two more Western European languages <?page no="36"?> 36 2 Text genre “comments” on the Social Web and corpus description (Italian and Portuguese) of relevance to the tourist industry. This methodological procedure is explained by the nature of our first goal itself: to offer a methodology for the contrastive analysis of customer comments in any (European) language. - The second set of data, which will be called “General Corpus” or GC henceforth, will provide the empirical data used to draw conclusions about the communicative macrostructure and text-grammatical structures in German, Dutch, Spanish and French hotel comments and reviews. Part of the collected data in this second corpus also appears in the first one. In the following, both corpora will be briefly presented. The Exemplary Corpus includes 1800 hotel comments or reviews retrieved from online travel and booking sites (900 comments), social networking services (300), video-sharing platforms (300), and Wiki discussion pages (300). This means that 300 comments/ reviews have been collected for each one of the six languages considered here (German, Dutch, Spanish, French, Italian and Portuguese). In contrast, the General Corpus is made up of 2000 comments and reviews (500 per language)� For each one of the four languages whose textual conventions are examined in chapter three (German, Dutch, Spanish and French), 300 comments on online travel and booking sites were collected, along with 100 from social networking services, and 50 from video-sharing platforms and Wiki discussion pages (table 8)� Exemplary Corpus General Corpus Social Web Applications No. of comments/ reviews No. of comments/ reviews Online travel and booking sites: - Booking.com - Expedia - TripAdvisor 150 - 50 - 50 - 50 300 - 100 - 100 - 100 Social networking services: - Facebook 50 - 50 100 - 100 Video-sharing platforms: - YouTube 50 - 50 50 - 50 Wiki discussion pages: - Wikipedia 50 - 50 50 - 50 Total number: 300 500 Table 8. Corpus size per language <?page no="37"?> 2.4 Corpus description 37 Although the General Corpus is not a large one, it is suitable for our contrastive purposes due to its design, with its main principles being neutrality and visibility at the moment of compilation� The one hundred customer comments per language for each one of the three more internationally renowned online travel and booking sites considered here (Booking.com, Expedia and TripAdvisor) include the first ten positive comments from the five highest-rated hotels (thus 50 comments), and the first ten negative comments from the five lowest-rated hotels (thus another 50 comments) located in one tourist destination per country� These towns (Bonn, Utrecht, Salamanca and Montpellier, as well as Verona and Braga for the Exemplary Corpus) have a population ranging between 100,000 and 350.000 people. On TripAdvisor and Expedia, not enough reviews of hotels in these towns were available that matched the retrieval criteria, so hotels in other tourist destinations had to be taken into consideration (Heidelberg, Berlin and Munich in Germany, Amsterdam in the Netherlands, Madrid in Spain, and Paris in France). On Booking.com, only positive comments rated over 8 and negative comments under 6 are included in the corpus� The clients who posted the comments are individuals living in their own country (Germany, the Netherlands, Spain, and France, as well as Italy and Portugal). This means that a comment made in German by a Swiss national, or a comment made in Spanish by a Mexican client is not considered, and the same applies to a comment made in German (or Spanish) by a German (or Spanish) client living in France. The origin of the customer is usually stated before or after the hotel review. Booking.com reviews were retrieved in July 2012. TripAdvisor reviews were retrieved in March 2013 (German, Spanish and Dutch) and in October that same year (French). The reviews posted on Expedia were compiled in March 2013 (German and Dutch), and in April (Spanish) and October (French) of that same year. When quoting hotel reviews posted on online travel and booking sites that belong to both corpora, the particulars stated before or after the review in the original context are given in brackets. For example, “(EC D Booking: Manfred, Älteres Paar, Nordholz, Deutschland, 21. August 2011)” is to be interpreted as follows: the hotel review belongs to the German Exemplary Corpus and was written by Manfred, who is currently living in (or comes from) Nordholz (Germany), and he wrote the review on 21 st August 2011. The one hundred Facebook comments per language that compose the General Corpus regarding comments on social networking services were collected in a similar way: only the first ten Facebook page results (and possible comments) displayed by Facebook’s search engine for the search parameter “hotel + capital (or most important city)” were considered. Each result is the Facebook page of a hotel. The hotel must have an active Facebook page and more than ten cus- <?page no="38"?> 38 2 Text genre “comments” on the Social Web and corpus description tomer and management comments. Only the first five posts written by the hotel management or the community manager and the first five comments written by other Facebook users (“others”) in the local language were considered. Due to the impossibility of storing Facebook comments adequately, we were forced to work with two different sets of comments, thereby building up two (online) subcorpora, one for the communicative macrostructure and another one for the text-grammatical structures� The General Corpus used for the empirical analysis of the communicative macrostructure includes Facebook comments up to 7 th March 2014, and has been labelled “Subcorpus I”. For the analysis of the text-grammatical structures, the date of retrieval was 1 st July 2015 (“Subcorpus II”). Quotations from these two sets of subcorpora are stated in brackets, and besides the name of the commenting individual or hotel they also include the name of the hotel on whose page or wall the comment was posted, for example “(EC D Facebook: Mövenpick Hotel, Berlin)”. As for the corpus building of video-sharing platforms and Wiki discussion pages, the General Corpus and the Exemplary Corpus are identical, and include the first ten results displayed on the YouTube and Wikipedia search engines, respectively� The search parameter was “hotel in + capital” in both cases� On YouTube, each result is a hotel’s YouTube video page. On Wikipedia, it is the Wiki Talk page of the respective entry. Only the first five comments on the video or on the Wikipedia article were included in the corpus� Since some video pages and Wiki Talk pages did not reach five comments, more comments were retrieved from the next results displayed on the search engines. On Wikipedia, articles on legendary upscale hotels (e.g., Hotel Adlon) were not considered. The YouTube corpus was retrieved in July 2012, and the Wikipedia one in July 2013. The quotation format is very similar to the one employed in the case of Facebook comments. With the exception of Facebook comments, both the General and the Exemplary Corpus were retrieved manually from the web and stored in two different electronic databases, in Excel, and in the freeware corpus analysis toolkit AntConc, which has been partially used for the text-grammatical analysis of customer comments. However, due the limitations of both Excel and AntConc (or other corpus software such as, for instance, WordSmith) concerning text functions, text actions and most text structuring devices, we were forced to analyse the data manually� <?page no="39"?> 3 Analysing social web conversations and discourses from a contrastive point of view 3.1 Communicative macrostructure: text functions and text actions in consumer comments The contrastive description of the text functions featured in multilingual customer comments will focus on revealing the purpose and text features of the informative, appellative (or argumentative) text functions established by authors such as Searle (1975), Longacre (1976) and Brinker (1983, 1985). When analysing a given customer comment, decoding the intentions of the commenting client should be the first step taken in the process of explaining and interpreting the hotel review, or any other comment on a given product or service. In this respect, this would be the first of four “functional planes” that would instantiate the text (Esser 2009: 167): the encoder’s textual intentions (first functional plane) would be conveyed by propositions (second functional plane), which in return would be formulated by words (third functional plane), and would subsequently constitute their final written representation (fourth functional plane). This chapter will carefully examine the textual intentions expressed by propositions� Our goal is to establish a working method for analysing the text functions present in customer comments in different social web environments: on the comments pages of online travel and hotel booking agencies, in social networking services, on video-sharing platforms and on Wiki discussion pages. Starting from a functional point of view, we understand communicative text functions to be text linguistic structures that are expressed by text actions; working with text actions rather than with text functions will allow us to set up a more detailed tertium comparationis. In return, this will allow us to conduct a truly in-depth practical contrastive analysis� This would be particularly important in the case of those social web applications that allow completely free postings, and do not standardise customer comments, such as social networking services or video-sharing platforms. Comments can be so varied, and common comparative grounds so exiguous, that in certain cases a contrastive analysis would not be possible without a unitary descriptive model� This certainly applies to examples (1) and (2): although both were posted on Facebook (i.e., the same social web application), the first German comment largely follows the tex- <?page no="40"?> 40 3 Analysing social web conversations and discourses from a contrastive point of view tual conventions of a letter, whereas the second one is simply a short text message written without any regard for the orthographic conventions of French: (1) Harald Ohr Hallo liebes Facebook-Team von Hotel Berlin, Berlin Heute ist der Gutschein für 2 Ü/ F für 2 Personen von eurem EM-Gewinnspiel angekommen� Ich wollte mich recht Herzlich dafür bedanken. Meine Frau und ich freuen uns schon auf den Besuch in Ihrem Hotel : ) LG H. Ohr (EC D Facebook: Hotel Berlin, Berlin) (2) Tiko Amg sublime hotel et tre covivialle jai kife grave tre tre bien! ! ! (EC F Facebook: Shangri-la hotel, Paris) When formulating the unitary description model that will be proposed for analysing textual intentions and actions from a contrastive point of view, some complications are certain to arise� One of the most awkward ones involves measuring text functions and actions. Since language utterances do not “fulfil only one function, to the total exclusion of others” (Brown/ Yule 1983: 1), customer comments usually pursue different aims or intentions, and those intentions may be expressed by just one text function and text action, or by several ones. Even a single sentence may convey information and persuasion, or information and description, etc. However, one text function or text action usually outweighs and dominates. In order to establish the dominant text function (or text action), the comment should be broken down into textual units of meaning. A textual unit of meaning could be defined in this respect as the semantic and textual expression of a single concept. As a general rule, a textual unit of meaning is related to only one text function or text action in the sense of Brinker’s “thematische Entfaltung” (1985: 59ff), so the assignment of a text function or action to a textual unit of meaning should not pose a problem. After segmenting a customer comment according to this criterion, we should be able to determine the whole comment’s dominant text function or the dominant text action of parts of it. If a hotel review like (3) were to be segmented by searching for textual units of meaning, we would have the intentional structure expressed in table 1: (3) Die Lage [unit of meaning 1], da wir Freunde besucht haben die in der Nähe wohnten [unit of meaning 2], deshalb haben wir das Hostel ausgewählt [unit of meaning 1]. Als wir um kurz vor 17 Uhr ankamen, bekamen wir den Schlüssel [unit of meaning 3], aber als wir aufs Zimmer kamen war es noch schmutzig vom <?page no="41"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 41 Vorgänger. Keine frisch überzogenen Betten, schmutzige Handtücher, voller Mülleimer usw [unit of meaning 4]. Nachdem wir uns beschwert haben, wurde das Bett frisch überzogen und Mülleimer geleert, aber Tisch und Boden waren immer noch schmutzig [unit of meaning 5]. Die Gemeinschaftsdusche und Toilette war schmutzig, wir haben vorgezogen nicht zu duschen [unit of meaning 4]. Wir empfehlen dieses Hotel auf keinen Fall weiter und werden auch nicht wieder dorthin gehen [unit of meaning 6] Ramona [unit of meaning 7] (EC D Booking: Älteres Paar, Karlsruhe, Deutschland, 15. Januar 2012). Unit of meaning Text function Text action 1 Informative function stating or commenting on the hotel location 2 Informative function stating or commenting on the hotel location 3 Informative function describing the check-in 4 Informative function assessing the cleanliness 5 Appellative/ informative function assessing the cleanliness/ commenting on the performance of hotel staff 6 Appellative function recommending or discouraging a stay at a given hotel 7 Contact function signing the comment Table 1� Textual units of meaning - example (3) and text functions/ actions (following Brinker 1992: 104) As can be inferred from the table, the dominant text function is clearly the informative one, as is apparently the case with most customer comments: the appellative and informative text functions seem to dominate� A further complication that will need to be tackled regards the theoretical framework used to develop the unitary description model. Various attempts have been made to describe text functions and text actions� Most of them ( Jakobson 1960, Gülich/ Raible 1975, Searle 1975, and Große 1976, only to quote the classic ones) are based on Bühler’s organon model. Three other authoritative studies on text linguistics -Longacre (1976), Wehrlich (1976) and Brinker (1985)also take Bühler’s communicative functions into account, while also seeking to offer a more homogeneous classification of text functions. As Brinker puts it (2005: 105): “Im Unterschied zu den besprochenen Klassifikationsansätzen soll unsere Abgrenzung von Textfunktionen auf einem einheitlichen <?page no="42"?> 42 3 Analysing social web conversations and discourses from a contrastive point of view Kriterium beruhen, und zwar auf der Art des kommunikativen Kontakts, die der Emittent mit dem Text dem Rezipienten gegenüber zum Ausdruck bringt”. Whereas Longacre’s approach (1976: 197ff) is still heavily influenced by Bühler, and distinguishes four basic text functions (expository, narrative, hortatory or behavioural and procedural text functions or “deep structure genres”, as he calls them), Wehrlich’s description of text functions (1976: 39ff) is based on five cognitive processes underlying five “text types”: description, narration, exposition, argumentation, and instruction. Brinker’s approach (1985: 104ff) is an extended revision of Searle’s illocutionary speech acts. Although Brinker’s illocutionary model is also described mainly from a theoretical point of view, the five proposed text functions are aimed at analysing text structures on an eminently practical level. This may be the reason why Brinker’s approach and his five text functions (informative, appellative, obligation, contact and declarative) have enjoyed such acceptance in practical contemporary studies. This is also the case with the textual description of our consumer comments. Due to the actual nature of contrastive studies, Brinker’s framework is a highly applicable one. Customer comments regularly use three of Brinker’s text functions, namely, the informative, appellative and contact ones. Propositions containing obligation and declarative functions are extremely rare in hotel reviews, so these two functions will not be taken into consideration� Text actions underpin text functions� Since text actions are determined by the intent of the commenting clients, it is they who dictate their comments’ text function or functions, being dependent upon “sprachlichen Handlungen desselben Sprechers oder verschiedener Sprecher” (Van Dijk 1980: 212). Text actions have been employed in recent studies on text linguistics (e.g., Bendel 1998, Schröder 2003, and Janich 2005) instead of Brinker’s “thematische Entfaltung”� One of the most extensive descriptive models of text actions is the “Handlungsmodell” proposed by Von der Lage-Müller (1995: 50ff), who identifies three different hierarchical levels of internal organisation for text actions: 1� The global textual intent of the sender is denoted by the main text actions or “Texthandlungen”� When the main text actions are characteristic for a text type (i.e., commenting on the hotel stay in the case of hotel customers), they are mandatory� 2� The main text actions are implemented by secondary text actions or “Teilhandlungen”� The dominant text action “commenting on the hotel stay” could, for example, be specified by the secondary text actions “describing the room and the hotel premises” or “assessing the cleanliness of the premises”� 3� Secondary text actions can be supplemented by an additional third level of text actions of a very different nature. For example, when assessing the <?page no="43"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 43 cleanliness of his room, a hotel client could elaborate on the particular circumstances related to the lack of toilet paper, or mention the moment when they found a dirty hand towel� Von der Lage-Müller’s approach allows a detailed analysis to be made of highly standardised text types, such as obituary notices (Von der Lage-Müller 1995) or advertisements (Bendel 1998, Janich 2005), but it needs to be reformulated if it is to be applied to customer comments� Since the main text action is clearly the same for all hotel reviews, and for most other comments about hotel stays, it can be argued that Von der Lage-Müller’s “Texthandlungen” are not relevant for an empirical analysis. They simply reveal the subject on which the hotel client is commenting, and could be compared to Brinker’s “Textthemen” within his theory of “thematische Entfaltung” (Brinker 1985: 55). The term “main text action” will be used here in a similar way to Von der Lage-Müller’s “Teilhandlungen”: it will denote the use of linguistic propositions to express a certain textual intention that helps to shape the form of a given text function. Exposing a comment’s main text actions will sometimes not be enough to allow an in-depth analysis. In that case, secondary text actions will be identified. The textual intention of the commenting hotel guest shown in the second unit of meaning observed in example (4), a Spanish Facebook comment, can be stated as “describing the room and the hotel premises”� (4) Magnífico hotel en la zona céntrica de Madrid [unit of meaning 1]... sus nuevas instalaciones y habitaciones MARAVILLOSAS ! ! ! Estuvimos el martes 3 de Julio y tuvimos la suerte de estrenar 2 habitaciones de la tercera planta que estaban comunicadas entre si [unit of meaning 2]. Todo salió perfecto! ! [unit of meaning 3] Gracias tambien a los trabajadores y dueños del hotel por lo bien que nos trataron! ! [unit of meaning 4] Seguro que volveremos! ! [unit of meaning 5] saludos desde Zarauz! ! [unit of meaning 6] Izaskun Ajates Garcia [unit of meaning 7] (EC E Facebook: Hotel Artrip, Madrid) The main text action present in the second unit of meaning helps to develop the informative and appellative functions dominating the comment, and consists of three or four secondary text actions: “sus nuevas instalaciones y habitaciones MARAVILLOSAS ! ! ! ” (evaluating the premises), “estuvimos el martes 3 de Julio” (specifying the date), “y tuvimos la suerte de estrenar 2 habitaciones de la tercera planta que estaban comunicadas entre si” (describing the location of the rooms)� <?page no="44"?> 44 3 Analysing social web conversations and discourses from a contrastive point of view Since different text types may have different textual intentions, and hence very disparate text actions, our practical contrastive approach to text actions will be outlined separately for each relevant text type in the corresponding section� The three communicative functions observed in hotel comments posted on the websites of online travel and hotel booking agencies, in social networking services, on video-sharing platforms, on Wiki discussion pages, and in blogs will be discussed after the following overview of the informative, appellative and contact functions� The overall purpose of the informative function essentially involves the communication of information about the hotel the commenting client has stayed at and/ or about the stay itself. In this way, they are helping other travellers to make a decision about booking a room at the hotel in question. Passing on information is not usually explicitly indicated by performative formulae in either text type of relevance to our study� Performative utterances are only common on Wikipedia discussion pages when a user asks for more information about the facts stated in an article: (5) Diskussion: Estrel Hierbei handelt es sich um einen offiziellen Pressetext des Estrel Berlin, der allen Internet-Usern auf der Unternehmens-Homepage frei zur Verfügung steht� Größtes Hotel Es wird jetzt nicht klar ob das Estrel das größte Hotel Deutschlands an sich ist oder das größte der 4+-Kategorie. Wie siehts aus? Und woran wird die Größe gemessen? Zimmerzahl, Quadratmeter, Umsatz ... ? Rainer ... 17: 39, 16. Sep 2005 (CEST) (EC D Wikipedia: Hotel Estrel Diskussion) The informative content of most hotel comments includes a major evaluative component. As long as the evaluation does not seek to influence the recipient, that is, information remains the main issue of the hotel review or comment, the informative function outweighs the appellative one. In fact, “diese thematische Einstellung ist kennzeichnend für die Textsorten ‘Gutachten’, ‘Rezension’, ‘Leserbrief ’ usw.” (Brinker 1985: 106), and thus also for hotel reviews. A good example is the following German comment posted on Booking: (6) Das Frühstück entsprach dem Preis-Leistungsverhältnis Die Sauberkeit lies sehr zu wünschen übrig. Im Bad hingen noch die Handtücher vom Vorgänger. Die Bodenfließen waren sehr verschmutzt von den Vorgängern. Die kleine Küche war noch mit Essensresten bestückt. Im Kühlschrank befanden sich noch angebrochene Lebensmittel <?page no="45"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 45 die auch bereits Schimmel anstzten. Auf unseren Hinweis dieser Mißstände hin wurde uns gesagt, daß die Monteure, die vorher das Zimmer belegten, ja wiederkommen würden. Wir haben daraufhin daws Hotel verlassen, ohne unsere 2. Übernachtung wahrzunehmen. Eine Entschädigung haben wir nicht bekommen� (EC D Booking: Manfred, Älteres Paar, Nordholz, Deutschland, 21. August 2011) However, many comments are open to interpretation, and whether the dominant function can be classified as informative or appellative depends heavily on the context. Such is the case of example (7). On the one hand, some pieces of information are given to the reader, while on the other, the comment may be read as a straightforward advertisement for the hotel (and thus with an appellative character): (7) La perfection n’existe pas, mais je dois dire que la propriete TOUTOUNE s’en approche, ses propriétaires sont sympas, serviables et surtout trés attentifs, Le dépaysement total, en centre ville etre en un instant dans une Yourte, on se croit en mongolie + le spa, jacuzzi, sauna, enfin génial, bon moment de détente����� (EC F Booking: Fabrice, voyageur individuel, montpellier, France, 2 mars 2012)� Although the informative function in comments usually reveals certain grammatical characteristics, such as the use of assertions or the indicative mood (often embodied in a past tense), it would be unwise to establish a list of prototypical grammatical features for this text function: the informative content of a unit of meaning is not necessarily linked to any grammatical treatment� The appellative function could be characterised as the linguistic expression of personal preferences on the part of the emitter in order to influence the recipient in a given way. In the specific case of hotel customer comments, reviewers report their experiences during a hotel stay so that potential hotel clients may adopt their opinions and attitudes towards the hotel and then proceed to book (or not) a room at this hotel. Thus, following and interpreting Brinker (1985: 108), the intentional force of the appellative function can be twofold: either the sender wants to convince the recipient to adopt their views about the hotel and its staff (appellative function by opinion building; the term Brinker coins for it is “Meinungsbeeinflussung”), or they would like to encourage the recipient to perform a specific action (appellative function by influencing behaviour or “Verhaltensbeeinflussung”). The second type of appellative function implies the first one: the first step to influence someone’s behaviour is to convince them to adopt <?page no="46"?> 46 3 Analysing social web conversations and discourses from a contrastive point of view your own views about something. The final units of meaning of the following comment about a Parisian hotel (8, in italics) are a good example of a hotel review containing an implicitly dominant appellative function (“en résumé”): after describing the featured hotel’s lack of cleanliness and poor conditions (informative function and/ or appellative function by opinion building), the author of the comment advises the booking agency to remove the hotel from the list of available hotels on the web (appellative function by influencing behaviour): (8) RIEN, la chambre était très très simple, pas propre de même que la salle de bain (douche avec des moisissures), équipements vieillot, seul la literie est confortable. La chambre est au rez de chaussée, donnant sur la cour de +/ - 5m2 du voisin donc beaucoup de promiscuité. De plus le groom de la porte d’entrée est cassé ou tellement vieux qu’il ne fonctionne pas correctement se qui fait que la porte claque à chaque passage et pour nous qui étions à coté cela a généré beaucoup de gène, surtout en pleine nuit. La salle du petit déjeuner est sale, le mobilier est désuet et bancale. La cuisine est aussi très sale. Le petit déjeuner dans son ensemble est insuffisant. En résumé vous feriez bien de supprimer ce prestataire de vos listing car il fait du tort à votre site internet. Ce serait bien également de vérifier la qualité des prestations proposés par votre site internet. (EC F Booking: Pierrette, couple d’âge mûr, GENISSIEUX, France, 20 juin 2012) When the commenting client is just expressing their opinions without recommending booking a room, the user is only seeking to influence the readers’ opinion about the hotel, as in (9): (9) ”¡Un Hotel de Ensueño! ” Ha sido una experiencia maravillosa. Es un hotel en pleno centro histórico de Salamanca que sorprende en todos los aspectos. Desde luego que este será el hotel donde me hospede la proxima vez que vaya. En general, todo fue excelente, desde el personal del hotel con una amabilidad y atencion exquisita, hasta el mas mínimo detalle, tanto en la bienvenida como en la habitacione, que por cierto son acogedoras y a la altura de un gran hotel, que te hacen sentir como en casa. El desayuno de buffet, fantastico y si lo deseas, te lo sirven en la habitación sin ningun coste adicional. (EC E Expedia: Jose Manuel R, Albatera, Comunidad, Valenciana, Spain) Distinguishing between the appellative function by opinion building and by influencing behaviour may not be important for most prototypical appellative text types such as advertisements, instruction manuals, recipes or sermons, but it is crucial for analysing online hotel customer comments from a contrastive <?page no="47"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 47 perspective. First, the use of text actions within an appellative context that aims to convince a potential hotel client to book (or not) is necessarily different from that intended solely to change an existing attitude about a hotel. Second, the distinction may be significant for explaining a possible divergent use of the appellative function by customers that belong to low and high context cultures in Hall’s sense of the term (Hall 1976). Units of meaning with contact function are not very common among most customer comments. However, a reader of hotel comments may come across expressions of interpersonal communication between reviewers or users in conversations taking place on Facebook, and sometimes on the comments pages of booking agencies (especially TripAdvisor). The purpose of the contact function is to establish or maintain contact between the emitter and the recipient� Since “Kontakttexte sind vielfach an feste gesellschaftliche Anlässe geknüpft, die den Ausdruck der psychischen Einstellung des Emittenten verlangen” (Brinker 1985: 118), in other words, both the commenting client and the prospective one fulfil a linguistic social expectation, this language function is a highly standardised one, and hence very easily recognisable. That includes greetings, forms of address, congratulations, excuses, etc. By way of illustration, let us consider the following Facebook conversation: (10) Kiki Nung liebe grüsse an das ganze maritim team [unit of meaning 1]. ich war vor langer zeit mit einer gruppe da und habe mich sehr wohl gefühlt [unit of meaning 2]. hoffentlich kann ich bald wieder kommen. : ) [unit of meaning 3] lg [unit of meaning 4] kiki [unit of meaning 5] Maritim Hotel Berlin vielen dank kiki [unit of meaning 1], das hören wir gern [unit of meaning 2] und grüßen ganz herzlich zurück [unit of meaning 3]. auf hoffentlich bald! [unit of meaning 4] das mari-team [unit of meaning 5] (EC D Facebook: Maritim Hotel Berlin, Berlin) Facebook user Kiki Nung contacts the hotel management on the hotel’s Facebook site and greets the hotel staff (unit of meaning 1). At the end of her conversational turn, she uses a leave-taking expression (unit of meaning 4) and signs off (unit of meaning 5). The answer she obtains from the hotel management is structured in nearly the same way: the only difference involves the first unit of meaning, an expression of thanks instead of a greeting. Units of meaning with a contact function are also sometimes observed in hotel reviews apart from Facebook (and TripAdvisor), mostly embedded within an appellative context, such as the Dutch thanking formula “bedankt voor” in (11): <?page no="48"?> 48 3 Analysing social web conversations and discourses from a contrastive point of view (11) Erg vriendelijk personeel, mooie groene omgeving op loop afstand van het centrum. Heerlijk gegeten in de brasserie buiten op het terras met uitzicht over de tuin, de zon scheen, mensen gezellig aan de wijn en hapjes ....echt vakantie gevoel� Bedankt voor een heerlijk ontspannen weekend! (EC NL Booking: Edward, Jong stel, Den haag, Nederland, 30 mei 2012) Having described the relevant text functions observed in hotel customers’ comments on the basis of our exemplary corpus and outlined the problems we will be facing when using the term “text action”, the next sections in this chapter will examine the text actions that constitute the communicative structure of the comments pages of online travel and hotel booking agencies, social networking services, video-sharing platforms, Wiki discussion pages, and blogs. 3.1.1 Communicative functions and text actions on the comments pages of online travel and hotel booking agencies Former or current customers who post a hotel review usually have the textual intention of informing or influencing the opinion or behaviour of prospective bookers. In the first case, the communicative function that dominates a comment, or a unit of meaning within it, is clearly informative: prospective clients are informed about the quality of the accommodation and the hotel services offered by a specific establishment. In the second case, the text function is overwhelmingly appellative� The contact function is never the main communicative function of a customer comment, but from time to time it can be observed in a review building up its own unit of meaning. This is true for all the German, Dutch, Spanish, Portuguese, French and Italian comments included in both our exemplary and general corpus. Western European hotel clients posting a comment on online travel and hotel booking agencies seem to share similar text actions that articulate the mainly appellative and informative text functions observed in the corpus. The statistically most significant text actions in terms of occurrences and importance are the following ones (always illustrated by at least one example): - Describing the room and the hotel premises: “Es ist sehr modern eingerichtet mit neuen Möbeln” (EC D Booking: Markus, Junges Paar, Böhmenkirch, Deutschland, 3. Juli 2012). This text action is usually appellative, although it can also include some neutral information, and hence be of an informative nature� - Assessing the cleanliness: “Wir waren sehr angetan von der Sauberkeit im Hotel” (EC D Booking: Elfriede, Älteres Paar, Reichenbach, Deutschland, 5. Juni 2012)� This text action is usually an appellative one� <?page no="49"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 49 - Stating or commenting on the hotel location: “Das Hotel liegt genau im Zentrum Bonns” (EC D Booking: Mandy, Junges Paar, Hohenleipisch, Deutschland, 12. Juni 2012). Although this example is clearly informative, and only states the location of the hotel, other comments on the hotel location are more expressive, and therefore appellative, such as “Un lieu improbable et exotique en pleine ville avec espace sauna, piscine, verdure, oiseaux” (EC F Booking: Saul, voyageur individuel, ELANCOURT, France, 11 juin 2012). - Assessing the price/ performance ratio: “Badkamers mogen moderner, als deze gemoderniseerd zijn dan is de prijs kwaliteit verhouding optimaal” (EC NL Booking: Jef, Ouder stel, Cadier en Keer, Nederland, 1 mei 2012). Most sentences referring to the price/ performance ratio are appellative, but sometimes some descriptive information is provided: “75 euro a notte (con gli sconti book ing...) per un hotel a 1 stella che ne vale mezza! ” (EC I Booking: Corrado, Coppia giovane, torino, Italia 03 aprile 2012). - Commenting on the performance of hotel staff: “Hotelmitarbeiter waren sehr nett, freundlicher Empfang” (EC D Booking: Familie mit älteren Kindern, Mühlhausen, Deutschland, 1. November 2011). This text action is usually appellative when it concerns the helpfulness and politeness of hotel employees� - Describing breakfast choices: “Koffie en suiker was er wel alleen geen koffiemelk, beetje jammer : )” (EC NL Booking: Marcella, Vriendengroep, Delft, Nederland, 25 april 2012). It can be informative and/ or appellative. - Indicating parking availability or commenting on parking-related problems: “Meter el coche en el ascensor del parking es bastante angosto” (EC E Booking: Alberto, Familia con niños pequeños, Madrid, España, 5 de junio de 2012). Like the last one, this text action may be informative and/ or appellative. - Commenting on quietness and privacy: “Han puesto doble ventana. Ya no se oye nada de ruido del exterior” (EC E Booking: Xavier, Pareja joven, Valencia, España, 21 de mayo de 2012). Nearly all the comments are appellative, with an implicit or explicit informative content� - Recommending or discouraging a stay at a given hotel: “Wir würden jederzeit wieder dort einchecken. Das Hotel kann man mit ruhigem Gewissen weiterempfehlen” (EC D Booking: Jens, Gruppe, Alsfeld, Deutschland, 6. April 2012)� This text action is appellative by its mere nature� - Expressing one’s gratitude: “(...) et merci encore pour l’excellent petit-déjeuner de loin supérieur à celui habituellement proposé dans les hotels impersonnels et de même prix .... merci ! ” (EC F Booking: Patrick, couple d’âge mûr, Chatellerault, France, 30 avril 2012). This text action belongs to a contact text function� - Congratulating the hotel: “Parabéns pela qualidade dos serviços! ” (EC P Booking: Ana, Viajante individual, Portela - LRS, Portugal, 26 de Abril de <?page no="50"?> 50 3 Analysing social web conversations and discourses from a contrastive point of view 2012). As well as the previous and ensuing text actions, this one also involves a clear contact function� - Making a direct request to the hotel: “(...) muchos hoteles deberían tomar nota y hacerles una visita para aprender como se gestiona un hotel. Continúen así caballeros” (EC E TripAdvisor: Jimbo Tarragona, opinión escrita el 8 febrero 2012) Although these text actions are commonly or very commonly used among German, Dutch, Spanish, Portuguese, French and Italian commenting hotel guests, their actual use varies greatly from one language to another. An efficient statistical method that can be applied to assess the differences in the multilingual use of text actions consists in calculating the percentage of occurrences for each text action. The comparative figure in the resulting rate of occurrences (%) can be either the total amount of hotel reviews or the total sum of units of meaning (Σ) found in a corpus. The following table includes all the relevant information that needs to be collected for an in-depth analysis of the text functions and text actions observed on the comments pages of online travel and hotel booking agencies� Informative text function - Σ-(%) Appellative text function - Σ-(%) - by-opinion-building-(bob) - by-influencing-behaviour- (bib) Contact text function---Σ-(%) Text action Σ % Σ % Σ % describing the room and the hotel premises - bob - bib - bob - bib assessing the cleanliness - bob - bib - bob - bib stating or commenting on the hotel location - bob - bib - bob - bib assessing the price/ performance ratio - bob - bib - bob - bib commenting on the performance of hotel staff - bob - bib - bob - bib <?page no="51"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 51 describing breakfast choices - bob - bib - bob - bib indicating parking availability or commenting on parking-related problems - bob - bib - bob - bib commenting on quietness and privacy - bob - bib - bob - bib recommending or discouraging a stay at a given hotel - bob - bib - bob - bib expressing one’s gratitude congratulating the hotel making a direct request to the hotel Total comments - Σ (100%) Table 2� Contrastive analysis of text functions and text actions The coloured cells of the table mean that the respective text functions and text actions are not compatible� The table chart will be used in this chapter to display the text functional and actional structures of hotel reviews for each language� This procedure will allow us to show the differences found in the quadrilingual general corpus. It could prove useful not only to determine the exact communicative structure of comments about hotels, as the descriptive model could also be applied to nearly all the text types that exist on the social web. In the following description of the communicative structure of Western European hotel reviews, we will take the total number of hotel reviews collected in the corpus (300 comments posted on the online travel agencies Booking.com, Expedia and TripAdvisor, that is, 100 comments per online booking agency) as the comparative figure for calculating percentage rates. The evaluation of the Spanish corpus data according to the model described above sheds interesting light on the commenting behaviour of Spanish hotel cli- <?page no="52"?> 52 3 Analysing social web conversations and discourses from a contrastive point of view ents. An overwhelming majority of comments reflect three text actions, namely, “describing the room and the hotel premises” (89.2% of all Spanish comments), “assessing the cleanliness” (70�2%) and “commenting on the performance of hotel staff” (65.2%). When describing rooms and hotel premises, Spanish customers usually resort to the appellative function (67.9%), be it by opinion building as in (12) or by influencing behaviour (39.3% and 28.6%, respectively). In this respect, only one fifth of all comments (21.3%) are mainly informative (see 13): (12) Este hotel de 4 estrellas debería haber perdido, al menos, 2 de ellas hace bastante tiempo porque las instalaciones, así como, las habitaciones están anticuadas, y solo se le ha dado un lavado de cara a algunas habitaciones, pero de manera muy superficial (GC E Expedia: Un viajero verificado, Publicado el 27 de septiembre de 2012 por un cliente verificado de Hotels. com) (13) Las habitaciones son pequeñas, pero la cama es comodísima y el baño tiene ducha con hidromasaje (GC E Booking: Carlos, Pareja mayor, Madrid, España 21 de febrero de 2012) The same is also true for the text action “assessing the cleanliness”: about 60% of Spanish comments reflecting this text action are appellative (from which slightly more than 42% are appellative by opinion building, and 17.3% by influencing behaviour), but just 10.6% are informative. The appellative nature of the text action “commenting on the performance of hotel staff” is even more salient: the appellative function is overwhelmingly dominant (61�2% of the corpus comments reveal such a function: 38.6% by opinion building, and 22.6% by influencing behaviour), whereas informative text actions are marginal (4%). The use of the mainly informative text action “stating or commenting on the hotel location” and the generally appellative text action “assessing the price/ performance ratio” is also very common (60.2% and 59.9%, respectively). Spanish commenting clients usually restrict themselves to stating the location of the hotel or commenting on the location from an informative point of view (42.6%). Nevertheless, appellative remarks like (14) can also be found in the corpus (17�6%): (14) La ubicación es inmejorable, a cinco minutos de la Plaza Mayor andando (GC E Booking: Juan, Pareja joven, MADRID, España, 19 de septiembre de 2011) When assessing the price/ performance ratio, there is a clear preference for appellative text actions (49.6%), particularly for appellative propositions by opinion building (35%), such as (15). The percentage of appellative text actions <?page no="53"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 53 whose goal is to influence the reader’s behaviour is still higher than the percentage rate of informative propositions (14�6% vs� 10�3%)� (15) Inmejorable relación calidad precio (GC E Expedia: Aacc210, Madrid, España, Opinión escrita el 11 mayo 2012) Nearly half the reviews also include comments about parking availability or elaborate on parking-related problems (47.2%, from which 24% are of an informative nature, and 23.2% are mainly appellative). The percentage of comments in which the former client recommends or discourages a stay at a given hotel is surprisingly low. Fewer than one third of the evaluated reviews (29%) incorporate such an appellative text action (e.g., see 16). One plausible explanation for this communicative behaviour may be the high-context profile of Spanish culture (Copeland/ Griggs 1985: 107): after reading the whole review, which typically includes a description of the room and the hotel premises, and comments on room cleanliness and staff performance, the prospective hotel client should be able to discern whether the former customer is recommending the hotel or not� (16) En definitiva no lo recomiendo y no volveremos (GC E TripAdvisor: Vicente_Palma34, Palma, España, Opinión escrita el 28 agosto 2010) Other mostly appellative text actions that seem to be neglected by commenting Spanish customers refer to the description of breakfast choices and the assessment of quietness and privacy. As for the first text action, breakfast choices are commented on in only 35�6% of the reviews� More than 23% of all comments including such a text action have an appellative nature� Many of them (15�3%) are appellative by opinion building, but the percentage rates of appellative propositions by influencing behaviour (8%, see example 17) and of informative text actions (12.3%, see example 18) are also significant: (17) (...) te mandan a la cafetería que hay al lado del hotel a desayunar y mientras ves que ponen tostadas grandes con tomate y aceite a la gente a ti por ser del hotel y tener el desayuno incluido solo te dan un cafe y una tistadita pequeña, ni zumo ni nada, súper cutre y encima el camarero un mal educado y borde, como que re estuviera haciendo un favor (GC E Booking: Miriam, Grupo de amigos, Valladolid, España, 29 de mayo de 2012) (18) El desayuno con gran variedad (GC E TripAdvisor: Andertxa, Lesaka, España, Opinión escrita el 12 abril 2012) <?page no="54"?> 54 3 Analysing social web conversations and discourses from a contrastive point of view The situation is very much the same regarding quietness and privacy: just 29.2% of all reviews comment on quietness and privacy, mostly from an appellative perspective (22.6%). Even though the informative function is not very common for this text action (6.6%), it can still be detected in the corpus: (19) Lo que menos me ha gustado es que las habitaciones delanteras son ruidosas porque a pesar de estar situado en una calle peatonal pasan coches y motos por allí, durante toda la noche (GC E Booking: Begoña, Persona que viaja sola, Gijón, España, 22 de diciembre de 2011) The three text actions that constitute the contact text function of online hotel reviews are too negligible to be mentioned in our corpus evaluation� The detailed results obtained from the evaluation of the Spanish corpus are shown in table 3� Informative text function - 300 (100%) Appellative text function - 300-(100%) - by-opinion-building-(bob) - by-influencing-behaviour- (bib) Contact text function - 8 (2.6%) Text action Σ % Σ % Σ % describing the room and the hotel premises 64 21�3% - bob 118 - bib 86 - bob 39�3% - bib 28�6% assessing the cleanliness 32 10�6% - bob 127 - bib 52 - bob 42�3% - bib 17�3% stating or commenting on the hotel location 128 42�6% - bob 40 - bib 13 - bob 13�3% - bib 4�3% assessing the price/ performance ratio 31 10�3% - bob 105 - bib 44 - bob 35% - bib 14�6% commenting on the performance of hotel staff 12 4% - bob 116 - bib 68 - bob 38�6% - bib 22�6% describing breakfast choices 37 12�3% - bob 46 - bib 24 - bob 15�3% - bib 8% <?page no="55"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 55 indicating parking availability or commenting on parking-related problems 72 24% - bob 56 - bib 14 - bob 18�6% - bib 4�6% commenting on quietness and privacy 20 6�6% - bob 43 - bib 25 - bob 14�3% - bib 8�3% recommending or discouraging a stay at a given hotel - bob - bib 87 - bob - bib 29% expressing one’s gratitude 4 1�3% congratulating the hotel 2 0�6% making a direct request to the hotel 3 1% Total comments - 300 (100%) Table 3� Text functions and text actions in Spanish customer comments on online travel and hotel booking agencies The data obtained by the evaluation of the German corpus could help us understand the communicative behaviour of German commenting hotel customers� The most relevant text actions for them are “describing the room and the hotel premises”, “assessing the cleanliness”, “commenting on the performance of hotel staff” and “describing breakfast choices”, which feature, respectively, in 72.5%, 71.3%, 67.2% and 63.5% of the entire corpus of comment. As in Spanish, the text action “describing the room and the hotel premises” has mainly an appellative nature. As such, it is observed in 58.9% of the hotel reviews. In this respect, many units of meaning are appellative by opinion building (39.3%), like (20). Nevertheless, the percentage of appellative propositions by influencing behaviour (28.6%, see example 21), as well as of those units of meaning embedded in an informative setting (21.3%, see example 22) cannot be overlooked. (20) Außerdem roch alles moderig, feucht, schimmelig, die Möbel sind offensichtlich noch original 70er, das Bett war vom restlichen Raum mit <?page no="56"?> 56 3 Analysing social web conversations and discourses from a contrastive point of view einem ekligen “Gardinen-Raumteiler” getrennt (GC D TripAdvisor: JK83, Karlsruhe, Bewertet am 6. November 2008) (21) Wer durchgelegende Matratzen mag ist da gut auf gehoben... (GC D Booking: Danie, Familie mit kleinen Kindern, Bad Salzuflen, Deutschland, 4. Januar 2012) (22) Jedes Zimmer einer Weltstadt gewidmet und entsprechend landestypisch eingerichtet (z.B. New York, Athen, Amsterdam, Rom, Sylt, Zermatt, etc.). Ungewöhnlich und unverwechselbar; mal etwasanders im Vergleich zu den internationalen Hotelketten (GC D TripAdvisor: Iberis_10, Schriesheim, Deutschland, Bewertet am 27. September 2010). According to the results extracted from the corpus analysis, Germans usually assess cleanliness by using the appellative function, and try to influence the reader’s opinion (37.3%) or even their behaviour (26%), as in (23): a total of 63.3% of all German comments are appellative. Just 8% reflect an informative text function, which is mostly linked to a positive review, and often only consists of a describing adjective (example 24): (23) So ein schlechtes Hotel habe ich noch nie gesehen es ist so dreckig unfassbar EKELHAFT! ! ! ! ! ! ! ! ! ! der emfang sowas von unfreundlich haben noch am selben Abend das Hotel verlassen werden jetzt ein Anwalt ein schalten da sie unser Geld nicht zurück geben wollen dieses Hotel gehört GESCHLOSSEN! ! ! ! ! ! das sollte mal das gesundheitsamt hin schicken ! ! ! ! ! ! ! ! ! ! Bitte nicht buchen! ! ! ! ! ! ! ! ! ! (GC D TripAdvisor: bzrf, Bewertet am 7. April 2012) (24) Sauber, ordentlich, gute Lage (GC D Booking: Lutz, Alleinreisender, Fürth, Deutschland, 25. Juni 2012) Another text action with a strong appellative nature is “commenting on the performance of hotel staff”. Almost 65% of all collected comments refer to staff performance in appellative terms. Within this text action, there is a clear preference for the appellative function by building opinion, as in example (25), (44.3% vs. 20.6% for the appellative function by influencing behaviour): (25) Die Mitarbeiter des Hotels haben wir als sehr freundlich, hilfsbereit und zuvorkommend erfahren. Sie waren mit praktischer Hilfe (Suche nach Änderungsschneiderei, Tipps,Hilfe mit Gepäck) jederzeit zur Stelle (GC D Expedia: Philip, Osterholz-Scharmbeck (bei Bremen), Gesendet am 24. Februar 2011) <?page no="57"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 57 Another very widespread text action in German is “describing breakfast choices”. In 53.9% of the reviews analysed, this text action has an appellative nature, and in one way or another aims to influence the reader’s opinion or behaviour as regards the hotel under review. For example, in (26) a negative opinion about having breakfast at the hotel may have a negative impact on the reader. An informative description of breakfast choices is not very common, and is present in just 9.6% of the comments (see example 27). When the unit of meaning is predominantly informative and positive, most German customers use the fixed expression “(sehr) gutes Frühstück”. (26) Aldi Wurst in kleinen Mengen zum Frühstück, Tütenkaffee oder Billigtee, was nach Heu schmeckt, Mineralwasserflasche steht zwar da war aber immer leer (GC D Booking: Dmitry, Alleinreisender, Wiesbaden, Deutschland, 25. März 2012) (27) Frisch gepresste Säfte, Brote, Brötchen in vielen verschiedenen Sorten, marmeladen, Wurst - es war einfach rundum alles dabei (GC D Booking: Mandy, Junges Paar, Hohenleipisch, Deutschland 12. Juni 2012) A similar fixed expression appears to be popular among German hotel clients for assessing the price/ performance ratio from an informative (and positive) point of view: “(sehr) gutes Preis-Leistungs- Verhältnis”. Not unsurprisingly, clients who single out positive aspects about hotel pricing policy usually make use of the informative function and very short assertions, mostly in the form of a nominal phrase, as in (28). This finding would explain the unusually high results for the informative function. However, when assessing the price/ performance ratio, Germans prefer the appellative function to the informative one, especially when they exercise criticism, as in (29). This text action is documented in 60.6% of all comments, but whereas 38.3% of them include appellative units of meaning (27% by opinion building and 11.3% by influencing behaviour), the text action we are dealing with is mainly informative in only 22�3% of the hotel reviews� (28) Lage, Preis, Frühstück, Freundlichkeit (GC D Booking: Volker, Älteres Paar, Rüthen, Deutschland, 31. Oktober 2011) (29) Das hatten wir wegen des günstigen Preises auch gerne in Kauf genommen, aber unter dem Strich ist geschenkt hier noch zu teuer (GC D Trip- Advisor: Sabine B, Augsburg, Deutschland, Bewertet am 2. Juni 2011) By contrast, the text action “stating or commenting on the hotel location”, observed in slightly more than 56% of all comments, is mainly of an informative nature (34%, see example 30). The appellative function is documented in 22.2% <?page no="58"?> 58 3 Analysing social web conversations and discourses from a contrastive point of view of the comments (17�6% by opinion building -see example 31and 4�6% by influencing behaviour). (30) Das Hotel liegt direkt an der Fußgängerzone (GC D Expedia: Brigitte, Mainz, Gesendet am 26. Juli 2011) (31) Wir waren im Januar dort, die Straßen lagen voller Schnee, aus unserem Zimmer konnten wir aufs Schloß schauen, es war sehr romantisch (GC D TripAdvisor: annaluise9, Berlin, Deutschland, Bewertet am 17. November 2012) A high degree of importance also seems to be attached in German to the text action “commenting on quietness and privacy”: 53.6% of the comments examined include some remark about this matter� Most of the observations regarding quietness and privacy are purely appellative (42%), be it by building opinion (29%, see example 32) or influencing behaviour (13%). An informative text action “commenting on quietness and privacy”, as in (33), is observed in only 11.6% of all comments� (32) Durch die Ecklage an einer nicht so ruhigen Straße war es etwas lauter als erwartet, aber auch für Lärm-empfindliche wie mich war’s noch gut auszuhalten. : -) (GC D Booking: Christina, Alleinreisender, Stuttgart, Deutschland, 7. Mai 2012) (33) Da das Zimmer zum Hof gelegen war, war es sehr laut. da aus dem Nachbarhof und Haus sehr viel Geräusch drangen (GC D Expedia: Hans-Werner, Herne, Gesendet am 16. Mai 2011) It is remarkable that German clients seem to recommend or discourage a stay at a given hotel far more often than the Spanish do: 41�3% of all comments have this appellative text action, which can be expressed explicitly by using the verb “(weiter)empfehlen”, an adverbial phrase like “immer wieder! ”, or indirectly by stating that the commenting client would book the same hotel again and again, as in (34): (34) Da wir zukünftig mindestens zweimal pro Jahr für ein Wochenende in Bonn verbringen werden und das Hotel fußläufig zu erreichen ist, werden wir wohl immer im Hilton einchecken (GC D Booking: Markus, Älteres Paar, Salzkotten, Deutschland, 29. Februar 2012) Indicating parking availability or commenting on parking-related problems is not so common among German hotel clients. Remarks made accordingly are found in just 28.6% of the entire corpus of comments. More often than not, such remarks are informative (17%)� <?page no="59"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 59 As in the Spanish corpus, the contact function is documented only in a handful of cases� The detailed results obtained from the evaluation of the German corpus are shown in the following table� Informative text function - 300-(100%) Appellative text function - 300-(100%) - by-opinion-building-(bob) - by-influencing-behaviour- (bib) Contact text function - 7 (? %) Text action Σ % Σ % Σ % describing the room and the hotel premises 41 13�6% - bob 94 - bib 83 - bob 31�3% - bib 27�6% assessing the cleanliness 24 8% - bob 112 - bib 78 - bob 37�3% - bib 26% stating or commenting on the hotel location 102 34% - bob 53 - bib 14 - bob 17�6% - bib 4�6% assessing the price/ performance ratio 67 22�3% - bob 81 - bib 34 - bob 27% - bib 11�3% commenting on the performance of hotel staff 7 2�3% - bob 133 - bib 62 - bob 44�3% - bib 20�6% describing breakfast choices 29 9�6% - bob 124 - bib 38 - bob 41�3% - bib 12�6% indicating parking availability or commenting on parking-related problems 51 17% - bob 28 - bib 7 - bob 9�3% - bib 2�3% commenting on quietness and privacy 35 11�6% - bob 87 - bib 39 - bob 29% - bib 13% <?page no="60"?> 60 3 Analysing social web conversations and discourses from a contrastive point of view recommending or discouraging a stay at a given hotel - bob - bib 124 - bob - bib 41�3% expressing one’s gratitude 2 0�6% congratulating the hotel 3 1% making a direct request to the hotel 2 0�6% Total comments - 300 (100%) Table 4� Text functions and text actions in German customer comments on online travel and hotel booking agencies According to the results obtained by analysing the corpus, Dutch hotel guests comment on the performance of hotel staff particularly often. A description of the room and/ or the hotel premises and the room’s cleanliness is also very common. Over 72% of all Dutch comments include some information about the performance of hotel staff. The information is usually given from an appellative point of view: the appellative function by opinion building and by influencing behaviour is present in 48.3% and 19.3% of all analysed comments, respectively. For an example, see (35): (35) Enthousiast en warm welkom door gastvrouw met uitleg van kamer. Vriendelijk personeel bij cocktailbar (GC NL Booking: Ellen, Ouder stel, Geervliet, Nederland, 2 mei 2012) The second text action found in the corpus in percentage terms (68�6%) refers to room cleanliness. This text action also has an eminently appellative nature, which can be observed in 57�6% of all comments� Only 11% of the comments with this text action are of an informative nature. Almost as often as they assess cleanliness, Dutch clients describe the room and the hotel premises. This is the case in 68.5% of all comments. Again, the appellative function (49.9%) outweighs the informative one (18.6%). In an international comparison, the text action “stating or commenting on the hotel location” can be detected especially often: 67.9% of all Dutch comments include such a text action, whose function is informative in 39�3% of all comments (see example 36) and appellative in 28�6% (example 37): <?page no="61"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 61 (36) Het hotel ligt aan de oude gracht en binnen 2 min ben je in het centrum van Utrecht (GC NL TripAdvisor: mirjamreis, Amsterdam, Nederland, Beoordeeld op 20 januari 2013) (37) Dit appartement is een fantastische beleving op een schitterende locatie midden in de binnenstad van Utrecht (GC NL Booking: Eric, Jong stel, Berkel-Enschot, Nederland, 11 juni 2012) Another three text actions with a higher occurrence rate in Dutch than in the other languages targeted in this study are “describing breakfast choices”, “assessing the price/ performance ratio” and “recommending or discouraging a stay at a given hotel”. More than 66% of all Dutch hotel reviews include observations on the breakfast service provided by the hotel� These observations are mostly appellative (55%-43% appellative by opinion building and 12% by influencing behaviour), as in the following example: (38) Avond eten was goed ontbijt wat minder de broodjes werden niet aangevuld de roerei was 1 grote drap en nog lang niet gaar (GC NL TripAdvisor: Jersam86, Beoordeeld op 2 februari 2011) The price/ performance ratio is a topic of comment in 61.5% of all hotel reviews, both from an appellative point of view (39�9%) and from an informative one (21.6%). The following sentence is a good example of this text action; its text function is appellative by opinion building: (39) Prijs kwaliteit vond ik slecht (GC NL Booking: Franca, Individuele reiziger, Leiden, Nederland, 20 oktober 2011) Given the name the Dutch have for being straightforward people, it will hardly surprise anyone that over half of the Dutch comments (52.3%) either include a hotel recommendation or urge not to book the reviewed hotel� When making a recommendation, the Dutch seem to have developed an idiomatic expression, “(echt) een aanrader! ”. When discouraging a hotel stay, the most usual phrases are “niet voor herhaling vaatbaar” and “(ik) Zou er niet meer heen gaan”. In any case, the text action is usually very explicit, as in (40): (40) Ga er lekker eten, feesten en drinken, maar slaap ergens anders (GC NL Booking: Gezin met jonge kinderen, Leiderdorp, Nederland, 27 oktober 2011) The detailed results for all Dutch (and French) text actions are shown in table 5, which sums up all the results obtained for the four languages� For French hotel customers, describing the room and the hotel premises and commenting on the performance of hotel staff and assessing cleanliness seem to <?page no="62"?> 62 3 Analysing social web conversations and discourses from a contrastive point of view be the most important text actions. A room or hotel description is observed in 83�5% of the entire corpus of comments� This text action is eminently appellative (65.9%), be it by opinion building (36.3%) or by influencing behaviour (29.6%). Nevertheless, the informative function is also present in a relatively high percentage of comments (17.6%). This is, for example, the case in (41): (41) Hôtel quasi-neuf et très propre, chambres très confortables avec tout ce qu’il faut dedans (cafetière, réveil, douche italienne très grande, TV grand écran...) (GC F TripAdvisor: 656guy30, ALES, Avis écrit le 21 octobre 2013) The performance of the hotel staff is assessed in 68% of all comments, overwhelmingly from an appellative point of view (65%)� When the appellative function is carried out by opinion building (38% of all comments), French customers tend to use the fixed expression “accueil (très) agréable”. In this respect, the text function attached to this text action is appellative by influencing behaviour in a remarkable 27% of the corpus comments, as in (42): (42) L’accueil y est remarquable, les propriétaires, qui il faut le reconnaitre, sont de vrais passionnés, sont à votre disposition avec un sourire sincère! (GC F Booking: Stéphane, famille avec enfants, Montmorency, France, 3 mars 2012) The text action “assessing the cleanliness” can be detected in two thirds of the analysed French comments (67.6%). As with the other two text actions commented above, the appellative function here also outweighs the informative one: 58% of the comments assessing cleanliness are appellative (38% by opinion building and 20% by influencing behaviour). Between 50% and 60% of the corpus comments also include the text actions “stating or commenting on the hotel location”, “assessing the price/ performance ratio” and “recommending or discouraging a stay at a given hotel”. Nearly 60% of the reviews comment on the hotel location: the customers make use of the informative function in 30�3% of all comments, and of the appellative one in 28.9% (see example 43, where the text action is mostly informative): (43) L’hôtel est idéalement situé (place de la comédie) (GC F Expedia: Un voyageur vérifié, Publié le23 août 2013) The price/ performance ratio is assessed in more than half the comments (56.9%), and from an eminently appellative point of view (47�6%)� One of the most interesting findings of the corpus analysis concerns the high percentage of comments recommending or discouraging a stay at a given hotel (52.3%). If a hotel is recommended, most French hotel customers just state their wish to stay at the hotel again sometime in the future, using verbal expressions like “nous revien- <?page no="63"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 63 drons” or “j’y retournerai”. If, on the other hand, the client issues a warning not to book the hotel, then fixed expressions like “(hotel) à ne pas recommander” or “à ne pas conseiller” are also used. However, in that case, French customers often resort to very direct and telling exhortations like (44), (44) or (45): (44) Odieux, evitez absolument (GC F TripAdvisor: Dom N, Paris, France, Avis écrit le 14 décembre 2012) (45) Je ne le recommanderai même pas à mes ennemis! ! (GC F Booking: Samantha, voyageur individuel, eynesse, France, 30 juin 2012) (46) Voilà un bon conseil, n’y allez pas (GC F Booking: Giraudot, famille avec adolescents, laxou, France, 11 juillet 2012) Other text actions are not very common among French customer comments. They comment on quietness and privacy in 40.2% of cases, and on breakfast choices in 33% of all comments. Both text actions are carried out essentially using the appellative function� The opposite is true with the text action “indicating parking availability or commenting on parking-related problems” (present in 33�3% of the comments): the informative function is slightly more common here. As is the case in the Spanish, German and Dutch corpus, the text actions considered with a contact function also record a very low percentage in French. It may be noteworthy that French has the highest percentage in this respect, especially when the French want to express their gratitude to the hotel staff or management (4.3%), as in (47): (47) (...) et merci encore pour l’excellent petit-déjeuner de loin supérieur à celui habituellement proposé dans les hotels impersonnels et de même prix .... merci ! (GC F Booking: Patrick, couple d’âge mûr, chatellerault, France, 30 avril 2012) As can be inferred from the results obtained by analysing our multilingual corpus (see also table 5), the text functions and text actions that structure hotel reviews from a communicative point of view are comparable in the four languages. Spanish, German, Dutch and French hotel customers seem to draft hotel reviews following similar communicative patterns� This is especially true for three mainly appellative text actions (“assessing the cleanliness”, “assessing the price/ performance ratio” and “commenting on the performance of hotel staff”), and for the three text actions expressing the contact function (“expressing one’s gratitude”, “congratulating the hotel” and “making a direct request to the hotel”). In this respect, cleanliness is nearly as important a factor for booking a hotel room for the Germans and Spanish as it is for the Dutch or French. This is verbalised by resorting to the appellative function. Yet even if there are almost <?page no="64"?> 64 3 Analysing social web conversations and discourses from a contrastive point of view no differences in the communicative orientation of the text action in the hotel reviews studied here, a minor divergence can still be observed when the internal nature of the appellative function is carefully examined: the number of comments that include this text function reflecting the appellative function by influencing behaviour is nearly ten percentage points higher in German than in Spanish, and five or six points higher than in Dutch or French. This means that German hotel customers have a greater intentional and persuasive force in this respect than Spanish or French reviewers. Quite the opposite is true for the text action “commenting on the performance of hotel staff”: In French and Spanish hotel reviews, the appellative function by influencing behaviour is more common than in German or Dutch ones. A similar phenomenon can be detected when we compare the text action “assessing the price/ performance ratio”� The text action records an almost identical percentage in the four languages. In all of them, the appellative function is clearly the one most used. However, whereas the informative function is present in one third of the comments with this text action in the two Germanic languages analysed (22.3% and 21.6% in German and Dutch, respectively, out of a total of 60.6% and 61.5% for the text action), the informative function prevails only in one sixth of the Spanish and French comments with the text action (10.3% and 9.3%, respectively, out of a total percentage of 59.9% and 56.9%). So when it comes to commenting on the price/ performance ratio, a considerably higher percentage of German and Dutch customers than Spanish or French ones prefer to resort to the informative function� Two other text actions (“stating or commenting on the hotel location” and “describing the room and the hotel premises”) present some slight differences in the four languages. Dutch customers state or comment on the hotel location more often than commenting guests of the other nationalities� The widest percentage gap is detected between Dutch and German customers (nearly 12 percentage points). It seems that knowing the exact location of a hotel (as well as its pros and cons) is a more important criterion in the decision-making process for the Dutch than for the Germans, Spanish or French. Although this text action is mostly informative in all four languages, the use of the appellative function is very different in French, on the one hand, and in Dutch, German, and especially Spanish, on the other: whereas more than two thirds of the Spanish reviewers using this text action in their comments restrict themselves to informing potential clients about the hotel location (42�6% of all Spanish comments with these text actions are informative, and only 17.6% are appellative), for those French clients commenting on the hotel location the appellative function is as important as the informative one (28�9% vs� 30�3%)� When describing the room and the hotel premises, the differences detected do not concern the language <?page no="65"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 65 function (which is mainly appellative in our multilingual corpus) but instead the percentage of reviews including this text action. For the Spanish and French, such descriptions seem to be essential� The text action is therefore present in 89.2% and 83.5% of all Spanish and French comments, respectively. This percentage is surprisingly much lower in German (72.5%) and in Dutch (68.5%). The contrastive conclusions extracted from the corpus data are very significant in the case of the four following text actions. Breakfast seems to be far more important to German and Dutch hotel clients than to Spanish and French ones. In fact, whereas only 33% of all French and 35.6% of all Spanish comments contain a reference to breakfast, breakfast choices are described in nearly twice that number of Dutch (66.3%) and German (63.5%) comments. Given the differences in breakfast habits between Northern and Southern Europeans, this is hardly surprisingly� Another telling difference that has been observed by analysing the quadrilingual corpus involves the text action “indicating parking availability or commenting on parking-related problems”. North of the Pyrenees, the total number of hotel reviews that include a comment on parking issues does not exceed 30% or 35% of all comments. The figures range from 20.3% in the case of the Dutch to 28.6% and 33.3% in the case of the Germans and French, respectively. However, parking for Spanish hotel customers seems to be a matter of concern, and therefore important when deciding to book a hotel: 47�2% of all Spanish comments raise the issue of parking availability and comment on parking-related problems. A possible explanation for this huge divergence (nearly 27 percentage points between Spanish and Dutch, and almost 20 points between Spanish and German) may be that our corpus consists solely of hotels in towns: since German, Dutch and to some extent also French towns have very good rail connections, German, Dutch and French city tourists prefer the train for travelling to their destination. By contrast, Spanish tourists rely heavily on their own car for city trips� A southern-northern divide can be observed when commenting on quietness and privacy. This usually appellative text action is present in a majority of German and Dutch comments (53.6% and 54.3%, respectively). The rate of French comments with this text action is nearly 15 points lower (40.2%), and that of Spanish hotel reviews even lower (29.2%). On the one hand, these results are very insightful. On the other, they tellingly help to corroborate the perception that German and Dutch clients grant more importance to quietness (and privacy) than Spanish and French hotel customers. The last text action that records a striking difference between the languages analysed has been labelled “recommending or discouraging a stay at a given hotel”. According to our data, in more than half of all their comments Dutch <?page no="66"?> 66 3 Analysing social web conversations and discourses from a contrastive point of view and French hotel customers include a direct hotel recommendation or advice against booking the hotel (52.3% and 51.3%). In the Spanish corpus, this is the case in only 29% of all comments� German hotel reviewers fall somewhere inbetween (41.3%). An attempt has already been made to explain this Spanish communicative behaviour by resorting to the high-context profile of Spanish culture (Copeland/ Griggs 1985: 107)� Table 5 summarises the empirical results of our multilingual corpus that have been presented and explained above� Text action Language German Dutch Spanish French describing the room and the hotel premises 72�5% - Inf. 13.6% - App. 58.9% 68�5% - Inf. 18.6% - App. 49.9% 89�2% - Inf. 21.3% - App. 67.9% 83�5% - Inf. 17.6% - App. 65.9% assessing the cleanliness 71�3% - Inf. 8% - App. 63.3% 68�6% - Inf. 11% - App. 57.6% 70�2% - Inf. 10.6% - App. 59.6% 67�6% - Inf. 9.6% - App. 58% stating or commenting on the hotel location 56�2% - Inf. 34% - App. 22.2% 67�9% - Inf. 39.3% - App. 28.6% 60�2% - Inf. 42.6% - App. 17.6% 59�2% - Inf. 30.3% - App. 28.9% assessing the price/ performance ratio 60�6% - Inf. 22.3% - App. 38.3% 61�5% - Inf. 21.6% - App. 39.9% 59�9% - Inf. 10.3% - App. 49.6% 56�9% - Inf. 9.3% - App. 47.6% commenting on the performance of hotel staff 67�2% - Inf. 2.3% - App. 64.9% 72�2% - Inf. 4.6% - App. 67.6% 65�2% - Inf. 4% - App. 61.2% 68% - Inf. 3% - App. 65% describing breakfast choices 63�5% - Inf. 9.6% - App. 53.9% 66�3% - Inf. 11.3% - App. 55% 35�6% - Inf. 12.3% - App. 23.3% 33% - Inf. 11% - App. 22% indicating parking availability or commenting on parking-related problems 28�6% - Inf. 17% - App. 11.6% 20�3% - Inf. 12% - App. 8.3% 47�2% - Inf. 24% - App. 23.2% 33�3% - Inf. 18% - App. 15.3% commenting on quietness and privacy 53�6% - Inf. 11.6% - App. 42% 54�3% - Inf. 12.3% - App. 42% 29�2% - Inf. 6.6% - App. 22.6% 40�2% - Inf. 11.3% - App. 28.9% <?page no="67"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 67 recommending or discouraging a stay at a given hotel 41�3% - App. 41.3% 52�3% - App. 52.3% 29% - App. 29% 51�3% - App. 51.3% expressing one’s gratitude 0�6% 3% 1�3% 4�3% congratulating the hotel 1% 0�6% 0�6% 1�6% making a direct request to the hotel 0�6% 1% 1% 2% Table 5� Text functions and text actions in customer comments on online travel and hotel booking agencies in German, Dutch, Spanish and French 3.1.2 Communicative functions and text actions in social networking services As noted above, comments posted on social networking sites such as Facebook fulfil different functions to hotel reviews on online booking agencies such as Booking.com. Whereas on Booking.com there is only one-way communication between clients and hotels, as is almost always the case on Expedia and TripAdvisor, posts on social networking services are characterised by their bi-directionality. For hotel clients, social networking services are not only a platform to be used for sharing their experiences, as they can also provide an easy and open tool for direct communication with the hotel management. From a hotel management’s point of view, a Facebook profile is a tool for interaction with former or would-be clients that is designed to improve branding and customer service (Chang at al. 2010: 59). As a result, there are two different types of postings: a first group includes postings drafted by hotel customers, with a second one comprising postings written by hotel staff. These considerations imply that the working method for analysing the communicative functions and text actions that has been applied to comments posted on online booking agencies needs to be revised and adapted� Hotel reviews posted on Booking.com and other such sites are predominantly informative and/ or appellative. However, for comments on Facebook, the contact function is as important as the informative and appellative ones. In fact, in a relatively <?page no="68"?> 68 3 Analysing social web conversations and discourses from a contrastive point of view high percentage of Facebook comments the contact function is the main communicative function of a customer’s comment, as in (48): (48) Agentur Stelzer Lieben dank für die Gastfreundschaft in Ihrem Haus. Wir haben am So. d. 23.2.14 eine Schulung für unsere Mitarbeiter bei Ihnen durchgeführt und haben einen tollen Raum und eine grossartige Verpflegung erhalten. Danke nochmals vom ganzen Team. Auf unserer Seite finden Sie auch einige Fotos darüber. Mövenpick Hotel Berlin Liebe Agentur Stelzer, wir möchten uns herzlich bei Ihnen für Ihre netten Worte bedanken. Wir freuen uns sehr über Ihr positives Feedback, dass Sie sich bei uns sehr wohlgefühlt haben freut uns natürlich umso mehr! Wir hoffen, Sie bald wieder in unserem Haus begrüßen zu dürfen, herzliche Grüße aus Berlin Ihr Mövenpick Hotel Berlin (EC D Facebook: Mövenpick Hotel, Berlin) On social networking services, the informative text function does not necessarily refer solely to the hotel’s premises and services. The main goal of a Facebook profile is not primarily to provide information, but to establish a communication channel with former and prospective clients and foster customer loyalty by strengthening ties between a hotel and its clients. In this respect, posting everything the targeted audience could find interesting is a proven strategy for achieving this goal. Thus, the intentional force of the informative function in social networking services like Facebook can be twofold: either the emitter may wish to inform the recipient about hotel-related issues such as services, special offers, etc., or some information is given about events that are essentially not related to the hotel. In the first case, the textual intention is informative by presenting information on one’s own behalf. In the second case, the informative function is performed by presenting external information. The French post (49) is a good example of an outweighing informative function by presenting information on one’s own behalf. The German post (50) announces a handball match that could be interesting for hotel customers (informative function by presenting external information): <?page no="69"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 69 (49) Shangri-La Hotel, Paris Bonne nouvelle! A l’occasion de la Journée de la Femme, La Bauhinia offrira samedi 8 mars au dîner, une coupe de champagne* Duval-Leroy - Femme de Champagne à toutes les femmes (...) (EC F Facebook: Shangri-La Hotel, Paris) (50) Hotel Berlin, Berlin Die Füchse Berlin treffen bei den Final Four auf MT Melsungen, wer unsere Jungs unterstützen möchte, hier gibt es alle Infos: (...) (EC D Facebook: Hotel Berlin, Berlin) The appellative function is also present in Facebook comments posted by the hotel management and even by hotel clients (see example 51), but it does not seem to be as prominent at it is in the case of hotel reviews posted on the webs of online travel agencies� Hence it will not be necessary to distinguish between different types of appellative sub-functions. (51) Roy Holfert Tolles Hotel mit super Zimmern und leckerem Frühstück! Einfach perfekt um einen kurzen Silvesterurlaub zu genießen! Danke! (EC D Facebook: Andel’s Hotel, Berlin) In our multilingual corpus, the informative, appellative and contact functions that make up the communicative structure of comments posted on social networking services, above all Facebook, are articulated around a relatively short list of text actions. The most significant text actions in terms of occurrences are the following ones (always illustrated by at least one example): - Asking for information about the hotel, its premises and its services: “¡Hola! Estoy valorando la posibilidad de pasar un par de noches en este céntrico hotel y realizar pensión completa, ya que en junio, pasé un par de noches en media pensión y quedé muy satisfecho, con su restaurante Amayra. ¿El día de Navidad, dan comidas? . Tengo unos días de vacaciones y quiero volver a Madrid y alojarme en este céntrico y agradable hotel” (EC E Facebook: Hotel Regina, Madrid). This text action is usually carried out by prospective (or former) clients, and involves mostly the contact function, although it can also include some neutral information, and hence be of an informative nature. - Giving some piece of information about the hotel, its premises and its services: “En el Hotel Europa puedes dormir en una doble superior como ésta con vistas a la Puerta del Sol por 119€” (EC E Facebook: Hotel Europa Madrid, Madrid)� Text actions such as this one can be informative or appellative depending on the writer’s intentional force. <?page no="70"?> 70 3 Analysing social web conversations and discourses from a contrastive point of view - Commenting on events that will take place or have taken place at the hotel� Like the last text action, this one can be either of an informative or appellative nature. It can be posted by the hotel management, by clients, or even by third parties. The following example announces the opening of a hotel’s bar. In this case the text action is clearly appellative: “Le Daniel’s OPENING (...) Tous les jeudis , notre petit bar à cocktails bien caché, intimiste vous invite dans une atmosphère feutrée et vintage, avec un bar à cocktails maison...Dress code chic, fun, nœuds-papillons, frous frous et dentelles bienvenus (...)” (EC F Facebook: Mon Hotel Paris, Paris). When the event has already taken place, the informative function usually prevails, as in the following example, where a third party informs about an event that was organised in an Amsterdam hotel: “Afgelopen vrijdag, Valentijnsdag, mochten we in samenwerking met Bad Manor 3 dames (plus een vriendin) een exclusieve akoestische showcase incl. overnachting in citizenM Hotel Amsterdam City aanbieden. Bekijk de aftermovie van deze mooie avond http: / / youtu.be/ KfaiFwWv1p” (EC NL Facebook: Citizen M Hotel, Amsterdam). - Commenting on news or events that will take place or have taken place outside the hotel� This text action may also be informative or appellative: “Unsere Kollegen Franziska und Christoph haben gestern am Pokertunier der Füchse Berlin teilgenommen und hatten eine Menge Spaß in der Spielbank Berlin. Vielen Dank für die Einladung. (EC D Facebook: Hotel Berlin, Berlin). - Assessing the hotel stay. The central objective of customers posting on Booking.com, Expedia and TripAdvisor is to assess the hotel stay; they therefore resort to different text actions. By contrast, writing a review of the hotel is not the main goal of Facebook comments on hotels. Nevertheless, the hotel is sometimes described and reviewed, albeit not in as detailed a way as on Booking.com or Expedia: “Heerlijk weekend doorgebracht bij Okura. Heerlijke ontspanning in de spa door de geweldige shiatsu behandeling en een verrukkelijke schoonheidsbehandeling. Ook nog genoten van een kookworkshop the taste of Christmas bij Okura. Aanrader! ” (EC NL Facebook: Hotel Okura, Amsterdam). This text action is usually appellative, although it can also contain some informative elements or be of an entirely informative nature� - Recommending or discouraging a stay at a given hotel. This text action, which is always appellative, does not seem to be as common as one might expect. It is very often carried out indirectly, as in: “In November komm ich wieder freu mich schon drauf (EC D Facebook: Hotel Mövenpick Berlin, Berlin). - Expressing one’s gratitude or congratulating the hotel: “Merci pour la nuit de rêve dans votre magnifique suite verte” (EC F Facebook: Hidden Hotel, Paris), “j’adore” and “superbe” (EC F Facebook: Kube Hotel, Paris). This text action belongs to a contact text function� <?page no="71"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 71 These text actions will help us to determine the communicative structure of our general quadrilingual corpus. The proposed working method is a slight variation of the one described above for Booking.com, Expedia and TripAdvisor hotel reviews� The following table chart will be used to describe the text functional and text actional structure of Facebook comments for each language. Informative text function - Σ-(%) - by presenting information on-one’s-own-behalf-(biob) - by presenting external information-(bei) Appellative text function --Σ-(%) Contact text function --Σ-(%) Text action Σ % Σ % Σ % asking for information about the hotel, its premises and its services giving some piece of information about the hotel, its premises and its services commenting on events that will take place or have taken place at the hotel commenting on news or events that will take place or have taken place outside the hotel assessing the hotel stay <?page no="72"?> 72 3 Analysing social web conversations and discourses from a contrastive point of view recommending or discouraging a stay at a given hotel expressing one’s gratitude or congratulating the hotel Total comments - Σ (100%) Table 6. Contrastive analysis of text functions and text actions - Facebook The general Spanish, German, Dutch and French corpus that will illustrate this working method and allow us to draw provisional conclusions about the communicative structure of social networking comments is composed of 400 Facebook comments - a hundred per language� Half of these comments are postings written by the hotel management and the reactions to these postings. Another fifty comments are postings composed by Facebook users interested in the hotel (“recent posts by others”) and the reactions to them� Since the comparative figure for calculating percentage rates and the total amount of occurrences of a given text action are in this case both 100, there will be no need to separate the cells labelled “%” and “Σ” in the table charts that sum up the results for each language� The evaluation of the Spanish Facebook corpus data throws up surprising results� The most common text action found in our corpus does not refer directly to the hotel and its services -as would have been expected-, but to events that are not primarily related to them. In fact, the text action “commenting on news or events that will take place or have taken place outside the hotel” is present in one third of all comments (33%), whereas events that will take place or have taken place at the hotel are only included in 24% of our corpus comments� The next two text actions by order of importance are “giving some piece of information about the hotel, its premises and its services” (20%) and “asking for information about the hotel, its premises and its services” (19%). According to the data extracted from the corpus, the text action “commenting on news or events that will take place or have taken place outside the hotel” is in all but two cases (2%) informative (31%)� The information is very often presented on one’s own behalf (23%): usually it is a third party, thus not necessarily a hotel client, who posts an advertisement on the Facebook profile of the hotel announcing an event they (or their company) will be holding or has held in the recent past (see the following example): <?page no="73"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 73 (52) ArteYmadrid AYm Visita guiada Museo del Prado con arteYmadrid. http: / / arteymadrid.com/ 2014/ 01/ 11/ museo-del-prado-obras-clave/ (GC E Facebook: Hotel Artrip Madrid, Madrid) Only in 8% of all comments does the hotel provide its Facebook followers with information about events that are unrelated to it, as in (53). It may be pertinent to mention that those comments are very often shared links: (53) Hotel Artrip Madrid shared a link Uno de los vídeos más “Happy” felices del mundo, hecho en Madrid Os dejamos un vídeo que esta siendo catalogado como el vídeo más feliz del Mundo y producido en Madrid. Abrazos y Feliz Viernes (GC E Facebook: Hotel Artrip Madrid, Madrid) Events that will take place or have taken place at the hotel are present in nearly one fourth of all comments. As is the case with the text action already discussed above, the informative function outweighs the appellative one: the informative setting is dominant in 21 out of the 24 comments in which this text action appears (the other three are of an appellative nature). Surprisingly, the percentage of comments posted by clients informing about hotel events or activities is higher (12%) than that of comments left by the hotel management (9%). Thus, the informative function by presenting external information (example 54) seems to be more frequent that the informative one by presenting information on one’s own behalf (55): (54) Alejandra Feldman Un domingo muy divertido! ! para niños y padres! ! ! El Hotel Silken Puerta América Madrid ofrece unos menús de buffet para niños y adultos! ! y además taller de Pizzas para niños! ! ! Hotel Silken Puerta América Madrid Gracias por compartir Alejandra Feldman! (GC E Facebook: Hotel Silken Puerta América, Madrid) (55) Hotel Silken Puerta América Madrid NUEVA CARTA INSPIRADA EN LOS PLATOS DE Martín Berasategui Hoy presentamos la renovada oferta gastronómica de Lágrimas Negras, de la mano del nuevo director gastronómico Raúl Cabrera y del nuevo chef Juan Carlos Delle Vedove.Se trata de (...) (GC E Facebook: Hotel Silken Puerta América, Madrid) Only 19% and 20% of the Spanish comments posted on Facebook hotel pages seem to seek and give information about the hotel. Asking for information about <?page no="74"?> 74 3 Analysing social web conversations and discourses from a contrastive point of view the hotel, its premises and its services is not always related to the informative function (this is the case in 11% of the comments, as in 56): the real intention of Facebook users asking questions about the hotel in 8% of the comments analysed is to contact the hotel management, as can be observed in 57: (56) Niceto Pacheco (...) ¿El día de Navidad, dan comidas? . Tengo unos días de vacaciones y quiero volver a Madrid y alojarme en este céntrico y agradable hotel. (GC E Facebook: Hotel Regina, Madrid) (57) Primera Escena ¿Hacéis bodas? Lo digo porque somos una empresa organizadora de bodas y nos interesa vuestro hotel. (GC E Facebook: Hotel Europa Madrid, Madrid) When giving some piece of information about the hotel, its premises and its services (20%), the informative function by presenting information on one’s own behalf is clearly the most dominant one (15%), as in the answer given to an enquiry made by a prospective hotel client in (58). However, in 5% of the comments, especially in those posted by satisfied customers, the appellative function prevails (59): (58) Claudia Rosa-López Buenas noches, me gustaría saber si hay lavandería en el hotel o en las cercanías del mismo. Gracias Hotel Europa Madrid Hola! Si disponemos de servicio de lavandería. Aquí puedes consultar otros servicios e instalaciones: http: / / www�hoteleuropa� eu/ madrid-centro/ Te esperamos! (GC E Facebook: Hotel Europa Madrid, Madrid) (59) Marta Belmonte Una fantástica habitación con maravillosas vistas desde el balcón... Y la cena, deliciosa. ¡Gracias! (GC E Facebook: Hotel Europa Madrid, Madrid) One of the most unexpected findings is the low rate of occurrence of the text action “assessing the hotel stay”. It is only documented in 12% of all comments, which are either appellative (7%, see example 60) or informative (5%, see 61): (60) Oscar Rosales Me encanta! (GC E Facebook: Hotel Único Madrid, Madrid) (61) Julian Madrid buen hotel (GC E Facebook: Hotel Silken Puerta América, Madrid) <?page no="75"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 75 This finding may be explained by the fact that some hotels have a separate posting area for hotel reviews on their Facebook page. The reviews posted there have not been included in the corpus� Expressing one’s gratitude or congratulating the hotel seems to be far more common on Facebook than on Booking.com and Expedia. Such a text action, which obviously implies the contact function, is present in 12% of all Spanish comments� The following post is a highly illustrative one: (62) Gloria Alcolea Santos Buenos días a todos/ as los/ as que formáis y sois parte de este entrañable Hotel Regina. Quería daros las gracias, y esta página me ha parecido la mejor propuesta para hacerlo. Gracias por vuestra paciencia, por vuestras atenciones, por vuestra generosidad y por vuestro saber estar y saber hacer. El nombre no hace ni a las personas ni a las cosas, somos las personas las que hacemos que las instituciones, organismos, empresas, industrias, comercios etc. Merezcan la pena y vosotros/ as hacéis que el Hotel Regina sea un referente y se pueda mantener desde 1918 en pie. Mi más sincero agradecimiento. Nos veremos pronto. Un abrazo para todos/ as. (GC E Facebook: Hotel Regina Madrid, Madrid) Leaving aside the posting area for reviews (which is not always present on Facebook hotel pages), communicatison on Facebook does not seem to encourage hotel clients to recommend or discourage a hotel stay� Only a marginal percentage (4%) of the Spanish comments analysed include such a text action� It may be concluded that Spanish hotel pages are seen by Facebook users as a platform for promoting unrelated hotel events� The detailed results obtained from the evaluation of the Spanish corpus are shown in the following table� <?page no="76"?> 76 3 Analysing social web conversations and discourses from a contrastive point of view Informative text function Σ-(%) - by presenting information on one’s own behalf-(biob) - by presenting external information-(bei) Appellative text function --Σ-(%) Contact text function --Σ-(%) Text action % and Σ % and Σ % and Σ asking for information about the hotel, its premises and its services 11 8 giving some piece of information about the hotel, its premises and its services - biob 15 - bei 0 5 commenting on events that will take place or have taken place at the hotel - biob 9 - bei 12 3 commenting on news or events that will take place or have taken place outside the hotel - biob 23 - bei 8 2 assessing the hotel stay 5 7 recommending or discouraging a stay at a given hotel 4 expressing one’s gratitude or congratulating the hotel 12 <?page no="77"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 77 Total comments - 100 (100%) Table 7. Text functions and text actions in Spanish Facebook comments Relatively similar results have been found for the German corpus. The most important text actions that seem to articulate German Facebook comments on hotels are, by far, “commenting on news or events that will take place or have taken place outside the hotel” (36%), “giving some piece of information about the hotel, its premises and its services” (32%) and “commenting on events that will take place or have taken place at the hotel” (29%)� Slightly more than one third of all comments (36%) focus on commenting on news or events that have no direct connection with the hotel and have taken place or are to take place elsewhere� Only 5% of all comments could be described as appellative, which means that this text action is overwhelmingly informative (31%). Although the hotel management posts information on external events in 11% of all the comments analysed, as in (64), the informative structure of the text action has to be ascribed mostly to other Facebook users that exploit the comment function to advertise their own activities (20%): they present information on their own behalf, as in (63). (63) MAAL - berufsbegleitender Masterstudiengang Ambient Assisted Living Jetzt für den kostenlosen Design Thinking Workshop an der HTW Berlin anmelden. Ziel des Workshops “Zukunftsprodukte” ist es Impulse für die nutzerorientierte Produktentwicklung im Bereich Wohnwelten zu entwickeln. Die Veranstaltung richtet sich an Entwickler, Anbieter und Unternehmer aus und für den Bereichen Hotellerie, Möbelherstellung und Tourismus http: / / maal.htw-berlin.de/ workshop/ anmeldung/ — at HTW- Berlin Campus Wilhelminenhof (GC D Facebook: Maritim Hotel Berlin, Berlin) (64) Hotel Berlin, Berlin shared a link. Heute startet die Leipziger Buchmesse, für alle Leseratten unter euch hat DER SPIEGEL die 20 besten Romane 2014 rausgesucht: (GC D Facebook: Hotel Berlin, Berlin) Information about the hotel is given in 32% of all comments. Two thirds of the posted comments (23%) with this text action are informative by presenting information on their own behalf (it is usually the hotel management that is responsible for proceeding in this way). A good illustrative example may be (65). However, in 9% of the entire corpus of comments, as in (66), an appellative undertone can be clearly detected� <?page no="78"?> 78 3 Analysing social web conversations and discourses from a contrastive point of view (65) Maritim Hotel Berlin Und auch morgen geht’s wieder rund in der “LE BAR” MARITIM BER- LIN! mit our very special Jerry Canizales & Latino DJ Fidel Zambrano! Cocktail des Monats: Latino Lover für 5.50€! (GC D Facebook: Maritim Hotel Berlin, Berlin) (66) Mövenpick Hotel Berlin PS: Dekoriert hat das Zimmer unsere Guest Relations-Beauftragte Sina Bartel. Ist es nicht fein geworden? ! (GC D Facebook: Mövenpick Hotel Berlin, Berlin) The third most common text action, “commenting on events that will take place or have tsaken place at the hotel” (29%), is also mainly informative. In 13% of all comments it is the hotel itself that provides information about its own events, as in (67). Only in 9% of the entire corpus of comments do clients or other Facebook users refer to events organised by the hotel from an informative point of view� The appellative function is not very common, but it can be observed in postings written by the hotel management (68), as well in a few reactions to comments on hotel events posted by customers (69)� (67) andel’s Hotel Berlin Packe, packe Tüte Die Vorbereitungen für die morgige Jubiläumsparty laufen auf Hochtouren� Jeder packt beim Packen mit an (���)� (GC D Facebook: andel’s Hotel Berlin, Berlin) (68) Mövenpick Hotel Berlin Tadaaa - unser erstklassiges Tatar ist wieder da! Kulinarisch steht der März im Zeichen unserer beliebten Klassiker, wie dem Kalbstatar aus sorgfältig ausgesuchtem Rindfleisch mit Olivenöl, Pfeffer und Zitrone oder dem Lachstatar aus köstlichem Lachs, welches von Crème fraîche, Kaviar und Kartoffelchips begleitet wird (siehe Foto). Bis bald im Hof zwei! (GC D Facebook: Mövenpick Hotel Berlin, Berlin) (69) Andel’s hotel Berlin Wir erheben die Gläser und stoßen an, auf unglaubliche 5.000 Fans! 6 March in andel’s Hotel Berlin. Und darum möchten wir uns bei euch bedanken: Für alle Gäste, die heute im Ristorante oscar’s oder im sky. café das Stichwort “High Five” nennen, gibt es ein Glas Prosecco aufs <?page no="79"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 79 Haus! Außerdem verlosen wir unter allen 5.000 Fans zwei Übernachtungen für zwei Personen (...) Sabrina Toprakhisar Super Medison II Wow, das ist ja mal eine super Idee. (GC D Facebook: andel’s Hotel Berlin, Berlin) The other text actions observed in the corpus are not nearly as important as the three preceding ones� The text action “assessing the hotel stay” is carried out in only 12% of all comments, and reflects a mostly appellative text function (see example 70). Asking for information about the hotel (9%), expressing one’s gratitude or congratulating the hotel (8%, see example 48) and recommending or discouraging a stay at a given hotel (7%, see example 71) are infrequent text actions in our corpus� (70) Jessica Vornkahl Hat mir sehr gut gefallen ���� (GC D Facebook: Hotel Berlin, Berlin) (71) Franz Daxauer Ein selten tolles Hotel - geniales Essen und superfreundliches Personal - nur zu empfehlen ! ! ! (GC D Facebook: Andel’s Hotel Berlin, Berlin) After describing the results obtained by the analysis of the Spanish corpus, the findings based on the German one are no longer surprising: the Facebook pages of German hotels seem to be a tool for sharing information and promoting all kinds of events, even those that are not related to the hotels. The following table sums up the German results� <?page no="80"?> 80 3 Analysing social web conversations and discourses from a contrastive point of view Informative text function Σ-(%) - by presenting information on one’s own behalf-(biob) - by presenting external information-(bei) Appellative text function --Σ-(%) Contact text function --Σ-(%) Text action % and Σ % and Σ % and Σ asking for information about the hotel, its premises and its services 4 5 giving some piece of information about the hotel, its premises and its services - biob 23 - bei 0 9 commenting on events that will take place or have taken place at the hotel - biob 13 - bei 9 7 commenting on news or events that will take place or have taken place outside the hotel - biob 20 - bei 11 5 assessing the hotel stay 3 9 recommending or discouraging a stay at a given hotel 7 <?page no="81"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 81 expressing one’s gratitude or congratulating the hotel 8 Total comments - 100 (100%) Table 8. Text functions and text actions in German Facebook comments The results obtained by the evaluation of the small Dutch corpus place the text actions “commenting on news or events that will take place or have taken place outside the hotel” and “commenting on events that will take place or have taken place at the hotel” as the most important ones for Dutch hotels and their clients. Comments on news and events that are not directly related to the hotel are present in 39% of all comments� Most of them (31%) have an informative function� Only 8% are appellative� The information provided is not necessarily posted by a third party seeking to promote an event of their own, as we can observe in (72), where the hotel management is referring to a television programme that may be of interest to its clients� (72) Hotel Okura Amsterdam Vanavond om 21.30 op RTL4 The Taste “Het Verre Oosten” De Japanse gastchef Masanori Tomikawa van sterrenrestaurant Yamazato uit Amsterdam beoordeelt blind de groepsopdracht waarin de chefs met hun team typisch Japanse ingrediënten moeten verwerken in hun gerechten. Een spannende strijd! (GC NL Facebook: Hotel Okura Amsterdam, Amsterdam) Facebook also appears to be used to comment on hotel-related events (35%). The cosmmunicative function is here again overwhelmingly an informative one (26%). A large number of comments with this text action are informative by presenting information on one’s own behalf: some are posted by the hotel management, as in (73), and some by a third party involved in the event that will be or has already been held at the hotel (see example 74)� (73) Mövenpick Hotel Amsterdam Hotelnacht 2014 in het Mövenpick Hotel Amsterdam Al een kleine impressie van de avond! — at Moevenpick Hotel Amsterdam City Center (EC NL Facebook, Mövenpick Hotel Amsterdam, Amsterdam) <?page no="82"?> 82 3 Analysing social web conversations and discourses from a contrastive point of view (74) 22tracks Afgelopen vrijdag, Valentijnsdag, mochten we in samenwerking met Bad Manor 3 dames (plus een vriendin) een exclusieve akoestische showcase incl. overnachting in citizenM Hotel Amsterdam City aanbieden. Bekijk de aftermovie van deze mooie avond http: / / youtu.be/ KfaiFwWv1p8 (GC NL Facebook: Citizen M Hotel Amsterdam, Amsterdam) Only 14% of the comments analysed include an assessment of the hotel stay� Most of these comments are appellative (9%), as in (75): (75) S/ B Styling Heerlijk uitzicht vanochtend bij de opnames’ Hotel Okura Amsterdam (GC NL Facebook: Hotel Okura Amsterdam, Amsterdam) Other text actions are not so frequent in the corpus. Information about the hotel is only given in 8% of all comments� The same percentage applies to the text action “recommending or discouraging a stay at a given hotel” (see example 76)� Surprisingly, very few comments provide any information about the hotel (6%) or ask for information (8%)� Thanking or congratulating the hotel is also rare (7%, see example 77). (76) Sanders Gert Vandaag gebeld door dhr. Rey, prima oplossing aangedragen door het hotel. Dik voor elkaar..... Amsterdam, soon we are back..... (GC NL Facebook: Hotel Victoria Amsterdam, Amsterdam) (77) Agnes Stikkelorum hoi leuke dag gehad en nog gefeliciteerd gr, (GC NL Facebook: Hotel Okura Amsterdam, Amsterdam) As a whole, Dutch hoteliers and hotel customers use Facebook to promote events that may or may not be related to the hotel. The findings concerning the Dutch (and French) corpus are shown in table 9, which sums up all the results obtained for the four languages� All but four text actions are irrelevant for the communicative structure of the French comments that form our general corpus. The most common text action, devoted to comment on events that will take place or have taken place at the hotel, is present in 39% of all comments. Although a slight majority of those comments (21%) are appellative, the informative function is also very much present (18%): in (78) the hotel management is advertising a soiree at its own hotel lounge (appellative function); in (79), Facebook users are being informed about a fashion shoot that took place the week before the comment was posted (informative function by presenting information on one´s own behalf)� <?page no="83"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 83 (78) Prêts pour ce soir? Nous sommes impatients de vous accueillir au #ShangriLaLounge! Visualisez les photos des plus belles soirées: https: / / www.facebook.com/ media/ set/ ? set=a.584044884977576.10737418 40.146928212022581&type=1 (GC F Facebook: Shangri-La hotel, Paris) (79) Mon Hotel Paris shared a link� Le joli shooting mode de la semaine dernière à Monhotel avec @the little world of fashion http: / / www.thelittleworldoffashion.fr/ 2014/ 03/ portraits-in-paris.html (GC F Facebook: Mon Hotel, Paris) Hotel clients and the hotel management are also very active on Facebook when providing other users with information about the hotel, its premises and its services. The figure for this text action is 30%. The appellative function is here again more common than the informative one (17% and 13%, respectively, see examples 80 and 81)� (80) KUBE Hotel - Paris changed their cover photo. KUBE Hotel - FR Paris 18ème +33 (0)1 42 05 2000 / events@kubehotel.com http: / / www.muranoresort.com/ — at KUBE Hotel - Paris (GC F Facebook: Kube hotel, Paris) (81) Les Pépites de Noisette Découvrez la recette du super cake au saumon du petit déjeuner du Hidden Hotel, juste ici! Un régal! http: / / bit.ly/ PmYEP4 (GC F Facebook: Hidden Hotel Paris, Paris) In just over one fifth of all French comments (21%) some kind of information about the hotel is requested by a third party. The informative function clearly prevails (11%, see example 82), but there are also a few examples with the appellative and even the contact functions (both 5%), as in (83): (82) Crotale Masson Bonjour il me faut une place. Kube (GC F Facebook: Kube hotel, Paris) (83) Nico Bens Bonjour, votre événement a été référencé dans notre page ARTS in Paris, qui avec entre 50.000 et 200.000 visites par mois permet de promouvoir <?page no="84"?> 84 3 Analysing social web conversations and discourses from a contrastive point of view un événement, un artiste. Nous vous invitons à vous y abonner et nous contacter pour toute recherche de communication (���) (GC F Facebook: Shangri-La hotel, Paris) French users do not seem to use Facebook extensively as a platform for commenting on news or events that will take place or have taken place outside the hotel. This text action is observed in only 16% of all comments, of which 11% are informative (84)� (84) Mon Hotel Paris shared a link� On fait quoi à Paris ce week-end? On va au Panthéon! ! ! Une étape de plus vous attend pour le projet «Au Panthéon! ». Venez vous faire photographier Du 5 au 29 mars, le camion photographique de JR vient gratuitement à votre rencontre... http: / / www�au-pantheon�fr/ fr/ (GC F Facebook: Mon Hotel hotel, Paris) From a purely contrastive point of view, some of the empirical results obtained for the four languages (see table 9) are very surprising: the communicative structure observed in hotel comments posted on Facebook by Western Europeans is not always similar. Contrary to expectations, users of the leading social networking service Facebook do not seem to use it to ask for information about the hotel and its services. In this regard, the figure is very low in the Dutch and German corpus (only 6% and 9%, respectively, of all comments ask in some way for information). Even when we consider that only 50% of the corpus postings analysed were posted by hotel clients or prospective ones (the other half were posted by the hotel management), the figure remains low. In comparison, Spanish and French users tend to see Facebook as a way of obtaining information about the hotel much more than their Germans or Dutch counterparts: 19% and 21% of all Spanish and French Facebook comments can be interpreted as queries for more information about the hotel facilities� Dutch and Spanish hotel managements seem to hold back from giving information that has not been requested by internet users: the text action “giving some piece of information about the hotel, its premises and its services” can be observed in 8% and 20%, respectively, of the Dutch and Spanish corpus of comments. The staff of the French and German hotels analysed who are responsible for managing the customer relationship over the internet tend to offer some kind of information on Facebook in most cases (the figures in the table -30% and 32% for the French and the Germans, respectivelyrefer to the entire corpus of comments, from which only 50% were posted by hotel staff), even when no information has been requested by the Facebook users commenting. Another <?page no="85"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 85 telling language and cultural difference in the commenting culture of Western European Facebook users concerns the text action “commenting on news or events that will take place or have taken place outside the hotel”� This text action, which is usually informative, should be understood as a means for turning the hotel’s Facebook profile into a cultural and leisure information platform for and by hotel clients, and hence it is part of the hotel’s customer loyalty strategy. German, Dutch and Spanish comments record a relatively high number of postings announcing activities that are not directly related to the hotel (36%, 39% and 33%, respectively). By contrast, posts containing this type of information are much less common in our French corpus (16%), fuelling speculation that commenting on events taking place outside the hotel is not welcomed by the management of French hotels. Good evidence of this assumption may be found in the fact that French comments record the highest percentage when addressing the text action “commenting on events that will take place or have taken place at the hotel” (39%, in comparison with 24%, 29% and 35% of Spanish, Dutch and German comments). It seems that only hotel-related events are allowed on French Facebook profiles. Contrary to what has been observed in the case of Booking.com, users do not usually assess their hotel stay on Facebook, except in the very few cases in which the Facebook profile explicitly allows the hotel to be assessed, but which have not been included in the corpus: only 14% of Dutch Facebook users, 12% of German and Spanish, and an exiguous 4% of French assess their hotel stay. Even less common are hotel recommendations. As for the Spanish corpus, which records a two-digit percentage, the text action “expressing one’s gratitude or congratulating the hotel” is also barely documented� <?page no="86"?> 86 3 Analysing social web conversations and discourses from a contrastive point of view Text action Language German Dutch Spanish French asking for information about the hotel, its premises and its services 9% - Inf. 4% - Con� 5% 6% - Inf. 4% - Con� 2% 19% - Inf. 11% - App. 8% 21% - Inf. 11% - App. 5% - Con� 5% giving some piece of information about the hotel, its premises and its services 32% - Inf. 23% - App. 9% 8% - Inf. 7% - App. 1% 20% - Inf. 15% - App. 5% 30% - Inf. 13% - App. 17% commenting on events that will take place or have taken place at the hotel 29% - Inf. 22% - App. 7% 35% - Inf. 26% - App. 9% 24% - Inf. 21% - App. 3% 39% - Inf. 18% - App. 21% commenting on news or events that will take place or have taken place outside the hotel 36% - Inf. 31% - App. 5% 39% - Inf. 31% - App. 8% 33% - Inf. 31% - App. 2% 16% - Inf. 11% - App. 5% assessing the hotel stay 12% - Inf. 3% - App. 9% 14% - Inf. 4% - App. 9% 12% - Inf. 5% - App. 7% 4% - Inf. 3% - App. 1% recommending or discouraging a stay at a given hotel 7% - App. 7% 8% - App. 8% 4% - App. 4% 2% - App. 2% expressing one’s gratitude or congratulating the hotel 8% - Con� 8% 7% - Con� 7% 12% - Con� 12% 1% - Con� 1% Table 9. Text functions and text actions in German, Dutch, Spanish and French Facebook comments <?page no="87"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 87 3.1.3 Communicative functions and text actions in video-sharing platforms and on Wiki discussion pages As already noted, customer comments on hotels are extremely rare on videosharing platforms such as YouTube and on Wiki discussion pages. Internet users do not seem to see these two social web applications as a valuable channel for conveying information about hotels� This is the reason comments are so very scarce in our exemplary corpus. As a logical consequence, this chapter will not draw any empirical conclusions about the occurrence of communicative functions and text actions in comments posted on YouTube and Wikipedia. Instead, we will make do with a very brief description of the informative and appellative text actions that have been encountered in the corpus� Despite the fact that most videos showing hotels on YouTube, which are usually posted by the hotel management or by travel agencies, are not commented on by users, there are a few audiovisuals that have indeed received comments from users� These comments are structured according to a few recurrent text actions: - Assessing the hotel stay: users tend to be relatively succinct when assessing the hotel stay on YouTube, as in the following postings (which are appellative): (85) Hotel Hessischer Hof Demiglacechef Sehr schönes Hotel. Kann ich nur bestätigen. Vor allem der Service ist immer wieder super! ! ! (GC D Youtube: Hotel Hessischer Hof) (86) Hotel Dómine 5* noe lia hace ahora 2 años estuve en ese hotel un par de noches, y es fantastico! ! , cenamos en la terraza y todo estaba buenísimo, y el servicio muy atento... (GC E Youtube: Hotel Dómine 5*) - Expressing one’s gratitude or congratulating the hotel: this text action is usually linked to another one in which the hotel stay is assessed, and it has a contact language function: (87) Andel’s Hotel Berlin, Germany welpen2006 Die sogenannten Fünf-Sterne-Kettenhotels können im Vergleich zum Andel’s einpacken (...) Das Video ist Ästhetik pur, ich dachte sowas gibt’s nur in Südostasien oder Dubai. Kompliment! (GC D Youtube: Andel’s Hotel Berlin, Germany) <?page no="88"?> 88 3 Analysing social web conversations and discourses from a contrastive point of view - Assessing the quality of the video posted: since most YouTube users have not stayed at that specific hotel, more often than not comments refer to the video itself and not to the hotel� This text action is usually appellative: (88) Kurhaus Binz - Rügen / Ostsee Chrfilmer Sehr gut gemachtes Video! Das kann ich hier mal sagen! Wobei es hier eher um Werbung geht! (GC D Youtube: Kurhaus Binz - Rügen / Ostsee) - Commenting on issues unrelated to the hotel: some comments have no apparent connection at all to the hotel, such as in (89), where a speech on the Spanish economic crisis and the European Union is delivered: (89) Hotel Melià Marbella Banús 4* Miguel Zeta variedades de clima por lo que si España se independizase de Europa. Perjudicaría a Europa pero no a España. Podemos incluso no comprar petróleo. Tenemos el suficiente para nos dejen en paz. Pobrecitos mis ángeles pobres y bellos que no tienen nada o casi nada. me refiero a Irlanda, Portugal y Grecia. No tienen recursos naturales y los tiburones se los han zampado (GC E Youtube: Hotel Melià Marbella Banús 4*) Customer comments on Wikipedia are even scarcer than on YouTube. In fact, very few hotels have an entry on the online encyclopaedia, and their discussion pages tend to be blank� The following are the most common text actions: - Suggesting the Wiki article is not relevant and/ or proposing its deletion: this text action is one of the most common among the very few Wikipedia comments on hotel discussion pages. It is usually motivated by the fact that the management of some hotels are using Wikipedia as a platform for advertising tourist services. In the following examples, Wikipedia editors are strict with the authors of the respective articles and either directly request the deletion of the article for being an advertisement (90) or ask for details that could confirm the relevance of the article before nominating the article for deletion (91, “LA” stands for “Löschantrag” or “deletion request”): (90) Hotel Berlin, Berlin Was hat denn diese Werbung im Wiki zu suchen? (nicht signierter Beitrag von 82.198.100.134 (Diskussion) 07: 57, 9. Mär. 2013 (CET)) (GC D Wikipedia: Hotel Berlin, Berlin) <?page no="89"?> 3.1 Communicative macrostructure: text functions and text actions in consumer comments 89 (91) Sofitel Berlin Kurfürstendamm Moin, bei dem Artikel erscheint die Relevanz sehr fraglich gemäß WP: RK für Bauwerke (7.2): • Historisch: nein • Wahrzeichen: nein • Kulturdenkmal: nein • Wichtiger Beitrag zur Architekturentwicklung: AD keine entsprechende Rezeptionsquelle Daher sollte relevante Rezeption im Artikel dargelegt werden, sonst scheint LA angezeigt. Gruß, --mpkmpk (talk, Beiträge) 17: 15, 5. Jul. 2012 (CEST) (GC D Wikipedia: Sofitel Berlin Kurfürstendamm) - Requesting the deletion of specific text sections or files of an article: (92) Hilton Amsterdam Hotel Eén of meerdere afbeeldingen die gebruikt worden op deze pagina of overlegpagina, zijn genomineerd voor verwijdering. Het gaat om Afbeelding: Hiltonhotel.jpg, zie Wikipedia: Te verwijderen afbeeldingen/ Toegevoegd 20071224. --E85Bot 25 dec 2007 02: 13 (CET) (GC NL Wikipedia: Hilton Amsterdam Hotel) - pointing out the necessity of creating a desambiguation page because there are more than one hotel with the same or a similar name and therefore there could be a conflict regaring the article title: (93) Hotel Palace Cuidado, también hay un Hotel Palace, por ejemplo, en Barcelona, véase su página web: www.hotelpalacebarcelona.com (GC EWikipedia: Hotel Palace) - Expressing that a certain text or file is explicitly released for Wikipedia: this text action is carried out by the hotel management, which is usually the copyright holder of the content it releases for inclusion in the edition of the article: (94) Estrel Hierbei handelt es sich um einen offiziellen Pressetext des Estrel Berlin, der allen Internet-Usern auf der Unternehmens-Homepage frei zur Verfügung steht. (GC D Wikipedia: Estrel) As can be inferred from these text actions, Wikipedia is not a valid channel for promoting a hotel, and nearly all the comments on hotels that are featured in this online encyclopaedia refer to the edition of the articles� <?page no="90"?> 90 3 Analysing social web conversations and discourses from a contrastive point of view 3.2 Text-grammatical structures in consumer comments Text-grammatical structures are at the very core of text linguistics. Any serious analysis of a text genre must include an in-depth linguistic study of the mostly grammatical phenomena that interconnect the different sentences and propositions composing a text. It is no surprise, therefore, that text linguistics usually refers to the fact that “sentences or utterances are linked together” as “the most salient phenomenon of discourse” (Renkema 2004: 103). In this respect, Brinker (1992: 21) defines the structure of a text as “Gefüge von Relationen, die zwischen den Sätzen bzw. den Propositionen als den unmittelbaren Strukturelementen des Textes bestehen und die den inneren Zusammenhang, die Kohärenz des Textes bewirken”. Although there is no commonly accepted opinion as to what can be specifically described as text-grammatical structures (Gansel/ Jürgens 2008: 55), the most influential theoreticians from Halliday/ Hasan (1976: 2ff), De Beaugrande/ Dressler (1981: 3ff) or Brown/ Yule (1983: 190) through to Vater (2001: 52-54) or Brinker (2005: 21-22) agree that out of the seven standards of textuality proposed by De Beaugrande/ Dressler (1981: 3) the first two, “cohesion” and “coherence”, are text-oriented, and hence focus on the text either as a product of grammatical dependencies on the surface structure or underlying it� Since the goal here is simply to provide a working linguistic instrumentarium for analysing a concrete type of microtext, no attempt will be made to challenge the generally accepted view that cohesion and coherence are the text-grammatical structures to be taken into account when examining texts� This chapter will therefore determine and explore the text-grammatical elements that create grammatical (and lexical) cohesion and coherence in online customer conversations, and are relevant for a bior multilingual contrastive analysis� This comparative study of customer comments will be conducted from an eminently practical point of view. Accordingly, our text-grammatical approach will be eclectic, and will not adhere to a particular theory or conceptual and theoretical framework. Thus, the working method in this chapter could be best defined as an adjustment of already existing text linguistic descriptive devices to our textual and contrastive needs. In this regard, some of the following pages will consider the pertinence of the inclusion of the already very well described cohesive and coherence devices in our contrastive model. Due to the fact that there is no unanimity concerning the question of how to analyse text and discourse configuration, this may be the only possible approach for such a practical undertaking as ours� Our contrastive approach will take into account the aforementioned textuality standards of cohesion and coherence, as well as the also very important <?page no="91"?> 3.2 Text-grammatical structures in consumer comments 91 phenomenon of deixis. Deixis is also cohesive in nature, but as an exophoric category it needs an extralinguistic interpretation to be decoded properly� Before we discuss in detail the cohesive, coherent and deictic devices that should be examined when conducting contrastive research of customer comments, it is essential to define or re-define the concepts of cohesion and coherence. De Beaugrande/ Dressler’s terminological and conceptual distinction between cohesion and coherence (1981: 3) is not universally accepted. In fact, for many authors it is a “nicht unumstrittene Differenzierung” (Storrer 2004a: 16), which may generate considerable terminological problems (Rickheit/ Schade 2000: 275). De Beaugrande/ Dressler (1981: 3) understand cohesion as “the ways in which the components of the surface text, i.e. the actual words we hear or see, are mutually connected within a sentence”. For them cohesion “rests upon grammatical dependencies”. On the other hand, coherence concerns “the ways in which the components of the textual world, i.e. the configuration of concepts and relations which underlie the surface text, are mutually accessible and relevant” (De Beaugrande/ Dressler 1981: 4). This rather cumbersome formulation of coherence has been understood differently by different authors, and since it has become polysemous in text linguistic research (Esser 2009: 140), the use of the term is troublesome for some authors (among them, for example, Schnotz 1994 and Storrer 2004b: 3), who prefer to merge the concepts of coherence and cohesion or completely refrain from using one or both concepts� The wide reception of Halliday/ Hasan’s standard work on cohesion has helped to create even more confusion. These authors, as well as Van Dijk (1977: 126), seem to use both terms as stylistic variants. Nevertheless, they argue that cohesion is at the basis of coherence (Halliday/ Hasan 1976: 9), and that would mean that they should be considered different concepts. This ambiguous approach has been severely criticised by other linguists, first and foremost by Carrell (1982: 479ff), and in the 1980s it triggered a major debate about the nature of the two concepts (see Fulcher 1989 for an overview). Although “cohesion” and “coherence” remain disputed terms, we will resort to them because they allow us to contrast not only straightforward grammatical and semantic characteristics on the text surface of our multilingual corpus, but also non-overt culture-related relations that link the different textual units of meaning to each other. As these relations might differ due to the cultural component inherent to each one of us when writing hotel reviews or assessing any other service or product, it could prove useful to uphold the initial distinction made by De Beaugrande/ Dressler. However, the main difference between cohesion and coherence will be redefined following Tanskanen (2006: 7): Cohesion will be understood as referring “to the grammatical and lexical elements on the surface of a text which can form connections between parts of the text”, <?page no="92"?> 92 3 Analysing social web conversations and discourses from a contrastive point of view whereas coherence “resides not in the text, but is rather the outcome of a dialogue between the text and its listener or reader”. Plainly speaking, it is all about explicitness and implicitness: “cohesion” could be described as explicit coherence, which could be grammatical and lexical, while the term “coherence” could be used solely for implicit coherence, and would hence be limited to inferred coherence� In the following, the cohesion, coherence and deictic devices of relevance to a contrastive study on online comments on products and services will be presented and illustrated with examples from our exemplary corpus� The following discourse connective devices will be examined: - Reference (cohesion by reference and substitution) - Reiteration and collocation (lexical cohesion) - Syntactic recurrence and partial morphological recurrence (morphosyntactic cohesion) - Ellipsis (cohesion by ellipsis) - Conjunctive adjunction (cohesion by conjunction) - Coherence - Deixis Cohesion by reference and substitution One of the most widely used cohesion devices in customer comments is in fact the reference. As will be explained later, we will not distinguish formally between reference and substitution, referring to the latter as part of the reference device. An analysis of how two textual items are linked by reference in different languages should bear in mind that this task has a cross-cultural perspective� Textual conventions are always the result of cultural conventions, and as such they are based on the preferences that native speakers have commonly developed when writing and reading a certain text genre� So it is to be expected that cohesion by reference (and substitution) will be established in different ways in online comments in Spanish, French, German and Dutch. Our approach to uncovering these differences is again based partly on Halliday/ Hasan (1976) and Brinker (1992: 26). Reference is hence defined as a textual cohesive mechanism that links two or more specific items within a text (or a part of a text or a discourse) by grammatical means. Nevertheless, contrary to what Halliday/ Hasan (1976: 89) have proposed, this definition implies that there must a direct relationship between linguistic items within the text, which is why substitution will be considered part of the reference device: reference is not only interpreted here as a mere “relation between meanings”, as Halliday/ Hasan, <?page no="93"?> 3.2 Text-grammatical structures in consumer comments 93 Renkema (2004: 103) or Esser (2009: 41) do, but also as a relationship between grammatical forms. Although the differentiation between the classical cohesion devices of reference and substitution has not been respected here according to the principle that “a strict classification of cohesive relations is not possible” (Vujević 2012: 407), the actual analysis that will be outlined in the following includes some of the substitution categories identified by Halliday/ Hasan (the nominal and the verbal substitution)� Reference is usually established by a pro-form, that is, by “expressions of identical or different morphological form designating the same object and replacing each other under certain conditions” (Rieser 1978: 7). The most common way of instantiating a pro-form is using a pronoun. However, as pro-verbs and pro-adjectives are also pro-forms, and could also appear in a contrastive study, they should also be considered when examining this cohesive device� The use of pronouns as a means of structuring texts is usually anaphoric: they refer back to a linguistic item that has already been introduced in the text� A quick glance at out examplary corpus will show that, in fact, most pronouns establishing a reference are back-referential. However, there are some examples where the pronominal reference is cataphoric, as in (95). In this Facebook example, the subject pronoun “sie” refers to the nominal phrase “die #Oberbaumbrücke”, which is situated after the forward-referential pronoun: (95) Hotel Berlin, Berlin Sie zählt zu den bekanntesten Brücken in Berlin: die #Oberbaumbrücke. Heute ist Sie die ganzen Tag gesperrt um die 12. Open Air Gallery 2014 Oberbaumbrücke zu beherbergen. Ein Besuch lohnt sich in jedem Fall. (EC D Facebook: Hotel Berlin, Berlin) Since the referential cohesion established by anaphoric and cataphoric pronouns is not a different one and pronominal cataphors are not very common, at least not in the Western European languages we are dealing with, anaphora and cataphora will not be differentiated in our contrastive analysis. Personal pronouns are, of course, widely used when referring back (or forward). Some of them are in the nominative, and hence the subject of the subsequent sentence, as in (96), where the Dutch first-person plural pronoun “we” refers back to “voor ons, fietsers”. Personal pronouns in an oblique form are also readily found in the corpus. Such is the case of the French and Spanish third-person singular accusative pronouns “le” and “lo” (examples 97 and 98)� The French pronoun links the adversative sentence “mais faut le chercher” with the subject of the previous one, “le bouquet Canal+ est présent”. In the Spanish example “lo” refers to the antecedent “ete hote” (sic): <?page no="94"?> 94 3 Analysing social web conversations and discourses from a contrastive point of view (96) Heerlijk rustig gelegen en voor ons, fietsers, lekker dicht bij het centrum van Utrecht. We vonden de aankleding van de kamer, de gangen, hallen en eetzaal echt goed en plezierig. (EC NL Booking: Elmar, Ouder stel, Wageningen, Nederland, 14 april 2012) (97) Les programmes/ manque France 2, France 3, m6, sur le réseaux hertzien, et plusieurs chaines sur la tnt. Le bouquet Canal + est présent, mais faut le chercher. (EC F Booking: Pascal, Groupe d’amis, Dijon, France.10 juin 2012) (98) Muy decepcionado con ete hote. No lo recomiendo para nada. Hay otros más centricos y coges un taxi y en el aeropuerto en 20 minutos. (EC E Expedia: Un viajero verificado, Publicado el 5 de enero de 2012 por un cliente verificado de Hotels.com) 5 Since online customer comments and reviews are personal experiences and also imply a hotel as the object of conversation, most personal pronouns are expected to be either firstor third-person. Possessive pronouns or adjectives can also be observed throughout our exemplary corpus to refer to previous textual elements, while linking this reference to a particular grammatical person, usually the first or third. Possessive adjectives or determiners (as in 99 “leur présence”, or in 100 “por mi habitación”) are no doubt more frequent than possessive pronouns (example 6: “la mía”). (99) Bravo aux propriétaires pour ce lieu et leur présence toute en discrétion. (EC F Booking: Laura, couple jeune, Nice, France, 17 avril 2012). (100) Otros amigos reservaron el mismo fin de semana que yo y por fortuna les asignaron una habitación bastante mas amplia que la mía así que no juzgaré a todo el hotel por mi habitación pero es injusto que esa habitación tenga el mismo precio que las otras o al menos, que no te lo adviertan. (EC E Booking: Luis, Grupo de amigos, Pozuelo de Alarcón - Madrid, España, 22 de agosto de 2011) Demonstrative pronouns and adjectives act as textual cohesive devices when they refer to identifiable grammatical elements mentioned earlier in the text. In all four languages being examined in this study, the demonstrative reference, which can be charaterised as a form of verbal pointing and is anaphoric, identifies the referent on a proximity scale with at least one distal and one proximal demonstrative. Therefore, demonstratives can also be seen as a realization of 5 As the reader may have noted, misspelings are not marked in any way in most examples. A comment is quoted exactly as it appears in the source. <?page no="95"?> 3.2 Text-grammatical structures in consumer comments 95 place deixis. Here, however, demonstrative determiners and independent demonstratives will be treated as primarily cohesive ties that refer to a nominal or verbal phrase that has already been introduced. Such is the case, for example, in (101) and (102): The demonstrative pronouns “diese” and “deze” refer to previous nominal phrases. A good example of a demonstrative adjective extracted from our corpus may be (103), where “este” on “regresaré a este Hotel” (sic) establishes an anaphoric relationship with the aforementioned nominal phrase “este hotel” 6 . It is noteworthy that German hotel reviewers may also resort to a series of colloquial demonstrative pronouns (or adjectives), such as “der/ die/ das”. In the last German example, the colloquial demonstrative pronoun “die” in “die ist essentiell” refers back to the nominal phrase “die Beleuchtung im Bad”: (101) Schlafen kann man nur bei geschlossenen Fenstern und mit Ohrenstöpseln. Diese lagen auf dem Nachttisch schon bereit. (EC D Expedia: Brigitte, Mainz, Gesendet am 24. Juli 2011) (102) Badkamers mogen moderner, als deze gemoderniseerd zijn dan is de prijs kwaliteit verhouding optimaal (EC Booking NL: Jef, Ouder stel, Cadier en Keer, Nederland, 1 mei 2012). (103) Este hotel me ha resultado de los mejores que he estado en relación calidad/ precio (...). Sin duda si vuelvo a Salamanca (que siempre vuelvo) regresaré a este Hotel. (EC E TripAdvisor: Beatriz_Algeciras, Algeciras, opinión escrita el 4 marzo de 2013) (104) Es gibt nur eine Kleinigkeit, die aber Frauen und auch Männern das gute Aussehen schwer macht: das ist die Beleuchtung im Bad. Die ist essentiell - nicht nur, um sich beim Rasieren nicht zu schneiden oder den Augenbraunstift präzise zu setzen. (EC D Booking: Heike, Alleinreisender, Konstanz, Deutschland, 9. Juli 2012) Pronominal adverbs are commonly used in German and Dutch to create cohesion in texts and discourses by condensing information (Böttger/ Probst 2001: 11 ff.). They belong to the relational adverbs (Krein-Kühle 2002: 48), and are usually inserted after the nominal phrase (or the entire sentence) they refer to (and are therefore anaphoric). There are no pronominal adverbs as such in the Romance languages: the Spanish and French equivalents would be a demonstrative (or personal) pronoun. A good example extracted from our corpus would be the Dutch sandwich constructions “waar...van” and “er...in” in (105): 6 Please note that this first “este” cannot be considered a cohesive tie: it has no text structuring function� <?page no="96"?> 96 3 Analysing social web conversations and discourses from a contrastive point of view (105) De kamer straalt comfort en luxe uit en wij hadden het gevoel om op een geheim plekje te slapen midden in de drukke stad. Waar je overigens bijna niets (wij niet) van merkt ondanks dat je er midden in zit. (EC NL Booking: Maaike, Vriendengroep, Rosmalen, Nederland, 14 september 2011). Pro-verbs and pro-adjectives are much less used cohesive devices (Willkop 2001: 315). They also have a clear reference function within the text. However, as already noted, pro-verbs and pro-adjectives can also be described as belonging to an independent cohesive device called substitution. According to De Beaugrande (1980: 147), pro-verbs can be employed “for reusing event-based knowledge”. They thus refer to a previous verbal phrase, as can be observed in (106), where the imperative form “tut” is both substituting and referring to “in diesem Hotel buchen” expressed by the relative clause without an antecedent “wer in diesem Hotel schon gebucht hat”: (106) Wer in diesem Hotel schon gebucht hat, den kann ich nur mein Beileid wünschen, allen anderen will ich sagen: “Tut es nicht! ” (EC D TripAdvisor: TheBinzi, Amberg, Deutschland, Bewertet am 20. Juli 2009). Pro-adjectives are forms that substitute adjectives or adjective phrases. Proadjectives very rarely fall within the scope of text grammar studies (Baker 2003: 129), possibly because it is unusual to refer to an adjective with a pro-adjective in both natural and written discourse. This is, however, the case in (107): the pro-adjective “solch” in “in solchen Betten” is acting as a cohesive device substituting “provesorisch repariert” (sic), which as the past participle in a passive structure could be reinterpreted as a participial adjective or “Partizipialadjektive” (“das provisorisch reparierte Bett”): (107) Er unterstellte uns wir hätten das Bett demoliert. Maßlose Frechheit! ! ! Am nächsten Tag wurde das Bett provesorisch repariert, es war eine unruhige Nacht! Darf man sich in solchen Betten einmal umdrehen! ! Also Vorsicht! ! ! (EC D Tripadvisor: 165dada, Schwarzenbach, Deutschland, Bewertet am 25. November 2012) The proposed text-grammatical analysis of the reference devices that create cohesion in customer comments include the following parameters (table 10): <?page no="97"?> 3.2 Text-grammatical structures in consumer comments 97 Using a pronoun Personal pronouns - Person and number: First person: singular/ plural Second person: singular/ plural Third person: singular/ plural - Case: Nominative form Oblique form Possessive pronouns and adjectives - Possessive pronouns: First person: singular/ plural Second person: singular/ plural Third person: singular/ plural - Possessive adjectives First person: singular/ plural Second person: singular/ plural Third person: singular/ plural Demonstrative pronouns and adjectives - Demonstrative pronouns Distal Proximal Colloquial - Demonstrative adjectives Distal Proximal Colloquial Pronominal adverbs Using a pro-verb Pro-verbs “tun”, “doen”, “hacer”, “faire”, etc. Using a proadjective Pro-adjectives “solch”, “tal”, etc. Table 10� Cohesion by reference and substitution Lexical cohesion Lexical cohesion is another potentially powerful connective device that may reveal important contrastive differences between languages. It is primarily based on the relationship between lexical items and no longer on grammatical structures, as is the case with the devices of reference and substitution. The concept of lexical cohesion as used in modern text linguistics was developed and defined <?page no="98"?> 98 3 Analysing social web conversations and discourses from a contrastive point of view by Halliday/ Hasan (1976: 274) as “the cohesive effect achieved by the selection of vocabulary”� This selection of vocabulary is mostly carried out “diskursfeldspezifisch” (Caro Cedillo 2004: 137). This means that the use of some sort of lexical ties depends on the text genre and text type. In this regard, lexical cohesion can be established either by reiteration or by collocation. Some authors, such as Tanskanen (2006: 46ff), refer to both terms as categories of lexical cohesion. By contrast, other authors regard reiteration and collocation as syntagmatic and paradigmatic relations (e.g., Steinbach 2007: 182 and Busse (2009: 102), but otherwise respect the twofold distinction made by Halliday/ Hasan� Lexical cohesion by reiteration refers to the systematic use of content words linking lexical items in subsequent textual segments. Halliday/ Hasan (1976: 318-319) define reiteration as “the repetition of a lexical item, or the occurrence of a synonym of some kind, in the context of reference; that is, where the two occurrences have the same referent”. Thus, it may also involve “the repetition of a sense or part of it” (Martínez-Cabeza 2002: 91). Lexical semantics sub-divides reiteration into different types of lexical relationships (e.g., Saeed 2009: 63ff). However, we will follow Renkema’s classification, which is specifically intended for text and discourse analysis (Renkema 2004: 105). The cohesive lexical relations established by reiteration that should be considered for the empirical analysis of customer comments are, therefore, the following ones: lexical recurrence, synonymy, hyponymy/ hyperonymy, meronymy and antonymy. Lexical recurrence is simply the actual repetition of expressions, that is, the straightforward repetition of elements or patterns. Although Camacho Adarve (2009: 28) is basically right when she observes that lexical recurrence is pragmatically “imposible desde el punto de vista del discurso, porque nunca se dice exactamente lo mismo que hemos dicho o que nos han dicho la segunda o sucesivas veces” and also because “se convierte en un bloque o unidad pragmática diferente”, the fact that word repetitions usually involve reference may justify the inclusion of this cohesive device in our textual contrastive instrumentarium� A good example is the following Dutch hotel review extracted from out Booking.com corpus; the word “hotel” is repeated here five times: (108) Mijn verblijf in de stone hotel voor 3 dagen. Ik heb bijna alle hotels bezocht hier in Utrecht, Rotterdam en Amsterdam. Ik hou van comfort en luxe! Ik zag niet het verschil van een Hotel en een Hostel. Het was meer een Hostel. Maar ik kan niet geloven dat ze zulke hoge prijzen vragen voor een hotel dat niet voldoet aan wat je mag verwachten van een hotel. (EC Booking NL: Shureska, Individuele reiziger, Rotterdam, Nederland, 18 juni 2012). <?page no="99"?> 3.2 Text-grammatical structures in consumer comments 99 Lexical relations of synonymy also appear to be very common in customer comments. Although it may affect nouns, synonymic relations between adjectives are far more frequent. This is the case in (109) between “sehr gut” and “ausgezeichnet”, as well as in (110) between “fantantisch” and “schitterend”: (109) Das Frühstück, das man in einem unglaublich schönen Ambiente zu sich nimmt, ist sehr gut! Der Kaffee ausgezeichnet! ! ! (EC D TripAdvisor: Anke4711, Berlin, Deutschland, Bewertet am 9. August 2012) (110) Dit appartement is een fantastische beleving op een schitterende locatie midden in de binnenstad van Utrecht. Echt een aanrader! (EC NL Booking: Eric, Jong stel, Berkel-Enschot, Nederland, 11 juni 2012) Hyponymy and hyperonymy are closely related lexical relations that contribute to structure customer comments in a very specific and hierarchical way. Since it includes “the meaning of a more general word” (Saeed 2009: 69), hyponymy may be defined as a relation of inclusion. The most general, and hence superordinate, term that includes several subordinated lexical items is called a hypernym� Hyponym and hypernym relations are readily found in our exemplary multilingual corpus. This taxonomic relation should thus be thoroughly analysed. In (111) “comestibles” is, for instance, the hypernym of its subordinate terms “conservas”, “aceites”, “dulces”, “quesos”, etc.: (111) Jronia k Jronia Esta tienda de comestibles griegos lleva en su nombre un guiño a un famoso anuncio televisivo de yogurt (griego, claro). En el (sic) conservas, aceites, dulces, quesos, patés, salsas, mieles, mermeladas, galletas y demás delicatessen del país helénico. (EC E Facebook E: Hotel Único Madrid, Madrid) Meronymy links different lexical items to each other and to a whole. Meronymic relations thus involve words that refer to parts of a whole. By way of example, “salle de bain” is in (112) the whole, of which “douche”, “toilettes” and “chasse d’eau” are part: (112) La chambre 33: odeur nauséabonde dans la salle de bain. Pas de pression d’eau à la douche et dans les toilettes, d’où le besoin de tirer 3 fois la chasse d’eau. (EC F Booking: Pascal, Groupe d’amis, Dijon, France, 10 juin 2012) Antonymy has the same structuring function as synonymy, but -at least in online customer commentsantonyms are not expected to be that common. Antonymic (as well as synonymic) relations can be established between the same or <?page no="100"?> 100 3 Analysing social web conversations and discourses from a contrastive point of view different parts of speech: in the Spanish example (113), the adjectives “pequeña” and “grande” are in a classical antonymic relation; the German comment (114), in which the urban Berlin dialect is simulated, is partially structured lexically by two antonyms, a noun (“Hitze”) and an adjective (“kühl”): (113) (...) pero nuestra habitación muy pequeña, y claustrofóbica a más no poder, no es lógico que sea del mismo precio (alto) que otras bastante más grande y espaciosas. (EC E Booking: Juan, Pareja mayor, Madrid, España, 22 de marzo de 2012) (114) Maritim Hotel Berlin puh, watt ne Hitze...heute solln wa uns wohl so richtich einfühlen in unsere Jungs uffm Platz, oda wie? (...) Ick habe jedenfalls een idealet Plätzchen jefunden hier in meem Zimma im Maritim, kannik sehr empfehln, hier is dit kühl und Jetränke sind ooch gleich zur Hand. (EC D Facebook: Maritim Hotel Berlin, Berlin) The other category of lexical cohesion is collocation� Collocation can be best described according to Renkema (2004: 105) as “the relationship between words on the basis of the fact that these can often occur in the same surroundings”� So it is a type of lexical cohesion “achieved by means of two lexical items that tend to co-occur” (Martínez-Cabeza 2002: 93). In semantics, this co-occurrence has lead to a slightly different use of the term “collocation”, denoting partly or fully fixed expressions. This may be the reason why Halliday/ Hasan (1976: 284) regard collocation as “the most problematic part of lexical cohesion”� The term will be used here as suggested by Halliday/ Hasan and Renkema. As a way of illustration, in (115) “buena compañia” (sic) usually implies “amigos”, and so they tend to co-occur. The same is true of “familiär” in relation to “nett”, “freundlich” and “gemütlich” in (116): (115) Ignacio Isla Moreno En buena compañia (sic) siempre con amigos en el Hotel Silken Puerta América Madrid (...) (EC E Facebook: Hotel Silken Puerta América, Madrid) (116) Das hotel (sic) wird sehr familiär geführt, die besitzer (sic) sehr nett und freundlich. Das hotel (sic) ist sehr gemütlich, dennoch modern und sauber. (EC D Booking: Kai, Alleinreisender, Krefeld, Deutschland, 19. März 2012)� The contrastive analysis of the lexical cohesion in customer comments presented in the previous pages includes the cohesive devices mentioned in table 11� <?page no="101"?> 3.2 Text-grammatical structures in consumer comments 101 Reiteration lexical recurrence synonymy hyponymy and hyperonymy meronymy antonymy Collocation Table 11. Lexical cohesion Morphosyntactic cohesion The term “morphosyntactic cohesion” is not commonly used in text linguistics to refer to morphological and syntactic cohesion devices� The devices referred to in the following lines are usually treated within lexical cohesion, or simply not taken into consideration in theoretical studies: for instance, Halliday/ Hasan (1976: 10), to quote one of the most important works on cohesion, explicitly exclude syntactic cohesion from their approach on text linguistics 7 . However, as syntactic and morphological cohesion is present in texts, and is clearly part of syntactic or morphological phenomena, the syntactic recurrence and the partial (morphological) recurrence will be described here separately as “morphosyntactic cohesion devices”� One of the simplest and most effective means of achieving cohesion in small texts, as in brief customer comments for example, is to use syntactic recurrence (Kallmeyer et al. 1974: 147). De Beaugrande/ Dressler (1981: 49) understand recurrence to be the “straightforward repetition of elements or patterns”� The most common device involving recurrent syntactic patterns is syntactic parallelism, which could be described as the repetition of structural elements in a text in order to produce cohesion� Parallelism creates cohesion by relating one clause to another. As a general rule, those clauses may be either nominal phrases, as in (117), or verbal phrases, as in the German example (118): (117) Accueil très aimable! Un oasis au milieu des immeubles, une belle surprise. Petite piscine sympa pour les enfants, le jacuzzi agrèable pour les parents. (EC F Booking: Mikko, Famille avec adolescents, Céret, France, 5 juillet 2012) 7 However, in a later study these authors included syntactic cohesion as a new sub-component of structural cohesion (Halliday/ Hasan 1989)� <?page no="102"?> 102 3 Analysing social web conversations and discourses from a contrastive point of view (118) Die Mitarbeiter waren sehr freundlich und hilfsbereit. Das Frühstücksbuffet war sensationell. Die Sauberkeit der Zimmer und Bäder waren vorbildlich. (EC D Booking: Jens, Gruppe, Alsfeld, Deutschland, 6. April 2012) Another means of creating syntactic recurrence is chiasmus, that is, the reversal of syntactic structures� Chiasmus is most certainly not a common feature of online customer comments, possibly because it is a rhetorical figure that appears in elaborate speech. In fact, no chiasmic structures were found in our exemplary corpus� Morphological recurrence can also help to structure a text� Since full morphological recurrence is identical to lexical recurrence, the term “morphological recurrence” will refer here exclusively to partial recurrence. It implies using the same language items in different morphological ways to structure the text. Partial morphological recurrence is mostly instantiated by derivation. De Beaugrande (1981: 54) describes this device as a “shifting of already used elements to different classes (e.g. from noun to verb)”. This is, for instance, the case in the following Facebook comments: in (119) the finite verb form “auscheckt” has the same basic word components as the verbal noun “(zum) Auschecken”; in (120) the noun “winnaars” is morphologically very closely related to the past participle “gewonnen”: (119) Marc Raschke Da sagt man einen Abend vorher, dass man nicht um 12, sondern gegen 13 Uhr auscheckt. Der Nachtportier sagt ok, kein Problem. Um 12.40 Uhr kommt man zum Auschecken - und man soll 10 € Überziehungsgebühr zahlen. Unfassbar! ! ! (EC D Facebook: Maritim Hotel Berlin, Berlin) (120) Hotel Okura Amsterdam We hebben de winnaars van het Beauty & Wellness weekend zojuist op de hoogte gesteld dat ze hebben gewonnen! (EC NL Facebook: Hotel Okura Amsterdam, Amsterdam) Another case of morphological recurrence is compounding, which for typological reasons is more likely to be present in German and Dutch customer comments. Its relevance for a contrastive analysis is, therefore, doubtful. We will, nonetheless, include it as one of our contrastive parameters for the sake of completeness. A good example may be the next one (“Frühstücksraum” - “Frühstücksbuffet”): <?page no="103"?> 3.2 Text-grammatical structures in consumer comments 103 (121) Die Fensterbänke im Frühstücksraum waren nicht gereinigt. Das Frühstücksbuffet war sehr ausgedünnt. (EC D Booking: Wolfgang, Alleinreisender, Artlenburg, Deutschland, 7. März 2012) The proposed variables for the contrastive analysis of morphosyntactic cohesion are summed up in the following table� Syntactic recurrence syntactic parallelism: - NP-NP (...) - VP-VP (...) - other phrases chiasmus Partial morphological recurrence derivation compounding Table 12� Morphosyntactic cohesion Cohesion by ellipsis Ellipsis is simply the omission of an item within the text, which according to Halliday/ Hasan (1976: 143) can be referred to as “substitution by zero”. As such, it generally involves the “continuation of a previously stated referent” (Campbell 1995: 48), so that it could also be characterised as a reference device. Cohesion by ellipsis should be fairly common in customer comments, particularly in those written in a more casual style that mimics the spoken language, where, as Warren (2009: 46) points out, the use of ellipsis is “more prevalent in spoken discourse than in written discourse”� Halliday/ Hasan (1976: 146ff) distinguish three types of ellipsis depending on the elided element: nominal ellipsis, verbal ellipsis and clausal ellipsis. In keeping with Halliday/ Hasan and many other authors, we will speak of nominal ellipsis when the nominal phrase is partially or completely elided, as in the Spanish example (122). Here, “hotel” is left unsaid, but it is understood from within the context in “Tampoco dispone de Aire (sic) acondicionado”: (122) (...) por lo tanto, no recomiendo el hotel en época de frio en Salamanca (lo que viene siendo de noviembre a abril) Tampoco dispone de Aire acondicionado, con lo cual, en época estival, Julio-Agosto tampoco lo recomiendo de ninguna manera (...) (EC E Booking: Raul, Pareja joven, Madrid, España, 10 de abril de 2012). <?page no="104"?> 104 3 Analysing social web conversations and discourses from a contrastive point of view For practical reasons, Halliday/ Hasan’s distinction between verbal and clausal ellipsis will not be upheld in our contrastive approach: clausal ellipsis also involves a verb, even though the verbal form is embedded within a clause. By way of example, in the following German comment the finite form “war” has been left out in “die Zimmer tob (sic) und modern eingerichtet”: (123) Sehr zentral gelegenes Hotel. Der Service war super, die Zimmer tob und modern eingerichtet und absolut sauber (EC D Expedia: Ein verifizierter Reisender, Gesendet am 15. Juli 2012) Cohesion by conjunctive adjunction Conjunction is one of the most widely used cohesion devices in text linguistics. Conjunctive elements create texture “not in themselves but indirectly, by virtue of their specific meanings; they are not primarily devices for reaching out into the preceding (or following) text, but they express certain meanings that presuppose the presence of other components in the discourse” (Halliday/ Hasan 1976: 226). Since the term “conjunction” may be used specifically to denote words that link syntactic constituents, we will use here the denomination “conjunctive adjunction”. This term could be described according to Butler (2003: 116) as connective elements that “relate the clause to the text which has gone before”� This broad description of conjunctive adjunction will allow us to extend our contrastive analysis of textual cohesion devices to include not only coordinating and subordinating conjunctions, but also connective adverbs, prepositional phrases and prepositional expressions. We will refer to them as “conjunctive adjuncts”. As Schiffrin (2003: 55) points out, these texture-creating conjunctive items are in fact discourse markers� A detailed contrastive description of the connective cohesive devices in customer comments should include a practical classification of conjunctive elements. Such a classification would allow the linguist working with a corpus to obtain detailed and contrastable information about the cohesive structure of the conjunctive adjunction in two or more languages. Taking into account the morphosyntactic and semantic dimensions of connective items, two classifications will be proposed here accordingly� From a morphosyntactic point of view, conjunctive adjuncts will be classified by partially following Eggler’s classification made for separating connectives in argumentative texts (Eggler 2006: 90) into conjunctions, prepositional phrases, prepositional phrases, adverbs and particles. Coordinating conjunctions linking clauses appear to be commonplace in our exemplary corpus, both on Booking. com and on Facebook. The most common ones are the additive or copulative <?page no="105"?> 3.2 Text-grammatical structures in consumer comments 105 “and” and the adversative “but”, as in (124) - French “et”and (125) - German “aber”-: (124) Mon Hotel Paris Réservez avant le 07 Juillet 2014 sur notre page (onglet Tarif Fan) et bénéficiez de nos tarifs très privés pour votre séjour entre le 11 Juillet et le 03 septembre 2014 (EC F Facebook: Mon Hotel Paris, Paris) (125) Einziger Wermutstropfen ist die Nähe zur relativ gut befahrenen Bahnlinie, zeitweise etwas laut, aber wir wurden trotz geöffnetem Fenster nachts nicht dadurch geweckt, warum auch immer. Allerdings sollte man den Zeitaufwand für die Warterei an der Schranke stadteinwärts für eventuell einzuhaltende Termine mit einplanen, manchmal 10-15 Minuten. (EC D Booking: Anonym, Gruppe, Deutschland, 12. Juni 2012) Subordinating conjunctions are also observed at first glance, particularly in German and Dutch comments. A good example may be the temporal subordinator “als” in (126): (126) Als wir um kurz vor 17 Uhr ankamen, bekamen wir den Schlüssel, aber als wir aufs Zimmer kamen war es noch schmutzig vom Vorgänger. (EC D Booking: Ramona, Älteres Paar, Karlsruhe, Deutschland, 15. Januar 2012) Prepositional phrases acting as connectives also help to create cohesion in customer comments. In the Spanish comment (127), “a pesar de ser un hotel de 2 estrellas” is one such example of a connective prepositional expression� (127) Sin duda lo mejor del hotel es la excelente ubicación y que a pesar de ser un hotel de 2 estrellas, dispone de minibar (con agua mineral gentileza de la casa), secador de pelo y wifi gratuito (...) (EC E Booking: Niceto, Pareja joven,Viladecans, España, 20 de julio de 2011) Adverbs also appear to be popular conjunctive adjuncts in the exemplary corpus. For example, in (125) the German “Konkunktionaladverb” “allerdings” expresses an adversative relation, and in (128) the Dutch adverb “dus” expresses a causal one: (128) Na boeking via Booking.com ontving ik van booking.com de bevestiging van de boeking. Daarna een mail van Jay; s met de mededeling dat bij hun van mijn interesse voor het appartement is gebleken en dat het appartement op de door mij gevraagde datum nog vrij is. Het antwoord had dus ook kunnen zijn dat het appartement op die datum niet vrij zou zijn. (EC NL Booking: Frederik, Individuele reiziger, Maastricht, Nederland, 20 april 2012) <?page no="106"?> 106 3 Analysing social web conversations and discourses from a contrastive point of view Particles may also be used as connective devices (Fernández 1994: 75). This is particularly true of Germanic particles, which have a special status as a part of speech in German and Dutch grammar. In Spanish and French, the concept of particle is very vague (Beerbom 1992: 51ff), whereby most grammars speak of adverbs. Nevertheless, and because they are common in German and Dutch customer comments, particles will be included in our analysis, even if the possible results obtained from their corpus examination are not directly comparable� An illustrative example of a cohesive particle is the German “Modalpartikel” “doch” in (129): (129) 25 Euro pro Tag für die Tiefgarage ist doch etwas überzogen! (EC Booking G, Lutz, Alleinreisender, Fürth, Deutschland, 25. Juni 2012) Conjunctive adjuncts can develop different semantic relations. The semantic classification of conjunctive adjuncts and connective relations proposed here is based on Rudolf (1988: 108-109), and has already been accepted and/ or adapted by several authors (e.g., Martínez-Cabeza 2002: 91). The following semantic parameters will be highlighted: - Addition: conjunctive adjuncts adding information are expected to be very common. This semantic type of connection is observed in examples (124), (127), and (128): “Réservez (...) et bénéficiez de nos tarifs très privés pour votre séjour”, “la intención, además de dicha celebración, giraba en torno a (...)”, “en dat het appartement (...) nog vrij is”. - Contrast: contrast adjuncts express adversative and concessive relations. This is, for example, the case in (125) with “allerdings”, or in (126) with “aber”. - Time: temporal conjunctive adjuncts may also be common when reporting hotel experiences, for instance in (126): “als wir um kurz vor 17 Uhr ankamen, (...)”. - Causality: following Rudolf, under causal conjunctive adjuncts we understand any connective item that expresses a conditional relation, a relation of purpose, a relation of result, or only a simple causal relation. A good example is “con motivo de” in (127)� A linguistic analysis that includes these cohesive parameters of connectivity (see table 13 for a recapitulation) could be very productive from a contrastive point of view� <?page no="107"?> 3.2 Text-grammatical structures in consumer comments 107 Parts of speech conjunctions: - coordinating conjunctions - subordinating conjunctions prepositional phrases adverbs particles Semantic relations adjuncts adding information contrasting adjuncts temporal adjuncts causal adjuncts Table 13. Cohesion by conjunctive adjunction Coherence As discussed earlier, texture is also achieved by coherence. Our text-grammatical contrastive approach will avoid distinguishing between stylistic coherence and semantic/ pragmatic coherence, as proposed by authors such as Van Dijk/ Kintsch (1983) and Esser (2009: 141). The reason for our procedure is that there is no common interpretation among text linguists as to what “stylistic coherence” actually means and represents� The authors that introduced and popularised the term in text linguistics define this device as follows: “Stylistic coherence would mean that the speaker or discourse makes use of the same register, in lexical choice, sentence complexity and length, etc.” (Van Dijk/ Kintsch 1983: 149). The implementation of this definition in a contrastive analysis of online customer comments would create several practical problems at various levels� The most difficult one would certainly concern the register and lexical choice used by commenting users: it is highly doubtful that a unified and standard language register has already emerged� The same should apply for the lexical choice� As a result, the term coherence will be understood here as semantic and pragmatic coherence, which following Esser (2009: 141) we will not differentiate. In our interpretation, which has practical purposes, coherence implies an inferred relationship between two (or more) units of meaning as a way of structuring the text� Several models have so far been proposed for the analysis of these inferred relations. In fact, there are “many classifications and a variety of labels for coherence relations” (Taboada 2009: 126)� The best-known proposal may be Mann and Thompson’s Rhetorical Structure Theory (RST), developed in Mann/ Thompson (1985) and (1987). Both authors propose a rather complicated set of <?page no="108"?> 108 3 Analysing social web conversations and discourses from a contrastive point of view over 20 pragmatic and semantic relations. Because of its unnecessary complexity and unrelatedness to cohesion devices, which to a certain degree help to create coherence, RST is not a suitable approach for our practical goals. We will avoid the problem of sticking to one particular description model by adopting Renkema’s basic types of discourse relations, namely, the additive relation and the causal relation (Renkema 2004: 108-109). The additive relation is related to coordination. In our multilingual corpus, only addition, contrast and disjunction seem to occur as additive relations. A good example of a purely additive relation may be (130): we could link the Spanish nominal phrase with the copulative conjunction “y”. In (131), the additive relation is created by contrast: (130) Excelente situación en el centro. Perfecto trato y servicio del personal. Buen precio. (EC E Booking: Adelina, Grupo de amigos, Madrid, España, 12 de junio de 2012) (131) Kube Paris Booba92i au KUBE Hotel - Paris. Le Ice Kube Bar va chauffer. Uniquement sur invitation! (EC F Facebook: Kube Paris) The causal relation can be “traced back to an implication” (Renkema 2004: 109), and is a subordinating one. Again following Renkema in part and traditional grammar, six causal relations have been found throughout the corpus: - Cause and reason� The poor condition of the bed sheets and armchair was the reason (and the cause) the room was perceived by the client as being grubby and sleazy in (132): (132) Das Zimmer wirkte billig und schmuddelig. Die Laken waren zwar gewaschen, aber mit Flecken. Der eine alte Sessel im Zimmer hatte schmutzige Armlehen. (EC D Booking: Kathrin, Familie mit älteren Kindern, Braunschweig, Deutschland, 12. Juni 2012) - Means. In the following Dutch hotel review, the wonderful ornaments and the elegant interior are the means employed by the hotel management to give the client the impression of entering a lovely place: (133) Bij binnenkomst voel je direct de liefde waarmee deze bijzondere plek is opgebouwd. Aan de muur prachtige ornamenten die je als een parelsnoer naar de prettige receptieen ontbijtruimte leiden. Het interieur is smaakvol met accenten van vintage en art. (EC NL TripAdvisor: Ludovicus000, Rotterdam, The Netherlands, Beoordeeld op 24 juni 2012) <?page no="109"?> 3.2 Text-grammatical structures in consumer comments 109 - Consequence. This is a very straightforward causal relation, as (134) and (135) can confirm. In (134) the hotel rooms are very noisy; this is because they are not sound-proof. In (135), the client comes to the conclusion (as a consequence) that she will not choose this hotel anymore because of its poor condition: (134) Las habitaciones no están nada insonorizadas. Mucho ruido. (EC E Booking: Anónimo, Grupo de amigos, España, 14 de junio de 2011) (135) Sterk verouderd motel. Kamer rook zeer muf. Slecht schoongemaakt. Kamer moet echt aangepast worden. Zou er niet meer heen gaan (EC NL Booking: Karin, Vriendengroep, Dronten, Nederland, 18 juni 2012) - Purpose. This causal relation does not seem to be as frequent as one might suppose. The French Facebook comment (136) is a good example: allegedly, the purpose of the new hotel offer is to thank the hotel’s Facebook fans for being so numerous: (136) Mon Hotel Paris 2687 fans... ça se fête! Pour vous remercier, profitez de notre offre exclusive de l’été Fans Monhotel! ! ! ! (EC F Facebook: Mon Hotel Paris, Paris) - Condition. This causal relation does not seem to be very common either. In the following German example, the necessary conditions for the applicant to gain an internship at the hotel are listed under “deine Qualifikation”: (137) Astrid Zölfel Der Ball kommt ins Rollen und mit ihm kommst Du ins andel’s Hotel Berlin! Wärme Dich unter hoher Ansteckungsgefahr mit dem WM-Fieber für Dein Spiel des Lebens in der Berliner #Hotellerie auf! Deine Qualifikation: Du absolvierst Deine erste Übungseinheit mit einem Lauftraining Du beweist Dein Geschick beim Mixen des Power Drinks “Green” (..) (EC D Facebook: Andel’s Hotel Berlin, Berlin) - Concession. Concessive relations are very closely linked to contrast, whereby they can be easily confused with the additive relation of contrast� Only very clear concessive relations (i.e., those in which the meaning “in spite of the fact that” is expressed) should be classified as such, as in (138). Here, in spite of the fact that the beds could have been larger and that breakfast was a little bit meagre, the positive aspects outweigh the negative ones: <?page no="110"?> 110 3 Analysing social web conversations and discourses from a contrastive point of view (138) die Betten könnten breiter sein - das Frühstücksangebot ein wenig größerder positive aspekt überwiegt (EC D Booking: Rainer, Alleinreisender, Marloffstein, Deutschland, 12. Oktober 2011) The coherence relations listed above are implicit ones. It goes without saying that coherence can also be explicit. In that case, coherence devices are introduced by adjuncts. In this regard, adjuncts adding information and contrastive adjuncts will relate to an additive relation, whereas causal adjuncts will obviously refer to one of the six causal relations. This being so, we do not see the need to conduct a second contrastive analysis regarding coherence: our proposal already includes the examination of the explicit coherence within the cohesive device of connectivity� Table 14 sums up the coherence devices proposed for the contrastive analysis� Additive discourse relations addition contrast disjunction Causal discourse relations cause and reason means consequence purpose condition concession Table 14� Coherence Deixis Deixis is the last texture-creating discourse device that will be included in our practical text-grammatical contrastive model. It is a contextual phenomenon that is grammatically represented in the text by different expressions, and may be understood as “part of the modality function in language, i.e. the situating of the information with respect to the current context” (Werth 1999: 157)� One of the best definitions of deixis has been provided by Lyons (1977: 636), for whom it is the function of certain expressions that “relate utterances to the spatio-temporal co-ordinates of the act of utterance”. Deictic elements in online customer comments always have a reference point that is usually determined by the commenting internet user. These reference points, called the “deictic center” <?page no="111"?> 3.2 Text-grammatical structures in consumer comments 111 by Fillmore (1975: 83), “zero-point” by Lyons (1977: 646) and “origo” by many others following Bühler (1934/ 1982: 107), as well as the relations established in connection to them, structure the personal, temporal and local axes upon which the comment is situated (prototypically “I”, “now” and “here”). Person deixis grammaticalises the role of participants in hotel comments from the perspective of the commenting hotel customer. In German, Dutch and French, it is usually realised using one of their personal pronouns and an inflected verb form. In Spanish, personal deixis is mostly expressed by verbal inflection alone. However, in all four languages, only personal pronouns are to be considered responsible for deixis when they are used in an oblique case. Deictic personal pronouns may have a reference function when they refer back (or, rarely, forward), and therefore be part of the reference cohesive device. Yet this is not always the case by far, particularly when they are used in the nominative: in the following Dutch Facebook posting, the unstressed personal pronoun “we” is not referring to anything mentioned before, it is simply the person marker for the first person plural of the verb. So it is a deictic pronoun, but not a referring one: (139) 22tracks Afgelopen vrijdag, Valentijnsdag, mochten we in samenwerking met Bad Manor 3 dames (plus een vriendin) een exclusieve akoestische showcase incl. overnachting in citizenM Hotel Amsterdam City aanbieden (EC NL Facebook: CitizenM Amsterdam, Amsterdam) Personal deictic elements, be it personal pronouns or verbal inflection, appear to be very common in our multilingual corpus. They should be classified according to the grammatical person and its number in order to gain an accurate picture of their use in customer comments. In this regard, one of the most interesting assumptions to be made from a contrastive point of view concerns the T-V distinction; that is, the distinction between familiar and non-familiar or polite forms of address (“tu” and “vos” in Latin), which Levinson (2003: 89) calls “social deixis”. At a first glance, Spanish and Dutch users seem to address potential readers with a familiar pronoun (e.g., example 140), whereas French and German ones seem to prefer a polite form (141): (140) Si reservas en Placentinos, asegúrate que no te dan la habitación número 12 porque es francamente diminuta (EC E Booking: Luis, Grupo de amigos, Pozuelo de Alarcón- Madrid, España, 22 de agosto de 2011) (141) Pour le même prix vous trouverez des hôtels au centre ville plus confortables. Fuyez : ( (EC F Booking: Marie, groupe d’amis, Merignac, France, 1 avril 2012) <?page no="112"?> 112 3 Analysing social web conversations and discourses from a contrastive point of view Place deixis allows the commenting hotel client to set up a situative frame around them or around the hotel being reviewed� Spatial deixis is expressed by grammatical items that act as place relators� Consistent with relevant studies on deixis, pragmatics and text linguistics, for example Meibauer (2008: 14), the most usual place relators in our corpus are demonstrative pronouns, demonstrative adjectives, demonstrative adverbs, prepositional phrases and even some noun phrases. For example, in (142) there is a nominal phrase (“eine hervorragende Lage”), a prepositional phrase (“mitten in der Fußgängerzone”) and a demonstrative adverb (“dort”) acting as place relators: (142) Das Hotel hat eine hervorragende Lage. In 5 Minuten ist man mitten in der Fußgängerzone. Die Mitarbeiter waren sehr freundlich und hilfsbereit. Das Frühstücksbuffet war sensationell. Die Sauberkeit der Zimmer und Bäder waren vorbildlich. Wir würden jederzeit wieder dort einchecken. (EC D Booking: Jens, Gruppe, Alsfeld, Deutschland, 6. April 2012) It is noteworthy that some deictic demonstrative pronouns and demonstrative adjectives may also be part of the cohesive device of reference. Both deictic elements, as well as the demonstrative adverbs, can be distal or proximal. Time deixis concerns the temporal orientation of the posting writer� Since tenses, and hence verbal forms, may sometimes be implied, they are “difficult to isolate” (Renkema 2004: 123). However, verbs do not seem to play an important role in temporal deixis in our multilingual corpus� Most deictic elements that express time are adverbs, adjectives, prepositional phrases, noun phrases and verb tenses. In the following German example, there are two time adverbs (“zuvor” and “schon”) and two prepositional phrases (“bis zum 07.09.2012” and “im Voraus”) with deictic meaning: (143) Wir haben vieles gesehen jedoch so etwas blieb uns bis zum 07.09.2012 erspart! Erschreckend das diese Hotel wirklich 2 Sterne hat! Wir kamen am Abend in das Hotel, zuvor schickten wir schon mal im Voraus eine Kostenübernahe die in jedem Hotel akzeptiert wird, diese wurde nicht berücksichtigt, die Dame an der Rezeption wurde sehr unfreundlich und ausfallend. (EC D TripAdvisor: katharina_mueller, München, Deutschland, Bewertet am 21. November 2012) For the sake of accuracy, time relators should be classified according to their temporal reference; that is, whether they are situated at or near the origo point (simultaneity), before it (anteriority), or after it (posteriority). Table 15 summarises all the deictic elements and relations that can be taken into account in a text-grammatical analysis of customer comments� <?page no="113"?> 3.2 Text-grammatical structures in consumer comments 113 Personal deixis Person: - First person - Second person (familiar) - Second person (polite) - Third person Number - singular - plural Place deixis Place relators: - demonstrative pronouns - demonstrative adjectives - demonstrative adverbs - prepositional phrases - noun phrases Place relations (only dem. pronouns, adj. and adv.): - distal - proximal Time deixis Time relators: - adverbs - adjectives - prepositional phrases - noun phrases - verb tenses Time relations: - simultaneity - anteriority - posteriority Table 15. Deixis 3.2.1 Text-grammatical structures on the comments pages of online travel and hotel booking agencies The texture-creating text-grammatical devices that have been described earlier may occur in customer comments posted on the comments pages of online travel and hotel booking agencies, as well as in Facebook postings and other text types on the social web� Contrary to what was proposed for text functions and text actions in subchapter 3.1., this is why we will use exactly the same description model and very similar table charts for all Web 2�0 text types that will be examined. In the following description of the text-grammatical structures of <?page no="114"?> 114 3 Analysing social web conversations and discourses from a contrastive point of view Western European comments on hotels, every cohesive, coherence and deictic device will be empirically compared� The empirical procedure will also be very similar to the one applied for analysing the functional and actional communicative structure: the comparative figure for calculating percentage rates will be the total amount of hotel reviews collected in the corpus (300 comments posted on the online travel agencies Booking.com, Expedia and TripAdvisor; that is, 100 comments per online booking agency)� Cohesion by reference and substitution As already mentioned, cohesion by reference and substitution should be one of the most common cohesive devices in customer comments� The evaluation of our corpus data confirms this hypothesis. The reference device is almost exclusively realised by pronouns. Pronominal adverbs, as well as pro-verbs and pro-adjectives 8 , are found in only a tiny percentage in all four languages (see table 7 at the end of this sub-section)� These types of reference devices are thus marginal ones in Western European hotel reviews and customer comments. Some kinds of pronouns are used in more than two thirds of all German (77.3%), French (72%) and Spanish (70%) comments. The figure for the Dutch corpus is slightly lower (64.6%). The most frequent manner of achieving cohesion by reference is by far the referential use of a personal pronoun. Referential personal pronouns are extremely common in German (71%)� They are also widely used in French (63.3%), but less so in Dutch (64.6%) and least of all in Spanish (54%)� Since Spanish verbs are only very rarely accompanied by personal pronouns, this last figure may still appear a high one. A closer look at the results obtained after the examination of the Spanish corpus will reveal the reason: a personal pronoun in the nominative form is found in only 8�6% of all 162 Spanish comments, where such a pronoun acts as a referring item. This is one of the most remarkable differences between Spanish and the other languages analysed: a personal pronoun acting as a reference device is used mostly in an oblique case, as in the following example: (144) El hotel nos encanto. Violeta fue un encanto. Nos gustaria quedarnos mas tiempo, pero ibamos con tiempo asignado. De seguro se lo recomendamos a amigos (GC E Expedia: Un viajero verificado, Publicado el 3 de junio de 2012) This may also explain why nearly one third of all Spanish comments (32�3%) include some pronominal reference in the third person singular: the oblique case 8 Pro-verbs, pro-adjectives and another pro-forms are defined following Sasse (2015: 192) as “substitutes which replace members of the corresponding full-word category”� <?page no="115"?> 3.2 Text-grammatical structures in consumer comments 115 form “lo” very often refers back to “hotel”. In some cases, the aforesaid NP does not appear in the comment, whereby texture is created by substitution and not by reference. A good example may the very short hotel review (145), where the accusative pronoun “lo” is clearly substituting “el hotel”: (145) El servicio y la ubicación al precio que lo ofrecían. (GC E Booking: Jose, Pareja mayor, Antequera (Málaga), España, 3 de junio de 2012) With the exception of Spanish, first person pronouns are the most common ones among those with a reference function. A pronoun of the first person plural can be observed in 37% of all German comments, 30.3% of all French ones, and in 29.3% of all Dutch ones (as well as in 23% of the Spanish ones). These figures are slightly higher than those including a first person singular pronoun: 24.3%, 22.6% and 20.6%, respectively, of French, German and Dutch comments use a referring “je”, “ich” or “ik”. The third person is relatively common in German comments (the first person singular and plural appear in 11% and 18% of them), as well as in the Dutch ones (especially in the plural: 14%). French hotel customers do not seem to resort to the third person as a reference device as often as their German and Dutch counterparts: only 2.6% and 8% of all French comments in our general corpus include a referring pronoun in the third person singular or plural� The Spanish results for the third person have already been discussed and explained: there seems to be a direct link between the high percentage of referring devices in the third person oblique case, and the usual lack of personal pronouns in definite verbal phrases: when personal pronouns accompanying the verb are not common, texture is also created by a third person in an oblique case. In a written contextual situation where no visible interlocutor is present or can be addressed, it is not surprising that the second person is among the less used referring devices. In fact, only French hotel reviewers tend to use the polite second person plural pronoun “vous” to some extent (8%) to refer back or forward� In both Romanic languages, as well as in Dutch, an impersonal pronoun, or an impersonally used pronoun, is also commonly used as a referring pronoun. Spanish hotel reviewers make use of the impersonal “tú” in 28�6% of all corpus comments, mostly in an oblique case as in (146): the second accusative pronoun “te” refers back to the first one, but it does not imply a contextual second person. It is just a colloquial means of creating an impersonal assertion: (146) Solo tiene planta baja para las habitaciones, al menos yo no vi más. Por la noche te despierta el timbre de la puerta del hotel. A no ser que lo pidas <?page no="116"?> 116 3 Analysing social web conversations and discourses from a contrastive point of view no te ofrecen nada, ni un mapa de la ciudad. (GC E Booking: Anónimo, Pareja joven, España, 18 de mayo de 2011) In French, 14% of the corpus comments include the impersonal pronoun “on”, which refers to a pronoun or a NP of the first person singular or plural, as is the case in (147), where “on” in “on se croit” refers back to “je” in “je dois dire”: (147) La perfection n’existe pas, mais je dois dire que la propriete TOUTOUNE s’en approche, ses propriétaires sont sympas, serviables et surtout trés attentifs, Le dépaysement total, en centre ville etre en un instant dans une Yourte, on se croit en mongolie (GC F Booking: Fabrice, voyageur individuel, Montpellier, France, 2 mars 2012) A very similar percentage (12.6%) is obtained from the examination of the Dutch corpus. The colloquial impersonal “jij” is here used far more than the proper impersonal pronoun “men”. In (148), the unstressed impersonal “je” in “wat je er voor betaalt” creates texture by referring back to “wij”: (148) Wij als gezin van 2 volwassenen en 1 kind waren verplicht om 2 kamers te boeken want ze hebben geen 3 persoonskamer en er kon geen bed bij geplaast worden. De kamer was in verhouding te klein wat je er voor betaalt. (GC NL Booking: Vanessa, Gezin met oudere kinderen, Lelystad, Nederland, 31 juli 2011) “Jij” is thus an impersonal substitute form for “wij” or “ik”, and can be used in the same posting by the commenting client together with the formal treatment “u”. In the following example, the second impersonal “je” refers to the first one. The second “wij” is also a reference device, but “u” is not. (149) Je verwacht de deuren van een oude kast open te doen, maar ineens kijk je in de strakke badkamer met gescheiden ligbad en inloopdouche (...). Wij hebben Mary K ervaren als een zeer bijzondere, liefdevolle, warme plek die wij aan iedereen (ook zakelijke gasten) zouden willen aanraden. U moet er rekening mee houden dat het hotel niet beschikt overeen lift. Parkeren kan het beste in garage Paardenveld (25 euro voor 24 uur, Kroonstraat 9) (GC NL TripAdvisor: Ludovicus000, Rotterdam, The Netherlands, Beoordeeld op 24 juni 2012) In German, the impersonal pronoun “man” is present in only 4% of all corpus comments, for instance in (150), where it refers back to the NP in dative “Frauen und Männern”: (150) (...) es gibt nur eine Kleinigkeit, die aber Frauen und auch Männern das gute Aussehen schwer macht: das ist die Beleuchtung im Bad (...) Mit <?page no="117"?> 3.2 Text-grammatical structures in consumer comments 117 Halogen - wie im Königshof - ist man denkbar schlecht beraten (EC D Booking: Heike, Alleinreisender, Konstanz, Deutschland, 9. Juli 2012) The logical conclusion to be drawn from these telling results is that hotel reviewers seem to comment on the basis of their own personal experiences, and therefore make use of referring devices that involve first person personal pronouns, be it plural if they were travelling as a couple or in a group, or singular if they were staying alone at the hotel. As an alternative, impersonal or impersonal pronouns are also used: they usually refer or imply a first person as well. The use of referring possessive pronouns and possessive adjectives is similar in all four corpus analysed� They can be found in 21% of all Spanish comments and in 18.6% of the German ones, as well as in 17% and 14%, respectively, of the Dutch and French hotel reviews. The only striking (but minor) difference between the four languages may be the much higher (albeit still not really significant) figure in the use of Spanish possessive pronouns: they appear in 6.6% of all Spanish comments, but only in 1.6% and 1% of the French and German ones, respectively. By way of example, “la nuestra” in (151) refers back to “la habitación”: (151) La limpieza del baño y de la habitación eran bastante escasas (...) Cuando avisamos que el baño estaba sucio subieron sin ningún problema a limpiar. Pero me pregunto si entraron a la nuestra o a otra porque el baño seguía dando el mismo asco. (GC E Booking: Susana, Grupo de amigos, Zaragoza, España, 14 de septiembre de 2011) Possessive adjectives with a reference function are slightly more common in both German (18.6%) and Dutch (17%) than in Spanish (16.3%) and French (14%). In both Germanic languages and in French, the possessive adjective of the first person plural is the most usual one when it comes to creating texture: it is present in 16% of all German comments, 13.6% of the Dutch, and 12% of the French ones. In the Spanish corpus, they appear in only 8% of all comments. The German and the Dutch examples (152) and (153) may be taken as prototypical ones: (152) Wir haben eine Nacht im Hotel verbracht. Unser Zimmer mit Himmelbett war wirklich sehr schön eingerichtet im Shabby Style (GC D TripAdvisor: dredg, Baden-Baden, Deutschland, Bewertet am 19. März 2012) (153) (...) wij twee fietsvrienden, fietsten van Zwitserland langs de Rijn terug naar nederland. Hotel Rhenen lag op onze route (GC NL Booking: Willem, Vriendengroep, Leidschendam, Nederland, 25 juni 2012) The possessive adjectives of the first person singular seem to be more common in Spanish (9.6%) and French (8.6%) than in Dutch (7%) and German (only <?page no="118"?> 118 3 Analysing social web conversations and discourses from a contrastive point of view 6%). In the following Spanish and French examples, the possessive adjectives “mi” in “mi mujer” and “mes” in “mes voisins” clearly refer to “yo (llegué)” and “j’(entendais)”: (154) Pero todo fué horrible desde el principio. Yo llegué antes que mi mujer, recogí la llave y fuí a buscarla. (GC E TripAdvisor: TCO76, Castro Urdiales, España, Opinión escrita el 4 marzo 2013) (155) (...) et les murs sont si fins que je n’ai guère dormi plus de 3h par nuit, étant donné que j’entendais la porte d’entrée de l’hôtel claquer, et mes voisins de chambre parler. (GC F Booking: Justine, voyageur individuel, Villeneuve, France, 19 novembre 2011) Thirdand second-person possessive adjectives are not really significant as reference devices. The only exceptions may be the Spanish “tu”, which has an occurrence of 6% based on its impersonal use, and the French “votre”, which in 4�3% of the comments refers to the deferential pronoun “vous”� In our multilingual corpus, cohesion by reference is achieved in some cases using demonstrative pronouns and demonstrative adjectives. In this regard, Spanish and French comments record the highest rates of 15.3% and 14.6%, respectively, followed by German hotel reviews, in which such a cohesive device is present in 11.3% of all comments. The Dutch seem to be somewhat more reluctant to use referential demonstrative pronouns and adjectives: they are only observed in 6�6% of all comments� Demonstrative pronouns are more common in German (9.3%) and Spanish (8.3%) than in French (5.6%) and Dutch (3.6%). In all four languages, most of these pronouns are proximal ones, like “diese” in (156), the neutral “dies” in (157), and “estos” in (158): (156) Im Zimmer sind die Kabel für Telefon, Fernseher und andere Geräte so wirr verlegt, dass man den Sessel leicht darüber rollen un diese beschädigen kann. (GC D Booking: Peter, Alleinreisender, Schwanstetten, Deutschland, 11. Februar 2012) (157) Dafür war dies auch das erste Hotel das es doch tatsächlich schaffte das Zimmer zu reinigen während wir zum frühstücken weg waren und uns nach dem Frühstück sofort wieder gereinigt zur Verfügung stand. (GC D TripAdvisor: Juergen_Cl, Greifenstein, Bewertet am 10. Januar 2013) (158) (���) existen carteles exigiendo silencio���pero para la gente mal educada estos son invisibles. (GC E Booking: Endika, Pareja joven, Bilbao, España, 26 de junio de 2012) <?page no="119"?> 3.2 Text-grammatical structures in consumer comments 119 However, distal pronouns, for instance “dat” in the following Dutch example, are also used in a few Dutch, Spanish and French comments: (159) Kamer erg gedateerd met name de badkamer, wel schoon. Het bed was goed en dat is toch het belangrijkste! (GC NL Booking: Wieke, Individuele reiziger, Zuidlaren, Nederland, 23 april 2012) Colloquial pronouns (for an illustrative example see 10 above) are only present in the German corpus� Demonstrative adjectives also appear to be more usual among French (12%) and Spanish (11.6%) hotel reviewers. In comparison, only 6.3% and 5% of all German and Dutch comments have a referential demonstrative adjective. With the exception of Dutch, such demonstrative adjectives are usually proximal, as in (160), where “cet” in “cet hotel” refers back to “l’hôtel”. Distal ones are only relatively common in Spanish (see example 161, “ese hotel”) and Dutch: (160) Ne connaissant pas du tout Montpellier, je craignais de ne pas trouver l’hôtel très facilement (...) Je conseille cet hôtel aux voyageurs en déplacement à Montpellier (EC F Expedia: Guitter, Publié le Jun 06, 2012) (161) Como solamente fui por razones laborales, estuve muy poco tiempo en el hotel, aun así, el bullicio de la calle llegaba a mi habitación, el quinto piso, dormí mal y en cuotas, honestamente no volvería a ese hotel (GC E Expedia: Un viajero verificado recomendado, Publicado el 2 de octubre de 2012) Table 16 sums up all the results commented on above. The main differences between the languages analysed concern the low occurrence of personal pronouns in the nominative form in Spanish for grammatical reasons in comparison with the other three languages, the use of a referential impersonal pronoun or impersonally used pronoun in nearly a third of all Spanish comments, and only in one seventh of all French comments and in 4% of all German ones, as well as the higher occurrence of referential demonstrative pronouns and adjectives in Spanish and French. Cohesive-device Language German Dutch Spanish French Σ (%) Σ (%) Σ (%) Σ (%) Using a pronoun 232 (77�3%) 194 (64�6%) 210 (70%) 216 (72%) Personal pronouns 213 (71%) 178 (59�3%) 162 (54%) 190 (63�3%) <?page no="120"?> 120 3 Analysing social web conversations and discourses from a contrastive point of view Person and number: - 1 st PS - 2 nd PS - 3 rd PS - 1 st PP - 2 nd PP - 3 rd PP - Impersonal 284 68 (22�6%) 2 (0�6%) 33 (11%) 111 (37%) 4 (1�3%) 54 (18%) 12 (4%) 238 62 (20�6%) 0 (0%) 6 (2%) 88 (29�3%) 2 (0�6%) 42 (14%) 38 (12�6%) 313 24 (8%) 0 (0%) 97 (32�3%) 69 (23%) 1 (0�3%) 36 (12%) 86 (28�6%) 263 73 (24�3%) 0 (0%) 8 (2�6%) 91 (30�3%) 25 (8�3%) 24 (8%) 42 (14%) Case - Nominative form - Oblique form 213 158 (74�2%) 55 (25�8%) 178 147 (82�6%) 31 (17�4%) 162 14 (8�6%) 148 (91�4%) 190 137 (72�1%) 53 (27�9%) Possessive pronouns and adjectives 56 (18�6%) 51 (17%) 63 (21%) 42 (14%) Possessive pronouns: - 1 st PS - 2 nd PS - 3 rd PS - 1 st PP - 2 nd PP - 3 rd PP 3 (1%) 1 (0�3%) 00 2 (0�6%) 00 0 (0%) 000000 20 (6�6%) 8 (2�6%) 3 (1%) 2 (0�6%) 7 (2�3%) 2(0�6%) 2 (0�6%) 5 (1�6%) 4 (1�2%) 0 1 (0�3%) 6 (2%) 2 (0�6%) 0 Possessive adjectives: - 1 st PS - 2 nd PS - 3 rd PS - 1 st PP - 2 nd PP - 3 rd PP 55 (18�3%) 18 (6%) 0 (0%) 8 (2�6%) 48 (16%) 2 (0�6) 5 (1�6%) 51 (17%) 21 (7%) 0 (0%) 5 (1�6%) 41 (13�6%) 0 (0%) 3 (1%) 49 (16�3%) 29 (9�6%) 18 (6%) imp 7 (2�3%) 24 (8%) 4 (1�3%) 3 (1%) 40 (13�3%) 26 (8�6%) 0 (0%) 4 (1�3%) 36 (12%) 13 (4�3%) 11 (3�6%) Demonstrative pronouns and adjectives 34 (11�3%) 20 (6�6%) 46 (15�3%) 44 (14�6%) Demonstrative pronouns - Distal - Proximal - Colloquial 28 (9�3%) 0 (0%) 25 (8�3%) 7 (2�3%) 11 (3�6%) 2 (0�6%) 10 (3�3%) 0 (0%) 25 (8�3%) 6 (2%) 23 (7�6%) 0 (0%) 17 (5�6%) 3 (1%) 16 (5�3%) 0 (0%) <?page no="121"?> 3.2 Text-grammatical structures in consumer comments 121 Demonstrative adjectives - Distal - Proximal - Colloquial 19 (6�3%) 0 (0%) 19 (6�3%) 0 (0%) 15 (5%) 12 (4%) 8 (2�6%) 0 (0%) 35 (11�6%) 14 (4�6%) 29 (9�6%) 0 (0%) 36 (12%) 5 (1�6%) 34 (11�2%) 0 (0%) Pronominal adverbs 5 (1�6%) 6 (2%) 0 (0%) 0 (0%) Using a pro-verb 3 (1%) 1 (0�3%) 0 (0%) 1 (0�3%) Using a pro-adjective 2 (0�6%) 0 (0%) 1 (0�3%) 5 (1�6%) Table 16� Cohesion by reference and substitution in customer comments on online travel and hotel booking agencies in German, Dutch, Spanish and French Lexical cohesion The data concerning lexical cohesion extracted from our multilingual corpus are surprisingly consistent for all four languages: lexical texture is achieved mostly by reiteration� This lexical cohesion device appears in nearly the same percentage in French (57.6%), Spanish (56.3%) and German (52.3%). The figure for Dutch is slightly lower, 40.6%. The most important lexical relation that creates texture by reiteration appears in all the languages, with the exception of Dutch lexical recurrence and synonymy� Lexical recurrence appears to be especially frequent in French: 40% of all French comments include some sort of texture-creating word repetition. For instance, a geographic denomination (“Montpellier”) is repeated up to three times in (162) in a single sentence: (162) Accueil gentil, mais personnel incapable de nous diriger dans Montpellier, en voiture avec le portable, “vous êtes où” “je ne connais pas ce boulevard? ” Arrivée à Montpellier vers 18 h, arrivée dans la chambre 3 h 30 + tard ! (Travaux importants dans Montpellier) (GC F Booking: Alain, couple d’âge mûr, Cusset, France, 1 septembre 2011) Lexical recurrence is also common in the other three corpora examined: it is present in 35.6% of all Spanish comments, as well as in 31.6% and 30.3% of the German and Dutch comments. In (163), the reiterated element is an adverb (“muy”), while in (164) it is a noun. (163) El trato es muy agradable, te orientan muy bien, y parece que les conoces de toda la vida, al lado hay un restaurante de la hosteria, que se come <?page no="122"?> 122 3 Analysing social web conversations and discourses from a contrastive point of view muy bien y muy econocmico (GC E Booking: Aida, Pareja mayor, Hazas de Cesto, Cantabria, España, 16 de junio de 2012) (164) Weiterhin ein Etagenbad, obwohl uns ein eigenens Bad versprochen wurde. Auf Nachfrage, warum wir kein eigenes Bad haben, sind wir für dumm verkauft worden: die Zimmer haben alle ein eigenes Bad. (GC D TripAdvisor: ReinholdJessen, Jessen, Bewertet am 3. Juni 2009) Synonymic relations are not as common as one might expect� This may be due to the rather short nature of most hotel reviews� Synonymy is only relatively frequent in French comments: nearly a quarter of them (23%) include one or more synonyms that refer back to a term. The figure for German and Spanish comments, 17% and 16.3%, respectively, is very similar, whereas the use of synonymic expressions seems to be very restricted in the Dutch corpus (only 8%). The reason for this different lexical behaviour may be found in the divergent lexical strategies used by speakers of the four languages. It seems that French hotel reviewers describe hotels at length and with a profusion of synonyms in order to convey their opinion about the hotel stay. On the other hand, the Dutch tend to describe their experiences in a more laconic way. Within this scale, German and Spanish hotel clients are situated somewhere in the middle. A very good example of the flowery use of lexical items by French commentators is (165). In this sequence, “idéal” is a synomym of “de rêve” and “parfait”. “Ville” and “espace urbain” are also related to each other here by a synonymic relation� The following Spanish and German examples (“muy bien situado” - “completamente centrico” -sic-, and “unsauber” - “dreckig”) are not so lexically dense: (165) Un lieu idéal pour une nuit de rêve, un lieu parfait pour se reposer et découvrir la ville pendant son séjour. Un conte de fées dans un espace urbain... (GC F Booking: Guilhem, couple d’âge mûr, Saint Feliu D’Avall, France, 5 mars 2012) (166) Buen hotel, muy bien situado, completamente centrico. (GC E Expedia: Jose Luis, Algeciras, Publicado el 17 de febrero de 2012 por un cliente verificado de Hotels.com) (167) Zimmerteppich, Dusche etc. Machten einen sehr unsauberen Eindruck. Wir haben nichts gegen eine einfache Unterkunft, aber gegen eine dreckig erscheinende durchaus etwas. (GC D Booking: Johannes, Junges Paar, Hamburg, Deutschland, 8. Mai 2012) Hyponymy and hyperonymy are not very common, but in contrast to the two preceeding lexical devices, they are more frequent in German (9.3%) and Span- <?page no="123"?> 3.2 Text-grammatical structures in consumer comments 123 ish (7.6%) than in French (6.3%). In (168), “Kinder” is the hyperonym of “Kleinkinder”: (168) Ich habe selber Kinder, aber wenn in einem Hotel von Freitags bis Montags Kleinkinder den Flur ununterbrochen lautstärkemäßig zur Spielwiese zu machen, da habe ich kein Verständnis mehr für! (GC D Booking: Uwe, Älteres Paar, Weilerswist, Deutschland, 11. Juli 2012) Meronymic relations are present with almost the same frequency in all four languages, ranging from 16.3% in the case of the Spanish corpus to 12% in the case of the German one. The following comment sequences are a good example: in the first one, meronymy is established, on the one hand, between “habitación” and “mobiliario”, “aire acondicionado”, “nevera” and “cocina”, and on the other, between “cocina” and “vitrocerámica, fregadero”, etc. In the second sequence, “kamer” includes a “muur”, a ”(fijn) bed” and a ”(plat) kussen”: (169) Mi habitación resultó sencillamente extraordinaria teniendo en cuenta la categoría del establecimiento, ya que además de estar limpia y con mobiliario nuevo, contaba con aire acondicionado, nevera y una pequeña cocina (vitrocerámica, fregadero, cubiertos, vajilla...). (GC E: Marino, Pareja joven, Oviedo, España, 18 de julio de 2011) (170) Prettige kamer: verschillende kunstwerken aan de muur, fijn bed en een plat kussen: heerlijk geslapen. (GC NL Booking: Individuele reiziger, Haarlem, Nederland, 15 september 2011) Antonomy is a means of creating lexical texture only in fewer than 5% of all comments in all the languages. In the few cases it appears, it affects not only adjectives, but also adverbs (as in 171: “buiten”-“binnen”): (171) Buiten kun je aan de gracht een drankje doen, binnen is de slaapkamer en achter de halfdoorzichtige glazen wand de badkamer (...) (GC NL Booking: Tijn, Jong stel, Amsterdam, Nederland, 28 augustus 2011) Collocation is especially frequent in German and Dutch. Collocations are twice as common in both Germanic languages as in the Romanic ones: 23% of all German comments present some sort of collocation� The same is true for 20�3% of all Dutch comments. In comparison, this lexical device is found in only 11.3% of the Spanish hotel reviews and in 12.3% of the French ones. A prototypical example of collocation is “trappen (erg) stijl en smal” in (172): (172) Prijs kwaliteit vond ik slecht, trappen erg stijl en smal, er hing een onfris luchtje (GC NL Booking: Franca, Individuele reiziger, Leiden, Nederland, 20 oktober 2011) <?page no="124"?> 124 3 Analysing social web conversations and discourses from a contrastive point of view As explained above, and as can be observed by studying table 17, there are no major differences concerning the use of the lexical cohesion device: lexical recurrence is the preferred lexical strategy in all cases. Lexical recurrence is followed in second place by collocation in the case of German and Dutch, and by synonymy and meronymy in Spanish and French. This is the main difference at this level of analysis� Cohesive- device Language German Dutch Spanish French Σ (%) Σ (%) Σ (%) Σ (%) Reiteration 157 (52�3%) 122 (40�6%) 169 (56�3%) 173 (57�6%) Lexical recurrence 95 (31�6%) 91 (30�3%) 107 (35�6%) 120 (40%) Synonymy 51 (17%) 24 (8%) 49 (16�3%) 69 (23%) Hyponymy and hyperonymy 28 (9�3%) 15 (5%) 23 (7�6%) 19 (6�3%) Meronymy 36 (12%) 39 (13%) 49 (16�3%) 42 (14%) Antonymy 10 (3�3%) 10 (3�3%) 14 (4�6%) 13 (4�3%) Collocation 69 (23%) 61 (20�3%) 34 (11�3%) 37 (12�3%) Table 17. Lexical cohesion in customer comments on online travel and hotel booking agencies in German, Dutch, Spanish and French Morphosyntactic cohesion Morphosyntactic cohesion is created in all four illustrative corpora mainly by syntactic parallelism. It is, in fact, the only syntactic recurrence device that has been found in the corpus� The texture-creating repetition of structural elements in hotel reviews with the aim of relating clauses to each other appears to be more common in German, French and Dutch than in Spanish. It is present in 63% of all German comments, as well in 58% and 57.3% of French and Dutch hotel reviews, respectively. The Spanish figure (45.3%) is clearly lower. A good deal of syntactic repetitions concern nominal phrases, such as the French example (117) or the almost prototypical Dutch comment (173). However, the repetition of verbal phrases with a cohesive aim is also common, particularly in Spanish, French and German. German example (118) and Dutch comment (174) below are repetitions of a simple syntactic structure. In (175), the syntactic recurrence <?page no="125"?> 3.2 Text-grammatical structures in consumer comments 125 not only includes a VP; an infinitive phrase (“para ser un cutro -sicestrellas”, “para subir a la parte de arriba”) is also part of the cohesion device: (173) Plezierige, ruime kamer goed bed, goed ontbijt (GC NL Booking: Harry, Ouder stel, Maastricht, Nederland, 25 juni 2012) (174) De kamer was nog niet af. De kamer was heel erg gehorig. (GC NL Booking: Valentijn, Jong stel, Eindhoven, Nederland, 20 november 2011) (175) Para ser un cuatro estrellas las instalaciones son muy mejorables. Por delante de las habitaciones que se encuentran es pequeñas casitas adosadas hay un pasillo de piedrecitas que son demasiado ruidosas para poder descansar cada vez que pasa alguien con una maleta. Para subir a la parte de arriba hay unas tremendas escaleras. (GC E Expedia: Un viajero verificado, Publicado el 19 de abril de 2011 por un cliente verificado de Hotels. com) Partial morphological recurrence can only be considered a cohesive device in German and Dutch. Although some derivations have been observed in the Spanish and French corpus, the rate is only 1% in the former case and under 1% in the latter one. In German and Dutch, derivation and compounding are not used very often as cohesive devices, but they can easily be spotted in the corpus. The total figure for both cases of partial morphological recurrence amounts to 9.6% in German, and to 8% in Dutch. In some cases, as in (176), complementary use is made of derivation (“Park-” as part of a compound and “parken” as an infinitive) and compounding (“Parkplätze”, “Parkhäuser”): (176) Das Einzige das man eventuell als nicht optimal bewerten kann sind die Parkplätze. Da das Hotel inmitten der Fussgängerzone liegt ist man gezwungen in den umliegenden Parkhäusern zu parken, was nicht gerade günstig ist. (GC D TripAdvisor: jojonline, Lindau, Deutschland, Bewertet am 22� Juni 2010) Overall, the text-grammatical analysis of morphosyntactic cohesion has not thrown up significant differences across the four languages. The (not really significant) exception could be the somewhat lower percentage rate of occurrences regarding syntactic parallelisms in Spanish. As a result of the different typological structures of German and Dutch, on the one hand, and Spanish and French, on the other, the data on partial morphological recurrence cannot be compared� Table 18 sums up the result of the contrastive analyses� <?page no="126"?> 126 3 Analysing social web conversations and discourses from a contrastive point of view Cohesive-device Language German Dutch Spanish French Σ (%) Σ (%) Σ (%) Σ (%) Syntactic recurrence 189 (63%) 172 (57�3%) 136 (45�3%) 174 (58%) Syntactic parallelism 189 (63%) 172 (57�3%) 136 (45�3%) 174 (58%) Chiasmus 0 0 0 0 Partial-morphological recurrence 29 (9�6%) 24 (8%) 3 (1%) 2 (0�6%) Derivation 20 (6�3%) 18 (6%) 3 (1%) 2 (0�6%) Compounding 24 (8%) 15 (5%) 0 (0%) 0 (0%) Table 18� Morphosyntactic cohesion in customer comments on online travel and hotel booking agencies in German, Dutch, Spanish and French Cohesion by ellipsis Although ellipsis is very common in hotel reviews, above all of the copula verb in short hotel reviews, cohesion by ellipsis appears to be rather rare in all four general corpora. This is especially true of the nominal ellipsis, which can be observed in 7.6%, 7% and 6.3% of all Spanish, Dutch and French comments, respectively, but only in 2.6% of hotel reviews that belong to the German corpus. Verbal ellipsis, which in our contrastive approach includes clausal ellipsis, is more common, particularly in Spanish (13.3%) and Dutch (11.6%), but also in French (8.6%) and German (8.3%). Although it mostly implies the copula verb, examples where a verb other than the copula one has been omitted can also be found in the corpus. In (177), the previously introduced modal verb “können” (“könnte...sein”) has been elided in “das Frühstücksangebot ein wenig größer-”: (177) die Betten könnten breiter sein - das Frühstücksangebot ein wenig größerder positive aspekt überwiegt (GC D Booking: Rainer, Alleinreisender, Marloffstein, Deutschland, 12. Oktober 2011) These results seem to disprove our initial hypothesis that cohesion by ellipsis would be common in customer comments because they are written in a colloquial language style. The data examined have provided evidence for the opposite in all four languages, in particular in the case of nominal ellipsis. Table 19 sums up the statistical results� <?page no="127"?> 3.2 Text-grammatical structures in consumer comments 127 Cohesive- device Language German Dutch Spanish French Σ (%) Σ (%) Σ (%) Σ (%) Nominal ellipsis 8 (2�6%) 21 (7%) 23 (7�6%) 19 (6�3%) Verbal ellipsis 23 (8�3%) 35 (11�6%) 40 (13�3%) 26 (8�6%) Table 19� Cohesion by ellipsis in customer comments on online travel and hotel booking agencies in German, Dutch, Spanish and French Cohesion by conjunctive adjunction As predicted, conjunctive adjunction seems to be one of the most common cohesion devices in all four languages. However, some considerable differences can be identified after a thorough analysis of the multilingual data extracted. One of the most striking discrepancies concerns the frequency of occurrence of the parts of speech used as connective elements. In this regard, conjunctions are used in more than half the comments in all four of the language corpora examined, but the percentage figure is much higher in Spanish than in the other three languages: whereas a conjunction is present in more than three fourths of the Spanish comments (76.6%), it appears in only 57.3%, 56% and 53% of the French, Dutch and German comments, respectively. The explanation for the high frequency observed in the Spanish corpus may lie in the very common (and colloquial) use of the conjunction “y” and other similar copulative connectives. These types of cohesive adjuncts, which have previously been defined as “adjuncts adding information”, are to be found in 66% of all Spanish comments, but only in 38% of the German ones and in slightly more than 30% of the Dutch ones (see table 11 for more details). Another telling finding that suggests that Western European hotel reviewers use somewhat different texture-creating adjuncts concerns the presence of subordinating conjunctions in the corpus. Despite the relatively low use of conjunctions in the corpus, German has the higher percentage of occurrence of subordinating conjunctions (32%), followed by Spanish (25.6%). French (10%) and Dutch (8.3%), on the other hand, tend to use much fewer subordinators. These results are rather surprising, as they imply that German -along with Spanishhotel reviewers prefer the verbal style in a good deal of comments (the verbal style is a necessary requirement for using subordinating conjunctions), and that Dutch (and French) commentators either do not use them or use shorter sentences with fewer clauses� <?page no="128"?> 128 3 Analysing social web conversations and discourses from a contrastive point of view Prepositional phrases acting as cohesive connective devices are only common in Spanish: 11�6% of all Spanish comments include some prepositional phrases with such a function� Connective adverbs are more common in all four languages, especially in German (12.3%), Spanish (10.6%) and Dutch (9.6%), but also in French (7.6%). In this regard, no real differences can be observed. Particles are not common as connective adjuncts, even in German and Dutch. Nevertheless, some of them can be found in both Germanic languages� From a semantic point of view, conjunctive adjuncts adding information are the most frequent ones in all four languages. However, as already pointed out, such connective elements are far more common in Spanish than in any other of the three languages. In fact, Spanish adjuncts adding information are twice, or nearly twice, the figure for Dutch or German adjuncts. This type of connection is not only very common for coordinating conjunctions, such as the copulative “y”, “et”, “und” and “en”. Adverbs adding information are also common. This is, for instance, the case of “sogar” in (178), embedded in a nominal style, of “además” in (179), and of “d’ailleurs” in (180), which from a morphological point of view could also be interpreted as a prepositional phrase: (178) Freundliches Personal - sogar sehr freundlich - sehr gutes Preis-Leistungs-Verhältnis. Sehr gutes Frühstücks-Buffet. (GC D Booking: Margot, Alleinreisender, Frankfurt am Main, Deutschland, 9. April 2012) (179) Tuvimos que dejar la llave en recepción al salir a cenar y, al volver, la atención fué pésima, atendiéndonos en pijama y de mala gana. Además, desde la habitación se escuchaba bastante el ruido de la calle y nos despertaron con una radio o TV a todo volumen. (GC E Booking: Anónimo, Pareja joven, España, 25 de mayo de 2012) (180) Sincèrement, je voyage souvent pour le travail et c’est la seule fois ou tant de prestations étaient proposé! (...) D’ailleurs, j’ai déjà réservé sur place une semaine en famille ! ! ! ! (GC F Booking: Stéphane, famille avec enfants, Montmorency, France, 3 mars 2012) Contrasting adjuncts are present, ranging from 23.6% of all comments in Dutch and French to 20.3% in German. The results are thus very close for all four languages. Contrastive relations, which are mainly adversative and concessive, are mostly expressed either by coordination conjunctions (see “maar” in example 181 and “sino” in 182) or by subordinators, as in examples 183-85 (German “obwohl”, Spanish “aunque” and French “même si”): (181) De brasserie heeft mooi eten maar de bediening kan beter. (GC NL Trip- Advisor: Mark V, Beoordeelt op op 11 Juli 2012) <?page no="129"?> 3.2 Text-grammatical structures in consumer comments 129 (182) Creo que este hotel debería ser revisado, no sólo en cuanto a servicios y estrellas, sino respecto a las normas de salubridad y limpieza. (GC E TripAdvisor: Aylabe, Asturias, Opinión escrita el 13 octubre 2009) (183) (...) zum anderen konnte ich nach dem Marathon noch duschen (obwohl die Checkout-Zeit schon überschritten war). (GC D Booking: Ludger, Älteres Paar, Münster, Deutschland, 24. April 2012) (184) En general bien aunque el desayuno un poco justo para un hotel de 4 estrellas (GC D Expedia: Un viajero verificado, Publicado el 12 de diciembre de 2012) (185) (...) boire un verre sur la terrasse panoramique autour de la piscine etait un vrai dépaysement complet même si nous sommes de la région. (GC F Booking: Christine, couple jeune, Castelnau le Lez, France, 18 avril 2012) Temporal conjunctive adjuncts record only a two-digit figure in German (14%), and are mostly subordinating conjunctions, being used mostly for the chronological description of personal experiences. In this regard, example (186) could be seen as an exception: here “sobald” is embedded in a more general (and not merely chronological) description of breakfast time at the hotel� Temporal connective adjuncts are not nearly as common in German as in Spanish (6.6%), French (6.3%) and Dutch (4%). These results suggest that, when using the verbal style, German reviewers proceed chronologically, whereas the Dutch, Spanish and French do not. It may also be noteworthy that most Spanish temporal adjuncts are infinitive phrases introduced by the preposition “a” (and hence considered morphologically as prepositional phrases). “Al llegar” in (187) may be a good example: (186) Frühstücksbuffet ausgesprochen vielseitig (...) - sobald etwas zur Neige geht, wird gleich wieder aufgefüllt. (GC D Booking: Petra, Gruppe, Daun- Neunkirchen, Deutschland, 6. Dezember 2011) (187) Las atenciones al llegar, los detalles de iluminacion en la habitacion,la calidad y elegancia de los sanitarios, lenceria,etc. El desayuno (GC E Booking: Mar, Familia con niños mayores, Alicante, España, 20 de agosto de 2011) Causal conjunctive adjuncts are more frequent than temporal ones. Giving the reason for a particular assertion with a causal adjunct is especially common in Spanish (24%), German (17.6%) and French (15%), but less so in Dutch (where only 8.3% of all comments include a causal connective adjunct). It seems that stating the reason is an argumentative line of Western European (and particu- <?page no="130"?> 130 3 Analysing social web conversations and discourses from a contrastive point of view larly Spanish) hotel reviewers -with the notable exception of the Dutch. In the Spanish corpus, subordinators such as “porque” or “ya que”, as well as a causal infinitive phrase introduced by the preposition “a” (see example 188), are extremely common� (188) Al ser tan céntrico, en las habitaciones que dan al exterior se oye mucho el ruido de la calle . . . (GC E Booking: Niceto, Pareja joven, Viladecans, España, 20 de julio de 2011) It is common in German and Dutch to subordinate conjunctions such as “weil”/ “omdat” and even some subordinators marked as belonging to refined language, such as “zumal” in (189) and “aangezien” in (190), as well as causal adverbs such as “also” (example 191), “daher” and “thus”. (189) Parken überteuert, zumal ein alternatives Parkhaus (Oper) direkt um die Ecke ist. (GC D Booking: Ronny, Junges Paar, Frankfurt, Deutschland, 10. Juni 2012) (190) Het is best gehorig, aangezien er verkeer ‘boven’ je hoofd rijdt. (GC NL Booking: Nicolien, Jong stel, Beverwijk, Nederland, 4 maart 2012) (191) Eine wunderschöne Einrichtung genau wie auf den Bildern. Der service war sehr gut und die Hotel Mitarbeiter freundlich. Es ist sehr modern eingerichtet mit neuen Möbeln und es war sehr sauber! ! ! Also Preis Leistung für Super und ich kann das Hotel nur weiter empfehlen! (GC D Booking: Markus, Junges Paar, Böhmenkirch, Deutschland, 3. Juli 2012) The French reviews have both causal coordinating and subordinating adjuncts. For example, in (192) the coordinating “car” is used: (192) La facturation de l acces a la piscine et hammam qui est une heresesie commerciale car il semble judicieux de l inclure au prix de la chambre (GC F Booking: Gérard, couple d’âge mûr, Donzenac, France, 6 janvier 2012) The results obtained by analysing the extracted data are detailed in table 20. It seems clear that cohesion by conjunctive adjunction is very common in all four languages, but it is realised in a different way in each corpus. <?page no="131"?> 3.2 Text-grammatical structures in consumer comments 131 Cohesive-device Language German Dutch Spanish French Σ (%) Σ (%) Σ (%) Σ (%) Parts of speech - Conjunctions: • coordinating conjunctions • subordinating conjunctions 159 (53%) 115 (38�3%) 96 (32%) 168 (56%) 156 (52%) 25 (8�3%) 230 (76�6%) 215 (71�6%) 77 (25�6%) 172 (57�3%) 161 (53�6%) 30 (10%) - Prepositional phrases 18 (6%) 6 (2%) 35 (11�6%) 19 (6�3%) - Adverbs 37 (12�3%) 29 (9�6%) 23 (7�6%) 32 (10�6%) - Particles 17 (5�6%) 12 (4%) 0 0 Semantic relations - Adjuncts adding information 114 (38%) 91 (30�3%) 198 (66%) 133 (44�3%) - Contrasting adjuncts 61 (20�3%) 71 (23�6%) 71 (23�6%) 66 (22%) - Temporal adjuncts 42 (14%) 12 (4%) 19 (6�3%) 20 (6�6%) - Causal adjuncts 53 (17�6%) 25 (8�3%) 72 (24%) 45 (15%) Table 20. Cohesion by conjunctive adjunction in customer comments on online travel and hotel booking agencies in German, Dutch, Spanish and French Coherence As discussed earlier, texture is also achieved by coherence. Achieving texture in hotel customer comments by coherence, in other words, without the help of any signalling device, allows the reader to infer a relationship between two (or more) units of meaning� Coherence does not appear to be a very common way of structuring this specific text subgenre. The range of percentage values does not differ considerably for the four partial corpora and is, in any case, between 13% (German) and 10% (Dutch). Causal discourse relations are even scarcer: among the 300 customer comments that make up each one of the four corpora, we find that only 4.3%, 5.3% and 5.6% of the Dutch, German and Spanish comments, respectively, signal some type of causal relations without a discourse <?page no="132"?> 132 3 Analysing social web conversations and discourses from a contrastive point of view marker. Only the French figure (7.3%) is a little higher. Overall these results are relatively consistent with the findings in Taboada (2013: 9), one of most extensive and most recent reports on the relative distribution of relations with respect to whether they are indicated by a discourse marker or by an inferred relation (i.e., coherence). This is particularly true of the distribution of causal discourse relations. The figure we have obtained for additive discourse relations is higher than Taboada’s, which is probably because online customer comments mostly have a spatial restriction on the number of characters the reviewer may use� This may have led to a reduction in the number of discourse markers� There does not seem to be any divergent uses of additive discourse relations in the four corpora� German hotel reviewers are certainly more prone to the use of purely additive discourse relations that are not signalled: 10�3% of all German comments have some kind of such purely additive relations. The figures obtained for the Spanish and the French corpora are very close (9.3% and 9%, respectively); while the Dutch one is slightly lower (7.6%). The following German and Dutch examples may illustrate how purely additive relations are constructed. In both of them, the copulative conjunctions “und” and “en” could have been used to signal the additive relation: (193) Sehr sehr kleines Bad; Teppichboden im Zimmer; zu weiche Matratzen (...) (GC D Booking: Roland, Älteres Paar, Grünkraut, Deutschland, 21. September 2011) (194) Erg mooi vormgegeven en ingericht hotel: ‘designhotel’, waardoor het een luxueus en persoonlijk tintje krijgt. Grote kamers die met recht de benaming suite verdienen. Erg fijn restaurant waar ook het uitgebreide, verse ontbijt wordt geserveerd. Je krijgt als gast 10% korting op het diner. (GC NL Booking: Natasja, Ouder stel, Amsterdam, Nederland, 28 juni 2012) The two other types of additive relations, particularly disjunction, are very rare and sporadic (see table 21)� Contrastive relations are at least present in 4% of all Dutch comments and in 3.6% of all German ones. To this effect, the lowest figure is the Spanish one (2.3%). In the following Dutch and German examples, the contrastive conjunctions “maar” and “aber” have been omitted. In example “197”, the adversative conjunction “no obstante” has been left out: (195) Het uitzicht vanuit het restaurant was prachtig, vanuit de kamer niks. (GC NL Booking: Peter, Individuele reiziger, Vlissingen, Nederland, 2 januari 2012) <?page no="133"?> 3.2 Text-grammatical structures in consumer comments 133 (196) Das Hilton verbindet man mit Glamour. Leider sind aber die Zimmer sehr einfach, das Parkhaus kostet 25EUR/ Nacht und besondere Extras bekommt man nicht (GC D Expedia: Gast, Veröffentlicht am Jun 13, 2012) (197) La habitación individual demasiado incomoda, pequena y bastante ruidosa. Hotel bastante deprimente, no lo aconsejo para nada. La ubicación excelente, el metro en la puerta del hotel. (GC E Expedia: Un viajero verificado, Publicado el 7 de febrero de 2012 por un cliente verificado de Hotels�com) None of the six causal relations indicated by Renkema (2004: 109) is common. In fact, most of them are few and far between. Even the indication of cause and reason without the use of a signalling device is very rare, and in no case exceeds 2%: it is present in only 1.6% of all Spanish comments, and it is even scarcer in the other three corpora. It is not usual in this text genre to omit the adjuncts when expressing cause and reason. Examples such as (198), where the subjunction “por lo que” has been left out, or (199), where a whole causal phrase such as “denn es liegt” is missing, are the exception rather than the norm: (198) Excelente hotel con extraordinarios servicios, bien ubicado y personal muy amable. Volveremos, sin duda. (GC E Booking: Enrique, Pareja joven, Guadarrama, Madrid, España, 24 de enero de 2012) (199) Die Lage des InterCityHotels ist fantastisch - mitten in der Innenstadt und direkt neben dem Bahnhof. (GC D TripAdvisor: Familie_Welchering, Pfaffenhofen an der Ilm, Deutschland, Bewertet am 30. August 2012) The only causal relation found in more than 3% of all comments is related to a semantic relation of consequence. French reviewers tend to use this type of coherence device slightly more often than German, Dutch and Spanish users. However, the low percentage rate of occurrence in all four languages does not allow drawing detailed conclusions regarding a possible divergent use of this device. In our corpus, this type of coherence is used to justify a decision or a suggestion (as in 200), or to expose a consequence derived from a fact (as in 201): (200) Het ontbijt is puur slecht (...) Ook de bedden waren van matige kwaliteit, de minibar was stuk en 1lamp deed het niet, de badkamer had te weinig licht en de trap om naar de eerste verdieping te komen is eng. Ik zal voortaan n ander hotel kiezen (GC NL Booking: IreneJong stelVlaardingen, Nederland10 juli 2012) (201) Han puesto doble ventana. Ya no se oye nada de ruido del exterior (GC E Booking: Xavier, Pareja joven, Valencia, España, 21 de mayo de 2012) <?page no="134"?> 134 3 Analysing social web conversations and discourses from a contrastive point of view As the detailed results from the corpus analysis in table 21 suggest, there do not appear to be any significant differences regarding the implicit coherence relations in German, Dutch, Spanish and French hotel customer comments. Coherence- device Language German Dutch Spanish French Σ (%) Σ (%) Σ (%) Σ (%) Additive discourse relations 39 (13%) 30 (10%) 33 (11%) 36 (12%) Addition 31 (10�3%) 23 (7�6%) 28 (9�3%) 27 (9%) Contrast 11 (3�6%) 12 (4%) 7 (2�3%) 10 (3�3%) Disjunction 4 (1�3%) 2 (0�6%) 3 (1%) 6 (2%) Causal discourse relations 16 (5�3%) 13 (4�3%) 17 (5�6%) 22 (7�3%) Cause and reason 4 (1�3%) 2 (0�6%) 5 (1�6%) 3 (1%) Means 3 (1%) 5 (1�6%) 2 (0�6%) 5 (1�6%) Consequence 12 (4%) 10 (3�3%) 9 (3%) 14 (4�6%) Purpose 4 (1�3%) 2 (0�6%) 6 (2%) 7 (2�3%) Condition 2 (0�6%) 1 (0�3%) 4 (1�3%) 4 (1�3%) Concession 2 (0�6%) 3 (1%) 4 (1�3%) 1 (0�3%) Table 21� Coherence in customer comments on online travel and hotel booking agencies in German, Dutch, Spanish and French Deixis As expected, deixis appears to be a very common texturing device in hotel customer comments. In our corpus, this is particularly true of person deixis, which is found in one way or another in more than 70% of all the comments analysed� In this respect, the highest figure is observed in the Spanish corpus, where 82% of all comments include a deictic element. The lowest figure corresponds to the Dutch corpus (71.3%). In Spanish, the deictic element is usually found only in verbal inflections. Some Dutch and German comments also omit the personal pronoun, which is a marker of colloquial speech in both languages. In (202), for instance, the Dutch pronoun “ik” is missing in “zou er net meer heen gaan”: <?page no="135"?> 3.2 Text-grammatical structures in consumer comments 135 (202) Sterk verouderd motel. Kamer rook zeer muf. Slecht schoongemaakt. Kamer moet echt aangepast worden. Zou er niet meer heen gaan (GC NL Booking: Karin, Vriendengroep, Dronten, Nederland, 18 juni 2012) With the notable exception of Spanish, first person deixis is the most common deictic element in our corpus. The figures range from 63.3% in French and 63% in German to 56.3% in Dutch, and only 52.6% in Spanish. There is another divergence between Spanish and all three other corpora: whereas in the German, Dutch and French comments the first person deixis is mainly both deictic and referential (“d&r” in the table), in Spanish hotel reviews the nature of the first person deixis can be deictic and referential, as well as only deictic (“only d” in the table). Both types of person deixis may sometimes be encountered in the same comment, as in (203), where the verbal form “fui” involves the first person in the utterance, and the personal possessive adjective “mi” refers back to the first person expressed in “fui”, being thus both deictic and referential: (203) Como solamente fui por razones laborales, estuve muy poco tiempo en el hotel, aun así, el bullicio de la calle llegaba a mi habitación (GC E Expedia: publicado el 2 de octubre de 2012 por un cliente verificado de Hotels.com) The second person deixis is relatively common in Spanish and French, but unusual in German. The data obtained for the Dutch corpus have to be handled with care, since most deictic second personal pronouns found in the comments only allow an impersonal interpretation: “je” is the colloquial impersonal pronoun in Dutch. The second person deixis is realised overwhelmingly by the colloquial form in Spanish and by the polite form in German and French (examples 204 and 205)� This is hardly surprising� (204) Nun müssen Sie nicht alle Badezimmer komplett sanieren ... - ich möchte Ihnen vorschlagen, eine Zusatzlampe zu installieren - sei es Linestra, led oder eine schlichte Sparlampe. (GC D Booking: Heike, Alleinreisender, Konstanz, Deutschland, 9. Juli 2012) (205) Vous devriez effacer cet établissement de votre liste! (GC F Booking: Maria, voyageur individuel, Vigny, France, 29 mai 2011) Third person deixis is far more common in the Spanish corpus than in the other three corpora� This may be because the third person plural is the preferred impersonal construction in Spanish. However, this cannot explain the large gap in the percentage between Spanish and the other three languages, so these findings should be treated with caution. It is noteworthy that there are some deviations between these results, which are detailed in table 22, and the data obtained for the cohesive device of refer- <?page no="136"?> 136 3 Analysing social web conversations and discourses from a contrastive point of view ence� These deviations are due to the fact we do not distinguish here between plural and singular, and that singular and plural deictic pronouns sometimes occur in the same review� Place deixis does not seem to be as common as personal deixis among hotel customer comments. Nevertheless, it is present in slightly more than 50% of all German, Spanish and French comments, and in 41% of all Dutch comments in the corpus� Since place deixis is closely linked to cohesion by reference when demonstrative pronouns and demonstrative adjectives are implied, the results obtained for that cohesion device have been partly extrapolated to the current analysis of place deixis. As table 13 shows, the use of demonstrative pronouns, demonstrative adjectives and demonstrative adverbs (e.g., German “dort” or Dutch “daar”) as place relators is not very high in percentage terms. Demonstrative adverbs are only of some significance in the German and French corpus. Particularly in German, their use seems to be linked to a final hotel recommendation, as in the following example: (206) Wer individuelle hotels mit eigeem charme und kreativ-hochwertigen zimmern will, ist dort sehr gut aufgehoben! (GC D TripAdvisor: MaryMillerLondon, Frankfurt am Main, Deutschland, Bewertet am 6. Januar 2013) In all four languages, place deixis is mostly realised by prepositional phrases that refer to a concrete location inside the hotel itself or nearby. Deictic prepositional phrases appear in around half the comments. In example (297), we come across two such prepositional phrases: “bij Spa Sereen in Maarssen” (a location nearby but outside the hotel) and “in een prachtige kamer” (a location inside the hotel itself): (207) Wij hadden een heerlijke dag doorgebracht bij Spa Sereen in Maarssen, daarna konden we in een heerlijk bed in een prachtige kamer duiken. (GC NL Tripadvisor: Wendy I, Veghel, Noord-Brabant, Netherlands, Beoordeeld op 10 september 2012) Some noun phrases also express place deixis� Their incidence varies from 15�6% in French and 14% in Dutch to 12% in German and 11.3% in Spanish. As to the place relations conveyed by place relators, the spatial perspective adopted by hotel commentators is mainly a neutral one� This means that place relation is not marked in any way� This is particularly common in prepositional phrases and noun phrases. In this respect, it is remarkable that a proximal place relation can be found in only 3% to 5.3% of all corpus comments. Distal relations are far more common, although they cannot be observed in more than 20% of the comments in any language: in 19�3% of all German comments and in 15�3% of all Dutch ones; the figures are even lower for Spanish and French. It is not <?page no="137"?> 3.2 Text-grammatical structures in consumer comments 137 uncommon to find distal (or proximal) place relators together with unmarked (and thus neutral) ones in the same comment, as in (208), where the prepositional phrase “in de stadstuin van het Hotel (sic)” is neutral, but “daar” is distal: (208) Daarna nog heerlijk borrelen in de stadstuin van het Hotel. Lekkere wijn met borrelnootjes die je erbij kreeg. Je kunt daar lekker zitten omdat het volledig beschut is met de avondzon. (GC NL Expedia: Petra uit Wildervank, geplaatst op Jun 07, 2010) Time deixis is not as common as had been assumed� There was no deictic time relator in over 3% of all the customer comments analysed. This is why the findings concerning time deixis have not been detailed in table 22� To sum up the findings revealed by our deictic analysis, the results show that person and place deixis are very common among German, Dutch, Spanish and French hotel customer comments. Nonetheless, whereas no really striking differences could be observed regarding the spatial anchoring of the corpus comments, person deixis seems to be realised in a somewhat different way in our four partial corpora: the first person is more common in German and French than in Spanish and Dutch, the second person deixis presents some differences due to the usual terms of address in Spanish, on the one hand, and in French and German, on the other, and third person deixis appears to be far more common in Spanish than in any one of the other three languages� Table 22 provides the detailed results for the deictic analysis of the general corpus� Deixis device Language German Dutch Spanish French Σ (%) Σ (%) Σ (%) Σ (%) Person deixis -221-(73.6%) -214-(71.3%) -246-(82%) -223-(74.3%) First person 189 (63%) - d&r: 164 (54�6%) - only d: 76 (25�3%) 169 (56�3%) - d&r: 142 (47�3%) - only d: 94 (31�3%) 158 (52�6%) - d&r: 89 (29�6%) - only d: 98 (32�6%) 190 (63�3%) - d&r: 156 (52%) - only d: 104 (34�6%) Second person (or familiar form) 6 (2%) - d&r: 6 (2%) - only d: 5 (1�6%) 59 (19�6%) - d&r: 2 (0.6%) - only d: 58 (19�3%) 62 (20�6%) - d&r: 1 (0.3%) - only d: 62 (20�6%) 5 (1�6%) - d&r: 2 (0.6%) - only d: 5 (1�6%) <?page no="138"?> 138 3 Analysing social web conversations and discourses from a contrastive point of view Second person (or polite form) 21 (7%) - d&r: 0 - only d: 21 (7%) 2 (0�6%) - d&r: 0 - only d: 2 (0�6%) 3 (0�9%) - d&r: 0 - only d: 3 (0�9%) 52 (17�3%) - d&r: 23 (7�6%) - only d: 48 (16%) Third person 67 (22�3%) - d&r: 49 (16�3%) - only d: 26 (8�6%) 59 (19�6%) - d&r: 46 (15�3) - only d: 22 (7�3%) 204 (68%) - d&r: 124 (41�3%) - only d: 149 (49�6%) 58 (19�3%) - d&r: 30 (10%) - only d: 45 (15%) Place deixis 164 (54�6%) 123 (41%) 168 (56%) 163 (54�3%) Place relators Demonstrative pronouns 28 (9�3%) 11 (3�6%) 25 (8�3%) 17 (5�6%) Demonstrative adjectives 19 (6�3%) 15 (5%) 35 (11�6%) 36 (12%) Demonstrative adverbs (“dort”) 36 (12%) 12 (4%) 16 (5�3%) 32 (10�6) Prepositional phrases 138 (46%) 150 (50%) 163 (54�3%) 145 (48�3%) Noun phrases 36 (12%) 42 (14%) 34 (11�3%) 47 (15�6%) Place relations Distal 58 (19�3%) 46 (15�3%) 23 (7�6%) 28 (9�3%) Proximal 11 (3�6%) 9 (3%) 16 (5�3%) 10 (3�3%) Neutral 114 (38%) 91 (30�3%) 141 (47%) 134 (44�6%) Table 22. Deixis in customer comments on online travel and hotel booking agencies in German, Dutch, Spanish and French <?page no="139"?> 3.2 Text-grammatical structures in consumer comments 139 3.2.2 Text-grammatical structures in comments on social networking services Most of the text-grammatical devices detected on the comments pages of online travel and hotel booking agencies can also be found in social networking services like Facebook. Since our operative working method has been presented (and applied) in detail in the previous section, the description of the empirical results extracted from our multilingual Facebook corpus will be restricted to those text-grammatical structures that are divergent in one or more of the languages studied� Cohesion by reference and substitution Cohesion by reference and substitution is also one of the most common cohesive devices in Facebook comments written by hotel managers and internet users. Cohesion is achieved almost solely through the use of pronouns. In this respect, pronouns are used as a cohesive device in one way or another in more than 90% of all Facebook comments in German, Spanish and French, while the figure for Dutch is slightly lower (88%). The most interesting findings concern the use of referential personal pronouns. The first person singular is widely used by all Facebook users, particularly by present or prospective hotel clients. However, as a quick glance at table 23 confirms, the figure recorded is slightly higher for Spanish and German. The first person plural is also relatively common, particularly among French and German hotel managers writing posts on behalf of the hotel or answering comments posted by other users. The use of the first person plural is much more restricted in Spanish and Dutch. The main differences concern the second person. In the singular, the second person is relatively common in all the languages except French: whereas it can be found in more than 28% and 26% of the comments in Spanish and Dutch, respectively, and even in 17.3% of all German comments in the corpus, but it is only present in 1.6% of the French ones. These results are not unsurprising, as French speakers do seem to respect the traditional sociolinguistic tu/ vous-distinction when addressing unknown people on Facebook or when trying to be polite, as in the following Facebook conversation between the hotel management and a prospective client (209): <?page no="140"?> 140 3 Analysing social web conversations and discourses from a contrastive point of view (209) Hidden Hotel **** Paris shared UNE beauty’s event. Le 27 juin, profitez d’un brunch bio 100% végétarien au Hidden Hotel ! Au programme: (...) Un Cactus Sur La Platine Vous en faites chaque mois? Hidden Hotel **** Paris Bonjour, nous en organisons de temps en temps (GC F Facebook: Hidden Hotel **** Paris, Paris) They are hardly surprising in Spanish either, as the Spanish tend to use the second person singular as the unmarked second person. The German and Dutch results, on the other hand, are very telling: the tu/ vos-distinction, which tends to be the norm in both colloquial and educated German speech (less so in Dutch), does not seem to be used in the Facebook comments analysed. Here at least, the colloquial nature of Facebook communication has led to a shift in register. Conversations between clients and hotel staff, as in (210), where only the second persons “du” and “ihr” (alongside the first person) are used, would be unimaginable in a face-to-face situation in reception: (210) Maike Fröse Stellt den Kuchen kalt, bin in 14 Tagen wieder bei euch und genieße den tollen Service Hotel Berlin, Berlin Wir freuen uns auf Dich Maike! Viele Grüße. (GC D Facebook: Hotel Berlin, Berlin) The results obtained for the second person plural need to be explained in some detail. The French figure (32.3%) is due to the use of the polite “vous”. By contrast, the German one (25%) is explained by the fact that both the hotel management and all other Facebook users tend to use the plural form rather than the singular one to address each other. The Spanish and the Dutch figures (12.6% and 14.6%, respectively) are lower: both the Spanish and the Dutch seem to prefer the singular form over the plural one� The third person is much less used in all the languages� The higher percentage rates are found in Spanish (211) and French: (211) Hotel Regina Madrid #Madrid se mueve en Verano. Mañana martes, a partir de las 22.00 horas, con un espectáculo de música, danza, luz y velas, para celebrar el solsticio de verano. La obra es abierta y gratuita para todos. (GC E Facebook: Hotel Regina Madrid, Madrid) <?page no="141"?> 3.2 Text-grammatical structures in consumer comments 141 Possessive and demonstrative pronouns and adjectives can also be identified in the multilingual corpora, but there do not seem to be any salient differences between the languages analysed when the previous results for personal pronouns are taken into account. As for the pronominal adverbs, pro-verbs and pro-adjectives, they are almost absent from the corpus, so they will not be considered. In conclusion, only the pronominal reference is common in all four languages. In German, the hotel management refers to itself predominantly in the first person plural, whereas other Facebook users tend to use the singular form. The favourite forms of address are “du” and “ihr”� This is also mostly true for Dutch. In Spanish, the second person singular is widely used by both hotel staff and clients. In French, the polite form is the only common one. As in German, Dutch and Spanish, the hotel management prefers to refer to itself in the plural, and other Facebook users do so in the singular. The detailed results are shown in table 23� Cohesive-device Language German Dutch Spanish French Σ (%) Σ (%) Σ (%) Σ (%) Using a pronoun 277 (92�3%) 264 (88%) 272 (90�6%) 270 (90%) Personal pronouns 263 (87�6%) 243 (81%) 257 (85�6%) 253 (84�3%) Person and number: - 1 st PS - 2 nd PS - 3 rd PS - 1 st PP - 2 nd PP - 3 rd PP - Impersonal 83 (27�6%) 52 (17�3%) 41 (13�6%) 69 (23%) 75 (25%) 26 (8�6%) 4 (1�3%) 75 (25%) 79 (26�3%) 49 (16�3%) 48 (16%) 44 (14�6%) 21 (7%) 12 (4%) 89 (29�6%) 85 (28�3%) 51 (17%) 52 (17�3%) 38 (12�6%) 29 (9�6%) 9 (3%) 68 (22�6%) 5 (1�6%) 58 (19�3%) 75 (25%) 97 (32�3%) 23 (7�6%) 20 (6�6%) Possessive pronouns and adjectives 73 (24�3%) 63 (21%) 76 (25�3%) 66 (22%) Person and number: - 1 st PS - 2 nd PS - 3 rd PS - 1 st PP - 2 nd PP - 3 rd PP 24 (8%) 13 (4�3%) 10 (3�3%) 29 (9�6%) 32 (10�6) 7 (2�3%) 25 (8�3%) 18 (6%) 8 (2�6%) 23 (7�6%) 17 (5�6%) 3 (1%) 34 (11�3%) 21 (7%) 10 (3�3%) 24 (8%) 11 (4�6%) 4 (1�3%) 20 (6�6%) 1 (0�3%) 8 (2�6%) 32 (10�6%) 31 (10�3%) 5 (1�6%) <?page no="142"?> 142 3 Analysing social web conversations and discourses from a contrastive point of view Demonstrative pronouns and adjectives 27 (9�3%) 20 (6�6%) 35 (11�6%) 29 (9�6%) Demonstrative perspective: - Distal - Proximal - Colloquial 0 (0%) 23 (7�6%) 8 (2�6%) 4 (1�3%) 18 (4�6%) 0 (0%) 12 (4%) 29 (9�6%) 0 (0%) 4 (1�3%) 25 (8�3%) 0 (0%) Table 23. Cohesion by reference and substitution in German, Dutch, Spanish and French Facebook hotel comments Lexical cohesion The results obtained for lexical cohesion do not vary greatly from language to language. The use of this device does not appear to be as common in Facebook as in Booking.com, Expedia and TripAdvisor. The reason may lie in the shorter nature of some Facebook comments. Even so, some sort of reiteration device is present in nearly half the German, Spanish and French comments. In the Dutch corpus, we encounter lexical reiterative mechanisms in just one in three comments. Lexical recurrence is by far the most common resource for achieving lexical texture: it appears in 29.6% and 35% of all German and French comments, respectively. The figures obtained for the Spanish and Dutch corpora are inbetween. A good example of lexical recurrence may be the following German comment from the hotel management (212), where two word items (the Anglicism “Newsletter” and the adjective “exklusiv”) are repeated. Furthermore, a synonymic relation comprising three nouns (another Anglicism, “Promotions”, “Angebote” and “Vorteile”) is readily apparent: (212) Mövenpick Hotel Berlin Melden Sie sich kostenlos zu unserem Newsletter unter www.hofzwei. de an und erhalten Sie einmal im Monat und zu speziellen Promotions exklusive Angebote und Neuigkeiten über das Restaurant Hof zwei. Den Newsletter-Abonnenten werden wir immer wieder exklusive Vorteile zukommen lassen� (GC D Facebook: Mövenpick Hotel Berlin, Berlin) In this respect, synonymy is not as common in Dutch as in the other three languages� This is also true for the occurrence of hyponymic and hyperonymic relations, which in the corpus are more common in Spanish (see example 213) and German than in Dutch and French: <?page no="143"?> 3.2 Text-grammatical structures in consumer comments 143 (213) Hotel Regina Madrid El festival Veranos de la Villa acogerá, a partir del próximo 1 de julio y hasta septiembre, espectáculos de danza, conciertos, teatro y sesiones de cine al aire libre en varios espacios culturales de Madrid� (GC E Facebook: Hotel Regina Madrid, Madrid) However, this type of lexical recurrence does not exceed 9% of the corpus comments in any language� The only other relevant lexical device for our corpus is collocation, which can be identified in 16% and 14.3% of all German and Dutch comments, but only in 9.3% and 8.3% of all Spanish and French ones, respectively. Although collocation, as it is understood here, is not as common in the two Romance languages, it can also link content words in Spanish and French, particularly in comments written by the hotel management, such as the following one. Here, “travail/ travaux” is closely related with “efforts”, as is “restaurant” with “réservation”, “déjeuners” and “diners”: (214) TERRASS HOTEL **** - PARIS Après plus de 6 mois de travaux, beaucoup d’efforts et de travail de la part de toute notre équipe, ENFIN, nous ouvrons les portes de notre restaurant bar du Terrass”! Nous vous accueillerons sur réservation du 23 au 27 juin pour les déjeuners, et à partir du 30 juin pour les déjeuners et les diners! (GC F Facebook: Terrass Hotel Paris, Paris) After a throrough study of lexical cohesion in our corpus, we may conclude, regarding collocation and synonymy, that there are no remarkable differences in the language use of German, Dutch, Spanish and French Facebook users commenting on hotels (see table 24)� Cohesive-device Language German Dutch Spanish French Σ (%) Σ (%) Σ (%) Σ (%) Reiteration 135 (45%) 102 (34%) 138 (46%) 140 (46�6%) Lexical recurrence 89 (29�6%) 93 (31%) 96 (32%) 105 (35%) Synonymy 45 (15%) 22 (7�3%) 44 (14�6%) 58 (19�3%) Hyponymy and hyperonymy 26 (8�6%) 16 (5�3%) 21 (7%) 15 (5%) Meronymy 34 (11�3%) 33 (11%) 40 (13�3%) 38 (12�6%) Antonymy 8 (2�6%) 9 (3%) 7 (2�3%) 12 (4%) Collocation 48 (16%) 43 (14�3%) 28 (9�3%) 25 (8�3%) Table 24. Lexical cohesion in German, Dutch, Spanish and French Facebook hotel comments <?page no="144"?> 144 3 Analysing social web conversations and discourses from a contrastive point of view Morphosyntactic cohesion As with lexical cohesion, morphosyntactic cohesion does not seem to appear in Facebook comments as frequently as in customer comments posted on Booking. com, Expedia or TripAdvisor. In fact, the figures are much lower, which may be partly due to the shorter length of the comments, as well as to the more colloquial and/ or direct nature of Facebook communication. Concerning the syntactic recurrence device, only syntactic parallelisms are significant in the corpus. However, they are present in only 10% to 13.6% in all partial corpora� This small percentage range also means that there do not appear to be any major differences across the compared corpora. The only significant difference may concern the nature of the sentence or phrase structures that are repeated: the structural repetition of noun phrases and assertive sentences is more common in German and Dutch (e.g., in 215), whereas Spanish and French Facebook users seem to have developed a weakness for the reiteration of imperative sentences: (215) Rene Paul Kremer & een overnachting in het art o’tel op 12 juli geef ik mijn ouders cadeau ze worden tenslotte maar 1 x in de 80 jaar - 80 jaar ! .... Etentje & volgende dag lekker lunchen ... Ze lezen dit toch niet want ze zitten niet op FB. (GC NL Facebook: Park Plaza Victoria Hotel Amsterdam, Amsterdam) As for partial morphological recurrence, the empirical results are highly consistent with those obtained for the websites of travel agencies: derivation and compounding are detected only in German and Dutch. By way of example, in (216) derivation is observed in the German noun “(Hotel)klassifizierung” and in the participial adjective “klassifiziert”; what’s more, the noun “Hotel” is present in “Hotelklassifizierung” as Bestimmungswort, or determiner, and in “Sternehotel” as Grundwort, or primary word: (216) Karl Prüfer FROHES NEUES JAHR Im Namen der Deutschen Hotelklassifizierung wünsche ich Ihnen und Ihren Gästen ein frohes neues Jahr 2015 und weiterhin viel Erfolg als klassifiziertes Sternehotel. (GC D Facebook: Berlin Plaza Hotel, Berlin) On the whole, and with the exception of the recurrence linked to word formation, it is safe to say that no major differences are to be found related to morphosyntactic cohesion (see table 25 for all details)� <?page no="145"?> 3.2 Text-grammatical structures in consumer comments 145 Cohesive-device Language German Dutch Spanish French Σ (%) Σ (%) Σ (%) Σ (%) Syntactic recurrence 39 (13%) 42 (14%) 30 (10%) 34 (11�3%) Syntactic parallelism 38 (12�6%) 41 (13�6%) 30 (10%) 34 (11�3%) Chiasmus 2 (0�6%) 1 (0�3%) 0 0 Partial-morphological recurrence 33 (11%) 27 (9%) 2 (0�6%) 5 (1�6%) Derivation 17 (5�6%) 17 (5�6%) 2 (0�6%) 5 (1�6%) Compounding 20 (6�6%) 13 (4�33%) 0 (0%) 0 (0%) Table 25. Morphosyntactic cohesion in German, Dutch, Spanish and French Facebook hotel comments Cohesion by ellipsis Cohesion by ellipsis is slightly more common in the Spanish corpus than in the others. The results appear to correlate with the data obtained for Booking. com, Expedia and TripAdvisor hotel reviews, albeit the percentages are slightly higher in nearly all the languages� Cohesive verbal ellipsis is particularly common in Spanish, in comments posted by both the hotel management and other users, as in (217), where “se mueve” is left out in the second sentence: (217) Hotel Regina Madrid #Madrid se mueve en Verano. Mañana martes, a partir de las 22.00 horas, con un espectáculo de música, danza, luz y velas, para celebrar el solsticio de verano� (GC E Facebook: Hotel Regina Madrid, Madrid) The higher incidence of both cohesive nominal and verbal ellipsis could be attributed to the fact that customer-to-customer or business-to-customer communication tends to be more colloquial on Facebook than on the websites of online travel and hotel booking agencies� <?page no="146"?> 146 3 Analysing social web conversations and discourses from a contrastive point of view Cohesive-device Language German Dutch Spanish French Σ (%) Σ (%) Σ (%) Σ (%) Nominal ellipsis 13 (4�3%) 26 (8�6%) 31 (10�3%) 22 (7�3%) Verbal ellipsis 32 (10�6%) 39 (13%) 49 (16�3%) 35 (11�6%) Table 26. Cohesion by ellipsis in German, Dutch, Spanish and French Facebook hotel comments Cohesion by conjunctive adjunction Conjunctive adjunction is less common in Facebook posts than in Booking. com or Expedia hotel reviews. Most of the conjunctive elements found in the corpus were conjunctions. They are present in slightly more than half of the Spanish comments (50.3%), but only in 38% and 39.3% of the Dutch and German ones� These results are consistent with those reported before for comments on online travel and hotel booking agencies. The use of coordinating conjunctions is particularly high in Spanish (47%) and French (40%). In some Spanish (and also French) comments, coordinating contrasting adjuncts and adjuncts adding information are used side by side, like “y” and “pero” in (218): (218) Handfie ¡El equipo Handfie te espera este fin de semana en el espacio Brico y Deco en Diy Show! Hotel Silken Puerta América Madrid Ven a conocernos, pero sobre todo, vente a los talleres para crear tus proyectos #DIY y llevártelos a casa....¡Te esperamos! (GC E Facebook: Hotel Silken Puerta América Madrid, Madrid) German and Dutch commentators do not seem to resort to coordination as often as the Spanish or the French. As for the subordinating conjunctions, the highest figure is to be found in German (nearly 15% of all German comments include a subordinating adjunct), and the lowest one is surprisingly in Dutch (5%). This could be explained by the very colloquial nature of the Dutch comments posted by hotel managements. In this respect, formal advice tends to be posted in English. Prepositional phrases, adverbs and particles are very scarce in our multilingual corpus� Most of these conjunctive adjuncts have the semantic function of adding information. This type of adjunct is eight percentage points more frequent in <?page no="147"?> 3.2 Text-grammatical structures in consumer comments 147 the Spanish corpus than in the German one (38% vs. 30%). By contrast, the occurrence of contrasting adjuncts in German is four and nearly three percentage points higher than in French and Spanish, respectively (12% vs. 8% and 9.3%). Causal adjuncts, however, are more frequent in Spanish (12.6%) and French (9.6%) than in German (8.3%) and Dutch (only 4%). Although this type of semantic relationship is usually expressed by conjunctions, causal adverbs are also common in the French, German and Dutch corpora. This is the case in the following French comment with “alors”: (219) KUBE Hotel - Paris Ce Dimanche 21 Juin, le #du Brunch KUBE Hotel - Paris est en fête! La Musique, l’Eté et surtout les Papas! Alors profitez de ce jour particulier pour vous retrouver autour du brunch et sur la très conviviale terrasse du Kube ! (GC F Facebook: KUBE Hotel - Paris, Paris) Temporal adjuncts are not common in any of the four corpora analysed. To sum up, the only relevant differences between the four corpora are restricted to the higher occurrence of coordinating conjunctions that add information in Spanish and French. Cohesive-device Language German Dutch Spanish French Σ (%) Σ (%) Σ (%) Σ (%) Parts of speech Conjunctions: coordinating conjunctions subordinating conjunctions 118 (39�3%) 93 (31%) 44 (14�6%) 114 (38%) 102 (34%) 15 (5%) 151 (50�3%) 141 (47%) 32 (10�6%) 130 (43�3%) 120 (40%) 28 (9�3%) Prepositional phrases 9 (3%) 5 (1�6%) 14 (4�6%) 12 (4%) Adverbs 14 (4�6%) 16 (5�3%) 11 (3�6%) 16 (5�3%) Particles 9 (3%) 6 (2%) 0 0 Semantic relationships Adjuncts adding information 90 (30%) 97 (32�3%) 114 (38%) 103 (34�3%) <?page no="148"?> 148 3 Analysing social web conversations and discourses from a contrastive point of view Contrasting adjuncts 36 (12%) 35 (11�6%) 28 (9�3%) 24 (8%) Temporal adjuncts 14 (4�6%) 6 (2%) 10 (3�3%) 11 (3�6%) Causal adjuncts 25 (8�3%) 12 (4%) 38 (12�6%) 29 (9�6%) Table 27. Cohesion by conjunctive adjunction in German, Dutch, Spanish and French Facebook hotel comments Coherence The coherence device is not widely used in all the comments examined. Furthermore, no significant differences could be found between the four partial language corpora. Additive discourse relationships are the most common way of achieving texture without any signalling device� They are present in a similar proportion in German, Dutch, Spanish and French. In this respect, a large number of them are purely additive relationships, such as (220), where three verbal phrases are implicitly connected by coherence without using a conjunction like “et”: (220) Mon Hotel Paris C’est bientôt l’heure des vacances d’été tant attendues! Farniente au soleil, faire le lézard à la plage, randonnez joyeusement en montagne... (GC F Facebook: Mon Hotel Paris, Paris) Causal discourse relations are slightly less common� The only relevant coherence relations here are “cause and reason” and “consequence”. It can sometimes be hard to determine whether such relations are purely causal, or instead imply a consequence. Such an example is (221): the second sentence could also have been linked to a subordinating conjunction like “porque”, which could be interpreted as both the reason and the consequence of the fact that the hotel is the best placed one in Madrid: (221) Mirador Madrid El Hotel Europa, ¿es el hotel mejor situado de Madrid? Te asomas a la Puerta del Sol (GC E Facebook: Hotel Europa Madrid, Madrid) On the whole, as can be inferred from table 28, the corpus has not shown any significant differences regarding the coherence device. <?page no="149"?> 3.2 Text-grammatical structures in consumer comments 149 Coherence- device Language German Dutch Spanish French Σ (%) Σ (%) Σ (%) Σ (%) Additive discourse relations 26-(8.6%) 22-(7.3%) 28-(9.3%) 33-(11%) Addition 22 (7�3%) 16 (5�3%) 24 (8%) 26 (8�6%) Contrast 9 (3%) 11 (3�6%) 5 (1�6%) 8 (2�6%) Disjunction 3 (1%) 2 (0�6%) 1 (0�3%) 3 (1%) Causal discourse relations 18-(6%) 15-(5%) 20-(6.6%) 22-(7.3%) Cause and reason 5 (1�6%) 3 (1%) 6 (2%) 3 (1%) Means 2 (0�6%) 4 (1�3%) 2 (0�6%) 3 (1%) Consequence 11 (3�6%) 8 (2�6%) 11 (3�6%) 10 (3�3%) Purpose 2 (0�6%) 2 (0�6%) 5 (1�6%) 4 (1�3%) Condition 1 (0�3%) 1 (0�3%) 2 (0�6%) 3 (1%) Concession 2 (0�6%) 1 (0�3%) 2 (0�6%) 1 (0�3%) Table 28. Coherence in German, Dutch, Spanish and French Facebook hotel comments Deixis Deixis does not appear to be as common in Facebook hotel comments as in customer comments on online travel and hotel booking agencies. However, it is still a widely used texturing device, particularly person deixis. In this respect, and according to our corpus, there seems to be considerable variation between the four corpora� Whereas this type of deixis has a very similar incidence in French and German (in both cases well above 60%), it is detected in only 53% of all Spanish comments and 47.6% of all Dutch ones. The first person is more common in French (53%), German (45.3%) and Dutch (41.3%) than in Spanish. This could be explained by the preferential use of the first person plural in the German and French comments posted by hotel managements, as in (222), where the possessive adjectives “unsere” and “unseren” refer back deictically to the “Andel’s Hotel Berlin”: <?page no="150"?> 150 3 Analysing social web conversations and discourses from a contrastive point of view (222) Andel’s Hotel Berlin Nur 3 Stunden und 12 Minuten von Nessi entfernt begegnen unsere Verkaufsleiterin Monika und Vertriebsrepräsentantin Zuzana aus unseren VI Schwesterhotels in Prag “people from Scotland and the world”. (GC D Facebook: Andel’s Hotel Berlin, Berlin) In comments posted by visitors, the first person singular is more common in all four languages. Although the familiar form of the second person is very widely used in all German, Dutch and Spanish comments, it rarely reveals a deictic or deictic and referential use: it is detected in only 23% of all Spanish comments and in 17% of all Dutch ones. The figure for German is even lower (14%), and insignificant for French. In return, a deictic polite form is frequent only in French corpus comments (22.6%). It is often used by both Facebook users and the hotel management to address one another. As a way of example, in (223) the hotel management directly addresses potential clients by referring to them with the oblique pronoun “vous”: (223) KUBE Hotel - Paris Un petit coup de chaud à Paris? On vous attend dans notre Ice Bar pour faire redescendre la température! (GC F Facebook: Kube Hotel - Paris, Paris) There are no major differences in the deictic use of the third person, which is slightly more usual in Spanish hotel comments posted by management. In (224), “su” refers back deictically to the restaurant “Tatel Madrid”: (224) Hotel Único Madrid Tatel Madrid Coctelería y multiespacio gastro con cocina de Nino Redruello y Patxi Zumárraga por lo que no falta boqueroncitos limpios al ajo y fritos al revés o la tortilla trufada. Carta de picoteo en su zona lounge así como una pequeña tienda gourmet (GC E Facebook: Hotel Único Madrid, Madrid) Place deixis, which can be realized by the use of demonstrative pronouns and demonstrative adjectives, demonstrative adverbs and prepositional and noun phrases, is relatively common in our corpus. It is observed in a similar frequency in all four languages, ranging from 37% in Spanish to 29.6% in Dutch. The most usual place relators are not the expected demonstrative pronouns and adjectives, closely linked to the cohesion device, but prepositional phrases and nouns introducing a geographic location, as in example (223) “Un petit coup de chaud à Paris ? ”. <?page no="151"?> 3.2 Text-grammatical structures in consumer comments 151 As for place relations, there were no big differences: most of them are neutral ones, and refer either to the concrete location (as in 223) or to the hotel itself, as in the following Spanish comment: (225) Recámara Producciones Entrevistando para Sol Música al grupo londinense Years & Years en el Hotel Silken Puerta América Madrid � (GC E Facebook: Hotel Silken Puerta América Madrid, Madrid) Proximal relations seem to be slightly more common in German than in the other three languages, and tend to be realized by adverbs, as in (226): (226) Mövenpick Hotel Berlin Unser Tipp: Am 3. und 4. Juli findet der “Rotary-Kindertag” im Berliner Zoo statt und unser Küchenchef Arne Buschow ist hier auch mit dabei� (GC D Facebook: Mövenpick Hotel Berlin, Berlin) Summing up, the results obtained for the deixis device (which are detailed in table 29) show some important differences between the languages analyzed, particularly concerning the second person deixis. In this respect, and as in European Spanish, German (and also Dutch), Facebook users seem to have given up the rigid du-Sie distinction they would uphold in other communicative situations. French users, however, still honour the formal manner of address on Facebook. Deixis device Language German Dutch Spanish French Σ (%) Σ (%) Σ (%) Σ (%) Person deixis -187-(62.3%) -143-(47.6%) -159-(53%) -194-(64.6%) First person 136 (45�3%) 124 (41�3%) 108 (36%) 159 (53%) Second person (or familiar form) 42 (14%) 51 (17%) 69 (23%) 9 (3%) Second person (or polite form) 18 (6%) 5 (1�6%) 7 (2�3%) 68 (22�6%) Third person 63 (21%) 59 (19�6%) 72 (24%) 63 (21%) Place deixis 104 (34�6%) 89 (29�6%) 111 (37%) 98 (32�6%) Place relators <?page no="152"?> 152 3 Analysing social web conversations and discourses from a contrastive point of view Demonstrative pronouns and adjectives 31 (10�3%) 22 (7�3%) 41 (13�6%) 29 (9�6%) Demonstrative adverbs 15 (5%) 8 (2�6%) 10 (3�3%) 12 (4%) Prepositional phrases and NP 78 (26%) 71(23�6%) 87 (29%) 80 (26�6%) Place relations Distal 6 (2%) 8 (2�6%) 12 (4%) 4 (1�3%) Proximal 41 (13�6%) 29 (9�6%) 29 (9�6%) 25 (8�3%) Neutral 64 (21�3%) 55 (18�3%) 81 (27%) 72 (24%) Table 29. Deixis in German, Dutch, Spanish and French Facebook hotel comments 3.2.3 Text-grammatical structures in comments on video-sharing platforms and Wiki discussion pages As already stated in the description of the subtext genre “comments in video sharing platforms, Wiki discussion pages and blogs”, Internet users do not seem to comment on hotel stays or hotels in general on social web sites like YouTube or Wikipedia. The few comments that could be collected, and which build up the Exemplary Corpus, are not enough to draw any valid conclusions regarding the text-grammatical structures employed in the corpus� This section will therefore simply seek to provide some very brief and general descriptions of the most relevant text-grammatical structures that can be found in comments posted on video-sharing platforms and Wiki discussion pages� The most widely used texture device in the few YouTube comments is undoubtedly the grammatical cohesive device of reference. Due to the very colloquial nature of YouTube comments, using a pronoun in the second person singular of the familiar form is by far the most usual way of addressing and referring to fellow commentators. This is not only true for Spanish, Dutch and German comments: even in the very few French postings, the familiar form of the second person outweighs the courtesy form. By way of example, in (227) the user AbejaCojonera refers back to the person that has uploaded the video and is also taking part in the conversation using the verbal inflection in “te habías hospedado”: <?page no="153"?> 3.2 Text-grammatical structures in consumer comments 153 (227) #OMG jajajaja por el nombre del hotel pensaba que te habías hospedado en París (porque uno de los dos aeropuertos que tiene se llama Orly). xDD Está chulo el baño. Tiene un estilo atemporal (que no pasa de moda y no cansa)� LAS VISTAS SON PRECIOSAS =D AbejaCojonera hace 2 años (GC E YouTube: TRYP San Sebastian Orly Hotel) In contrast, lexical cohesion, which can be found in a relatively high percentage of Facebook, Booking.com, Expedia and TripAdvisor comments, is extremely rare on YouTube. On Wikipedia discussion pages, both lexical cohesion and cohesion by reference are also very rare� We may encounter certain reiteration devices such as lexical recurrence, hyponymy/ hyperonymy and meronymy in articles related to hotels, but not on the discussion or Talk pages. The cohesion by conjunctive adjunction on YouTube and on Wiki discussion pages is restricted to some (few) coordination conjunctions that tend to be either additive or contrastive adjuncts. As for the coherence device, both additive and causal discourse relations seem to be relatively common on YouTube, possibly even more common than in Facebook hotel comments� Coherence is achieved here mainly by addition and by cause and reason, as in the following, where a conjunction such as “denn” could be introduced between the false interrogative clause “warum nur der komische Name” and the copulative sentence “marketingtechnisch wäre (...)”: (228) Warum nur der komische Name, marketingtechnisch wäre ein englischklingender Name angebrachter IMHO. Andel’s klingt nach Christian Anders (Bruder vom Modern Talking-Duo) und der hat ja ziemlich unästhetische Sachen gemacht� welpen2006 vor 1 Jahr (GC D Youtube: andel’s Hotel Berlin) On the whole, no real differences between the four languages could be detected. This may of course be due to the lack of sufficient data. In any case, a through text-grammatical analysis of hotel comments on YouTube and on Wiki discussion pages cannot be successfully carried out with such scant data. It seems that both YouTube and Wikipedia are not the channels hotel customers and internet users employ to talk about hotels and hotel stays� <?page no="155"?> 4 Conclusions This work has presented a practical text-linguistic approach to the contrastive analysis of online customer comments that should be understood as a multidisciplinary attempt to contribute to marketing studies from a linguistic point of view. The main aim of the proposed contrastive model was to provide efficient working tools for analysing the linguistic practices in online customer comments, particularly in hotel reviews, in different social web applications. The analytical method that has been explained in-depth in this monograph will hopefully enable linguists and experts in international marketing to evaluate and compare product and service customer comments on the social web written in two or more European languages. The proposed contrastive model has been successfully tested on two relevant text genres for the tourism industry (comments on online travel and hotel booking websites and comments on social networking services) in four languages: German, Dutch, Spanish and French. Two levels of analysis have been used� On what has been called the macrocommunicative level, our approach has focused on analysing the textual intentions and actions that can be found in customer comments. As for the text functions, a well-established and productive framework such as Brinker’s has been adapted to the contrastive needs arising in the study of hotel reviews on the social web. In a second analytical step, a differentiated study of the text actions underpinning text functions has been presented� This procedure will hopefully help interested text researchers to determine and explain the textual communicative structure of online hotel reviews and other computer-mediated texts on the social web� On a purely text-grammatical level, those cohesion and coherence devices that are relevant to a contrastive study of customer comments have been detected and examined. As a result, it is imperative to conduct a text-grammatical analysis of referential, lexical, morphosyntactic and conjunctive cohesion, as well as of certain coherence and deictic devices� The proposed analytical method has been tested on a multilingual corpus comprising 2000 hotel comments and reviews retrieved from online travel and booking sites, social networking services, and other social web applications. The application of our text-linguistic working tool has allowed us to draw some interesting conclusions about the communicative macrostructure and the textgrammatical structures in German, Dutch, Spanish and French hotel comments and reviews� <?page no="156"?> 156 4 Conclusions Where communicative functions and text actions on online travel and hotel booking websites are concerned, no relevant differences could be found for most text actions and functions. The only striking differences detected concern the mainly appellative or informative text actions “describing breakfast choices”, “commenting on quietness and privacy” (both occur frequently in the German and Dutch corpora), “recommending or discouraging a stay at a given hotel” (more frequent in Dutch and French hotel reviews), and “indicating parking availability or commenting on parking-related problems” (more common in the Spanish corpus). The differences are greater on the social networking service Facebook: “asking for information about the hotel” is more common among the French and Spanish, “giving some piece of information about the hotel” is only commonplace in German and French reviews, and the text action “commenting on news or events that will take place or have taken place outside the hotel” is more frequent in German, Dutch and Spanish (while “commenting on events that will take place or have taken place at the hotel” is very usual in the French corpus). Facebook is not used in any one of the four languages to assess hotel stays or give hotel recommendations� As for text-grammatical structures, the reference device is very commonly used in all four corpora in both hotel reviews and comments retrieved from online travel and hotel booking websites and from Facebook. In hotel reviews on online travel and hotel booking websites, the main differences between the corpora are related to the higher occurrence of a referential impersonal pronoun or impersonally used pronoun, and to the lower incidence of first person pronouns with a reference function in Spanish for grammatical reasons� The main difference on Facebook involves the forms of address: the pronominal reference of the second person is expressed mostly in the familiar form in all corpora except the French one, where the polite form is the only common one. These results are also consistent with those obtained for the deixis device: only French users still uphold the formal manner of address on Facebook. Nevertheless, the person deixis appears to be realized in a different way in the corpus: apart from the obvious differences concerning the second person, a deictic first one is more common in German and French than in Spanish and Dutch, while the third person deixis only has a relatively high incidence in Spanish. Although cohesion by conjunctive adjunction is very common in hotel reviews and comments retrieved from online travel and hotel booking websites in all four languages, two divergent features have been detected: firstly, the rate at which conjunctions are used as connective elements is much higher in the Spanish corpus than in the other three, and secondly, German and Spanish customers operate with three times more subordinating conjunctions than their French and Dutch counter- <?page no="157"?> 4 Conclusions 157 parts. On Facebook, the most significant difference concerns the higher occurrence of coordinating conjunctions that add information in Spanish and French. The results obtained for lexical and morphosyntactic cohesion, as well as for coherence devices, do not vary greatly from one language to another. Despite the obvious limitations of the working tools used in this study, these empirical results show that the proposed approach is operable for determining some of the most relevant text-linguistic structures of customer comments on the social web� <?page no="159"?> 5 References Adamzik, Kirsten (2001): Kontrastive Textologie. Tübingen: Stauffenburg. Adamzik, Kirsten (2004): Textlinguistik. Eine einführende Darstellung. Tübingen: Niemeyer. Baker, Mark (2003): Lexical categories: verbs, nouns and adjectives. Cambridge: CUP. Beerbom, Christiane (1992): Modalpartikeln als Übersetzungsproblem. Eine kontrastive Studie zum Sprachenpaar Deutsch - Spanisch. Frankfurt: Peter Lang. Bendel, Sylvia (1998): Werbeanzeigen von 1622-1798. Entstehung und Entwicklung einer Textsorte. Tübingen: Niemeyer. Böttger, Claudia/ Probst, Julia (2001): Adressatenorientierung in englischen und deutschen Texten� Hamburg: Sonderforschungsbereich 538� Brinker, Klaus (1983): Textfunktionen. Ansätze zu ihrer Beschreibung. In: Zeitschrift für Germanistische Linguistik 11/ 83, 127-148. Brinker, Klaus (1985, 1992, 2005): Linguistische Textanalyse. Eine Einführung in Grundbegriffe und Methoden. Berlin: ESV. Brown, Gillian / Yule, George (1983): Discourse analysis. Cambridge: CUP. Bühler, Karl (1982): Sprachtheorie. Die Darstellungsfunktion der Sprache. Stuttgart: UTB. First published in 1934. Busse, Dietrich (2009): Semantik. Paderborn: Fink. Butler, Cristopher (2003): Structure and function: a guide to three major structuralfunctional theories. Amsterdam: John Benjamins. Camacho Adarve, María Matilde (2009): Análisis del discurso y repetición: palabras, actitudes y sentimientos. Madrid: Arco Libros. Campbell, Kim (1995): Coherence, continuity and cohesion: theoretical foundations for document design. New York: Routledge. Caro Cedillo, Ana (2004): Fachsprachliche Kollokationen: ein übersetzungsorientiertes Datenbankmodel Deutsch-Spanisch. Tübingen: Narr. Carrell, Patricia (1982): Cohesion is not Coherence. In: TESOL Quarterly 16/ 4, 479-488. Chang, William et al. (2010): Transforming enterprise cloud services. Doordrecht: Springer� Copeland, Lennie/ Griggs, Lewis (1985): Going International. New York: Random House� Coseriu, Eugenio (1980): Textlinguistik. Eine Einführung. Tübingen: Narr. De Beaugrande, Robert (1980): Text, discourse and process: Toward a multidisciplinary science of texts. London: Longman. De Beaugrande, Robert/ Dressler, Ulrich (1981): Introduction to text linguistics. London: Longman. Douyère, Christel/ Sosthé, Franck (2014): E-Reputation Management and Strategic Business Development Using Web 2.0 Tools: The Case of the Hotel Industry. In: Mariani, <?page no="160"?> 160 5 References Marcello et al. (eds.): Tourism Management, Marketing, and Development: The Importance of Networks and ICTs. New York: Palmgrave, 99-112. Ebel, Bernd (2007): Kompakt-Training. E-Business. Ludwigshafen: Kiehl Verlag. Eggler, Marcel (2006): Argumentationsanalyse textlinguistisch. Tübingen: Niemeyer. Eisenlauer, Volker (2013): The true colours of Facebook. London: Blommsbury. Esser, Jürgen (2009): Introduction to English text-linguistics. Frankfurt: Peter Lang. Fernández, Jocelyne (1994): Les particules énonciatives. Paris: PUF. Fillmore, Charles (1975): Santa Cruz lectures on deixis. Bloomington: Indiana University Linguistics Club. Fix, Ulla (2006): Was heißt Texte kulturell verstehen? Ein- und Zuordnungsprozesse beim Verstehen von Texten als kulturellen Entitäten. In: Blüdorn, Hardarik et al. (eds.): Text - Verstehen: Grammatik und darüber hinaus. Berlin: De Gruyter, 254-276� Fulcher, Glenn (1989): Cohesion and coherence in theory and reading research. In: Journal of Research in Reading 12/ 2, 146-163. Gaddy, Michelle et al. (2001): The influence of text cues on the allocation of attention during reading. In: Sanders, Ted et al. (eds.): Text Representation: Linguistic and Psycholinguistics Aspects. Amsterdam: John Benjamins, 89-110. Gansel, Christina/ Jürgens, Frank (2002): Textlinguistik und Textgrammatik. Wiesbaden: Westdeutscher Verlag. Golonka, Joanna (2009): Werbung und Werte: Mittel ihrer Versprachlichung im Deutschen und im Polnischen. Wiesbaden: VS Verlag. Große, Ernst Ulrich (1976): Text und Kommunikation: eine linguistische Einführung in die Funktionen der Texte. Stuttgart: Kohlhammer. Gülich, Elisabeth/ Raible,Wolfgang (1975): Textsortenprobleme. In: Moser, Hugo (ed.): Linguistische Probleme der Textanalyse. Düsseldorf: Schwann, 144-197. Gunter, Barrie (1987): Poor Reception: Misunderstanding and Forgetting Broadcast News. Hillsdale: Erlbaum. Hall, Edward (1976). Beyond culture. New York: Anchor Books. Halliday, Michael/ Hasan, Ruqayia (1976): Cohesion in English. London: Longman. Halliday, Mchael/ Hasan, Ruqayia (1989): Language, Context and Text: a social semiotic perspective� Oxford: OUP� Jakobson, Roman (1960): Linguistics and Poetics. In: Sebeok, Thomas (ed.): Style in Language. Cambridge: M.I.T. Press, 350-377. Janich, Nina (2005): Werbesprache. Ein Arbeitsbuch. Tübingen: Narr. Kallmeyer, Werner et al. (1974): Lektürekolleg zur Textlinguistik. Frankfurt: Athenäum. Kozminsky, Eh (1977): Altering Comprehension: the effects of biasing titles on text comprehension. Memory and Cognition 5, 482-490. Krein-Kühle, Monika (2002): Cohesion and Coherence in Technical Translation: the Case of Demonstrative Reference. In: Van Vaerenbergh, Leona (ed.): Linguistics and Translation Studies. Antwerpen: Hogeschool Antwerpen, 41-53. Levinson, Stephen (2003): Pragmatics. Cambridge: CUP. Longacre, Robert (1976): An Anatomy of Speech Notions. Lisse: De Ridder. <?page no="161"?> 5 References 161 Lyons, John (1977): Semantics. Cambridge: CUP. Lüger, Heinz-Helmut/ Lenk, Hartmut (2008): Kontrastive Medienlinguistik. Ansätze, Ziele, Analysen. In: Lüger, Heinz-Helmut/ Lenk, Harmut (eds.): Kontrastive Medienlinguistik. Landau: Verlag Empirische Pädagogik, 11-28. Mann, William/ Thompson, Sandra (1985): Assertions from discourse structure. Marina del Rey: Information Sciences Institute. Mann, William/ Thompson, Sandra (1987): Rhetorical structure theory: A theory of text organization. Marina del Rey: Information Sciences Institute. Martínez-Cabeza, Miguel Ángel (2002): The study of language beyond the sentence. From text grammar to discourse analysis. Granada: Comares. Meibauer, Jörg (2008): Pragmatik. Tübingen: Stauffenburg. Nord, Christiane (2005): Text Analysis in Translation: Theory, Methodology, and Didactic Application. Amsterdam: Rhodopi. Polenz, Peter von (1980): Möglichkeiten satzsemantischer Textanalyse. In: Zeitschrift für Germanistische Linguistik 8, 133-153. Renkema, Jan (1993): Discourse Studies: An Introductory Textbook. Amsterdam: John Benjamins Renkema, Jan (2004): Introduction to discourse studies. Amsterdam: John Benjamins. Rickheit, Gert / Schade, Ulrich (2000): Kohärenz und Kohäsion. In: Brinker et al. (eds.): Text und Gesprächslinguistik. Ein internationales Handbuch zeitgenössischer Forschung. Berlin: De Gruyter, 275-283. Rieser, Johannes (1978): On the development of text grammar. In: Eckart, Rolf (ed.): Die Funktionen der Gebrauchstextsorten. Berlin: De Gruyter, 6-20. Römer, Ruth (1968): Die Sprache der Anzeigenwerbung. Düsseldorf: Schwann. Rudolf, Elisabeth (1988): Connective relations - connective expressions - connective structures. In: Petöfi, Janos (ed.): Text and discourse constitution. Berlin: De Gruyter, 97-133� Saeed, John (2009): Semantics. Oxford: Wiley-Blackwell. Sánchez Prieto, Raúl (2011): Unternehmenswebseiten kontrastiv. Eine sprachwissenschaftlich motivierte und praxisorientierte Vorgehensweise für eine kontrastive Analyse deutscher, spanischer und französischer Unternehmenswebseiten.Unternehmenswebseiten kontrastiv. Tübingen: Narr. Sánchez Prieto, Raúl (2015): Wie kommentieren Kunden (Fernseh-)Werbung im Netz? Zu einer deutsch-romanischen Diskursanalyse der Kommentarfunktion auf Video- Sharing-Plattformen. In: Lebsanft, Franz/ Schrott, Angela (eds.): Diskurse, Texte, Traditionen: Modelle und Fachkulturen in der Diskussion. Göttingen: Vandenhoeck & Ruprecht, 289-306. Sánchez Prieto, Raúl (2016): Marca España: Titulares de artículos periodísticos y encabezamientos de comentarios sobre ellos en la prensa online española, alemana y holandesa. Technical report for the research Project “Talking about Spain”. Unpublished work� Sasse, Hans-Jürgen (2015): Syntactic Categories and Subcategories. In: Jacobs, Joachim/ Wiegand, Herbert Ernst: Syntax - Theory and Analysis. Vol. I. Berlin: De Gruyter. <?page no="162"?> 162 5 References Schriffrin, Deborah (2003): Discourse markers: Language, meaning and context. In: Schriffrin, Deborah et al. (eds.): The handbook of discourse analysis. London: Blackwell, 54-75. Schnotz, Wolfgang (1994): Aufbau von Wissensstrukturen. Untersuchungen zur Kohärenzbildung bei Wissenserwerb mit Texten. Weinheim: Beltz. Schröder, Thomas (2003): Die Handlungsstruktur von Texten. Ein integrativer Beitrag zur Texttheorie. Tübingen: Narr. Schütte, Daniela (2004): Homepages im World Wide Web. Eine interlinguale Untersuchung zur Textualität in einem globalen Medium. Frankfurt: Peter Lang. Searle, John (1975): A Taxonomy of Illocutionary Acts. In: Günderson, Keith (ed.): Language, Mind, and Knowledge. Minneapolis: Minneapolis University Press, 344-369. Spillner, Bernd (1981): Textsorten im Sprachvergleich. Ansätze zu einer Kontrastiven Textologie. In: Kühlwein, Wolfgang et al. (eds.): Kontrastive Linguistik und Übersetzungswissenschaft. München: Fink, 239-250. Stahlheber, Eva (1992): Die Fachtextsorte Zeitschriftenartikel im Deutschen und Adress/ Article im Amerikanischen: Popularisierungsgrad und Diachronie von Funktionen und Strukturen. In: Baumann, Klaus-Dieter (ed.): Kontrastive Fachsprachenforschung. Tübingen: Narr, 162-189. Steinbach, Markus (2007): Semantik. In: Meibauer, Jörg (ed.): Einführung in die germanistische Linguistik. Stuttgart: Metzler, 183-206. Storrer, Angelika (2004a): Text und Hypertext. Mit einem Exkurs zu XLink und XPointer von Eva Anna Lenz. In: Lemnitzer, Lothar/ Lobin, Hennig (eds.): Texttechnologie. Tübingen: Stauffenburg, 13-50. Storrer, Angelika (2004b): Kohärenz in Hypertexten. In: Zeitschrift für germanistische Linguistik 31/ 2, 274-292. Taboada, Maite (2009): Implicit and explicit coherence relations. In: Renkema, Jan (ed.): Discourse, of course. Amsterdam: John Benjamins, 125-138. Taboada, Maite (2013): Explicit and implicit coherence relations: a corpus study. In: Shan, Luo (ed.): Proceedings of the 2013 annual conference of the Canadian Linguistic Association. Toronto: Canadian Linguistic Association, 1-15. Tanskanen, Sanna-Kaisa (2006): Collaborating towards coherence. Lexical Cohesion in English Discourse. Amsterdam: John Benjamins. Van der Eijk, Pirn (1994): Comparative discourse analysis of parallel texts. Technical report cmp-lg/ 9707022. Digital Equipment Corporation. Van Dijk, Teun (1977): Text and Context: Explorations in the Semantics and Pragmatics of Discourse. London: Longman. Van Dijk (1980): Textwissenschaft: Eine interdisziplinäre Einführung. München: Dtv. Van Dijk, Teun (1985): Structures of news in the press. In: Van Dijk, Teun (ed.): Discourse and Communication. Berlin: De Gruyter, 69-93. Van Dijk, Teun/ Kintsch, Walter (1983): Strategies of discourse comprehension. New York: Academic Press. Vater, Heinz (1994, 2001): Einführung in die Textlinguistik: Struktur, Thema und Referenz in Texten. München: Fink. <?page no="163"?> 5 References 163 Von der Lage-Müller, Kathrin (1995): Text und Tod. Eine handlungstheoretisch orientierte Textsortenbeschreibung am Beispiel der Todesanzeige in der deutschsprachigen Schweiz. Tübingen: Niemeyer. Von Stutterheim, Christiane (1997): Einige Prinzipien des Textaufbaus. Empirische Untersuchungen zur Produktion mündlicher Texte. Tübingen: Niemeyer. Vujević, Vera (2012): Ellipsis and substitution as cohesion devices. In: Radovi filosofskog fakulteta 13/ 1, 407-416. Warren, Martin (2009): Cohesive chains and speakers’ choice of prominence. In: Flowerdew, John/ Mahlberg, Michaela (eds.): Lexical cohesion and corpus linguistics. Amsterdam: John Benjamins, 45-64. Wehrlich, Egon (1976): A text grammar of English. Heidelberg: Quelle und Mayer. Werth, Paul (1999): Text worlds: representing conceptual space in discourse. London: Longman. Willkop, Eva-Maria (2001): Linguistische Analyseverfahren von Texten. In: Helbig, Gerhard et al. (eds.): HSK Deutsch als Fremdsprache. Berlin: de Gruyter, 314-323. Zidrasco, Tatiana et al. (2010): Building and analyzing corpus to investigate appropriateness of argumentative discourse structure for facilitating consensus. In: García Pedrajas, Nicolás et al. (eds.): Trend in applied intelligent systems. Berlin: Springer, 575-584�
