Power and status by definition Preliminary theoretical considerations on a study of defining as a metadiscursive strategy in lay-to-lay interaction Georg Marko This paper outlines the theoretical framework for a study of definitions as metadiscursive moves and their social functions and effects in discussion forums on health. Defining a lexical element is metadiscourse because a lexical element used elsewhere in the current speech event is defined, i.e. its meaning is provided. Definitions imply that the speaker or writer knows the lexical element’s meaning and at the same time that the other participants in the interaction do not (otherwise there would be no need for a definition). This creates a gap in knowledge (however real and however temporary) and thus a hierarchy and possible power imbalance between interlocutors. To contextualize the main issue, the article starts by discussing essential features of the phenomenon, also pointing to their relevance for definitions. The article then goes on to describe various features of definitions. These include questions such as whether the definiendum (= the term to be defined) occurs before the definiens (= the meaning), whether the definiens is elaborate or just an alternative term, whether the definiens defines the term exhaustively and unambiguously, generically or with reference to individual events or experiences, etc. I also evaluate these different features with respect to whether they could be argued to enhance or mitigate the social effects mentioned above. 1. Introduction (1) Meds like sertraline aka zoloft (a SSRI) can help with migraine. (2) With your vision problems, a VER / VEP (Visual Evoked Potentials) which is a painless test to test the optic nerves, would be routine too. AAA - Arbeiten aus Anglistik und Amerikanistik Band 42 (2017) · Heft 2 Gunter Narr Verlag Tübingen Georg Marko 244 (3) I have elevated CRP levels, which is C-Reactive Protein, another word for inflammation. These three examples, all taken from lay discussion forums on health, have something in common: they contain definitions, i.e. the posters use terms whose meanings they immediately spell out and explain (in all three sentences this is even done in two different ways, viz. by giving the meaning and an alternative expression). Definitions are metadiscursive (being about language used in the same discourse) and play a crucial role in interactively negotiating claims to knowledge and expertise and thus indirectly relations of power in an environment - a lay discussion forum - which does not predetermine these aspects. This article will examine the metadiscursiveness of definitions and their social interactive functionality with special emphasis on their role in Internet forums on health. The approach taken here is abstract in nature, outlining how the question of metadiscourse and social function more generally and defining technical terms and the negotiation of power more specifically can and should be researched. The paper, however, will not present a comprehensive study of the extent, the forms and roles of the definitions in the discourse under scrutiny. The emphasis here is on a theoretical discussion of metadiscourse and its essential features (laying the foundation) and a description of the main categorial variables of definitions as metadiscourse and their relevance for a socially-oriented study of the use of definitions. 2. Data As mentioned, this article outlines the background of and the approach to a potential study of the use of definitions in lay health forums rather than presenting the details and results of such a study. However, as this requires examples for illustrating the analytical categories to be introduced below, I nevertheless need data even at this stage. Since I am generally interested in the structural and functional patterns in which definitions are employed by participants in forum interactions and thus in recurrence across individual texts, I will take a corpus analytical approach. My data therefore needs to be a corpus, i.e. a large computer-held collection of texts. My main interest is metadiscourse in health forums, where the difference between lay vs. expert plays a particular important role. So the corpus consists of postings to such forums. I chose four conditions, viz. cardiovascular diseases (CVDs), multiple sclerosis (MS), headaches and migraines, and depressions, in order to cover a wide range of different topics on health. Then I selected three platforms for each of these four conditions (to preserve some anonymity Power and status by definition 245 to posters, I will not name these) and copied and pasted all postings from all threads till I reached a certain target number of words. Table 1 below contains the sizes of the whole corpus and its components. The four topics are not represented equally, quantitatively speaking, because the components have originally been produced for different projects (cf. Marko 2010, Kettemann/ Marko/ Triebl 2010, Marko 2012) (that is also the reason for different word number targets). Overall CVD MS Headaches Depression Number of words 2,309,812 530,968 396,343 830,289 552,212 Table 1: Size of the health forum corpus (and its components) (in number of word tokens). 3. Metadiscourse: conceptions, structures and functions Metadiscourse has received a lot of attention in linguistics in the past 30 years and has been studied from different angles. What becomes clear from this research - and readers can also find evidence of this in the present volume - is that metadiscourse is a concept difficult to define in a way that would make its identification in authentic data easy and straightforward. I will therefore spend a considerable time describing what I mean by it (section 3.1) and what its functions could be (3.2), also explaining what the aspects mentioned mean for definitions as metadiscursive moves. 3.1 Conceptions of metadiscourse The basic definition of metadiscourse - “discourse about the evolving [or current] discourse, or the writer’s [or speaker’s] explicit commentary on her own ongoing text” (cf. Ädel 2006: 2) - serves as a starting point. But we need to specify the necessary features of metadiscourse in order to really be able to use the concept and distinguish it from related ones. These features are concerned with: ‣ Explicit metalinguistics ‣ Relationship between metadiscourse and discourse ‣ Relationship between metadiscourse and propositional level ‣ Intratextuality ‣ Mode ‣ Fuzziness Georg Marko 246 3.1.1 The structure of metadiscourse According to Ädel (2006: 22-23), metadiscourse is a functional and not a formal category. This view is supported by the fact that while there are meaning elements that are defining and necessary features of metadiscourse, e.g. a reference to an aspect of the ongoing discourse, there is some variation in how metadiscourse is formally realized, i.e. which linguistic constructions can be used for this purpose. While this is certainly true, we need to bear in mind that functional is concerned with semantics and meaning here rather than with any social functions of metadiscourse - as will be pointed out below, metadiscourse is not coherent and consistent with respect to these functions. Besides, although there is some formal variation in the realization of the said functional categories, the fact that the latter need to be realized formally and are not completely unpredictable, i.e. there is a limited set of options, suggests that we may also speak of structural-functional categories. What are now these structural-functional categories required in metadiscourse? I stipulate that metadiscourse necessarily includes three essential components, viz. the discourse referent, the metadiscursive reference, and the metadiscursive comment. 1. Discourse referent The linguistic element or structure in the ongoing discourse that I refer to or comment on. So if a thread in a discussion forum starts with a question and someone refers to it (as question), then the question becomes the discourse referent. Anything linguistic can become a discourse referent, whether individual phonemes, grammatical structures, words, rhetorical strategies, or interactive behaviour. 2. Metadiscursive reference The metalinguistic expression 1 that creates an explicit link to the discursive referent. In the above example, using the word question, for instance, is the metadiscursive reference. To be precise, the metadiscursive reference covers two aspects, viz. a metalinguistic expression and a discourse deictic relation, i.e. the knowledge that the metalinguistic expression refers to the ongoing speech event. This may be explicated by using demonstrative pronouns, e.g. this question, possessive pronouns, e.g. your question, or spatial or temporal deictic expressions, e.g. the question that you asked above/ earlier, etc. In some cases, however, the metalinguistic dimension is implied, e.g. when referring, and thus somehow citing, elements - whether words or longer constructions - used in the current 1 As linguistic could mean ‘pertaining to language’ and ‘pertaining to the study of language’, metalinguistic and linguistic in many contexts are interchangeable. Power and status by definition 247 discourse, e.g. How are you does not really sound nice. People often use typographical devices (italics, underlined, inverted commas, quotation marks, etc. 2 ) to mark the element as metalinguistic and to make it clear that they refer to it, but do not use it. 3. Metadiscursive comment What I am saying about the discursive referent. If someone, for instance, says I don’t like your question, then the metadiscursive comment is that the writer does not like something occurring in the same discourse, viz. the question that the participant directly addressed has asked. Below, you find a coherent example of a short exchange in a health forum which illustrates the above categories. What poster A says in her post by itself is not metadiscursive because she is just describing her experience of certain neurological-psychological symptoms. However, the metadiscursive reference what you described (shaded) used by speaker B turns A’s posting into a discursive referent (underlined), i.e. something that (in this case) someone else chooses to talk about, saying that the symptoms A described sound like a panic attack, which then is the metadiscursive comment (in bold). (4) A: […] Last nite it felt like I was dying. I couldnt sleep and when I did close my eyes for a few minutes it felt like my right side was numb. when I got up it lasted for 2 minutes, then went away. […] B: […] I’ve had panic attacks all my life and what you described sounds EX- ACTLY like one. I will now discuss the six features mentioned above. 3.1.2 Explicit metalinguistics Discourse in ‘about discourse’ in the above definition is a linguistic unit. It is language in use or a particular instance of language in use (a speech event). Speaking about discourse thus means speaking about language, as used in the current speech event. Metadiscourse thus is necessarily metalinguistic, i.e. we are not talking about discourse as sets of emotions evoked, a systems of ideas standing behind a text, or the participants’ identities, but only as discourse qua language or language use. This metalinguistic dimension should be explicit. Emotions, systems of ideas, and identities are often associated with particular passages in texts and discourses without being the explicit topic. So if they are mentioned later in the same text, they could be argued to be implicitly metadiscur- 2 For ease of reference, I will subsume punctuation under typographical in this article even though the two are, of course, distinct aspects. Georg Marko 248 sive, e.g. if someone in an interaction asks us, in reaction to something we have just said (or an implication thereof): Why are you so angry? (→ emotion), Do you really subscribe to Marxism? (→ system of ideas), or Do you consider yourself an expert in the field? (→ identity). But as Ädel (2006: 27) puts it, “the reference to the world of discourse [must] be overtly stated,” which means there must be an explicit element that makes reference to the language or the language use of the current speech event, e.g. a verb of saying, e.g. as mentioned before, I’m not saying that…, or a noun denoting a linguistic unit, e.g. the last sentence in the last reply was exactly my point, etc. This is the metalinguistic component of the metadiscursive reference, i.e. the way we refer to a discourse referent. We have to take into account that referring to someone’s language or language use by quoting passages or elements is also a legitimate form of metadiscursive reference. Such references are usually marked by prosodic or typographical devices - italics, underlined, inverted commas, quotation marks, etc. - which all are supposed to indicate that a linguistic element is metalinguistically referred to rather than ordinarily used (e.g. metalinguistically referred to here means… - the italics in this example clearly mark the expression as metalinguistic). Oddly enough, Ädel (2006: 28) discards such formal marking as not being explicit, as the metalinguistic reference is not done in words. My view is that it is done in words, it is just the shift from the object to the meta-level that is marked. However, there is a greater problem involved here because the metalinguistic markers mentioned help us in our interpretation, but they are not a requirement, i.e. their absence does not mean that a passage is not metadiscursive. If I write I will not use absence again in this article without setting off absence from the rest of the sentence, the context (perhaps also the syntactic form, i.e. especially the lack of a determiner such as my absence) makes it clear that I am still referring to an element of my own discourse, viz. the word absence just used in the first sentence of the current paragraph. This means we are dealing with metadiscourse in the narrow sense here. This, however, entails that we either have to drop the criterion of explicit metalinguistics or we have to re-interpret it to include direct quotes, even if not metalinguistically marked. The second option appears to make sense and it is the one I opt for, even though it will have no serious implications for the current article. 3.1.3 Relationship between metadiscourse and discourse If we look more closely at the word metadiscourse itself, we will see that it is ambiguous or at least semantically odd. Firstly, the element discourse seems to fulfil two functions in the word: it denotes what something is about, viz. discourse, but also what this something is, viz. discourse. The word thus actually is metadiscourse discourse, i.e. literally ‘discourse about Power and status by definition 249 discourse’. 3 Secondly, as has been rightfully criticized by Sinclair (2005), the definition ‘discourse about discourse’ suggests that the former discourse is distinct from the latter one, so metadiscourse would then be a discourse A talking about a discourse B. However, metadiscourse is necessarily integrated into, or is part of, the other discourse as its function can only be seen in relation to the latter, e.g. it organizes or clarifies it. If someone, for instance, says: (5) Thanks for your reply. then they (no gender-specific user name found in the corpus) show their gratitude for a response to an initial question. This metadiscursive act (referring to a speech act performed by another participant in the same speech event), however, is embedded in an - in this case - lengthy exchange of objective and subjective information on a health condition. And thanking someone for what they have done in this exchange, of course, has a function for at least the social atmosphere of the very interaction. Being only a part of the ongoing discourse means that the main speech event is not exclusively - or even primarily - concerned with talking about itself, but mainly with some other object (= the object level of discourse). This implies that metadiscourse always marks a shift from the object level to the meta-level, with the discourse normally shifting back to the object level afterwards. I refer to such elements of a discourse that have a specific strategic function with respect to the ongoing discourse and mark a shift from the main strand of the discourse as moves (the term is often used for general elements in the rhetorical structure of a text, cf. Swales 1990). The shift becomes obvious in example (6) below, where a poster talks about a diagnosis he received, using the technical term for a particular pathological phenomenon (first occurrence of the term, the first shaded element in (6)). He then goes on to ask what the term means, thus moving from the ordinary use of the term to describe what happened to referring to the term qua linguistic expression (second occurrence of the term, the second shaded element in (6)). Mind that all examples - below and elsewhere - are SIC! , this means informal spellings and grammatical, lexical, and orthographic errors have not been changed or corrected. (6) Apparently, they’ve ignored the presence of the white matter lesions and have classified them as peri-ventricular non specific white matter lesions which are 3 The same argument could also be made for metalanguage, which is also ‘language about language’ or meta-level, which is also ‘the level above a specified level’. Georg Marko 250 not significant according to this consultant! Could some one please tell me what peri-ventricular non specific white matter lesions means? 4 In many other cases, though, the shift is hardly noticeable because it is so tightly integrated into the ongoing discourse, both formally and semantically. This applies to the following example (7), where metadiscourse is confined to the relative clause that the above poster said, referring to a statement made in the same thread (and thus within the same discourse). As this relative clause is just used to specify the noun things in a clause that presents the writer’s opinion (I will disagree with) and is thus not metadiscursive by itself, the metadiscursive shift is backgrounded. (7) I will disagree with some of the things that the above poster said. 3.1.4 Relationship between metadiscourse and propositional level In Systemic Functional Linguistics (= SFL), language structure is assumed to be informed by three metafunctions, the ideational one, concerned with representing, or providing information about, the world, the interpersonal one, concerned with evaluating this information according to likelihood and desirability and with integrating it into communicative interactions with other people, and the textual one, concerned with organizing the information according to features such as importance, newness, etc. (cf. Halliday 1994, Eggins 1994, Thompson 1996). Now there is a tradition in the study of metadiscourse that starts from the assumption that language use and thus discourse mainly means exchanging information about content. The ideational - or propositional - function is thus central since the other two functions are dependent on the former, adding to it or organizing it rather than presenting something completely unrelated. This has led to a conception of the interpersonal and the textual functions as being about what is essential to a speech event (i.e. information) and thus as being about discourse. As a consequence, such approaches conceive of metadiscourse as mainly covering the nonpropositional dimensions of language use (cf. Ädel 2006: 16-17, Mauranen 2010: 14-15). This becomes clear in the definitions of metadiscourse given by Crismore, Markkanen and Steffensen (1993: 40): Linguistic material in texts, written or spoken, which does not add anything to the propositional content but that is intended to help the listener or reader organize, interpret and evaluate the information given. 4 Note that even though the discourse referent is not metalinguistically marked here, the grammar of the last clause points to metalinguistics. The fact that the word lesions, which is plural, goes together with the singular verb means shows that the expression is meant - the word lesions is plural on the object language level, but singular on the meta-level, after all it is just a single word - and not the pathological condition. Power and status by definition 251 and Hyland (2005: 14): [Metadiscourse covers] aspects of a text which explicitly organize a discourse or the writer’s stance towards either its content or the reader. This conception of metadiscourse is relatively broad since it covers a wide range of phenomena, from the use of modality, discourse markers, hedging, theme/ rheme, topicalization, etc. (cf. Mauranen 2010: 14). In the alternative perspective, metadiscourse is limited to instances of explicitly referring to and commenting upon any aspect of the current discourse. Discourse here, however, covers all levels of what is happening in the current speech event. The range of phenomena nevertheless is not as wide and the definition of metadiscourse consequently is narrower. As metadiscourse here reflects upon the ongoing discourse, this conception is also called discourse reflexivity (cf. Mauranen 2010). This article will understand metadiscourse as discourse reflexivity. I will not look into phenomena typically covered by the broad, SFLinspired approach mentioned above. 3.1.5 Intratextuality The scope of metadiscourse was above defined as the ongoing discourse. Now while it is certainly debatable in certain instances what constitutes the beginning and the end of a speech event (e.g. in connection with online discussion forums, the boundary might be a single post or a thematic thread or the whole host site), it will be clear in most cases. Referring to linguistic aspects of other speech events thus means transgressing the boundaries of the current discourse and should therefore be treated as intertextuality, rather than intratextuality, the latter being a necessary criterion of metadiscourse. 5 The following two examples illustrate the distinction. In the first one, the writer mentions that his doctor presented a diagnosis (atrial fibrillation) in a consultation with him. This interaction happened before and outside of the health forum and participants therefore have no knowledge of what happened. The main piece of information consequently has to be mentioned. In the second example, the writer draws a conclusion from another participant’s posting, which is also part of the same thread in the forum. Participants can therefore rightfully be expected to know the content of the posting and the writer therefore does not need to repeat it, but only metadiscursively refers to it. 5 Traditionally, discourse (at least in one of its senses) is not only equated with speech event, but also with text. This is due to the fact that earlier research focused on written texts, where discourse and text were not so easy to distinguish. This is the reason why the terms used are based on textuality rather than on discursivity (as a matter of fact, interdiscursivity and intradiscursivity would be more precise, in my view). Georg Marko 252 (8) Dr said atrial fibrillation. (9) From what you’ve said, it doesn’t sound like your illness is something that will be easy to manage without professional help. The two examples also show why the distinction between intraand intertextual is not arbitrary, but rather is central to the nature of metadiscourse as discourse reflexivity. Metadiscourse is supposed to have an immediate impact on the speech event in which it is embedded, connecting aspects (as in the example above), organizing, clarifying, and securing comprehension of the current communication. Intertextuality, on the other hand, does not have this function. So asking What do you mean by a scan? (and the answer to the question) is pivotal for participants to understand each other in the situation in the health forum. Asking What does your doctor mean by that? (and the answer to the question) may be important to understand what happened in a consultation, but not for the processing of the current interaction. If we postulate that immediate relevance for the current interaction - as an implication of the condition of intratextuality - be a necessary feature of our conception of metadiscourse, there is a consequence that is never addressed in the literature, viz. the question of how we could classify metalinguistic references to aspects of the ongoing discourse that have no immediate relevance for the latter. This would be the case if such a move opened a new, albeit probably temporary, topic not functionally related to the rest of the discourse. I have not found any examples in my corpus (which is not to say that there aren’t any, they might just be difficult to track). But we could imagine a spoken interaction where someone suddenly says I love the way you pronounce the word ‘intricacies’ (if another participant in the conversation has just used the very word). As a matter of fact, as linguists we often feel tempted to act in exactly this manner, mentioning that a certain exchange is a nice example of a flouting of a Gricean maxim, that a certain word occurring in someone else’s turn is etymologically related to some other word, that a sentence used contains an aliteration, or - reacting to the last remark - that alliteration is, as a matter of fact, spelt with double ll. Are we ready to discard such examples as non-metadiscursive? I think that claiming apriori knowledge of whether such moves have an immediate impact on the current discourse is problematic. We could very well argue that such utterances are intended to change the topic, at least for a moment, which might in turn serve the less immediate purpose of providing relief, avoiding conflictual subjects, offending or comforting someone (as a matter of fact, Mauranen (2010) postulates a topicchanging springboard function for metadiscourse; and Fill (1993: 68-69) mentions that metadiscursive shifts, e.g. saying something like You’re constantly interrupting me? or Keep calm! (reacting to the anger and excitation in another participant’s utterance), often lead to, or have the purpose Power and status by definition 253 of, creating, aggression). I will therefore not treat this as a prerequisite for metadiscourse. 3.1.6 Mode Metadiscourse is not a phenomenon restricted to either the written or the spoken mode. As has been pointed out and shown in various studies (cf. Schiffrin 1980, Ädel 2006, Mauranen 2010, Penz 2016), the functions of metadiscourse in spoken and interactive - the interactive aspect might very well be the more important dimension here - language might be different, with the local co-construction of discourse comprehension taking a more central position as opposed to the structuring of one’s text. These, however, are tendencies rather than categorial differences. Although multimodality is hardly ever mentioned in connection with metadiscourse, the phenomenon also occurs in discourses combining different semiotic modes. Any online interaction, for instance, that includes photos could lead to verbal comments on the picture itself (e.g. This is an interesting camera angle or It is a bit blurry) rather than about what you can see in the picture. Multimodal metadiscourse has not been found in my health forum posts and I will therefore not have anything further to say about it. 3.1.7 Fuzziness For Ädel (2006), metadiscourse is inherently fuzzy as “it is difficult to make categorical distinctions between what is and what is not metadiscursive” (Ädel 2006: 22). The question here is whether this is just a methodological problem - if confronted with an example, it simply is not easy to distinguish different categories - or whether there are any inherent theoretical aspects that make distinctions difficult. There is, in my view, a theoretical source for the fuzziness of metadiscourse, viz. the fact that we cannot clearly distinguish between the meaning of a text, on the one hand, and the content and information that it conveys. As discourse is always concerned with content and information, we also relate and react to the latter. And of course, content and information are more or less directly linked to texts and elements in the texts as the latter ‘created’ them in the first place. However, in most cases this relation to text and language is irrelevant so that utterances such as I get that [referring to a pathological experience someone else has just described] too mate or I agree with you 100%, while obviously reacting to something that was said in the same thread of a forum, seem to focus on content and thus not really on the language used, i.e. on the linguistic forms and/ or linguistic meanings. The metalinguistic dimension of metadiscourse thus seems to be missing. However, whether or when something is no longer metalinguistic is a matter of degree and often also a matter of interpretation. For Schiffrin Georg Marko 254 (1980: 201, 219), for instance, operators like true, false, or agree are indicators of metadiscourse, while I would - and will - exclude them from consideration because they are only concerned with content. Definitions in the sense of providing a meaning of an expression used in the immediate co-text clearly are metadiscursive moves: they necessarily occur in a discourse where the expression also occurs, possibly more than once (defining an expression not used would not make any sense) and their metalinguistic dimension is usually explicitly marked (even though this is more complex than this simple statement might suggest, see below). 3.2 Functions of metadiscourse As indicated above, the overall function of metadiscourse is to direct addressees through a discourse and secure their comprehension (in monologic texts) or to co-operatively co-construct a comprehensible discourse (in interactive scenarios). Most theoretical accounts go on to propose sets of more specific functions. However, these often seem to be specific to the domain examined (e.g. Mauranen’s (2010) functions are particularly relevant to the use of English as a lingua franca as this is what she investigates). No more systematic taxonomy of functions has been developed so far, to the best of my knowledge. I will not be able to achieve this either, but in the following I will at least propose the very general structures for such a framework. On a first plane, we can distinguish between primary functions and secondary functions of metadiscourse. The former are concerned with achieving the superordinate objective mentioned above, the latter with social goals that are only indirectly connected to the primary functions. Primary functions can be categorized according to whether they provide information on an aspect of the ongoing discourse, whether they evaluate the latter, or whether they are concerned with organizing information. We may call them ideational function, evaluative function, and structuring function of metadiscourse. They clearly correspond to the three metafunctions of SFL introduced above (I do not use the same terms, though, because I do not believe they are ideal descriptors). The ideational function seems to be the most common one, as any information on something in the current text or interaction such as a particular lexical item or a whole utterance would be subsumed under this category. Writers and speakers provide such information if they think and/ or have evidence that there is or could be a problem with comprehension. Ideational metadiscourse thus often serves the purpose of clarification by adding, elaborating on, or modifying information. This may be done because the speaker/ writer foresees a problem, as in the following example, where the poster defines two technical terms, viz. cervicogenic Power and status by definition 255 and cervical, by providing the respective meanings between brackets (the discourse referent shaded, the information in bold). (10) Pain in the back of the head is often referred to as a cervicogenic (meaning: from the neck) headache because it often originates from an irritation in the upper cervical (neck) spine. Clarifications could, however, also be triggered by questions such as: (11) A: What do you mean by a scan? I hope you mean MRI. […] B: yes i did mean an MRI, In some cases, they are provided by other participants, who offer their own interpretations of what another speaker/ writer has said (the term formulations, adopted from Garfinkel/ Sacks (1986/ 1970), is often used for such moves where a speaker metadiscursively offers an understanding of what has previously/ so far happened in a conversation, cf. Childs 2015) (12) A: Prayer I have had the same problem, The lord Jesus saved me, i couldn't sleep and was scared all day long i couldn't take it no more and i thought i would die. […] B: You are saying this was cured with prayer? Ideational function does not always mean clarification, though. In some cases, a discourse referent is simply integrated into another piece of information, as in the following example, where the poster uses a metadiscursive move (what you said) just to connect what someone else said in the thread to some other information. (13) In regards to what you said about the neurologist - I have a letter, as well as a list of every medication I have taken, which ones have worked, which didn’t, what triggers my migraines, etc. While metadiscourse seems to be concerned with meanings in the widest sense, it can also refer to formal aspects such as pronunciation or spelling. In the first example below, for instance, a poster provides information on how to pronounce the word Sjögren’s, using an orthographic form of transcription (SHOW grins) (this, by the way, would be an example of a metadiscursive move without any immediate function for the ongoing speech event; see above). In the second example, the poster uses (sp) between brackets to indicate that she (rightly) doubts her own spelling of oesophagus and that she invites others to give the correct orthography (this is common practice in discussion forums). She thus could be argued to request information about spelling and thus about a metalinguistic aspect of an expression used in the current discourse. (14) With the dry eyes, your doctor may look at the possibility of Sjögren’s syndrom (pronounced “SHOW grins”) Georg Marko 256 (15) I would feel like something was caught in my asophogus (sp) almost like heart burn Metadiscursive moves may also have evaluative functions, which means that some aspect of the current discourse is assessed with respect to (grammatical/ linguistic/ other) correctness, aesthetic or ethical value, importance, etc. and/ or is integrated into an evaluative dimension of the interaction (e.g. thanking, praising or criticizing someone for something they have said) in the ongoing speech event. In the first example below, the poster metalinguistically refers to the word intelligence (putting it between quotation marks), used in a preceding contribution, and adds an evaluative comment, assigning the adjectives sticky, vague and broad, all meant in a negative way, to it. In the second example, the poster refers to nicknames of participants in the evolving discussion, whose meanings (structure-sensitive seems to mean ‘semantically transparent’) are so clearly negative and ambiguous - the names in question are no-one and no-hope - that the writer - evaluatively - suggests that such names should not be used (the discourse referent shaded, the evaluation in bold). (16) I think using “intelligence” is a sticky word. It is vague and broad in it’s different meanings. (17) And it’s true people should not use “structure-sentisive” (or whatever you could call it) nicknames. In many instances of evaluative metadiscourse, a metalinguistic reference to an element in the ongoing speech event is integrated into an emotional utterance or a speech act with an evaluative dimension. What is particularly common in the health forums is showing empathy with the help of metadiscourse by using words such as understand, relate to, or hear, all in the sense of ‘emotionally and empathetically relating to’. Here are two examples of this pattern. (18) I completely understand what you mean by “deserving” to shower. (19) i can so relate to what u are saying Equally often we find metadiscourse in evaluative speech acts such as thanking, praising, criticizing, correcting, etc. Showing gratitude and/ or appreciation for an answer is by far the most common subcategory here, as in the examples below. In all three cases, the discourse referent is a response to an initial question/ problem description, metalinguistically represented as your reply/ answer. This is then embedded in a positive act (thank you for) in (21), and descriptions of positive emotions (loved and made me feel so much better) in (21) and (22). Power and status by definition 257 (20) Thank you for your reply and sharing your insight about searching for a possible surgeon. (21) Loved your answer as I had just been prescribed Wellbutrin and found DF as I was researching it (via Google of course! ). (22) Now your reply has made me feel so much better. As far as the third class mentioned above is concerned, metadiscourse has a structuring function if it organizes discourse, dividing it into sequences of language-related components, most commonly speech acts. The structuring function may refer to what has been said or what will be said, the latter being more common. In the simplest form, metadiscursive moves with this function just name one act that is to be performed or has just been performed. In (23), for instance, a poster announces a speech act, i.e. an answer, thus preparing the addressee (and other readers) for what to expect (here also linking the speech act to another act, viz. the original question). In (24), the poster uses the metadiscursive reference my question to announce the nature of the following act. Back to also links this to a previous passage in his post where he already started to formulate this question. The metadiscursive move thus forms a connection between what happened in the speech event to what is going to happen. In the concrete example, this is very important because neither is there any indication that what the poster said before could be a question, nor does anything he says afterwards sound like a question until the very end of his post. Without the metadiscursive move, readers would not expect the poster’s description of his experience to finally end in a question. (23) Honestly, I will answer it this way … IF you have to ask this question then your previous drug use has definately done something to that noggin“ of yours (and it is NOT MS related) … (24) Anyway back to my question, yesterday i had a strange experiiance, i was sat down and all of a sudden, it felt like both my eyes were going from side to side really fast, also both my legs started to kind of “jump about” like they were beeing hit with electric shocks, i have vertigo on and off at different times, but this didnt feel the same, has anyone had this happen to them? In more complex cases, metadiscourse mentions several acts and also orders them by using ordinals (1./ 2. or firstly/ secondly/ thirdly), as in the example below, where the poster announces two questions and then uses an ordered list to present them. (25) I have two questions here: 1. How long does a clean/ normal Cardiac Catheterization and Echo Cardiogram last....(in other words, how long is the diagnosis good for? ) 2. Can anxiety really cause someone to have all those physical symptoms? Georg Marko 258 Structuring metadiscursive moves can refer to the future, but theoretically also to the past, e.g. by ending a post with These are my two replies. To a certain extent, the above example with back to my question does partly have a backward orientation. But apart from this post, I have only found forward-oriented instances of this function. Judging from my informal explorations of my corpus, the structuring function is mostly limited to announcing a series of two or more questions, but may not play as important a role as in written monologic texts, probably because the individual contributions to the interaction are relatively short. As definitions provide the meanings of lexical elements, they are clearly ideational in function and cannot really be claimed to do anything else besides this on the primary plane. By providing information, evaluation or organization, metadiscursive moves can indirectly also fulfil further, i.e. secondary, functions that go beyond what has been described above. These are mostly social functions concerned with the management of the participants’ social identities or social relationships. 6 Whether we can provide information about linguistic elements in the ongoing discourse, whether we criticize others for the linguistic elements they are using, or whether we give five answers to certain questions - all aspects that may be part of metadiscourse - will inform the ways others perceive us and our relationship with them. Definitions are very important in this respect. Obviously, speakers who define a term that they are using themselves indirectly signal two things to their addressees: firstly, the speakers know the meaning of the term; secondly, the addressees are not assumed to know - or are positioned in a way as if they did not know - the meaning of the said term; as a consequence, a difference in knowledge (often called competence gap, cf. Lupton 2003: 117) is constructed between speakers and addressees, which in turn will create an at least temporary hierarchical relationship between knowing and unknowing participants. This may be in line with the apriori perception of such a competence gap. In a university context, for instance, I as the linguistics teacher can be expected to know more about language than my students and I can therefore also be expected to provide most definitions in a course. However, in an interaction in a lay health forum - the discourse this article focuses on - the situation is different because it is a discourse in which normally only lay people participate and which therefore does not feature an apriori hierarchical relationship. Such hierarchical relationships may have further implications, enhancing some participants’ status and credibility, while undermining others’. As status and credibility are associated with power, what has been said may also lead to the unequal distribution of power in a presumably or theoretically egalitarian context, viz. a lay discussion forum. 6 There may also be other secondary functions. These, however, will not concern us here. Power and status by definition 259 However, definitions do not create a competence gap, hierarchies and power relations automatically and for all times. Participants using definitions may make claims, but in an interactive discourse such as a discussion forum these claims may be confirmed, reiterated, refuted, negotiated - in other words, these aspects are managed locally, i.e. by the very utterances that participants use in the concrete situations. Whether, and if so, to what extent and how this is done is the subject matter of the study the approach to which this article outlines. Suffice to say that such processes are not restricted to definitions, but could occur in connection with other forms of metadiscourse, too. 4. A structural-functional account of definitions As mentioned in the introduction, books and articles on metadiscourse often discuss many aspects relevant to the study of the phenomenon, but normally not in a systematic way, i.e. not as a framework of variables whose values may be central to the description and interpretation of the use of metadiscourse. In the previous chapter, I tried to at least point to areas relevant to a theorization and description of metadiscourse in general. This chapter will provide a more detailed framework of structuralfunctional - i.e. focusing on form and meaning - variables applicable to the analysis of definitions in interaction (with special emphasis on discussion forums on health). Many of the variables are also relevant to research into other forms of metadiscourse. I will, however, confine myself to definitions in my discussion and also in my selection of illustrative examples, which are all taken from my corpus. With the exception of an outline of the basic structure of definitions, all sections in this chapter introduce aspects in the form of variables. Variables can be represented as simple questions, e.g. in which order do the expression to be defined and its meaning occur. Variables have two or more values. These are the different answers that could be given to the initial question, e.g. the possible orders are [expression > meaning] or [meaning > expression]. The values might be equally important, but in some cases there is one that is more prototypically associated with definitions. We have to bear this in mind when assessing their functions. The variables described below are: ‣ Order of elements ‣ Formal elaboration ‣ Semantic exhaustiveness ‣ Genericness ‣ Interactiveness Georg Marko 260 ‣ Explicitness ‣ Metalinguistics All of these variables and their values will be assessed with respect to their relevance to a study of the secondary social functions of metadiscourse mentioned above, with special emphasis on the local management of the competence gap and concomitant social hierarchies and power relations. However, as mentioned, I will start with looking at the basic structure of definitions, naming and describing their essential components. 4.1 Basic structure of definitions The three elements introduced above in connection with metadiscourse in general are also part of definitions, even though the case of the third one, i.e. metadiscursive reference, is more complex than suggested in the general discussion. The discursive referent, i.e. the element from the ongoing discourse I want to talk about, is a simple (e.g. hemophilia) or complex (e.g. polycystic ovarian disease) lexical element. It is called the definiendum (Latin, ‘something to be defined’). The metadiscursive comment in definitions is the meaning of the above-mentioned lexical element, here called the definiens (Latin, ‘something that defines’). The metadiscursive reference, i.e. the element that marks a word as a lexical element, consists at least of one of the following two elements: a metalinguistic marker of the lexical element, which could be either lexical, e.g. using expressions such as word, term, expression, etc., or prosodic/ typographical, e.g. using italics, inverted commas, quotation marks or corresponding prosodic means for the word (as done in this very sentence) and/ or a semantic connector that indicates that the metadiscursive comment is concerned with the meaning of a lexical element, which again could be either lexical, e.g. using expressions such as define, mean, understand (something by X), or prosodic/ typographical, e.g. using inverted commas, quotation marks, brackets or corresponding prosodic means. The discourse deictic aspect could be represented by demonstrative pronouns, e.g. this word means, or other deictic elements, e.g. the paragraph above is saying. In some cases, the deictic relationship is simply established by the repetition of the lexical element to be defined. I will give a few examples from my corpus to illustrate these categories. As in all examples below, the definiendum will be shaded, the definiens will be in bold, semantic connectors and metalinguistic markers will be underlined (except for typography/ punctuation, as underlining brackets or inverted commas would be hardly visible). (26) Meds like sertraline aka zoloft (a SSRI) can help with migraine. Power and status by definition 261 In example (26), for instance, the definiendum is the term sertraline, the definiens zoloft and - here, we even have two meanings - a SSRI (‘selective serotonin re-uptake inhibitors’, a common class of antidepressants). The definiendum is not marked, but we find a semantic connector in the acronym aka ‘also known as’ and the brackets, which indicate that what we find between them is the meaning of the preceding term. (27) A: Can someone define “nystagmus” for me? […] B: Yes, nystagmus is involuntary eye movement. In the exchange above, the definiendum is the term nystagmus, the definiens - requested by poster A, provided by poster B - is involuntary eye movement (poster B uses affirmative Yes because poster B already offers this as a possible meaning, when saying Is this what nystagmus means - a uncontrollable movement of the eyes? at the end of his/ her post (not included above)). Poster A uses quotation marks as metalinguistic markers of nystagmus, and there are also two semantic connectors, viz. define in the question and is in the response (copula be is probably sufficient because A has already used the verb define to indicate that we are concerned with a relation between a word and its meaning). (28) The term paradoxical septal motion (PSM) refers to movement of the interventricular septum towards the right ventricle rather than the left ventricle in systole (during the contraction phase), with normal thickening. In this example, the definiendum is paradoxical septal motion, the definiens is, on the one hand, the acronym PSM, and, on the other hand, the whole passage from movement of the interventricular septum to the end of the sentence. The definiendum is metalinguistically marked by the word term. There are also two semantic connectors, viz. the brackets around PSM and the verb refers to. There are no restrictions on what qualifies as definiendum. There is, however, a tendency to define (technical) terms, i.e. expressions with a very specific meaning within a particular domain which are not normally used in everyday language, have no evaluative connotations and are primarily known to experts in the domain (hence the need for the definition in contexts involving non-experts). In some cases, however, it makes perfect sense to use definitions also for non-technical expressions if the latter are pivotal for the understanding of an utterance. This becomes obvious in example (29), which demands a definition of the non-expert word ugly, which the poster deems central to someone else’s post (to be precise, it is, as mentioned, demanding rather than providing a definition, but it supports the point). (29) define the word “ugly” Georg Marko 262 There are of course also mock definitions, which treat ordinary terms or artificial terms for very mundane aspects as if they had some special status. In (30), for instance, totally ridiculous amounts of candy is given as the meaning of the invented technical-looking acronym t.R.A.C. (30) I took my children to church last night for a t.R.A.C. (totally ridiculous amounts of candy) party. As can be seen in the discussions below, there is also some variation in what appears as definiens in definitions. It is important to stress at this point that I will only be concerned with meaning as the socially-shared sense that an expression has, but not with any personal connotations that individuals associate with such an expression. So (31), while looking like a definition formally at the beginning, is a typical example of what will not be included in the theory below. (31) […] so to me the word Migraine means extreme pain! ! ! 4.2 Order of elements The fundamental question in connection with the order of elements is: In which order do definiendum and definiens occur? The variable has two values, viz. anaphoric and cataphoric. These will be explained and illustrated below. 4.2.1 Anaphoric In anaphoric definitions, the term to be defined, i.e. the definiendum, occurs before its meaning, i.e. the definiens. This is the traditional sense of what we understand by definition, i.e. have a term and then provide its meaning. The sentences below are just two of the many examples to be found in my corpus of health forums. As can be seen, (32) is a definition in a more straightforward sense as the discourse referent limbo is metalinguistically marked by the word word and quotation marks and we also have a clear semantic connector in is defined as. The second sentence is less clear since neither is the discourse referent, i.e. the word colostomy, metalinguistically marked, nor is the copula is a clear semantic connector. We may, however, still interpret the copula as a semantic connector of a definition because the neoclassical compound (a word composed of elements from Ancient Greek) colostomy looks like a technical term - and thus a term which a lay audience may not know - and the rest of the sentence with its generic character resembles the formulation we find in definitions. (32) The word “limbo” is defined in Wikipedia as this: Latin limbus, edge or boundary, referring to the “edge” of Hell Power and status by definition 263 (33) A colostomy is a surgical operation that creates an opening from the colon to the surface of the body. 4.2.2 Cataphoric In cataphoric definitions, the order is reversed, i.e. we first encounter the description of a phenomenon and are only then given the term for it. We may classify this as labelling (or naming), i.e. we provide a label for something, rather than as defining. Strictly speaking, however, this is a definition because we have all the required components. The only difference is whether the expression is the point of departure or the phenomenon. The sentences below are examples of cataphoric definitions. The definiendum is not metalinguistically marked, but we have semantic connectors - is known as and is called - in both examples, which indicate the status of the sentences as cataphoric definitions. (34) Stroke induced dementia is known as vascular dementia. (35) This condition, dehydration plus hyperthermia is called heat stroke. Why is the distinction between anaphoric and cataphoric definitions relevant for the study of social functions of definitions? Cataphoric definitions have a secondary function with respect to textual organization. Having a term for a phenomenon which may be clumsy to describe allows speakers and writers and other participants in the interaction to refer back to this phenomenon with a single lexical element. A term might also be provided if this very expression has already been used elsewhere in the current speech event, thus forming a cohesive tie. Unlike anaphoric definitions, which are crucial for comprehension - without them, we would not understand what a particular term means - cataphoric definitions, however, are redundant for comprehension as not having a term for a phenomenon does not mean that we do not understand the very phenomenon. Being a redundant move, cataphoric definitions may thus even more strongly than anaphoric definitions be motivated by the secondary goal of presenting oneself as a ‘know-it-all’ and of creating a competence gap between speakers/ writers and addressees. This probably must be taken with a grain of salt since this redundancy of cataphoric definitions may also mean that they and their implications are not interpreted as relevant by others. So a more thorough analysis of this variable is certainly necessary. (I tend to favour the former interpretation.) Georg Marko 264 4.3 Formal elaboration The fundamental question in connection with formal elaboration is: How elaborate is the linguistic structure of the definiens? In other words: if we provide meanings, how do we formulate these meanings linguistically? There are two values, viz. propositional and lexical, both with two subclasses. These will be explained and illustrated below. 4.3.1 Propositional A definition is propositional, if the definiens is represented by a structure that explicitly or implicitly describes a situation in the world and could theoretically be true or false. Propositional definitions may be:  Phrasal The definiens is a phrasal structure, normally a (complex, i.e. consisting of more than a single word) noun phrase. In the example below, the definiendum arrhythmia is defined by the noun phrase an irregular heartbeat, which implicitly expresses the proposition ‘if/ when one’s heartbeat is irregular’. (36) You also may need a stress test if, during exercise, you […] have other symptoms of an arrhythmia (an irregular heartbeat).  Clausal The definiens is a clausal structure, often a subordinate clause starting with when or if. In the sentence below, the definiens is represented by the when-clause. (37) Chiari Malformation is when your cerebellar tonsils protrude down into your spinal cord […] 4.3.2 Lexical A definition is lexical if the definiens is simply an alternative lexical element. This element may be a(n):  Synonym The definiens is an alternative term that has the same (or almost the same) meaning as the definiendum. In the example below, only the single word clot is given as the meaning of the word emboli (which is further metalinguistically characterized as stemming from Greek here). (38) emboli is from the Greek and means clot. Power and status by definition 265  Allonym The definiens can be regarded as an alternative realization of the same (or almost the same) lexeme (= allonym). This almost exclusively applies to abbreviations and the corresponding full forms. In (39) below, the definiendum AP, an acronym, is thus defined by the spelt-out allonym Advanced Placement. This also applies to (40), which does, however, not include a semantic connector, but is just metalinguistically marked by brackets. Interestingly enough, the term is further defined by the phrasal definiens a progressive neurological disorder and a form of dystonia. (39) AP means Advanced Placement […] (40) Benign essential blepharospasm (BEB) is a progressive neurological disorder and a form dystonia. As far as the social implications of the distinction introduced above are concerned, it seems plausible to assume that lexical definitions require less real expertise and may therefore be used by posters who are not really interested in providing their addressees with the meanings of terms. This could, in turn, mean that for them the social function may have more significance. However, we do not know how other participants react to different linguistic realizations of definitions. Perhaps more elaborate definitions, particularly those using whole clauses, might be perceived as more knowledgeable. This could then create a larger competence gap than the shorter forms of defining. (I tend to prefer the latter option.) 4.4 Genericness The fundamental question in connection with genericness is: How general is the definiens? The variable has two values, viz. generic and exemplary. These will be explained and illustrated below. 4.4.1 Generic Generic definitions describe the meaning of an expression as a set of general, timeless and non-individual features. As definitions are prototypically generic, almost all of the examples given above and below belong to this category and further ones are not required. Georg Marko 266 4.4.2 Exemplary In exemplary definitions, the definiens consists of the anecdotal description of a specific event, experience or phenomenon (or a series or group of these), usually with the implicit invitation to generalize this description (often with discourse markers indicating exemplification such as for example or e.g. - the very fact that something is labelled an example implies the superordinate generic category that this something is an instance of). Exemplary definitions are clearly the non-prototypical case with respect to genericness and they are therefore relatively rare (and difficult to track in a corpus). (41) could, however, be considered an instance of this category. The poster defines the expression big events by giving an example, viz. her grandmother’s death (to be frank, she seems to be wandering off-topic here so that the relation between the dream of her grandmother’s death and the concomitant hope that it is not a bad omen, on the one hand, and big events (and her failing memory), on the other, remains unclear). (41) One last problem is my memory, I can never remember both big events or simple little things that I see, or was told, or read. By big events I mean, for example, I had a dream that my grandmother passed away and the next day I was rethinking the dream and told my father about how I don’t like having dreams about family dying because what if it were to really happen. Providing an example as a definiens appears to be more common with cataphoric definitions. This does not really come as a surprise as describing something concrete and then adding a label for the general category under which this can be subsumed does not seem to be an unfamiliar pattern (albeit occurring not nearly as frequently as other forms of definitions). The exchange in (42) is an example of a cataphoric exemplary definition. Poster A describes the (recurrent) experience of a whooshing sound in her right ear. Poster B refers to her description and says that her experience (when you can hear it [the whooshing]) is called intracranial hypertension or pseudotumor cerebri. (42) A: […] I also occasionally and not necessarily concurrently hear a whooshing rhythmic (to my heartbeat) in my right ear. […] B: […] The whooshing is your blood. When you can hear it, it is called intracranial hypertension, or occasionally called pseudotumor cerebri. […] We do not need to discuss the difference between generic and exemplary definitions with respect to their social functions. The latter do not occur so often and if, they will probably not be perceived as proper definitions. This may, in turn, mean that associations with knowledge and expertise will be weaker. Power and status by definition 267 4.5 Semantic exhaustiveness The fundamental question in connection with semantic exhaustiveness is: How much semantic detail does the definiens provide? To put it differently, we are interested in whether a definition provides the full and unambiguous meaning of an expression or only a partial version of it. The variable has three values, viz. exhaustive, categorial (nonexhaustive), and allonymic. These will be explained and illustrated below. 4.5.1 Exhaustive In exhaustive definitions, the definiens provides the full distinctive meaning of an expression, i.e. the meaning that is specific to the latter and which it does not share with any other expression (with the exception of synonyms). The definitions in the following two examples are exhaustive as the two meanings exactly describe what sclerosis and palinopsia are, respectively, neither being more general, nor more specific. (43) The “sclerosis” [this part of the word multiple sclerosis] means scars […] (44) I have had visual snow and palinopsia (afterimages) for over a year now. 4.5.2 Categorial (non-exhaustive) In categorial definitions, the definiens only contains reference to a superordinate category without giving more specific details that would distinguish a concept (as the meaning of an expression) from other members of this category. They therefore are non-exhaustive and also ambiguous (there is more than one expression that would be defined by the given meaning). The following two examples feature categorial definitions as Naproxen is not the only non-steroidal anti-inflammatory drug (NSAIDs are, as a matter of fact, the standard class of painkillers) and palinopsia is just one of many eye problems (note that in (44) palinopsia is defined exhaustively). (45) I have taken naproxen (a non steroidal antiinflammatory drug) for just over a month. (46) I also […] developed palinopsia (eye problem) Categorial definitions cannot only be anaphoric, but they may also be cataphoric, as in (47) below, where the definiendum metroprolol is added to a categorial definiens a beta blocker. (47) The last thing I’d recommend is a beta blocker called metoprolol […] Georg Marko 268 4.5.3 Allonymic As already mentioned in connection with formal elaborations above, allonymic definitions are restricted to expressions that could be argued to be realized in two different ways. This primarily applies to abbreviations and acronyms and their spelt-out forms. Providing an explicit version of such a term is, to a certain extent, an exhaustive definition, albeit a very special one. The two examples below show the difference between an ordinary definition and an allonymic definition. The acronym TIA is defined by the explicit three-word allonym transient ischaemic attack (spelt correctly) in (48), and in terms of its conceptual content - a minor stroke (i.e. a condition caused by a brief interruption of blood flow to certain brain regions, which causes stroke-like symptoms, which however disappear again without causing any permanent damage) - in (49). (48) If stroke was excluded other possibilities could be TIA (transitory ishemic attack), brain infection or some poisoning. (49) I had a tia (minor stroke). The three types of semantic exhaustiveness described above also differ with respect to the knowledge they (seem) to indicate. Categorial and allonymic definitions are certainly easier to provide - a quick WWW search on our smartphones will do the trick - than exhaustive ones, which might mean that the latter may be more effective in creating a competence gap. However, not giving a full definition could also be interpreted as a sign that the current speaker or writer does not think that her addressees and the other participants in a discussion forum are able to understand a more detailed account. This could then also lead to a widening of the perceived competence gap. (I tend to favour the former interpretation.) 4.6 Interactiveness The fundamental question in connection with interactiveness is: Is the definition the product of a single speaker or writer or is it accomplished interactively and collaboratively by several participants in a speech event across two or more turns? While we could differentiate between who first uses an expression, who initiates a definition, and who provides the definition, I do not want to overcomplicate the matter and therefore postulate just two values for this variable, viz. speaker-definitions and collaborative definitions (speaker encompassing writers here). These will be explained and illustrated below. Power and status by definition 269 4.6.1 Speaker-definitions If writers (more rarely, speakers) of monologic texts feel that their readers are not familiar with a lexical element being used, they will define it. They thus provide all elements of the definition, the definiendum, the definiens, the metalinguistic marking and/ or the semantic connector themselves. This might also be the case in interactive discourses if one participant introduces and uses a term that she then defines herself. It is not necessary to give further examples as most of those cited above are as a matter of fact speaker-definitions (with the exception of passages involving speakers A and B). 4.6.2 Collaborative definitions In interactive scenarios, there is also the possibility of collaboration, where we distinguish between using an expression, initiating a definition (e.g. by requesting one - What does X mean? or Could you define X? - or by evoking one - I’m not sure I understand this.), and providing the meaning of the expression, and where these acts are performed by at least two different participants. We might thus have one writer using a term, but at the same time asking for a definition. This may at first seem odd, but the situation is common if people talk - as they often do in health forums - about statements by third parties, e.g. a diagnosis by a doctor or even other people’s posts. In example (50) below, for instance, poster A speaks about what others talk about to then ask what they mean by the expression episode (in connection with multiple sclerosis) (interestingly, she also proposes a potential definition herself). Poster B, in the following turn, gives a sentencelength explanation in response to the request for a definition. (50) A: I always read about people having episodes. What is an episode exactly? I’m guessing that on May 5th when I experienced Vertigo for 24 hours, that was an episode? […] B: MS attack / episode is generally considered when it lasts for at least 24 hours and leaves when it decides to go. The request for a definition may also come from the second poster as in example (51) (already cited as (27) above, at least in a shorter version). A introduces the term nystagmus, B asks for a definition - the request is not directly addressed to A, but to anyone in the community - also, as in the example above, offering a plausible hypothesis concerning the meaning of the term (Does this have something to do with…). Finally, A provides the demanded definition. (51) A: I have mild nystagmus while tracking … like during reading. Mine does not get better no matter what I do but it is quite mild. Georg Marko 270 B: Can someone define “nystagmus” for me? Does this have something to do with eyes jumping around or not staying still? A: Yes, nystagmus is involuntary eye movement. Collaborative definitions may also be more complex than the two examples given, with more than two people involved and the whole process extending beyond two or three turns. The category described here is not restricted to anaphoric definitions, but also cataphoric definitions may be collaborative if writer or speaker A describes a phenomenon and speaker/ writer B then provides an expression for the latter (in the sense of What you are describing is called X.). Strictly speaking, A’s utterance is only turned into a definiens by B’s mentioning of the expression. Example (52) contains an interactively created cataphoric definition. Poster A describes an episode of severe headaches. Poster B refers back to this description and says that the term for this type of headaches is ice pick headaches (the expression is metalinguistically marked by quotation marks; this is relevant because sounds like is not normally a semantic connector pointing to a definition). Interestingly enough, B goes on to provide a definition of the very term herself (between brackets and introduced by i.e.), indicating that she feels that a personal anecdote - and it is this rather than a generic description of a phenomenon - is not sufficient as the meaning of a (medical) term. (52) A: [… description of an episode of headaches] B: Some of what you’re describing sounds like "ice pick headaches" (i.e. painful, stabbing headaches that only last for a few seconds at a time that are NOT accompanied by any other neurological symptoms). A comparison between speaker-definitions and collaborative definitions with respect to their effects on the creation of a competence gap and, as a consequence, of hierarchies and power relations in discussion forums - or elsewhere - suggests that the former category may play a bigger part in this. In collaborative definitions, the individual poster does not autonomously decide what to define and when and how to define it. She will thus not really be able to use definitions - consciously or not - to create the aforementioned competence gap. As a consequence, speakerdefinitions may be more interesting for a social investigation of definitions in discourse. 4.7 Attribution The fundamental question in connection with attribution is: Is a definition attributed to someone who is not a participant in the ongoing discourse? Power and status by definition 271 The two values of this variable are self-attribution and otherattribution. These will be explained and illustrated below. 4.7.1 Self-attribution In self-attributed definitions, the source of knowledge underlying the act of defining a lexical element is the speaker or writer herself. She may explicitly appear in the definition, e.g., if she uses a construction such as I define X as or as far as I remember, X means…. The fact that I have not found any examples of this in my corpus suggests that it is not very common for definitions to include direct references to the speaker herself. However, self-attribution is the default option, i.e. without any evidence that the definition given is actually someone else’s - and not the speaker’s - we assume that the speaker has the necessary knowledge for the definition. All examples cited above are actually self-attributed definitions. 4.7.2 Other-attribution In other attributed definitions, the source of knowledge underlying a definition is someone other than the speaker herself, generally someone with a high credibility, e.g. a scientist or more generally, any kind of expert. In some cases, it is not a person, but rather a source (text), e.g. the Oxford English Dictionary or Google, because the information obtained from there cannot easily be personalized and because the source might carry more authority than a concrete person (e.g. the OED is the authority, not the ‘unknown’ lexicographer). In the context of health forums, other-attribution usually involves general or specialist medical doctors. (53) and (54) below are examples of other-attributed definitions. In (53), the definition (MS means “multiple scars”) is embedded in a superordinate clause (doctors feel X) (which is itself embedded in another clause, but this is just an aside). This superordinate clause indicates that the definition in this case is assigned to doctors and there is therefore no selfattribution. In (54) (already quoted as (32)), the source to which the definition is explicitly (is defined in) attributed is Wikipedia. The definition itself is represented as a literal quote (the passage following the colon). (In both examples, the other source of knowledge is in recte, i.e. not in italics.) There is one important difference between (53) and (54), viz. that the way the attribution is formulated in the former suggests that multiple scars is not the real meaning of MS. The other-attribution thus serves to cast doubt on a definition and the whole sentence cannot be interpreted as the poster providing readers with the information that MS really means ‘multiple scars’. In (54), the reference to Wikipedia is supposed to add authority to the definition and the sentence can certainly be understood as the poster telling readers about the meaning of the term limbo. Georg Marko 272 (53) It seems that doctors feel that MS means "multiple scars". (54) The word „limbo“ is defined in Wikipedia as this: Latin limbus, edge or boundary, referring to the "edge" of Hell It might be objected that other-attributed definitions, especially those relativizing the meanings given, are cases of reporting a definition (from another text) rather than providing a definition and that they are therefore not metadiscursive, but rather intertextual. I agree that this type of definition certainly has different functions from self-attributed ones. This, however, does not change their fundamental metadiscursive nature - after all, speakers and writers always mention a meaning to a term that is being used elsewhere in the ongoing discourse, whether they at the same time undermine this definition or whether they actually corroborate it. Self-attributions are not only the prototypical form of definitions and the default option, any function with respect to the local management of the competence gap, hierarchies and power can only be fulfilled by selfattribution as other-attribution clearly denies any claim to special knowledge or expertise of the speaker. 4.8 Explicitness The fundamental question in connection with explicitness is: Is there a semantic connector, i.e. an element that explicitly signals that the current move is a definition? The two values of this variable are explicit and implicit. These will be explained and illustrated below. 4.8.1 Explicit Explicit definitions include semantic connectors, i.e. expressions overtly denoting the word-meaning relationship. These include, for instance, means, is defined as, is a (in anaphoric definitions, i.e. in defining in the narrow sense), and is called, is named, aka (‘also known as’) (in cataphoric definitions, i.e. in labelling). Here are a few examples (semantic connectors are underlined, as elsewhere in this chapter). In the anaphoric examples (55) and (56), the connectors are defined as and is another word for, in the cataphoric examples (57) and (58), the connectors are is called and aka. (55) […] men who exercised vigorously defined as breaking into a sweat more than five times a week - […] (56) He said I have thickening of the anterior leaflet of my Mitral valve with mild submitral chordal, which is another word for calcification build up. Power and status by definition 273 (57) […] blood pressure between 120/ 80 and 139/ 89 is called “prehypertension” […] (58) Yep, if you have migraine without headache with aura (aka painless or acephalgic migraine) 4.8.2 Implicit In implicit definitions, overt semantic connectors are missing. The meaning of a lexical items is usually marked by punctuation: it may occur in parentheses (separated by dashes or commas), between brackets or quotation marks/ inverted commas. In cataphoric definitions, where the lexical item follows its meaning, it is the item that is thus marked. Examples (59) and (60) are anaphoric definitions, in the former the meaning of diverticulitis occurs between brackets directly following the definiendum, in the latter the meaning of photophobia occurs after a comma also directly following the definiendum. In the two cataphoric cases (61) and (62), the definiens is put between brackets, bradycardia in the one case, LDN in the other. (59) We found out that she has diverticulitis (inflammation of the colon) (60) Have you been told that you have photophobia, sensitivity to fluorescent lighting? (61) In fact, low blood pressure can be a symptom of heart attack, as well as an abnormally low heart rate (bradycardia), problems with heart valves, and heart failure. (62) tried Low Dose Naltrexone (LDN) for first time - - good so far! It is difficult to assess the effect that explicit and implicit definitions have with respect to the secondary social functions of definitions. I assume that attention plays a role here and that constructions that background the defining process have a greater impact. The speaker’s or writer’s claim to a special knowledge that allows her to provide meanings to what are often very technical expressions will be more difficult to reject or relativize, if it is made in passing. Implicit definitions would be the more effective tool for underlining a competence gap and creating an imbalance of power. We have to treat this conclusion with caution, though, because we could very well argue the opposite, too, viz. that an explicit definition foregrounds the fact that the writer or speaker has a special knowledge. (I favour the former interpretation.) Georg Marko 274 4.9 Metalinguistics This variable is different because it is not actually concerned with different types of definitions but with a feature that distinguishes definitions from cases where we simply provide information about a phenomenon. In a broad sense, we could also classify this as definition, after all these phenomena are also normally represented by particular expressions. This means there is a difference between metalinguistic definitions and nonmetalinguistics definitions. 4.9.1 Metalinguistic definitions As this article is mainly about definitions as forms of metadiscourse, I have only talked about metalinguistic definitions. The class therefore does not need to be described any further. 4.9.2 Non-metalinguistic definitions In non-metalinguistic definitions, neither the phenomenon which a speaker or writer is talking about nor the information being provided about it is metalinguistically marked as a linguistic element or as linguistic meaning, i.e. no italics, quotation marks, etc. Grammatically speaking, the definiendum is used like an ordinary word, taking, for instance, articles (e.g. a heart attack) or modifiers (e.g. a serious heart attack) and appearing in the plural (e.g. heart attacks). The definiens is not limited to, or may not even contain, essential information; so it can be more elaborate than ordinary semantic meanings and/ or be concerned with nonessential, accidental aspects. Any semantic connector is not metalinguistic, but as general as possible, most commonly as the copula in is a/ an. Here are two examples of non-metalinguistic definitions. In both, the phenomenon is used in a generic noun phrase with the indefinite article, viz. a stent and a fever. While the information given in (63) on a stent resembles an ordinary formulation used in connection with the linguistic meaning of an item, the information provided in (64) mentions nonessential aspects of a fever, saying what one function could be - to be an indexical sign of a problem in the immune system - rather than mentioning what it really is. (63) A stent is a stainless steel coil inserted into a coronary artery to help hold it open. (64) A fever is a sign that the immune system is “working”. What I have said so far on non-metalinguistic definitions might suggest that there is a clear boundary to metalinguistic ones. To be honest, I do not believe that this is the case. I think that the distinction is scalar, i.e. a matter of more or less, rather than categorial, i.e. a matter of either-or. Power and status by definition 275 (64) thus is clearly a non-metalinguistic definition (if we can really use the concept of definition here at all). In (63), this is much less obvious and we could very well interpret it as a definition of the word stent. This section is just supposed to highlight the fact that it is sometimes very difficult to study definitions exhaustively, especially because of this fuzziness. Apart from this, there is nothing important to say about the opposition between metalinguistic and non-metalinguistic definitions. This fuzziness certainly poses a methodological problem, but the distinction is irrelevant for a discussion of the social effects of definitions. 5. The (hypothetical) study: outlook and conclusion Instead of a proper “Conclusion” chapter, I will here present the condensed outline of the study of the use of definitions in health forums to and for which this article was supposed to lay the foundation and set the agenda. All these consideration should form the foundation of, should feed into, and should be considered in, a study of the social functions and effects of definitions in the lay-to-lay interaction in health forums on the worldwide web. The research questions of this study could be: How, to what extent and by which linguistic means do posters to lay health forums use definitions and which sociocultural implications does their use of definitions have, with a particular emphasis on the construction of social hierarchies and power relations related to a competence gap? As patterns cannot be established by looking at individual texts, these questions require the analysis of a corpus of health forum postings (such as the one used as a source of examples in this article). Methodologically speaking, the questions require that we find, identify, classify, and quantify definitions in such a corpus. Based on the discussions of social effects of different types of definitions in the previous chapter, we could postulate the following principles. ‣ The use of definitions helps to create a competence gap between the person defining a term and other participants in the interaction. Implication:  Definitions >no definitions 7 7 Using the greater-than-sign means that the more definitions I find in the corpus, the more participants in the generally egalitarian scenario of the health forum appear to be interested in creating a competence gap and in creating hierarchies and Georg Marko 276 ‣ The less central a definition is for the understanding of the ongoing discourse, the greater its social effects (with respect to the aspects mentioned). Implication:  Cataphoric definitions (labelling) > anaphoric definitions (definitions in the narrow sense) 8 ‣ The more detailed a definition is, the more it indicates that a writer or speaker has expert knowledge in the field, which in turn means a greater impact on its social effects. Implication:  Exhaustive definitions > categorial definitions  Propositional definitions > lexical definitions  Generic definitions> exemplary definitions ‣ The less attention a definition attracts, the less participants in an interaction will see its possible social implications, which in turn means a greater impact on its social effects. Implications:  Lexical definitions > propositional definitions 9 ‣ The more speaker-centred a definition is, the more the speaker is in control and the more aspects of knowledge or power are associated with her directly, the greater the impact on its social effects. Implication  Speaker-definitions > collaborative definitions  Self-attributed definitions > other attributed definitions Taking these into consideration should help in answering the research questions. power relations. In this case, this is particularly relevant if I compare the data from my corpus to data from other corpora. I will also use this form of representation in the following. It always means that I have to look at the quantitative relations between different types of definitions and that the category (or the feature) on the left has a greater impact on the competence gap, hierarchies and power than the one on the right and that therefore the quantitative relation between the categories is relevant. 8 Alternatively, we may also argue that the exact opposite holds, i.e. the more central a definition is and thus the more important the speaker’s knowledge becomes in the given situation, the greater its role in achieving a secondary social function 9 The last two principles may form an opposition and it is difficult to judge which one is more important and whether we should claim that propositional definitions are more effective with respect to the competence gap and power (because of the detail they provide) or lexical definitions (because they can be given in passing and do not normally attract a lot of attention). Power and status by definition 277 Studying the local management of the competence gap, social hierarchies and power relations means that we should not stop at the definitions themselves, but should also take a closer look at the interactive cotext to see whether the implied claims to knowledge and power are accepted (e.g. by thanking a participant for a definition), enhanced (e.g. by praising a participant for a definition - You really know a lot about this stuff. or Wow, that’s so interesting), mitigated (e.g. by reclaiming knowledge for oneself and thus rejecting the unknowing role - I know (in reaction to a definition)), or rejected (e.g. by saying that a definition is wrong). These would have to be taken into account for a full appraisal of the social effects of definitions. 5.1 A methodological aside Let me end with describing some ideas concerning the main methodological challenge in a corpus-based study of definitions - or of metadiscourse, more generally, for that matter - viz. how to track them in a corpus. Considering that a definition consists of four components, the definiendum, the definiens, the semantic connector, and the metalinguistic marker, we could focus on any of these in our search for definitions. I will very briefly discuss the pros and cons of approaches based on these components. ‣ Definiens-oriented As an obligatory element of a definition, the definiens appears to be a good starting point for a search. However, there are no formal restrictions on how to formulate the meaning of a term so that there are no common formal features of this component. A definiens-oriented search thus seems infeasible. ‣ Definiendum-oriented Like the definiens, the definiendum is an obligatory element. Unlike the definiens, however, it is a short element normally consisting of a single word or a relatively short sequence of words, from a relatively limited set of candidates, e.g. loan words from Latin or Greek (e.g. vena cava), neoclassical compounds/ forms (e.g. allopathy, nephritis), and acronyms (e.g. TIA ‘transient ischaemic attack’, CABG ‘coronary artery bypass graft’) are typical examples. Definienda are, however, not fully predictable, i.e. focusing on these three types of word formation would still mean that we miss out on many expressions that are defined in the corpus. What is more, not all tokens of the above classes appear inside a definition. As a matter of fact, we would expect most words to be defined only once in a text, but to occur several times elsewhere. This means that a corpus search would yield many false positives, examples that fulfil Georg Marko 278 the search criteria without belonging to the target set (i.e. definitions in this case). ‣ Semantic connector-oriented The major argument in favour of starting our search at the semantic connectors of definitions is that this is a limited set. So searching for defined, define, definition, means, has the meaning/ with the meaning, called, named, aka, etc. should produce an output including a large group of definitions in a corpus. However, the results will still represent only a sample of the definitions actually used in a corpus because firstly, many definitions are implicit and therefore lack an explicit semantic connector and secondly, words such as especially means are too general. They are thus not only used in definitions, but also in a lot of other contexts. A search would thus again yield many false positives. ‣ Metalinguistic marker-oriented The arguments for and against looking at metalinguistic markers - e.g. italics, brackets, inverted commas, quotation marks - are similar to those used in connection with semantic connectors. These markers are often part of definitions so focusing on any of the devices mentioned will produce a large number of definitions. However, any of the above, whether italics, brackets, or quotation marks, are also commonly used for other purposes. Besides, not all definitions necessarily contain a metalinguistic marker. Possibly only a combination of these approaches could prove successful and would yield at least the vast majority of instances of definitions in a corpus. But whatever methodological focus we choose, the efforts appear to be worth our while as only a systematic search for definitions and their categories will allow deeper insights into the social implications of these acts. References Ädel, Annelie (2006). Metadiscourse in L1 and L2 English. Amsterdam & Philadelphia. PA: John Benjamins. Childs, Carrie (2015). “Formulations.” In: Todd Sandel/ Cornelia Ilie/ Karen Tracy (eds.) (2015). The International Encyclopedia of Language and Social Interaction. 634-638. Crismore, Avon/ Raija Markkanen/ Margaret Steffensen (1993). “Metadiscourse in persuasive writing: a study of texts written by American and Finnish university students.” Written Communication 10(1). 39-71. Eggins, Suzanne (1994). Introduction to Systemic Functional Linguistics. London: Pinter. Fairclough. Norman (1992). Discourse and Social Change. Cambridge [etc.]: Polity Press. Power and status by definition 279 Fill, Alwin (1993). Ökolinguistik. Eine Einführung. Tübingen: Narr. Garfinkel, Harold/ Harvey Sacks (1986, originally 1970). “On formal structures of practical actions.” In: Harold Garfinkel (ed.) (1986). Ethnomethodological Studies of Work. London: Routledge & Kegan Paul. 157-189. Halliday, Michael A.K. (1994). An Introduction to Functional Grammar. 2 nd edition. London [etc.]: Edward Arnold. Hyland, Ken (2005). Metadiscourse. Exploring Interaction in Writing. London & New York: continuum. Kettemann, Bernhard / Georg Marko/ Eva Triebl (2010). “‘I have MS. MS doesn’t have me.’ Social identity construction in the discourse of multiple sclerosis forums.” In: Rudolf de Cillia/ Helmut Gruber/ Michał Krzyżanowski / Florian Menz (eds.) (2010). Diskurs - Politik - Identität/ Discourse - Politics - Identity. Festschrift für Ruth Wodak zum 60. Geburtstag. 355-367. Lupton, Deborah (2003). Medicine as Culture. 2 nd edition. Los Angeles [etc.]: Sage. Marko, Georg (2010). “‘Your story has made me feel angry on your behalf.’ The roles of empathy and equality in the construction of health identities in lay-tolay interaction on chronic diseases.” Paper presented at CADAAD 2010 (Critical Approaches to Discourse Analysis Across Disciplines). University von Łódź (Polen) (13.-15. September 2010). Marko, Georg (2012). “My painful self. Health Identity Construction in Discussion Forums on Headaches and Migraines.” AAA - Arbeiten aus Anglistik und Amerikanistik 37/ 2. 245-272. Mauranen, Anna (2010). “Discourse Reflexivity - A Discourse Universal? The Case of ELF.” Nordic Journal of English Studies 9(2). 13-40. Penz, Hermine (2016). “The uses and functions of metadiscourse in intercultural project discussions on language education.” In: Nikola Dobrić/ Eva-Maria Graf/ Alexander Onysko (eds.) (2016). Corpora in Applied Linguistics Current Approaches. Newcastle upon Tyne. Cambridge Scholars Publishing. 95-120. Schiffrin, Deborah (1980). “Meta-Talk. Organizational and Evaluative Brackets in Discourse.” Sociolinguistic Inquiry: Language and Social Interaction 50. 199-236. Sinclair, John (2005). “Language as a string of beads: Discourse and the M-word.” In: Elena Tognini- Bonelli/ Gabriella Dell Lungo Camiciotti (eds.) (2005). Strategies in Academic Discourse. Amsterdam & Philadelphia. PA: John Benjamins. 163-168. Swales, John M. (1990). Genre Analysis. English in academic and research settings. Cambridge. CUP Thompson, Geoff (1996). Introducing Functional Grammar. London [etc.]: Edward Arnold. Georg Marko Department of English Studies Karl-Franzens-University Graz Austria