…or not to be. The Strategic and Non-Strategic Use of Negative Identifiers in Online Forums Eva Triebl Negative identifiers, that is, expressions of the type [I + copula + not + indefinite NP], serve to defeat explicit or implicit identity claims present in the immediate co-text, the situational context or the wider cultural context of the utterance. This means that on the one hand they can serve as situational strategic moves in a conversation, supposed to guide the interpretation of the speaker’s utterance, and that on the other hand they might have implications for identity construction that go beyond the immediate conversational situation and can thus be related to the wider, socio-cultural context of selfreflexive identity construction and individualization characterizing late modern society. In this paper, I will set up a formal-functional framework of negative identifiers, arguing that they are functionally similar and can thus be approached like discourse markers. Based on a pilot study of a derived corpus of 85 negative identifiers in context, I will then demonstrate how such a framework could be operationalized in a corpus-based Critical Discourse Analysis of negative identification. 1. Introduction I’ve never been a conformist, I’m by no means a racist and I wouldn’t consider myself an expert. Why do people state what they are not when they could just positively assert what they are? And how does that relate to the cotext and context of particular conversational situations or even the broader context of changing conceptualizations of identity in late modernity (cf. Giddens 1991, Beck 2002)? In this paper I will present a formalfunctional framework for the analysis of such utterances which can be AAA - Arbeiten aus Anglistik und Amerikanistik Band 40 (2015) · Heft 1-2 Gunter Narr Verlag Tübingen Eva Triebl 248 applied when addressing social questions about beliefs, identities, and relationships. Negatives, rather than just being the pragmatically less favoured counterpart to affirmatives (cf. Leech 1983: 100), ―display significant contextual and interpersonal meanings in addition to their ideational sense‖ (Jordan 1998: 706) in that they are used to defeat explicit or implicit identity claims present or felt to be present in the immediate co-text, the situational context or the wider cultural context of the utterance. Indeed, negative identifiers do much more than just provide (non-) information: for instance, while the utterance I‟m not a hipster could be used assertively, as a response to a question by another speaker for instance (e.g. Are you a hipster? - No, I‟m not a hipster), it could also serve as a disclaimer modifying the interpretation of some other part of the speaker‘s turn (e.g. I‟m not a hipster, but I like my moustache), as a defeat of identity claims felt to be present in the situational context of the conversation (for instance, in a discussion in a beard forum, the speaker may want to pre-emptively distance himself - or, more rarely, herself - from possible identity claims evoked by his participation in such a forum), or even to dissociate the speaker from identity categories perceived to be of relevance in the wider cultural context (if I am wearing a shirt saying I‟m not a hipster, this could suggest that ‗hipsterness‘ is a culturally salient identity category from which I want to distance myself). In the following I will, firstly, develop a taxonomic framework of the forms and functions of negative identifiers. On the one hand, negative identifiers, except if used assertively, function like discourse markers in that they can serve to strategically structure and modify ongoing discourse (cf. Aijmer & Simon-Vandenbergen 2004) rather than to provide information, which means that their interpersonal function overrides their ideational function (cf. Kim 2010: 86). On the other hand - and this is where the broader socio-cultural context comes in - negative identifiers can indicate how speakers, by contrasting themselves with social groups, ideologies or other categories in particular communicative situations, conceptualize their self-identity beyond the immediate conversation. In the second part of this paper, I will present a pilot study on negative identifiers in which I operationalize this framework by qualitatively and quantitatively analysing a derived corpus of 85 instances of negative identifiers used by participants in online forum discussions. 2. Why no(t) Viewed in isolation, negative utterances are an odd thing: they are grammatically more complex and consequently harder to process, some see them as ‗non-events‘ which are cognitively less salient than their positive counterparts (cf. e.g. Givón 1993: 191), and in terms of Grice‘s Coop- …or not to be 249 erative Principle, they are less informative than affirmatives (cf. Leech 1983: 100). Still, it seems that people routinely use negative identifiers in authentic discourse contexts - for example, 17% - i.e. one in six - of all occurrences of the construction I‟m a in the COCA (Corpus of Contemporary American English) are not-negated. The reason for this is that while talking about what is not the case is, indeed, not very informative, negative identifiers have interesting functions for their textual and contextual environments. Linguistic expressions whose function is largely non-propositional in that they do not convey any information themselves but signal how the speaker wants an utterance to be understood in relation to the text preceding or following it and manage the interaction between speaker, hearer and message, are dealt with under the label of discourse markers in pragmatics (cf. Noveck & Sperber 2004). Discourse markers have - under myriad different names 1 and with diverse analytical categories - been extensively studied by linguists with different research interests (cf. Fung & Carter 2007: 410). Among the works that have paved the way for a wealth of research on the topic are Schiffrin (1987), Blakemore (1992), Fraser (1996), Brinton (1996), Jucker and Ziv (1998), and Aijmer (2002, 2013). As Brinton (1996: 30) explains, there are many different definitions of discourse markers depending on which aspect is deemed most relevant. The following criteria for categorizing a word or a particle (or, as in my argumentation, a clause) as discourse marker stand out as important according to Brinton‘s overview of the concept: firstly, discourse markers generally have little semantic or truth-conditional meaning, but are used for pragmatic reasons and occur mainly in spoken discourse. Popular examples of these ―seemingly empty expressions‖ are actually, oh, right, well, I mean and you know (cf. Brinton 1996: 29). There is agreement that discourse markers signal how an utterance is to be related to its co-text or its context: they can, for example, indicate the relevance of one utterance in relation to another, they can serve to focus the attention on the upcoming utterance or ‗comment on‘ a sequential relationship between two utterances. While for some the structuring or demarcating function of discourse markers is the most important (cf. Erman 1986), others mainly see them as conventionalized response signals (cf. Schourup 1985), as continuity and floor keeping devices (cf. Crystal & Davy 1975), or as means of establishing interpersonal relations (cf. Stubbs 1983). 1 As Brinton (1996: 29) explains, there is a ―plethora‖ of over twenty different names used to designate the ―seemingly empty expressions found in oral discourse‖, of which the term discourse marker is the most frequently used. As Jucker and Ziv (1998: 2) argue, this term is the one with the ―widest currency and with the least restricted range of application‖, which is why I will adopt this term in my study, too. Eva Triebl 250 As for the functions of discourse markers, Aijmer and Simon- Vandenbergen (2004: 1782) remark that despite the constantly growing scholarly interest in discourse markers, there is still a ―lack of generally accepted terminologies and classifications, [and] uncertainty regarding essential formal, semantic, and pragmatic characteristics […].‖ 2 Generally, most functional models distinguish between either two or three main functions of discourse markers. Aijmer (2013: 17) introduces Östman‘s (1995: 99) three functional categories, viz. discourse organizing (i.e. moving a text forward and structuring it: for example, Bob Ross, painter and host of the legendary TV show The Joy of Painting, often uses the discourse marker now to indicate the upcoming description of a new step in the painting process 3 ), interaction-signalling (e.g. signalling understanding or eliciting responses through tags such as right), and attitude/ involvement-signalling (i.e. signalling epistemic and deontic commitment to or evaluation of the relevance of the proposition expressed). The two characteristics of discourse markers most relevant regarding the analysis of negative identifiers are, firstly, their function as metalinguistic monitors used by speakers to reflexively manage how their utterances fit with what they have said or what they are about to say on the one hand, and with the context - conceived here cognitively, i.e. in terms of what information is assumed to be given at a particular moment in a particular communicative situation - on the other hand. Thus, discourse markers ―emerge as overt indicators of (or windows on) ongoing metalinguistic activity in the speaker‘s mind‖ (Aijmer 2013: 4). Secondly, discourse markers serve as contextualization cues, which Gumperz (1996: 379) defines as a ―cluster of indexical signs […] that jointly index, that is, invoke, a frame of interpretation for the rest of the linguistic content of the utterance.‖ This means that discourse markers, to take a Relevance Theoretical perspective (cf., e.g. Blakemore 2002: 47), signal to hearers what inferential route they should take in a particular communicative situation, providing a kind of ‗catalyst‘ restricting the interpretatory options of an utterance in a given context. Thus, discourse markers ―do not have a fixed meaning but a meaning potential‖ 4 (Aijmer 2013: 12), but can index various dimensions of a communicative situation depending on the context. 2 For a comprehensive overview of the features and functions that have been claimed to define discourse markers, see Brinton (1996: 32-35). 3 As can be seen, for example, in this video: http: / / www.youtube.com/ watch? v =MghiBW3r65M 4 Although I am not sure if this property can be claimed to be specific of discourse markers - after all, it is a general assumption in theoretical pragmatics that linguistic elements or structures are ―systematically underdetermined and must be contextually enriched‖ to arrive at their fully-fledged meaning (Grisot and Moeschler 2014: 8). …or not to be 251 Negative identifiers share these two important properties with discourse markers, i.e. they are both reflexive and indexical. Crucially, as Ochs (1996) suggests, discourse markers - and negative identifiers - can ―reach beyond the utterance‖ (Aijmer 2013: 13) and index other dimensions, among them social identity. This leads me to the questions I seek to address in my study: Is it possible to identify formal-functional regularities in the use of negative identifiers to manage the ongoing conversation on the one hand and to index aspects of the speaker‘s identity on the other? How does that relate to the value of the indefinite NP in the structure [I + copula + not + indefinite NP]? And, ultimately, can we relate the situated use of this structure in context, if examined across a large sample of authentic data, to broader social research questions like the reconceptualization of self-identity in late modernity? In the rest of this paper, I will present an approach that might help find answers to these questions. 3. From form to function 3.1. Methodological challenges In the following, I will set up an initial framework of formal and functional categories of negative identifiers and apply it to the analysis of authentic data. I am drawing on other studies that have investigated the relation between lexico-grammar, utterance function and discourse context as well as on a pilot study on negative identification with feminism in online forums. 5 The initial difficulty of conducting an empirical study to yield insights about the functions of negative identifiers in authentic informal discourse is that, since they work similarly to discourse markers and thus have a contextually variable meaning potential, they can only be accounted for by explicating how their linguistic properties interact with the co-text and contextual factors to serve particular discourse functions (cf. Noveck & Sperber 2004). However, as Adolphs (2008: 8) explains, a study that has the goal of empirically exploring the patterns emerging from the use of a particular linguistic structure necessarily starts with form, and not with function. Taking linguistic forms as departure point to analyse functions sounds like a relatively straightforward thing to do at first and has proven to work well with questions that can be answered by looking at forms alone: for instance, studies such as Louw (1993) and Sinclair (1991) have used corpora to analyse semantic prosodies, i.e. recurrent associations, of lexical items whose meaning is also defined by the - mostly evaluative - 5 This study examined a derived corpus of 126 instances of negative identification with feminism in online discussion forums and was presented at the CADAAD Conference 2014 in Budapest. Eva Triebl 252 meanings of elements they collocate with. An example of such an item is set in, the (subject) collocates of which are usually negative, e.g. pain, rain or panic. In the case of negative identifiers, things are a bit more complicated, because we can neither determine their function by looking at their form or forms in their immediate co-text alone, nor postulate formal categories without making a priori assumptions about their functions. As for the first difficulty, negative identifiers are often used non-ideationally to guide the interpretation of what is explicitly stated in the proximate co-text or - and this is the methodological challenge - in the more distal co-text or even implied by the context. For example, while I‟m not a feminist can be a response to an identity claim raised by another participant in the previous turn, i.e. within the proximate co-text of the negative identifier, it can also be used to dissociate the speaker from an identity category that is part of the situational context of the conversation. To give an example, the forum in which a particular discussion takes place might be a feminist discussion board, which could be reason enough for a speaker to distance heror himself from feminism and which should therefore be considered in the analysis. But to take this contextual aspect into account, looking at the immediate co-text of the utterance alone is obviously not enough. The key importance of co-textual and contextual aspects for creating functional profiles is a theoretical and methodological challenge in corpus pragmatics and has already been recognized as such by other scholars. For instance, Carretero et al. (2014), who study the functions of expressive speech acts used in a corpus of e-forum history logs of students, stress ―the need to have access to non-immediate co-text in order to assign a specific pragmatic value to a given sentence.‖ For the same reason that it is difficult to determine the function of negative identifiers without taking their co-texts and contexts into account, it is hardly possible to separate their formal properties from their functions: for example, the position of a negative identifier within a sentence can already be seen as having functional implications (for the interdependence of form and function, cf. Hasan 1999 and Swales 2004). What this means for my study is that it requires the a priori postulation of an analytical framework which defines textual, co-textual and contextual aspects that guide the corpus design and structure the analysis. The initial formal-functional framework which I will propose in the following should thus be seen as a first analytical grid whose categories are likely to be adapted, modified or omitted once it has been applied to the analysis of actual language in use. 3.2. Three analytical levels My research on negative identifiers seeks to answer questions on three levels of discourse, which correspond to those established by Fairclough …or not to be 253 (1989: 25) and further developed by Marko (2008: 101) to describe the methodological practice in Critical Discourse Analysis. I will adapt this model to operationalize the objectives of my study, which can be located at the level of textual form, the level of situational context and the level of wider social significance of the discourse practice under scrutiny. Description of forms and local functions of negative identifiers in their proximal co-text and context Interpretation of (patterns of) meanings created by the use of negative identifiers in authentic discourse contexts Critical evaluation of how these patterns of meanings can be related to social research questions on language and identity Fig. 1: Methodological practice of CDA, adapted for my research purposes. While the impetus for conducting research in CDA is usually situated at the third level of this model, i.e. at the level of the wider socio-cultural significance of a particular discursive phenomenon, the actual analysis in my case starts with the description of a concrete linguistic structure in use, which is examined in its co-text and proximal context, and, by abstracting from the concrete by means of a quantitative analysis, works its way up to the evaluation of socio-cultural significance. In other words, contrary to methods in CDA such as Fairclough‘s (2003, based on Bhaskar‘s (1986) ‗explanatory critique‘), which proceed top-down by identifying a ―social problem which has a semiotic aspect‖ (ibid.: 209) and work their way down to textual analysis, I am reversing this order by taking a corpus-pragmatic approach, using a ―speech function that is generally realized in a small number of variant patterns‖ (Jucker, Schreier & Hundt 2009: 4) as my departure point. What this means for the operationalization of my research goal - to start with a concrete form to address social questions about beliefs, identities, and relationships - is that my analysis takes place on three analytical levels which differ in their degree of abstraction: at the most basic, specific level, I qualitatively analyze the forms and local textual and interpersonal metafunctions of negative identifiers in their immediate coand context. By taking a quantitative approach, examining which patterns of using this structure in particular discourse contexts emerge across larger samples of data, I abstract from the concrete, situated use of a linguistic structure to recurrent relations between form and context. By examining in which discourse contexts negative identifiers become a salient linguistic choice, I can eventually ―draw conclusions about the socio-cognitive resources standing behind these [recurrent linguistic choices]‖ (Marko 2008: 79). Eva Triebl 254 I have formulated the following research questions on the three analytical levels just introduced. On the first plane, i.e. at the level of textual form, my analysis describes the formal properties of negative identifiers and their interaction with their co-text. As mentioned earlier, it is difficult to neatly separate form from function in the case of negative identifiers, because while the form of the structure per se is more or less fixed, describing its textual links with the co-text has functional implications. The research questions at this analytical level are: 1. Description of textual form in its proximal co-text: What are the formal properties of negative identifiers? I.e.: ‣ The structure itself ‣ The structure in relation to its co-text, i.e. what role do negative identifiers play for the local information structure of the conversation? On the second plane, the textual meanings created by negative identifiers are examined, which requires taking into account the context, i.e. aspects of the concrete communicative situation as a dynamic event with particular semiotic output (the text) and specific contextual configurations (who is talking to whom about what, where, when, and why) (cf. Marko 2008: 79). At this analytical level, features of online forum discussions as my main data are taken into account (e.g. the topic of the forum) because, as Adolphs (2008: 31) stresses, interpreting the functions of negative identifiers as discourse markers ―relies heavily on the context in which it is used.‖ 2. Interpretation of discourse meanings at the level of situational context: What are the functions of negative identifiers in authentic communicative situations? I.e.: ‣ Strategic functions: How do speakers use negative identifiers strategically to manage possible interpretations of what they say in concrete communicative situations? ‣ Non-strategic functions: In how far do negative identifiers index something about the speaker‘s social identity relevant beyond the immediate conversational situation? On the third plane, i.e. at the level of wider socio-cultural significance, I will critically evaluate whether and, if so, how patterns of conceptualizations of self-identity can be viewed in the light of wider socio-cultural questions of identity construction in late modernity. Since this paper presents a pilot study for a larger project on negative identification, I will not draw any major conclusions about the socio-cultural implications of the discourse meanings identified at this point: instead, the following …or not to be 255 should be seen as an exercise in testing and modifying a framework so that it can be applied in Critical Discourse Analysis - so it is a paper about asking better questions rather than as a presentation of definite answers, which I do not have at this point. 3.3. An initial formal-functional framework 3.3.1. The structure and its co-text As for the formal properties of negative identifiers per se, we can consider possible variations of the structure [I + copula + not + indefinite NP] on the one hand, and the conceptual category of the indefinite noun phrase (e.g. the category the speaker dissociates her-or himself from) on the other. The former can appear in the variants listed below. It needs to be mentioned, though, that it is one goal of my study to find out if and to what extent formal variations of the negative identifier can be related to functional differences:  Contraction: I‟m not  Tense/ aspect: e.g. I‟ve never been  Modality: e.g. I may not be  No-negation: I‟m no  Adverbs: e.g. I‟m not really, I‟m definitely not As for the conceptual category of the noun phrase, the following categories might play a role:  Profession: I‟m not a lawyer.  Lifestyle: I‟m not a vegetarian.  Ideology: I‟m not a feminist.  Ascribed negative ideology: I‟m not a racist. 6  Evaluative: I‟m not a hater. A first step in creating a formal-functional profile of a discourse marker is to establish the various positions in which it occurs (cf. Cuenca & Marín 2009, Erman 2001). Arguably, the position of negative identifiers in a conversation is a formal feature likely to correlate with their function. Consider, for example, the following examples taken from the corpus of instances of negative identification with feminism: (1) I‟m not a feminist per se, but I am a liberal white woman. 6 This category includes attitudes that people cannot positively associated with due to their inherently negative connotations. It is, for instance, odd - if not impossible - to say I‟m a racist. Eva Triebl 256 (2) It‟s interesting to see that you guys have listed so FEW female actors as potential legends - probably due to that lack of great roles for women in modern films no, I‟m not a feminist. While in (1), the negative identifier is reflexively used as an anticipatory disclaimer, supposed to pre-emptively avoid possible implications of the coordinate clause I am a liberal white woman, the negative identifier in (2) refers back to and modifies the interpretation of the speaker‘s utterance preceding it. Concerning the relation between the negative identifier and its co-text, I think the categories defined below will be relevant in my study. As can be seen, the local function of the identifier closely relates to its position in the sentence and in the turn, i.e. to the information structure of the utterance. Also, the framework already shows how difficult it is to neatly separate co-text from context. 1. Endophoric reference: The negative identifier refers to something internal to the speaker‘s utterance(s). a. Anaphoric reference: the negative identifier refers back to something the speaker her/ himself has said  Within the same sentence: as a subordinate or coordinate clause (e.g. I respect women, but I‟m not a feminist.)  In a previous turn b. Cataphoric reference: the negative identifier anticipates something the speaker is about to say  Within the same sentence: as a superordinate or coordinated clause (e.g. I‟m not a feminist, but I respect women.) 2. Exophoric reference: The negative identifier refers to something external to the speaker‘s utterance(s). a. Utterance-referring: to something another speaker has said b. Paratext-referring: to text belonging to or surrounding the current text without really being part of it (e.g. forum/ thread titles) 7 7 The terms endophoric and exophoric are used here for the relationship between an identifying phrase and anything that relates to it in content. It does not refer to the relationship between two identifying phrases. Endophoric thus does not mean that, for instance, the NP a feminist has been used in the same utterance, nor does exophoric mean that it has been used outside of this utterance. …or not to be 257 3.3.2. Contextual aspects As Aijmer (2013: 9, referring to Verschueren 1999: 111) states, ―what makes pragmatic markers unique is that they ‗allow context into the linguistic analysis‘.‖ Likewise, Adolphs (2008: 31) stresses that the ―interpretation of a particular function of a speech act expression relies heavily on the context in which it is used.‖ Having ‗context‘ figure as an important criterion in a linguistic analysis has to be taken with a grain of salt because, as Aijmer (2013: 9) remarks, discourse markers, precisely because of their context-dependence, provide a ―testing ground for contextual theories of meaning.‖ While I will not go into much detail about different theories of context at this stage, I would like to briefly explain how I conceive of and how I will handle aspects that are ‗contextual‘ in that they are external to the structure of interest, viz. negative identifiers. The notion of context in my analysis and in the model I introduced above firstly refers to aspects of the communicative situation that are not part of the proximate or more distal co-text, but elements of the situation that might be of import when analysing negative identifiers. For instance, the purpose of the discussion forum in which the discussion takes place or aspects of the participants‘ social identity feature in this category (e.g., it might make a difference whether a discussion on feminism takes place in an academic forum or on the website of Men‟s Health). Secondly, the level of context in my study is the level of analysis that examines the meanings that negative identifiers create as instances of language in use in authentic communicative situations, i.e. in interaction with their context. It is thus at this second analytical level where I want to find out what communicative work negative identifiers do, i.e. which pragmatic functions they fulfil in the conversation on the one hand and which identity-constituting functions they might serve beyond the immediate conversational situation on the other hand. While it is the aim of the qualitative analysis of my data to find out what pragmatic and nonpragmatic functions negative identifiers fulfil and how they can be related to formal aspects, the goal of the quantitatively-oriented part of the analysis is to reveal if patterns of form-function correlations can be detected across a larger sample of data. Context, in pragmatics and probably in all fields of linguistics, is a much theorized and contested concept. As Adolphs (2008: 31) explains, mainstream pragmatics has traditionally mainly relied on invented contexts for the interpretation of utterance functions. The analysis of authentic language in use has led to the development of various theories of context, which is one of the reasons why there are now quite heterogeneous notions of context in pragmatic research. Along with Aijmer (2013: 15), I will draw on Ochs‘ (1996) model of context, which posits five sociocultural dimensions (social identity, social act, activity, affective stance and epistemic stance), allows for the indexing of more than one dimen- Eva Triebl 258 sion at a time, and recognizes the possibility of utterances having implications beyond the immediate situational context. For example, a speaker might primarily dissociate himor herself from a particular social category to take epistemic stance in the sense of ―I‘m not an expert in this matter,‖ and this might secondarily index his or her social identity (being a layperson might be important for this person‘s social identity in a particular context - imagine, for instance, medical discussion forums which can serve to create an in-group of non-experts). This broad understanding of context seems important for the purposes of my study: after all, it will eventually feature a critical element, trying to explain how situated linguistic choices form patterns that might have socio-cultural implications far beyond the immediate conversational situation. For the purposes of my pilot study, the following elements of the situational context seem to be important and will therefore be considered in the corpus compilation and analysis. Medium of communication As already mentioned, the data I will use for my analysis are discussions taking place in forums and similar online communication platforms. The reason why I decided on this type of data in the first place is that these platforms allow participants to interactively negotiate paradigmatically defined - i.e. with reference to shared features rather than social relationships - social categories, such as particular lifestyles (cf. Marko 2012: 249). According to Hodkinson (2007: 625), ―online communications have proved to be a significant factor in the development of fragmented, fluid patterns of individual identity.‖ While I will not elaborate in much detail on the issue of new forms of identity in this paper, suffice to say here that online forums, because they enable people to construct a global identity (cf. Arnett 2002: 777) appear to be an ideal site for studying how new forms of identity are created and negotiated in discourse. Online discussions are also an interesting medium of communication from a linguistic perspective: for example, recent studies such as Landone (2012), which investigates discourse markers and politeness in a digital forum, have observed that written dialogues, though different from oral conversations because of their asynchronous nature and the absence of signals such as body language and intonation, share many properties of spoken interaction, among them the use of discourse markers. In the case of negative identifiers, I would even assume that they occur more frequently in online discussions, because making statements about one‘s (non-identity) seems to be an important feature of Internet communication in general (just think, for example, of forum signatures 8 ) - a reason might be the absence of other linguistic and non-linguistic identity-signalling elements 8 A forum signature is the message that always appears at the bottom of one‘s posts. (cf. http: / / wiki.scratch.mit.edu/ wiki/ Scratch_Discussion_Forums) …or not to be 259 such as dialects, clothes etc. and the larger numbers of interactants not knowing each other (outside the forum). Finally, online discussion forums ideally lend themselves as data for my pilot study because they are easily accessible and seemingly endless in number, which is advantageous for a quantitatively-oriented study. Paratextual elements Another contextual aspect I am taking into account in my pilot study is the paratext, i.e. text belonging to or surrounding the current text without really being part of it. In my pilot corpus, I include information about the website, the topic of the discussion and the title of the thread in the header of the file. The reason for this is that dissociation from particular social categories might be influenced or even prompted by the topic of a discussion or the purpose of the website in general. Participants Theoretically speaking, who speaks to whom is a key contextual feature: for example, whether a discussion about feminism takes place among women only or in a mixed group might have an impact on the participants‘ self-representations as reflected in their arguments, their style etc. Likewise, features such as age, ethnicity, etc. seem to be relevant information about the participants. While I will control the larger corpus to be used for the main study of negative identification for these features, the only information about the participants included in this pilot corpus is via the forum name, purpose and target audience. 9 3.3.3. Meanings in context: functional aspects As mentioned earlier, postulating functional categories before actually carrying out a study is not unproblematic because it means imposing categories on the data which might turn out less relevant than previously assumed. Still, based on previous research on discourse markers and my (preliminary) study on negative identification with feminism, the following functions of negative identifiers might play a role in my analysis: 9 This has practical and ethical reasons. Practically speaking, this is a pilot corpus serving to explore the kind of data to be analysed for the main study and controlling Internet data for these features is nearly impossible. For the main study, I am therefore planning to create a derived corpus from existing corpora of online discussions, which include more detailed information about the speakers. Ethically speaking, creating a corpus from online forum discussions requires maximal anonymity of the speakers whose utterances I am examining, which is another reason why I decided not to control this pilot corpus for the mentioned features (cf. Fielding, Lee & Blank 2008). Eva Triebl 260 Endophoric negative identifiers Relation of negative identifier to co-text Example of co-text Relation of co-text to negative identifier Cataphoric modification: the negative identifier guides the interpretation of the co-text following it. but I support equality. 10 Contrast because I don’t believe people can be bundled into categories. Reason so I don’t share your opinion on this subject. Consequence In fact, I hate the word with a vengeance. Intensification Anaphoric modification: the negative identifier modifies co-text preceding it. I'm for equal rights, but Contrast I like men’s attention, so Reason I can’t say I support your views here because Consequence Exophoric negative identifiers Referring to another speaker’s utterance A: You are basing all your arguments on the feminist assumption I presented to you in my first post […] B: I don’t subscribe to any particular group. I’m not a feminist. Rejection of another participants’ identity claim. Referring to paratext (i.e. to an identity category present in the context of the communicative situation) Example: a user of the explicitly feminist forum www.reddit.com initiates her turn by stating that she is not a feminist as such. Disclaimer: the speaker, aware of identity claims potentially implied in the context of conversation, uses a negative identifier as a disclaimer. Table 1: Initial formal-functional framework of negative identifiers. 10 The examples in this table are mostly taken from the pilot study on feminism mentioned earlier, but were partly modified for reasons of space. …or not to be 261 4. The pilot study I am currently testing the initial formal-functional framework against a larger sample of authentic data in order to find out which categories actually play a role and if negative identifiers can be seen as cues to identity concepts beyond the immediate conversational situation - and how such a correlation between the situated use of a concrete linguistic form and broader social questions can be explicated. Eventually, my study seeks to demonstrate how we can bridge the gap between micro-linguistic analysis of a concrete structure and critical analysis of the wider sociocultural implications of the use of this expression in context. The data I am using for my pilot study is a derived corpus of currently 85 instances of negative identification used by participants in online discussion forums (I am constantly enlarging this corpus, so this paper is a snapshot). A derived corpus is a corpus which consists of the target structure, i.e. negative identifiers and the co-text that is relevant for their interpretation and which is, in my case, enhanced with contextual information required to arrive at a sufficiently full scenario of the communicative situation in which the structure is used. It is derived because it consists of - variously enhanced - concordance lines (or their equivalents) yielded by online searches rather than of full - whatever this may mean - texts. Creating such a corpus is somewhat unusual in the study of discourse markers: firstly, most studies use existing corpora that were not specifically compiled for the study of the said structure (e.g. the CANCODE corpus in Fung and Carter 2007, the Bergen Corpus of London Teenager Language (COLT), and the London-Lund Corpus (LLC) in Erman 2001; the CAP (Corpus Audiovisual Plurilinguë) in Cuenca & Marín 2009). Secondly, studies in pragmatics in general and of discourse markers in particular have examined corpora of spoken language, which is unsurprising since discourse markers prototypically occur in spoken language. There are, however, some studies which have investigated Computer Mediated Communication to answer pragmatic questions and which have convincingly argued for the suitability of online discussions as data for this kind of study. For example, Carretero et al. (2014) conducted a corpus-based study of expressives in online communication. They explain that social computing media such as blogs and e-forums are invaluable sources of information about online communication and ideally suited for pragmatic studies because on the one hand, they share many features of oral discourse including the use of discourse markers, and on the other hand, they are ―complete transcriptions of the linguistic production of the participants‖ (ibid.: 262). Generally speaking, it is more the interactive di- Eva Triebl 262 mension that makes people use discourse markers than the medium of communication by itself. As for the criteria guiding the compilation of my corpus, I am using customized Google searches which allow me to collect occurrences of the structure [I + copula + not + indefinite NP] (and its formal variants, see 3.3.1.) in forums and threads. In contrast to the previously mentioned first study on negative identification with feminism, which I used to set up the initial formal-functional framework introduced above, I am not specifying the value of the indefinite NP for this study - after all, I want to find out what identity categories people dissociate themselves from, and if there are patterns of categories recurring in connection with particular functions of negative identifiers. To prepare my data for the quantitative part of my analysis, which employs the concordancing software WordSmith Tools 5.0, I annotated my samples to be able to easily retrieve the following information:  Different speakers  The negative identifier itself  The indefinite NP the speaker negatively identifies with  Name of the forum and the topic of the discussion/ title of the initial thread To provide an example of my annotation system, let us look at the following extract randomly selected from the mass of hits my search yielded: <forum subject: diabetes.co.uk: the global diabetes community> <thread>I'm cold! </ thread> Since going low carb and losing a lot of weight. The winter has well and truly arrived. I've worked outside for a number of years and have never felt the cold even when it's freezing. I'm feeling it now even the thermals aren't working! Has anyone noticed this? <NI>And no I'm not a <IC>southern softie! </ NI> As can be seen, what I did with this sample was to add information about the forum subject in the file header (which I framed with triangular brackets so that it is easily searchable with a concordancer, but not treated as part of the text to be analysed by the software). I also indexed the title of the thread starting the discussion (in this case, the speaker himor herself initiated the discussion) and tagged the negative identifier as well as the identity category it is used to dissociate the speaker from. This makes it possible to search, for example, for all categories speakers …or not to be 263 negatively identify with by means of a concordancer, i.e. just by looking for the tag. 4.1. Qualitative analysis I will now look in more detail at the example presented above and see how it could be qualitatively analysed according to the provisional formal-functional framework I set up earlier. The negative identifier is used as an assertion, and the fact that it is initiated by ―and no‖ is interesting because no does not only intensify the negative identification by cataphorically referring to the polarity of the clause, but also makes the negative identifier sound like the rejection of an explicit identity claim which cannot be found in the immediate coand context. In other words, the speaker rejects an identity category which s/ he her/ himself just brought up. The negative identifier can thus be classified as endophoric, anaphorically referring back to and modifying possible interpretations of what the speaker has said before (namely that s/ he is constantly cold etc.). The category of the indefinite NP is interesting because it can, as Ochs‘ (1996) model of the socio-cultural dimensions of context predicts, be seen as indexing two aspects of the speaker‘s concept of self-identity at the same time: the negative identifier is used as anaphoric disclaimer supposed to avoid association with ‗southern softiness‘. This, on the one hand, shows that he (I suppose the speaker is male) thinks that complaining about feeling cold might expose him as ‗softie‘ and, on the other hand, shows awareness of stereotypes relating to geographic and climatic affiliation (and acting upon the presupposition that ‗people from the South are softies‘). What I also consider relevant for the interpretation of this statement in this context is that the forum in which this discussion takes place is a diabetes forum which, we might assume, mainly contains posts from people affected by this condition. Contrasting oneself with ‗southern softies‘ in this context is interesting because it might be interpreted as showing that the speaker, by invoking two identity categories which have nothing to do with his condition (qualities he considers to present a contrast to the concept of a ‗softie‘ and being ‗cool‘ as opposed to other people from the South), could suggest that having diabetes is not an identity category he perceives as central enough to bring it up (maybe just because it is evident from his participation in the forum anyway). This brief exploratory analysis of one instance of a negative identifier in use already has interesting implications for the formal-functional framework I postulated earlier because it shows that data is hardly ever as neat and easily categorizable as taxonomies suggest. In this case, for example, the co-text of the negative identifier (i.e. what the speaker says about feeling cold) is as important for the interpretation of its functions as the situational context (i.e. that we are dealing with a diabetes forum). By contrasting himself with a ‗southern softie‘, the speaker indexes differ- Eva Triebl 264 ent aspects of his identity, depending on what is seen as the identity category ‗southern softies‘ can be assigned to: is it a gender identity category (‗softies‘ as opposed to, say, ‗real men‘), a geographic (‗cool northerners‘ versus ‗softies from the south‘), or a climatic stereotype (‗weatherresistant northerners‘ versus ‗weather-sensitive southerners‘), and how does the speaker‘s negative identification with these identity categories relate to his (presumable) condition of having diabetes as another possibly relevant identity category? 4.2. Quantitative analysis My pilot study has clearly underlined the merits of combining qualitative and quantitatively-oriented analysis: undoubtedly, examining individual instances of negative identification such as in 4.1 is vital to grasp the conceptual complexity of the topic - as we have seen, contrasting oneself with a particular identity category can have various reasons and implications, which is why rigid either/ or-taxonomies alone bear the risk of missing out on interesting aspects that only come to light by means of indepth analysis. On the other hand, looking at the data from a bird‘s eye view by examining negative identifiers quantitatively has been extremely revealing in this study, because while the data seems to speak a clear language, it does not seem to say what I assumed it would after having qualitatively analysed a few instances of negative identification. What I did was to analyse my corpus of 85 negative identifiers with WordSmith to find out whether and, if so, how often the functional categories specified in 3.3.3. occur and how they are related to each other. The first step was to search for the values of the indefinite NPs speakers negatively identify with, with the result that most NPs refer to professions or levels of expertise, followed by lifestyle categories, as can be seen in table 2 below: Identity category Examples Tokens Professional magician, poet, artist, doctor, marketing expert, keyboard player, mechanical engineer, pianist, IT professional, real estate mogul, CAM designer 32 Expertise newbie, expert, novice, beginner 18 Lifestyle off-piste fan, trekkie, addict, abuser, MP player, Zuul player 16 Evaluative Jerk, nice person, idiot, grammar nazi, attention seeker 9 Ideology Pro-lifer, feminist, racist, religious nutter, elitist 7 Table 2: Identity categories. …or not to be 265 It should be mentioned that the categorization of NP values is by no means a straightforward either/ or-task, because categories may change depending on what aspect defining the category is seen as most important. For example, it is debatable whether terms such as elitist or religious nutter can be unproblematically classified as referring to people‘s ideological position: for example, the concept of an elitist, which refers to a person‘s beliefs about how the social world should be structured and can thus be assigned to the category of ideology, can also have implications for the referent‘s lifestyle (an elitist lifestyle might also involve consuming particular ‗exclusive‘ products etc.). The fact that professionalism features so prominently as dissociated identity category alone is interesting, because I had expected a greater variety of different identity categories: if late modernity is marked by a pluralisation of choices to make for the individual (cf. Giddens 1991), who has become the central unit of social life (cf. Beck 2002), and if authenticity is the ―pre-eminent value and framework for self-actualization‖ (Harris n.d., online), I thought that this might be reflected linguistically in a greater variety of categories people contrast themselves with. The clear predominance of the categories of professions and expertise made me wonder about the functions of speakers‘ frequent dissociation from these categories - and indeed, the functional analysis was quite revealing. In terms of my differentiation between endophoric and exophoric reference, the large majority (76 of 85) of negative identifiers were endophoric, i.e. served to modify parts of the speaker‘s own utterance, which is why I will focus on this type of negative identifier in the following. My data shows a strong correlation between the value of the indefinite NP and the type of reference, as there is a clear dominance of cataphoric negative identifiers used to contrast the speaker with a particular level of expertise or professional categories. It seems that speakers engaging in online forum discussions predominantly use negative identifiers epistemically to cataphorically mitigate utterances (most often introduced by but) that could be interpreted as claiming expert knowledge. The results are summarized in table 3 below (for reasons of space, I only include the categories with more than one token assigned to them): Function Identity category Total: 19 Example Contrast Expertise 7 I'm not aware of variation to window positioning (although I'm no expert on MM4) Contrast Lifestyle 7 I realize Fentanyl is a powerful drug, but I'm just not an abuser. Contrast Ideology 2 Women have been objectified for many years. I'm no feminist, well, not a militant feminist. reference Anaphoric Eva Triebl 266 reference Function Identity category Total: 57 Example Contrast Professional 21 I’m no IT professional but I know my shit. Contrast Expertise 12 I wouldn’t call myself an expert but I’m also not a novice. Contrast Lifestyle 8 I’m not Trekkie. But I really enjoyed this Prelude to Axanar. Contrast Ideology 4 I'm not a men's right activist myself, but these days, I can't ignore the feeling that we're living more and more in female paradise. Contrast Evaluative 3 I’m no white knight. But thankfully I’m also not an attention seeker. Table 3: Functions of negative identifiers. Admittedly, I was disappointed upon first realizing that negative identifiers are obviously most often used strategically as epistemic disclaimers in my corpus: my assumption was that the interesting cases would be the ones where negative identifiers ‗do more‘ than ‗just‘ modify their co-text, because I thought that these would be the instances of negative identification that might be relatable to broader late modern social trends like individualization associated with greater pressure to be ‗authentic‘ in an increasingly competitive global marketplace. However, being convinced that the ordinary, typical is usually more revealing than the rare, extraordinary, even though this often means the collapse of initial hypotheses, I decided to step back and look at my frequencies again. While I initially distinguished between strategic and non-strategic uses of negative identifiers, assuming that the non-strategic cases would be the ones with implications beyond the immediate conversational situation, the predominance of endophoric epistemic disclaimers have made me wonder if the answer - or, at least, new interesting questions - might be in front of my eyes: so speakers are very careful about voicing their opinions on things, or giving advice, without previously disclaiming expertise. The fact that having or not having expertise is an important issue in my corpus certainly relates to the text type examined: online forums are often used to get advice easily, fast and for free. Then again, if so much knowledge is circulating on the Internet anyway, then why do speakers feel the need to - out of their own volition - fend off associations with expert identities and, thus, responsibilities? Who would expect them to be experts? What this suggests is that, whatever the cause is, being an expert or not is an important source of identity in the speakers‘ conceptualizations of self. To refer back to the research questions I initially formulated, I had assumed that it would be possible to distinguish between a strategic, i.e. discourse marking use of negative identifiers to modify the interpretation of the rest of the utterance, and a non-strategic use of negative identifiers, Cataphoric …or not to be 267 serving to index aspects of the speaker‘s identity beyond the immediate conversational situation, and that this difference would correspond to the difference between exophoric (referring to utterance-external text) and endophoric (referring to parts of speaker‘s own utterance) negative identifiers. However, the results of the pilot study imply that formal differences between the strategic and the identity-indexing use of negative identifiers might not be as important as assumed: on the contrary, it seems that one form may simultaneously fulfil local, discourse-managing functions and more global, identity-indexing functions. After all, whenever we choose to express ourselves, ―we draw on discursive resources which have already been shaped to represent the world in particular ways‖ (Sealey 2009: 196). In other words, just because negative identifiers are predominantly used as discourse markers and, thus, for their textual and interpersonal rather than for their ideational metafunctions (cf. Bloor & Bloor 2004: 11), this does not mean that they do not give cues to the speakers‘ social identity. 5. Conclusion In this paper, I presented my approach to negative identifiers, which I am analysing on three levels to find out why and how speakers use them in concrete communicative situations, if there are patterns of usage manifesting themselves across larger samples of data and whether and, if so, how this can be related to social research questions in CDA. Arguing that negative identifiers are functionally comparable and can thus be analysed like discourse markers, I introduced my initial formal-functional framework and then demonstrated how I am applying this framework to the qualitative and quantitative analysis of negative identifiers in authentic discourse contexts. I wanted to find out what the main functions of negative identifiers are and if it would be possible to identify a ‗non-strategic‘ function with implications beyond the immediate conversational situation. The quantitative analysis of a pilot corpus of 85 negative identifiers showed that speakers predominantly use this structure to disclaim expertise and, related to this, that they most frequently dissociate themselves from professional or expert identity categories. These findings challenged my initial assumption that only the ‗non-strategic‘ use of negative identifiers would have implications for people‘s self-concepts. On the contrary: qualitatively analysing negative identifiers suggests that negative identification can be multifunctional and simultaneously have different implications for the speaker‘s self-representation and self-concept. Thus, while I still think that the initial distinction between strategic and non-strategic uses of negative identification is reasonable, the pilot study implies that negative identifiers can at the same time serve as discourse markers and give cues to more permanent concepts of self-identity. Eva Triebl 268 Acknowledgements I would like to thank my PhD supervisor, Alison Sealey, and my mentor and dear colleague Georg Marko, for their invaluable feedback on this work in progress. References Adolphs, Svenja (2008). Corpus and Context. Investigating pragmatic functions in spoken discourse. Amsterdam/ Philadelphia: John Benjamins. Aijmer, Karin (1996). Conversational Routines in English. London: Longman. Aijmer, Karin (2002). English Discourse Particles: Evidence from a Corpus. Amsterdam/ Philadelphia: John Benjamins. Aijmer, Karin (2013). Understanding Pragmatic Markers. A Variational Pragmatic Approach. Edinburgh: Edinburgh University Press. Aijmer, Karin & Anne-Marie Simon-Vandenbergen (2004). „A model and a methodology for the study of pragmatic markers: the semantic field of expectation.‖ Journal of Pragmatics 36. 1781-1805. Arnett, Jeffrey Jensen (2002). ―The Psychology of Globalization.‖ American Psychologist 57 (10). 774-783. Beck, Ulrich (1986). Risikogesellschaft. Auf dem Weg in eine andere Moderne. Frankfurt a.M.: Suhrkamp. Blakemore, Diane (1992). Relevance and Linguistic Meaning: The Semantics and Pragmatics of Discourse Markers. Cambridge: Cambridge University Press. Brinton, Laurel J. (1996). Pragmatic Markers in English. Grammaticalization and Discourse Functions. Berlin/ New York: Mouton de Gruyter. Brown, Peter & Stephen Levinson (1977). ―Universals in language usage: Politeness phenomena.‖ E. Goody (ed..). Questions and Politeness: Strategies in Social Interaction. Cambridge: Cambridge University Press. Carretero, Marta, Carmen Maíz-Arévalo & Maria Ángeles Martínez (2014). ―‘Hope This Helps! ‘ An Analysis of Expressive Speech Acts in Online Task-Oriented Interaction by University Students.‖ In: Jesús Romero-Trillo (ed.). Yearbook of Corpus Linguistics and Pragmatics 2014: New Empirical and Theoretical Paradigms. 261-290. Conrad, Susan (2002). Corpus linguistic approaches for discourse analysis. Annual Review of Applied Linguistics 22. 75-95. Coulmas, Florian (1979). ―On the sociolinguistic relevance of routine formulae.‖ Journal of Pragmatics 3. 239-266. Cristina Grisot & Jacques Moeschler (2014). ―How Do Empirical Methods Interact with Theoretical Pragmatics? The Conceptual and Procedural Contents of the English Simple Past and Its Translation into French.‖ In: Jesús Romero-Trillo (ed.). Yearbook of Corpus Linguistics and Pragmatics 2014: New Empirical and Theoretical Paradigms. 7-34. Crystal, David & Derek Davy (1975). Advanced Conversational English. London: Longman. Cuenca, Maria-Josep & Maria-Josep Marín (2009). ―Co-occurrence of discourse markers in Catalan and Spanish oral narrative‖. Journal of Pragmatics 41. 899- 914. Erman, Britt (1986). ―Some pragmatic expressions in English conversation.‖ In: Gunnel Tottie & Ingegerd Bäcklund (eds.). Pragmatic expressions in English: A …or not to be 269 study of “you know”, “you see” and “I mean” in face-to-face conversation. Stockholm: Almqvist and Wiksell. 131-147. Erman, Britt (2001). ―Pragmatic markers revisited with a focus on you know in adolescent talk‖. Journal of Pragmatics 33. 1337-1359. Fairclough, Norman (1989). Language and Power. London: Longman. Fraser, Bruce (1996). ―Pragmatic Markers.‖ Pragmatics 6(2). 167-190. Fung, Loretta & Ronald Carter (2007). ―Discourse Markers and Spoken English: Native and Learner Use in Pedagogic Settings.‖ Applied Linguistics 28 (3). 410- 439. Giddens, Anthony (1991). Modernity and Self-Identity. Self and Society in the Late Modern Age. Cambridge: Polity. Givón, Talmy (1993). English Grammar: A Function-Based Introduction. Amsterdam/ Philadephia: John Benjamins. Grisot, Cristina & Jacques Moeschler (2014). ―How do empirical methods interact with theoretical pragmatics? The conceptual and procedural contents of the English simple past and its translation into French‖. In: Jesus Romero-Trillo (ed.). Yearbook of corpus linguistics and pragmatics 2014. New empirical and theoretical paradigms. Dodrecht: Springer. 7-33. Gumperz, John J. (1996). ―The linguistic and cultural relativity of inference.‖ In: John J. Gumperz & Stephen C. Levinson (eds.). Rethinking Linguistic Relativity. Cambridge: Cambridge University Press. 374-406. Halliday, Michael (1994). An Introduction to Functional Grammar (2 nd ed.). London/ New York et al.: Arnold. Hammond, Jennifer. & Derewianka, Beverly. (2001). ―Genre.‖ In: Ronald Carter & David Nunan (eds.). The Cambridge Guide to Teaching English to Speakers of Other Languages. Cambridge: Cambridge University Press. Harris, Dave (n.d.). ―Reading Guide to Giddens (1991). Modernity and Self-Identity: Self and Society in the Late Modern Age. Cambridge: Polity Press. [online]. www.arasite.org/ giddmody.htm (5 July 2014). Hasan, Ruqaiya (1985). ―The structure of a text.‖ In: M.A.K. Halliday & R. Hasan (eds.). Language, Context and Text: Aspects of Language in Social-semiotic perspective. Cambridge: Cambridge University Press. Hodkinson, Paul (2007). ―Interactive Online Journals and Individualisation.‖ In: New Media and Society 9 (4). 625-650. Hymes, Dell (1972). ―On communicative competence.‖ In: John Bernard Pride & Janet Holmes (eds.). Sociolinguistics. Harmondsworth: Penguin. 269-293. Jordan, Michael (1998). ―The power of negation in English: Text, context and relevance.‖ Journal of Pragmatics 29. 705-752. Jucker, Andreas H. & Yael Ziv (1998). Discourse Markers: Descriptions and Theory. Amsterdam/ Philadelphia: John Benjamins. Jucker, Andreas H., Daniel Schreier and Marianne Hundt (eds.). (2009) Corpora: Pragmatics and Discourse. Papers from the 29th International Conference on English Language Research on Computerized Corpora (ICAME 29). Ascona, Switzerland, 14-18 May 2008. (Language and Computers: Studies in Practical Linguistics 68). Amsterdam: Rodopi. Kim, Mira (2010). ―Translation Error Analysis. A Systemic Functional Grammar Approach.‖ In: Caroline Coffin, Theresa Lillis & Kieran O‘Halloran (eds.). Applied Linguistics Methods. A Reader. Oxon/ New York: Routledge. Landone, Elena (2012). ―Discourse markers and politeness in a digital forum in Spanish.‖ Journal of Pragmatics 44. 1799-1820. Eva Triebl 270 Lawrence Schourup (1985). Common Discourse Particles in English Conversation. Garland: New York. Leech, Geoffrey (1983). Principles of Pragmatics. London: Longman. Lenk, Uta (1998). Marking Discourse Coherence: Functions of Discourse Markers in Spoken English. Tübingen: Gunter Narr. Louw, Bill (1993). ―Irony in the text or insincerity in the writer? The diagnostic potential of semantic prosodies.‖ In: Mona Baker, Gill Francis & Elena Tognini-Bonelli (eds.). Text and Technology: In Honour of John Sinclair. Amsterdam: John Benjamins. 157-176. Marko, Georg (2008). Penetrating Language. A Critical Discourse Analysis of Pornography. Tübingen: Gunter Narr. Noveck, Ira & Dan Sperber (2004). Experimental pragmatics. Basingstoke/ New York: Palgrave Macmillan. Ochs, Elinor (1996). ―Linguistic resources for socializing humanity.‖ In: John J. Gumperz and Stephen C. Levinson (eds.). Rethinking Linguistic Relativity. Cambridge: Cambridge University Press. Östman, Jan-Ola (1995). ―Pragmatic particles twenty years after.‖ In: Brita Wårvik, Sanna-Kaisa Tanskanen & Risto Hiltunen (eds.). Organization in Discourse. Turku: University of Turku. 95-108. Schiffrin, Deborah (2007). Discourse Markers. Cambridge: Cambridge University Press. Sealey, Alison (2012). ―‗I just couldn't do it‘: representations of constraint in an oral history corpus‖. Critical Discourse Studies 9(3). 195-210. Sinclair, John M. (1991). Corpus, Concordance, Collocation. Oxford: Oxford University Press. Stubbs, Michael (1983). Discourse Analysis. The Sociolinguistic Analysis of Natural Language. Oxford: Blackwell. Verschueren, Jef (1999). Understanding Pragmatics. London: Arnold Publishers. Swales, John M. (2004). Research Genres: Explorations and Applications. Cambridge: Cambridge University Press. Wierzbicka, Anna (1976). ―Particles and linguistic relativity.‖ In: International Review of Slavic Linguistics 1(2-3). 327-367. Eva Triebl English Department University of Graz