1 Occasioned Semantics: A Systematic Approach to Meaning in Talk Jack Bilmes University of Hawaii (This paper was published in Human Studies, 34:129-153, 2011. The final publication is available at link.springer.com.) Abstract: This paper puts forward an argument for a systematic, technical approach to formulation in verbal interaction. I see this as a kind of expansion of Sacks’ membership categorization analysis, and as something that is not offered (at least not in a fully developed form) by sequential analysis, the currently dominant form of conversation analysis. In particular, I suggest a technique for the study of ‘‘occasioned semantics,’’ that is, the study of structures of meaningful expressions in actual occasions of conversation. I propose that meaning and rhetoric be approached through consideration of various dimensions or operations or properties, including, but not limited to, contrast and co-categorization, generalization and specification, scaling, and marking. As illustration, I consider a variety of cases, focused on generalization and specification. The paper can be seen as a return to some classical concerns with meaning, as illuminated by more recent insights into indexicality, social action, and interaction in recorded talk. Sequential Analysis and Membership Categorization Analysis The work of Harvey Sacks resulted in two organized traditions of research— sequential analysis (SA) and membership categorization analysis (MCA). After Sacks’ death, SA emerged as by far the dominant approach. This is due to a number of factors. SA has the more elegant and, many would argue, the more rigorous methodology. Moreover, it has a broader and more flexible approach than MCA. And we should not underestimate the influence of Emanuel Schegloff, the most prominent practitioner of conversation analysis (CA) since Sacks’ death in 1975. Schegloff has been a firm advocate of SA. However, in recent years, MCA has begun, in a modest way, to flourish again. A major turning point was the publication of the edited volume Culture in Action (Hester and Eglin 1997 ). Perhaps the most telling indication of the increased interest in MCA is the recent publication of two papers on the topic by Schegloff (2007a , b ). One gets the impression (I am going out on a limb here) that these papers were written less from Schegloff’s own interest in the topic than as a response to others’ interest. A central question with regard to the re-emergence of MCA is why it is happening at all. What analytic need has SA left unfulfilled? To begin with, SA tends to avoid (explicit) reliance on, or analysis of, cultural resources (but see, e.g., Turner 1971). The notion is that the analysis should be demonstrable from the data at hand. This approach has obvious virtues. But cultural resources (language to begin with) are brought into play in SA in a generally unexplicated way. Our cultures undoubtedly are a major factor in the way we talk, and the refusal to deal with culture explicitly can be argued to be a weakness of SA. The central tenet of 2 SA, to my mind, is the determination to attend to meaning in terms of the understandings of the participants in the conversation, since it is, after all, they who are constructing the interaction. This, of course, requires recordings of actual situations of talk. The way that this sort of analysis is achieved is through attending to participant responses. The response to a prior utterance demonstrates the interlocutor’s understanding of that prior utterance, and that proposed understanding is, in its turn, subject to acceptance or contradiction by the original speaker (Moerman and Sacks 1988 ). In addition, the CA approach to evidence has been more or less the opposite to that proposed by Karl Popper (1959 ). A claim in CA is not falsified by the fact that, in some particular instance, a thing does not occur where it ‘‘should’’. CA is not after laws but looks instead for resources and sensitivities. If a phenomenon sometimes, intelligibly, occurs, that is proof of its availability as a resource for speakers.1 Falsification would occur in CA only in response to a claim that something always or never (intelligibly) happens. These were brilliant innovations, and they have resulted in new approaches to such fundamental matters as context (Schegloff 1991 ), description (Schegloff 1988a ), rules and analytical categories (Bilmes 1988 ), and social structure (Drew and Heritage 1992 ), as well as methodological innovations; but, I would like to argue, they constitute only a partial approach to meaning in conversation. It is a result-oriented approach. It deals with what utterances end up meaning for conversational purposes. There is no comparable systematic methodology in SA for dealing with speakers’ choices, that is, with the rhetorical aspects of talk, or with how certain understandings are promoted by those choices. This inevitably involves a consideration of culture and particularly of linguistic resources and of possible but nonoccurring alternatives. Thus, such an analysis, while taking the actual talk as its object, must, to some extent, depart from the data at hand. In addition, it involves dealing with such matters as categorization, ‘‘scripting’’ (Edwards 1994 , 1995 ), and other features of presentation, that is to say, with semantics and rhetoric. Not, to be sure, with semantics and rhetoric as general, contextless topics, but with the semantics and rhetoric ‘‘of the moment.’’ In addition, SA’s handling of conversational structure is somewhat limited. It deals with beginnings and endings, sequencing of actions, insertion sequences, preliminaries, prefaces, and such Schegloff 2007c ), but it falls short in analyzing the structures of meaning that are built up over the course of a conversation. A conversation is an ongoing construction of interrelated agreements and understandings, of claims and presuppositions. We require a methodology that is capable of 1 CA does look for normative generalizations, but, again, the fact that a norm is sometimes violated, and that sometimes such violations are not sanctioned, is not taken as falsification. Rather, the occasional occurrence of sanctions in response to violations is taken as proof that participants are, or may be, oriented to such norms and, therefore, that the norms exist. It is true that conversation analysts sometimes make quasi-statistical empirical generalizations (Schegloff 1993), but the lack of adequate sampling procedures makes these claims less than scientific. 3 dealing in a systematic way with the construction and transformation of meaning over the course of interaction, and that is capable of assessing the final product as a produced structure of meaning. I am not saying that practitioners of SA have not dealt with any of these matters. One thinks, for example, of Heritage and Raymond’s (2005 ) and Raymond and Heritage’s (2006 ) work on epistemics or Stivers on marking (2007 ) or Charles Goodwin on a variety of subjects, such as highlighting (1994 ). Ideas such as scripting and epistemic authority are contributions to the wider field of formulation, but not to occasioned semantics. They are themselves made up of meanings and meaning relationships. What is lacking is a broad framework, a methodology, for dealing with the organization of meaning in conversation, and it is this, I think, that accounts for lasting appeal of MCA. MCA, while rather limited in its scope, deals with some of the shortcomings of SA as an approach to meaning in conversation, in that it attends to choices of categories in actual situations of talk and the effects of those choices in promoting certain understandings. In addition, it asks how categories are invoked and related to one another. Sacks had a lively interest in categories in general, but he developed his analytical ‘‘machinery’’ in terms of categories of persons, and his successors, by and large, have stayed within those limits. The best known of Sacks’ work on MCA, and the major charter for the movement, is his paper analyzing ‘‘The baby cried. The mommy picked it up’’ (1972 ). Although the analysis is, in my view, seriously flawed, the ideas he presents are seminal.2 I cannot see any cogent reason why these ideas cannot be applied to the analysis of all types of categories. Furthermore, limiting the analysis to categories seems unnecessarily restrictive. We can aspire to deal with all signifying expressions (I will return to this), including those which are arguably not categories at all. A clear example of the is the name of an individual, such as, ‘‘President Barack Obama’’. ‘‘New York City,’’ although it consists of a variety of more specific places, such as Times Square, does not seem to qualify as a category. In many cases, it is not clear what is to be counted as a category. Housley and Fitzgerald (2009 ) treat expressions such as ‘‘untrustworthiness’’ and ‘‘incompetence’’ as categories. Perhaps more questionably, Billig (1996 ) writes that ‘‘To say that ‘this cup is red’… is to make a categorization. … This cup has been categorized as belonging to the wider category of red objects’’ (151). Although this transformation from description to set may be logically unassailable, I’m not sure that the set is being relevantly invoked. It seems simpler to just say that ‘‘red’’ is a signifying expression. What is needed, I am proposing, is a methodology for analyzing all signifying expressions, including categories, references, and descriptions. I will, shortly, propose an approach, which I am calling ‘‘occasioned semantics,’’ 2 See the appendix to this article for a critique of Sacks’ paper. 4 that deals with the limitations, mentioned above, of MCA and offers a greatly expanded methodology. It can be viewed as an attempt (a partial one, to be sure) to deal with conversational meaning and rhetoric in a systematic way. Formulation Signifying expressions are formulations in themselves, and units of larger, complex formulations. I stipulate at the outset that I am not talking about formulation in the sense proposed originally by Garfinkel and Sacks (1970 ) and adopted by various other conversation analysts (e.g., Antaki 2008 ; Arminen 2005 ; Barnes 2007 ; Drew 2003 ; Gafaranga and Britten 2004 ; Heritage and Watson 1979 ; Walker 1995 ); that is, as a statement of the gist, meaning, or upshot of previous conversation. In fact, Sacks himself, in his lectures, uses the term ‘‘formulation’’ frequently, and not in this sense. What Garfinkel and Sacks are talking about is more properly called ‘‘reformulation,’’ because the previous conversation already consists of formulations. So, then, what is a formulation? We can start by examining some examples of Sacks’ (1992 ) usage of the term. formulating one’s present state (V. 1: 69) he formulates his remarks in terms of ‘you’ (V. 1: 166) there are alternative ways that he and those he is dealing with… may be categorically formulated (V. 1: 205) formulating the session as a ‘group therapy session’ (V. 1: 515) assign name-formulations to the actions (V. 1: 515) In these passages, and many more, Sacks usage of the term is consistent with the dictionary definition—‘‘to put into words’’ (wordnetweb.princeton.edu). To be sure, Sacks sometimes uses the term in somewhat divergent and idiosyncratic ways (‘‘we could formulate the omnirelevance of patient-therapist’’ [V. 1: 515]), but, in general, he uses formulation in its common meaning. Now, ‘‘to put into words,’’ implies a ‘‘something’’ that is being put into words, an object, concept, attribute, situation, action, etc. That is, a formulation has a referent, a signification in Saussure’s sense. I want to stress that I am using ‘‘referent’’ in a broad sense; whatever can be referred to, from an object to a concept to a state of affairs, is a referent. ‘‘Hello,’’ as opposed to ‘‘hi’’ or ‘‘how do you do,’’ is a rhetorical choice but not a formulation, because it has no referent.3 A formulation may do pure reference, as with a proper name.4 Or, it may have a descriptive, reality constructive aspect. The notion of formulation bridges Schegloff’s (2007b ) distinction between category and reference. I am not questioning the utility of 3 It is to be noted that an expression may signify by virtue of its placement. So, ‘‘yes’’ or ‘‘he did’’ has a particular signification when preceded by ‘‘Did he do his homework?’’ 4 I am not really sure that there is such a thing as ‘‘pure reference’’. Even the use of a proper name is a choice from among other referential possibilities and may therefore have descriptive implications or overtones. 5 this distinction, but, given that we may speak of someone as ‘‘John Smith’’ or ‘‘the guy with the hat’’ or ‘‘the mailman,’’ we will need a term that covers the possibility of these alternative ways of speaking. In it’s minimal form, a formulation is a single meaningful item, such as a word. However, it may also be a much longer unit. A narrative, by my definition, may be considered a formulation insofar as it may be said to have a referent (e.g., what happened at the ballgame). A formulation is always usable as an utterance but is somewhat different than an utterance. An utterance must actually be uttered.5 A formulation may occur as writing as well as speech. When a formulation is in fact uttered, we include as elements not only words but also features of performance, such as gesture,6 stress, ‘‘tone of voice,’’ etc. (In fact, such elements may be included even in unuttered formulations, insofar as they are indicated. For example, a written formulation may include underlining to indicate stress.) The way an utterance is formulated is obviously (partially) determinative both of its meaning and its character as social action. More nebulously, but perhaps most importantly, when we use ‘‘utterance’’ in analytical talk, we tend to be focusing on its role in interaction, whereas ‘‘formulation’’ stresses ways of identifying, categorizing, describing, and persuading, that is, the rhetorical and reality-constructive aspects of linguistic action. (This is not to deny that formulations occur in some context and derive part of their situated meaning from the context.) Looking further into this notion of formulation, here is another passage from Sacks: Members can’t do pure formulating. That is to say, you can’t be engaged in ‘merely’—non-consequentially, non-methodically, non-alternatively—saying ‘This is, after all, a group therapy session’. To do that—even though you’re merely invoking one thing that’s true about this—is to do other things as well, e.g., put somebody down for something they said, propose special relevancies, propose that some topic ought to be discussed or not be discussed, invoke a status hierarchy, etc. At any rate, in each case that a formulation of a setting, or an identity, is done, that’s something that has some line of consequences, and some analyzable bases, for participants, which can be one differentiated from another possible formulation, and also from not doing it at all (1992 , Vol. 1: 516). Sacks is making two major points about formulations. One, of course, is that they are inevitably a way of doing, not merely saying, something, a notion familiar to us from speech act theory, although Sacks’ notion of ‘‘doing’’ is broader. But he makes 5 I have run across definitions that seem to include written language, but this, I think, does not represent the common understanding. 6Some gestures are significant, in the sense that they have referents and may substitute for words. They are, like signifying words, to be considered, in themselves, formulations. Thus, the notion of a formulation as ‘‘putting into words’’ is too narrow. 6 another crucial point with ‘‘non-alternatively’’ and ‘‘differentiated from another possible formulation,’’ namely, that a formulation is a choice from among a number of alternative ways of identifying or describing the referent or producing the conversational action. So, a formulation is a consequential choice, what the discursive psychologists call ‘‘rhetoric’’ (Billig 1996 ; Edwards 1997 ; Potter 1996 ). Presumably, the particular choice made forwards some line of argument, defends some position, or otherwise serves the purposes of the speaker, but it is sufficient, and preferable, to simply say that it has consequences. It follows that a formulation is always local—it cannot be fully accounted for in cultural or semantic terms. Central to the subject of formulation and occasioned semantics is the concept of choice. When describing a person, we may say ‘‘He is tall’’ or ‘‘He is smart’’. Both may be true, but they are very different. That is, we are ‘‘putting into words’’ two different ‘‘somethings’’. But in saying ‘‘He is smart,’’ we are also choosing to say that in preference to ‘‘He is clever,’’ which may express essentially the same concept, although with perhaps differing connotations. So, in describing him as smart, we are making, simultaneously, two types of choices. Sometimes these choices are made obvious in the talk, as when an item is recategorized. Sometimes it is implicit—it is up to the analyst to imagine the possible alternatives. What, then, is a ‘‘possible alternative’’? I would suggest that what makes an alternative ‘‘possible’’ within a particular conversation is similar reference, plausibility, relevance, and non-contradiction. If two formulations do not refer to the same thing, they are not alternatives. The plausibility criterion provides that a formulation that is clearly false cannot be offered as an alternative to one that is possibly true, given the recipient’s knowledge. So, for example, ‘‘I saw a short man’’ cannot (normally) be alternatively formulated by ‘‘I saw a man who was about three inches tall,’’ because the first is possibly true, that is, plausible, and the second is transparently false. The third criterion is relevance. Consider a math teacher talking to John’s parents at a teacherparent conference. ‘‘John is tall, but he doesn’t try’’ is not a relevant alternative to ‘‘John is clever, but he doesn’t try’’. Perhaps we could say that the relevance criterion is simply a way of claiming that the context constrains the possible referent—so, in this case, the referent is not John in toto but John’s intellectual ability. In any event, a fact may be free-floating truth in theory, but in practice it is bound to a context. The final criterion is non-contradiction: ‘‘John is short’’ and ‘‘John is tall’’ are not alternative formulations. They both refer to John’s height, they may both be plausible and relevant, but they are contradictory—they cannot both be true. Formulations are not, in the first instance, ways of doing things (e.g., eliciting offers), although they may serve such purposes. Saying is doing, to be sure, but saying is also saying. (So, for instance, asserting something to be true, an illocutionary act, requires the communication of that which is being asserted.) I would argue that even performatives, such as ‘‘hello,’’ say before they do. We must understand them as language before we can see what act they are performing.7 7 I am indebted to Arnulf Deppermann for reminding me that this is not necessarily true in the case of 7 This may seem a regression to a pre-Austinian, pre-Wittgensteinian mentality, but we will retain the conversation analytic focus on recordings and transcripts of actual occasions of talk, on indexical meaning, and on participant interpretations, and we will retain a sensitivity to interaction as an analytical resource. And we are not losing sight of the fact that formulations do something; we are merely insisting that they also say something. In addition, as with CA in general, we will avoid theoretical preoccupations and mentalistic explanations. We will look for what the data has to offer. Occasioned Semantics Occasioned semantics deals with a subset of formulation. It deals with words or phrases (or signifying gestures), what I will call ‘‘expressions,’’ as opposed to longer units, such as sentences, narratives, or accounts. My approach has been to look at various properties or operations or dimensions associated with signifying expressions. An initial, nonexhaustive list includes. 1. Co -categorization and contrast . I place these together because they are opposite sides of the same coin. Co-categorization emphasizes the similarity of constituent formulations. Contrast also implies that the items belong to the same overarching category, but emphasizes their differences. 2. Generalization and specification . Frequently, a formulation in conversation is later reformulated at a more general or specific level. Even when this is not so, many formulations may be said to be given at some particular level of generality. Thus, when we say ‘‘tree,’’ we have said something more general than ‘‘pine’’ and more specific than ‘‘plant’’. The choice of any particular level of generality is a rhetorical choice, with certain consequences (see Deppermann 2011 , this volume; Hauser 2011 , this volume). 3. Scaling . This involves the arrangement of a set of items from less to more extreme or intense. My interest is primarily in ‘‘implicative scales’’ in actual talk (see Bilmes 2010 for an initial treatment). Two general characteristics of implicative scales are that at least some of the items are not mutually exclusive and that the use of a less extreme item implicates the absence of the more extreme items. So, for example, offering a weak excuse implicates, but does not logically imply, that one does not have a stronger excuse available (see Bilmes 1993 , 1995 , on response priority). CA already incorporates ideas of upgrading and downgrading, but these notions have been used in an ad hoc manner and have not, to my knowledge, been topicalized or given technical specification. My approach is clearly related to Gricean analysis and to so-called Horn scales (Grice 1975 ; Horn 1972 ) but there are a number of differences, including attention to actual data, on-the-spot creation of scales, and sensitivity to cultural, indexical, and sequential—not just logical or semantic, in-the-language—factors. language acquisition. Perhaps in the case of formulaic performatives, to understand what act they perform is to understand them as language. 8 4. Marking . In linguistics, marking consists of several loosely related criteria. In my usage, though, it refers to an opposition between a formulation that would be normally expected in a particular situation and one that has the same reference but is not, in some sense, normal. Thus, when a wife says to her husband, ‘‘Do you know what your son did today,’’ referring to their child as ‘‘your son’’ rather than by his name, this marked usage will give rise to inferences. In this case, we can be fairly certain that their son did something wrong (see Geoghegan 1973 ; Stivers 2007 ). The approach is basically structural—it deals with expressions in relation to other expressions used in the conversation and in relation to other expressions that might possibly have been used. I have, thus far, relied not only on Sacks’ work on categories, but also on adaptations of ethnosemantic taxonomic analysis (see Conklin 1969 ; Frake 1969a , b for representative examples) and implicature theory (Levinson 1983 , Chap. 3). The result is an analysis which is systematic and technical rather than ad hoc and intuitive. I view occasioned semantics as a step toward a more inclusive methodology of formulation analysis, midway, as it were, between MCA and the analysis of broader units such as accounts and narratives. Most of my work so far has dealt with generalization and specification, and that is what I will focus on in the remainder of this paper. Although my general approach to talk and data is primarily rooted in CA and, in particular, in Sacks’ work on categories, it was anthropological ethnosemantics and, especially, a fugitive paragraph in an article by Charles Frake (1961 ) that started me in the particular direction that I took. At a time when anthropologists were interested in taxonomies primarily as abstract, decontextualized, static cognitive or cultural ‘‘maps,’’ Frake recognized that taxonomic hierarchies gave individual speakers choices in regard to how they might categorize particular phenomena, and that these choices were strategic resources that could be used in conversational environments. Noting that skin disease terms may enter into situations such as bride-price negotiation, competitive joking, and maligning, Frake observes that ‘‘in many of these situations it is imperative to speak at just the level of generality that specifies the pertinent information but leaves other, possibly embarrassing, information ambiguous’’ (122).8 I chose to use the conventions of taxonomic analysis rather than a Sacksian presentation for heuristic reasons. Taxonomic analysis deals with multilevel hierarchies. Although, as Watson (1978 ) pointed out, we can conceive of ‘‘devices within devices’’ (in Sacks’ terms), MCA has not capitalized on that possibility. MCA-oriented work has dealt with a category or set of related categories and a more general term that includes the subordinate categories. It is not very well adapted for 8 Categorizing, in this quotation from Frake, involves withholding of information. Typically, within the conversation analytic tradition, the emphasis is on meaning or relevancies that are added in the process of categorization. That is, the choice of this category rather than that is informative. By categorizing, we reveal the nature of the thing as we would have it known on this occasion. 9 dealing with deeper taxonomic relations because it does not use diagrams. It is not visual. In dealing with multilevel taxonomies, one would have to narrate one’s way through a thicket of relations of generality and contrast. When dealing with questions concerning level of generalization, a taxonomic presentation seems clearly superior. In addition, the diagrammatic visualization forces one to consider items that were not mentioned but might be implicated by the talk. Case 1: Generalization and Implication I will illustrate with material from a previously published article (Bilmes 2009a). Although many expressions can be seen to be produced at a particular level of generalization, in preference to being more or less specific, I have looked particularly at what Garfinkel has called ‘‘perspicuous settings,’’ (2002: 182) that is, settings that are especially amenable to analysis. I have examined exchanges where a signification appears and then reappears at a more general or specific level. Such settings made visible the act of generalization (or specification) and its conversational functions. The following data, first presented by Sacks (1992, Vol. 1: 113–118; also 184–185 and 411–412) and elaborated by Edwards (1997: 96–98), presents such a setting. The police have been called to intervene in a domestic situation. They instruct one of the parties (‘B’ below) to confer with a social worker. Sacks provides us with an excerpt from B’s discussion with the social worker (‘A’). Here is the excerpt, in Sacks original format: (Sacks 1992, vol. 1: 113) (1) A: Yeah, then what happened? (2) B: Okay, in the meantime she [wife of B] says, ‘‘Don’t ask the child nothing.’’ Well, she stepped between me and the child, and I got up to walk out the door. When she stepped between me and the child, I went to move her out of the way. And then about that time her sister had called the police. I don’t know how she…what she… (3) A: Didn’t you smack her one? (4) B: No (5) A: You’re not telling me the story, Mr B (6) B: Well, you see when you say smack you mean hit (7) A: Yeah, you shoved her. Is that it? (8) B: Yeah, I shoved her Sacks’ major point is that A, not being a party to the events in question, is nevertheless able to see that B is not telling ‘‘the story,’’ and B can recognize A’s grounds for doing so, because B has not provided an adequate warrant for the police being called. I was particularly interested in line 6. It is phrased as an explanation, but what does it explain? Of course, when A says ‘‘smack,’’ he means ‘‘hit’’. What else would he mean? What is accomplished by saying this? We begin by noting that ‘‘hit’’ is more general than ‘‘smack’’. The relationship is illustrated in Fig. 1. 10 Fig. 1 ‘‘Hit’’ (The lines indicate ‘‘kind of’’ relationships; e.g., smack is a kind of hit. Actually occurring items are in bold) By ruling out ‘‘hit,’’ B is instructing A that (for example) ‘‘punch’’ would not be an appropriate second guess. ‘‘Punch’’ is never mentioned but it is nonetheless implicated. ‘‘Smack’’ is proposed by A as part of ‘‘the story’’ and, since ‘‘hit’’ is suggested ascategorizing ‘‘smack,’’ it too is a possible version of the story. This gives us Fig. 2 . Fig. 2 ‘‘The story’’ (The arrow indicates a ‘‘part of’’ relation) This is what I have called a ‘‘hybrid taxonomy,’’ in that it combines two types of relations. “Hit” is part of ‘‘the story,’’ whereas “smack” is a kind of hit. The crucial point for my purposes is that each ascending level is more general than the last. Once we are thinking in multilevel terms, we are led to ask why B’s claim was not generalized further. He might have said ‘‘I didn’t do anything to her’’ or ‘‘That is the whole story’’. If there were no conceivable warrant for his sister-in-law’s call to the police, one might expect a broader denial than the one he actually produces. We evaluate what people say by reference to what they might, relevantly, have said. When someone generalizes, we may ask, why did he generalize just so far and no further? The very act of generalization invites such speculation. Thus, B’s turn 6 can be heard as an invitation to A to look for some act other than hitting that would warrant the call to the police. He doesn’t deny that there is a warrant for calling the police, just that smacking or hitting is that warrant. A responds by guessing at ‘‘shove,’’ an action that has not been ruled out and that is a subcategory of B’s initial description (‘‘move’’). This analysis suggests that the degree of generalization may generate certain implicatures that have consequences for the further course of the conversation. The full analysis of the Sacks’ segment (not presented here) produced a complex taxonomy. Some of the relations in that taxonomy were wholly indexical rather than being based in a pre-existing linguistic semantics. (The relation of ‘‘smack’’ to ‘‘hit’’ 11 is explainable on linguistic grounds, but ‘‘smack’’ as, potentially, part of ‘‘the story’’ is a product of the particular circumstances.) The taxonomy is a field of meaning, co-constructed in and for the moment, with sequential implications. Case 2: Generalization and Discursive Environment In another article (Bilmes 2008 ), I dealt with constraints on the level of generalization. The setting was a discussion of a passage in a draft memo in a US regulatory agency. The passage offered a very specific reason for not enforcing a certain regulatory provision. The provision required that loan companies inform female applicants that they need not mention income from alimony or child support. In recommending that the agency not enforce this provision, the memo authors suggested that giving applicants the required notice might have the effect of discouraging them from providing full information on their income, which might, in turn, prevent them from getting the requested loan. Ben, one of the discussants, reformulated this reason as ‘‘We’re not enforcing this portion of Reg B because we think it’s bad public policy’’. This was, of course, a very generalized version of the original reason. It would have been possible to generalize less radically; for example, instead of ‘‘bad public policy,’’ he might have said ‘‘harmful to female loan applicants’’. However, as I tried to show, a less radical generalization would not have made sense in terms of the then-current discursive environment of the agency or in terms of the argument that the speaker was constructing. The politically appointed administrative officers of the agency were free market advocates and were hostile to what they considered overregulation. Thus, their idea of what was bad public policy was rather different from the generally more activist legal staff. The staff countered the new agency policies by insisting that their job was to enforce the law. So, by saying ‘‘We’re not enforcing this portion of Reg B because we think it’s bad public policy,’’ Ben was pointing to an inconsistency in the staff rhetoric. He was also suggesting that the argument legitimized administration obstructionism in regard to enforcement of the law. This become apparent in his follow-up statement: ‘‘If Ted [a member of the administration] didn’t enforce everything he thought was bad public policy where would you be’’. Given these considerations, a less extreme generalization would not have served Ben’s purposes, that is, it would not have constituted a coherent argument against keeping the targeted passage in the memo. The level of generalization was fitted to the administration’s economic/legal arguments and to the staff’s counterargument. Sometimes we need to consider not only possible alternatives, but also those which are related but impossible. This is what I did with ‘bad public policy’. In particular, it may be profitable to look at upgradings and downgradings along various scales (intensity, generality, etc.) Case 3: Generalization and Implied Categorization A focus on generalization and specification, as it turns out, yields a variety of insights into the structuring of meaning in conversation. The following extract is from a mediation session at mediation agency known as the Neighborhood Justice 12 Center. L, the landlord, and T, the former tenant, have come for mediation of a dispute. After T moved out of the house which he and a roommate had been renting from L, L refused to return any of their deposit. The session began with L, T, and M, a trained mediator, and myself as nonparticipating observer, sitting around a table. L and T described the situation and their positions to M. Then M had L leave the room, so that she could speak privately to T, after which she spoke privately to L. They then reassembled as a group. M stated that they seemed to be at an impasse, and L and T reiterated their positions. Then L asked T to have his former housemate call her ‘‘so that I can explain’’ and T replied ‘‘Yah (whatever) I can explain it (too)’’. M seizes on this as an agreement and suggested that they ‘‘write it up’’. (NJC) 1. T: Aoh:: that's s'ch a um: a poor: um: y'know uh- 'at's 2. not even worth the time t' write it up y'know I c'n 3. gi[ve ya John's uh: (.5) John's (work) phone number is:= [( ) 4. ?: 5. T: =if you'd like [( ) [No- b't- b't Lloyd (how I feel) 6. M: 7. about it is this= 8. T: =I do hav I do have 'n appointmenttuh: y'know I I 9. feel th't uh .hh John Seiden's number at work is 10. ((telephone number)) an:d ah: (.) don' need any 11. signatures on anything. 12. M: Wull (2.5) okay [is is [Yeah (.) ('n) I have another meeting 13. T: 14. too so I- I can' see ( [) (even spending) around uh= [Okay 15. M: 16. T: =spending the time to write anything down (there). 17. (1.3) 18. M: Okay so: that you'll give Linda the (ud number) 19. c'd- could we- (.) have that number again: jus' so th' 20. I c'n (.) communicate that to ( )? 21. T: ((repeats the telephone number)) (1) 22. M: four two nine six. 23. T: °umhm 24. M: .hh Well (.) (bLloyd) I'd like y' ta- (.8) would 25. you: be willing t' take a minute (.) just to: (.) let 26. us put that down in writing? 27. T: No I can't see any reason f'r that.=It's (j's a) 28. phone number it's available y'know: uh: (.4) he works 29. at Pearl Harbor 'nd uh (.7) so on 'n' so forth (.5) 30. .hh .hh °y'know° I din' come in here 'n ho- hour 'nd 31. a half 'r whatever j'st ta j'st t' give a phone 32. number 'n': 'n'put something down 'n writing (.) 13 33. 34. M: 35. T: 36. 37. M: 38. M: 39. 40. T: like that [(y'know)= [Okay (.7) wull: =(.) I come in here f'r: substantial uh (1.2) [reasons [Okay 'n' I c'n see how- (.) y'know y' could be disappointed= =resolution yeah My interest, for present purposes, is in lines 28–41. T is rather specific in saying what he did not come here for. He mentions two items (give a phone number and put something like that in writing)without offering a categorization of those items. One might expect the contrast to be at a similar level of specificity (I came to get my money back). However, the contrast is actually made at a more general level (substantial reasons). The direct contrast to substantial reasons is insubstantial (i.e., trivial) reasons. He has thus, by implication, categorized his original two items. This way of doing it makes a certain amount of sense, given that he begins his turn with ‘‘No I can’t see any reason for that’’. But then he mentions giving a phone number (which M has brought up earlier as an agreement and, therefore, something which should be recorded) and putting something down in writing (which M later mentions as worthwhile in itself, since it acknowledges them both for attending the mediation). Having first claimed that he could see no reason to accede to M’s request, and then mentioned two possible reasons, T uses the procedure I have described to more or less negate the two reasons by classifying them as trivial. This analytic outcome (i.e., the implied categorization of the two items as trivial) is, I think, fairly obvious. What is significant here is the mechanism by which the outcome is produced. When item(s) A is contrasted with a more general item B, this suggests a categorization of item A. The situation is illustrated in the following diagram (Fig. 3 ). trivial reasons (implied contrast) substantial reasons (contrast) give a phone number put something in writing get my money back Fig. 3 ‘‘Trivial reasons’’ (Actually occurring items are in bold) Case 4: Tactical Generalization and Specification I turn now to a short example of the rhetorical possibilities of generalization and specification. Sara Taylor, a high ranking member of President George W. Bush’s Justice Department, has been summoned to appear before a congressional committee. 14 Taylor testimony (originally transcribed by Matthew Prior) 1. S: Did you recei:ve (.) telephone calls or other forms 2. of communication (1.6) in regards (.) to (.) the (.) 3. US attorneys (.) that were fired. 4. T: I don’t recall. (.) I don’t recall getting 5. communications about them. (3.0) 6. S: hhhh (2.0) hh (.) You don’t reca:ll if (.) someone 7. called you to complain about a US attorney? (.07) 8. T: Uh (.) Senator (.) uh- (.) I’m sure you can 9. appreciate (.) uh (.) that somebody (.) who (.) uh 10. (0.9) i- u was in my position (.) who got (.) and I’m 11. estimating here (.) who got roughly twenty phone calls 12. a day (.) roughly three hundred emails (.) a day. Each 13. and every day (.) uh (.) about a myriad of topics (.) 14. uh (.) any and everything you could probably (.) uh 15. (.) uh (0.9) would not recall (.) conversations (.) uh 16. or uh (.) phone calls that came to her (.) I-I (.) 17. senator I-I can’t remember what I had for breakfast 18. last week=I just don’t recall any of those 19. conver[sations] 20. S: [I] I assume what you had for breakfast 21. last week has not been the subject of considerable 22. national attention In lines 1–2, the senator asks about ‘‘telephone calls or other forms of communication in regards to the US attorneys that were fired’’. When Taylor says that she does not recall such communication, the senator becomes more specific (lines 6–7). He narrows his question to calls and specifies the subject matter (complaints about state attorneys). This seems to be based on an intuitive notion of what one would be likely to remember. Taylor responds by dealing with the matter in more general terms. She gets a large number of communications each day and so cannot reasonably be expected to remember their content. Instead of mentioning complaints about state attorneys specifically, she speaks only of a nondescript ‘‘myriad of topics’’. She thus generalizes telephone calls to communications and omits the feature that might make certain calls memorable. In structure and function, this is rather similar to Frake’s skin disease example. The level of generalization is manipulated to fit the argument that they are making and, more particularly, to bolster the credibility of the participants’ claims. Case 5: Specification and Meaning Another aspect of the relationship between generality and specificity is revealed in 15 the following extract, from Lawrence 1996 . The manager of a brothel in a midWestern town is being interviewed on the radio. (Lawrence 1996) 1. IR: And you were providing a service 2. IE: .h Yes services were provided .h Mister Mellish I 3. can tell you (.) I can remember a time when fathers .h 4. brought their sons there .h to keep them from uh getting 5. young girls pregnant (.) .hhh uh picking up u:h (.) 6. something off of the street (0.6) picking up a venereal 7. disease (0.8) .hhh u:h I can remember a motherh that had 8. a somewhat retarded so:n (1.3) that waited across the 9. tra:ck (.) wh- while he came down to my house .h my 10. particular house .hhhhhhhhhh u:h I can remember picking 11. up the social section of the newspaper and seeing young 12. men that I knew as customers .hh that married nice 13. gir:ls .hh uh but while they were courting .h um while 14. they were (0.3) into u:h .hh the relationship .h they 15. didn't want they didn't want to go out on these girls 16. (1.1) they used the house and its facilities 17. (1.3) 18. IR: .hh Rather than: (0.3) what is the saying violate 19. the one that they love 20. IE: Yes or as opposed to going to a bar and maybe 21. getting mixed up with a bar gir:l .h o:r u::h (0.8) .h 22. (0.2) u:h there is a difference (.) between a house 23. girl and a street girl an' anything .h we- u- let me 24. define what street means that means any .h girl that 25. workd on the street .h or in a tavern (0.2) or- (.) 26. massage parlor .h where uh: (0.8) she doesn't have a 27. place to live right along with it (.) those are street 28. girls My attention was drawn to the word ‘‘service’’. One might think that the claim that a brothel provides a service requires no discussion, since prostitution is, more or less by definition, a service. It turns out, though, that she is talking about service to the community and not simply the sexual services that are provided to her customers. One such service was to help prevent unwanted pregnancies and venereal disease. Note that it is the father who brings his son—she is providing a service to the family. The second story is similar in that it is the mother who brings her retarded son for a life experience that he might otherwise have no access to. She provides a service to courting couples, in the third example, by providing the male an alternative to what she calls street girls. She goes on to point out the way in which house girls are superior to street girls. Thus, each of these narratives represents a kind of service (Fig. 4 ). 16 Fig. 4 Service Note that this taxonomy combines a single word (‘‘service’’) with more complex discursive units. That is, the taxonomy includes both an expression and several narratives, thus crossing the line between occasioned semantics and formulation in the wider sense. The narratives help to define what IE means as ‘‘service,’’ which might otherwise be understood to refer to a purely physical service—giving the customer what he wants. The work of defining is frequently done, as it is here, through specification. At the same time, IE reflexively classifies these narratives as service, thus instructing us on how they are to be understood. Note, for instance, that the first two narratives could be offered, in other circumstances, as examples of child abuse, or at least poor parenting. Case 6: Specification and Sequence This example is taken from a phone conversation between Linda Tripp and Monica Lewinsky. They are discussing the fact that Lewinsky intends to lie in court about her relationship to the President. Tripp says that, if she is called to testify, she will feel compelled to tell the truth, which will contradict Lewinsky’s account. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Tripp 6-1.1, 4+ minutes into the recording. T=Tripp, L=Lewinsky T: (*) Look Monica: .hh We already know: that you're gonna lie under oath. .hh We also know that I want out of this big time. (.8) I mean (.) awh: how else is this gonna go.=If I have to testify, (.6) if I am forced to answer questions and I answer truthfully, it's gonna be the opposite of what you say so therefore it's a conflict right there. (.8) L((plaintively)): But it doesn't have t'be a conflict.= T: =Whataya mea:n. (1) Ho:w. Tell me ho:w. (2) Tell me how. .hh I'm supposed to say=if they sa:y has Monica Lewinsky ever said to you (.6) tha::t she:: (.3) is in love with the president or is having a physical relationship with the president=if I say no that is (*) perjury .hh (.) that's the bottom line. L: (****) T: I will do everything I can: (.) not to be in that position. (.6) That's what I'm trying to do. But I don't know how else to do this. .hh I'm coming up with foot surgery in California.=I will do anything= 17 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 L: =I ^kno:wˇ (.5) T: But no you really don't know cause you don't believe me. (.5) I think you really belie:ve (.) [that this is= L: [(*) T: =very ea:sy –an' I should just say ((one or two syllables deleted)) if they can't prove it L: I believe you but obviously: -I don't have the same feelings about the situation that you do so I [(*) T: [What do you mean. L: .hh Because i- if I had the same feelings that it was so .hh (.5) wron:g (.5) t'do: (.) t t t deny- to deny something (.8) then: then I would not be doing it. Ya see what I mean? T: .hh I think down deep ya don't like t having to lie. (.5) L: ^Of course I don't think anybody ^likes to:.^ (.5) I don't think anybody likes to –but if they [(**)= T: [but the scary thing to me is L: =this is how family is (.) This is- I would- lie on the stand for my family. That is how I was raised as family. (.5) T: I don't think I- ya know what Monica you're gonna die here an' I would do almost anything for my kids but I don't think I would lie on the stand for them. L: Wull- I mean I would just that's just- for me: ya know that's just how: (1.5) T: I mean: [(.) I have (this) L: [(an') I'm sure you were raised too in a very: (.) sort of honorable: (.5) [(**) T: [I would have b- I would have probably been: (.5[) tarred and feathered (.) if= L: [(*) T: =I even considered saying something like that in my house.= L: =Right (.) [(*) an' I'm- an' I was brought up with lies.= T: [(look) L: =(.) all (.) the time. (.5) So tha- that was how: (.) that was how: you got along -in life -was by lying. T: I don't believe that. Is that true? L: ^Yes it's true. Y I wanted something from my dad, well first my parents were divorced.=If I wanted money from my dad, I had to make up a story. .hh When my parents were married .hh my mom would –always (.) lying to my dad (.) for everything. (.5) Everything. My mom helped me sneak out of the house (1) I mean that's just how: (.) that's just how I was raised. (2) 18 68 69 70 71 72 73 74 75 76 77 T: Well (.) [In in the Catholic religion there are white= L: [(*) T: =lie:s and there are black lies.=Those are white lies. Those ar- (.) those are like white lies.=Those are kind of like lies wL: You look great. T: When you really look (.) t.hhh That's .hh that's not what I'm talking about here. L: I mean I have outright (.) I lie: to my dad. I have lied my entire life 19 When T asserts that she would not lie on the stand even for her own children (lines 44–45), L begins to make a direct contrast between herself and T, but aborts this in favor of a comment on T’s upbringing. This sets up a contrast with her own upbringing. Her argument, as it develops, is not that she is lying for her family, but rather that she does not mind lying because of her family. (It appears that her ‘‘family,’’ in this case, is her mother. Her father is the one who was lied to.) T agrees that she was raised in an ‘‘honorable’’ way, thus providing the first item in the contrast.9 L continues: L: =Right (.) [(*) an' I'm- an' I was brought up with lies.= T: [(look) L: =(.) all (.) the time. (.5) So tha- that was how: (.) that was how: you got along -in life -was by lying. When T expresses doubt (line 61), L offers specifics: L: ^Yes it's true. Y I wanted something from my dad, well first my parents were divorced.=If I wanted money from my dad, I had to make up a story. .hh When my parents were married .hh my mom would –always (.) lying to my dad (.) for everything. (.5) Everything. My mom helped me sneak out of the house (1) I mean that's just how: (.) that's just how I was raised. (2) This instance of specification is somewhat different from that offered in case 5. In the radio interview the categorization and the events categorized are discrete items. That is, the narratives exist independently of the way they are categorized. On the other hand, in case (6), the category ‘‘lie’’ more or less constitutes the examples rather than telling us how we are to interpret them. The events, as formulated, are not only categorized by the concept of lying; also they consist of lying. We have seen in the radio interview example one function of offering specifics. It helps to define what one is talking about. Also, it seems that somehow specifics stand as a form of evidence, even when they are no more proven than the original assertion. Therefore, expressions of doubt regularly elicit specifics. Perhaps the specifics show how the general statement could be true, thus making it more credible. At any rate, it is T who seizes on the defining function of L’s specifics. She no longer resists L’s assertion that she was brought up with lies, but she interprets the statement in a way that suits her. The conversation to this point might be (partially) conceptualized in the following diagram (Fig. 5 ). 9 This is a common way to do certain kinds of tactical contrast—get the other to provide or accede to the first term. It is also a way to introduce an item into a conversation. In this case, show a proper (and, at least in this case, sequentially motivated) interest in the other’s upbringing, then reciprocate with information about one’s own upbringing. 20 Fig. 5 Lies T uses L’s specifications to introduce a new categorical distinction: white lies vs. black lies. She associates this with the Catholic religion, perhaps to explain why she is making a distinction which seems to have escaped L. L may have grown up lying but her lies were white lies and therefore insignificant. That is not, as T says, what is being discussed here (i.e., lying under oath). Therefore, L’s upbringing, although it may be as she describes it, has no application to the present situation. The diagram, incorporating L’s proposal now looks like this (Fig. 6 ). her father Fig. 6 White and black In line 73, L offers an understanding of ‘‘white lie,’’ which T validates in line 74. But in lines 76–77, L asserts that she has ‘‘lied my entire life’’. She appears to be ignoring the distinction that T is making, or at least rejecting the significance that T assigns to it. We have here a clear example of how categorical distinctions are sequentially motivated. L’s offering of details follows T’s expression of doubt. And T’s introduction of the distinction between white and black lies utilizes materials supplied by L’s attempt at specification. Conclusions In the above examples, I have illustrated various manifestations, implications, and functions of a general feature of expressions in conversation—generalization/ specification. Some findings: 1. The level of generalization may limit what can be said next and may supply a 21 basis for advancing the conversation in particular directions by providing for certain implicatures or inferences. 2. The direction and level of generalization may be, at least in part, determined by such matters as the kind of argument being constructed and the discursive environment. 3. The use of a generalized contrast may provide for the categorization of some original item or set of items. 4. Generalization and specification may be used strategically to achieve rhetorical effects; to persuade, justify, excuse, and so forth. 5. The relation between a general item and its specifications may reflexively reveal the nature of the items concerned. That is, the meaning of the general expression may be defined by its specifications at the same time that the meaning of the specific items are defined by the general expression. 6. Categorizations and specifications are sequentially motivated. 7. Specification provides a kind of quasi-evidence.10 If nothing else, I hope that I have demonstrated the value of studying this conversational phenomenon. I have chosen diverse examples from a variety of settings in an attempt to demonstrate the richness of the subject. However, I have presented my analyses as exemplifications of one possible aspect of an occasioned semantics and, ultimately, an approach to formulation. Other features of conversational meaning, such as contrast (Deppermann 2005 ) and scaling, presumably have a similar diversity of function. Whether a fully formed, systematic approach to formulation will take the form I have envisioned and partially exemplified remains to be seen. But I am convinced that the current analytical boundaries of conversation analysis need to be broadened to include matters which traditionally concerned linguistic pragmatists, semanticists, linguistic philosophers, and other students of meaning and rhetoric, even including scholars practicing so-called critical discourse analysis and so-called postmodern approaches. In recent decades, there has been a turn, led by Wittgenstein (1958 ), Austin (1962 ), Bakhtin (1986 ), Garfinkel 1967 ), Sacks(1992 ), and others, from language as saying something to language as doing something and from semantics to indexicality. It is time to unite the old concerns with the new, as Sacks began to do in his discussions of categories. We cannot go back to abstract discussions of meaning in language—when we speak or write, we are doing something, and we are doing it in a context that partially determines what we are taken to mean. But we cannot lose sight of the fact that the doing is accomplished by means of saying, or, to put it in Austinian terms, illocution is accomplished through 10 I have argued in another publication (Bilmes 2009b) that at least some metaphors may also be considered generalizations of a sort, but I have chosen not to consider metaphor in this article, since it raises issues that are not especially relevant to the current discussion. 22 locution. We need to concern ourselves, once again, with the ‘‘what’’ as well as the ‘‘how,’’ with content as well as operations. We have, on the one hand, the technical, methodologically-informed studies of conversational interaction; on the other, we have the concern with the topics and meanings expressed through language, approached largely through ‘‘interpretation,’’ akin to literary criticism. In the former case, the focus has been primarily on conversation and related forms of verbal interaction; the latter tends to focus primarily on texts and monologs (although Bakhtin has shown us that these too are interactive). In terms of this distinction, linguistic pragmatics may be said to occupy a middle ground—it deals with the meanings of utterances and textual passages in a technical way. In fact, some pragmatists (e.g., Labov and Fanshel 1977 ) even work with extended stretches of actual conversation. But their analyses and results— which frequently draw on unexamined ‘‘mentalistic’’ resources, such as intention and motivation—tend to be dominated by their theoretical orientations, which limits their approach to meaning, interaction, and conversational (or textual) structure and constrains their sensitivity to indexical matters in favor of general rules (Streeck 1980 ; also Bilmes 1988 ; Schegloff 1988b ). I am suggesting, therefore, that we find a new way of bridging the gap between action and meaning, between doing and saying, an approach that maintains something of the CA sensibility exemplified in Sacks’ lectures. I propose that this approach consist of a methodologically-informed study of formulation, rhetoric, and meaning structure. I view this as an attempt to enrich CA, to broaden its reach. SA is, and always has been about more than the ‘‘clacking of turns’’ [in Moerman’s (1988 : xi) memorable phrase]. Nevertheless, SA has been primarily concerned with the organization of conversational interaction, with the organization of turns and actions. I am proposing a parallel concern with the organization of meaning in conversation. Appendix Three of Sacks’ major points on ‘‘The baby cried. The mommy picked it up’’. are: (a) Sacks asks how we know that the mommy is the mommy of the baby. ‘‘Baby,’’ he claims, belongs to two ‘‘devices’’—a device being a collection (i.e., a set of categories that go together, that is, a set of categories which are themselves members of a more general category, as ‘‘male’’ and ‘‘female’’ are members of the category ‘‘sex’’) and its rules of application—a stage-of-life device and a family device. ‘‘Baby’’ is a member of the category ‘‘age group’’ and also a member of the category ‘‘family member’’. The ambiguity is resolved by the use of ‘‘mommy’’. Sacks proposes three rules to explain this. According to the ‘‘economy rule,’’ ‘‘a single category from a membership categorization device can be referentially adequate’’. So, for instance, the category ‘‘baby’’ is sufficient reference for the purposes of this story. The consistency rule states ‘‘If some population of persons is being categorized, and if a category from some device’s collection has been used to categorize a first member of the 23 population, then that category or other categories of the same collection may be used to categorize further members of the population’’. So, if ‘‘oak’’ is mentioned, there is a distinct possibility that ‘‘pine’’ will be mentioned also. There is also a ‘‘corollary’’ of the consistency rule which holds that ‘‘if two or more categories are used to categorize two or more members of some population, and these categories can be heard as categories from the same collection, then: Hear them that way’’. So, given that both ‘‘mommy’’ and ‘‘baby’’ can be heard as members of the family category, that is how we will hear them. Moreover, since families, like teams, are ‘‘duplicatively organized,’’ we will hear them as belonging to the same family. I think that Sacks’ argument on this point is flawed. To begin with, ‘‘baby,’’ unlike ‘‘child,’’ is not ambiguous as between age and family membership. ‘‘Baby,’’ in its literal usage, refers to an age-group. For this very reason, one cannot literally refer to an adult as a baby, although an adult can, literally, be someone’s child. (The online Miriam-Webster dictionary supports me on this.) Many languages have two words that would be translated as ‘‘child,’’ one for child as offspring and one that refers to an individual as a member of an age group. Furthermore, Sacks ignores a crucial aspect of the word’s context—its situation within a nominal phrase. Since, I have argued, ‘‘baby’’ is not ambiguous to begin with, let’s use the semantically ambiguous ‘‘child’’ as our example. The phrase ‘‘the child’’ signals, in most contexts, that we are referring to age, whereas ‘‘my child’’ refers to family membership. So, even if ‘‘baby’’ were ambiguous, the phrase ‘‘the baby’’ would make it clear that the reference is to age. (Of course, the phrase ‘‘my baby’’ does invoke family membership, but so does ‘‘my teen-ager’’ or ‘‘my little fireman’’. It is the possessive pronoun that recruits teen-agers and firemen—and babies—into the family.) Why, then, do we hear ‘‘the mommy’’ as the parent of ‘‘the baby’’ in this story? Given that the adult in question could be formulated as, e.g., woman or lady, we are led to ask why this more marked formulation. ‘‘Mommy’’ is unambiguously a family member. We assume, therefore, that she is formulated in this way because she is the mother of the baby who cried. (Consider: ‘‘The teen-ager was rude. The mother scolded him.’’) But, again, the nominal phrase is crucial. As Arnulf Deppermann has pointed out to me, if it were ‘‘a mommy’’ who picked the baby up, we would probably understand that it was not the baby’s mother. We require a linguistic, as well as categorial, analysis. One further thought: If the story was ‘‘The baby cried. The nanny (or the babysitter) picked it up,’’ would we not be likely to hear ‘‘the nanny’’ as the nanny of the baby? How does Sacks’ analysis apply to this case? I think, as in the case of ‘‘mommy,’’ that the interpretation is guided by the categorization of the adult. Why formulate the woman as a nanny, unless she is the nanny of the baby? Despite these flaws in Sacks’ analysis, his economy and consistency rules are powerful. I am interested in particular in the corollary to the consistency rule. Take, for instance, the statement ‘‘He hates rats and squirrels.’’ These animals 24 can be heard as belonging to the same collection—rodent or small mammal or some such—but I don’t think that we necessarily hear them in that way. It may be that he doesn’t like rats and he doesn’t like squirrels, period. In contrast, ‘‘He hates rats and squirrels and so forth’’ calls for the inference of a superordinate category, with ‘‘rodents’’ as one possibility. Compare ‘‘He hates rats and cockroaches and so forth,’’ which might lead us to infer ‘‘vermin’’ as the collection. (b) The second major point of Sacks’ analysis is that membership categories are associated with particular activities. He calls these ‘‘category-bound’’ activities. [Watson (1978 ) has proposed the phrase ‘‘category-bound predicate,’’ to accommodate the fact that categories may have associated attributes other than activities.] Thus, the category ‘‘baby’’ is associated with crying. (Note that this lends further weight to my contention that ‘‘baby’’ in this story refers unambiguously to the age-group category.) When a category is invoked, it brings with it a set of possibly relevant predicates. And mention of an activity or attribute, in turn, may invoke the associated category. The concept of category-bound activity is not exactly original—one thinks, for example, of role theory—but Sacks’ use of it to analyze conversation was an innovation, and subsequent studies in MCA rely heavily on this concept. (c) Sacks points out that the baby’s crying and the mommy’s picking it up are ordered acts. First the baby cried and then the mother picked it up. He proposes that this perception of sequence is produced not so much by the ordering of the sentences (cf., Labov and Waletzky 1967 ) as by our notion that a mother picking up a baby is commonly occasioned by the baby crying. (See, especially, Sacks 1992 : 244–245.) That is, we see not merely a sequence of behaviors, but events linked by causation. We suppose that the mother’s action was consequent on the baby’s. I think Sacks is not entirely correct on this point. How would we hear ‘‘The mommy picked the baby up. It cried.’’? It would seem to me that the ordering of the sentences is, in this case, determinative. I think this story would normally be heard as suggesting that the mother, by picking the baby up, caused it to cry. Perhaps the more crucial point that Sacks is making, though, is that, when we hear of some sequentially ordered sequence of events, we infer, when necessary, some mechanism— causation, purpose, etc.—that connects the events and accounts for their ordering. One can go to other sources (e.g., Lepper 2000 ; Schegloff 2007a ; Silverman 1998 ) for more complete (and less critical) expositions of MCA. Some, most notably Jayusi (1984 ), have tried to elaborate Sacks’ notions regarding categories. Others (e.g., Watson 1997 ) have argued for a more complete integration of MCA with SA. References Antaki, C. (2008). Formulations in psychotherapy. In A. Perakyla, C. Antaki, S. 25 Vehvilainen, & I. Leudar (Eds.), Conversation analysis and psychotherapy (pp. 26–42). Cambridge, England: Cambridge University Press. Arminen, I. (2005). Institutional interaction: Studies of talk at work. Aldershot, England: Ashgate Publishing. Austin, J. L. (1962). How to do things with words. Oxford: Clarendon. Bakhtin, M. (1986). Speech genres and other late essays. Austin: University of Texas Press. Barnes, R. (2007). Formulations and the facilitation of common agreement in meetings talk. Text & Talk, 27(3), 273–296. Billig, M. (1996). Arguing and thinking: A rhetorical approach to social psychology (2nd ed.). Cambridge, England: Cambridge University Press. Bilmes, J. (1988). Category and rule in conversation analysis. Papers in Pragmatics, 2(1), 25–59. Bilmes, J. (1993). Ethnomethodology, culture, and implicature: Toward an empirical pragmatics. Pragmatics, 3(4), 387–409. Bilmes, J. (1995). Negotiation and compromise: A microanalysis of a discussion in the Federal Trade Commission. In A. Firth (Ed.), The discourse of negotiation: Studies of language in the workplace (pp. 61–81). Oxford: Pergamon Press. Bilmes, J. (2008). Generally speaking: Formulating an argument in the US federal trade commission. Text & Talk, 28(2), 193–217. Bilmes, J. (2009a). Taxonomies are for talking: A reanalysis of a Sacks classic. Journal of Pragmatics, 41(8), 1600–1610. Bilmes, J. (2009b). Kinship categories in a Northern Thai narrative. In H. Nguyen & G. Kasper (Eds.), Talk-in-interaction: Multilingual perspectives (pp. 29–56). Honolulu: National Foreign Language Resource Center. Bilmes, J. (2010). Scaling as an aspect of formulation in verbal interaction. In Language learning and socialization through conversations (pp. 3–9). Conference proceedings. Center for Human Activity Theory, Kansai University, Osaka, Japan. Conklin, H. (1969). Lexicographical treatment of folk taxonomies. In S. Tyler (Ed.), Cognitive anthropology (pp. 41–59). New York: Holt, Rinehart and Winston. 26 Deppermann, A. (2005). Conversational interpretation of lexical items and conversational contrasting. In A. Hakulinen & M. Selting (Eds.), Syntax and lexis in conversation: Studies on the use of linguisticresources in talk-in-interaction (pp. 289–306). Amsterdam: John Benjamins. Deppermann, A. (2011). Notionalization: The transformation of descriptions into categorizations. Human Studies, 34(2). doi:10.1007/s10746-011-9186-9. Drew, P. (2003). Comparative analysis of talk-in-interaction in different institutional settings: A sketch. In P. Glenn, C. LeBaron, & J. Mandelbaum (Eds.), Studies in language and social interaction: In honor of Robert Hopper (pp. 293–308). Mahwah, NJ: Lawrence Erlbaum Associates. Drew, P., & Heritage, J. (1992). Analyzing talk at work: An introduction. In P. Drew & J. Heritage (Eds.), Talk at work: Interaction in institutional settings (pp. 3–65). Cambridge, England: Cambridge University Press. Edwards, D. (1994). Script formulations: An analysis of event descriptions in conversation. Journal of Language and Social Psychology, 13(3), 211–247. Edwards, D. (1995). Two to tango: Script formulations, dispositions, and rhetorical symmetry in relationship troubles talk. Research on Language and Social Interaction, 28(4), 319–350. Edwards, D. (1997). Discourse and cognition. London: Sage. Frake, C. (1961). The diagnosis of disease among the Subanun of Mindanao. American Anthropologist, 63(1), 113–132. Frake, C. (1969a). The ethnographic study of cognitive systems. In S. Tyler (Ed.), Cognitive anthropology (pp. 28–41). New York: Holt, Rinehart and Winston. Frake, C. (1969b). A structural description of Subanun ‘‘religious behavior’’. In S. Tyler (Ed.), Cognitive anthropology (pp. 470–487). New York: Holt, Rinehart and Winston. Gafaranga, J., & Britten, N. (2004). Formulation in general practice consultations. Text, 24(2), 147–170. Garfinkel, H. (1967). Studies in ethnomethodology. Englewood Cliffs, New Jersey: Prentice Hall. Garfinkel, H. (2002). Ethnomethodology’s program: Working out Durkheim’s aphorism. (edited and introduced by A. Rawls). Lanham, Maryland: Rowman & Littlefield. 27 Garfinkel, H., & Sacks, H. (1970). On formal structures of practical actions. In J. McKinney & E. Tiryakian (Eds.), Theoretical sociology: Perspectives and developments (pp. 337–356). New York: Appleton-Century-Croft. Geoghegan, W. (1973). Natural information processing rules: Formal theory and applications to ethnography. Monographs of the Language-Behavior Research Laboratory, University of California, Berkeley. Goodwin, C. (1994). Professional vision. American Anthropologist, 96(3), 606–633. Grice, H. P. (1975). Logic and conversation. In P. Cole & J. Morgan (Eds.), Syntax and semantics 3: Speech acts (pp. 41–58). New York: Academic Press. Hauser, E. (2011) Generalization: A practice of situated categorization in talk. Human Studies, 34(2). Heritage, J., & Raymond, G. (2005). The terms of agreement: Indexing epistemic authority and subordination in talk-in-interaction. Social Psychology Quarterly, 68(1), 15–38. Heritage, J., & Watson, D. R. (1979). Formulations as conversational objects. In G. Psathas (Ed.), Everyday language: Studies in ethnomethodology (pp. 123–162). New York: Irvington. Hester, S., & Eglin, P. (1997). Culture in action: Studies in membership categorization analysis. Washington, DC: University Press of America. Horn, L. (1972). On the semantic properties of logical operators in English. PhD thesis, University of California, Los Angeles. Housley, W., & Fitzgerald, R. (2009). Membership categorization, culture and norms in action. Discourse and Society, 20(3), 345–362. Jayusi, L. (1984). Categorization and the moral order. London: Routledge and Kegan Paul. Labov, W., & Fanshel, D. (1977). Therapeutic discourse: Psychotherapy as conversation. New York: Academic Press. Labov, W., & Waletzky, J. (1967). Narrative analysis. In J. Helm (Ed.), Essays on the verbal and visualarts (pp. 12–44). Seattle: University of Washington Press. Lawrence, S. (1996). Normalizing stigmatized practices: Achieving co-membership by 28 ‘‘doing being ordinary’’. Research on Language and Social Interaction, 29(3), 181–218. Lepper, G. (2000). Categories in text and talk: A practical introduction to categorization analysis. London: Sage. Levinson, S. (1983). Pragmatics. Cambridge: Cambridge University Press. Moerman, M. (1988). Talking culture: Ethnography and conversation analysis. Philadelphia: University of Pennsylvania Press. Moerman, M., & Sacks, H. (1988). On understanding in the analysis of natural conversation. In M. Moerman (Ed.), Talking culture: Ethnography and conversation analysis (pp. 180–186). Philadelphia: University of Pennsylvania Press. Popper, K. (1959). The logic of scientific discovery. London: Hutchinson. (First published in German in 1935). Potter, J. (1996). Representing reality: Discourse. rhetoric, and social construction. London: Sage. Raymond, G., & Heritage, J. (2006). The epistemics of social relations: Owning grandchildren. Language in Society, 35(5), 677–705. Sacks, H. (1972). On the analyzability of stories by children. In J. Gumperz & D. Hymes (Eds.), Directions in sociolinguistics: The ethnography of communication (pp. 329–345). New York: Holt, Rinehart and Winston. Sacks, H. (1992). Lectures on conversation. Edited by G. Jefferson. Oxford: Blackwell. Schegloff, E. (1988a). Description in the social sciences I: Talk-in-interaction. Papers in Pragmatics, 2(1/2), 1–24. Schegloff, E. (1988b). Presequences and indirection—applying speech act theory to ordinary conversation. Journal of Pragmatics, 12(1), 55–62. Schegloff, E. (1991). Reflections on talk and social structure. In D. Boden & D. Zimmerman (Eds.), Talk and social structure (pp. 44–70). Cambridge, England: Polity Press. Schegloff, E. (1993). Reflections on quantification in the study of conversation. Research on Language and Social Interaction, 26(1), 99–128. Schegloff, E. (2007a). A tutorial on membership categorization. Journal of Pragmatics, 29 39(3), 462–482. Schegloff, E. (2007b). Categories in action: Person-reference and membership categorization. Discourse Studies, 9(4), 433–461. Schegloff, E. (2007c). Sequence organization in interaction: A primer in conversation analysis 1. Cambridge, England: Cambridge University Press. Silverman, D. (1998). Harvey sacks: Social science and conversation analysis. New York: Oxford University Press. Stivers, T. (2007). Alternative recognitionals in person reference. In N. Enfield & T. Stivers (Eds.), Person reference in interaction: Linguistic, cultural and social perspectives (pp. 73–96). Cambridge, England: Cambridge University Press. Streeck, J. (1980). Speech acts in interaction: A critique of Searle. Discourse Processes, 3(2), 133–154. Turner, R. (1971). Words, utterances, and activities. In J. Douglas (Ed.), Understanding everyday life: Toward the reconstruction of sociological knowledge (pp. 169–187). Chicago: Aldine. Walker, E. (1995). Making a bid for change: formulations in union/management negotiations. In A. Firth (Ed.), The discourse of negotiation: Studies of language in the workplace (pp. 101–140). Oxford: Pergamon Press. Watson, D. R. (1978). Categorization, authorization and blame: Negotiation in conversation. Sociology, 12(1), 105–113. Watson, R. (1997). Some general reflections on ‘categorization’ and ‘sequence’ in the analysis of conversation. In S. Hester & P. Eglin (Eds.), Culture in action: Studies in membership categorization analysis (pp. 49–75). Washington, DC: University Press of America. Wittgenstein, L. (1958). Philosophical investigations (2nd ed.). Oxford: Blackwell.