Some approaches proceed in their analyses from sharper starting points than others. The objectives of interactional sociolinguistics, for example, seem more clearly defined than those of speech act theory. Broadly construed, interactional sociolinguistics views "language as indexical to a social world."1 By analyzing discourse, interactional sociolinguistics aims to reveal the social meaning of conduct in a particular context and of the interaction between self and other associated with the conduct. On the other hand, speech act theory, though not initially formulated as an approach to discourse, concerns itself "with what people do with language."2 Such a starting point may be overly broad to be of insightful use in discourse analysis. Schiffrin, for instance, points out that such all-inclusive functional approaches threaten "to submerge discourse analysis within the broader and more general analyses of language functions, without leaving a space within which discourse analysts can formulate a clear set of principles, goals, topics, and methods specific to their own enterprise."3 Although Schiffrin does not explicitly place speech act theory in such a category during her discussion of the definition of discourse analysis in Chapter 2, she implicitly does so because speech act theory is, in effect, an approach that takes language use as its central domain. Thus, the focus of speech act theory may address only the surface characteristics of the language used in a sentence within a wider discourse, providing little more than a sophisticated taxonomy of the obvious. As such, a speech act theory analysis may provide only superficial descriptions of the functions of words or sentences in a discourse while lacking the machinery to connect those functions with not only with the rest of the discourse but also with the sociological and cultural influences that surface in language and often decisively affect meaning. Interactional sociolinguistics may be better positioned than speech act theory to reveal these cultural influences and their effect upon meaning.

As I mentioned above, questions must be narrowly formulated, objectives clearly laid out. Taken together, questions and objectives help govern the choice of approach. On closer inspection, however, a problem soon emerges from such a point of departure. It leads to a theoretical morass, a circle of determination from which it is impossible to escape with objectivity intact: the research agenda determines the approach; the approach singles out certain phenomena over others; the analysis of the singled-out phenomena that the approach generates answers the questions or meets the objectives specified at the outset. The circle is now complete: The questions we ask produce the answers we seek. The findings are radically determined by the perspective.

There are a few ways out of this predicament. The simplest is to acknowledge that a certain amount of subjectivity, of methodological determinism, is inherent in whatever choice of approach is made. First heeded by the sociologist Max Weber, this position maintains that the researcher's objectives and subsequent analysis are necessarily influenced by his or her values, which, Weber believed, could not be justified "scientifically," that is, through value-free analysis.

Thus, in comparing different approaches to discourse, one system cannot be chosen over another without taking a value or end -- such as reducing miscommunication between teachers and pupils -- into consideration. (Even though Schiffrin analyzes the same text and reaches somewhat different conclusions using different approaches, the subjectivity inherent with approaches to discourse does not seem to be made explicit in Approaches to Discourse, but is made early on in Brown and Yule's Discourse Analysis : "The perception and interpretation of each text is essentially subjective."4)

A second, more rigorous retreat from the circle of methodological determinism would be to maintain that the results of an approach's analysis must be (more or less) reproducible by at least one but preferable two other approaches. Such reproduction could be taken as partial but not absolute verification.

A third possibility would be to apply several approaches to the discourse in question and closely compare the results, drawing conclusions only after divergent or competing explanations are resolved.

In any event, all three retreats necessitate that conclusions are seen as explanations, not truths. A complete comparison of the three possibilities and their methodological consequences lies beyond the scope of this essay.

Speech act theory, Schiffrin writes, "focuses upon knowledge of underlying conditions for production and interpretation of acts through words."7 From this characterization, I expect the approach to be particularly helpful in discovering the underlying conditions of the production of (2) and its interpretation as borne out by the response in (3).

At the core of speech act theory are explicit and implicit performatives, sentences that "are not used just to say things, i.e. describe states of affairs, but rather actively to do things."8 To succeed, performatives must meet certain felicity conditions, outlined by Stephen Levinson in Pragmatics :

A. (i) There must be a conventional procedure having a conventional effect

(ii) The circumstances and persons must be appropriate, as specified in the procedure

B. The procedure must be executed (i) correctly and (ii) completely

C. Often, (i) the persons must have the requisite thoughts, feelings and intentions, as specified in the procedure, and (ii) if consequent conduct is specified, then the relevant parties must so do.9

What is James doing in (2)? What does (2) count as in the context given by the exchange? At face value, (2) could be taken merely as a literal report about James's internal state: He does not know. The problem with this interpretation, however, is that it does not correspond well with the teacher's response: There is a gap between the literal meaning of (2) and the effect it brings about in its hearer. The teacher apparently takes James's (2) as meaning that James does not feel like trying. At least from the teacher's perspective, which is assumed to be informed by James's previous behavioral patterns, James, in uttering (2), is doing something more than merely reporting his inner state; he is avoiding the question. James is uttering what John Austin would call an implicit performative. The teacher gives little weight to the rising intonation of James's response, an intonation pattern that opens the possibility of a third analysis of James's speech act.

The rising intonation of James's utterance, as Schiffrin points out in her chapter on interactional sociolinguistics, can be interpreted as indicating a need for encouragement. Seen in this way, the utterance meets most of Austin's felicity conditions. Briefly, James's utterance is a conventional procedure, at least within his African American community, and is supposed to have a conventional effect, the elicitation of encouragement. The procedure seems to be executed completely and correctly. The circumstances seem appropriate enough, but the persons as specified in the procedure may not be: The teacher, who is white, does not belong to James's African-American community. Thus, James can be seen as performing the act of requesting encouragement by uttering (2) but it misfires because it is directed to an inappropriate person. James's speech act is a conventional procedure only within his community.

To summarize, under the theory of speech acts at least three possible explanations, given the data above, can be drawn:

(i) James is giving a straightforward report of his inner state, which is misinterpreted by the teacher.

(ii) James is trying to avoid the question, and is rightly interpreted by the teacher as trying to do so.

(iii) James is asking for encouragement but the act misfires because it violates the felicity condition stipulating that the persons involved fit the procedure.

We have arrived at the heart of the problem with speech act theory. There are three alternative explanations, all equally plausible. Two of the explanations involve speech acts. The other views the utterance as merely a literal report. The explanatory power of speech act theory stops here. It provides no useful criteria for determining whether James's utterance in fact a speech act or merely a report. If we concede that speech act theory "ends up as a general theory that pertains to all kinds of utterances,"10 not just explicit and implicit performatives, then we can at least say that all three of the utterances are speech acts. But this gets us nowhere: We are left with three alternative acts and no criteria, perhaps save felicity condition C above, for determining which procedure is actually being undertaken by James. We cannot further establish how to categorize James's utterance in the taxonomy of speech acts. Condition C -- "often, (i) the persons must have the requisite thoughts, feelings and intentions, as specified in the procedure" -- is of no help, not only because we don't even know the procedure he is undertaking but also because we cannot know what he is thinking, feeling or intending.

In summary, speech act theory provides only possible descriptions of the meaning of the pupil-teacher exchange with no criteria other than C, which is empirically unacceptable, for choosing among the alternative interpretations. Beyond trying to determine James's thoughts, feelings, and intentions by interviewing him or experimenting on him, which may be empirically impractical or unreliable (James may be unavailable for comment, may be unwilling to cooperate, may give a false impression of his thoughts, may be misled by the questions, etc.), we cannot even begin to choose among the three alternatives of what he may mean or intend by uttering "I don't know" in the classroom context of our teacher-pupil example. Applying speech act theory to the pupil-teacher example shows that what x counts as in context y is often indeterminable because the "requisite thoughts, feelings and intentions" of the speaker are frequently neither known nor ascertainable. The connection to conventions that speech act theory establishes in felicity condition A begins to shed some light on the matter but fails to provide the necessary connection to culture that could help determine whether James's rising intonation indicates a need for encouragement or is merely a noncontrastive element of the utterance.

The conjecture that a strong connection to culture could advance the interpretation of James's remarks is supported by the would-be failure of applying a Gricean pragmatic analysis to his utterance. Indeed, such an analysis, when based upon conversational implicature, would likewise get us nowhere, though an analysis based on conventional implicature begins to point in the right direction, just as the use of felicity condition A did in the speech act theory analysis. Yet the same interpretations remain equally probable, and Gricean pragmatics would provide little basis for choosing one interpretation over another in a principled way.

Speech act theory does have a certain utility, however. The approach reveals disjunction between two of the three possible interpretations of James's utterance and the teacher's response. But the theory stops short of providing a principled method by which to identify the nature of the utterance and to pinpoint the reason for the disjunction where it arises.

Interactional sociolinguistics, Schiffrin says, takes from John Gumperz a focus "on how people from different cultures may share grammatical knowledge of a language, but differently contextualize what is said such that very different messages are produced."11 As such, Gumperz aims "to develop interpretative sociolinguistic approaches to the analysis of real time processes in face to face encounters."12 The constructs that Gumperz's interactional sociolinguistics uses to analyze discourse include contextualization cues, which are signaling mechanisms "of language and behavior (verbal and nonverbal signs) that relate what is said to the contextual knowledge ... that contributes to the presuppositions necessary to accurate inferencing of what is meant."13 The definition of contextualization cues, Schiffrin points out, contains the notion of contextual presupposition, the assumed knowledge that permits the inferencing of two related levels of meaning: the kind of communicative activity being undertaken in the utterance and the conventional communicative force that the speaker intends by making the utterance.

Schiffrin's chapter on interactional sociolinguistics uses the teacher-pupil exchange to demonstrate how Gumperz's contextualization cues can affect the interpretation of a message. The teacher's response, Schiffrin says, indicates that the teacher is interpreting James's "I don't know" not only according to its literal meaning but also as suggesting that James does not want to try to answer the question. Yet in James's African-American community, rising intonation conveys the desire for encouragement. Thus, the teacher, in Gumperz's view, "did not retrieve the contextualization presuppositions needed to accurately interpret Freddy's message from his use of rising intonation."14 In this way, contextualization cues can affect the message's meaning. "The methodological consequence of this," Schiffrin writes, "is that one can discover shared meaning by ... using the reaction that an utterance evokes as evidence of whether interpretive conventions were shared."

In the pupil-teacher exchange, there may be a dislignment between the pupil's remark and the teacher's reply, as the speech act theory analysis showed, depending on what James actually means. Interactional sociolinguistics also exposes this disjunction: The meaning of James's utterance, except if viewed as avoidance, does not seem to be shared by both the teacher and him.

At least in light of the data given, however, interactional sociolinguistics also falls short of giving the meaning of James's remark, though the approach does provide a principled method by which, with further study of the reactions that James's "I don't knows" receive in his usual cultural setting, conclusions could be drawn about the likelihood of what James meant.

The empirical methods are these: We could follow James around in his community for several days, pay particular attention to the intonation of such utterances as "I don't know," and gauge whether the utterances, when made with rising intonation, elicit a response of encouragement. If such a response predominates, we can assert that in his conversation with the teacher, James was in all likelihood seeking encouragement. Yet this investigation still does not show us what James was thinking at the time; the meaning of that particular utterance remains, in the end, indeterminable, save another method of analysis that can tell us with some degree of certainty what he was thinking at the time.

Such an empirical line of investigation begins to lead away from interactional sociolinguistics and into the ethnography of communication, which quite generally puts forth "a methodology by which to discover `what counts' as communicative events" using the distinction between emic and etic.15 In other words, by following James around in his community, we could determine whether, in his culture, rising intonation in such responses as "I don't know" is "emic," or classifiable as contrastive for meaning. The approach would also use the methodology of Labovian sociolinguistics, whereby a statistical probability would be established for whether James sought to avoid the question or receive encouragement with his reply to the teacher's question. Thus we have arrived at a more principled theoretical approach and a tenable empirical method for interpreting James's remark.