conversation, they are often forced to feel their way -- to try to determine as best they can the kind of the game that is being played, its rules and goals, its boundaries and players. Yet precisely how interpretation and response are carried out in the game of conversation remains a puzzle. The goal of discourse analysis, as broadly conceived, is to unravel this mystery: To describe the game, to illuminate its often obscure rules, to clearly mark out its boundaries and to identify its players, coaches and referees.

Although discourse analysis has come to be seen as a subdiscipline of linguistics, the roots of several of the seven established approaches to

discourse have grown out of philosophy, and at least two of them are based directly in the writings of prominent philosophers. The approach known as speech act theory was formulated by the philosopher John L. Austin and developed by John Searle. A second approach, often called pragmatics, has its foundations in the writings of H.P. Grice. Both approaches have been influenced, at least on the margins or in their maturation, by Wittgenstein's later writings, especially Philosophical Investigations. There are especially strong parallels between speech act theory and Wittgenstein's emphasis on usage and language-games.2

This essay seeks to take Wittgenstein's influence on discourse analysis a step further by using his writings as the theoretical foundation for an approach to analyzing discourse that is distinct from speech act theory,

which stems from the analytic tradition in philosophy, and to suggest that a Wittgenstein-inspired approach may actually be closer in spirit and content to that of an unlikely candidate whose views, in contrast to the analytic school, harbor a distinctly Continental flavor which has come to influence critical theory: Mikhail Bakhtin.

The essay begins by attempting to outline in fairly broad strokes an approach to discourse analysis based on Wittgenstein's philosophy of language in Philosophical Investigations. The approach will appeal to the following Wittgensteinian views and constructs: a focus on ordinary language, meaning as use, the language-game and context, function, speech activities, the connection of language to life, the role of customs and rule- governed activities, the indeterminacy of meaning, an antipathy to reductionism, and a focus on moves in a game. As I develop a Wittgenstein-driven approach to discourse, I will point out in passing several ways in which it differs from a speech act approach.

As the essay progresses, I will increasingly turn toward analyzing a particular kind of discourse from a Wittgensteinian perspective: psychoanalytic conversation. Even though I do not realize this ambition here, it is my hope that, as an extension of Wittgenstein's views on the interpretation of an utterance's meaning in a particular context, a Wittgensteinian approach to discourse will provide the basis for analyzing the rhetoric that surfaces in psychoanalyst-patient dialogue. Toward this end, the essay will close by briefly testing the approach's explanatory yield by applying it to an early psychoanalytic conversation.

The Basis for a Wittgensteinian Approach
Some aspects of Wittgenstein's thought as laid out in Philosophical Investigations readily lend themselves to analyzing discourse. His focus on ordinary language, rather than on such ideal or logical languages, is, for instance, a rather obvious starting point for any meaningful analysis of conversation or text. Such logic-based approaches to studying meaning as Tarskian truth-conditional semantics, while they have their place in postulations about how semantic meaning may be represented in the mind or brain of a speaker, are of little help in analyzing the meaning of utterances in actual conversation. Quite obviously, the analysis of meaning in the context of social discourse must be addressed not through formal logic or the modeling of cognitive structures but through attention to the influence of cultural and social factors on the use and interpretation of language across contexts.

Wittgenstein's appeal to ordinary language leads him to abandon the tendency to attribute the meaning of a word to the object it names, offering instead a nonreductionist principle that forms the foundation of a Wittgensteinian approach to analyzing conversation: "For a large class of cases--though not for all--in which we employ the word `meaning,'" he writes, "it can be defined thus: the meaning of a word is its use in the language."7 Thus, a participant in a conversation interprets a word's meaning in accord with its use. This principle is a fine starting point, but as a theoretical construct it may be so broad that its application to conversation may be limited, leaving some common signs accounted for. For instance, it cannot account for the inferences and implicatures that people commonly make in conversation. Furthermore, appealing to the use of a word may capture its direct meaning but leave untouched meanings that manifest themselves in the tone or inflection with which the word is used. Wittgenstein is not unaware of this objection. He quickly moves to delimit his definition of meaning more sharply by asking how to explain the difference in meaning between a report and a command that employ the same words. The answer: "It is the part which uttering these words plays in the language- game."8 With this remark Wittgenstein slightly restricts his earlier remark on meanings while expanding it from the level of the word to that of the utterance, producing a principle with greater utility for an approach to conversation, where the emphasis must be more on the interpretation of utterances than single words. Wittgenstein's remark can be seen as accomplishing two moves crucial to extending his philosophical perspective to the analysis of discourse: First, it expands the focus of analysis from the use of words alone to their use in an utterance; second, it connects use to not only the language in general but the particular context in which it is being used at the time of the utterance.

After introducing meaning as use and the notion of a language-game in Philosophical Investigations, Wittgenstein quickly adds another angle to the point of view he is developing: function. For Wittgenstein, the function of language goes beyond the mere conveyance of thought. He exhorts us to "make a radical break with the idea that language always functions in one way, always serves the same purpose: to convey thoughts."9 For Wittgenstein, language, including discourse in both conversation and text, can serve a multiplicity of functions or purposes. Language's function, in any given case, can be just as much to do something as to convey a thought. In this way, Wittgenstein's notion of function can, I believe, be distinguished from his notion of use. A function of a word or sentence is tied up more with what it is used to do, in contrast to its use, which is its meaning: A word is used to do something; the use of a word or a sentence has a function: A word or sentence is used to carry out a function.

A problem that arises by associating a word's function with what it is used to do is how to differentiate Wittgenstein's notion of the function of language from that of speech act theory, which also focuses on what an utterance is used to do. The difference lies in the connection that Wittgenstein makes between activity and the function of language -- both of which, in his view, may take an infinite variety of forms. Further: Activity and language are in turn closely connected to life, a relation that seems to be missing in speech act theory.

Speech Acts and Activities
In accord with Wittgenstein's emphasis on the use and function of words, a Wittgensteinian approach to discourse includes an acknowledgement that words are used as much to do things as to make statements. That words are used to do things is also the central tenet of John L. Austin's speech act theory. In remarks made to a Harvard audience in 1955 and later published

Bakhtin derives his claim from an analysis of 's work, which was also a favorite of Wittgenstein. Bakhtin stood in awe of the multiplicity of voices and languages (metaphorically, within say, Russian), and it was in the work of Dostoyevsky that Bakhtin found the ultimate celebration of the multiplicity of languages, a multiplicity that he believed did "a kind of justice to life itself."16

In his turn and from his own perspective, Wittgenstein exhorts us to "review the multiplicity of language-games in the following examples, and in others," after which he makes a list of such starkly different language activities as those quoted in (i) and (ii) below:

(i) "giving orders"
"asking, thanking, cursing, greeting, praying"

and

(ii) "describing the appearance of an object"
"reporting an event"

The activities in (i) and (ii) are seen by many of today's semantic

theorists as being distinct: The meaning of expressions that fall under set (i) is dependent in large part on what it is used to do; the meaning of set (ii), on the other hand, is often formulated in reference to the truth conditions of the expressions. But Wittgenstein makes no bones about placing both kinds of utterances into the same broad category: The uttering of sentences -- whether statements or performatives -- is "part of an activity."

In this way Wittgenstein's view of language closely mirrors Austin's. By the end of How to Do Things with Words, Austin has expanded his argument beyond explicit performatives to include all kinds of utterances, including statements, for "stating," he says, "is performing an act."17 Austin sums up his argument with the following conclusion: "The truth or falsity of a statement depends not merely on the meanings of words but on what act you were performing in what circumstances."18 Recall, however, that Wittgenstein gets one up on Austin: Like M. Bakhtin, Wittgenstein makes a connection between life and language: Speaking is an activity, a form of life.

Word and Original Action
Considering that I am ultimately interesting in answering the question, What takes place in psychoanalytic dialogue?, connecting language and action may provide a particularly powerful framework for viewing psychoanalytic discourse. As Jean Piaget puts it, "the psycho-analysts have shown us how close in their opinion is the bond which originally connected word and action, words being so packed with concrete significance that the mere fact of uttering them, even without any reference to action, could be looked upon as the factor in initiating the action in question."19 In fact, as Piaget tells us, "the word, [psychoanalysts] say, having originally formed part of the act, is able to evoke all the concrete emotional contents of the act."20 Thus, the connection Wittgenstein makes between language and action may prove particularly valuable when a Wittgensteinian approach to discourse is applied to psychotherapeutic conversation, because, as Bakhtin puts it, dialogue "is not the threshold to action, it is the action itself."21

Piaget's citation of the bond that originally connected word and action brings us to another aspect of Wittgenstein's perspective that may prove useful when directed toward discourse: his heuristic of appealing to primitive states. "It disperses the fog," Wittgenstein says, "to study the phenomena of language in primitive kinds of application in which one can command a clear view of the aim and functioning of the words."22 And, when I turn to an analysis of psychotherapuetic discourse later in this essay, I will do just that: I will examine a historic pre- Freudian exchange between psychiatrist and patient.

Culture and Convention
Another element of a Wittgensteinian discourse analysis is the connection that Wittgenstein repeatedly makes in Philosophical Investigations between language and culture, with societal conventions, norms, and rules falling under the general rubric of culture: They are, for Wittgenstein, customs. "To obey a rule, to make a report, to give an order, to play a game of chess, are," he writes, "customs (uses, institutions)."23

Wittgenstein is not, of course, the first or only language analyst to make such connections with regard to discourse: That culture heavily influences conversational styles in systematic ways is the central tenet of the approach to discourse analysis known as the ethnography of communication, which examines how culturally generated rules determine the underlying structure of conversation. For these ethnographers, "culture encompasses or embraces a totality of knowledge and practices," including speech acts.24 As such, the ethnography of communication subsumes speech act theory. Perhaps, then, a Wittgensteinian approach to communication that not only acknowledges the totality of cultural knowledge and practices involved in discourse but also emphasizes that language is itself a form of life becomes superior to both the ethnography of communication and to speech act theory.

Cultural Rules
As the ethnographers of communication and other linguists have shown in their studies, the use of rule-governed activities can often be empirically tested and verified. An empirical analysis of the role of rule-governed behavior in conversational interaction can stand near the center of a Wittgensteinian approach; its empirical basis endows it with a robust methodological power because hypotheses about conversation that take place in one interaction can be verified by examining other conversational interactions. The findings, that is, are reproducible. As a result, William Labov points out, many linguists and sociologists "have been focusing on the rule- governed character of conversation and uncovering preliminary principles which suggest that this activity may be as well formed as the production of sentences."25 Such comments as Labov's, however, should not be taken as meaning that there is already a systematic rule-governed approach to discourse, whether based on the writings of Wittgenstein or anyone else.

On the other hand, Wittgenstein's rule-based approach to explaining meaning has been criticized as too broad, as too encompassing, to be useful. The danger of the rules-for-use approach to meaning, Janet D. Fodor writes as she considers a Wittgensteinian approach to generative semantics, is that "it might be too broad to be useful." Thus, she continues, if the rules-for-use approach "is to serve as the basis for a theory of meaning, the notion of a use must be constrained in some way."26

Wittgenstein's view that interpretations themselves do not determine meaning points to another strength of his approach: Its antipathy to reductionism, its resistance toward assimilation, its aversion toward attempts to develop an all-embracing theory. A monolithic, unified theory need not -- indeed cannot -- account for all aspects of conversational meaning and understanding. And attempts should not be made to treat conversation thusly, for meaning and understanding vary greatly, as Wittgenstein suggests, from case to case, context to context. Neither rule-governed use nor other forms of interpretation can account for the infinite variety of things people do with language and for the functions that utterances serve. No unified theory can account for the variety of linguistic life. Any attempt to do so will be futile.

Conversational Moves
Finally, the crowning metaphor in a Wittgensteinian approach to discourse is the attention to the moves that speakers make in conversation. A focus on moves in a game can be mined not only from Wittgenstein's metaphor of the game in general and of chess in particular, but also from such typical entreaties as to consider "the part which uttering these words play in the language-game."34 In fact, the notion of a move in a language- game becomes explicit when Wittgenstein is ruminating in §22 over Frege's view that every assertion contains an assumption, which Wittgenstein says rests on the possibility of writing every statement as "it is asserted that such and such is the case." Wittgenstein retorts: "But `that such-and-such is the case' is not a sentence in our language -- so far as it is not a move in the language-game."35

The construct of a move closely mirrors the notion of footing that Erving Goffman employs in his interactional sociolinguistics approach to conversation: Footing, Goffman says, concerns "the alignments we take up to ourselves and the others present as expressed in the way we manage the production and reception of an utterance."36 Goffman's focus on alignments is quite similar to Wittgenstein's notion of communication as moves in a game: "In terms of the chess analogy, communication is a matter of the players' appropriate responses to each other's moves in accordance with the rules of the game," Roy Harris writes of Wittgenstein's language-game metaphor in Language, Saussure and Wittgenstein: How to Play Games with Words.37

By using the utterances "come to trouble you" and "crave your indulgence," Golyadkin also seems to be acknowledging his subordinate position in the game; as such, the use of these two utterances is linked closely to the social status of each participant, with it perhaps being customary to address doctors with deference. To summarize, the meaning of these utterances, in addition to the straightforward meaning generated by the use of their composite words, is the expression of deference, whereas their function is the indication of Golyadkin's purpose.

But next, Golyadkin, having begun his spiel, abruptly halts and finds himself at a loss for words. Perhaps he does not want the doctor to listen to him after all. The shift from seeking a listener for his words to having no words to say at all takes place at Dostoyevsky's insertion of the ellipses.

This appraisal of the first paragraph establishes two categories of constructs: the notion of the language-game is a macro-level analytical tool, whereas meaning as use, whether applied to words or utterances, is a micro-level construct.

Before Golyadkin has even begun to disclose what's troubling him, Christian Ivanovich, taking up the role of authority that Golyadkin has bestowed upon him, admonishes Golyadkin "to follow my instructions." The language-game has shifted. It is now an authority figure giving some rather stern counsel to a listener, and no longer a speaker seeking a sympathetic listener. But again, at the ellipsis, a change in the voice occurs: After the ellipsis Christian Ivanovich's advice loses its stern, monologic style and is supplanted by a more congenial, more suggestive dialogic voice -- a voice that more intimately addresses Golyadkin, a voice that seems to sense that Golyadkin needs sympathy, or at least a softer tone. How can a Wittgensteinian approach to discourse account for this subtle shift in voice? Seeing both segments as speech acts seems to be of little help: Both before and after the ellipsis, the doctor is engaging in the same general linguistic action: giving advice. Likewise, the language-game seems to shift little, if at all, in the segments before and after the ellipsis. Perhaps a customary rule dictates that an interlocutor shall soften the delivery of his admonitions after the first volleys have been fired. This possibility, however, does not account for my intuition that Christian Ivanovich is beginning to sense the need for sympathy in the other and to address him accordingly. Thus, perhaps Christian Ivanovich is making a subtle conversational move with the other in mind; that is, he is shifting his alignment with regard to the other.

This odd turn of events can be accounted for quite generally by arguing that both the doctor and Golyadkin are making a move - - concession -- at the same time. They are realigning their utterances to those of the other, a realignment that may be necessitated on both of their parts by certain cultural conventions that are at play. For instance, one such convention may be to not continue for too long any conflict that arises in a conversation in a particular language game, such as between doctor and patient. Or, to put it another way, the degree of sustainable conflict may be establishing a boundary for this conversation. With both Golyadkin and Ivanovich bumping up against the boundary, neither is certain what rules, if any, dictate the appropriate response. But without knowing more about the conversation rules and social mores of the times, this explanation cannot be confirmed. As it stands, then, the motivation for their mutual realignment is indeterminate. And, as we have seen above, the fact that a Wittgensteinian approach to conversation can leave meaning indeterminate can be seen in cases like this one as a strength.

Some Strengths and Limitations of the Approach
Although Wittgenstein's metaphor of conversation as ball game is a particularly useful lens through which to view conversation, there are, as my analysis of the excerpt from The Double illustrates, several weaknesses of an approach that uses Wittgenstein's philosophy as the basis for an analysis of discourse. Many of the shortcomings stem from the macro-level analysis that the approach generates: The approach's orientation may be too wide to be used successfully in a line-by-line analysis of a conversation. When applied to actual conversation, the notion of a move in a language-game, for instance, is so broad as to be nearly useless: It can account for just about anything done in conversation but only in quite general terms. Without knowing precisely what the rules of conversation are before conducting the analysis, we find ourselves left in the position of ascribing a move to an unknown or as yet undetermined rule. On the other hand, the analysis of a move may help unearth a rule or custom.

The connection that Wittgenstein makes between linguistic activities and language as a form of life is likewise problematic. It is unclear how such a broad declaration, while interesting in theory, can be applied in any specific way to analyzing conversation. With regard to psychoanalytic discourse, however, the connection does have a certain utility: A person's way of life and governing perspective are embedded in his or her manner of speaking. Thus, to shift that way of speaking significantly can be seen as a way to alter his perspective -- and his life.

Another aspect of conversation that seems difficult to account for within a Wittgensteinian perspective is the dialogic nature of the utterance -- the fact that any one utterance or any conversational turn may be marked by a multiplicity of voices coming from a single speaker. Furthermore, as the analysis of the passage from The Double attests, it is similarly difficult to account for subtle differences in the way that one speaker addresses another. In other words, it is difficult to find a principled place in Wittgenstein's perspective for the roles played by sympathy, empathy, and other emotional orientations to the other that take place at the margins of much discourse, especially psychoanalytic discourse. Indeed, it is difficult, given Wittgenstein's view of language combined with his belief that everything is open to view, to see how a Wittgensteinian approach to conversation can account for times when a person has said "one thing, but seemed to imply something quite different by the same words,"39 as the

Conclusion
From Philosophical Investigations I have mined those aspects of Wittgenstein's perspective toward language, meaning, and understanding that lend themselves to analyzing conversational discourse. In doing so, I have also revealed some of Wittgenstein's own ideas about meaning and understanding in conversation and of how they differ from Austin's theory of speech acts. I have also suggested that Wittgenstein's views on language in general and utterance meaning in particular are closer to the likes of M. Bakhtin than many in the analytical school typically recognize. More importantly I have shown how a Wittgensteinian approach to discourse can provide an account of what takes place in conversation. But whether a Wittgensteinian form of discourse analysis can compete with approaches like Dell Hyme's ethnography of communication or Gumperz and Goffman's interactional sociolinguistics -- especially in analyzing discourse, including psychotherapuetic conversation, on a microscopic level -- remains in question.