Language without Languages?  Donald Davidson

on Interpretation and Understanding



I.  Davidson on the concept of a language


Donald Davidson has argued for a view of language use that dispenses with the idea of a language, understood as an abstract system of words with specifiable meanings and rules for combining and using them.[1]  The occasion for making this claim is his analysis of the necessary elements of successful communication.  That is to say, in seeking to explain the human ability to employ language, Davidson has come to the conclusion that the very idea of a language to be learned and spoken should be avoided.  Among the many questions that such a conclusion might stimulate, at least two are prominent.  First, in what sense, if any, can the idea of a language be understood according to Davidson’s perspective?  This question will be important to those of us who recognize the enormous importance of various languages to the lives and beliefs of those who speak them.  Second, what plausible account can Davidson give of the distinction between successful and unsuccessful communicative exchanges between people?  Anyone who views linguistic competence as a distinctive capacity of humans is likely to find this question relevant.

            In response to the latter question, Davidson offers a detailed analysis of a listener’s successful interpretation of a speaker who utters a malapropism or slip of the tongue (“ ‘Lead the way and we’ll precede’ ” (NDE 158)).  The starting point for this discussion, and what makes these mistakes of usage useful for the analysis, is the observation that having a grasp of conventional word meanings and semantic rules does not, in itself at least, explain the ability of the hearer to understand what the speaker actually meant.  Furthermore, other examples (such as the case of two speakers of different languages able to correspond back and forth, each in her own tongue) illustrate that such a grasp of a language is not necessary for successful communication.

            Instead, says Davidson, what each interlocutor brings to such an encounter is a “prior theory” according to which she expects to be able to correctly interpret the speaker (NDE 168).  This theory is fleeting:  the hearer applies it only to the encounter at hand, and the theory is subject to revision or replacement as new information (including the utterance itself) emerges.  Davidson calls the actual means by which the hearer interprets the speaker a “passing theory,” and it may or may not be the same (or relevantly similar) to the prior theory.  For his part, the speaker approaches the encounter with a prior theory (“what he believes the interpreter’s theory to be”) and a passing theory (what “he intends the interpreter to use”) (NDE 168).  Communication is successful on those occasions when the passing theories of the speaker and hearer coincide.  This account begs for more specificity, but the general point is that it does not require reference to a distinct set of conventions previously learned by the participants and governing linguistic practice (NDE 169). 

            The purpose of this paper is to give a more thorough explanation of Davidson’s claim by defending it against two criticisms.  The first argues that knowledge of a language is the only plausible explanation of the capacity for interpreting another’s meaning, so Davidson’s rejection of languages as conceptual schemes fatally undermines the whole project of understanding linguistic competence.  From the other side, it is argued that the proper conclusion to draw from Davidson’s attack on abstract languages is that all talk of theories of language, be they prior, passing, or otherwise, should be rejected; they have no relevance to the practical activity of communication.


II.  Languages as shared conventions

A significant element of many debates regarding language and meaning centers on Wittgenstein’s ‘paradox of interpretation’.  Meredith Williams explains that this argument establishes that interpretation cannot ground the meaning of utterances because any number of possible theories of meaning would be compatible with a given utterance.  The result is that an interpreter, lacking any other grounding of meaning, would have no way to distinguish between correct and incorrect interpretations.  This issue of “normative contrast” is thus applied to Davidson’s account of passing theories, which are constantly being modified so as to make sense of utterances.[2]  According to Williams, Davidson attempts to circumvent this problem by appealing to a speaker’s intention as the standard for judging the correctness of an interpretation.[3]  This sounds consistent with Davidson’s claim, noted earlier, that successful communication occurs when the interpreter’s passing theory matches the speaker’s passing theory (her intention) in relevant ways.  Unfortunately, says Williams, the argument is circular in the following way.  To have intentions and other propositional attitudes, one must be a speaker; that is, linguistically competent.  To be linguistically competent means to be able to interpret (and make) linguistic utterances.  If correct interpretation is dependent on intentions, and intentions presuppose linguistic competence, we have grounded nothing.  Furthermore, argues Williams, to serve as the basis for judging interpretations, a speaker’s intention must reference a particular theory of meaning.  That theory, however, can have no justification as the ‘correct’ theory, even as it is supposed to serve as the standard for correctness when evaluating interpretations of the utterance.[4]

            On its own, this criticism of Davidson’s view fails for at least two reasons.  First, it mischaracterizes the role of intentions in Davidson’s explanation of successful communication.  He is quite explicit that the speaker’s intentions alone cannot ground meaning.  Instead, those intentions are constrained by the speaker’s expectation of what the hearer will take her to mean (NDE 160).  Williams acknowledges this ‘thin intention’, but treats it as an either/or alternative to a ‘thick intention’ which specifies meaningful content (along with an implicit theory of meaning), rather than engaging with Davidson’s description of how these intentions overlap.[5] 

            Second, and more importantly, Williams presumes a notion of correct interpretation to which Davidson would surely object.  Her analysis suggests that successful communication must involve the interpreter getting the ‘right’ theory among all possible contenders.  Davidson’s notion of the ‘right’ theory applies to the set of all possible theories that allow for successful interpretation in the circumstances at hand.[6]  This approach redirects our attention to the ability of speaker and hearer to “go on in the same way.”  Understanding “going on in the same way” to mean having the same theory of meaning, or speaking the same language, is to presuppose the very understanding of these concepts that Davidson seeks to undermine.  Without an additional argument connecting correct interpretation to an abstract language structure, this criticism cannot carry any weight.

            Williams takes up this challenge when she discusses the “instability of language” embedded in Davidson’s perspective.[7]  As we have seen, Davidson imagines participants in linguistic encounters continually adjusting their expectations of how to interpret their interlocutors, on the one hand, and of how their interlocutors are likely to interpret them, on the other.  Furthermore, as individuals move from context to context, encounter to encounter, they continue this process of adopting new expectations and intentions.  Davidson describes this whole process in terms of constructing theories, passing and prior, each of which is temporary and entirely customized to fit the immediate situation.  An overarching, fixed (“right”) theory of language or meaning plays no part in the account. 

            For Williams, this account of an endless procession of passing theories leaves no room for relationships between them, instead forcing an understanding of them as individual, isolated entities:[8]

No sense can be made of going on in the same way with such a picture of the rapid succession of different languages being instantiated, each one intended for the sake of momentary communication.  The temporal dimension of language use is unintelligible on this picture.


These circumstances, she argues, reveal the inability of Davidson’s reliance on speakers’ intentions to make sense of an individual’s development of long-term, complex intentions.  She goes on to observe that Davidson’s discussions of malapropisms, in which his ‘principle of charity’ is employed to explain judgments of correct interpretation, cannot rescue his position from either of the criticisms she levels.  The reason is that charity, whereby an interpreter assumes the correctness of the greatest possible number of the speaker’s assertions, can only occur against a background of overwhelming shared understanding. [9]  Davidson’s choice of a situation involving two competent language users obscures this fact, whereas Williams argues that looking at a learning context reveals the need for shared background, which she identifies with the conventions of a shared language.

            There are really two claims here.  First, successful communication requires a background shared by participants.  Second, that background consists of shared understanding of meanings embedded in a language.  The notion of shared background does bear on this issue (and Davidson acknowledges this), but it should not obscure the analytical work that consideration of malapropisms does for Davidson.  His discussion is meant to establish that a language, thought of as shared conventions, is not sufficient to explain successful communication.  According to Davidson, something like his account of prior and passing theories is necessary to explain communicative exchanges.  Williams is right to point out that this leaves no room for languages as conceptual schemes.  The question remains, must the shared background of a linguistic encounter include a language in this sense? 

            In the case of a learning situation, such as Davidson’s example of a child who responds similarly to objects that an adult recognizes as tables, Williams argues that it is only because the adult has access to the convention of naming these objects “tables” that the two are able to correlate their responses to them and thus achieve the possibility of communication.  The concept of a language, then, gives the adult “the ability to incorporate that utterance [“table”] into a network of inferential connections with other sentences and actions.”[10]  Davidson acknowledges part of this picture.  The child and the adult must respond to the objects in similar ways, and they must recognize each other to be so responding.[11]  They must also possess the mysterious ability to interpret the other, as exemplified in the case of malapropisms.  Williams wants to add conventions of a language to the picture because some norm must serve as the standard by which to judge correct and incorrect interpretations.  Davidson, on the other hand, insists that each participant must make those judgments according to the passing theories they construct.  These, if false, will only be revealed as such when the participants’ responses fail to be correlated. 

            Williams’ reliance on the learning situation undermines her position.  What distinguishes the learning context from other examples is that the normative standard is available only to one participant, the adult.  For the adult to successfully induct the child into “correct” participation in the language game, she must rely on the child’s ability to grasp the meaning of “table.”  Given this ability of the child to interpret prior to having achieved linguistic competence, however, we must ask what work the language is really doing here.  If the child interprets the term broadly, for example, and applies it also to desks and stools, the adult can easily (exercising interpretive ability) recognize this, and adjust his use of the term (as, indeed, parents often do).  Likewise, if the adults use of the term is unusual (“incorrect,” by Williams’ lights), this too need not be a barrier to communication.  When we try to pin down the role of language distinct from its practice, we seem to consistently return to Davidson’s view, or something like it.  At the least, the burden is shifted to those who would argue for the necessity of languages as conceptual schemes to show that communication is impossible without them. 


III.  Language without theory

If Davidson has convincingly discredited the view that languages exist as abstract social conventions that govern and give meaning to linguistic behavior, a question emerges regarding what role theory can have in relation to linguistic practice at all.  From a practical perspective, his analysis of a multitude of passing and prior theories seems unwieldy and intuitively absurd.  From this perspective, Martin Gustafsson has argued:[12]

…one would expect him to draw the straightforward, Wittgensteinian conclusion that faithful descriptions of linguistic abilities can only have the form of piecemeal, non-systematic remarks, and that it is simply false to represent speakers as being armed with systematic meaning theories.


Davidson, however, does not take this route, instead attempting “to compensate for the irregularities of actual linguistic practice by multiplying the number of systematic theories that the speakers are supposed to have.”[13]  From this perspective, Davidson’s aim is to salvage some kind of role for theory when the obvious conclusion to draw is that language or meaning theory has no place in discussions about linguistic practice.

            This charge disregards Davidson’s own repeated qualifications concerning the role of theory and conceptual talk generally.  His claim in these instances is that a full-blown theory of language need not be present in a speaker or interpreter’s consciousness every time a linguistic exchange takes place.  Rather, it is when observers seek to describe such interactions that theory-talk is necessary.  On this view, something like a theory must be available to account for the sense or meaning of an utterance, if it is to count as linguistic practice.  In discussing a speaker’s intentions (in a context where ‘having a theory of language’ would also apply), Davidson explains this idea by analogy (SAL 13-14):

Suppose I put one foot in front of the other in the course of walking to the kitchen to get myself a drink.  I give the motion of my foot no thought whatever, I don’t ask if it is an appropriate means for achieving my purpose.  I am just walking as I habitually do.  But if I were to decide I didn’t want the drink after all, or that the door I was approaching was locked, I wouldn’t take that step.  I had reasons for taking the step and would not take it without the reasons.  Similarly, it seems to me obvious that I would not speak the words I do if I thought I would not be understood.


This is what Davidson means when he speaks of someone “having a reason,” “having an intention,” and in our case, “having a theory.”  It is a way of describing what is unarticulated, but necessarily available when successful communication occurs.

            Of course, this way of putting things does not close the matter.  Even if we accept Davidson’s disclaimers, critics like Gustafsson might ask, why should we think that concepts like theories, intentions, etc… must even be implicitly available in this way?  After all, even Davidson acknowledges that it is an empirical assumption that individuals need a rule or norm in order to grasp meaning in communication, and it is Davidson who is so keen on avoiding talk of empirical issues.[14] 

            Still, this approach seems implausible, especially from someone who claims to be responding from a Wittgensteinian perspective.  After all, it was Wittgenstein who made so much of  “following a rule” in the first place.  If this is an empirical assumption, it is Wittgenstein’s.  Furthermore, Wittgenstein had reasons for making this claim.  Here we can recall our earlier reference to Williams’s description of the “normative contrast,” where some standard is required for distinguishing correct from incorrect interpretations.  If there is an assumption here, it is the assumption that speech has meaning.  The assertion that understanding involves following a rule (where the nature of such a rule is not specified) is carried along with the meaningfulness of speech.  This assumption, in turn, is taken to be a necessary component of communication.

            For Davidson, the combination of assertions, (1) that each new linguistic encounter brings with it a new set of expectations on the parts of speaker and hearer and (2) that each set of expectations must be described on the theoretical level by reference to a full-blown theory of language, results in the crowded picture of passing and prior theories we have been considering (NDE 170).  But what has been neglected in the exchange of the last several paragraphs is the great degree to which Davidson sympathizes with Gustafsson’s perspective.  He does believe, and takes great pains to argue, that inquiry into language and communication should avoid concern with the concept of an abstract structure called a language, and turn instead toward how people come to understand each other (NDE 174). 

            This shift in emphasis does highlight one merit of Gustafsson’s critique:  the concern that philosophical analysis should have some bearing on empirical research.[15]  It is not clear to me whether or not the universal pragmatics advanced by Jurgen Habermas and others is what Davidson has in mind here.  However, I do think that Habermas develops his version of that theory with a very similar understanding of the relationship between language and human interaction.  On the one hand, Davidson’s distinction between the perspective of a participant in communication and that of a philosopher analyzing communication suggests Habermas’ account of the Janus-faced notion of truth that operates as behavioral certainty at the level of action but is subject to rational discourse at the level of analysis.[16]  This preserves Davidson’s understanding of truth both as a primitive concept, irreducible to others in the way it is understood by people in normal contexts, and also as a concept embedded in language and subject to theoretical analysis in that context.[17]  On the other hand, thinkers like Habermas and Charles Taylor demonstrate a willingness to pursue their inquiry into language into the empirical realm by exploring various powerful properties it seems to possess, such as its way of opening up an understanding of the world for cultures and groups,[18] or the implicit validity claims that might ground a discourse-based theory of ethics.[19]  The point here is that this discussion lends support to Habermas’ claim that the linguistic turn in philosophy, rather than narrowing the field of inquiry and ruling out traditional questions about truth, goodness and meaning, instead shifts the ground under which those questions might be explored, “disclosing” rich new ways of understanding, as it were.


[1] Most prominently in ‘A Nice Derangement of Epitaphs,’ Philosophical Grounds of Rationality:  Intentions, Categories, Ends (Oxford:  Clarendon Press, 1986) Richard E. Grand and Richard Warner (eds.) and ‘The Social Aspect of Language,’ The Philosophy of Michael Dummett (Dordrecht, Netherlands:  Kluwer Academic Publishers, 1994), Brian McGuinness and Giandhigi Oliveri (eds.).  Further references to these papers will be noted in the text as NDE and SAL, respectively.

[2] “Wittgenstein and Davidson on the Sociality of Language,” Journal of the Theory of Social Behaviour (2000) 30:3, p. 303.

[3] ibid., p. 304.

[4] ibid., p. 305.

[5] That is to say, with the intention to be understood (“first meaning,” in Davidson’s terminology) directing the subsequent (thicker) meanings, on one hand, and operating at the crossroads where speaker and hearer’s passing theories intersect, on the other.

[6] “The Second Person,” Subjective, Intersubjective, Objective (New York:  Oxford University Press, 2001) p. 110.

[7]“Wittgenstein and Davidson on the Sociality of Language,” p. 305.

[8] ibid., p. 306.

[9] ibid., p. 306.

[10] ibid., p. 314.

[11] “The Second Person,” p. 120.

[12] “Systematic Meaning and Linguistic Diversity:  The Place of Meaning-Theories in Davidson’s Later Philosophy,” Inquiry (1998) 41:4, p. 444.

[13] ibid., p. 444.

[14] ibid., p. 448.

[15] “Systematic Meaning and Linguistic Diversity:  The Place of Meaning-Theories in Davidson’s Later Philosophy,” p. 445.

[16] Habermas, “Richard Rorty’s Pragmatic Turn,” On the Pragmatics of Communication, p. 363.

[17] Davidson, “Epistemology and Truth,” Subjective, Intersubjective, Objective (New York:  Oxford University Press, 2001).

[18] Habermas, “Communicative Rationality and the Theories of Meaning and Action,” On the Pragmatics of Communication, p. 193.

[19] Habermas, “Richard Rorty’s Pragmatic Turn,” p. 371.