Davidson’s approach to content is based on the thought experiment of radical translation. In order to clarify what we understand when we understand our home language, Davidson considers the conditions of possibility of the radical interpretation of a foreign language. Radical interpretation is supposed to be interpretation from scratch [Davidson 1993: 77]. It is a philosophical abstraction from the kind of interpretation undertaken by a field linguist having first contact with an alien tribe. Such interpretation - it is assumed - cannot appeal to bilingual speakers or dictionaries. It precedes those resources. Furthermore, according to Davidson, it cannot make substantial use of the content of the mental states of speakers. Whatever the connection between mental content and linguistic meaning, radical interpretation must earn access to, and cannot simply assume, facts about both. The intentional contents to which Grice appeals in the analysis of linguistic interchange, for example, cannot be identified prior to the interpretation of the agent’s language. Thus they cannot simply be appealed to in radical interpretation.
Instead, interpretation must rely only on the evidence of correlations between utterances and the circumstances which prompt them:
[The radical interpreter] interprets sentences held true (which is not to be distinguished from attributing beliefs) according to the events and objects in the outside world that cause the sentence to be held true. [Davidson 1983: 317]
Davidson’s account of radical interpretation is a development from Quine’s account of radical translation. But one key difference is its characterisation of the evidence available:
The crucial point on which I am with Quine might be put: all the evidence for or against a theory of truth (interpretation, translation) comes in the form of facts about what events or situations in the world cause, or would cause, speakers to assent to, or dissent from, each sentence in the speaker’s repertoire. We probably differ on some details. Quine describes the events or situations in terms of patterns of stimulation, while I prefer a description in terms more like those of the sentence being studied; Quine would give more weight to a grading of sentences in terms of observationality than I would; and where he likes assent and dissent because they suggest a behaviouristic test, I despair of behaviourism and accept frankly intensional attitudes toward sentences, such as holding true. [Davidson 1984: 230]
Davidson takes the evidence available to radical interpretation to be worldly facts and events in the environment of speakers together with the occasion of their utterances. The role that evidence plays is important and I will return to this issue having sketched in the underlying purpose of radical interpretation.
Davidson’s methodological claim for the philosophy of content is that one can clarify the nature of both linguistic meaning and mental content more generally by examining how it is determined in radical interpretation. ‘What a fully informed interpreter could learn about what a speaker means is all there is to learn; the same goes for what the speaker believes.’ [Davidson 1983: 315] Because it is intended to serve this philosophical purpose, Davidson concentrates on clear instances of radical interpretation - interpretation by field linguists - rather than the ‘interpretation’ that, he claims, takes place in daily life:
All understanding of the speech of another involves radical interpretation. But it will help keep assumptions from going unnoticed to focus on cases where interpretation is most clearly called for: interpretation in one idiom of talk in another. [Davidson 1984: 125-126]
Nevertheless, Davidson also thinks that everyday understanding of language involves radical interpretation. That claim puts some strain on the initial characterisation of radical interpretation as interpretation from scratch because it undermines the contrast that such a description presupposes. If everyday ‘interpretation’ is also really from scratch, what example could there be of interpretation which was not? But whilst Davidson makes this claim in part to defend his radical thesis that communal language plays no explanatory role in human understanding, it can also be seen as a reminder of the purpose of considering radical interpretation. (The radical claim is defended in [Davidson 1986a].) That is to shed light on what is understood when we understand speech and action generally. (I will return to the question of whether everyday understanding can be called interpretation at all.)
Seen in this light, Davidson’s account of radical interpretation serves as an example of reconstructive epistemology. It does not matter that our everyday understanding of other speakers does not proceed using the tools that Davidson describes. One might argue that everyday understanding works on the implicit and tacit assumption that others speak the same language as oneself. But radical interpretation does not aim at phenomenological accuracy. Similarly, it would not matter if real field linguists made use of interpretative heuristics less minimal than those Davidson describes. An example of that might be the assumption that any newly-encountered human language has a good chance of being related to some previously-encountered language. Such a principle would be useful if it turned out that all human languages sprang from a common source. Since radical interpretation is really a piece of reconstructive epistemology, it concerns the ultimate justification of ascriptions of content whatever the actual process of reasoning which gives rise to them. It concerns the evidence that could be used to justify both the possible heuristic suggested above and also our everyday methods of understanding. Radical interpretation is supposed to explain what the assumption that other speakers speak the same language amounts to. (According to Davidson, one of its consequences is that such talk of shared languages is of no philosophical significance.) It is precisely because it plays a clarificatory - via a justificatory - role that radical interpretation is characterised in the austere terms that it is.
The early Davidson sometimes suggests that progress can only be made if the evidence used in radical interpretation is not described in question begging-terms:
[U]ninterpreted utterances seem the appropriate evidential base for a theory of meaning. If an acceptable theory could be supported by such evidence, that would constitute conceptual progress, for the theory would be specifically semantical in nature, while the evidence would be described in non-semantical terms. [Davidson 1984: 142]
Since one of the obstacles to using Davidson’s account of language to shed light on Wittgenstein’s is precisely the question of evidence, it is worth taking this point slowly.
What reason did Davidson have for thinking that the evidence for a theory of meaning should itself be described in non-semantic terms? Two sorts of consideration suggest two related motivations. From the perspective of providing a general philosophical account of the nature of meaning, the motivation might run like this. If a philosophical account is to shed light on meaning by connecting it - albeit holistically - to the evidence which determines it, then that evidence should not be described in content-laden terms. If it is, no light will have been shed on what meaning is since facts about meaning will have been presupposed in the description of evidence.
The other consideration depends more specifically on Davidson’s chosen approach to the general philosophical account of meaning. Light is shed on meaning in general by reflecting on how radical interpretation is possible. But what is the final justification there is for ascribing meanings to other people? One method an interpreter could adopt for testing her interpretations - which Davidson assumes will form part of an interpretative theory or ‘theory of meaning’ - would be to see whether in individual cases her interpretations agreed with those given by dictionaries or bilingual speakers. But in radical interpretation no such resources can be assumed since they merely postpone the question of how ascriptions of meaning can be justified. In this context it may seem that the evidence which is available must be characterised in non-semantic terms:
In radical interpretation, however, the [interpretative] theory is supposed to supply an understanding of particular utterances that is not given in advance, so the ultimate evidence for the theory cannot be correct sample interpretations. To deal with the general case, the evidence must be of a sort that would be available to someone who does not already know how to interpret utterances the theory is designed to cover: it must be evidence that can be stated without essential use of such linguistic concepts as meaning, interpretation, synonymy, and the like. [ibid: 128 italics added]
If either or both of these arguments were compelling, then Davidson’s approach to the philosophy of content would run counter to that of Wittgenstein. One of the morals of Wittgenstein’s discussion is, as I have described, that no analysis of meaning can be given which turns on the interpretation of otherwise meaningless noises or movements. I will argue shortly, however, that whatever Davidson’s view of the matter, the thought experiment of radical interpretation can be pruned of any commitment to the neutral description in non-semantic terms of the evidence which supports interpretation. Briefly:
· The results of radical interpretation can still be supported or criticised by appeal to evidence even if that evidence cannot be described (qua evidence) in non-semantic terms. The evidence can still be used to test interpretation because of the systematicity and holism implicit in radical interpretation. Just because it is described in semantic terms does not undermine its independence of the theory for which it is evidence. Construing an utterance as a particular assertion will have consequences for how similar-sounding utterances are construed. (As will become clearer later in the chapter, Davidson’s philosophical monism implies that evidence could also be described in non-semantic, meaning-free terms. But this would obscure its role as evidence for an interpretation.)
· It is wrong to assume that radical interpretation only sheds light on meaning if the connection between meaning and evidence is explicitly represented in the results of that interpretation. The nature and limits of possible evidence can play an implicit role. Davidson’s key claim is that there are no facts about meaning which are inaccessible to the radical interpreter. But this connection between meaning and interpretation need not be encoded in the results of radical interpretation. Thus there is no need to require that the input can be characterised in non-semantic terms.
In fact Davidson himself realises that his project cannot escape all intentional notions and in later accounts drops the requirements about its non-semantic nature:
My way of trying to give an account of language and meaning makes essential use of such concepts as those of beliefs and intention, and I do not believe it is possible to reduce these notions to anything more scientific or behaviouristic. What I have tried to do is give an account of meaning (interpretation) that makes no essential use of unexplained linguistic concepts. (Even this is a little stronger than what I think is possible.) It will ruin no plan of mine if in saying what an interpreter knows it is necessary to use a so-called intensional notion - one that consorts with belief and intention and the like. [ibid: 175-6]
Given the strategic options for explicating linguistic meaning and mental content described in the first chapter, this distinction between intentional concepts and semantic concepts may suggest that Davidson advocates something like intention-based semantics. It may seem, in other words, that he is suggesting that linguistic meaning can be analysed in terms of mental content. But that is not the strategy he follows. The distinction between the narrowly semantic and the more broadly intentional marks his preferred strategy for coping with a general difficulty for radical interpretation.
Davidson thinks that, ultimately, facts about mental content have to be determined in the same way as facts about linguistic meaning. Meanings and contents are interdependent. This presents a principled difficulty for radical interpretation:
A speaker who holds a sentence to be true on an occasion does so in part because of what he means, or would mean, by an utterance of that sentence, and in part because of what he believes. If all we have to go on is the fact of honest utterance, we cannot infer the belief without knowing the meaning, and have no chance of inferring the meaning without the belief. [ibid: 142]
Thus the interpreter faces the task of unravelling two sets of unknowns - facts about meaning and facts about beliefs - with only one sort of evidence: linguistic actions which depend on both meaning and belief. How can the interpreter - to change the metaphor - break into this interdependent set of facts?
Davidson’s solution has two ingredients. Firstly, he takes the evidential basis of radical interpretation to be the prompted assent of a speaker, which he characterises as ‘the causal relation between assenting to a sentence and the cause of such assent.’ [Davidson 1983: 315] The reason for this is that it is possible to know that a speaker assents to a sentence without knowing what the sentence means and thus what belief is expressed by it (or vice versa). Characterising a speaker as holding a particular sentence true is an intentional interpretation of what is going on - the speaker is described by relation to a propositional content - but it does not presuppose a semantic analysis of the sentence. That will be derived later.
The second step is to restrain the degrees of freedom of possible beliefs in order to interpret linguistic meaning. The interpreter must impose his or her own standards of truth and coherence on ascriptions of beliefs and meanings. There must be a presumption that any utterance or belief held true really is true. Further, in a significant range of cases, the interpreter must assume that the object of an utterance, and the belief the utterance expresses, is the cause of the utterance and belief. (As Davidson remarks in a passage quoted above, the relevant cause is a worldly state of affairs rather than, as Quine suggests, proximal stimulation at the boundary of the body.) This complex of related assumptions governing the rationality imputed - generally briskly labelled the Principle of Charity - enables interpretation to get off the ground. If utterances are assumed by the interpreter to be generally true and to concern the worldly states of affairs which prompt them, then they can be correlated with those observed states of affairs. Their meaning can thus be determined. Given an overall interpretation, exceptional false beliefs can then be identified.
These a priori constraints on interpretation operate in a general manner but allow exceptions. Thus even the basic datum that a speaker holds a particular utterance true can be revised in the light of the subsequent interpretation of their other beliefs and meanings. The epistemology of interpretation is fallible and holistic. So the appeal to evidence should not be regarded as a foundational or reductive account of meaning. The earlier prescription that evidence should be describable in theory-free non-semantic terms (whilst still being represented as evidence) does not fit easily with the holism which Davidson more generally emphasises.
Davidson’s basic strategy can now be summarised as follows. On the assumption that radical interpretation has access to all the facts about content, content can be explicated by examining the conditions of possibility of radical interpretation. Thus Davidson assumes that content can be captured by a third person perspective and that it can be fully analysed through its connection to the action of agents in the world. In the weakest sense of the term, Davidson can be seen, in his philosophy of content at least, as promoting a form of philosophical behaviourism providing that this is not construed in its Quinean and reductive sense. Meaning is explicated through its role in human behaviour. (In fact, he adds to this picture of content a token identity theory in order to explain the causal role of content to which I will turn in the next chapter.)
Rorty describes this basic approach as the ‘philosophy of language of the field linguist’ [Rorty 1991: 132]. This is the aspect of Davidson’s philosophy of content which he shares with Wittgenstein. I will return to discuss the similarities which result. But in addition to this general approach to the philosophy of content, Davidson advocates a particular formal structure for the results of radical interpretation which has nothing in common with Wittgenstein’s approach.