PHILOSOPHY OF LANGUAGE

Transcription

1 PHILOSOPHY OF LANGUAGE DAVIDSON ON TRUTH AND MEANING LECTURE PROFESSOR JULIE YOO Theory of Meaning 1 vs Theory of Meaning 2 From Verificationism to Truth-Conditional Semantics Problems with Verificationism Truth-Conditional Semantics Davidson s Three Adequacy Conditions Extensional Adequacy: Convention T Compositionality Interpretation Davidson on Truth and Meaning Page 1 of 8

2 THEORY OF MEANING 1 VS THEORY OF MEANING 2 This expression is unfortunately used in two different ways: one as giving a theory about the nature of meaning, and the other as a way of assigning a specific meaning to a sentence of a language. So an example of a theory of meaning in the first sense is Frege s theory of sense whereas an example of a theory of meaning in the second sense is something like a Quinean translation manual, which is ideally some translation scheme that associates a sentence of the foreign language with a sentence of your own language that means the same as the foreign one. (Quine s theory of meaning in the first sense is that there is no such thing as meaning!) FROM VERIFICATIONISM TO TRUTH-CONDITIONAL SEMANTICS Verificationism In giving an account of the meaning of a sentence, verificationists looked to the set of experiences that would tend to verify its truth: S means that p IFF S expresses the observable conditions for knowing that p. 1 But we saw that this theory ran into a number of problems provocatively raised by Quine. First, verificationism presupposes an analytic synthetic distinction, which Quine argued was untenable, second, it assumed that confirmation or verification is atomistic, contrary to Quine and Duhem s argument that it is holistic. But most importantly from the point of view of giving a theory of meaning 2, the theory yields the wrong content; in other words, a verificationist construal of a (synthetic) sentence ends up attributing an inappropriate meaning to it. This is especially acute when sentences involve theoretic entities: not only do statements about electrons gets reformulated as statements about cloud chambers and microscopes, statements about mental states get reformulated as statements about bodily movements (Behaviorism). In fact, even statements about non-theoretical entities, like tables and chairs, get reformulated as statements about sensory data (Phenomenalism): the statement, There is a table in front of me ends up getting translated as, I am currently having sensory data of a hard surface supported by. Truth-Conditional Semantics Davidson suggests that we can get a better theory of meaning 2 if we look elsewhere for the meaning of a sentence, namely the conditions under which a sentence is true. This approach to giving a theory of meaning for a language is called Truth-Conditional Semantics. The rough idea is fairly intuitive: the meaning of a sentence is the conditions under which it is true. 1 This corresponds to what Foster calls Ayer s content principle, according to which the factual significance of a statement lies in its observational content that is, in its contribution to the deduction of observation-statements (Foster 1985: 17) and a statement has factual significance if and only if its content is purely observational, that is, if and only if the statement falls within the scope of an observational language (Foster 1985: 22) (Miller: 88) Davidson on Truth and Meaning Page 2 of 8

3 S means that p IFF p is the condition under which S is true. We already saw this the truth-conditional approach to a theory of meaning 2 in action in Russell s Theory of Descriptions. Remember that Russell thought he could cull the real meaning, the logical form, of a DD, by stating all of the conditions under which a sentence containing a DD would be true. On Russell s approach, we see the connection explicitly made between the real meaning of a sentence with its logical form, and the logical form of a sentence with its truth conditions. Davidson, who is the main proponent of truth-conditional semantics, explains these connections in greater detail. According to Davidson, a theory of meaning can be given by a theory that generates, for each sentence of the target language, a theorem that states its truth conditions. To understand the gist of this project, think of what we would have to do to write up a languagetranslating computer program, a program that could give the meaning of any incoming sentence of a certain (foreign) language L F. The program would enable a computer to generate, for each sentence S F of the language L F, some sentence of our language that tells us the meaning of the foreign sentence. This would not be one of those clunky programs that only do phrase translations, but a comprehensive program that can take in any novel in-coming sentence of the target (foreign) language and give you its meaning in your home language. Such a program would have to contain, at a minimum, two fundamental categories of things: a vocabulary or lexicon, and a set of rules for combining the vocabulary or syntax. DAVIDSON S THREE ADEQUACY CONDITIONS Davidson attempts to defend this approach by considering certain conditions of adequacy for a theory of meaning. By adequacy conditions, philosophers generally mean those conditions a theory must satisfy in order for it to be acceptable. For instance, no theory about how to cook a fluffy omelet should be circular: to cook a fluffy omelet, you must cook the omelet so that it is fluffy. In fact, this is true of any theory. We can add other adequacy constraints: an adequate theory should not contain an infinite number of instructions; an adequate theory should not appeal to supernatural powers; an adequate theory should be stated in a form that can be comprehended; an adequate theory should be illuminating; an adequate theory should be simple, in the sense that it shouldn t be more complicated than it needs to be. Now, these conditions are general in that they apply to all theories. Different subjects call for different constraints for the theories that are to explain them, and the ones Davidson mentions are these: An adequate theory of meaning must be: Extensionally adequate: The theory must generate a theorem that gives the meaning of each sentence of L, but in a way that does not presuppose meanings (in the Fregean sense). Compositional: The theory has finitely many axioms from which we can generate the meaning-giving theorems in a way that displays its semantic structure how the meaning of the whole sentence depends upon the meaning of its parts. Davidson on Truth and Meaning Page 3 of 8

4 Interpretive: The theory must enable us to interpret correctly the linguistic output of its speakers, in accordance with the constitutive principles of interpretation, such as the Principle of Charity. Extensional Adequacy: Convention T Davidson considers a number of approaches to giving a theory of meaning that conforms to the above constraints. According to Davidson, the theory that can satisfy these conditions while giving a theory of meaning for each of the sentences in the target language, is a truth theory, or T-theory, adapted from Tarski s theory of truth. A T-theory is always a theory of a particular language L; it has the capacity to generate shed light on the meaning of sentences of L by giving us their truth values. These are represented as the T-sentences of L: (T) S is true-in-l if and only if p. Compare (T) with our original schema: (F) S means m. A schema like (F) represents the Fregean approach to giving the meaning of a sentence, where one specifies what m is in terms of certain abstract entities senses and propositions. But not only did Davidson find the appeal to such entities ontologically troubling, he thought that instances of (F) were ultimately unilluminating. As Davidson explains at the beginning of his paper, Truth and Meaning : A Fregean answer [to the question, what is the meaning of the sentence Theatetus flies ] might go something like this: given the meaning of Theatetus as argument, the meaning of flies yields the meaning of Theatetus flies as value. The vacuity of this answer is obvious. We wanted to know what the meaning of Theatetus flies is; it is no progress to be told that it is the meaning of Theatetus flies. This much we knew before any theory was in sight. In the bogus account just given, talk of the structure of the sentence and the meanings of words was idle, for it played no role in producing the description of the meaning of the sentence. (Davidson 1984: 20) To construct a more illuminating theory of meaning, Davidson suggests that we look to Tarski s Theory of Truth, and the type of truth-definition found in it to give us what we want to know about the sense of a linguistic expression. The value of such a theory is that it gives us a way to match the appropriate truth conditions for each of the sentences of L. The matches, which take a sentence of L and give you information about what it means, are called the theorems of the T-theory for L. Instances of T are called theorems, as theorems are things that are provable in a deductive system strictly on the basis of that system s explicit assumptions, and the T-sentences are generated by the corresponding T-theory in exactly this fashion. Notice that the T-schema does not make any reference to meaning, as it is couched strictly in terms of the material bi-conditional. Davidson on Truth and Meaning Page 4 of 8

5 Now, one might ask why we don t use a schema like this: (M) S means that p. Unlike (F), (M) does not postulate entities to give the meaning of S, but rather states it in the form of that p. (M) certainly seems more straight-forward than (T), so why not use it? The reason Davidson does not use it is because it presupposes exactly what he wants to explain. Davidson s project is a reductive one: to give an account of the meaning of the sentences of the target in terms of something that does not presuppose their meanings. An approach to a theory of meaning using (M) is an intensional theory of meaning; Davidson wants to give us an extensional theory instead. Theory of Meaning for L F AXIOMS THEOREMS OBJECT LANGUAGE (foreign language) S F Kmj Kjm Lmj Ljm m refers to Mary j refers to John VOCABULARY L is satisfied by x loves y K is satisfied by x kissed y... COMBINATION RULES Lxy is true IFF x denotes a person, y denotes a person, and x loves y. Kxy is true IFF x denotes a person, y denotes a person, and x kissed y. For any sentence P, ~P is true IFF P is not true. For any sentences P and Q, P & Q is true IFF P is true and Q is true.... METALANGUAGE (our language) S F is true IFF p Kmj is true IFF Mary kissed John Kjm is true IFF John kissed Mary Lmj is true IFF Mary loves John Ljm is true IFF John loves Mary Compositionality This is a condition that is motivated by certain facts about our linguistic competence, and about some aspects of the nature of language: 1. Infinity: The set of sentences in a natural language has an infinite number of members. A person has only finite capacities: there are only a finite number of things a given person can learn, remember and process. But a person can understand any one of the infinite members in the set. How is this possible? Davidson on Truth and Meaning Page 5 of 8

6 2. Novelty: We understand sentences that we have already come across. But we can also understand sentences that we have never heard or seen before. 3. Learnability: Natural language is learnable. But how can we learn it on the basis of a command of only a finite number of sentences we have so far encountered? According to Davidson, these are data that any adequate theory of meaning needs to accommodate and explain. Accommodating and explaining these data is part of the adequacy condition upon a theory of meaning. What, then, must be true of the nature language so that it allows for these facts? Davidson s answer is that language is compositional. In other words, a language complies with a principle we have seen before in the work of Frege, the principle of compositionality: The meaning of a sentence is determined by the meaning of its parts. Consider this very hum-drum sentence: Mary kissed John. We all know the strict literal meaning of this sentence. This sentence means that there was some action going on, namely, kissing, and that the one who did the kissing is a person who goes by the name of Mary, and the one who got kissed is a person who goes by the name of John. The meaning of this sentence depends upon the meanings of the words, Mary, John, and kissed. None of these words have a meaning individually, out of the context of appearing in a sentence. The reason is that word order is crucial ( Mary kissed John is one state of affairs, while John kissed Mary is quite another state of affairs), and words come in an order only within a sentence. However, it s undeniable that this sentence is composed of distinct expressions, each of which makes a distinct contribution to the total meaning of the sentence. The meaning of a sentence is composed out of the meanings of its parts. This is what it means to say that a language is compositional. If language is compositional, then a theory of meaning for that language must reflect that feature: A theory of meaning is compositional IFF a. the theory has only a finite number of axioms i. one axiom for each member of the vocabulary ii. one axiom for each rule for combining the vocabulary b. the theorems clearly display the semantic structure of each sentence (i.e., the display the way in which the meaning of the whole sentence depends upon the meanings of its parts). The finitude of the number of axioms, the finite vocabulary and rules of combination, along with the recursiveness of some of the rules, makes it possible for us to account for the three questions that motivated the adequacy condition of compositionality. If a language is made up of a finite vocabulary and a finite number of sentence forming rules, then we, with our finite capacities, can learn the language. And if some of these rules are recursive, then we can generate an infinite Davidson on Truth and Meaning Page 6 of 8

7 number of sentences using only finite ingredients. Roughly, a recursive rule (definition or procedure) is a rule that can be reapplied to its instances. For instance, the following is a recursive rule: Negation Rule P is a sentence. If P is a sentence, then ~P is a sentence. With this rule, we can plug ~P into P get ~~P, plug in ~~P and get ~~~P, and so on till your heart s content, Here s another example. Suppose we want to define a positive integer. We can define it recursively: Positive Integer Rule 1 is a positive integer. If n is a positive integer, then n + 1 is a positive integer. Each application of the rule gives us an output. This output can then be used as the input for the next application of the very same rule, which will then give us a fresh output. This fresh output can then be used as the input for the next application of the rule, and so on: 1 is a positive integer; thus 1 + 1, which is 2, is a positive integer. 2 is a positive integer; this 2 + 1, which is 3, is a positive integer. 3 is a positive integer; thus 3 + 1, which is 4, is a positive integer.... Recursive rules are stated in a finite way, but they can be reapplied infinitely. And we thus get infinity from finitude! Interpretive Adequacy So how do we go about constructing a T-Theory for a language. To answer this question, Davidson follows Quine in his theory of radical interpretation. And this addresses a major concern that arises with the use of an extensional theory of meaning. Instances of (T) run into the problem of rogue theorems: R1. Snow is white is true-in-l IFF grass is green. R2. Snow is blue is true-in-l IFF grass is red. Material bi-conditionals are true just in case both components are true, or both are false. How are we to get at the T-sentences of L that give us their correct interpretation? Davidson argues that we solve this problem by using the Quinean technique of radical interpretation, in strict conformity to the Principle of Charity: Principle of Charity: Interpret the sentences of L as mostly true and rationally coherent. Davidson on Truth and Meaning Page 7 of 8

8 This is not just a convenient methodological injunction, but rather a metaphysical precept underlying the very subject-matter under study. If it is languages that we are studying, then we have to see it as the product or partner of cognition, and to the extent that it is cognition we are talking about, it is a rational phenomenon we are necessarily dealing with. If the T-sentences we come up with render our subject completely irrational or delusional, we cannot make sense of the linguistic output of our target speakers as being speakers of a language. To the extent that we understand someone as having false beliefs or irrational inferences, it is thanks to the largely rational background we have already attributed to the target. As Quine has insisted that scientific confirmation is holistic, Davidson also maintains that the construction of one s T-theory for L is a holistic enterprise. This means that how we interpret a basic lexical entry e for our target is influenced by the network of lexical entries in which e is located. Davidson on Truth and Meaning Page 8 of 8