Page 194 - Encyclopedia of Philosophy of Language

Truth and Meaning
various tenses), and simple syntactic rules (e.g., noun and verb must agree in number). Rules of this kind are explicitly stated, used in language teaching, applied as standards of correct linguistic usage, and, usually in contradistinction to much else included in grammars, called rules.
What should properly be called a linguistic rule, however, is another matter. To the extent that the required generality of a rule concerns its relation to behavior, the statement that killed is the past tense form of the verb kill is a statement of a rule, since it is general with respect to agents and highly general with respect to particular speech acts governed by it. Seen in this light the general rule about past tense forms of regular verbs appears to be a more factual or technical rule, a general guide or recipe for speaking in accordance with the more particular rules (gov- erning individual verbs) that may be regarded as nor- mative. In this way normativity, or the degree thereof, may be inversely related to the degree of generality.
Stating that in English gold is a noun, or a mass noun, is not generally considered as stating a rule, but rather as stating a fact. To the extent that the property of being a mass noun in English is a conventional one, however, this is a fact about a convention, and the statement may then also be regarded as a statement of the convention, or rule, itself. Together with, say, the rule that mass nouns do not take the indefinite article, the rule (about gold) so stated also has a share in syntactic standards of linguistic behavior. Indeed, from a formal point of view it may be regarded as a rule of higher order, implicitly laying down what other (lower order) rules apply to gold (namely the rules governing mass nouns).
On the other hand, however, the statement that gold is a mass noun may also be regarded in a number of other ways. The situation is quite unlike that with respect to formal languages. The class of sentences of a language for predicate logic is determined by a few simple clauses, stating on the one hand the basic vocabulary of signs of various categories, and on the other hand the formation rules, which comprise rules for forming atomic sentences (by way of joining terms and predicate letters) and for forming complex sen- tences out of these and the logical symbols (forexam- ple, if A and B are sentences, then A&B is a sentence). These rules are easily stated, and learned, and define the language in question.
2.1 Syntactic Rules in Generative Grammar
Specifying the class of sentences of a natural language, like English, on the other hand, is the task of modern generative grammars. The syntactic part of such a grammar may consist of a set of phrase structure rules, a set of lexical insertion rules, and a set of transformation rules. Rules of the first kind produce so-called deep structures of sentences (the most basic being S -* NP VP, which, roughly stated, produces
the category structure NOUNPHRASE-VERBPHRASE out of the category SENTENCE). Rules of the second kind provide for inserting linguistic expressions (lexical items) into structures, at appropriate places, depend- ing on their respective categories (such as NOUN). Rules of the third kind are rules for transforming results of applying rules of the former kinds by way of operations such as reordering and deletion (as with the rule of equi NP deletion, for removing a repeated occurrence of a particular noun phrase).
Such systems are readily understood as systems of rules for producing sentences. They are stated and can be followed. It is another question in what sense, if any, they are rules of a particular natural language. On the one hand the grammar may be incorrect in the sense of producing sentences which are not recognized as well-formed by speakers of the language. Even if correct its rules are clearly not, in any strict sense, followed by the speakers. Neither is it generally clai- med by linguists that such rules are subconsciously operative in actual practice. It is, on the other hand, claimed, for example, by Chomsky (1976, 1980), that a grammar which is adequate in a stronger sense rep- resents the linguistic competence of the speakers, their knowledge of the language. If this claim is good there is a sense in which the rules of such a grammar are rules of the language, but in that sense they can hardly be said to define it.
2.2 Semantic Rules
With respect to formal languages semantic rules, con- cerning meaning, can be stated and regarded as just as normative/defining as formation rules. Carnap (1956) distinguishes between rules of designation for simple expressions (predicates—'H' designates Human—and individual constants—'s' designates Walter Scott) and rules of truth (e.g., A&B is true if and only if A is true and B is true). In addition to rules of this kind he also proposes so-called meaning postulates, such as a rule to the effect that the formal counterpart of Bachelors are not married is true, in order to capture nonlogical conceptual (analytical) truths. In Montague semantics (Montague and Thomason 1974) there are meaning postulates as well as semantic rules corresponding to Carnap's rules of truth (and also for other operators and functions of higher types), but they belong to a grammar for (a fragment of) natural English. Thus, they are presented as in some sense being rules of English.
So-called truth theories of natural languages, according to Davidson's conception (1984), are to contain statements of virtually the same kind as Car- nap's statements of rules of designation and rules of truth. However, to the extent that they are rule state- ments the rules are just rules of the theory; the claim that they govern the practice of the speakers is not part of the theory and it is also rejected by Davidson. In the conception of Lewis, on the other hand, the
172