DICTIONARY ENTRIES

If dictionary entries are to be the objects of any formal calculation (by some apparatus such as the ‘projection rules’), they must be given in a carefully controlled format.¹ KF proposes the following normal form: every entry contains (i) a syntactic categorization, (ii) a semantic description, and (iii) a statement of restrictions on its occurrences. The syntactic categorization (i) consists of a sequence of one or more ‘syntactic markers’ such as ‘Noun’, ‘Noun Concrete’, ‘Verb-^ Verb Transitive’, etc. The semantic description (ii) consists of a sequence of semantic markers and, in some cases, a semantic distinguisher. Semantic markers contain those elements of the meaning of an entry for which the theory is accountable. The semantic markers constitute those elements of a meaning upon which the projection rules act to reduce ambiguity; they are, accordingly, the elements in terms of which the anomalous, self-contradictory, or tautologous nature of an expression is represented. Polysemy of an entry appears in the normal form as a branching in the path of semantic markers (SmM), e.g.:

Correspondingly, reduction of ambiguity is represented as the selection of a particular path (e.g. SmM₁ → SmM₂ → SmM₄) out of a set of alternatives. The distinguisher contains all the remaining aspects of the meaning of an entry - those, in effect, which do not figure in the calculation of ambiguity reduction. The selection restriction (iii) at the end of an entry (or, in the case of polysemous entries, at the end of each of its alternative paths) specifies the context in which the entry may legitimately appear. The context of an entry W is described in terms of syntactic and semantic markers, either positively (i.e. markers which must appear in the paths of entries in the context of W) or negatively (i.e. markers which may not appear in the paths of context entries). But the selection restriction does not, of course, refer to distinguishers, since these, by definition, play no role in the distributional potential of the word.

Somewhere in the generative process, the words of a sentence would also have to have their phonological form specified. The omission of such a step in KF is presumably due to reliance on an earlier conception of linguistic theory as a whole which did not anticipate a semantic component and in which the grammar included, as a subcomponent, a lexicon that stated the phonological form and the syntactic cate¬ gory of each word. In an integrated theory, the existence of a lexicon separate from the dictionary is a vestigial absurdity, but one which can be removed without difficulty.² We therefore pass over this point and take up the KF conception of normal dictionary entries in detail.

¹ On canonical forms of lexicographic definition, see Weinreich (1962: 31 ff).

² Katz and Postal (1964: 161) postulate a ‘lexicon’ (distinct from the dictionary!) which presumably specifies the phonological form of morphemes. Chomsky (1965) has the underlying phonological shape of morphemes specified by the same component - the lexicon - as the syntactic features.