Logical form: Types of evidence

Abstract

The notion of 'logical form' (LF), like other linguistic constructs, includ ing 'phonological segment' or 'noun', is an abstraction derived from the examination of primary linguistic data, and as such requires empirical justification. Such a level requires the same justification that is neces sary to establish other linguistic constructs: that we cannot do without it and maintain an insightful, productive, and correct theory of language. Logical form, however, represents a much more tenuous construct than, say, the notion of 'word', in that the arguments for its existence are few and far between.' In a linguistic system, LF may play two distinct roles. On the one hand, it may function as a level of semantic representation in the sense that such notions as entailment and logical equivalence may be directly defined on these structures. On the other hand, it also may represent a level of syntactic description, capturing generalizations which do not appear to be capturable elsewhere in the syntax. The first function faces what I feel are insurmountable difficulties, if we consider the alternative of doing the semantics model-theoretically via interpretations rather than forms. The second function faces the difficulty of deciding whether a given level of structure should in fact be labelled LF. What types of syntactic structures could, and which could not, qualify as a level of LF? These two functions of LF are considered in turn below. In the first