Chomsky is certainly free to prefer using“syntax”in the broader sense. This is apparently the sense intended when he
characterizes the syntactic levels PF (phonetic form) and LF (logical form) as“direct representations of sound on the
one hand and meaning on the other”(Chomsky 1986: 68). Phonologists, phoneticians, and semanticists might be
justifiably excused if they have been baffled by this characterization, given that the actual formal theory invoked in all
Chomsky's writing since 1970 has been that of syntax in the narrow sense. The statements quoted above help clarify
Chomsky's intention; they do not, however, clarify his practice.
In particular, consider Chomsky's example of referentialdependency. He is of course correct that the relationbetween
a pronoun and its antecedent is“in the head,”and therefore syntactic in the broad sense. But since the 1960s the
literature has been full of discussion about whether this relation is directly encoded in narrow syntax or whether it is
encoded partlyor evenprimarilyinsemantic/conceptual structure(Lees and Klima 1963; Ross 1967; Jackendoff 1972.
; Chomsky 1981; Lasnik 1989; Kuno 1987; Van Valin1994; Van Hoek 1995; Culicoverand Jackendoff1995; Levinson
2000, to mention only a few of hundreds of references). To just say,“Well, I prefer to call it all syntax,”without
distinguishing broad fro mnarrow syntax, is the rhetorical counterpart of the move by the opponents of generative
grammar,“It's all semantics.”The effect is to make it impossible to clearly articulate the issues at stake.
Let us look also at what Chomsky wishes to call“semantics.”As far as I can understand this characterization, he is
gropingfor thenotionofan interfacecomponent:“semantics”is a relationbetweenonekind ofstructureand another,
just what interface components are designed to instantiate. But it is unclear what interfaces he has in mind. In one
passage above, he speaks of the connection between language and conceptual systems and then, in the next breath, of
the connection between language and his computer; in another passage, he speaks of the relation between“language
(or concepts)”and the“outer and inner world.”Is semantics supposed to be the connection of language to concepts,
or thatof conceptstosomething else?And nowheredoes headdress whatI take tobethecentral question, theformal
organizationoftheconceptualsystem. Heis, however, correctinbeingdeeplyskepticaloftheordinarynon-mentalistic
notion of“thing,”an issue to which we return at length in the next chapter.
Jerry Fodor (e.g. 1975; 1983; 1990; 1998) has spent much of his career addressing some of the problems faced by a
serious theory of meaning. His approach concurs with conceptualist semantics in insisting on the importance of
placing this theory in a mentalistic framework. He argues that meanings must