Formatting Lexical Entries : Interface Optionality and Zero

Transformationalists (including minimalists) currently assume that (i) particular grammars reduce to the store of closed class lexical items, and (ii) syntactic structures project directly from lexical items. But they seem satisfied with common sense specifications of "possible lexical items." This study moves beyond this vague pre-theoretical stage, focusing on how best to lexically notate optimality and null realizations at both the PF and LF interfaces. It argues that both the symbol 0 and the parenthesis notation express linguistically significant generalizations in each of the phonological, syntactic, and contextual parts of lexical entries. As discussed here, their proper definitions allow us to construct simple and in principle easily learned lexical entries which fully explain many alternations between null and non-null PF allomorphs and between distinct yet partly similar interpretations of other grammatical morphemes such as Englishof, to, there, from, it, -ing, -en and Japanese –(r)are . 1. Some issues in lexical formalisms One can imagine a grammatical model for natural language in which a lexicon plays little or no formal role. Chomsky (1957) made a revolution in linguistics by presenting such a system. But current models of natural language, for example that of Minimalism in Chomsky (1995) and succeeding works, now crucially depend on the form and content of lexical entries, basing themselves on concepts such as lexical arrays and the strong features of various grammatical items. In fact, the dependence of recent transformational models on the lexicon is almost total. Their advocates currently typically assume that (i) particular grammars reduce to nothing more than the store of closed class lexical items in a language, and (ii) all syntactic structures project directly from lexical items. These claims sound very restrictive and scientific— until one realizes that most of these syntacticians are operating with essentially no theory about the form of closed class lexical items.1 There is not even any agreed on model of mechanisms for the more widely studied open class lexicon; lexical semantics is usually Theoretical nd Applied Linguistics atKobe Shoin 5, 1-22, 2002. © Kobe Shoin Institute for Linguistic Sciences. *I thank Ludmila Veselovska for carefully reading and commenting on a draft of this paper . 1 This state of affairs has a rather dismaying implication: transformational generative grammar studies tructures that are fundamentally intuitive and unformalized. In studies panning more than 40 years, one could hardly find for all languages taken together a dozen closed class items whose lexical entries have been analyzed in terms of