Lexical Selection, or How to Bridge the Major Rift in Language Processing

Abstract

The claim that a language's mapping of sound to meaning is (by and large) arbitrary needs no defense nowadays. But it would be false to conclude that the claim is a trivial one. It isn't. An explicit theory of language should give an account of this arbitrariness. Where in the system is it to be located? Is it limited to a single module, such as the lexicon, or is it distributed all over the system? Are there boundary conditions on arbitrariness? If so, are they universal? And additional questions should be raised by those (like me) who consider language to be a mental faculty: What kind of learning mechanism is involved in the acquisition of arbitrary sound/meaning relations? Is this arbitrariness reflected in language processing, in particular in the processes of speaking and of language comprehension? Is it, in particular, the case that processing is different where relations are arbitrary than where they are systematic or rule-governed? It is this latter question that will be the focus of the present paper, After some introductory remarks about what I will call the major rift in the system, the main locus of arbitrariness, I will discuss two central properties of lexical processing, activation and selection. I will then argue that the relation between activation and selection is a different one across this major rift than it is at either side of it. The claim will be supported by data on lexical access in speech production.