ABSTRACT
According to the modularity hypothesis, different components of the language apparatus obey different operating principles (cf. Crain & Shankweiler, 1991; Crain & Steedman, 1985; Fodor, 1983; Shankweiler & Crain, 1986). In this chapter, we argue that two subcomponents of the language apparatus are autonomous because of their different operating characteristics. One is the component responsible for language acquisition (i.e., the language acquisition device [LAD]). The other is the component used in resolving ambiguities (i.e., the sentence-parsing mechanism). We identify instances in which operating characteristics of these two components conflict due to the different demands placed on them. The principles of the LAD must be responsive to demands of learnability. To achieve learnability, the LAD must constrain learners’ hypotheses to guarantee that linguistic representations that are not derived in the target language will not be formulated or, if formulated, can be disconfirmed by readily available evidence. By contrast, the sentence-parsing mechanism selects among the competing linguistic representations that are derived in a language. Selection is based on considerations of simplicity. It turns out that certain initial representations that are favored on learnability grounds are ones that are dispreferred by the sentence-parsing mechanism. Putting it the other way around, certain linguistic representations that are preferred by the sentence-parsing mechanism are ones that would create problems of learnability if they were initially adopted by learners. The upshot of these deliberations is that the principles of parsing must not guide learners in their formulation of grammatical hypotheses. Therefore, the principles of parsing and the principles of learning must be kept in distinct modules of the language faculty.
