Crain, S., Ni, W., Shankweiler, D. P., Conway, L., & Braze, D. (1996). Meaning, Memory and Modularity. Proceedings of the NELS 26 Sentence Processing Workshop (pp. 27-44). MIT Working Papers in Linguistics 9. Cambridge, MA.

Given that meaning in language is conveyed by sequences of words occurring one at a time, questions of timing in studies of sentence processing are inescapable. The modularity hypothesis (Fodor, 1983) generates predictions regarding the relative timing of the availability and use of various sources of information in the resulution of structural ambiguities. A fundamental tenet of the modularity hypothesis is that plausibility considerations do not influence structure building operations of the parser. The operating characteristics of several parsing models are consistent with this assumption.In this paper, we test the implications of one such model, the referential theory (Crain & Steedman, 1985; Altmann & Steedman, 1988; Ni, Crain & Shankweiler 1996). The results of two eye-movement experiments (combined N=64) are reported. Two factors influenced how rapidly plausibility information can be accessed: memory capacity and the temporal availability of the information. Plausibility does not influence structure-building operations. Findings as a whole are consistent with the referential theory of sentence processing and the modularity hypothesis.