When adults encounter a novel syntactic pattern in their native language, they will initially have difficulty processing the "deviant" sentence. But, what happens when adults repeatedly encounter the novel pattern? Do they learn to process the novel sentence pattern with ease, or do they continue to struggle? And, if they learn to process the new pattern with ease, what (if any) consequences does this have for their understanding of other kinds of sentences in their language? I address these questions in a series of experiments that introduce adult English speakers to a novel syntactic pattern: the Needs construction (e.g., "The floor needs cleaned."). In three experiments, I show that adults can quickly learn a new syntactic pattern in their native language, and that experience with the novel pattern can have a strong influence on the way that readers process more familiar English sentence structures. The findings are discussed with regard to their implications for theories of sentence processing, and with regard to their possible import for the development of theories of linguistic change.

Is the processing of discourse a matter of satisfying abstract and arbitrary constraints, or is it grounded in recognized cognitive and communicative principles? In my talk I will look at one aspect of discourse, the processing of referential expressions, and argue that this processing reflects constraints on the activation and processing of semantic information in working memory. The theory that I propose, which is based on the linguistic theory of pragmatics, views discourse processing as an optimization process wherein processing cost, defined in terms of activating semantic information, should serve some discourse function-identifying referents and/or adding new information. My talk will review the computational principles underlying the theory and present results from a computational implementation of the theory in a restricted referential domain and two empirical studies, one with healthy young subjects and the other with Alzheimer's patients. The results of the two experimental studies corroborate the computational work in supporting the principle of cost and function balance that is advocated by the theory. I will finally discuss the implications of this work for how seemingly abstract linguistic principles could be grounded in psychological mechanisms.

In virtually every language, there are a set of processes that are clearly morphological, processes that are clearly phonological, and a residual set of 'morpho-phonological' processes that fall somewhere between the two extremes. For example, ablaut in English past tense formation, e.g., sing --> sang, is morpho-phonological (m-p); it is an essentially phonological process because it works on phonological structures, but its distribution is governed by morphology. The first aim of this talk is to argue that the inherent principles of Optimality Theory (OT, Prince & Smolensky 1993) provide a basis for classifying m-p processes and identifying their underlying functional motivation. A formal framework is developed and applied to three classes of m-p processes, 'non-structure preserving' m-p processes (processes whose outputs step outside of the legal phonological structures of a language), phonologically arbitrary neutralizations, and chain-shifting morpho-phonology, and it is shown that these phenomena must be morphological in OT. The second goal of the talk is to illustrate that the architecture of OT grammars can be naturally extended to explain the properties of these morphologically motivated processes. In particular, a theory of Transderivational Anti-Faithfulness (TAF) is proposed that accounts for the morphological function of m-p processes as a development of the theory of faithfulness constraints in OT (Alderete 2001ab). It is argued that TAF has both the descriptive power needed to account for m-p processes that have proven difficult to analyze in other theories, and significant restrictiveness, because the core ideas of TAF require m-p processes to have a cluster of properties that rule out certain logically possible process types.

References (available at http://roa.rutgers.edu/ )

Alderete, John. 2001a. Morphologically Governed Accent in Optimality Theory, Outstanding Dissertations in Linguistics. New York: Routledge. ROA-309.

Alderete, John. 2001b. Dominance effects as Transderivational Anti-Faithfulness. Phonology 18: 201-253. ROA-407.

Prince, Alan & Paul Smolensky. 1993. Optimality Theory: Constraint Interaction in Generative Grammar. RuCCS-TR-2, Piscataway, NJ: Rutgers Center for Cognitive Science. ROA-537. [Forthcoming MIT Press]