A central goal of my research is to use both production and comprehension paradigms not only to extend work in each domain, but to unify theoretical and conceptual work in both areas. This work is done within a constraint-based framework, which is used to investigate how multiple sources of information are integrated during processing, with specific attention to the role of distributional constraints. For the most part, recent work falls into one of the following categories:
I. Agreement Processes in Production and Comprehension
One recent project has focused on agreement phenomena in English. The majority of the world's languages exhibit agreement phenomena, meaning that the forms of two or more words in a sentence match in some way. For example, in English, many verbs agree with their subjects. Because agreement is such a fundamental aspect of language, it has played a central role in recent production models. The dominant view of production is that agreement is a purely grammatical phenomenon, and early agreement results supported this view by demonstrating that semantic and phonological variables had little effect on agreement error rates. Some recent studies, however, have suggested that semantic variables may play a role in agreement production, but those semantic effects have been equivocal. We have conducted a series of related experiments that demonstrate robust semantic effects on agreement production, which run contrary to the standard account. Moreover, our data suggest that both the production and comprehension of agreement are governed by similar, constraint-based mechanisms. Another aspect of this work has used corpus data to explain a puzzling asymmetry in agreement error data by appealing to a distributional asymmetry in the input. This work is part of a larger constraint-based framework of language production that is central to my research program.II. Cross-Linguistic Variation in Sentence Processing
A second question that my research has addressed is cross-linguistic variation in sentence processing. Until recently, most comprehension research has been conducted using only English materials. This raises important questions about the relevance of extant English-based results for other languages. In other words, do current psycholinguistic findings inform us about general cognitive principles, or do they simply reflect language-specific properties? My work has explored this question by examining differences and similarities in processing syntactic ambiguities in Spanish as well as English. Another aspect of this cross-linguistic approach, which is just under way, investigates the universality of central theoretical concepts from the production literature by examining production phenomena in Japanese, which is a very typologically different language than English.III. Length and Distributional Constraints on Processing
Another research question investigates the effects of phrase length and distributional constraints on sentence comprehension. Hypotheses regarding distributional constraints are currently quite controversial and represent a major claim about how knowledge is represented. Examining these constraints is also particularly interesting because of increasing evidence that infants are sensitive to such information as early as can be tested. One source of distributional constraint involves phrase length. Several recent papers have suggested that length plays a central role in comprehension. Some of my empirical work has tested these length-based claims and has related the results to production work on length. This work supports a distributional account of processing, in which independently motivated principles of production are used to explain comprehension data.IV. A Connectionist Model of Sentence Processing
Mike Harm, Maryellen MacDonald and I have been developing a connectionist model that embodies a distributional perspective to examine the timecourse and integration of lexical, pragmatic, and distributional factors in the processing of locally ambiguous phrases, such as "the desert trains", in which "trains" can be a noun ("the desert trains were late") or a verb ("the desert trains the soldiers to be tough").V. Prosody in Language Processing and Acquisition
This is work in an early stage of progress, done with Suzanne Curtin, Toby Mintz, and Maryellen MacDonald. We are interested in how children, as well as adults, use prosody to resolve syntactic ambiguities. The questions this work intends to address are (a) do speakers reliably produce prosodic cues to disambiguate material and (b) are children able to use prosodic cues interpret temporarily ambiguous input.In sum, my work to date has focused on a specific domain of cognitive functioning in humans: language processing. Not only is this research intrinsically interesting in that it provides important insights into one crucial facet of human behavior, but it holds additional merit to the extent that findings contribute to our understanding of human psychology in general. As psychology is becoming an increasingly diverse science, I believe that it is important to take an interdisciplinary approach to research. To this end, I attempt to use a wide range of concepts and methodologies from psychology, linguistics, computer science, and neuroscience to study language processing and I am enthusiastic about making stronger links with these other disciplines.
R. Thornton--Last modified Thu Jan 23 13:02:40 PST 2003