Joan W. Bresnan - US grants

ABSTRACT This project is a comparative study of the ways in which the syntax of subjects and objects interacts with semantic roles and discourse functions in several Bantu languages. The objective is to test and further develop the theory that asymmetries in the syntactic behavior of verbal arguments universally arise from underlying semantic and pragmatic hierarchies which are formally associated with underspecified syntactic functions, following a line of research initiated in Bresnan and Kanerva (1989). This theory combines aspects of both the functionalist and the formalist linguistic traditions in explaining argument asymmetries in Bantu and other languages. If it is correct, the theory has significant consequences for understanding the relations of discourse, syntax, and morphology and implies that the sharp division of labor that current linguistic practice continues to impose between "sentence grammar" and "discourse grammar" has led to uninsightful or incomplete accounts of important linguistic phenomena. In research over three years the project team will systematically analyze subject and object properties in eight Bantu languages that vary in the extent of asymmetry among verbal arguments. The phenomena analyzed will include locative inversion, unaccusativity, and presentational focus; interactions of argument-adding and argument-suppressing processes; agreement, word order, and extractions; and topic and focus marking, including phonological marking.

9818077
BRESNAN
The importance of markedness hierarchies in morphosyntax has been well-documented in functional and typological work. Hierarchies of person, animacy, definiteness, grammatical function, semantic role, etc. underlie grammatical asymmetries which are found both within particular languages, and in the cross-linguistic distribution of various grammatical phenomena. Although universal tendencies in the form of these hierarchies are discernible, this domain is characterized by a great deal of language-particular variation. A central problem then is how to integrate markedness hierarchies into morphosyntactic theory in a way that acknowledges what is universal and at the same time permits for a wide range of language-particular variation.

The variability of hierarchies -- both across and even within languages -- has impeded recognition in formal generative syntax of their explanatory importance as universals. While hierarchies have played a central role in functional and topological syntax, they have remained on the margins of formal description and theory. The emergence of Optimality Theory (OT), however, has introduced new ways of thinking about the relation between the universal and the language-specific, as well as a formal theory of markedness. In OT, all constraints are conceived as universal but violable, with language-particular difference the consequence of different rankings of universal constraints. These developments provide the opportunity to reconsider the role of markedness hierarchies in syntax and to build a bridge between the insights and results of functional/typological work on the one hand, and the explicit modeling techniques of formal grammar on the other. The goal of this collaborative project is to develop a fully explicit OT approach to markedness hierarchies in syntax, and to test it against both crosslinguistic typological research and language-internal studies of syntactic structures.

This project will study crosslinguistic and language-specific asymmetries along several dimensions of prominence, including person, animacy, definiteness, and obviation. These asymmetries manifest themselves in various subsystems of grammar. To be studied here are voice, direction (i.e. inverse systems), case marking, agreement, and pronominal inventories, focusing on the Mayan and Bantu languages for in-depth linguistic study. A comparative survey of Australian languages will also be undertaken in the subsystems of interest. In addition, the role of markedness hierarchies in pidginization will be examined. This broad empirical scope is demanded by the theoretical approach (for which a major empirical test is the generation of typologies by reranking), and made possible by the combined expertise of the principal investigators and their assistants.

The use of language is central to human social dynamics. Current techniques for modeling human language are widely based on simplifying assumptions that underestimate human language capacities by ignoring statistical patterns. The present project starts from a strongly contrasting approach which can lead to a breakthrough in the understanding of linguistic dynamics by linking changes in language across the different time scales of speech, the human life cycle, and the history of communities. This approach models grammar as a highly plastic cognitive system sensitively tuned to the probabilities of the environment. The project combines detailed statistical analyses of patterns found in recorded speech with laboratory experiments on language production in adults and language learning in children. The phenomena considered include high-level grammatical choices (such as whether to say "give someone a job" or "give a job to someone"), low-level pronunciation choices (such as where "to" is reduced to "tuh"), and overregularizations (such as when children say "goed" instead of "went"). Studies will involve work with data from multiple dialects of English, as well as German and Chinese. The project involves cross-disciplinary collaboration among researchers trained in Linguistics, Psychology, and Computer Science. This research will lay the groundwork for improved language technologies, and may find other applications, for example in the treatment of language disorders.

Mastering spoken and written discourse is vitally important in human cognitive development, and it is increasingly recognized that this mastery depends on the implicit knowledge of "linguistic probabilities"--the likelihoods of types of expressions occurring in ordinary language use. For example, psychologists have shown that while reading or hearing a sentence, people instantaneously anticipate the more likely of two types of continuations, even when they are semantically equivalent. For another example, researchers of the historical and social dimensions of language have demonstrated that speakers of different regional varieties of the same language (such as British and American English) are characterized by different models of their speech probabilities, and that these probabilities have been changing over historical time. Finally, developmental studies have revealed children's sensitivity to the statistical regularities of their linguistic environment and their mastery of the variable higher-level linguistic structures that occur in spontaneous discourse. Combing these psychological, historical/social, and developmental perspectives, the present project will investigate how the implicit knowledge of linguistic probabilities develops in the individual and in historically diverging groups of speakers. The project will make use of a common theoretical framework for studying linguistic probabilities as well as a common set of semantically equivalent types of linguistic expressions which differ in their linguistic probabilities. The latter, called "syntactic alternations", are alternative ways of paraphrasing the same message (such as "give her a book/give a book to her" or "the woman's shadow/the shadow of the woman"). The project will enlist an international team of experts to conduct on-site field research of the same syntactic alternations in three suites of studies: (i) parallel observational studies and experiments across groups of speakers of closely related dialects of English to investigate the varying probabilities of syntactic alternations in speaking or writing and their effect on understanding, (ii) studies of how probabilistic models of higher-level linguistic structures from the spontaneous spoken language of children and their caretakers develop over the time-course of primary language learning, and (iii) studies of how the predictors of probabilistic changes in the same high-level linguistic structures develop in historical time.

The project has unusual intellectual scope, because it applies multiple methods of expert collaborators that are seldom brought together within an integrated theoretical approach. New datasets built for this project will be made publicly available to all researchers. This work also has potential applications in reading, second language education and language impairment. Working as a multidisciplinary team on this central set of fundamental linked problems in the comprehension, production, and development of spoken and written discourse will more rapidly advance the growing convergence of probabilistic approaches to language in all of the language sciences, including computer science, communication engineering, psychology, and linguistics.