Amy Weinberg - US grants

ABSTRACT The project is an experimental investigation into the use of distinct types of information during sentence comprehension. Pilot work supports the view that syntactic processing is "informationally encapsulated" in that certain types of information fail to influence or guide syntactic processing, even in contexts where such information may be useful. Such encapsulation allows for a computationally efficient parse (in linear time). Specifically there is a sharp contrast between the utilization of information such as category, subcategory, number, and case specification and information such as animacy and abstractness features. The investigators will use four experimental tasks: (i) lexical decision (Wright & Garrett 1984); (ii) Word-by-word syntactic decision (Stowe 1989); (iii) word-monitoring (Marslen- Wilson & Tyler 1980), and (iv) the probe task. Both the word-by- word syntactic decision task and the word-monitoring task involve problems which limit their usefulness in experiments attempting to shed light on the issue of encapsulation. The issue of encapsulation is related to the issue of computational efficiency because the domains in which concord relations are established varies from feature to feature. The probe task will be used to investigate the role of the c-command domain in syntactic processing. In general, the aim is to conduct experiments which pilot work indicates will support a particular view of human sentence comprehension, i.e. the process of human sentence comprehension distinguishes the use of lexically-specified and general syntactic information from the use of semantic or pragmatic information, even when syntactic information is ambiguous.

9422138 Dorr This award is to purchase equipment (cost-shared with The Institute for Advanced Computer Studies (UMIACS) and Department of Linguistics at the University of Maryland) dedicated to support specific projects in the laboratory for Computational Linguistics and Information Processing (CLIP). These projects are in the areas of information mediation, language translation and tutoring, and text filtering and retrieval. The software requirements include large dictionaries, corpora, data models and databases, high-level database query languages, and a Prolog interpreter. Since many of these resources are available only on CD ROMs, the hardware requirements include an optical disk controller and large disks for storage. Also there is a need for two workstations with significant computing power for indexing and processing of text and installation of an object server. Finally, four Xterminals for extensive prototype development and an eye-tracker for testing hypotheses related to the development of psycholinguistically-grounded NLP models are needed. ***

EIA-0130422
Louiqa Raschid
University of Maryland College Park

CISE Research Resources: Infrastructure to Develop a Large Scale Experiment Testbed of Multi-model Resources

The use of the widely distributed collections of structured and unstructured information expressed in multiple languages or modalities provided by the Internet, requires production of scalable, robust algorithms for the discovery of replicated content, determination of delay or access latency of sources, and the confrontation of the inherently dynamic nature of the Internet.

This project's objective is to establish a laboratory testbed providing a controlled environment that captures structural, content, and latency characteristics of the (publicly accessible) Web. This will stimulate collaboration between researchers whose interests range over natural language applications, language independent processing of scanned documents, analysis of video information sources, information retrieval, and wide area applications and resource discovery across heterogeneous servers.

The testbed will support the development and testing of: (1) tools for broad-scale, cross-linguistic analysis and discovery of relevant information across languages and modalities, (2) cost models and access cost catalogs for wide area environments, reflecting the temporal variability in access latency, (3) distributed content based indexing and association of media clips for resource discovery, (4) transcoding and scheduling of multimedia resources for delivery any time and anywhere to disparate clients; from mobile wireless to high speed optical links.

The National Science Foundation will support a workshop on The Relation between Parsing and Production, to be held at the University of Maryland, College Park, as a part of the 17th Annual CUNY Conference on Human Sentence Processing, March 25-27, 2004. The workshop will bring together leading researchers on language production and language understanding from psychological, computational and neuropsychological perspectives, with the goal of generating a new synthesis of the parallels and differences between production and comprehension. Until recently, research on speaking and understanding investigated largely different phenomena, using largely different methodologies; this led to the view that sentence comprehension and sentence production are largely independent cognitive systems. Recent advances in both fields now suggest a different picture: parallel experimental and computational investigations of speaking and understanding indicate that there is substantial overlap in the mechanisms used for both tasks. The specific questions addressed at the workshop will include the following: What do sentence comprehension and production have in common, beyond the fact that they use the same language? Do comprehension and production operate on the same time-scale? How can experimental findings of close parallels between speaking and understanding be reconciled with the finding of dissociations between speaking and understanding in language disorders? What methodologies will allow closer comparison of comprehension and production?

The workshop will have broad impacts for both research and education. A proper understanding of the relation between language production and language comprehension is important for a number of different fields, including language learning and especially the study of language disorders, where language comprehension and language production have often been found to be differentially impacted. An improved understanding of the relation between comprehension and production also has implications for the development of computational models of interactive natural language agents. The workshop will foster interaction among researchers in different fields, and by supporting the participation of students, it will enhance the training of a new generation of researchers. Results of the workshop will be made available through publication in a special journal issue.

Human language is both universal within the species and highly variable across populations. This Integrative Graduate Education and Research Traineeship (IGERT) project will train young scientists and engineers to understand language diversity by combining the tools of behavioral, computational and biological research, drawing upon an extensive collaborative network that spans nine departments in five colleges at the University of Maryland. The project aims to promote sustainable change in the science of language by building infrastructure for interdisciplinary research on diverse languages through local and international collaborations and outreach efforts, by strengthening links between basic science and clinical and engineering applications, and by building awareness of the science of language through high school and undergraduate partnerships. The training plan provides coursework, research training, and an environment geared towards preparing students for interdisciplinary research and equipping them to build collaborative networks in their future careers. Preparation for interdisciplinary research will be provided by broad coursework, integrative pro-seminars and a post-candidacy lab rotation that will pair trainees with students from other disciplines. A central component of the project is the Winter Storm, an intensive two-week workshop that will provide foundational skills training, research planning, and professional development. The project will enhance the use of computational and neuroscientific techniques in studies of atypical language and second language learning, and will partner with an NSF-supported Science of Learning Center based at Gallaudet University that focuses on visual language. IGERT is an NSF-wide program intended to meet the challenges of educating U.S. Ph.D. scientists and engineers with the interdisciplinary background, deep knowledge in a chosen discipline, and the technical, professional, and personal skills needed for the career demands of the future. The program is intended to catalyze a cultural change in graduate education by establishing innovative new models for graduate education and training in a fertile environment for collaborative research that transcends traditional disciplinary boundaries.