January
19 Richard Aslin
26 Leonard Talmy
27 Philosophy Seminar

February
2 Charles Duffy
3 Computer Science Seminar
9 Business Meeting
16 Donald Pollock
17 BLC
23 Jennifer Stolz
24 BLC

March
1 Student Poster Session
8 Spring Break
15 CogSci Symposium
22 Paul Luce
23 BLC
29 Mark Turner
29 Poetic Talk
30 BLC

April
4 Stephen Palmer
5 Stephen Palmer
12 Nicholas Cercone
19 Peter Jusczyk
26 David Eddins

Regular colloquia are Wednesdays, 2:00-4:00 p.m., 280 Park Hall, North (Amherst) Campus, and are open to the public. Refreshments are served.

RICHARD ASLIN

Department of Brain and Cognitive Sciences
Center for Visual Science

"Statistical Learning in Linguistic
and Non-linguistic Domains"

Statistical approaches to language learning have generally been understudied because distributional information was thought to be inconsistent in the child's input and because learners were thought to be incapable of extracting many key consistencies that are present. A series of studies of statistical learning in the domain of word-segmentation from fluent speech will be reviewed. These studies demonstrate that adults, children, and 8-month-old infants are exceptionally adept at extracting some forms of distributional information. The statistical learning mechanisms that enable some forms of on-line distributional analysis are domain-general, as evidenced by similar learning of tone-sequences, and species-general, as evidenced by similar performance in Tamarin monkeys. The constraints on statistical learning have implications for the evolution of natural languages.

Back to Top

CHARLES J. DUFFY

Department of Neurology
Center for Visual Science
University of Rochester

"Neuronal and Perceptual Mechanisms
of Spatial Orientation"

Single-cell recordings in monkey cerebral cortex have shown that MST neurons respond selectively to optic flow, the patterned visual motion that is seen during observer self-movement. These optic flow responses reflect the direction of simulated self-movement and the three-dimensional structure of the environment. MST integrates these responses with vestibular signals about self-movement and spatial orientation cues that can be derived from moving objects.

The importance of visual mechanisms for spatial orientation is revealed by their impairment in the spatial disorientation of Alzheimer's disease. Patients with this syndrome show a selectively elevated perceptual threshold for the patterned visual motion of optic flow. This perceptual deficit may be linked to lesions in posterior association cortex that impair spatial navigation. Together, these findings suggest that extrastriate visual areas process optic flow and other self-movement cues to support spatial orientation. The failure of these mechanisms may result in debilitating spatial disorientation.

Back to Top

JENNIFER STOLZ

Department of Psychology
University of Waterloo

"On the Joint Effects of Attention and Word Recognition: The Relations between Resources and Meaning"

The study of attention and that of visual word recognition have both resulted in large literatures. Interestingly, despite the fact that common sense dictates that attention is involved in word recognition, there is very little work at the intersection of these two literatures. The present work addresses this intersection by examining the joint effects of attention, viewed as a resource, and a key variable important in word recognition, semantics. Two central questions are pursued. First, is attention necessary for semantics to be activated? This question is asked by examining the semantic priming effect under dual task conditions. Second, does previewing a word's meaning result in fewer resources being required for the word's subsequent recognition? This question is addressed by investigating the effects of priming a word presented in the context of an attention-demanding tone discrimination task. The results reveal a rich pattern in which resource attention affects, and is affected by, the activation and maintenance of meaning during word recognition.

Back to Top

Paul Luce

Language Perception Laboratory Department of Psychology
University at Buffalo

"Probabilistic Phonotactics, Neighborhood Activation,
and Spoken Word Recognition:
An Adaptive Resonance Perspective"

Recent work investigating the role of probabilistic phonotactics in spoken word recognition suggests the operation of two levels of representation, each having distinctly different consequences for processing. The lexical level is marked by competitive effects associated with similarity neighborhood activation, whereas increased probabilities of segments and sequences of segments facilitate processing at the sublexical level. I will discuss a series of studies that provide support for the hypothesis that the processing of spoken stimuli is a function of both facilitative effects associated with increased phonotactic probabilities and competitive effects associated with the activation of similarity neighborhoods. I will also describe recent extensions of this work aimed at evaluating two hypotheses regarding the segmentation of words from fluent speech, one phonotactic (the trough hypothesis) and one lexical (the lexical burst hypothesis). Finally, I will describe our attempts to account for effects of neighborhood activation and probabilistic phonotactics from the perspective of Grossberg's adaptive resonance theory (Grossberg, Boardman, and Cohen, 1997).

Back to Top

MARK TURNER

Department of Linguistics
University of California, Stanford

"Conceptual Compressions and Decompressions"

Everyday conceptual life is based on integrating clashes and compressing vital relations. Life is various and often diffuse. Its many lines of connection run simultaneously over large expanses of time and space and involve complicated relations of change, cause and effect, intentionality, identity, analogy, and representation. To form a conceptual apparatus and use it requires constant and frequent compressions over these vital relations. Compression is so natural to us that when literature uses them to compress large worlds of life into a few pages, we hardly notice.

When we look at the Persian rug in the store and imagine how it would look in our house, we are compressing over two different physical spaces: the physical space with the rug and the physical space where we live. When we imagine how we would now answer a criticism directed at us several years ago, we are compressing over times. We compress over time when we tell someone our life story in three minutes. We compress over space when we draw the Empire State building on the back of an envelope.

Conceptual blending is an unrivaled tool of compression.

Back to Top

NICHOLAS CERCONE

Department of Computer Science
University of Waterloo

"Natural Language Access to Relevant
Information on the Internet"

Information available on the World Wide Web (WWW) has grown enormously, thus rendering difficult the retrieval of relevant information. To be an unqualified success on the internet, data mining, electronic commerce, etc. will depend, in part, on the successful retrieval of relevant information from the Internet. Several tools have been developed for browsing and searching these collections of highly unstructured and heterogeneous data. These tools organize web pages into listings and allow users to search these listings to find required information. Each of these tools has its own listing or catalogue for searching. Based on how the new Uniform Resource Locators (URLs) are added to the catalogue, they can be classified as a directory or as a spider or robot. We generally refer to both types as search engines. We propose to use natural language (English) to access information on the WWW and illustrate this process with two interesting prototype systems. NLAISE is our initial prototype implementation of natural language access to internet search engines. EMATISE is our second prototype implementation, English Meta Access to Internet Search Engines. Both systems employ Head Driven Phrase Structure Grammar (HPSG) implementations for reasons explained in the talk. Initial experiments with these prototypes will be presented and discussed.

Back to Top

PETER W. JUSCZYK

Department of Psychology
Johns Hopkins University

"Infants' Use of Multiple Cues to Segment
Words from Fluent Speech"

Several years ago, Jusczyk and Aslin (1995) reported that infants first display some abilities to segment words from fluent speech at around 7 months of age. Many subsequent studies have focused on the nature of the information that infants rely on to find words in fluent speech. The kinds of cues investigated include prosodic, phonotactic, allophonic, and statistical cues to word boundaries. English-learners appear to develop sensitivity to some of these types of cues earlier than they do for others. I will review some of these findings, along with some more recent attempts to investigate the relative weighting that infants give to the different types of cues. Although much remains to be learned about how infants come to integrate these different types of cues, it is clear that word segmentation abilities evolve considerably in the second half of the first year.

Back to Top

Wednesday, April 26, 2000
280 Park Hall
2:00-3:30 p.m.
North Campus

DAVID EDDINS

Department of Communicative Disorders and Sciences
University of Buffalo

"A linear systems approach to the study
of sensory processing"

The basic principles underlying the perception of auditory temporal and spectral patterns are not well understood despite the fundamental nature of such patterns in auditory perception. This likely reflects the lack of a unifying framework for understanding the processing of acoustic features in either domain. Borrowing from principles of spatial vision, a single global process will be outlined by which the auditory system might process both temporal and spectral features of simple and complex acoustic stimuli. Support for such a process will be gathered from recent psychophysical and physiological studies and compared to relevant analogs in spatial vision.

Back to Top