Paper presented at the Workshop on Distributional Methods in Language Modelling, School of Computer Science, University of Birmingham, U.K.

Informing Connectionist Psycholinguistics through Corpus Analyses


Morten H. Christiansen


Abstract

Connectionist psycholinguistics is an emerging approach to the modeling of human language processing using artificial neural networks (Christiansen & Chater, 2001b). Whereas early connectionist modeling demonstrated that neural networks could in principle capture important aspects of language, more recent work has moved towards capturing detailed psycholinguistic data (Christiansen & Chater, 2001a). In this talk, I outline three ways in which corpus analyses can be employed to inform connectionist psycholinguistics. I first discuss how corpus analyses can be used to improve the representativeness of the input provided to connectionist models, using recent simulations of syntactic acquisition via multiple-cue integration as an example (Christiansen & Dale, 2001). I then consider how corpus analyses can be employed as baseline models against which network performance can be compared, pointing to syntactic processing comparisons between N-gram models and neural networks as an example (Christiansen & Chater, 1999; Christiansen & Dale, 2001). Finally, I argue that corpus analyses can be used to further explore and substantiate network predictions, providing an example from the modeling of multiple-cue integration in infant speech segmentation (Christiansen et al., 1998) along with corroborating infant data (Curtin et al., in preparation).


References

Christiansen, M.H., Allen, J. & Seidenberg, M.S. (1998). Learning to segment speech using multiple cues: A connectionist model. Language and Cognitive Processes, 13, 221-268.

Christiansen, M.H. & Chater, N. (1999). Toward a connectionist model of recursion in human linguistic performance. Cognitive Science, 23, 157-205.

Christiansen, M.H. & Chater, N. (Eds.) (2001a). Connectionist psycholinguistics. Westport, CT: Ablex.

Christiansen, M.H. & Chater, N. (2001b). Connectionist psycholinguistics: Capturing the empirical data. Trends in Cognitive Sciences, 5, 82-88.

Christiansen, M.H. & Dale, R.A.C. (2001). Integrating distributional, prosodic and phonological information in a connectionist model of language acquisition. In Proceedings of the 23rd Annual Conference of the Cognitive Science Society (pp. 220-225). Mahwah, NJ: Lawrence Erlbaum.

Curtin, S., Christiansen, M.H., Mintz, T.H., Byrd, D. & Werker, J.F. (in preparation). Stress changes the representational landscape: Evidence from word segmentation.


rule.gif (155 bytes)

Home | People | Research | Links | Contact | Publications | Presentations
Cognitive Neuroscience Laboratory

Please email suggestions/errors to mhc27@cornell.edu