For adults, the segmentation of continuous speech into words is generally effortless, but it is not so for infants. In this paper we discuss an algorithmic-level model of word segmentation, developing a specific, cognitively plausible approach to segmentation that is better integrated with research in infant development than previous models. We use experimental evidence to suggest the operations the segmenter should perform and show that a simple learning algorithm predicts the type of developmental changes observed in child language development.
Proceedings of the 30th West Coast Conference on Formal Linguistics
edited by Nathan Arnett and Ryan Bennett
Table of contents