Presentation 2009-03-12
Role of statistical learning in grammar-like rule acquisition from speech streams
Osamu WATANABE, Miki HAYASAKA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) When infants are acquiring their mother tongues, they have to segment continuous speech streams at first and then discover grammatical structures that rule out how these words are connected. The phonetic information, such as an intonation, is an important cue for language acquisition, but would not be the only one. Words and syllables are not connected randomly in spoken language, and the statistical information, or the transitional probabilities of words and syllables, could provide cues for word segmentation and grammatical rule extraction. However, Pena et al. (2002) reported that the statistical information does not contribute to acquire grammatical-like rules from continuous speech streams. Here we show the result of Pena et al. can be explained by the framework of statistical learning with a computer simulation and a psychological experiment. The present result suggests that the statistical information is utilized for grammatical rule acquisitions as well as word segmentations.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) language acquisition / temporal pattern / grammatical rule / neural network / Elman net
Paper # NC2008-134
Date of Issue

Conference Information
Committee NC
Conference Date 2009/3/4(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Role of statistical learning in grammar-like rule acquisition from speech streams
Sub Title (in English)
Keyword(1) language acquisition
Keyword(2) temporal pattern
Keyword(3) grammatical rule
Keyword(4) neural network
Keyword(5) Elman net
1st Author's Name Osamu WATANABE
1st Author's Affiliation Department of Computer Science and Systems Engineering, Faculty of Engineering, Muroran Institute of Technology()
2nd Author's Name Miki HAYASAKA
2nd Author's Affiliation Department of Computer Science and Systems Engineering, Faculty of Engineering, Muroran Institute of Technology
Date 2009-03-12
Paper # NC2008-134
Volume (vol) vol.108
Number (no) 480
Page pp.pp.-
#Pages 6
Date of Issue