Presentation 2004/6/17
Neural Network Model of Syntactic Knowledge Acquisition Based on Data-driven Learning Theory
Daichi MORIFUJI, Toshio INUI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Recent developmental studies support the concept of "data-driven learning theory", which suggests that the properties of linguistic stimuli play important roles in syntactic knowledge acquisition. Based on this theory, a neural network model of syntactic knowledge acquisition was constructed. The model also implemented functions that have been separately suggested to be related to linguistic ability by both developmental and brain imaging studies. During the learning phase, the model appeared to extract regularities, such as word order and lexical co-ocurrence relationships, from linguistic stimuli that did not explicitly include categorical information, and then categorized words according to these regularities. As a result of these functions, the model demonstrated a word acquisition curve whose characteristics displayed quantitative similarities to those of children. Thus, we supported the validity of "data-driven learning theory", from a connectionist perspective.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) language development / neural network model / data-driven learning theory
Paper # NC2004-13
Date of Issue

Conference Information
Committee NC
Conference Date 2004/6/17(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) Neural Network Model of Syntactic Knowledge Acquisition Based on Data-driven Learning Theory
Sub Title (in English)
Keyword(1) language development
Keyword(2) neural network model
Keyword(3) data-driven learning theory
1st Author's Name Daichi MORIFUJI
1st Author's Affiliation Graduate School of Informatics, Kyoto University()
2nd Author's Name Toshio INUI
2nd Author's Affiliation Graduate School of Informatics, Kyoto University
Date 2004/6/17
Paper # NC2004-13
Volume (vol) vol.104
Number (no) 139
Page pp.pp.-
#Pages 6
Date of Issue