?url_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Adc&rft.title=Modeling+the+development+of+lexicon+with+a+growing+self-organizing+map&rft.creator=Farkas%2C+Igor&rft.creator=Li%2C+Ping&rft.subject=Language&rft.subject=Semantics&rft.subject=Psycholinguistics&rft.description=We+present+a+self-organizing+neural+network+model+that+can+acquire+an+incremental+lexicon.+The+model+allows+the+acquisition+of+new+words+without+disrupting+learned+structure.+The+model+consists+of+three+major+components.+First%2C+the+word+co-occurrence+detector+computes+word+transition+probabilities+and+represents+word+meanings+in+terms+of+context+vectors.+Second%2C+word+representations+are+projected+to+a+lower%2C+constant+dimension.+Third%2C+the+growing+lexical+map+(GLM)+self-organizes+on+the+dimension-reduced+word+representations.+The+model+is+initialized+with+a+subset+of+units+in+GLM+and+a+subset+of+the+lexicon%2C+which+enables+it+to+capture+the+regularities+of+the+input+space+and%0Adecrease+chances+of+catastrophic+interference.+During+growth%2C+new+nodes+are+inserted+in+order+to+reduce+the+map+quantization+error%2C+and+the+insertion+occurs+only+to+yet+unoccupied+grid+positions%2C+thus+preserving+the+2D+map+topology.+We+have+tested+GLM+on+a+portion+of+parental+speech+extracted+from+the+CHILDES+database%2C+with+an+initial+200+words+scattered+among+800+nodes.+The+model+demonstrates+the+ability+to+highly+preserve+learned+lexical+structure+when+100+new+words+are+gradually+added.+Implications+of+the+model+are+discussed+with+respect+to+language+acquisition+by+children.%0A&rft.date=2002&rft.type=Conference+Paper&rft.type=PeerReviewed&rft.format=application%2Fpostscript&rft.identifier=http%3A%2F%2Fcogprints.org%2F2150%2F1%2Fijci2002.ps.gz&rft.identifier=++Farkas%2C+Igor+and+Li%2C+Ping++(2002)+Modeling+the+development+of+lexicon+with+a+growing+self-organizing+map.++%5BConference+Paper%5D+++++&rft.relation=http%3A%2F%2Fcogprints.org%2F2150%2F