Synergies in learning words and their referents

Mark Johnson, Katherine Demuth, Michael Frank, Bevan K. Jones

Research output: Chapter in Book/Report/Conference proceedingConference proceeding contributionResearchpeer-review

Abstract

This paper presents Bayesian non-parametric models that simultaneously learn to segment words from phoneme strings and learn the referents of some of those words, and shows that there is a synergistic interaction in the acquisition of these two kinds of linguistic information. The models themselves are novel kinds of Adaptor Grammars that are an extension of an embedding of topic models into PCFGs. These models simultaneously segment phoneme sequences into words and learn the relationship between non-linguistic objects to the words that refer to them. We show (i) that modelling inter-word dependencies not only improves the accuracy of the word segmentation but also of word-object relationships, and (ii) that a model that simultaneously learns word-object relationships and word segmentation segments more accurately than one that just learns word segmentation on its own. We argue that these results support an interactive view of language acquisition that can take advantage of synergies such as these.

LanguageEnglish
Title of host publicationAdvances in Neural Information Processing Systems 23: 24th Annual Conference on Neural Information Processing Systems 2010, NIPS 2010
EditorsJ. Lafferty, C. K. Williams, J. Shawe-Taylor, R. S. Zemel, A. Culotta
Place of PublicationUnited States
PublisherNeural Information Processing Systems (NIPS) Foundation
Pages1018-1026
Number of pages9
ISBN (Print)9781617823800
Publication statusPublished - 2010
Event24th Annual Conference on Neural Information Processing Systems 2010, NIPS 2010 - Vancouver, BC, Canada
Duration: 6 Dec 20109 Dec 2010

Other

Other24th Annual Conference on Neural Information Processing Systems 2010, NIPS 2010
CountryCanada
CityVancouver, BC
Period6/12/109/12/10

Fingerprint

Linguistics

Cite this

Johnson, M., Demuth, K., Frank, M., & Jones, B. K. (2010). Synergies in learning words and their referents. In J. Lafferty, C. K. Williams, J. Shawe-Taylor, R. S. Zemel, & A. Culotta (Eds.), Advances in Neural Information Processing Systems 23: 24th Annual Conference on Neural Information Processing Systems 2010, NIPS 2010 (pp. 1018-1026). United States: Neural Information Processing Systems (NIPS) Foundation.
Johnson, Mark ; Demuth, Katherine ; Frank, Michael ; Jones, Bevan K. / Synergies in learning words and their referents. Advances in Neural Information Processing Systems 23: 24th Annual Conference on Neural Information Processing Systems 2010, NIPS 2010. editor / J. Lafferty ; C. K. Williams ; J. Shawe-Taylor ; R. S. Zemel ; A. Culotta. United States : Neural Information Processing Systems (NIPS) Foundation, 2010. pp. 1018-1026
@inproceedings{e809516b32244a13ab66a6b242da40b8,
title = "Synergies in learning words and their referents",
abstract = "This paper presents Bayesian non-parametric models that simultaneously learn to segment words from phoneme strings and learn the referents of some of those words, and shows that there is a synergistic interaction in the acquisition of these two kinds of linguistic information. The models themselves are novel kinds of Adaptor Grammars that are an extension of an embedding of topic models into PCFGs. These models simultaneously segment phoneme sequences into words and learn the relationship between non-linguistic objects to the words that refer to them. We show (i) that modelling inter-word dependencies not only improves the accuracy of the word segmentation but also of word-object relationships, and (ii) that a model that simultaneously learns word-object relationships and word segmentation segments more accurately than one that just learns word segmentation on its own. We argue that these results support an interactive view of language acquisition that can take advantage of synergies such as these.",
author = "Mark Johnson and Katherine Demuth and Michael Frank and Jones, {Bevan K.}",
year = "2010",
language = "English",
isbn = "9781617823800",
pages = "1018--1026",
editor = "J. Lafferty and Williams, {C. K.} and J. Shawe-Taylor and Zemel, {R. S.} and A. Culotta",
booktitle = "Advances in Neural Information Processing Systems 23: 24th Annual Conference on Neural Information Processing Systems 2010, NIPS 2010",
publisher = "Neural Information Processing Systems (NIPS) Foundation",

}

Johnson, M, Demuth, K, Frank, M & Jones, BK 2010, Synergies in learning words and their referents. in J Lafferty, CK Williams, J Shawe-Taylor, RS Zemel & A Culotta (eds), Advances in Neural Information Processing Systems 23: 24th Annual Conference on Neural Information Processing Systems 2010, NIPS 2010. Neural Information Processing Systems (NIPS) Foundation, United States, pp. 1018-1026, 24th Annual Conference on Neural Information Processing Systems 2010, NIPS 2010, Vancouver, BC, Canada, 6/12/10.

Synergies in learning words and their referents. / Johnson, Mark; Demuth, Katherine; Frank, Michael; Jones, Bevan K.

Advances in Neural Information Processing Systems 23: 24th Annual Conference on Neural Information Processing Systems 2010, NIPS 2010. ed. / J. Lafferty; C. K. Williams; J. Shawe-Taylor; R. S. Zemel; A. Culotta. United States : Neural Information Processing Systems (NIPS) Foundation, 2010. p. 1018-1026.

Research output: Chapter in Book/Report/Conference proceedingConference proceeding contributionResearchpeer-review

TY - GEN

T1 - Synergies in learning words and their referents

AU - Johnson, Mark

AU - Demuth, Katherine

AU - Frank, Michael

AU - Jones, Bevan K.

PY - 2010

Y1 - 2010

N2 - This paper presents Bayesian non-parametric models that simultaneously learn to segment words from phoneme strings and learn the referents of some of those words, and shows that there is a synergistic interaction in the acquisition of these two kinds of linguistic information. The models themselves are novel kinds of Adaptor Grammars that are an extension of an embedding of topic models into PCFGs. These models simultaneously segment phoneme sequences into words and learn the relationship between non-linguistic objects to the words that refer to them. We show (i) that modelling inter-word dependencies not only improves the accuracy of the word segmentation but also of word-object relationships, and (ii) that a model that simultaneously learns word-object relationships and word segmentation segments more accurately than one that just learns word segmentation on its own. We argue that these results support an interactive view of language acquisition that can take advantage of synergies such as these.

AB - This paper presents Bayesian non-parametric models that simultaneously learn to segment words from phoneme strings and learn the referents of some of those words, and shows that there is a synergistic interaction in the acquisition of these two kinds of linguistic information. The models themselves are novel kinds of Adaptor Grammars that are an extension of an embedding of topic models into PCFGs. These models simultaneously segment phoneme sequences into words and learn the relationship between non-linguistic objects to the words that refer to them. We show (i) that modelling inter-word dependencies not only improves the accuracy of the word segmentation but also of word-object relationships, and (ii) that a model that simultaneously learns word-object relationships and word segmentation segments more accurately than one that just learns word segmentation on its own. We argue that these results support an interactive view of language acquisition that can take advantage of synergies such as these.

UR - http://www.scopus.com/inward/record.url?scp=84860635395&partnerID=8YFLogxK

M3 - Conference proceeding contribution

SN - 9781617823800

SP - 1018

EP - 1026

BT - Advances in Neural Information Processing Systems 23: 24th Annual Conference on Neural Information Processing Systems 2010, NIPS 2010

A2 - Lafferty, J.

A2 - Williams, C. K.

A2 - Shawe-Taylor, J.

A2 - Zemel, R. S.

A2 - Culotta, A.

PB - Neural Information Processing Systems (NIPS) Foundation

CY - United States

ER -

Johnson M, Demuth K, Frank M, Jones BK. Synergies in learning words and their referents. In Lafferty J, Williams CK, Shawe-Taylor J, Zemel RS, Culotta A, editors, Advances in Neural Information Processing Systems 23: 24th Annual Conference on Neural Information Processing Systems 2010, NIPS 2010. United States: Neural Information Processing Systems (NIPS) Foundation. 2010. p. 1018-1026