Articles

The company that words keep: comparing the statistical structure of child- versus adult-directed language  ^*

THOMAS HILLS 

University of Warwick

ABSTRACT

Does child-directed language differ from adult-directed language in ways that might facilitate word learning? Associative structure (the probability that a word appears with its free associates), contextual diversity, word repetitions and frequency were compared longitudinally across six language corpora, with four corpora of language directed at children aged 1 ; 0 to 5 ; 0, and two adult-directed corpora representing spoken and written language. Statistics were adjusted relative to shuffled corpora. Child-directed language was found to be more associative, repetitive and consistent than adult-directed language. Moreover, these statistical properties of child-directed language better predicted word acquisition than the same statistics in adult-directed language. Word frequency and repetitions were the best predictors within word classes (nouns, verbs, adjectives and function words). For all word classes combined, associative structure, contextual diversity and word repetitions best predicted language acquisition. These results support the hypothesis that child-directed language is structured in ways that facilitate language acquisition.

(Received August 25 2010)

(Revised November 24 2011)

(Accepted April 10 2012)

(Online publication May 14 2012)

Correspondence

Address for correspondence: Thomas Hills, University of Warwick, Department of Psychology, Gibett Hill Road, Coventry CV4 7AL, UK. tel: +44-(0) 24-7652-3183. e-mail: t.t.hills@warwick.ac.uk

Footnotes

^[*]  I thank Linda Smith, Josita Maouene, Brian Riordon, Peter Todd, Thorstun Pachur and Sara Hills for suggestions and comments on the research and manuscript. I also thank Fionniain Hills for the banjo–violin example. This work was supported by a grant from the Swiss National Science Foundation (100014 130397/1).