PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

Mutual Information and Semantic Similarity as Predictors of Word Association Strength: Modulation by Association Type and Semantic Relation
Anat Prior and Maayan Geffet
EuroCogSci 2003. Osnabruck, Germany 2003.


Associated word pairs differ in their degree of Association Strength, e.g., how commonly the target is given by subjects as a response to the cue. In the present work we investigated the degree to which Association Strength can be predicted by a measure of Mutual Information of the word pair, by the Semantic Similarity of the words, or by both factors jointly. We examined this issue in two compilations of free association norms, one in Hebrew and one in English, and analyzed circa 6,000 associated pairs in total. Further, English associated pairs were classified as syntagmatic or paradigmatic associates, while Hebrew noun pairs were classified according to the semantic relation between them (i.e. synonyms and antonyms). We found both Mutual Information and Semantic Similarity were significantly correlated with Association Strength for the English associates of both types. For the Hebrew associates, Mutual Information was significantly correlated with Association Strength for noun pairs related idiomatically, funcionally and heirarchically, while Semantic Similarity was found to reliably predict Association Strength only for antonyms. The importance of these results for understanding the cognitive operations underlying the free association task is discussed.

PDF - Requires Adobe Acrobat Reader or other PDF viewer.
EPrint Type:Article
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Computational, Information-Theoretic Learning with Statistics
ID Code:379
Deposited By:Maayan Geffet
Deposited On:18 December 2004