Correlation between Word Association Strength and Associative Overlap

Abstract

Two measures of natural, or pre-experimental, language habits which have received considerable attention in the recent verbal learning literature may be described as word association strength (Russell & Jenkins, 1954) and associative overlap (Deese, 1962). The two indices have seldom been compared. The intent of the present srudy was to determine the gross relationship between them by correlaring values obtained with the two techniques across a large number of word pairs. 128 word pairs were drawn from a large body of normative associative overlap data in preparation by the authors. The assessment technique employed in the collection of this normative data is fully described by Garskof and Houston (1963). Briefly, associarive overlap increases ar; the number and strength of common word associates increases. The 128 word pairs chosen for the present study represented a wide range of overlap values. 38 Michigan undergraduates associated to the words which comprised the 128 pairs in a standard word association task. The sum of the numbers of times the members of a given pair evoked each other in the association task was taken as the word-association value for that pair. These 128 sums, or the association values, were correlated with the 128 overlap values obtained from the normative overlap data. The Pearson productmoment correlation was .59 (p < .01), indicating a significant positive relationship between word association strength and associative overlap. Care should be taken in interpreting this result. The correlation does nor suggest that the two indices reflect the same underlying process. In fact, there is some evidence for the conclusion that the two techniques measure relationships which can operate independently (Garskof, Houston, & Ehrlich, in press; Houston & Garskof, 1963). The positive correlation merely implies that the words which evoke each other also tend to elicit common associates. This result argues for the gross co-variation of associative strength and associative overlap but not for their identity.