The extent to which semantic differential ratings of trigrams remain consistent over a four-week retention period following pairing with positive or negative attitudinal words was determined. The trigrams initially served as stimulus items in a paired associate list consisting of both positive and negative evaluative response terms; and rating and recall periods were presented to five independent groups at the end of zero, one, two, three, and four-week retention intervals. The mean ratings for syllables paired with positive words and the syllables paired with negative words tended to converge across the four-week retention period. However, when the ratings were adjusted for the number of associates recalled, either statistically or experimentally, no evidence of systematic change in the ratings across time was obtained. These results were interpreted as supporting a classical conditioning interpretation of attitude formation as opposed to a word association interpretation. (Author)