Subjects performed for five trials on a free recall list containing both rhyming word pairs and word-associate pairs, with either visual or auditory presentation. High-anxiety subjects recalled and clustered less than low-anxiety subjects for both rhymes and associates. There was little effect due to input modality, nor did modality interact with anxiety.Recent research has shown that high-anxiety subjects do not perform as well as low-anxiety subjects in free recall (cf. Mueller, 1976). Specifically, although highanxiety subjects may recall as many words, their recall is less organized; for example, there is less clustering of words in terms of intralist relationships such as taxonomic category membership. One interpretation of this deficit is that high-anxiety subjects engage in a more restricted encoding of the items, and thus are less able to utilize various word features to advantage in organizing material in memory.A question of interest concerns the generality of this cue utilization deficit. It has been observed (Broadbent, 1971;Schwartz, 1975) that states of high arousal seem to orient the subject's attention to the physical characteristics of linguistic materials, at the expense of semantic content. One way to conceptualize this is in terms of the depth-of-processing analysis (Craik & Lockhart, 1972). This view assumes that "shallow" processing of a word involves attending to physical features, that is, its orthography or sound, with the result being an unstable or transitory memory trace. "Deep" processing is said to occur when semantic content is examined, and it results in a stronger memory trace. One hypothesis, then, might be that high-anxiety subjects have a deficit in processing deep (semantic) features, but no deficit in terms of processing shallow (physical) attributes. While arousal and anxiety may not be identical, the arousal results noted above are consistent with this interpretation.A test of this hypothesis should be possible in free recall, given that the word list involves both shallow and deep dimensions for organization. The present experiment used a list construction with both rhyming (shallow) and word-associate ( deep) relationships in the same list (cf. Dolinsky, 1972). With such a list, subjects can cluster their recall by associates, rhymes, or both. The anxiety-depth hypothesis suggests that high-anxiety subjects would show a deficit in terms of associative organization, as has been found previously, but no deficit should be observed for acoustic organization. As a methodological point, the method of list presentation was either visual or auditory, to determine whether this had any effect on the perception of the presence of the acoustic relationships.
METHOD Subjects and DesignSixty-four students in introductory psychology courses were selected on the basis of test anxiety scores (Sarason, 1972). High-anxiety subjects had scores greater than 19 (mean = 28.1), and low-anxiety subjects had scores below 10 (mean = 7.2). The basic design was a 2 by 2 factorial for test anxiety (h...