Grammatical and associative constraints in sentence retention

Abstract

The free learning technique was used to test several hypotheses about sentence processing in recall. Sentence complexity (mean depth), syntactic type. (active, passive), and subjectobject association direction were found to be significant factors in over-all sentence recall. Recall error patterns showed differences between syntactic type and between association conditions. In addition, Ss showed evidence for an incomplete transformational process. The present experiment was conducted against a background of controversy about both the psychological structure and the mode of processing sentences. The aim of these studies was not so much a resolution of the controversies as an exploration of relevant variables and effects that require explication before a resolution can be proposed. One variable frequently emphasized in previous research has been syntactic type. Mehler (1963) has shown that when length and sentence complexity are uncontrolled, passive sentences are more poorly recalled than actives, with a tendency for passives to be recalled in their active form. The explanation of these phenomena has been couched in terms of transformational rules and s’s memory capacity. The nature of the processes by which S exercises these rules is as yet unclear. Nor is the case for this hypothesis furthered by the fact that, when complexity and length are controlled, passives are recalled as well as, or better than active sentences (Martin and Roberts, 1966). An alternative proposal is relevant: It has been hypothesized that the speaker is sensitive