Academic journal article Canadian Journal of Experimental Psychology

Strategies of Text Retrieval: A Criterion Shift Account

Academic journal article Canadian Journal of Experimental Psychology

Strategies of Text Retrieval: A Criterion Shift Account

Article excerpt

Abstract This study scrutinized people's ability to apply different strategies to randomly intermixed immediate and delayed test items. In three experiments, participants first read one set of stories. Later, they read more stories, and after each one, answered inter-- mixed questions about that story and one of the earlier ones. The experiments cumulatively manipulated amount of delay, test probe plausibility, probe relation (explicit, paraphrase, inference), and testing procedure (mixed versus uniform delay). Using signal detection response criterion as the index of strategy, we contrasted the single criterion hypothesis, according to which one text retrieval criterion is applied to all test items, and a multiple-criterion hypothesis. The results consistently favoured the multiple-criterion hypothesis. The results also indicated that the presence of immediate and delayed probes mutually influence one another: Less extreme signal detection criteria were adopted under mixed than uniform testing. It was concluded that text retrieval strategy is continually calibrated with reference to the quality of the test probes.

Theories of the cognition of text retrieval and question answering routinely posit distinct processing stages, including the encoding and categorization of the question, and the resulting memory search. Several theorists have proposed that strategy selection constitutes a distinct processing stage of question answering and text retrieval (Graesser & Murachver, 1985; Reder, 1982; Singer, 1990). Two prominently discussed text retrieval strategies are direct retrieval and plausibility judgement (Camp, Lachman, & Lachman, 1980; Lehnert, 1977; Reder, 1982, 1987, 1988; Singer, 1991a). Suppose that after reading a story about a man-eating tiger, one were asked, Were the villagers afraid of the tiger? One could attempt to directly retrieve this statement from the episodic memory representation of the text. Alternatively, one could compute its plausibility, either with reference to general knowledge or to other statements in the story (e.g., The villagers hired a hunter to help them, so they probably were afraid).

In one empirical test of these proposals, people read stories and were instructed to evaluate test questions either by recognizing them, a retrieval orientation, or by judging their plausibility (Reder, 1982). The recognizers answered accurately more quickly than the plausibility judges in immediate testing, but the reverse was true after two days. This outcome confirmed people's ability to apply different strategies in text retrieval. More specifically, Reder reasoned that, in immediate testing, the recognizers benefited more than the plausibility judges from the verbatim details of the memory representation. After two days, when those verbatim details faded (Clark & Sengul, 1979; Jarvella, 1971), the recognizers were hampered by their attempts to retrieve the test statements, resulting in longer correct answer times than those exhibited by the plausibility judges.

The present study focused on people's ability to adjust their text retrieval strategy one trial at a time. For example, Reder (1987, Experiment 3) presented advice to her participants, before each trial, to use either the direct retrieval strategy or the plausibility judgement strategy. Answer times were lower when the advice was helpful than when it was misleading, confirming that the participants implemented the suggested strategies.

Even more pertinent to the present study was Reder's (1988) report that item characteristics can also affect strategy selection on an item-by-item basis. Reder's participants viewed randomly intermixed sets of test sentences. Half of the items referred to an immediately preceding story (e.g., Man-Eating Tiger), and the others referred to a story read two days earlier (e.g., Steamboat Race). Each participant received instructions either to recognize or to judge the plausibility of the test sentences. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.