Academic journal article Memory & Cognition

Repeating Phrases across Unrelated Narratives: Evidence of Text Repetition Effects

Academic journal article Memory & Cognition

Repeating Phrases across Unrelated Narratives: Evidence of Text Repetition Effects

Article excerpt

Research has shown that text repetition effects are limited to conditions in which the context remains consistent across the two processing episodes, particularly when readers are focused on comprehension. Despite this, we found evidence of transfer effects across unrelated narratives. In a repeated condition, an ambiguous phrase appeared in two consecutive stories. In Story A, the phrase was presented in a sarcasm-biasing context, and in Story B, the phrase was presented in a neutral context. The pattern of findings from an offline measure (Experiment 1) and a reading time measure (Experiments 2, 3, and 4) indicated that participants were more likely to interpret the phrase in Story B as sarcastic in the repeated version than in a nonrepeated version, in which the phrase was absent from Story A. We conclude that during the reading of Story B, the phrase was reactivated from memory, even though the two stories were unrelated.

An important goal for a theory of discourse comprehension is to determine what information is included in a reader's memory representation. One of the ways this question has been addressed has been through studies of rereading, in which participants are asked to read a text two times. The logic of rereading studies is that only information that is included in a reader's memory representation from the original reading can facilitate rereading. Facilitation, defined as an increase in speed during a second reading or improved performance on a secondary task such as a probe task, is commonly referred to as a rereading benefit or text repetition effect (see, e.g., Carr, Brown, & Charalambous, 1989; Levy & Bums, 1990; Raney & Rayner, 1995; Raney, Therriault, & Minkoff, 2000).

An interesting finding from a large number of studies is that repetition effects are often limited to conditions in which the context remains consistent across the two processing episodes. Consider a few examples: Levy et al. (1995) found rereading benefits, measured by shorter reading times on the second reading, in a related condition in which two passages came from the same novel and, thus, shared characters, a story line, and a significant number of content words. In contrast, they found no rereading benefit in a condition in which the two passages shared neither characters nor a theme, but shared the same number of content words as the related condition. Similarly, Levy and Burns ( 1990) presented multiparagraph passages that were identical across two readings, had the individual paragraphs reordered, had the sentences reordered, or had the words reordered. Although there were rereading benefits for the identical condition and the paragraph-reordered condition, there were limited rereading benefits for the sentence-reordered condition and no rereading benefits for the word-reordered condition. And finally, using ERPs as the dependent measure, Besson and Kutas ( 1993) found repetition effects (indicated by a decrease in N400 amplitude during a second reading) when words were presented a second time in their original sentence frame but not when they were presented in a new sentence.

Studies using word lists have also found that repetition benefits tend to be limited to conditions in which the context remains consistent across the two processing episodes. For example, using a perceptual identification task, Levy and Kirsner (1989) found a benefit for words that were both studied and tested as a part of a word list, but found no benefit for words that were originally seen in the context of a text and subsequently identified in a word list. Oliphant (1983) used a lexical decision task and found a reprocessing benefit when words were repeated within the lexical decision task, but found no reprocessing benefit when the words were first processed as part of the instructions for the experiment. And Jacoby (1983) presented a test list that had either 90% of the words in common with the study list or 10% of the words in common, and found significantly reduced repetition effects when there was only 10% overlap. …

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.