Rereading text: Words and their context.

Two experiments examined recent claims thai rereading benefits are mediated by abstract word-level representations (Carr, Brown, & Charalambous, 1989) and are subject to the reader's focus of attention (Carr & Brown, 1990). The effect of prior reading history was measured when readers reread a normal text and when they reread a scrambled word version of the texl. The prior reading history was either with the normal or the scrambled versions. In Experiment I readers were oriented to surface level processing by a Greek letter detection task while reading. In Experiment 2 they were oriented to message level processing by a later sum- marizing requirement. Results indicated that transfer to normal text rereading is context specific, and this specificity is resistant to influences by focus of attention. Transfer to rereading scrambled word displays is variable, usually nonspecific, and perhaps open to attentional strategy. We argue that lexical contributions to transfer while reading normal text cannot be assessed by use of measures at different linguistic levels. RESLIMK Nous avons examine dans deux experiences des propositions rcccntcs selon lesquelles les effets benefiques de la relecture seraient medics par des repre- sentations abstraites au niveau du mot (Carr, Brown, & Charalambous, 1989). Les effets d'une experience prealable de lecture etaient mesurcs quand les lecteurs reli- saient un tcxte normal et quand ils rclisaicnt unc version du tcxtc dans Icqucl les mots etaient brouilles. L'cxpcricnce prealable de lecture ctait acquisc soit avec la version normale ou avee la version brouillcc. Dans l'cxpcricncc I, les lecteurs etaient diriges vers un niveau dc traitcment dc surface par lc biais d'unc tachc dc detection d'une lettre grecque pendant la lecture. Dans l'expcrience 2, ils etaient oricnlcs vers un niveau dc traitement de message par une demande plus lardive de resumer le texte. Les rcsultats ont montrc quc lc transfcrt a la rclccturc du tcxtc normal cst specifiquc au contcxtc ct cette spccilicilc est resislante aux influences grace au foyer d'attcntion. Le translert a la relecture du lexte brouille est variable, habiluellenient non specifique et peut-etre ouvert a unc strategic attentionnelle. Nous proposons quc les contributions lexicales du transfer! pendant la lecture d'un textc normal ne peuvent elre evaluees par I 'utilisation dc mesures de diffcrcnts niveaux linguistiques. Rereading tasks provide an avenue for exploring the perceptual and conceptual processes involved in reading. Savings in reading time between the original and later reading trials act as indirect indicators of memory for the initial reading encounter. Like other indirect memory measures, reading time does not require the reader to actively remember the previous experience, yet it reflects the memories that are avail- able to aid later performance. Recent studies have indicated that perceptual, lexical,