Lexical access during eye fixations in reading: Are word access codes used to integrate lexical information across interword fixations?

Abstract

Eye fixation records during reading were obtained to test whether lexical access of a word is distributed across successive interword fixations. Specifically, prelexical analyses discerning the word access code could be performed during the parafoveal processing of a word and lexical access could be completed during the word's subsequent fixation. A series of sentences was read with each sentence containing either a 6-letter compound word (e.g., cowboy), a 6-letter pseudocompound word (e.g., carpet), or a 6-letter control word (e.g., mirror). Reading occurred under four parafoveal text preview conditions affording either no usable preview, preview of a word's beginning three letters, preview of a word's beginning four letters, or preview of a complete parafoveal word. Previews of three and four letters either corresponded to linguistically defined subword units (putative word access codes) or violated these units. The results showed that parafoveal preview benefits from 3- and 4-letter sequences were unaffected by the linguistic status of previewed letters, indicating that word access codes are not used to integrate lexical information across interword fixations. Parafoveal previews of whole words revealed, however, effects of linguistic status. Preview benefits were largest for pseudocompound words which comprised high frequency beginning and ending subword constituents and smallest for control words which did not contain subword constituents with lexical representations. These results support models of parafoveal word recognition which posit, first, that all letters of a parafoveal word contribute to the early phase of word identification and, second, that these letters form mutually reinforcing letter coalitions.