Serial and parallel processing in reading: Investigating the effects of parafoveal orthographic information on nonisolated word recognition

Abstract

We present a novel lexical decision task and three boundary paradigm eye-tracking experiments that clarify the picture of parallel processing in word recognition in context. First, we show that lexical decision is facilitated by associated letter information to the left and right of the word, with no apparent hemispheric specificity. Second, we show that parafoveal preview of a repeat of word n at word n + 1 facilitates reading of word n relative to a control condition with an unrelated word at word n + 1. Third, using a version of the boundary paradigm that allowed for a regressive eye movement, we show no parafoveal “postview” effect on reading word n of repeating word n at word n – 1. Fourth, we repeat the second experiment but compare the effects of parafoveal previews consisting of a repeated word n with a transposed central bigram (e.g., caot for coat) and a substituted central bigram (e.g., ceit for coat), showing the latter to have a deleterious effect on processing word n, thereby demonstrating that the parafoveal preview effect is at least orthographic and not purely visual.

Keywords:

• Reading
• Word recognition
• Eye tracking
• Parafoveal preview
• Parallel processing
• Serial processing

Acknowledgments

The first author was supported by an EPSRC (Engineering and Physical Sciences Research Council, UK) studentship. The second author was supported by ESRC (Economic and Social Research Council, UK) Project Grants R39195 and R39942.

Notes

¹ Parallel processing theorists such as Kliegl and colleagues instead characterize attention paid to the left of the fixated word as a normal part of the pattern of information gathering and processing, implying that there is useful information for researchers in the previously discarded fixations either following or preceding a regression. (It would be worthwhile for future research on the topic of processing of word n – 1 to include an analysis of those discarded fixations and of the regressions that are made, as it is possible that the regression patterns recorded are more than an anomaly; eye-movement corpora from text-reading studies may be the best, richest source for such data.)