Reading strategy modulates parafoveal-on-foveal effects in sentence reading

Abstract

Task demands and individual differences have been linked reliably to word skipping during reading. Such differences in fixation probability may imply a selection effect for multivariate analyses of eye-movement corpora if selection effects correlate with word properties of skipped words. For example, with fewer fixations on short and highly frequent words the power to detect parafoveal-on-foveal effects is reduced. We demonstrate that increasing the fixation probability on function words with a manipulation of the expected difficulty and frequency of questions reduces an age difference in skipping probability (i.e., old adults become comparable to young adults) and helps to uncover significant parafoveal-on-foveal effects in this group of old adults. We discuss implications for the comparison of results of eye-movement research based on multivariate analysis of corpus data with those from display-contingent manipulations of target words.

Keywords:

• Fixational selectivity
• Parafoveal-on-foveal effects
• Reading strategy
• Distributed processing
• Eye movements in reading

Acknowledgments

This research is part of Christiane Wotschack's dissertation (supported in part by a doctoral fellowship of the University of Potsdam) and was carried out in the context of a European Collaborative Research Project grant (PIs: Reinhold Kliegl, Alan Kennedy, Richard Shillcock, and Heinz Wimmer), with the German stage funded by Deutsche Forschungsgemeinschaft KL 955/7 and KL 955/14; PIs: Reinhold Kliegl and Ralf Engbert). We would like to thank Erik D. Reichle and two anonymous reviewers for their helpful comments on an earlier version of the manuscript. Data and R scripts are available at Potsdam Mind Research Repository (http://read.psych.uni-potsdam.de/pmr2).

Notes

¹ Data from the easy-question groups of young and old readers were included in the analyses in Kliegl et al. (2006) and Kliegl (2007), labelled as Group 4 and Group 9, respectively.

² The 10 subject-level predictors in the PCA were mean incoming and outgoing saccade amplitude, mean skipping probability of the previous and next word, and mean frequency, mean length, and mean proportion of fixated content words of the previous and fixated words.

³ Erik Reichle suggested this explanation.