The Return Sweep in Reading

A comprehensive model of eye movement control in reading will have to take into consideration not only progressive and regressive inter- and intraword saccades, but also the large return sweep saccades that bring the eyes from the end of one line of text to the beginning of another line (Suppes 1994). In normal text, about 20 percent of all fixations are preceded or followed by a return sweep and the contribution of these fixations to total reading time is substantial. From the perspective of text design, the optimal line length is assumed to depend on the reader’s ability to programme and execute an accurate return sweep and to avoid the costs of additional corrective saccades (Tinker 1963, Heller 1982). Given both theoretical and practical importance, surprisingly little is known about the parameters characterising return sweeps and the factors that control them. This chapter is an attempt towards closing this gap. We will first consider return sweeps as a special case of large goal-directed saccades and briefly look at previous findings. We will then present descriptive data on return sweep landing positions and the occurrence of corrective saccades and explore the effects of line length and text justification on these variables.