Laterally presented words: Orthographic analysis and serial, parallel or holistic modes of processing

Abstract

In a series of four experiments, manual reaction times were measured when subjects judged whether or not laterally-presented, single-syllable four-letter words were the same as a prespecified target. Nontargets differed systematically from targets by a single letter change at one of the four letter positions. Right visual field (RVF) superiorities were generally stronger with targets, with nontargets where letter changes occurred in the middle rather than the initial or final positions, and with upper or lower case typeface. A weak or nonsignificant RVF superiority occurred with cursive script (Experiment 2), which may have required a visuospatial preprocessing contribution from the right hemisphere, and where word shape (inclusion of ascending and descending letters) was changed between targets and nontargets (Experiment 4), which may also have invoked right hemisphere mechanisms. A change at an initial letter position was detected fastest, with either letter position 3 slowest, or all letter positions except the first equally slow. This suggests that processing can proceed largely in parallel for all except the first and possibly the last letter position. These findings were discussed in the context of lateral masking of internal by initial and final letters, of gradients of retinal acuity and of a possible periphery-to-fovea or ends-first processing gradient. We found little evidence of a serial left-to-right processing scan, or of global wholeword recognition based on outline word shape, or of a qualitative (as opposed to a quantitative) difference between the hemispheres in word recognition strategies.