Low-Vision Reading Speed: Influences of Linguistic Inference and Aging

Abstract

Reading is a dynamic task involving both linguistic and visual analysis. In this study, we asked how two types of linguistic information--characters used in segmenting words from one another, and sentence context--differ in their usefulness for people with normal and low vision. Given evidence for age-related differences in some forms of cognitive processing, we also investigated the effect of age. There were four groups of 10 participants: vision status (normal, low) crossed with age (young, <35 years; old, >65 years). Reading speeds were compared for regularly spaced text and text in which the spaces were removed, a manipulation intended to eliminate local cues for text segmentation and force attention to clusters of letters or whole words. We also evaluated the effect of sentence context by comparing reading speeds for regular sentences and sentences in which word order was scrambled. Removal of spaces had a greater impact on low vision than normal vision, reducing average speeds to 45% and 66% of speeds for regularly spaced text, respectively. We interpret this to mean that people with low vision have less access to spatially distributed linguistic regularities of text such as prefixes, suffixes, or word length. Removal of sentence context through scrambling had a greater impact on normal vision than low vision, reducing mean reading speed to 53% and 66%, respectively. Finally, comparison of our young and old readers showed no major differences in the use of sentence context or in the impact of removing spaces between words. People with low vision appear to rely more on spacing information in sentences, whereas people with normal vision appear to make better use of sentence context, irrespective of age.