Brandon Hansen, Utah Valley University
Social and Behavioral Sciences
Measuring eye movements during reading is an effective and ecologically valid way to investigate dynamic changes in human cognition. The boundary technique (Rayner, 1975) is often employed during experimental reading research by changing text in real-time between eye movements—permitting subtle manipulations which are not obvious to the reader. Frequently, these manipulations include the replacement of a target word (beach) by words that are homophones (beech), semantically related (shore), orthographically related (bench), or by random letter masks (hxnzt). Understanding the differences between text in which the target word is available (beach) and denied (hxnzt) allows a deeper understanding of dynamic cognitive processes. In a real-world sense, such investigations yield findings that assists doctors, clinicians, and educators as they create interventions for those, for example, with learning disabilities such as dyslexia.
Our research question addressed the amount of information available parafoveally, the area outside of what we clearly see while reading, while using the boundary technique. Of particular interest was how random letter masking of word shape information disrupted parafoveal preprocessing of upcoming words. This is theoretically and methodologically interesting as it impacts current reading models. We evaluated peer-reviewed journal articles in which participants read single sentences in English using meta-analysis. All studies included used eye movements as dependent measures (e.g., saccade latency). The results indicated several important methodological implications. First, the author’s theoretical orientation yielded significant differences in effect sizes. Next, a significant relationship was demonstrated between studies with high and low amounts of excluded data (those with higher exclusion rates tended to have the lowest effect sizes). Finally, there were no studies that investigated the nuances of random letter masks (e.g., similar v. dissimilar). Taken together, findings in this vein will have important methodological implications in a field which contains statistically significant differences of 30ms to 50ms.