Abstract
Background: Reading may be processed at either the level of the whole word or its individual letters, and the word-length effect may provide an index of serial letter processing versus rapid parallel or holistic processing. How reading is performed under various perceptual transforms and whether a word inversion effect is specific for normal text (as predicted by the expertise hypothesis) is not clear. Objective: We measured the word length effect in normal text or two transformations, mirror reflection (in which the form of the whole word is preserved) or written backwards, in both upright and inverted orientation. Methods: We measured verbal response time of 12 healthy subjects reading 3- to 9-letter words presented one at a time in random order, with transformations and orientations in different, counterbalanced blocks. Results: There was a main effect of transformation (F(2,55) = 39.52, p <.0001), with Tukey's HSD test now showing differences between all three transformations. Mirror text had a larger word-length effect than either backwards text (F(1,55) = 6.56 , p <.003) or normal text (F(1,55) = 9.68, p <.003), while backwards text also had a larger word-length effect than normal text (F(1,55) = 9.68, p <.003). There was a trend to an interaction between orientation and transformation (F(2,55) = 2.77, p <0.07). Tukey's HSD test showed that the inversion effect was significant for normal text (F(1,55) = 9.68, p <.003), but not for mirror or backward transformed text. Conclusion: Reading of perceptually difficult transformed text uses primarily local letter processing, consistent with predictions that rapid parallel or holistic word processing is acquired through experience and therefore limited to familiar text formats. The inversion effect suggests that the word-length effect is a more effective index of this expert process than mean response time. Funding: Canada Research Chair and Marianne Koerner Chair in Brain Diseases (JB)
Meeting abstract presented at VSS 2014