Abstract
Reading is a recent cultural invention that exploits the intrinsic recognition capacities of our visual system. But does reading consist only of learning letter-sound mappings, or does it also fundamentally alter letter representations? This question has been difficult to address because (1) illiterates and literates everywhere differ along socioeconomic and cognitive dimensions that confound all comparisons and (2) in the Western world, nearly all languages use nearly the same Latin letters. We addressed this question by exploiting the orthographic diversity of Indian languages. Specifically, we identified two distinct groups of students (both English-literate) but with one group fluent in reading the Telugu script but not the Malayalam script, and the other group fluent in reading Malayalam but not Telugu. To probe letter representations without reading, we used oddball visual search as a natural index of similarity between shapes. In Experiment 1, both groups of readers searched for a Telugu letter among Telugu letters, or a Malayalam letter among Malayalam letters. The main result is a double dissociation: in both groups, searching among familiar letters was more efficient than searching among novel letters. However, letter representations were only subtly altered because search times were strongly correlated across literates and illiterates. In Experiment 2, we asked whether reading alters the relationship between familiar and novel letters. To this end, both groups of subjects performed searches with the target from one script and the distractors were from the other. The main result was again a double dissociation: in both groups, finding an unfamiliar letter was easier among familiar letters than vice-versa. Taken together, our results show that reading alters letter representations through a somewhat paradoxical effect of familiarity: familiar letters are both more discriminable, yet less salient compared to novel letters.
Meeting abstract presented at VSS 2017