[PS-2.26] Transposed letter effects within an artificial language

Lally, C. , Taylor, J. & Rastle, K.

Royal Holloway University of London

Successful reading requires not only identification of a word's constituent letters but also analysis of their positions. Evidence from Indo-European languages suggests that position is represented flexibly, such that readers find items with transposed letters (TLs) perceptually similar. However, ground-breaking new evidence indicates that this tolerance does not extend to Semitic writing systems. This suggests that flexibility in letter position coding is not hardwired in the brain, but influenced by aspects of the writing system.

We tested this hypothesis using a powerful new method, in which adults learned novel words in an artificial script, constructed to be sparse (few anagrams) or dense (many anagrams). This allowed us to test how statistical properties of writing systems influence letter position coding, in a manner that could never be achieved in typical cross-linguistic comparisons.

Participants learned to read twenty-four items, before completing lexical decision tasks with TL, substitution-letter, and control distractors. In sparse orthographies, participants found TLs harder to reject than substitution distractors, indicating flexibility in position coding. Analyses from dense orthographies indicate greater position restrictions. For the first time, we have demonstrated that TL effects can be observed in artificial orthographies and revealed how orthographic density influences the precision of position coding.