Extending models of visual-word recognition to semicursive scripts: Evidence from masked priming in uyghur

One basic feature of the Arabic script is its semicursive style: some letters are connected to the next, but others are not, as in the Uyghur word /ya x∫i/ ("good"). None of the current orthographic coding schemes in models of visual-word recognition, which were created for the Roman script, assign a differential role to the coding of within letter "chunks" and between letter "chunks" in words in the Arabic script. To examine how letter identity/position is coded at the earliest stages of word processing in the Arabic script, we conducted 2 masked priming lexical decision experiments in Uyghur, an agglutinative Turkic language. The target word was preceded by an identical prime, by a transposed-letter nonword prime (that either kept the ligation pattern or did not), or by a 2-letter replacement nonword prime. Transposed-letter primes were as effective as identity primes when the letter transposition in the prime kept the same ligation pattern as the target word, but not when the transposed-letter prime didn't keep the ligation pattern. Furthermore, replacement-letter primes were more effective when they kept the ligation pattern of the target word than when they did not. We examined how input coding schemes could be extended to deal with the intricacies of semicursive scripts.