Abstract
Three experiments were conducted to investigate letter migration in Japanese Kana word recognition. The participants were presented two brief masked “source” words (e.g., _??__??_ _??__??_) each made up of two components (letters), followed by a “probe” word (e.g., _??__??_). The probe word in the critical trials was a blend of two letters, one each from the two source words. The task was to decide whether the probe was one of the source words. The results indicated that the proportion of false positive responses depends on consistency of positions of the components in probe word with those in source words set (global consistency), and on consistency between the first letter of source and probe words (left local consistency). The results also showed that statistical properties of letters, i.e., the number of companions (adjoining letters) of a letter influenced the false response. These results were compared with those with a Kanji word suggesting that knowledge of conjunctive properties of word-components affects word recognition irrespective of scripts.
