Hatta, Katoh and Kirsner (1984) presented stimuli (Japanese words, foreign loan words and nonwords) written in Hiragana or Katakana and tested native English readers who have studied Japanese for several years and native Japanese speakers in a lexical decision task. A model of Japanese Kana processing was postulated to account for the data obtained. Two problems are discussed with respect to their paper. First, the conclusions drawn may not be consistent with the data analyses performed. Second, the rationale behind some of the processing assumptions of the model should be clarified.