(1) 読字課題では仮名は音節文字であるために機能的にも連鎖反応が容易で,音韻の符号化も単一処理ですむために,音訓両音を持つ漢字よりも音韻の転換速度が速いことがわかった。 (2) 再認課題では音韻にのみ依存する仮名は系列的に処理されるために,文字間の弁別が難しく,意味を持つ漢字にくらべて再認が悪いことが考察された。また,同じ漢字でも意味が手がかりとして有効に働かない時は仮名と同様に再認が悪くなった。 (3) 自由再生課題ではリストの長短にかかわらず,直後再生では仮名(化)群の方が漢字群よりも良く,遅延再生では逆に漢字群よりも良くなった。これは短期の記憶では音韻情報が有効に働き,音連鎖の方略が取りやすかったのに対し,長期の記憶では逆に意味情報が有効に働いたためと考えられる。また,同じ漢字でも音韻に依存して符号化する場合と意味も同時に符号化する場合とでは結果が異なることが考察された。

Japanese orthography is of two types, kana and kanji. Kana is phonogram, and Kanji is considered as idiogram. The present experiments were conducted with the purpose of examining differences of functions in regard to verbal materials in Japanese orthography. Verbal materials in experiments were of several single letters (kana and kanji), each corresponding to one-syllable sound ; kana corresponding to the used kanji was of one syllable (phonem) without a porticular meaning, and kanji was of one syllable and one meaning. The results of experiments were as follows. 1) In preliminary examination, the subjects were required to read a list of kana or kanji (100 letters) as fast as they could. It was found that reading kana was faster than kanji. The results suggested that as kana had just a single coding, there was linking of several letters into single unit or a string of the pronounceable letters, but as kanji had multicoding (phonetic and semantic), there was less linking. 2) EXP. I investigated which was easier to recognize, kana or kanji. A learning list of 10 letters was made together with a test list of 5 letters of old items and 5 letters of new items. In recognition task, kanji was found to be better than kana. The present studies suggested that semantic information was much more effective than phonemic information in achieving correct recognition, since more features were involved and each collection of features was highly distinctive. 3) In EXP. II, the study was to clarify the phonemic and semantic information processes in STM or LTM by free recall task. In method 1, material was of 6 lists of 10 letters (kana or kanji), and in method 2,4 lists of 16 letters. Free recall tasks consisted of the presence or absence of delay after presentation of stimulus. In immediate recall (STM), kana was better than kanji, but in delayed recall (LTM), kanji was better than kana. When paticular phonemic code in kanji was selected by instruction, kana and kanji had the same result. Results indicated that in STM, when the letters were encoded in terms of its phonemic features, it was better recalled than when it was encoded in terms of the semantic features ; in LTM, semantic information was better recalled than phonemic information. The present study suggested the importance of S's strategy in processing of letters' information.