Abstract
Two experiments were designed to examine the way Chinese characters (Kanji) are identified. In the first experiment, subjects reported whether the stimulus was a correct character or a pseudo-character. The decision latencies were longer when parts of characters (Bushu) were rotated than when whole characters were rotated. If whole-shape cues of Chinese characters contribute to identification, distortions of these shapes by partial rotation should have negative effects on the recognizability of characters. The results of the first experiment supported this expectation. In the second experiment, subjects reported whether the stimulus was a two-letter-word or a nonword. In this case, the latencies when the whole stimulus was rotated did not differ from the latencies when only one character was rotated. These results suggested that identification of words written in Chinese characters did not rely on word-shape cues.
