The visual span for letter recognition in japanese and american subjects

Purpose: We have investigated the cultural difference in flanked-letter recognition by the Americans and the Japanese subjects. Method: We measured percent correct letter identification as a function of three factors: 1) Eccentricity, subjects fixated a central point and the letters were presented at one of 9 locations along the horizontal meridian (0, ±2.5, ±5.0, ±7.5,±10°). 2)Character sets, 25 letters were used from each of upper-case Times-Roman, Hiragana(syllabic letters) and Kanji(ideograms) of the Japanese. Character width was 26 arc min. 3)Native language, 5 subjects were native English speaking American students, and 5 were Japanese scholars or students studying in the USA for 1 or 2 years. Three flanking conditions were used: no mask, horizontal masks(a flanking pattern// was presented on both sides of the target letter), and vertical masks. Results: We defined the visual span to be the horizontal extent within which letter recognition was z50% correct. With no flanking masks, visual span(expressed in letter width) for Japanese subjects were 43 for Times, 40 for Hiragana, and 31 for Kanji. For American subjects visual spans were 42 (Times), 33 (Hiragana), and 23 (Kanji). With flanking, the visual span for the Japanese observers was reduced by a factor of 2 for all character sets. For American subjects flanking reduced the visual span by a factor of 3. Conclusion: Compared to American subjects, the Japanese readers are less susceptible to crowding effects caused by flanking. Japanese characters typically involve important details at high spatial frequencies. Over a life-time of experience, the letterrecognition mechanism may become tuned to respond to the details in Japanese characters(for example, shifting to a higher spatial-frequency band).