This study is to (1) investigate the effects (if any) of L2 acquisition age, the length of exposure to L2, and gender on the bilingual dual coding hypothesis proposed by Paivio and Desrochers (1980) and (2) verify whether the bilingual dual coding effect in incidental recalls would be the same in Japanese-English as in Indo-European languages.
Balanced Japanese-English bilingual subjects are presented with (1) pictures to be labelled in English, (2) Japanese words to be translated into English, and (3) English words to be copied as they are. Later without warning they are tested to recall the generated English words.
The results showed a 3.7 : 3.2 : 1.0 ratio for pictorial : translation : copy encoding conditions, which is supportive of the bilingual dual coding hypothesis. Both small pictorial-translation difference and the high ratio for translation were interpreted as caused by the Japanese language specific effects-logographic features. No length of residence (LOR) or gender effects were observed. The onset age of L2 acquisition proved to be a significant factor, which added an extension to Arnedt and Gentile's (1986) ‘manner’-proficiencies of L1 and L2 prior to formal schooling as well as the language sequence in schooling (L1 to L2 or L2 to L1) should be considered in the bilingual dual coding framework. Thus, this experiment seems to reveal that the bilingual dual coding hypothesis is generalizable across (1) bilinguals not only in Indo-European alphabetic languages, but also in alphabetic/non-alphabetic languages, and (2) bilinguals in their childhood as well as adulthood.