In this paper the contribution or impact of sub-word levels in the computation of word phonology was evaluated for the two very different orthographies of alphabetic English and logographic Japanese kanji. In particular, the studies of Wydell, Butterworth and Patterson (1995), Fushimi, Ijuin and Tatsumi (1996a, b), and Wydell, Butterworth, Shibahara and Zorzi (1997) are reviewed. These studies investigated the sub-word level (i.e., the level of constituent characters in two-character kanji words) as well as whole-word level of contribution in the computation of phonology to the compound kanji words. It is concluded that the available data suggest some involvement of sub-word level processing in the computation of word phonology in kanji, though to a much lesser extent than in English. More importantly, it is suggested that the structural differences between On-reading words (of Chinese origin) and Kun-reading words (of Japanese origin) may prove to be important factors when evaluating the speed and accuracy in the computation of the phonology of kanji words.