Odlin (1989) defines linguistic transfer as ‘the influence resulting from similarities and differences between the target language and any other language that has been previously acquired.’ ‘Reverse transfer’ happens when the direction of linguistic transfer is the reverse (Cook, 2003). In the field of Third Language Acquisition (TLA), language transfer could be more complicated than that of SLA because it involves three different languages. Different studies have already investigated L1/L2/L3 transfer factors (e.g. Cenoz et al., 2001; Hammarberg, 2009; Amaro, 2012) and it has been widely shown that L2 is preferred to L1 as a transfer source in the process of learning L3. However, most of these previous researches were performed on the transfers among European languages, with English being one of them, from the viewpoint of learners’ metalinguistic awareness of L1 and L2 syntax and lexicon.

This study attempts to ‘visualize’ cross-linguistic transfer among L1, L2, and L3 as well as to test the validity of Jim Cummins’s (1980) Cognitive Academic Language Proficiency and Common Underlying Proficiency (CUP) in L3 Japanese acquisition. Specifically, third/forth-language Japanese learners at the University of Tsukuba are asked to participate in a language background survey, reading/writing tests, and follow-up interviews. By examining the unconscious structure of L1/L2/L3 transfer through KH Coder network analysis and Naito’s PAC analysis, the study will also try to contribute to the development of new perspectives and strategies for forthcoming Japanese-language teaching and learning in the world.