Linguistics in Japan has failed to develop corpus-based language studies into corpus linguistics, inspite of the long history of computer-based mathematical linguistics dated from the 1960s and sporadic contacts with English corpus linguistics since the 1980s. This is contrastive to the situation in Britain, where corpus linguistics has been established since the early 1980s, with grammatical and lexicological studies as main foci of interest. It is noteworthy that there is no Japanese corpus, available to researchers, which could be safely claimed as representative, so that researchers are now obliged to use a haphazardous collection of electronic texts as a corpus. Usefulness of such a corpus is evident, as is shown in a tentative case study, but inevitably limited. A representative corpus would serve better to linguistic research. The project of Balanced Corpus of Contemporary Written Japanese, now being undertaken by the National Institute for Japanese Language, is expected to fill the need and this is evidently welcome. It should be noted, however, that, in order to gain full advantage of a corpus, users will have to make efforts to acquire knowledge on techniques and basic facts in text processing.