Abstract
This paper aims to test the general assumption that as a new concept is imported into another
culture, several translation equivalents co-exist but eventually converge into a few. Specifically, it traces how selected culture-specific terms have been translated, drawing on a bilingual corpus of the United States Constitution consisting of 86 Japanese translations dating from 1866 to 2015. These translated terms are then classified according to their goshu（etymological type, here native word, Chinese word, loanword or hybrid）to detect any patterns in the diachronic changes. Attention is also paid to the use of furigana（smaller syllabic characters printed alongside Chinese characters primarily to indicate their pronunciation）occasionally attached to the translated terms to indicate the transliterations. The results show that translated terms do indeed converge; more precisely, four patterns of change are identified in the examined corpus, which can be grouped broadly into those preferring Chinese or native words.
