Some new hypotheses concerning word families in Chinese
汉语词族的一些新假设
E. G. Pulleyblank 蒲立本

Abstract 摘要
It has long been apparent that Old Chinese had sets of words which were related in meaning and similar, but not identical, in sound. B. Karlgren conjectured that such ‘word families’ represented the relics of morphological processes but concluded that it was impossible to reconstruct what these processes had been. Advances in the reconstruction of Old Chinese allow us to see the nature of at least some of these processes. The best established affix is (a) the suffix *-s, which left its reflex in Middle Chinese as the departing tone. One may also recognize: (b) prefix -, cognate to Tibetan a-čhung, giving rise to alternations of voiced and voiceless obstruent initials, (c) prefix *s-, (d) prefix *r-. There were also (e) ablaut between close and open vocalic nucleous (ә/а), (f) alternation between accent on the first half or the second half of the syllable. More remote word family relationships, which cannot easily be accounted for by such morphological processes, may reflect a stage in which there were uniconsonantal root morphemes which could combine to form syllabic units. Comparisons are made to Northwest Caucasian and Indo-European.