This paper was written during my 1999 sabbatical leave at the University of Hong Kong supported by a grant from the Stanford Center for 'the Study of Chinese Language and Culture.' I first presented part of it at a colloquium (University of Hong Kong) and then at the 8th Annual Conference of the International Association of Chinese Linguistics (University of Melbourne, July 1999). The data of Oracle Bone Inscriptions and Bronze Inscriptions are mostly taken from Huang (1964). Other historical data used in this study are primarily taken from the Chinese Vernacular Corpus at the Academia Sinica in Taiwan. I