This work presents and defends the hypothesis that all wordbases in Proto-Austronesian and its early descendants were coined exclusively from CVC morphemes.

All wordbases in the most common form, CVCVC, that have been analysed otherwise by other writers are here analysed as having been coined by merger of two CVC morphemes, i.e. by overlap of the final consonant of the first and the initial consonant of the second so that those two (nearly-)identical consonants are expressed as one. It is seen that in every case each of the two morphemes thus identified is well evidenced also in other wordbases. It is also seen that there are only a few canonical forms other than CVCVC, each produced by a respective simple coining method. Also identified are various phonological processes that deleted or modified phonemes.

An appendix provides analyses of over 3,700 reconstructed wordbases and some 800 attested ones. Another appendix indexes the reconstructions, showing the published source(s) for each; it thus serves as a dictionary which in itself is an invaluable resource for research. Yet another appendix lists the more than 200 CVC morphemes that figure in the analyses, and groups them into cognate sets.