Reconstructions

Note

Both Wolff (1973) and Dahl (1976) reconstruct PAn *-in- as a past tense marker which co-occurred with the active infix *-um-, the instrumental passive *Si- and the local passive *-an. In the direct passive the relationship of *-in- to the verb is more complex. Whereas the direct passive of non-past verbs was formed with *-en, the direct passive of past verbs was formed with no suffix. As a result the infix *-in- in the past tense of direct passive verbs acquired a kind of portmanteau function, signalling both the past tense and the passive voice inseparably. This portmanteau function of *-in- is widespread in Austronesian languages, and was noted early in the 20th century by the Dutch linguist Hendrik Kern (Kern 1913-28).

In contrast to Wolff and Dahl, Starosta, Pawley, and Reid (1982) argue that *-in-, along with other reconstructed affixes that function both as voice markers and as nominalizers in contemporary languages, were originally markers only of various types of nominalization. To them it follows, then, that the widespread verbal uses of these affixes are historically secondary.

In my view there is merit both in the position taken by Wolff and by Dahl (that the affixes in question were verbal inflections in PAn), and in the position taken by Starosta, Pawley, & Reid (that the same affixes were derivational devices for creating deverbal nouns). If there is a shortcoming that is shared by the two approaches it is perhaps the implicit assumption that the PAn affixes *Si-, *-in-, *-an, and *-en were exclusively verbal or nominal. In fact, straightforward application of the comparative method seems to require the reconstruction both of verbs and of nouns marked with the same set of affixes in PAn. This leaves open the question whether at some still earlier period one of these functions might have developed out of the other.

In PAn and PMP it is clear that at least pronominal agents (if not also nominal agents) and possessors were marked the same: e.g. PAn *k-um-aen aku ‘I am eating’, but *kaen-en (ni) ku ‘is eaten by me, my eating’, *k-in-aen (ni) ku ‘was eaten by me, my eating’, *maCa (ni) ku ‘my eye’. Given this identity of marking it would not be surprising for pattern pressure to favor the reinterpretation of non-AV verbal constructions as possessive nominals. Since no such pressure would favor the reinterpretation of possessive nominal constructions as non-AV verbals, I cannot agree with Starosta, Pawley, & Reid that the focus systems of modern Formosan and Philippine languages probably evolved from earlier systems of nominalization.