Reconstructions

Note

Also Bontok etéy ‘death’, pátay ‘any sacrifice performed on behalf of the village’, Aklanon ka-mátay-on ‘death’, Manggarai mate ‘the late, the deceased’, Ngadha mate ‘die; dead; the deceased; kill, slay’, Lamaholot maté-n ‘already dead, not moving any more’, Fordata matta ‘die; dead; go out (fire, light)’, Kowiai/Koiwai matata ‘corpse’, Marshallese mij ‘dead; numb; death’, Niue mase ‘die’. Since all of the forms that Dempwolff compared contain a reflex either of *p- or of *m-, he reconstructed both *patay and *matay ‘die, dead’. However, rather than list these reconstructions with the relevant supporting evidence separately, as was his practice with doublets, he posited paired etyma *patay, *matay at the head of a single mixed comparison, with an empty cross-reference to *matay. The explanation for this treatment is given in Dempwolff (1934-38:sect. 47e), where it is suggested that *matay is derived from *patay by nasal substitution, but is sufficiently widespread (being found in ‘Indonesian’, ‘Melanesian’ and Polynesian languages) to require a separate reconstruction. This was perhaps the closest Dempwolff came to the reconstruction of morphology.

The evidence now available shows that the unaffixed PAn base in question was *aCay. Dempwolff's *patay and *matay can most convincingly be derived from *pa-aCay ‘kill’ (with causative *pa-) and *ma-aCay ‘dead’ (with stative *ma-). That the base began with a vowel rather than *q- is clear from such reflexes as Bunun atađ ‘kill’, Kalamian Tagbanwa patay ‘dead’, Moken matay ‘die; dead’, Muna mate ‘die; dead’, Lou mat ‘die; dead’, and Tongan mate ‘die; dead’. The merger of *q- and zero in many attested languages would have led to homophony between reflexes of PAn *aCay ‘death’ and *qaCay ‘liver’, both of which figured in numerous expressions relating to the emotions. Possibly to avoid such potentially awkward homophony the unaffixed base was lost in the great majority of languages, leaving reflexes of *pa-aCay and *ma-aCay as effective secondary base forms for further morphological elaboration. Thus, next to PAn *pa-aCay and PMP *pa-atay ‘kill’ there is a priori evidence for e.g. PMP *pa(ka)-m-atay ‘kill’ and PWMP *ka-m-atay-an ‘affected by a death’, where the outer affixation evidently was added to an already prefixed base.

The alternative hypothesis, that forms such as Bunun atađ ‘kill’, Casiguran Dumagat atë ‘die; dead; death’ or Kelabit ate are products of analogical back-formation from a base *paCay or *maCay is more difficult to maintain in view of the fairly clear (though by no means universal) association of reflexes of the former with the causative ‘kill’ and of reflexes of the latter with the stative ‘dead’. Partly by analogy with *ma-qudip ‘living, alive’, I have attributed all reflexes that Dempwolff would have assigned to *matay to PAn *ma-aCay, PMP *ma-atay, PCEMP *matay, POc *mate. It is possible that these forms actually represent a conflation of *ma-aCay ‘dead’ (stative) and *um-aCay ‘die’ (active), since the initial vowel in reflexes of *in-atay generally disappeared even though it was retained in most other forms, and we would expect a parallel treatment of the initial vowel of *um-atay. Nonetheless, pairs such as Kelabit mate ‘die; dead’, in-ate ‘killed’ suggest that the former does contain a reflex of *ma- rather than *um-. The same apparently must be assumed in PMP *m-in-atay ‘person who has died, corpse’ (= *m-in-a-atay, with contraction of the sequence of like vowels), despite the fact that the co-occurrence of *ma- ‘stative’ and *-in- ‘perfective’ is otherwise rare.

Because of occasional ambiguities in the morphology it is possible that I have posited some morphologically complex forms of *aCay and its reflexes which are in reality products of convergent innovation. Examples include *m-atay-an ‘lose someone through death’, *paŋ-m-atay ‘means of killing(?)’, *ka-p-atay-an ‘end of death of something(?)’ and *ka-p-atay-en ‘death(?)’. As implied in passing above, it appears likely that the prefix of PAn *ma-aCay was reinterpreted as part of the stem in PCEMP *matay. The only known counterevidence to this claim is Bimanese h-ade ‘kill, extinguish a fire’ and Soboyo n-ate ‘2sg, 3sg and 3pl form of m-ate ‘die’. Finally, although Dempwolff reconstructed both *patay and *matay with the meaning ‘die, dead’ it is clear from the data assembled here that by at least PMP times a far richer complex of meanings was associated with *atay, whether alone, affixed, or in combination with various other free morphemes. These meanings included not only ‘death’, ‘dead’, ‘die’, ‘kill’, and ‘corpse’ but also the following more figurative senses: