Some Observations on Proto-Austronesian *t to k

1. Introduction.¹

This note provides information on another language that exhibits the PMP *t > k change discussed in Blust (2004), namely Makuva of East Timor, and suggests that this change (which also involves an apparent merger of original *k and *t) has not come about through substratum influence from the neighboring dominant non-AN language Fataluku. The author also takes issue with some misunderstandings of Caddoan data in the subsequent squib by Donohue (2006), and presents the correct facts.

2. PMP *T > K: Another Instance.

The recent squib by Mark Donohue (Donohue 2006) addresses some of the issues raised in the account provided by Blust (2004) of the change from PMP *t to /k/ in some 43 daughter languages. This change actually represents at least 20 historically independent instances of *t > k change, including three historically separate instances of this in Polynesian. Striking sound-changes in Austronesian are not infrequent, as the extensive treatment of ten such changes in Blust (2005) makes clear,² but the change of *t > k is unusually widespread within Austronesian, and the nature of its geographical and genetic distribution makes it clear that this change is not an exclusively shared innovation that serves to mark off a single cluster of related languages. Donohue points out that the reverse change, that from *k to t in the world's languages, is even rarer, though it does occur, for instance, in some young children's speech output.³

In this paper I wish to add an example of a further Austronesian language that exhibits the *t > k change, and also to correct some observations by Donohue on stop systems in the Caddoan languages of North America, in particular Pawnee.

The language that is to be added to Blust's collection as showing further examples of PMP *t > k is Makuva, formerly also known as Maku'a, Lóvaia, or Loikera. This is a Central Malayo-Polynesian language closely related to Meher and other languages of Kisar that is spoken in far eastern East Timor near Tutuala and Lospalos and that is arguably the easternmost indigenous language of East Timor. Speakers of Makuva (who in any case are dominant in Fataluku, a non-Austronesian language that may be a member of the Trans New Guinea phylum) are few, data are scarce, and Makuva data to which I have had access are scarcer yet. My own knowledge of the language is confined to the report in van Engelenhoven and Cailoru (2006), and to the set of cardinal numerals in Hull (2004).⁴ Indeed, van Engelenhoven and Cailoru (2006) suggest that Makuva, whatever its origins, is now functionally a language "in coma," one that is simply known to a minority of mature speakers of Fataluku as a ritual language for use on certain social occasions.

In Makuva, PMP k apparently remains as k at least in some cases (n-kaku 'afraid' < PMP *takut 'fear', ulke 'brain' < PMP *kulit 'skin') but t also becomes k (o-kelo '3' < PMP *telu, Kikuola for the place-name Tutuala),⁵ though with a tendency word-finally to drop altogether, as the shapes of the words for 'afraid' and 'brain' suggest. Furthermore, s becomes t (itetla '1' < PMP *isa/esa plus an affix, or perhaps modified by reduplication). The latter sound-change (PMP *s > t) is not rare in Austronesian and is parallel with the same change in Palauan (Josephs 1990), where, as has long been known, PMP *s became t (tut 'breast' from PMP *susu) and PMP *t became ð, written <d> in the modern orthography, except when it was preserved intact postconsonantally (so that both reflexes are presented in Palauan dakt [ðakt] 'afraid' < PMP *takut). Any idea of the change from Makuva alveolars to velars being the outcome of a Fataluku substrate acting upon Makuva is unfounded, because it is clear that the sounds k t s are all phonemically distinct in Fataluku. For instance, they all contrast before a in Lospalos Fataluku ikafa '8', ta'ane '10', sapa 'skin disease' (data from Fataluku Language Project 2006, a web resource which is directed by Aone van Engelenhoven).

3...