Reconstructions

Note

Also Tagalog poʔ ‘honorific, Sir, Madam’. This comparison is complex in a number of respects. First, although it is the most common, it is only one of four stem variants: *ampu, *empu, *impu, and *umpu. Second, an unusual number of affixed forms can be reconstructed next to the simple base. Some of these, as *empu-ŋ, *empu-q, and *t-empu evidently involved distinctions of vocative/address and referential forms (Blust 1979). Another, *empu-ni-a, involves what appears initially to be a possessed kinship term (cp. *ama-ni-a ‘his/her father’) which on closer inspection turns out to function very differently from other kinship terms followed by the genitive marker *ni and the 3sg. object pronoun *-a. The common reflex of *-i ‘local transitive’ in Malay mem-pu-ña-i and Makassarese punnà-i ‘have, possess’, suggests an etymon *empu-ni-a-i. However, since *empu-ni-a evidently meant ‘its owner’ (lit. "owner-of-it"), the Malay and Makassarese constructions in -/i/ must have arisen after the reanalysis of *empu-ni-a as a verb *empuña, and hence is a product of independent parallel changes. Likewise, the morphological similarity of Malay per-empu-an ‘woman; feminine; womanly (courtly, and in titles)’ and Mongondow pog-ompu-on ‘children and grandchildren’ appears to have ho historical basis (cp. Malay (Jakarta) perempuan ‘woman’, with last-syllable vowel indicating *-an, not *-en).

Perhaps most problematic are the forms assigned to *maki-empu and *paki-empu. With regard to the first of these Mongondow clearly reflects a prefix *maki- which remains a functional (non-fossilized) affix in such languages as Ilokano (/maki/- ‘prefix used to indicate that something is done with others, etc.’), Tagalog (/maki/- ask for or make a request for; join in company; imitate’) and Murut (Timugon) (/maki/- ‘petitive’; also /amaʔ/ ‘father’:/maki-amaʔ/ ‘subject calls object "father" ‘, cited in Prentice 1971:135). Since the Mongondow form is supported by the parallel but unglossed Itbayaten form machi-apo, cited under apo ‘GP/GC’ (reflecting the variant *ampu), the total comparison appears to be explained most simply by assuming loss of the last prefixal vowel independently in POc *mokompu ‘grandchild’ (with regressive assimilation of the remaining first syllable vowel), and in such WMP forms as Bare'e makumpu opu ‘GGC’ and Dampelas maʔupu ‘GC’. Foreshortened forms such as Karo Batak k-empu ‘GC’, Rejang k-epew ‘GC’ and Uma k-umpu ‘GC’ can be assigned arbitrarily to forms that begin either with *maki- or with *paki-. Semantically, *empu and its variants also present a complex picture.

Among relatively uncontroversial PMP meanings for all four variants are: (1) GP/GC (recipr.), (2) ancestor, and (3) lord, master, owner. Meaning (1) places *empu among the set of kinship terms, but it is the only reciprocal term that can be reconstructed for PMP. Meaning (2) appears to underlie references to totemism in Manggarai empo, Numfor k-epu, and perhaps in some other languages (see below), while meaning (3) probably underlies at least two seemingly unrelated semantic developments: the meaning ‘parent-in-law’ and the meaning ‘crocodile’. Reflexes of *empu or one of its variants mean ‘parent-in-law’ in such widely separated languages as Ngaju Dayak ((h)empo), Sangir (empo), Kédang (epu), Tolai (tabu-na), and Kilivila (tabu). Among the Kédang, who practice matrilateral cross-cousin marriage, the term also refers to the mother's brother, and among the Kilivila, who practice patrilateral cross-cousin marriage, it refers to the father's sister.

I have found no generally accepted explanation for the grouping of terms for grandparents and parents-in-law under reflexes of the same etymon in Austronesian languages, but the alignment follows naturally from a hypothesis that PMP speakers practiced matrilateral cross-cousin marriage, since under such an arrangement the wife-giving group is universally regarded as ‘lord, master, owner’ of the wife-taking group (Blust 1993, and references therein). The term *empu (and its variants) in meaning (3), then, could have acquired its affinal sense through its use as a title reflecting the higher status of the descent group represented by the MB/WF as against that of the ZS/DH. The recurrent association of grandparents or ancestors with crocodiles and/or other large, dangerous animals is seen in e.g. Tiruray bébéʔ ‘GP/GC; kind of large, black, extremely poisonous snake, considered by the Tiruray to be the original kind of snake, and to have been born of human parents as a twin to a Tiruray child’, Manggarai empo ‘nickname for the crocodile’ (in West Manggarai = ‘crocodile’), Rembong empoʔ ‘GP/GC; crocodile’, Kedayan bapa ‘term of address for the category epu; crocodile’ (Barnes 1974:246-47), and in general anthropological references to the Berawan (Metcalf 1982:254, 257), Malagasy (Frazer 1960:601ff), Mentawai (Loeb 1972 [1935]:166), Atoni (Schulte Nordholt 1971:323), where the crocodile is regarded as the ancestor of the ruler of Kupang), and various societies of the southeastern Solomons (Codrington 1891:178ff). Such widespread associations may be products of convergent development, but are more likely to reflect totemic ideas that were present in PMP social organization, together with a mixture of fear and respect for the crocodile as the ‘lord, master, owner’ of the all-important rivers around which Austronesian speakers so often settled. Finally, the semantic connection of Sasak peŋ-empo-n with Old Javanese paŋ-empw-an is clarified by the discussion of the ‘rice mother’ in Frazer (1960:479-87).