Still on the Komuz thread: while looking into work on its western neighbors, I have accidentally run into what must be one of its more recent published defenses: M. Lionel Bender’s “The Limits of Omotic”, from the 1990 collection Omotic Language Studies (pub. Routledge, ed. Richard Hayward). This focuses on the classification of the Mao languages and Gumuz, both of which were at the time still poorly known — concluding that the former would be indeed Omotic but that Gumuz is not, and is instead akin to Koman.
The latter point builds on an earlier 1983 reconstruction of Proto-Koman also by Bender. Even without looking this up in full, it appears on the net clearly inferior to Otero’s. For one, he is still unaware of Dana (which contributes substantially to PK reconstruction, e.g. as the one language consistently distinguishing PK dental and alveolar stops), though keeps Gule data in the loop. Also aspiration, tone and the ±ATR distinction in high vowels, the last two of which condition some consonant shifts especially in Gwama, seem to have been disregarded or missed. From Gumuz he presents only vague “Common Gumuz” forms; again without tone, also without vowel length, and often clearly inconclusively phonologically analyzed. Occasional [ə ɔ ʊ] appear in forms like bəga ‘person’, žɔgwa ‘bone’, fʊčʼ- ‘to blow’ (Ahland’s ɓaga, ʒákwá, fwítʃʼ-); some forms are left with cover-letter vowels, e.g. mEta ‘bird’, kuwA ‘breast’ (Ahland: mátá, kúá); some forms show affricate/fricative or tenuis/ejective vacillation, e.g. (t)sʼea ‘ear’, ts(ʼ)ia ‘tail’ (Ahland: tsʼéa; SG tsia ~ NG tsía); the alveolar implosive ɗ is attested thruout, but labial ɓ seems to have gone amiss (cf. before in ‘person’). [1] All this naturally leaves also his comparative phonological analysis more vague than mine.
With a basic 300-word list for Gumuz, Bender ends up finding ca. 50 common Koman–Gumuz items, i.e. about the same number as I had so far too. The overlap in our lists is only 21 items though! Of the remainder, two are based only on Gule, not mainline Koman: Gule ufun ~ D mfoʔo, SG mfá, NG ofá (Bender: ʷofa) ‘ash’; Gule is ~ D íʃ, SG NG iʃa (Bender: iš) ‘egg’. 22 other items look good to me too and were simply not in the data available to me from Ahland and/or Otero: e.g. PK *kʰáɗ(a)- ~ Bender’s common Gumuz kɔɗɛs ‘open’; D mété, SG NG mátá ‘bird’ ~ Bender’s PK *mbit (no reflexes given; though I would predict that this is really *ɪ). I would only dispute or leave so far aside five of his comparisons:
- K *kʼɔlɔ (Otero), *kʼol- (Bender) ~ G *ela (> D éé, SG ela, NG éʔa; Bender: ela, era) ‘hand’: I see no correspondence here other than -l-. This is also one of a few items where Bender does not venture to present even any rough Proto-Komuz reconstruction. [2]
- K *kʼɔ/ʊʃ- (Otero), *kʼoš- (Bender) ~ G ʃákʼw- (Ahland), šokʼw- (Bender) ‘to kill’: requires metathesis and not as such very reliable for establishing a relationship (however accepted also in Ahland’s 2013 presentation, alongside several other similar metathetic comparisons).
- K *du (Bender) ~ D ndihi, SG andííá (Ahland), andiya (Bender) ‘liver’; no correspondence other than -d-.
- K *cwálá (Otero), *sua (Bender) ~ G *gʲá (> D gí, SG NG ɟá; Bender: gya, ǰa) ‘tree/wood’; some vague similarity, but already *c ~ *g is too far off from what other evidence shows so far.
- K *mbar (Bender) ~ G *bee-kʼwa (> D bekʼó, SG NG béékʼwá; Bender: bee-) ‘hair’; this Proto-Koman form seems to correspond most closely with Otero’s Komo–Uduk *mùr, while I would compare here instead Gwama bàk (plausibly < PK *bakʼ(a); coda deglottalization on Gwama is not entirely regular but has precedents, e.g. pák ‘shoe’ < PK *pʰákʼá > Opo pʰákʼ, pʰákʼá).
Altogether I do not think these issues undermine Bender’s conclusion of the Koman–Gumuz relationship too much either — he might draw this conclusion more weakly than I did, but still with good reason.
This seems to me to be a fairly typical situation for pioneering comparative work at medium-level separation. Poorer quality and quantity of base data makes reconstruction much less reliable — but as long as the primary units of comparison are themselves self-evident and phonologically reasonably compact units, the establishment of cognates remains almost as possible! There will be slightly elevated risks of false positives and false negatives, but already focusing on basic vocabulary and demanding exact or near-exact semantics will mostly weed out problems like adventurous speculation and external loanwords. This is after all what we find also in the research history of various long-established families like Uralic. The majority of the family-wide basic etymologies like *kala ‘fish’, *pesä ‘nest’, *ćilmä ‘eye’, *tälwä ‘winter’ are identified already in the “prescientific” period of comparison, long before the explicit demonstration of precise regular correspondences or their reconstruction in either Proto-Uralic (a task that in several cases still continues) or in the branch-specific proto-languages; the largely trivial basic correspondences that motivate them have mostly proven to be correct, and have been since then fortified with a still growing number of less trivial ones; and hence the “prescientific” work proves to have also been of real scientific value after all. Clearly not every proposal for a language relationship pans out or should be trusted, but here too, many basic signs are fairly clear if you know what to look for.
[1] However we do have in Daatsʼiin mété ‘bird’, tsʼíŋtsʼíŋ ‘tail’. I cannot guess if Bender’s first-pass Gumuz fieldwork may have covered also Daatsʼiin, or if forms similar to these exist also somewhere within the narrow Gumuz dialects.
[2] I wonder also (though Bender states nothing of the sort) if any ghost of “Nilo-Saharan movable k” has informed the comparison. — Amusingly similar also to the attempts of old Ural-Altaicists to relate Uralic *kätə ‘hand’ and Turkic *elig id., also eventually abandoned even by them for e.g. no parallels for *k- ~ zero.
Freelance reconstruction
Second postscript: Bender on Komuz
Still on the Komuz thread: while looking into work on its western neighbors, I have accidentally run into what must be one of its more recent published defenses: M. Lionel Bender’s “The Limits of Omotic”, from the 1990 collection Omotic Language Studies (pub. Routledge, ed. Richard Hayward). This focuses on the classification of the Mao languages and Gumuz, both of which were at the time still poorly known — concluding that the former would be indeed Omotic but that Gumuz is not, and is instead akin to Koman.
The latter point builds on an earlier 1983 reconstruction of Proto-Koman also by Bender. Even without looking this up in full, it appears on the net clearly inferior to Otero’s. For one, he is still unaware of Dana (which contributes substantially to PK reconstruction, e.g. as the one language consistently distinguishing PK dental and alveolar stops), though keeps Gule data in the loop. Also aspiration, tone and the ±ATR distinction in high vowels, the last two of which condition some consonant shifts especially in Gwama, seem to have been disregarded or missed. From Gumuz he presents only vague “Common Gumuz” forms; again without tone, also without vowel length, and often clearly inconclusively phonologically analyzed. Occasional [ə ɔ ʊ] appear in forms like bəga ‘person’, žɔgwa ‘bone’, fʊčʼ- ‘to blow’ (Ahland’s ɓaga, ʒákwá, fwítʃʼ-); some forms are left with cover-letter vowels, e.g. mEta ‘bird’, kuwA ‘breast’ (Ahland: mátá, kúá); some forms show affricate/fricative or tenuis/ejective vacillation, e.g. (t)sʼea ‘ear’, ts(ʼ)ia ‘tail’ (Ahland: tsʼéa; SG tsia ~ NG tsía); the alveolar implosive ɗ is attested thruout, but labial ɓ seems to have gone amiss (cf. before in ‘person’). [1] All this naturally leaves also his comparative phonological analysis more vague than mine.
With a basic 300-word list for Gumuz, Bender ends up finding ca. 50 common Koman–Gumuz items, i.e. about the same number as I had so far too. The overlap in our lists is only 21 items though! Of the remainder, two are based only on Gule, not mainline Koman: Gule ufun ~ D mfoʔo, SG mfá, NG ofá (Bender: ʷofa) ‘ash’; Gule is ~ D íʃ, SG NG iʃa (Bender: iš) ‘egg’. 22 other items look good to me too and were simply not in the data available to me from Ahland and/or Otero: e.g. PK *kʰáɗ(a)- ~ Bender’s common Gumuz kɔɗɛs ‘open’; D mété, SG NG mátá ‘bird’ ~ Bender’s PK *mbit (no reflexes given; though I would predict that this is really *ɪ). I would only dispute or leave so far aside five of his comparisons:
Altogether I do not think these issues undermine Bender’s conclusion of the Koman–Gumuz relationship too much either — he might draw this conclusion more weakly than I did, but still with good reason.
This seems to me to be a fairly typical situation for pioneering comparative work at medium-level separation. Poorer quality and quantity of base data makes reconstruction much less reliable — but as long as the primary units of comparison are themselves self-evident and phonologically reasonably compact units, the establishment of cognates remains almost as possible! There will be slightly elevated risks of false positives and false negatives, but already focusing on basic vocabulary and demanding exact or near-exact semantics will mostly weed out problems like adventurous speculation and external loanwords. This is after all what we find also in the research history of various long-established families like Uralic. The majority of the family-wide basic etymologies like *kala ‘fish’, *pesä ‘nest’, *ćilmä ‘eye’, *tälwä ‘winter’ are identified already in the “prescientific” period of comparison, long before the explicit demonstration of precise regular correspondences or their reconstruction in either Proto-Uralic (a task that in several cases still continues) or in the branch-specific proto-languages; the largely trivial basic correspondences that motivate them have mostly proven to be correct, and have been since then fortified with a still growing number of less trivial ones; and hence the “prescientific” work proves to have also been of real scientific value after all. Clearly not every proposal for a language relationship pans out or should be trusted, but here too, many basic signs are fairly clear if you know what to look for.
[1] However we do have in Daatsʼiin mété ‘bird’, tsʼíŋtsʼíŋ ‘tail’. I cannot guess if Bender’s first-pass Gumuz fieldwork may have covered also Daatsʼiin, or if forms similar to these exist also somewhere within the narrow Gumuz dialects.
[2] I wonder also (though Bender states nothing of the sort) if any ghost of “Nilo-Saharan movable k” has informed the comparison. — Amusingly similar also to the attempts of old Ural-Altaicists to relate Uralic *kätə ‘hand’ and Turkic *elig id., also eventually abandoned even by them for e.g. no parallels for *k- ~ zero.
Related
Posted in Commentary