As a follow-up to my earlier post, Heinz Fähnrich‘s Kartwelisches Etymologisches Worterbuch (2007, p. 119-120), which had existed in Georgian for 15 years before being translated into the more accessible German, contains an interesting etymological analysis of the Svan terms for ‘man’s sister’ and ‘woman’s sister’.
Svan is unique among Kartvelian languages in having special terms to denote siblings depending on Ego Gender. The pattern of contrasting man’s brother, woman’s brother, man’s sister, woman’s sister is very rare cross-linguistically. Unlike Klimov, Fähnrich sees the same root da– in both Svan u-d-il ‘woman’s sister’ (*udild, with the diminutive *-ild) and da-chwir ‘man’s sister’ (also in da-j ‘husband’s sister’). According to Klimov, the more specific meaning ‘woman’s sister’ found in Svan is original to the generalized meaning ‘sister’ found in the other Kartvelian languages. Another apparent archaism is the presence of the “frozen” obligatory possessive affix u– in the Svan form. We don’t know what the formant –chwir in ‘man’s sister’ comes from.
Since cross-linguistically (see The Genius of Kinship) sibling sets tend to lose semantic distinctions, rather than gain them, Svan must have lost the original term for either ‘man’s sister’ or ‘woman’s sister’. Root *da– came to replace it but it’s unclear which term is a survivor and which one is a replacement.
It’s possible that the underlying root *da– is further related to Kartvelian *ded- ‘mother, grandmother, woman, wife, mother-in-law’, which Fähnrich (pp. 128-129) presents in the following way:
If the Kartvelian term for ‘mother’ in its unreduplicated form is the ultimate source for the Svan terms for ‘man’s sister’, ‘woman’s sister’ and ‘husband’s sister’, then it’s easy to understand how the original independent stems for ‘man’s sister’ and ‘woman’s sister’ got replaced by a single-stem term. The system pulled a female term lacking Ego Gender semantics from the upper generation to create a new Ego-Gender-neutral form in the Ego generation. The morphology of u-d-il vs. the morphology of da-chwir are so different that it suggests that, since the replacement of the original terms for ‘man’s sister’, both u-d-il and da-chwir have undergone significant changes.
A close semantic parallel to the Kartvelian situation is furnished by Indo-European. Lith mosha ‘husband’s sister’ is a diminutive derivative of mote ‘mother’, while Albanian motre ‘sister’ goes back to IE *meH2ter ‘mother’. In the case of Albanian, it’s likely that the lowering of the ‘mother’ form to denote ‘sister’ (also Alb vella ‘brother’ comes from *awentlo-, which is found in Lat avus ‘grandfather’, Lat avunculus, Lith avynas and Breton eontr ‘mother’s brother’, see Huld, Martin E. 1984. Basic Albanian Etymologies. Columbus.) springs from an Omaha-type skewing in Proto-Indo-European (PIE) associated with patrilineal social organization. As part of this generational skewing, PF = MB = MBS and M = MZ = MBD. A later shift to ‘Hawaiian” or Generational nomenclature in Ego generation resulted in the form *awentlo– to acquire the meaning ‘mother’s brother’s son, mother’s sister’s son’, brother’. Finally, the shift from Generational to Lineal terminology led to the narrowing of the MBS = MZS = B semantic cluster to just refer to ‘brother’ (historical vella). It’s possible that Kartvelian has undergone a similar change and that the cognation of *ded– ‘mother’, *u-d-il ‘woman’s sister’ and da-chwir ‘man’s sister’ indicate that Proto-Kartvelian had an Omaha-type generational skewing, too.
Another Eurasian language that has the same 4-way division of sibling categories is Burushaski. But the neutralization woman’s sister ~ man’s sister in Svan is different from the neutralization found in Bur –co ‘a man’s brother; a woman’s sister’. In the Svan case, Ego Gender is neutralized, in the Burushaski case Referent Gender is neutralized.
Finally, Basque is the third language in Eurasia that displays the same pattern of classifying siblings. In all the Basque dialects but Biscayan (anaya mB, arrabea mZ, neba wB, aiztia wZ) Ego Gender is neutralized in the term for ‘brother’, so that Labourdin, Navarese, Guipuzcoan, Souletin have anay B, arreba mZ, ahizpa wZ.
While the three languages modify the original 4-term set in three different ways and the specific lexical forms don’t seem to be related, it’s possible that the rare sibling terminology attests to an ancient macrophylic unity underlying West Nostratic and West Dene-Caucasian.