Notes :SDR forms, on one hand, and Telugu, PKG, and PGn forms, on the other, can hardly be traced back to the same stem; the former look like quadrisyllabic compounds (*aḍí-KarV, with the vowel in a weak position) whereas the latter can only go back to *áṭikV, a trisyllabic stem with the vowel in a strong position and a nominal suffix tacked on at the end. Nevertheless, if Burrow & Emeneau are correct about comparing the forms to PDR *áṭ- ~ *aḍú- "to cook", the two variants may indeed share the same archaic root. // Present in 5/6 branches.
Proto-Dravidian :*átai
Meaning :female relative (father's sister; grandmother)
Notes :The meaning 'father' (somewhat expectable from a root like ata-) is actually present in Tamil and Malayalam, but nowhere else in Dravidian. The archaicness of this meaning in Tamil cannot be excluded; more probable, however, is the idea that in Tamil the word has developed a parallel masculine stem by analogy with *ápa- q. v. (the latter does have 'masculine' meanings outside of Tamil). // Present in 4/6 branches.
Notes :A clear distinction between the two stems can be demonstrated on Kannada evidence: aḍu = "to cook, boil", aṭṭu = "to evaporate, dry up". However, in Central Dravidian languages only the monosyllabic variant has been preserved. It is not excluded that the two variants represent originally different roots; a final decision cannot be reached without external comparison. // Present in 4/6 branches.
Proto-Dravidian :*áṭa-
Meaning :inner side of roof (loft, attic, supporting bamboo framework etc.)
Notes :While some of the forms adduced by Burrow & Emeneau may look like later reborrowings, the root as a whole is widely represented in Dravidian and is most probably PDR in origin. The corresponding IA root *aṭṭa- "high" (Turner No. 180) has no Indo-European etymology and must have been borrowed from an early Dravidian dialect. // Present in 5/6 branches.
Notes :The stem is occasionally used with pronominal prefixes (Tamil, Gondi), but their usage is too localised to suggest that the situation was the same in PDR. The original meaning of the root is hard to establish; it could hardly have been simply 'mother' (as there already are two basic terms for 'mother' in PDR - the honorific *áma- and the neutral *aj-). More often than not, the semantics of the stem is closely linked to the idea of old age or being elder (cf. 'old woman'; 'grandmother'; 'matron'; 'elder brother's wife'). The relations of this root with PDR *átai are also unclear. I would venture to guess that the original differentiation is seen in Kolami-Gadba, where *átai = 'father's sister', whereas *ávai = 'father's mother', but this is very flimsy evidence. // Present in 4/6 branches.
Notes :The variation in close, but different meanings suggests that the root may have been open to semantic contamination with several others (cf. in particular Tamil forms with initial n- which are contaminated phonetically as well). Konḍa namb- 'to cool down (fire)' is given by Burrow & Emeneau as a questionable cognate, but is certainly a different root (there is no way whatsoever that it could correspond to Gondi aviyānā 'to be tired'!). // Present in 3/6 branches.
Notes :The connection of the SDR and NDR roots (separated in DEDR) is possible if Malto -w- is primary and goes back to *-b- < *-v-. // Present in 2/6 branches.
Notes :Inlaut *-v- is primary in this root despite DED (Tamil aviṛ hardly < *agiṛ, whereas in Kannada and Telugu such developments are common). Etymology unclear; maybe a connection can be seen with *aví- "to press, subdue, diminish'? // Present in 2/6 branches.
Notes :A highly irregular (and therefore) dubious Central Dravidian-NDR isogloss. The only way to make all these forms fit together is to assume their descent from an old complex stem like *aẓ[V]-d- with non-trivial cluster development in daughter languages. However, if that is really so, the forms also beg comparison with PDR *aẓ[ú]nd- 'to press down; to dip, immerse' ('immerse' > 'pervade'). // Present in 3/6 branches.
Notes :Cf., perhaps, PGn *eṛi- 'to burn'. Burrow & Emeneau unhesitatingly join the Gondwan forms with the SDR-Telugu item in one entry, but if the root vowel in PGn has indeed to be reconstructed as *e (based primarily on Gondi reflexation, since all the other forms are syncopated), the etymology becomes somewhat problematic. The Gondwan forms themselves are questionable inasmuch as every subgroup has its own derivational suffix. // Present in 3/6 branches.
Notes :Gadba addap- is quite probably a borrowing from Telugu, which makes the stem reconstructible on the SDR-Telugu level. See also notes on PDR *aẓ- 'to pervade (as taste)'. // Present in 3/6 branches.
