Notes :The root is only represented in SDR and Telugu. Bongard-Levin & Gurov  have pointed out the phonetic similarity to *var- 'paddy', which is supposed to be a borrowing from an Austronesian source. However, a development like *v- > 0- is impossible in Dravidian, and no Austronesian languages have a zero anlaut when this particular root (or any other with the meaning 'rice', for that matter) is concerned. This does not exclude the possibility of borrowing altogether, but leaves us without any clues as to the possible source. // Present in 2/6 branches.
Notes :Cf. also Kuwi (Fitzgerald) aṛvū, (Schulze) aruwu 'loan' (< Tel.). The root is hardly of Proto-Dravidian origin; I suppose that it might be etymologically identical with *ari 'rice' (= 'means of payment'). // Present in 2/6 branches.
Notes :The Telugu-PKG forms, on one hand, and the Gondwan forms, on the other, are somewhat hard to reconciliate. PGn shows a bisyllabic root (not necessarily with an auslaut -i-, since the latter is only tentatively set up based on forms like Kuwi rī-ali, etc.; in reality this could have been any vowel but a labialised one) with a vocalic auslaut, whereas Telugu and PKG reflect an older variant like *arál-. Of course, if we assume the suffixal nature of the resonant, the question is resolved, although the meaning of the suffix is still unclear. No traces of the root are seen in SDR, where the main word for 'fall' is *vīẓ- - which, in its turn, has been quite poorly preserved outside of SDR. Some overlapping, however, exists, which leads us to surmise that the difference between the two roots in PDR must have been stylistic - *arí-/*arál- being the neutral, unmarked 'fall', and *vīẓ- being its "Magn" correlate, i. e. "to topple over, crumble", whence the figurative development > "die" in some languages. In SDR both roots eventually converged into one. // Present in 3/6 branches.
Notes :Not represented outside SDR and Telugu. The final resonant may be an old suffix, but this is hard to prove (the connection with Tamil ārvam, supposed by Burrow & Emeneau, is quite questionable). // Present in 2/6 branches.
Notes :A NDR-only root; the tentative match with Tulu arumbu is possible, but hard to accept because of different suffixation and the isolated character of the word in Tulu. Phonetically the NDR form is also a good match for *arú- 'to approach', but the semantics raises certain doubts. // Present in 2/6 branches.
Notes :The development *-ai > -i (as in Kan. and Tel.) is probably due to the influence of the inlaut sibilant fricative. PKG *aj- is a rather weak match; however, semantically 'sweep' certainly can be traced back to 'stir', and phonetically PDR *asái-k- > pre-PKG *as-k- > PKG *aj-k- is a possible development considering that we do not know the exact behaviour of PDR *-s- in such an environment. // Present in 4/6 branches.
Notes :The irregular correspondence PSDR *-r_- - Tel. -l- in the 2nd syllable is not unique (cf., for instance, PDR *pesaL- 'green gram'). The monosyllabic version of the root may actually not belong here - or might be an archaism. // Present in 2/6 branches.