Notes :The most common, and most certainly, primary meaning of the stem is 'mother'; however, much like *ápa-, this is the "formal" root for this relative, and is in that function opposed to *aj- q. v. // Present in 5/6 branches.
Meaning :to fit together; to be useful; to prepare
Notes :Outside SDR and Telugu only found in Kuwi (ambrinai 'to suit'), where it is most likely borrowed from Telugu (unless other parallels in Central Dravidian eventually show up). The root is completely homonymous with *amai- 'to be calm, settled', right down to the same opposition of the -ai-stem and the -r-stem; however, it does not seem prudent to join the two under one entry, although occasional semantic contamination may be present. A particular problem is the root's relationship with *sav- (*sam-?) 'to be ready'; I find it hard to trace both variants to a single PDR root and would suggest phonetic contamination (i. e. *sav- > *sam- in many dialects under the influence of *amai-) between two roots with similar meanings. // Present in 2/6 branches.
Notes :A rather local SDR root but certainly quite "Dravidian" in form. Burrow & Emeneau compare it with *am- 'to be pressed down, crushed' (> 'to settle'), but I would rather opt for homonymity here (occasional semantic contamination with *amai- 'to fit together' is also possible). As always it is possible to derive *ama-r- from *amai-r-. // Present in 1/6 branches.
Notes :Forms like Tamil appu-kkaṭṭu 'sheaf of arrows'; Telugu ampar_a (pl.) 'arrows'; and possibly Gondwan plural forms in ap- hint at the PDR status of the morphological gradation *amb- (sg. stem) ~ *amp- (plural and derivative forms). // Present in 4/6 branches.
Notes :SDR displays a compound with *kāl- 'leg, foot', but the Telugu form is apparently simple. In DEDR the items are tentatively compared with Kur. umbāxārnā 'to walk on hands and knees', but vocalic correspondences are impossible. // Present in 2/6 branches.
Notes :The presence of the root in North Dravidian clearly validates its PDR status (otherwise one could argue that all Central Dravidian languages - as well as some South Dravidian ones - have borrowed the form from Telugu). // Present in 5/6 branches.
Notes :A common PDR stem commonly displaying two variants, with *anḍ- being the exclusive one in all of Central Dravidian and *aṇái- safely reconstructed only for PSDR (and Telugu, if anãgu 'to mix' really belongs here). Some of the forms with -ṇṭ- may be due to contamination with *anṭ- 'to stick'. The two stems, bisyllabic and monosyllabic with retroflex cluster, may well be related on a higher level (with the original root being *aṇ-, which then may have been preserved as an archaism in Tam. aṇṇu), but separate origins cannot be excluded either. // Present in 6/6 branches.
Notes :Cf. also Betul Gondi an-c-ānā "to press". Considering that one of the meanings in Tamil is "be slain", it is possible to suggest historical derivation from P(S)DR *aṇai- 'to extinguish, die' (in which case the Gondi match hardly fits, but it is an isolated one-dialect form anyway). In DEDR the Kolami-Gadba forms are grouped with PDR *aẓí- 'to perish, be ruined', but the nasal consonant in PKG is too well defined to ignore; the current etymology seems to be just as plausible semantically and better phonetically. // Present in 3/6 branches.
Notes :Compared with Old Indian añcala- "edge" (< añc- "to bend, etc."), but it is strange that the resonant suffix is not found in Dravidian. Either the Dravidian forms reflect a different verbal derivative that has not been preserved in Indo-Aryan or the resemblance is accidental. // Present in 2/6 branches.
Notes :Despite being represented fairly well in SDR and even having possible Nostratic links, the root's position within Dravidian is not very strong since there exists a strong possibility of its being borrowed from IA: cf. Hindi ā̃ṭhī 'kernel or pip of fruit', Bengali, Oriya ā̃ṭhi id., etc. < IA *aṣṭi-; Turner 955). Definitely borrowed from Oriya are the NDR forms (Kur. aṭhū, Mlt. aṭi, as evidenced by the aspirated retroflex in Kur.). Nevertheless, coincidental resemblance cannot be excluded either. // Present in 1/6 branches.