Re: Substrate language which contributed sarSapa to

However let me put up again my arguments on zarkara in a bit more elaborate manner, to know how convincing they are.

Let us look at certain words in DEDR (Dravidian Etymological Dictionary, 1984):

DEDR 1265 Tamil karaṭu roughness, unevenness, churlish temper; karaṭṭu rugged, uneven, unpolished; karaṇ uneven surface in vegetables and fruits, scar; karu prong, barb, spike; karumai, karil severity, cruelty; karukku teeth of a saw or sickle, jagged edge of palmyra leaf-stalk, sharpness. Malayalam karaṭu what is rough or uneven; kaṟu rough; kaṟuppu roughness; karuma sharpness of sword; karukku teeth of a saw or file, thorns of a palmyra branch, irregular surface; karukarukka to be harsh, sharp, rough, irritating; karikku edge of teeth; kari-muḷ hard thorn; projecting parts of the skin of custard-apples, jack-fruits, etc.; kari-maṭal rind of jack-fruits. Kota karp keenness or harshness (of wind); ? kako·ṭ hoe with sharp, broad blade (for -ko·ṭ, see 2064). Kannada karaḍu that is rough, uneven, unpolished, hard, or waste, useless, or wicked; kaṟaku, karku, kakku, gaṟaku, garaku, garku, garasu a jag, notch, dent, toothed part of a file or saw, rough part of a millstone, irregular surface, sharpness.

...

Let us look at another closely related word:

DEDR 2354 Tamil caracara (-pp-, -tt-) to be rough of surface; caracarappu roughness of surface or edge; caruccarai roughness, ruggedness; caral, caraḷ, caraḷai gravel, laterite; curacura (-pp-, -tt-) to be rough, have a rough surface; curacurappu roughness as of woollen cloth. Mlayalam caral, carakkallu gravel. Kannada (Hav.) caralu small rounded pebbles. Tulu caraṭè what is coarse, leavings or stalks; jari grit, granule, sand. Kui srogu a rough surface, coarse sand or pebbles; rough, coarse, uneven; jrogu rough, gravelly; srogu srogu inba to be rough, coarse, uneven, pebbly; srāmbu gravel.

These words are again from the same stem ‘kar-’ formed by the change ka > ca.

The Dravidian words change C for different reasons than in Skt. Not only k / t / s / c but k / p / m / v (and kk / mp / pp , etc.) alt. are seen (with opt. V changes > i after k/c, > u after k/p, showing a cluster like ky ( > ty ( > cy) ) or kw ( > pw > p / v), such as in:

cilucilu -pp/tt- = rain gently, cilumpu -i- = flow out, jilïpu = ooze Kan; ciluku \ cilacila \ cilikincu \ cilukarincu = sprinkle, cilupu = pond, ciNka = small water channel Tel;

mair(U)suttu \ mairUtuttu = blue vitriol Tulu; tuttu \ tutta \ tutte, tukku = rust Kan; turucu \ turuci = blue vitriol / spot/stain/rust, turicu = b v / fault/crime, tuppu \ tukku = rust, turu = r/verdigris/flaw Tam; truppu = r, trukku = r/v Tel; turumpu \ turuvu = r Mlym;

Now, if something like * (calu-)cilukina > * cilkna > * cilnka > ciNka happened, it alone would be clear that optional shortening could occur to a great extent (seen in many other words, too). If a word at least as long and complex as cilukarincu is the basis of 1, the other * tirincumpa > turumpu \ -c- \ etc. seems as likely. The only reasonable way for the C alt. and disappearance seems to be as in other l. families with opt. nkxy > ntsy , mqXw > mpFw , etc., with many outcomes, often qX > X > h > 0 , making their prior existence in most words only seen in variants or closely related dia. (such as -mús^ = nose Yasin, -múpus. Hunza–Nagar (both in Burushaski) ). For 'stain/rust', a rec. at least as complex as * atsyarwankxyampwinqXya is needed.

C voicing alt. in an intervoc. cluster like tr or nc or mk can explain Dravidian / Tamil, c- \ j- , k- \ g- , etc.

[the copies of previous messages removed -- Sergei]