Tamil aaytam[aaydam] identified as PDlaryngeal

[Sudalaimuthu Palaniappan]

In a message dated 97-10-10 13:02:44 EDT, you write:

<< From a comparative study of
 definitely attested cases like Old Tamil ahtu 'that one', ihtu, 'this one',
 and pahtu/pattu 'ten' , h was traced to a PD laryngeal *H. By examining  a
 number of cases in Dravidian with similar phonological behaviour, I have
 shown that a PD laryngeal *H  would explain   several   lexical and
 grammatical items with aberrant phonology better than heretofore. These
 include: (a) the root for  '3' (*muH-); (b) irregular verbs like  caH-'die',
 *taH- 'bring, give to 1st or 2nd person', *waH- 'come' (all with *H in
 root-final position); (c) five verb roots involving *H in its interior
 structure, viz. *aHn-'say', *tiHn- 'eat', *uHn(subdot) - 'drink, eat',
 *kaHn(subdot) - 'see, eye', *weHn- 'hear' ;  >>

Without having access to your full paper and some Dravidian linguistics
publications,  here are some comments.

Coincidentally, I have been thinking about the radical vowel length variation
in connection with my research related to pA, pan2uval, pAr, para etc. In one
of my earlier postings, I said, "Another word with the meaning
�extend/spread� is �para/paravu� (DED 3255), which, I strongly feel, is
related to the root �pA�. (I think the process is similar to tA-taru, and
vA/varu.)"  There are other verbs like no/nO which could  belong to this
category. V. S. Rajam discusses such words in her book "A reference grammar
of Classical Tamil Poetry", p.69.

I think when you say "*wehn- 'hear'", you actually refer to the verb root
often transliterated as "ve" meaning 'heat'. If that is right, there is no
need for an 'n' in the root. Following your model, veh- should be enough.

Based on some words like pA, pan2uval, kaHsu, kaHRen2a, your model needs to
consider alternation of H with l and r also.

Also, if the morphophonemic rule concerning words ending in l or L as in kal
or muL followed by a dental t changing to R and T respectively applied to all
words, then in your model won't all words with roots with a short radical
vowel and ending in l or L have  the "H" inserted between the vowel and l/L?
For instance kaL (toddy) will have to be kaHL. (Actually, there is a word
cited by V. S. Rajam "kaHTu" meaning toddy which occurs in PuRanAn2URu 319.4)
 How does this affect the case of paHtu (ten)?

The medieval grammar nan2n2Ul addresses this problem in one of its puNarcci
rules. Although it does not reconstruct a H, the problem it deals with is the
same thing, although it does it in relation to paNpuppeyarp puNarcci. An
example is to combine pacumai (green) and ilai (leaf). There are two
alternate results possible, paccilai and pAcilai. In connection with deriving
the pAcilai form, the grammarian gives a rule to be applied "Ati nITal".. The
steps work out like this.

pacumai + ilai > pacu + ilai (rule: IRu pOtal)
pacu + ilai > pAcu + ilai (rule: Ati nITal)
pAcu + ilai > pAcilai (with the radical vowel being lengthened, u becomes
extra-short "u" which gets dropped when combining with a word beginning with
a vowel)

It is the second stage which relates to your model. I think, in your approach
the root meaning green should be modeled "paHc".

By the way, is using "w" instead of "v" a new convention in Dravidian
linguistics? I have seen your own "Telugu Verbal Bases" and works by P.S.
Subrahmanyam, and others always using "v".

Regards

S. Palaniappan