On 'patra/pattra' again!

[Madhav Deshpande]

        Zydenbos has brought up an important issue.  Prosodically there is
no difference in tatva and tattva, and this allows different readings,
which further can allow different etymologies.  Let me give a more ancient
example: Rigveda 1.24.11c reads: tat(t)vaa yaami brahma.naa.  Here, the
pronunciation of the first segment tat(t)vaa was flexible enough to allow
different analyses.  The RV Padapatha takes it as two words: tat + tvaa.
However, the same line is cited in Yaaska's Nirukta as an example of
Var.nalopa 'deletion of a sound in derivation'.  There is substantial
debate about how exactly this line serves to illustrate this concept.
But to shorten the comment, I will point out that the commentary of
Skanda-Mahezvara on Nirukta says that the form is tatvaa, and that it is
derived from the root -tan- after deleting the final 'n', as is done in
the form matvaa derived from -man-.  Whether this interpretation is
ultimately correct is besides the point.  The very fact that erudite
commentators like Skanda-Mahezvara could read this segment as tatvaa,
especially when the Padapatha reads tat+tvaa, suggests that the
pronunciation of the segment was not clearly indicative of whether there
were one or two 't's.
        A stop-geminate in pronunciation is produced in such a way that
the contact of the articulator-organ is held twice as long without
releasing the contact after the so-called first consonant. This phenomenon
of unreleased first consonants (abhinidhaana) in consonantal clusters is
discussed in detail in the Praatizaakhyas, Cf. Zaunakiiya-
Caturaadhyaayikaa 1.2.2 (vyanjanavidhaara.nam abhinidhaana.h) and 1.2.4
(sparzasya sparze 'bhinidhaana.h) (my edition in HOS 52).  Thus, in
reality, it is not two distinct consonants, but the contact held longer.
Only when one looks at this analytically, in contexts such as preparing
the Padapatha or in explicitly writing down the oral phenomenon, that the
representation has to make a choice of one or two consonant-signs.  The
prosodic value of the previous vowel does not change beyond if two
consonants, rather than three follow it. The value of consonantal doubling
in syllalbication has been discussed at length by Siddheshwar Varma in his
classic work: Critical Studies in the Phonetic Observations of Indian
Grammarians, pp. 61ff.
        Best,
                                                Madhav Deshpande

On Tue, 16 Feb 1999, Robert Zydenbos wrote:

> At 12:08 14.02.99 -0500, Madhav Deshpande wrote:
> >[...] In my opinion, the phonological variation
> >is in all likelihood the more primary fact, while the etymologies are
> >simply rationalizations, after the fact.
>
> This reminds me of words like 'tattva' and 'sattva', which in books printed
> in Kannada script almost always become 'tatva' and 'satva'. Prosodically
> there is no difference. Could this prosodical consideration have played a
> role in the change in spelling from 'pattra' to 'patra'?
>
> RZ
>