Re: origin of the first alphabet

From: Neeraj Mathur (neemathur_at_hotmail.com)
Date: 02/11/05 

Date: Fri, 11 Feb 2005 13:59:15 -0000

<ranjit_mathews@yahoo.com> wrote in message
news:1108082602.374273.202210@c13g2000cwb.googlegroups.com...
> How were vocalic <R:> (<kR:>,<pR:>) and vocalic <L> (<klp>) pronounced?
> If it is of any help in guessing the answer, the passive form of <kR:>
> is <kIrya>.
> www.svbf.org/sringeri/journal/vol2no4/sansk1.pdf

This is a very good question; not one, however, with a simple answer.

The vocalic <l> is usually assumed to have been pronounced something like
its counterpart in the English word 'bottle'; the short vocalic <r> is
usually to be like the American English pronunciation of 'butter' (I'm not
convinced; I think it's more likely to have been a trill, since the
consonant was a short trill, but I have no sources / proof etc.).

The long versions are a different story. In Indo-European terms, the
so-called 'long resonants' /r: l: m: n:/ are actually sequences of syllabic
resonant plus laryngeal.

Perhaps an example would explain that better. The stem for 'be born' in IE
is *genH1-. If you add the suffix -tor (which creates an agent noun), you
get *genH1tor, which is reflected in Greek 'genetor', Latin 'genitor' <
*genator (by vowel weakening), and Sanskrit 'janitar-'. If, on the other
hand, you add the suffix -tos, which gives you a verbal adjective, the stem
loses its *e (that is to say, you get the zero grade), leaving you with a
form *gnH1tos. Trapped between consonants it is less sonorous than, the *n
must serve as the vowel, which it does happily. However, the following
laryngeal causes it to act in funny ways. The expected reflex of a syllabic
IE *n is Grk a, Ltn en, Skt a. The forms found, however, are in fact as
follows: Grk gne:tos (found in the word 'kasigne:tos' meaning 'brother'),
Ltn gna:tus (later na:tus), Skt ja:tah.

Before laryngeal theory was developed, these correspondences were thought to
reflect an Indo-European 'long syllabic resonant', *n:. Once laryngeals were
'discovered' the question became whether or not 'long resonants' could
actually be said to exist in Indo-European. Much of the argument is based
around what to make of the evidence in Greek: most people are willing to
accept that the vowel could differ based on the laryngeal in Greek (so with
H1 we get (kasi)gne:tos, but with H2 we have thna:tos, and H3 gives pro:tos)
and if this is correct we must assume that there was no such thing as a
'long resonant' in IE but rather that the sequence was preserved as a
sequence into the daughter languages and then differentiated - or propose
three different versions of each of the four long resonants, an unwieldy
complication. On the other hand, not everybody accepts the Greek
developments as stated (they think they are due to analogy in some cases and
propose different etymologies in others), in which case it would be
reasonable to think of long resonants being a late Indo-European set of
phonemes, which had arisen from a sequence in early Indo-European. This is
more or less a minority view and not the one I hold.

Right, so now we get back to Sanskrit.

The developments of the 'long resonants' or vocalic liquid + laryngeal
sequences of IE in Sanskrit were as follows: *nH > a:, *mH > a:, *rH > i:r,
*lH > *rH > i:r. There is an occasional reintroduction of the nasal in the
first two; this appears irregularly in Vedic but its distribution becomes
fixed by the Classical period - in any case, it seems to be secondary. The
stated i:r could sometimes be u:r in the vicinity of a labial consonant.

Now, you will have noticed that none of these reflexes actually give the
long r: of the grammarians. Where does this come from, then, you ask?

In actual Sanskrit, the r: vowel occurs in two places only: the first is in
the stated forms of verbal roots; the second is in the accusative plural of
r-stem nouns (eg. pitr:n, acc pl of pitar- 'father'). This latter case is
easily explained: it is secondary, analogical to the endings of all the
other vowel-stems, which form their acc pl with a lengthened vowel followed
by -n (so ka:ta:n, s'uci:n, etc.).

The verbal roots, on the other hand, are not really words: they are simply
grammatical abstractions, created by the Indian linguists, not a part of the
natural language. If you look at the conjugation of a verb whose root ends
in a long r:, you will not find this long vowel anywhere - thus the odd
forms of the roots you quoted (kr:, passive ki:rya-; pr:, p.p. pu:rna-)
which are simply the reflexes of the sequence as mentioned above.

So, to come back to your question. The pronunciation of vocalic r: needs to
be considered in only one case, because it only existed in one form ever,
and that is the acc pl ending -r:n. It is thus not really overwhelmingly
possible to prove its pronunciation: it is an incredibly limited sound. I
would think that, if the analogical origin is correct, it should be a
prolonged version of whatever sound you want to posit for the short vocalic
r. One idea of how this might sound would be if you contrast the American
pronunciations of 'that coat's made of fur' and 'I've already forgotten!'.
If you think it was a trill, I suppose the long version would just be a
prolonged trill. In any case it wouldn't be very different, since the
syllable will always be closed by Sanskrit sandhi rules; you could argue
that as a phoneme r: doesn't really exist at all and is maybe a slightly
different, morphologically-conditioned allophone of the singular form.

> Is it [j^e:tum] because of a rule applied to [j^ajatum] making it
> [j^e:tum] or is it [j^e:tum] because that is authors' preferred usage?

I'm not entirely sure what you mean here - why are you expecting 'jayatum'?

Consider: from the root 'bhr', we have present 'bharati', infinitive
'bhartum', past participle 'bhrtah'; these are from Indo-European
*bher-e-ti, *bher-tu-m, *bhr-to-s respectively. In other words, the past
participle has a zero grade root, the infinitive has a full-grade root
(which the Sanskrit grammarians called 'guna'), and the present has a
full-grade root plus a thematic vowel. I picked 'ji' because it shows a
similar pattern. Remembering that the root is essentially quoted in its
zero-grade, we can predict some of its forms: zero-grade p.p. 'jitah', full
grade infinitive *jaitum, full grade present with a thematic vowel *jaiati.
I have no doubt that these forms were indeed used; however, all records show
these two developments: the -i- in between two vowels becomes -y-, and the
two vowels -ai- when between two consonants always become -e-. Thus the
forms I quoted above: jayati, jetum. These are the actual attested forms
foun everywhere. There are other places where you can see these kinds of
processes at work - leaf through a Sanskrit grammar and you'll find them
everywhere, particularly in the declension of ablauting i- and u-stem nouns.
As I said, the Indian grammarians recognised this and treated -e- and -ay-
as identical, also -ai- and -a:y- etc.

Neeraj Mathur

Quantcast