Re: Sanskrit nasalized L

From: Shriramana Sharma <>
Date: Sun, 26 Jun 2011 15:32:09 +0530

On Sat, Jun 25, 2011 at 10:31 PM, Peter Constable
<> wrote:
> The interesting thing about candrabindu with C2-conjoining forms, when the conjoining form is positioned directly below the base, is that there's no visual distinction between <C, candrabindu, C-vattu> and <C, C-vattu, candrabindu>.

From the encoded POV this means no visual distinction between C1 + Vir
+ Candrabindu + C2 and C1 + Vir + C2 + Candrabindu. Yes, this is an
inherent ambiguity in the abugida nature of the script as I have noted
in L2/09-372. However, if C2 were to be followed by a vowel sign and
the vowel sign were to be rendered to the right of the base then there
would be slight difference in that the Candrabindu would either be
centered directly over the consonant or the consonant+vowel-sign
cluster. But really, C1 + Vir + C2 + Candrabindu doesn't occur in

> As I look at the Devanagari examples with candrabindu on a three-consonant cluster, it looks as though the candrabindu is always on the middle consonant, and that suggests to me that the candrabindu is being applied to the entire cluster and not specifically the second consonant. Is that the case?

No -- perhaps I should have explained it more clearly. In 10-392 I only say:

"There are also some rare forms of writing Vedic Sanskrit in which the
chandrabindu is
placed on non-nasal class consonants to denote a nasal release:"

The base of the candrabindu is actually the vowelless *middle*
consonant, not the initial consonant:

So the encoded sequence would go: VA + VS-Voc-R + KA + Vir + KA + Vir
+ Candrabindu + NNA + VS-AA.

It is relatively easy to render this naturally in Devanagari (with
C1-conjoining behaviour) because one can just map KA + Vir (both
instances) to the half-form KA and allow the candrabindu to be placed
before the immediately preceding half-form (so there is no question of
centering over the cluster).

However, in South Indian scripts like Grantha or Telugu with
C2-conjoining behaviour, it would be not so natural. To unambiguously
identify the nasality as belonging to the second consonant, one has to
break the cluster and render it so:

KA-with-explicit-virama, KA-with-explicit-virama, Candrabindu, NNA, VS-AA

There is no other unambiguous option. As for the nasal-Y/L/V +
non-nasal-Y/L/V ... cases, the placement of the candrabindu above the
stacked cluster is possible in these cases presumably because we are
supposed to relate the candrabindu only to the consonant that it is
immediately placed above. Observe the four-level stack (actually only
three level if you count only the consonants) of Candrabindu + YA + YA
+ VA in the Grantha sample of TAA(NJ)CHAY~YVANTA -- though there are
three consonant sounds and one (inherent) vowel sound, the candrabindu
only modifies the sound indicated by the first consonant.

BTW as for the legitimacy of such sequences as K + nasal-K + NNA, see
Taittiriya Pratishakhya 21-12 "sparshaad anuttamaat uttamaparaad
aanupuurvyaat naasikyaah'"
( These sequences are
also discussed in Paan'iniiya Shiks'aa and Siddhaanta Kaumudii.

> Where I'm going is that I'm OK with enabling a rendering engine for Devanagari to display <C1, virama, candrabindu, C2> with the candrabindu over the half-form C1, and so give that a distinct presentation from <C1, virama, C2, candrabindu>. But in the C2-conjoining cases, it's seems to me that <C1, virama, candrabindu, C2> does not need to be supported with a distinct rendering from <C1, virama, C2, candrabindu>.

This is true, except when in such a South Indian script C1 is preceded
by another consonant C0 and is hence actually in the middle of a
cluster, in which case the cluster has to be broken up so as to
unambiguously indicate which consonant the candrabindu applies to.

> I'm not an expert on Sanskrit or other S. Asian languages, so there may well be important details I'm not aware of.

I am willing to provide any information I know, and in such matters
that I am also not aware of (and there well may be many) I can inquire
with Sanskrit scholars here.

Shriramana Sharma
Received on Sun Jun 26 2011 - 05:02:09 CDT

This archive was generated by hypermail 2.2.0 : Sun Jun 26 2011 - 05:05:11 CDT