Question regarding sorting order & Visarga

[Gérard Huet]

The nagging question of the place of visarga in the sorting order of  
Sanskrit arises in two subtle problems.

1. Should the visarga assimilated to a sibilant count as this  
sibilant in the sorting order ?
This seems logical, and parallel to the ranking of anusvaara as the  
homophonic nasal of the following consonant,
in the case of "non-original" anusvara.
Thus in my lexicon, I put <du.h"sala> after <du"sce.s.taa>, not  
before, as would be suggested by the rules just given
by Ashok Aklujkar.
Monier-Williams avoids the difficulty, by having an entry du.h-  
carrying compounds such as <du.h"sala>.

2. Where should one list <du.hkara>, obtained by sandhi from dus- 
kara, whereas its usual spelling is <du.skara>?
Some arbitrary choice has to be made, it seems, since the 2 spellings  
are at different places in the sorting order.
Monier-Williams indicates this dilemma at entry du.h- p. 483, where  
he signals that du.h-k and du.h-p are to be
looked at du.s- p 487. Thus for him the word <du.h.sama> is listed  
(under du.h-) BEFORE <du.hkara> listed as <du.skara>.
But he does not apply the same reasoning to <du.skha>, which he  
refers under du.s- as link to <du.hkha> listed
not in du.h- but as an independent entry, which he begins with some  
discussion on the etymology of the word.

If we want to standardize this issue, the first question to ask is  
whether the position of a string of letters/phonemes
is determined strictly by lexicographic order induced by the standard  
collating sequence of the alphabet,
or whether it is the position of some standardized form where  
anusvara (and presumably visarga) is to be taken
as the possible homophonic nasal (resp. homophonic sibilant) of the  
next consonant. This second choice, the usual one,
is context-sensitive. It is contradictory with Monier-Williams' use  
of an entry <du.h-> which is schizophrenic with
respect to ordering.

Then, if we do decide to assimilate say .m to n in <sa.mdhi>, as is  
usual, does that mean that the reverse is true,
i.e. is it legal to write say <sasa.mja> instead of <sasa~nja> for  
the perfect of root <sa~nj> ? This is debatable,
since here anusvara does NOT come from morphological sandhi, but  
merely as an abbreviation for the nasal savar.na with j.
Thus, it is not obvious to me that issues of sorting order are  
decided just at the phonemic level, without considering
how the phonemes arise from morphological operations. This issue is  
polluted by typographical considerations, since
anusvaara avoids complex ligatures with nasals.

In the same vein, there are many representational issues with hiatus  
(eg tita_u pra_uga pra_ucya in morphology,
ka.ta_aaste in syntax), with candrabindu versus anusvaara, and with  
the elusive avagraha. All these problems
arise in classical Sanskrit, even if we ignore Vedic difficulties.

Gérard Huet

Le 11 avr. 07 à 00:13, Deshpande, Madhav a écrit :

> Dear Indologists,
>
>      Harry Spier asked me the following question:  "where does  
> visarga come in the sorting order of
> Sanskrit letters?"  Is there a standard placement for the Visarga  
> in the sorting order  that is generally followed these days?
>
> Madhav M. Deshpande
>
>