'Vedic .l'

[Arlo Griffiths]

Dear Indologists,

Can anyone help remove my confusion about the following graphemic and
phonetic issues pertaining to the sign/sound often referred to as 'Vedic .l'
(i.e. the intervocalic allophone for .d)? The information provided by
Wackernagel, AiGr. I §222, seems quite insufficient to me.

1) What do the Praati;saakhyas (.Rg-, Vaajasaneyi-) say, precisely, about
the phonetic difference between this .l and normal .d? Deshpande,
;Saunakiiyaa Caturaadhyaayikaa p. 145 speaks of "flaps". Does this imply
that the phonetic value that the ancient Indian phoneticians had in mind can
be equated with the (near-)allophonic intervocalic variant of /.d/ in many
NIA languages?     
    Cf. Masica, The Indo-Aryan Languages, 146f.: "The favorite diacritic of
the ''Northern'' scripts is the subscript dot (.) ... . It is used for the
near-allophonic intervocalic flaps [.r, .rh] corresponding to /.d, .dh/ in
Hindi, Bengali, and Oriya ...; in Marathi and Gujarati it is ignored; ...''.
    If so, would it not be most consistent with conventional
transliterations for the writings systems as they are used for NIA to render
'Vedic .l' by r-underdot as well? At least in Oriya and ;Saaradaa scripts,
the mss. in fact write agnim ii.re purohitam, not agnim ii.le purohitam
(even though the Oriya language does have an /.l/ phoneme and a
corresponding separate .l-ak.sara, this is NOT the sign that is used here).

2) What is the basis of Witzel's assertion  "Kashmiri Manuscripts and
Pronunciation" (in Ikari (ed.), A Study of the Niilamata), p. 46 n. 67: "The
_l [i.e. .l AG] used in Vedic MSS now and in printed editions is a Marathi
invention expressing one of their .l-sounds"? As far as I know, the roughly
ƒ (infinity)-shaped sign of Devanaagarii as used for writing Marathi
represents a retroflex lateral phoneme, NOT a retroflex flap. Do .Rgvedins
in the Hindi-belt, where no /.l/ phoneme exists in the vernacular, also use
the same Devanaagarii .l-sign for writing their .Rgveda mss.? Is there any
reason to assume that Marathi .Rgvedins (etc.) pronounce their .l-sign
differently when reading Marathi than when reading from a .Rgveda ms.? Or is
there reason to assume that the Vedic allophone in question received
different phonetic realizations ([+flap]/[+lateral]) in different parts of
India in the course of time?

3) What sign do South Indian mss. use for the sound in question: a
palaeographically l-related sound (as in Marathi mss.) or a .d-derived sound
(as in Oriya and ;Saaradaa mss.), or something else?

Arlo Griffiths