The vowel .r/ri in Sanskrit and Hindi

[cardona at unagi.cis.upenn.edu]

>I suppose the vowel .r in Sanskrit must have started to cause difficulties
>in pronunciation as soon as in the mothertongues of the speakers this
>sound did cease to exist. Already in Pali this sound was not there and
>various vowels had taken place of it, as is well known. Well and the sound
>of a syllable forming r in between consonants is indeed difficult to
>pronounce - in my mothertongue, where this sound occurs quite frequently,
>small children and people with speech defects tend to pronounce a "mixed
>vowel" at its place. Now, knowing that a sort of r-sound must be heard
>there in Sanskrit (and not just a/i/u) the MIA speakers in attempt to
>pronounce it had to combine it with a vowel.
>  So the "ri"/"ru"/"ra" pronunciation must have gained currency pretty
>early. After all, in some Asokan inscription I recall to have seen the
>word for v.rk.sa written so, that it might well be read as vraccha (and
>not only vaccha), I think some editors also did read it so. In that case
>we would have evidence for [r+vowel] from 3rd BC in Prakrit. And ofcourse,
>speakers tend to pronounce a later learned language as they do their
>mothertongue. (Few people bother not to pronounce their Sanskrit as their
>Bengali, Malayalam, Hindi nowadays...)
>  For [ri] in particular I can't recall some evidence right now.
>Only in Jayadeva's Gitagovinda, which is too late (there a .r si clearly
>rhymed with ri)
>
>The pronunciation ri appears earlier also, as pointed out by Wackernagel
>(Ai.Gr. 31 [par. 28], Nachtraege p. 19).  The Pratijnasutra recognizes re
>for the Suklayajurveda, which goes along with the fact that in this
>tradition the svarabhakti in words like zIrSa is e (zIrekha) and not a.
>The a-coloring is also known early, with evidence from the Rkpratisakhya,
>where vocalic R is classed as jihvAmUlIya (RPr. 1.41) and the sound list
>for which has R after a, not after u.  Originally, though, R must have
>begun with a vowel, as evidenced not only by Avestan but also by the fact
>that the Rgveda has sequences like pUrvebhir RSibhir; if R began with r-,
>one would have deletion of preceding r with lengthening of the vowel
>preceding this original r.  Other factors also point to a variety of vowel
>colors in early R, comparable to the variety in modern pronunciations.
>Regards, George CArdona