Basically, the solution to the problem evoked by Michael Witzel and others is evidently correct: laryngeals which were still actively used and perceived ("sprachwirklich") in an earlier stage of the language had their influence on sandhi-phenomena in Vedic and Sanskrit: this explains both, in large outlines, the prag.rhya vowels ("irregular" external sandhi) and a form resulting seemingly from an "irregularity" in internal sandhi, such as aorist áana.t (LIV2 p. 282-283; vs. the regular augment, á-, before other dhaatus starting in n-, as in ánijam).
Between the earlier stage of the language and the stage observed in Vedic and Sanskrit we can posit, as usual,
A. some "sound laws" -- being, as posited laws, "ausnahmslos", i.e., without exception --
B. AND changes and adaptations by the speech community through analogy (on "soundlaw" and "analogy" in language change see for instance R.S. Beekes, Comparative Indo-European Linguistics, chapter on Language Change),
C. AND (AND THIS IS UNIQUE FOR THE VEDIC AND SANSKRIT TRADITION) THE INTERFERENCE SINCE A VERY EARLY DATE, somewhat annoying for the pure indo-europeanist, OF CONSCIOUS GRAMMATICAL REFLECTION AND SYSTEMATIZATION starting with the pada-kaaras, some of whom preceded Paa.nini, who refers to some of them; and of course Paa.nini himself (not much later or earlier than 350 B.C.) who (chronologically) followed but definitely also preceded some authorial and editorial activity in Vedic literature (as demonstrated by Johannes Bronkhorst and others); hence, Patañjali finds it necessary to state in the Vyaakara.na-Mahaabhaa.sya that it is the padakaaras who have to follow grammar, not the other way round :
na lakṣaṇena padakārā anuvartyāḥ | padakārair nāma lakṣaṇam anuvartyam.
(So, yes, at least according to Patañjali, "sandhi" has definitely dependence on "grammar")
In any case, in the context of internal vs. external sandhi we have to be aware that the padapāṭha of the R̥gveda, attributed to Śākalya, treats as words the forms which correspond to the Pāṇinian pada (AA 1.4.14-17) not only in the restricted sense (sup and tiṅ -- in Pāṇini's definition this includes forms where a supposed sup disappears, as in avyayas, AA 2.4.82), but in the broader sense as well, which includes the nominal stem before endings meeting certain conditions (not meeting the conditions for applying the technical term bha: AA 1.4.18-20). What modern Vedic grammar regards as different rules of internal sandhi when joining a nominal stem and an ending is according to both Śākalya and Pāṇini partly identical, partly not-identical to external sandhi, simply depending on whether the pre-vibhakti stem is a pada or bha.
For possible enthusiasts I refer to
https://www.academia.edu/39108127/Jan E.M. Houben 2016 -- From Fuzzy-Edged "Family-Veda" to the Canonical Śaakhas of the Catur-Veda: Structures and Tangible Traces
(on the activity of padakaaras in Vedic India but
not in the Avesta -- two cultural-linguistic domains which were GEOGRAPHICALLY not just adjacent but partly overlapping)
and
Jan E.M. Houben, Conférences de l’année 2011-2012, in: « Sources et Histoire de la tradition sanskrite », Annuaire de l'École pratique des hautes études (EPHE), Section des sciences historiques et philologiques [En ligne], 146 | 2015 (inter alia on pada 'word' in later Vedic texts -- not yet in the RV, where pada has a different meaning -- and in Pāṇini).
Further: Jan E.M. Houben, “Studies in India’s Vedic Grammarians, 1: Nārāyaṇa Bhaṭṭa’s Prakriyā-sarvasva and Pāṇini’s Śe.” In: Studies in Sanskrit Grammars : Proceedings of the 14th World Sanskrit Conference, ed. by G. Cardona, A. Aklujkar, H. Ogawa : 163-194. New Delhi : D.K. Printworld. 2012.
The basic reason for the special behaviour of dual endings might rather be that some of them originally had a disyllabic structure *-ahiH, *-iHiH so that it was not reduced to a simple
diphthong or vowel in Sandhi, and by analogy this was extended to other duals.
Malzahn assumes this also for the neuter in *-o-jh₁#V > *-aiH#V > *-ayy#V but here I would doubt that the laryngeal had such an impact since it clearly did not in cases of *ayHV > Skt.
ayV in internal position. But feminine forms in *-aH-iH and *-iH-iH are a different case. I discussed this in a review (Kratylos 49, 2004, pp. 53-55).
Best wishes,
Martin
Thanks, Dieter, for this helpful reference.
If I read Malzahn’s arguments regarding the dual endings correctly, she suggests that their exceptional behavior can be attributed to the frequency of pre-pausal forms in numerals and vocatives, and that in the latter
case prosodic factors may have played a role.
I’m not convinced, however, that this will explain the lack of sandhi; if anything, a short final vowel might be expected to undergo contraction even more readily than a long final vowel. Moreover, there are many forms
in short final vowel (e.g. the vocative in -a of a-stems) that do not exhibit exceptional sandhi behavior. So the situation still looks murky to me.
On the sandhi behavior of the dual endings and the linguistic history of their pragRhyatva, see Melanie Malzahn's dissertation, esp. ch. 2:
_______________________________________________
INDOLOGY mailing list
INDOLOGY@list.indology.info
https://list.indology.info/mailman/listinfo/indology
--