Dear Sri Misra,
I wanted to put the following on record before I forgot about it. Regarding pronunciation of सहस्रशीर्षा as सहस्रशीरषा or वर्षमिषवस्तेभ्यो as वरुषमिषवस्तेभ्यो -
There is a taittirIya prAtiSAkhya rule that there is a svarabhakti in the conjunction of repha and UshmANa-s. What is heard as if it is a र or a रु before the sha-kAra is nothing but a reciter's attempt at svarabhakti. I have even seen written notation where the symbol ऋ has been used, e.g. सहस्रशीऋषा, to depict this svarabhakti. Similarly, the word maharshi, in the Veda, it is pronounced as if there is a ऋ sound in there and not as a conjunct consonant involving the repha and an UshmANa. This goes directly counter to the classical vyAkaraNa rule for this word. Now, as for ideal vs. actual pronunciation of ऋ, that is a different matter. In whichever manner a reciter pronounces ऋ, it is in that manner that he renders svarabhakti in rephoshma conjunction.
All the other instances cited from the aupasana website, e.g. िपन्व॑न्न॒जरं॑ सु॒वीरम् vs. िपन्व॑न्न॒जर सु॒वीरम्, etc, pertain to a difference in approach in how to pronounce the anusvAra. They are contingent upon whether to take the anusvAra as a separate vowel that takes its own svarita or whether the svarita applies to the sound preceding the anusvAra. There is a taittirIya prAtiSAkhya rule that counts anusvAra as a hrasva vowel, indicating that rendering an anusvAra independently requires a time mAtra equal to that of any other hrasva vowel. Also, in taittirIya recitation, before repha and UshmANa, the anusvAra retains a glottal component to it, which actually does make it a hrasva sound separate from its preceding akshara. This glottal component is lost when sandhi converts anusvAra to an anunAsika as per the varga of the following consonant and then the whole sound becomes a conjunct consonant, losing its separate time mAtra. And as with the time required to render it, so also with whether the svarita is applied to it separately or not. This glottal feature of the anusvAra is not present in the Rk SAkhA recitations and conseuquently, in what I have heard of the Rgveda, the anusvAra never gets its own svarita. Similar considerations apply to sentence ending -n and to svarabhakti. The problem, if any, is one of how to capture these features consistently in writing. It is not a problem in recitation or in vedAdhyayana per se, so long as adhyApaka and adhyAyI are both careful.
The other problem highlighted in the
aupasana.com site, namely the mixing up of dIrgha svarita and glottal anusvAra specialities from Rg and yajus traditions, within one verse, is indeed one of faulty reciters. Again however, these issues affect the amateur and partial adhyAyIs much more than dedicated reciters of the vedas. These usually affect only the recitation of a handful of Rk verses that are not already in the yajurveda, but have been borrowed by yajurvedins for pUjA and pArAyaNa purposes. And nowadays, many people recite the popular sUktas, the namaka-camaka and other sundry veda verses, without proper training. In my experience, the collective drag inflicted by a host of mispronunciations and misapplied svara-s can be heavy even on seasoned and well-trained reciters. However, these stray instances cannot be laid at the doors of all the oral transmission traditions.
Regards,
Vidyasankar