The most recent reissue from Sino-Platonic Papers is On Attitudes towards Language in Ancient India, by George Cardona of the University of Pennsylvania. Cardona discusses how grammar “became part of a soteriological system, with linguistic usage a means to acquiring merit and the ultimate good.” [I had to look that one up.]
“In this respect,” he concludes, “the Indian attitude towards language is probably unique.”
He gives several examples from early writings, including this one:
“The demons, with speech taken away from them, said he’lavo he ‘lavaḥ and were thus vanquished. They uttered this indistinct speech there. That is barbaric speech (mlecchaḥ). Therefore, a Brāhmaṇa is not to utter barbaric speech (na mlecchet), for this speech is of the demons. One who know thus takes the speech of his competitors who hate him; they are vanquished with their speech taken from them.”
Now, the contrast here is not between Ārya and non-Ārya pure and simple. Instead, the emphasis is placed on usage that is correct according to an accepted norm and usage that is considered barbaric because of its deviation from the norm. Thus, the demons are said to have been vanquished because, incapable of uttering the correct form he3arayo he3arayaḥ, they said he’lavo he’lavaḥ. That is, instead of the accepted form arayaḥ, with –r– and –y-, they used a dialectal and unacceptable form alavaḥ, with -l- and -v-; and instead of using a prolated (trimoric) -e3 that is exempt from phonological alternation, they used an ordinary vowel -e and followed the rule of phonologic alternation whereby word-final -e and word-initial a- together give -e-.
Although some of that may sound complicated, depending on your familiarity with that field, the essay as a whole is aimed at nonspecialists.
This was first published in January 1990 as issue no. 15 of Sino-Platonic Papers.