Nyāya Sūtras 170

And this sound is of two kinds: having the nature of speech-sounds, and merely unarticulated sound. Among them, first with regard to what has the nature of speech-sounds—

Nyāyasūtra 2.2.40: Doubt arises from the teaching of both modification and substitution.

In 'dadhy atra' ('curds here'), some consider it a modification, thinking that the vowel 'i' abandons its 'i'-ness and attains 'y'-ness. Others say that when the use of 'i' has found its scope, in the place that 'i' abandons, the use of 'y' occurs. In the scope of continuous euphonic combination (saṃhitā), the vowel 'i' is not used; in its place the semivowel 'y' is used—this is substitution. Both of these are taught. Therefore, it is not known what the reality is.

The teaching of substitution is the reality. For in the teaching of modification, since continuity is not grasped, modification cannot be inferred: only if there were continuity could modification be inferred, in the form: 'something ceases and something arises.' And no continuity is grasped; therefore there is no modification. And when two speech-sounds have different vocal articulators, the use [of one] is tenable upon the non-use [of the other]: the vowel 'i' has an open articulator, the semivowel 'y' has a slightly contacted articulator; these two are to be pronounced by an effort known as distinct articulation; when one of the two is not used, the use of the other is tenable. And if there is no modification, there is no difference: where these two, 'i' and 'y', are not modifications—as in 'yatate', 'yacchati', 'prāyaṃsta', 'iti', 'ikāraḥ', 'idam'—and where they are [allegedly] modifications—as in 'iṣṭyā', 'dadhy āhara'—in both cases the speaker's effort is identical and the hearer's hearing is identical; hence substitution is established.

Also because that which is being used is not grasped [transforming]: indeed, 'i' being used is not grasped as turning into 'y'. What then? In place of the use of 'i', 'y' is used; therefore there is no modification. And if there is no modification, grammatical description is not invalidated. If it is said: 'speech-sounds are not modified,' it is not the case that grammatical description is impossible on this view, such that we would have to accept modification of speech-sounds. Indeed, one speech-sound is not an effect of another speech-sound; for 'y' is not produced from 'i', nor 'i' from 'y'. For these speech-sounds are produced by separate points of articulation and efforts; that one is used in place of another is reasonable. And modification could only be this much: either a transformation or a relation of cause and effect. And neither exists; therefore there are no modifications of speech-sounds. And just as the modification of a collection of speech-sounds is untenable, so is the modification of a single speech-sound untenable: just as in 'aster bhūḥ' (Pāṇini 2.4.52) and 'bruvo vaciḥ' (Pāṇini 2.4.53), in a certain context, another collection of speech-sounds constituting a verbal root is not a transformation nor an effect of the collection of speech-sounds, but rather another word is used in place of another word, so too with another speech-sound in place of a speech-sound. (40)