Phonemes are real entities and the most basic units of speech. Mīmāṃsā authors distinguish phonemes (varṇa) from their phonic manifestation, i.e., sounds (also called `phones’ in contemporary linguistic theory).
Linguistic communication is possible exactly because phonemes are unchangeable. Sounds, however, are changeable and are considered the cause of differences in local accents, etc. Words are strings of phonemes, arranged in specific sequences, just like sentences are string of words.
Mīmāṃsā authors are involved in an ongoing polemics with Nyāya authors about the nature and characteristics of phonemes. Traces of this polemics can be detected already in the Mīmāṃsāsūtra, in a section (MS 1.1.6–23), which has been suspected (see Frauwallner 1961) to be a slightly later addition, insofar as it deals with the nityatva `fixedness’ of phonemes and not just with the topic of the fixedness of the word-meaning relation, which is the only element being really essential for the Mīmāṃsā theory of language. In fact, MS 1.1.5 and then MS 1.1.24 deal with the necessary fixedness of the relation linking linguistic expressions and their meanings. Why would Mīmāṃsā authors also enter into the moot topic of the fixedness of phonemes? It is possible that, as reproduced in the very Mīmāṃsāsūtra, the topic is indeed triggered by (proto-)Naiyāyika objections aiming at showing that phonemes are nothing but sounds, produced each time by the speakers, just like any other sound.
Thus, the polemics with Nyāya (or proto-Nyāya) authors starts with a different understanding of two crucial terms, namely nitya and varṇa. For Mīmāṃsā authors, varṇa indicates the phoneme, as opposed to its phonic realisation, called dhvani. Nitya indicates, as usual in Mīmāṃsā, the fixed nature of phonemes, which cannot be altered by human interventions. By contrast, for Nyāya authors, varṇa indicates exactly the phonic realisation and it is distinguished from dhvani only insofar as groups of phonemes are conventionally linked to a meaning. Nitya indicates for Naiyāyikas temporal eternity, which cannot be attributed to sounds, since one knows out of experience that sounds cease soon after they are produced.
Mīmāṃsā authors reply that phonemes can’t be ephemeral, since if phonemes were nothing but ephemeral sounds, one could not recognise them and therefore linguistic communication —which depends on a fixed link between signifier and signified— would be impossible. Naiyāyikas answer that what one recognises is the universal shared by all single ephemeral sounds