ṣ̌ versus rt

Avestan ṣ̌ continues Indo-Iranian *-rt-. Its phonetic value and its phonological status (one or two phonemes) are somewhat unclear. The conditions under which change from -rt- to -ṣ̌- occurs are fundamentally ill-defined. Thus, for example, Gathic/Younger ərəta/arəta ('establish') is a variant of aṣ̌a but is consistently written with r t/. Similarly, arəti ('portion') and aši ('recompense'). But aməṣ̌a ('immortal') is consistently written with ṣ̌, while marəta ('mortal') is consistently written with r t. In some instances, a change is evident in only Younger Avestan. For example, the Gathic Avestan word for "bridge" is pərətūm, while in Younger Avestan it is pəṣ̌ūm. Both are singular accusative forms, but when the word is singular nominative, the Younger Avestan variant is again (and all but once) with r t.

Benveniste suggested ṣ̌ was only a convenient way of writing /rt/ and should not be considered phonetically relevant.^[1] According to Gray, ṣ̌ is a misreading, representing /r r/, of uncertain phonetic value but "probably" representing a voiceless r.^[2]

Miller follows the older suggestion that Avestan ṣ̌ represents a phoneme of its own, for which he introduces the symbol "/Ř/" and identifies phonetically as [r̝̥] (the voiceless allophone of Czech ř). He goes on to suggest that in writing, -rt- was restored when a scribe was aware of a morpheme boundary between the /r/ and /t/.^[3]

Consonants history