While looking up Jackson's section on epenthetic i in An Avesta Grammar, I rediscovered his section on prothesis (adding a sound to the beginning of a word):

*ri- > iri-: e.g., irinati 'lets go' (cf. Skt riṇakti)

*ru- > uru-: e.g., urūpayeiṇtī 'they cause pain' (cf. Skt ropayanti)

*rv- > urv-: e.g., urvan- 'soul' (cf. Persian روان ravān; Jackson has ruvān!)

In short,

*r + high vowel or corresponding glide > high vowel + r + high vowel or corresponding glide

But if initial r- was to be avoided, why not avoid it before all vowels: e.g., why Rašnu instead of Arašnu?

Moreover, why was "[a]n instance [of prothesis] found also before þ"?

*ty- > iþyejō 'destruction' (cf. Skt tyajas)

3.4.1:40: Prothesis before r- is also found in Mongolian which has no initial r- in native words: e.g., Oros 'Russia'.**

But is there a language with prothesis before þ = θ?

*3.6.3:36: Oros was borrowed into Mandarin as 俄羅斯 *Olosï which has been abbreviated to 俄 E [ɤ] in modern standard Mandarin. So Mandarin 俄 E 'Russia' contains nothing corresponding to any part of Russian Россия <Rossija> 'Russia'!

The graph 俄 originally represented an unrelated word 'in a moment' which was *ŋa in 10th century southern Chinese and borrowed into Vietnamese as nga. Hence the Vietnamese word for 'Russia' is Nga. AN IMPENETRABLE M-EMBRANE?

Maybe I've just been looking in the wrong places, but I've found more free online resources on long-dead and lesser-known Avestan than on Ukrainian, spoken by over 40 million. I was recently pleased to find new posts on ukrainiangrammar.com after a year-long hiatus. If that site can come back, so can mine, though I've only taken a week off. I have an ever-growing backlog of topics, and I fear that I'll forget about them if I don't cover them soon.

In lesson 4 of his Young Avestan course, Prods Oktor Skjærvø explained that in general,

- i and palatalized preceding consonants:

- an i was written before these palatalized consonants

- the palatalization was then lost, but the added i remained

This sequence of changes can be symbolized as

*VCY > *VCY (palatalization) > *ViCY (epenthesis: i.e., vowel insertion) > ViCY (depalatalization)

V = vowel

C = consonant

= palatalization of preceding consonant

Y = either i or

Examples (with intermediate steps omitted)

*api > aipi

*abi > aiβi (with lenition of b to β)

*aēbiš > ibiš (without lenition of b to β)

*aēti >iti

*āhuri >āhuiri

(I don't understand why Skjærvø regards sure as an exception. It doesn't have either i or i̯, so I wouldn't expect it to become *suire.)

I would expect an i to develop before *mY since it developed before other labial + Y sequences: e.g., pi and bi above. Yet according to Skjærvø,

Not all consonants show palatalization by epenthesis. Thus, the palatal consonants c and j and the sibilants (s, z, š, ž), as well as m and h, never do.

For instance, ahi 'thou art' did not become *aihi.

Why is m an exception? What makes it like an impenetrable membrane for palatalization? m has more in common with p and b which are not exceptions than with palatals or sibilants or h (which is from an earlier sibilant s: ahi is cognate to Skt asi). In other words, m does not form a natural class with the other exceptions which could be symbolized as S.

Oddly, Skjærvø mentions one case of an epenthetic i before m:

*arme > airime 'in peace'

(Why the superscript *? Because a nonetymological * is needed to account for the epenthesis? The locative suffix *-e corresponds to Sanskrit -e [not -ye] and is from Proto-Indo-European *-oi, not  *-i̯oi.)

So apparently epenthesis could occur before (and within!) the cluster -rm- but not simple -m- or most clusters: e.g, ahmi 'I am' did not become aihimi.

Tangut fonts by Mojikyo.org
Tangut radical and Khitan fonts by Andrew West
Jurchen font by Jason Glavy
All other content copyright © 2002-2012 Amritavision