It's a common Slavic thing. Original velars palatalized to affricates
before *i. Later *ɨ fronted after velars. Decreased markedness if you will.

Den 5 sep 2017 23:52 skrev "Galen Buttitta" <
[log in to unmask]>:

I think Polish disallows the central high unfounded vowel after velars,
which messes with certain inflections that produce an alternation where it
fronts to [i] in that environment. (Did I get it right? I can't remember
the exact inflection other than that it was something to do with noun case.)

Sent from my iPhone

> On Sep 5, 2017, at 17:37, Jyri Lehtinen <[log in to unmask]> wrote:
> In one of my less developed conlang projects, Janti, I've worked on
> including a process of vowel antiharmony. In essence, the language has a
> vowel inventory of /a e i o ɵ ɨ/ but adds to this a seventh antiharmonic
> morphophoneme //E//. This vowel surfaces usually as an /e/ but if the
> previous syllable already has an /e/, it surfaces as an /a/ instead. Hence
> you get variation like
> tis-nE > tisne
> sawa-nE > sawane
> koe-nE > koena
> I derive this pattern synchronically from a contraction of an earlier
> diphthong *aj into *ɛ that then went on to split and merge with either /a/
> or /e/. The motivating principle behind this split is pressure for
> dissimilation with the previous vowel. The usual outcome was *ɛ > e, but
> when the previous vowel was already an original *e, the language rather
> went to preserve the lowering vowel height contour and ended up turning
> *e(C)ɛ into e(C)a, or *ɛ > a.
> The change continued by levelling the resulting /e~a/ alternation in most
> cases into either an unalternating /e/ or /a/ but the alternation will
> still appear in a few reasonably common inflectional suffixes.
> Now, such a dissimilatory process seems pretty natural to me after working
> out it's details and I wouldn't be at all surprised to get a case or
> with it. I just can't think of any language that would have such
> antiharmony. Perhaps some of you on the list will know better than me.
>   -Jyri