This post is intended only for linguistic interest. It is pure speculation on my part.
I've been thinking about the palatalization of /t/ into [s] after the so-called, 'strong i', in North Baffin (don't remember whether the /t/ has to be between vowels or not, like so: itV...). Let me first clarify my assumptions here:
Assumption 1. that the proposed reconstruction of Proto-Eskimo as a four-vowel phonological system is correct and mathematically sound (ie, that there was, at least, a phonetic distinction between ə (an unstressed mid vowel) and i (ie, the strong i));
Assumption 2. that the notions of 'mora' as an analytical basis has predictive consequences;
Assumption 3. at the syllable level, the empty onset slot of the first mora in question contains, at least, an unrealized consonant-like feature(s).
The analysis itself is simple enough to follow (if one keeps in mind the three basic assumptions).
The Proto-Eskimo hypothesis is that, in the North Baffin dialects, the tendency to 'palatalize' /t/ into [s] after strong i (pronounced me) is a remnant of the 'original' (Proto-Eskimo) phonology.
As a practical illustration one may say, in Inuktitut,
itigak 'foot' in South Baffin and elsewhere
isigak 'foot' in North Baffin uniquely.
another example would be:
itiq 'to enter' in South Baffin and elsewhere
isiq 'to enter' in North Baffin uniquely.
In the literature I've read so far this phonological phenomenon is said to be due to the 'strong i' of Proto-Eskimo phonemic system. Period, full stop.
Now, I've been out of linguistics for a while, but I've not come across any analysis that I'd consider satisfactory. The theory seems contrived to fit the data from the get-go, and there seem to have been no proposals to transcend this glaring problem.
-What I propose here is unsubstantiated by any research and based solely on my intuition of Inuktitut as a linguist. But its elegance has captured my heart and soul. And, I think this may suffice.
I start out by assuming that a purely phonological analysis is insufficient to capture a theoretical basis that would approach any semblance of viability. So let us look, instead, at this problem as requiring a syllable-dependent analysis: like so,
for the itigak vs isigak ('foot') variation, we analyse the problem this way:
First, we analyse the original morpheme as
-the Greek letter: μ (or mu) stands for 'mora' (a mora may be regarded as the 'weight' of sound segments within a single syllable structure);
-the Greek letter: σ (or sigma) stands for 'syllable'. Simple enough. Now...
-notice that the onset slot in the first syllable above is empty. Going by Assumption 3 (that the empty slot has 'inherent /s/-like quality'), we then try and account for the phonological change between North and South Baffin dialects. We do this by recalling to mind the notion of 'metathesis'.
In other words, we propose that the first and second onset slots 'transposed' with one another in the process of historical change which resulted in
The second example looks almost self-explanatory now:
/t/ and [s], again, metathesize.
The above dialectal variation problem seems intimately tied into a metathesis phenomenon that affects specific morphemes like:
siti vs tisi 'den'
sitamat vs tisamat 'four'
sitijuq vs tisijuq 'it is hard (in substance)'