In his paper called, The Aims of Autosegmental Phonology, John Goldsmith forwards a lot of very convincing arguments for the autosegmental phonology program that he helped initiate in the 1960-70s (more like invented). But what caught my eye was what he said about language acquisition towards the end of his paper . I couldn't figure out who published the paper and when it was published but here is a link: http://hum.uchicago.edu/jagoldsm/Papers/AimsAutosegmental.pdf
In effect, in the section on language acquisition, he says (citing research by Liza Menn (1977)) that a child learns a language first by treating all perceived phonological features of an utterance as important and significant to meaning (autosegmentalization), then a process of "systematization" (de-autosegmentalization) over time begins to defoliate and pare down these (what turned out to be superfluous) features until a proper level of linguistic competency is acquired. -Makes total sense to me.
I guess "proper level of...competency" is a rather loaded phrase because the real process of language acquisition (and, more generally, linguistic change) is rarely neat and ideally linear but always contains in it certain randomness and idiosyncrasy. And this is precisely why language changes over time.
The Inuit language is an ideal subject of autosegmental study for this phenomenon. The diachronic literally exists in tandem with the synchronic (ie, contemporaneously related though apparently distinct forms that are spread out geographically) where phonological and morphological changes can be completely described and analyzed in real time.
There is this concept that Goldsmith invokes called "stability" in reference to "the resistance of the tonal features of a vowel to deletion, even when the vowel that bore the tonal features is deleted or desyllabified" (Goldsmith, p. 205) that may be recruited and retooled for our present purposes to describe semantic stability or resistance to consonantal changes that occur naturally synchronic-wise in the process of de-autosegmentalization in Inuktitut. To illustrate:
ivvit tavva tukisijutit "you, this, (do) understand"
may hypothetically be rendered as
illit tagga *dugisivutit
(*denotes a hypothetical rendering by changes in voicing and morphemic simplification but totally plausible while the first two stem from two different dialects and really do exist)
The ivvit/illit change can be explained most elegantly without reference to autosegmental phonology (still being clumsy when it comes to autosegmental analysis, I will just assume it). In eastern arctic Inuktitut the form ivvit came to exist from the reconstructed proto-Eskimo *ilvit whose form exists in Greenlandic as illit.
In Canada, a regressive assimilation occurred before metathesis could happen so we end up with present-day ivvit (ie, *[ilvit] became [ivvit]), whereas in Greendland a metathesis might have occurred first (*ivlit) before the /l/ finally regressively assimilated the /v/ ending up as illit. The accusative case of "you" was retained in either form.
Even nasalization of final segments tend not to affect semantic content of morphemes: sivumun and sivumut mean the exact same thing "forward". Though this nasalization process still occurs phonetically in many eastern arctic dialects the "proper" rendering of the orthographic conventions demand that we spell "sivumut". This is one of the hardest concepts to convey in the exercise of looking into orthographic reform.
Many people insist we spell words exactly the way we pronounce them.
But it is my view (technically speaking of course) there is something to be learned from the way French, English and Hebrew make use of different symbols (or sequences of symbols) to distinguish (especially) homonyms or dissimilar vowel qualities: bear; bare; bear, for eg.
I think these types of distinguishing markers (often silent but act as vital visual cues to pronunciation) need to be artificially introduced into Inuit language orthography(-ies) in order to stabilize the standardization the Inuit Language orthography across different dialects.