One of the hardest things to master (for Inuktitut speakers) is distinguishing the voiced vs unvoiced segments (called finals in the syllabic writing system). This is not surprising as these variations are usually allophonic and not phonemic in nature. An allophonic variation makes no meaningful difference for it is just a manner of articulation or follows a standardized writing convention (ie, not psychologically real to the speaker), whereas a phonemic different is psychologically real - as one can see below:
iglumi 'in the house'
iglumit 'from the house'
iglumik 'a house'
- the absence of a final in the first instance [-mi] denotes a locative case; the second [-mit] denotes an ablative case; the third [-mik] is an accusative case, and these all make a meaningful difference; but an allophonic variation (as per the writing convention):
nunaqaqtunga 'I live in...'
is sometimes written in syllabics as
*nunaqartunga 'I live in...'
doesn't make a meaningful difference being that /q/ and /r/ are articulated in the same place in the mouth cavity and only differ in the voicing - /q/ is voiceless; /r/ is voiced. Following what is called the ICI syllabic writing conventions (which Nunavut follows) the first example above is the correct one because the voiceless uvular segment /q/ occurs before a voiceless initial segment of the following morpheme [-tunga] - voiceless segments go with voiceless segments; voiced with voiced.
*Inuuvikmut 'to Inuvik' is not well-formed;
Inuuvingmut 'to Inuvik' is well-formed (as per writing convention).
-Again, /k/ and /ng/ occur in the same place of articulation (velar) but differ in voicing.
This system, despite my poor attempts at explaining it, is actually quite elegant for it is merely based on voicing of the first segment of the following morpheme and nothing else. In Inuktitut, the assimilation in manner and place of articulation is usually regressive or 'backward' (as above examples), but progressive assimilation also occurs:
pisuktunga 'I am walking'
the assimilation of the voiceless segment /t/ is progressive because the preceding morpheme ends in a voiceless segment /k/, whereas [-tunga] below becomes [-junga] - again following the voicing rule:
anijunga 'I am exiting'
and not *anitunga
-the last two examples occur at the phono-morphemic level [-tunga] and [-junga] where the assimilation tends to be progressive, but as in the first few examples regressive assimilation tends to occur for the segmental level (ie, just one consonant) voicing assimilation rules.
Older Inuit tend to be masters of the language, and tend also not to follow much the standardized final systems. The absence of the niqqud dots (ie, finals) does not bother the mature reader and errors in pronunciation tend to be neglible because they follow intuitively the grammar and not the spelling conventions (bare syllabics are just mnemonics for them). For younger Inuit who've been taught the whole-language approach (ie, not the grammar but solely the spelling system) the syllabic system is often read torturously syllable by syllable because they've not been taught to master the grammar.
It is for this reason I've often spoken out against the whole-language approach and advocate for grammar-based teaching. The importance of the narrative is key to grammar-based teaching (in direct opposition to rote memorization of 'spelling' lists which has sadly been the practice for a long time now). But I've heard that Nunavut Education HQ unofficially regards Inuit legends as religious material so the logical end result for Inuktitut is language death as things now stand.