Inuktitut Computing

The UQAILAUT Project

Inuktitut Linguistics for Technocrats


Mick Mallon

Ittukuluuk Language Programs

Iqaluit 2000


A: History of Syllabics

B: Linguistic Definitions

C: Static Phonology

D: Syllabic Structure

E: Morphology

F: Morphophonology

G: Systematic Dialectal Differences

H: My Blunder


I can give you all the dates and personalities by cribbing from Kenn Harper's work, but the most important detail to remember is that the syllabic system was originally devised by English clergymen in the last century. They were Godly men, but they weren't trained linguists. The system they developed filtered Inuit sounds through English ears, and we are still struggling with the infelicities that flowed from that.

So much for the last century. Then in 1976 came the meeting at the ICI (Inuit Cultural Institute) in Arviat (then called Eskimo Point). That's when we produced the new, modern, low-fat salt-free dual orthography. We missed a few opportunities, mostly through ignorance.


Phonetics is simply the study of the sounds themselves. The phonetician wants to know precisely how the sounds are produced, or what their precise audio qualities are.

Phonology is the study of the sound system of a language. The phonologist takes the results of the phonetician's work and uses it to see how these sounds interact. To the phonetician all sounds are equal. The phonologist considers phonemes (the basic sounds of the particular language: see below) to be of greater import than allophones (variants of phonemes).

Morphology is the study of the word-building processes of the language. English has a complicated, rather illogical morphology, but the bulk of meaning in English communication is transferred by the way we arrange words in sentences. Inuktitut has a logically simple morphology, which carries a far greater load in communication in Inuktitut than morphology does in English. An Inuktitut word can replace a whole English sentence. Parimunngauniralauqsimanngittunga I never said I wanted to go to Paris.

A morpheme is "a minimal unit of meaning". There are three morphemes in the English word unwarlike. There are eight (or so) morphemes in the Inuktitut word above: Paris + mut + nngau + niraq + lauq + sima + nngit + junga (Inuit morphemes often change shape slightly as they get tacked on).

Morphophonology is the study of how the rules of the sound system (the phonology) affect the word-building process (the morphology). In many languages you don't just jam the bits together: you have to modify them to fit the sound pattern of that particular language. Even in English we say a situation that is not "tolerable" is "intolerable", but one that is not "possible" is "impossible", (m and p are both formed in the same part of the mouth: English likes geographical propinquity in its sounds.) The morphophonology of Inuktitut is much more developed than that of English. The fontographer has to be aware of this. It's not enough for him to know the sounds of the language: he should be aware of their possible combinations.


We'll start with static phonology, a simple listing of the phonemes of something I'll call Standard Inuktitut, in other words, Inuktitut without the dialectal variations.

The layout is pretty standard.

Across the top is the Place of Articulation, i.e., the place in the mouth where the sound is produced. The Inuktitut table differs from the English table in one particularly important point: Inuktitut has uvular sounds; English doesn't. That difference is why, as explained in the history section above, we have that awkward syllabic digraph for qi qu qa q: ᕿ ᖁ ᖃ ᖅ.

Along the side is the Manner of Articulation, i.e., the manner in which we modify the air as it emerges. (This is all explained much more clearly in the PowerPoint presentation you have, the one on Phonology.) The three manners are voiceless, voiced, and nasal. This is of much more than academic interest, because in Inuktitut in any cluster of two consonants ... and you can't have more than two (see next section) ... both consonants must have the same manner. Both must be voiceless, or voiced, or nasal. A sequence like the mp of the English word impossible is impossible in Inuktitut, because m is nasal, and p is voiced. This has an enormous effect on how the finals behave when writing in syllabics, with additional dialectal subtleties, such as gemination (see ... below).

So far we have followed the standard linguistic approach to any language. However, I will eventually be adding another dimension: Flow of Articulation: i.e., is the sound a continuous one, like f s v or l, or is it a stop, like p t k q. We'll get into that later.

Now for the table, on the next page. If my colour printer is still working, you will see the voiceless sounds in green, the voiced sounds in red, and the nasal in blue. That way you can think of the consonant clusters as having to be in matching colours ... rather twee, but it works.


Place of Articulation
labial alveolar palatal velar uvular
Manner of Articulation Voiceless stops p t   k q
  fricatives   s ɬ      
Voiced   v l j g r
Nasal   m n   ŋ [ɴ]

Rules of syllable finals in their basic form

My belief is that in its basic form, every syllable in Inuktitut ends either in:

Rules of Combination: Colour with colour

Voiceless with Voiceless:   ps kp qq tq
  but not mp qv
Voiced with Voiced: vv rl gj gg
  but not vp mj
Nasal with Nasal: mŋ nm ŋŋ nm
  but not vn ml

This simple feature affects many of the combinations of morphemes.

inuk + lu     inuglu   and a person
ᐃᓄ + ᐃᓄᒡᓗ

inuk + mut     inuŋmut   to a person
ᐃᓄ + ᐃᓄᖕᒧ


An open syllable is one that ends in a vowel.

I've been told that Hawaian is a language where all the syllables are open. No syllable ends in a consonant. Oahu lei

Japanese is also fond of open syllables: arimasu kimono fujiyama

English is anything but an open-syllabled language. Even when you allow for our awkward spelling system, we abound in closed syllables, often with a cluster of consonants doing the closing instead of just one: strength enchantment switch Christlike

Now let's cut to the chase for Inuktitut. Here is the formula. Brackets mark options.


The essential element in a syllable is a single vowel:
    V una this one
That essential vowel may be doubled:
  VV uuma this one's
The syllable may begin with a consonant:
  CV siku ice
  CVV taina ᑕᐃ that one
And the syllable may end with a consonant:
  VC iglu ᐃᒡ house
  VVC uattiaq ᐅᐊᑦᑎᐊᖅ that's it (finished)
  CVC tuktu ᑐᒃ caribou
  CVVC quaq ᖁᐊᖅ frozen food

We've already made the classic schoolmarm's distinction between open and closed syllables. For the sake of our discussion on syllabics, I'm going to add these terms:

The complication with syllabics begins when a closed syllable is followed by a headed one, as in natsiq seal. That juxtaposition automatically creates a consonant cluster.

Notice, before we go on, that the syllabic structure of Inuktitut does not permit the conglomeration of consonants that we find in English. If you apply the formula, you will never have a sequence of more than two consonants, or more than two vowels. In fact, with the exception of a few exclamations, such as uaik!, expressing moderate admiration of someone's accomplishment, and of a few uai combinations in the Pond Inlet subdialect, (quaittuq she slips), Inuktitut does follow this rule rigourously.

Also, Inuktitut automatically considers a single consonant in the middle of a word to be the initial consonant of a headed syllable. A word like aiviq walrus, is always analysed as ai / viq, never as *aiv / iq.

So, a closed syllable only occurs at the end of a word, or in front of a consonant in the middle of a word.

Now, finally, we see where the finals come in.

A syllabic final marks the final consonant of a closed syllable.

Except for the two digraphs discussed below (ᖅ ᖕ), the final is the a-syllable symbol written small and high.

Before we get into the complications caused by the phonological system, we should clear up the complications caused by history.

Remember the two digraphs: the first caused by the use of as a diacritic to change k to q, in ᕿ ᖁ ᖃ ᖅ, and the second caused by the use of as a diacritic to change g to ŋ, in ᖏ ᖑ ᖓ ᖕ.


The symbol is inelegant enough in words like uqaqtuq, ᐅᖃᖅᑐᖅ she speaks. But it would be unacceptably awkward to apply it automatically in words with a double qq, such as utaqqijuq she waits, which would come out as *ᐅᑕᖅᕿᔪᖅ. So we have established a convention that double qqs are written in syllabics (but not in roman), as ᖅᑭ ᖅᑯ ᖅᑲ. Therefore the roman utaqqijuq appears as ᐅᑕᖅᑭᔪᖅ.

That creates a problem for transliteration programs, which I believe you have solved. Unless you write a subprogram, correctly written roman utaqqijuq comes out in syllabics as incorrectly written *ᐅᑕᖅᕿᔪᖅ, while in the other direction, correctly written syllabic ᐅᑕᖅᑭᔪᖅ, comes out as incorrectly written roman *utaqkijuq. There's some more history to this. It turned out that the use of q in the middle of words was a mistake, a psychological rather than a linguistic one, (for which I was largely responsible: see the last page). Anyway, Kativik decided to use only r medially: a good decision. (They made another decision, not so good ... more on that later.)


A similar, but not absolutely identical, problem to that caused by is the problem caused by . That symbol is inelegant enough in words like umianga, ᐅᒥᐊᖓ his boat. Once again there is a problem in words with a double ŋŋ, such as umiaŋŋuaq toy boat, which would come out as *ᐅᒥᐊᖕᖑᐊᖅ in syllables, and even more clumsily, as umiangnguaq in roman. The convention here is to drop the first g of the sequence in roman. This gives us the roman umiannguaq, and, to be consistent, the syllabic ᐅᒥᐊᙳᐊᖅ.

Once again, a problem, albeit a minor one. I'll leave the explanation for my discussion of dialectical variations.


Because of the syllabic structure of Inuktitut, finals occur only at the end of closed syllables.

iglu ᐃᒡ house
aullalaaq tut ᐊᐅᓪᓛᖅ ᑐᑦ they will depart

A diacritic can occur anywhere.

However, since the sound ŋ never appears at the beginning of a word, neither does the diacritic .

qajaq ᔭᖅ kayak

That's the template for the use of finals. But in practice the rules of Inuktitut morphophonology, and the differences between dialects, affect what actually appears. We'll look at morphology next, and then go on to morphophonology.


English morphology is incredibly complicated. First of all, you can have both prefixes and suffixes: like, dislike, likeness.

There is the fact that we have four parts of speech to play around with: nouns, grief; verbs, grieve; adjectives, grievious; adverbs, grieviously; plus all the combinations, like an adjective formed by combining a noun with a verb form: grief-stricken.

Then you have all the clutter caused by the inconsistent heritage of Latin and French: deceive, deceit, deception; receive, receipt, reception.

Inuktitut morphology is much simpler in its approach than English. First, with one exception, there are no prefixes, only suffixes. Second, there are only two major classes, Nouns and Verbs. So we'll start:

  Noun Roots Verb Roots  

The Noun Roots can exist on their own, as separate words:

  Noun Roots Verb Roots  

Verb Roots cannot exist on their own; they always need an ending. So I always write Verb Roots with a dash after them:

  Noun Roots Verb Roots  
  iglu taku- see
  qajaq sinik- sleep

There are eight noun endings:

The first three are grammatical (like the -m and -se pernickety speakers put on who "Whom do you wish to see?" "Whose little girl are you?

  noun endings  
  1  ø (i.e., no ending at all) subject of nonspecific verb, object of specific
  2 -up possessor, subject of specific verb
  3 -mik object of nonspecific verb

The next four deal with spatial relationships: some of them have secondary meaning as well.

  4 -mi in, on
  5 -mit from
  6 -mut to
  7 -kkut through

The last one is the "aequalis" case

  8 -tut like

Some examples:

Note: in this section I want to concentrate on the morphology. Therefore most of the examples will be given morpheme by morpheme, with the morphemes separated by +. In the morphophonological section I'll show how the morphemes actually combine.

  nuna + up nunaup the land's
  nuna + mi nunami on the land
  siku + mut sikumut to the ice
  inuk + tut inuktut like a person

(The plural of inuktut is inuktitut, like people.
That is where the name of the language comes from. If you speak Inuktitut you speak like people ... real people.)

Bad News.

There are, by one count, 699 verb endings in the North Baffin dialect, (only 609 in South Baffin).

Why so many?

First of all, Inuktitut nouns and verbs can be singular, dual and plural.

takujunga takujuguk takujugut
I see we two see we several see

Second, instead of using words like because, if, whether, whenever, Inuktitut uses different sets of verb endings:

takugama takugunnuk takungmangaatta
because I see if we two see whether we several see

Third, Inuktitut has one set of verb endings for nonspecific situations, and another for specific situations. Compare:

takujunga takujunga takugama
I see I see because I see
takujagit takujara takugakku
I see you I see him because I see him

Fourth ... no, forget the fourth. It's too complicated. Just trust me. So, anyway. 699 verb endings.

So far:

  Noun Roots Verb Roots
  noun endings verb endings

Nouns can have noun chunks attached to them.

The resultant word is a noun, and has all the privileges of one.

Some noun chunks act like adjectives:

umiaq boat
umiaq + juaq big boat (ship)
umiaq + juaq + mi in the ship

Some noun chunks expand on the meaning of the noun:

umiaq boat
umiaq + lik boat-owner
umiaq + lik + mut to the boat-owner

You can pile up noun chunks

umiaq + juaq + lik + viniq + mit from the former ship-owner

Verbs can have verb chunks attached to them.

The resultant word is a verb, and has all the privileges of one.

Some verb chunks act like adverbs:

taku- see
taku + ttiaq- see clearly
taku + ttiaq + jara I see it clearly

Some verb chunks indicate the time of the action:

taku- see
taku + lauq- saw
taku + lauq + gakku because I saw it

Other verb chunks act like English auxiliary verbs:

taku- see
taku + juma- want to see
taku + juma + jagit I want to see you

And of course, you can pile verb chunks up too:

taku + ttiaq + juma + lauq + gakkit because I wanted to see you clearly

At this stage:

Noun Roots Verb Roots
noun chunks verb chunks
noun endings verb endings

All we have done so far is create bigger and bigger nouns and verbs. The next two types of affixes enormously increase the flexibility of Inuktitut.

Verbs can have Noun-Makers attached to them.

A Noun-Maker turns a verb into a noun, with all the privileges of one.

-ji is a Noun-Maker which has roughly the same meaning as the -er of worker, helper, teacher.

sana- make, work
sana + ji construction worker
sana + ji + tut like a construction worker

-vik is a Noun-Maker which means place where the verb happens.

ani- go out
ani + vik exit (noun)
ani + vik + kkut through the exit

-suuq is a Noun-Maker which means one who carries out the action (habitually)

tingmi- fly
tingmi + suuq airplane
tingmi + suuq + juaq big plane
tingmi + suuq + juaq + mit from the big plane

Nouns can have Verb-Makers attached to them.

A Verb-Maker turns a noun into a verb, with all the privileges of one.

-liuq is a Verb-Maker which means make.

umiaq boat
umiaq + liuq- make a boat
umiaq + liuq + juma + junga I want to make a boat

-siuq is a Verb-Maker which means look for.

tuktu caribou
tuktu + siuq- look for caribou
tuktu + siuq + lauq + junga I looked for a caribou

It's quite all right to switch back and forth, like this:

umiaq + juaq + liuq + vik + mi in the ship yard

... or this:

anaq- defecate
anaq + vik + siuq + junga I'm looking for a defecatorium

Here are the categories so far.

Notice that at this stage they are symmetrical.

Noun Roots Verb Roots
noun chunks verb chunks
noun endings verb endings
Noun-Makers Verb-Makers

However there are three more categories that fit under the Noun heading, and then a final ubiquitous one.

Nouns can have possessive markers attached to them.

A possessive marker indicates that the noun belongs to someone.

panik daughter
panik + ga my daughter
panik + it your daughter
panik + nga her / his / its daughter
Kubluup paninga Kublu's daughter

Possessive markers can take noun endings.

panik + nga + mut to her daughter
nuliaq + viniq + nga + mit from his ex-wife

The Inuktitut Demonstrative Pronouns form a much more complex and subtle category than English ones. They distinguish not simply between this and that, but among this, that over there, that up there, that down there, that inside, that outside, that unseen. They distinguish between stationary and mobile objects. They indicate (by use of the only prefix in Inuktitut: ta-) between objects newly introduced to the conversation and objects already in context. And they have their own set of pronoun endings paralleling the noun endings. Just a couple of examples (given in their full combined forms):

paksuma that one's up there (mobile: new information)
takatutunaq like that one down there (stationary: in context)

And then we have the category I have named Whereats in honour of Newfoundland ("Say where you're to and I'll come where you're at.") The Root form is an exclamation:

tavva! Here it is! (stationary and expected)
avva! There it is over there! (mobile and unexpected)

The form of stem plus one of four endings gives you spatial information

tavvani at this (expected) spot
maangat from this (unexpected) area
tappaunga to that (expected) area up there
kanuuna through that (unexpected) spot down there.

The penultimate category consists of tails, little snippets appended to a word already grammatically complete:

uvanga I, me
uvanga + lu me too
tukisi- understand
tukisi + juq she understands
tukisi + juq + guuq she understands, she says

That's it for morphology.

We'll summarize the morphology on the next page, and then move on.


Noun Roots Verb Roots
noun chunks verb chunks
noun endings verb endings
Noun-Makers Verb-Makers
Pronouns (+endings)  
Whereats (+ endings)  

mit + vik + liaq + juma + lauq + juq + guuq
VR NM VM vc vc ve tail
land place go to want past he he says
he said he wanted to go to the landing strip

However . . .

What happens when we get rid of our + signs and join these morphemes together?

mit + vik + liaq + juma + lauq + juq + guuq
VR NM VM vc vc ve tail

Morphophonology happens.


Every affix has its own phonological behaviour. Every time you add an affix you have to be prepared for it either to affect the preceding consonant (the final consonant of the preceding syllable) or to adjust itself. In addition, there are other phonological behaviours such as gemination, or the Nunavik Law of double consonants, that can add a dialectal flavour to the finished word.

There are seven categories of the preliminary phonological behaviour of affixes, which I shall now list and then, in a moment, describe.





Consonant Alternating

Uvular Alternating

Vowel Heading

Every affix except the solitary prefix ta- must include its phonological behaviour in its description. Examples follow.

The template:

affix grammatical category meaning
phonological behaviour


-viniq noun chunk former, ex-
umiaq Noun Root boat
umiaviniq   former boat, wrecked boat

-niaq- verb chunk general future
sinik- Verb Root sleep
siniŋniaqtuq   she will sleep

-gama/nama/kama/rama verb ending because I
  becausative, nonspecific  
  1st person singular  
Uvular Alternator
tukisi- Verb Root understand
tukisigama   because I understand
tikit- Verb Root arrive
tikinnama   because I arrive
naalak- Verb Root listen
naalakkama   because I listen
utiq- Verb Root return
utirama   because I return


Preliminary note:

With the exception of Vowel Heads, affixes do not usually affect preceding vowels, only preceding consonants. The exceptions to this general rule are noted.

1: Deletion

The simplest of all, and the most numerous category. Deleters delete preceding consonants. Generally, they are unpredictable from their appearance. However, because the syllabic structure of Inuktitut blocks the occurrence of three consonant clusters, any affix beginning with two consonants is automatically a deleter.


umiaq NR boat -ksaq nc potential, material for
umiaksaq boat under construction
umiaq NR boat -liuq VM make, build
umialiuqtut they're building a boat
  -lauq- vc past
umialiulauqtut they built a boat

Neutral, Voicing, Nasalization

These are best understood as three examples of regressive assimilation of Manner. To understand that, we should go back to our table of consonants, repeated next.

In what follows, focus on the Manner, but note how the Place stays the same.

Place of Articulation
labial alveolar palatal velar uvular
Manner of Articulation Voiceless stops p t   k q
  fricatives   s ɬ      
Voiced   v l j g r
Nasal   m n   ŋ [ɴ]

In Assimilation, a consonant becomes similar to its neighbour in one or more of its features. (Remember that there can be only two consonants in an Inuit cluster.)

In Regressive Assimilation, the assimilation operates backwards, from the second consonant to the first.

In Regressive Assimilation of Manner, the first consonant assumes the manner of the second consonant. In conservative dialects, the first consonant changes its manner, but retains its place (Remember that final consonants are always voiceless in their basic form.)

kjgj knŋn

The k becomes voiced, but remains velar.

The k becomes nasal, but remains velar.

2: Neutral

In a regressively assimilating affix, if the second consonant is itself voiceless, then of course the first consonant does not undergo a change of manner. Affixes with this behaviour, or lack of behaviour, we label Neutral.

-siuq- Verb-Maker look for, hunt
aiviq Noun Root walrus
aiviqsiuqtuq   he's looking for walrus

So ... no change in the syllabic final:



3: Voicing

In a regressively assimilating affix, if the second consonant is voiced, then the first consonant becomes voiced also. Affixes with this behaviour we label Voicers.

-vik Noun-Maker place for
niuviq- Verb Root trade
niuvirvik   store

So ... change in the syllabic final:



4: Nasalizing

In a regressively assimilating affix, if the second consonant is nasal, then the first consonant becomes nasal also. Affixes with this behaviour we label Nasalizers.

-mut noun ending to
umiaq- Noun Root boat
umiarmut   to a / the boat

So ... change in the syllabic final:



Technical Note (but it's important):

You have probably noticed that in roman and syllabics we have use the same symbol for both voiced and nasal uvulars. The phonetic symbol we have used in the table for the nasal uvular is [ɴ] in square brackets. [ɴ] is the nasal allophone of q. Unlike /r/, it is not a phoneme: it is not a basic sound in the language. There is therefore no need to give it a separate symbol. No Inuk would ever pronounce the r in umiarjuaq (big boat, ship) as if it were nasal, or the r in umiarmut (to a boat) as if it were simply voiced.

There is, however, another source of confusion here. We'll leave that till later ... the last page.

One last complication with some voicers and nasalizers coming up.

Some odd complications after some Voicers

The affix which we have so far written as -juaq for some reason inserts r after vowels. So ...

-(r)juaq noun chunk big
Voicer: insert r after vowels
iglu + (r)juaq iglurjuaq big house ᐃᒡᓗᕐᔪᐊᖅ
kuuk + (r)juaq kuugjuaq big river ᑯᒡᔪᐊᖅ
tasiq + (r)juaq tasirjuaq big lake ᑕᓯᕐᔪᐊᖅ

In North Baffin and west of there the affix which we have so far written as -vik inserts v after vowels. So, for North Baffin, the Keewatin and west...

-(v)vik Noun-Maker place for
Voicer: insert v after vowels
kati + (v)vik kativvik meeting-place ᑲᑎᕝᕕᒃ
pisuk + (v)vik pisugvik walkway ᐱᓱᒡᕕᒃ
niuviq + (v)vik niuvirvik store ᓂᐅᕕᕐᕕᒃ

In South & East Baffin, and in Nunavik, the affix is simply -vik.

kati + vik kativik meeting-place ᑲᑎᕕᒃ

Similar complications after some Nasalizers

Compare these two affixes, both of which are Nasalizers beginning with m.

-mut noun ending to
nuna Noun Root land ᓄᓇ
nunamut   to the land ᓄᓇᒧᑦ


-mat verb ending because she / he / it
Nasalizer: insert ŋ after vowels
sana- Verb Root make  
sanaŋmat   because he makes ᓴᓇᖕᒪᑦ

Compare these two affixes, both of which are Nasalizers beginning with m.

-mi noun ending in, on
nunami   on the land ᓄᓇᒥ


-mi- verb chunk also, too
Nasalizer: insert ŋ after vowels
sanaŋmijunga   I work too ᓴᓇᖕᒥᔪᖓ

5: Consonant Alternating

So far the affixes we have looked at affect their neighbours, actively. Now we have an affix which reacts to its neighbour. These affixes take one form after vowels, and another after consonants. Disregarding some dialectal flavours, we can state this.

There are two sets of Consonant Alternators.

Note that we give both forms in the dictionary listing.

-ji/ti- Noun-Maker -er
Consonant Alternator
ilisai- Verb Root teach
ilisaiji   teacher
niuviq- Verb Root trade
niuviqti   trader

Syllabics: the vowel form has, of course, no final:

ᐃᓕᓴᐃ- ᓂᐅᕕᖅ-
ᐃᓕᓴᐃᔨ ᓂᐅᕕᖅᑎ

Consonant Alternators abound in Statement and Interrogative verb endings:

tukisijunga I understand naalaktunga I listen
tukisivit? do you understand? naalakpit? are you listening?
ᑐᑭᓯᔪᖓ ᓈᓚᒃᑐᖓ
ᑐᑭᓯᕕᑦ? ᓈᓚᒃᐱᑦ?

6: Uvular Alternating

Now it starts to get complicated. Not to begin with. Some Uvular Alternators are (comparatively) simple:

-ga/ra- possessive marker my
    1st person
    singular possessor
    singular possession
Uvular Alternator
after vowels: add -ga  
after k: fuse k + ga to ga  
after q: fuse q + ga to ra  
ataata Noun Root father
ataataga   my father
panik Noun Root daughter
paniga   my daughter
irniq Noun Root son
irnira   my son

Syllabics: note the final fusing into the possessive:

ᐊᑖᑕ ᐸᓂᒃ ᐃᕐᓂᖅ
ᐊᑖᑕᒐ ᐸᓂᒐ ᐃᕐᓂᕋ

However ...

In conservative dialects some Uvular Alternators have a different form for every possible stem. (Note that Noun Roots end only in vowels, k or q, but Verb Roots end in vowels, t k or q.) Here is one common form from North Baffin west to Siglitun (in the extreme west of Canada.)

-juma/tuma/guma/ruma- verb chunk want to
Uvular Alternator
after vowels: add -juma
after t: add -tuma to the stem, retaining the original t
after k: fuse k + juma- to guma-
after q: fuse q + juma- to ruma-
ani- Verb Root go out
anijumajunga   I want to go out
ingit- Verb Root sit down
ingittumajunga   I want to sit down
sinik- Verb Root sleep
sinigumajunga   I want to sleep
utiq- Verb Root return
utirumajunga   I want to return


ᐊᓂ- ᐃᖏᑦ- ᓯᓂᒃ- ᐅᑎᖅ-
ᐊᓂᔪᒪᔪᖓ ᐃᖏᑦᑐᒪᔪᖓ ᓯᓂᒍᒪᔪᖓ ᐅᑎᕈᒪᔪᖓ

You would be forgiven to think that if you memorized that pattern you could apply it to all the other multiple Uvular Alternators, but unfortunately they are unpredictable. Look at these two sets of verb endings. Notice that despite the obvious similarities they have different forms for Verb Stems ending in t.

I'll spare you the examples. The pattern is all we are after.

-gama/nama/kama/rama verb ending because I
    Becausative, nonspecific
    1st person singular
Uvular Alternator
after vowels: add -gama
after t: nasalize the t to n: add -nama
after k: add kama to the stem, retaining the original k
after q: fuse q + gama to rama


-guma/kuma/kuma/ruma verb ending if I
    Conditional, nonspecific
    1st person singular
Uvular Alternator
after vowels: add -guma
after t: change the t to k: add -kuma
after k: add kuma to the stem, retaining the original k
after q: fuse q + guma to ruma

7: Vowel Heads

At first sight any affix that begins with a vowel seems to be simply a deleter. Take the affix which would seem to be -innaq(-). I call it a double chunk, since it acts as both noun chunk and verb chunk (hence the bracketed dash).

Don't get tangled up in its meaning: it's very subtle.

siku sikuinnaq (it's) all ice
qungattuq qungainnaqtuq she hasn't stopped smiling
inuk inuinnaq a total person
imiqtuq imiinnaqtuq he hasn't stopped drinking

However, there is a potential problem lying in ambush. Remember: the syllabic structure of Inuktitut (C)V(V)(C) prohibits a succession of three vowels. So if you want to say a total qallunaaq or you want to use the verb tuksiaq- to say he hasn't stopped praying you will face a problem:

qallunaaq *qallunaainnaq a total qallunaaq
tuksiaqtuq *tuksiainnaqtuq she hasn't stopped smiling

The answer, in this case, is insertion. We insert the sound ŋ to break up the sequence.

qallunaaq qallunaaŋinnaq ᖃᓪᓗᓈᖏᓐᓇᖅ
tuksiaqtuq tuksiaŋinnaqtuq ᑐᒃᓯᐊᖏᓐᓇᖅᑐᖅ
-(ŋ)innaq(-) double chunk nc: total
    vc: still engaged in
Vowel Head
delete final consonant, then insert ŋ after two vowels

ŋ is not the only sound used to break up the three-vowel sequence:

-(ra)aluk noun chunk large, big
    (often used pejoratively)
Vowel Head
delete final consonant, then insert ra after two vowels
qimmiq qimmialuk a (nasty) big dog
ᕿᒻᒥᖅ ᕿᒻᒥᐊᓗᒃ  
qallunaaq qallunaaraaluk a (nasty) big qallunaaq
ᖃᓪᓗᓈᖅ ᖃᓪᓗᓈᕌᓗᒃ  
umiaq   boat
umiarjuaq   big boat, ship
umiarjuaraaluk   big ship

Insertion is not the only solution used by Vowel Heads to avoid three-vowel sequences. In some cases Vowel Heads go to the extreme of Self-Decapitation. Here is the affix for the plural ending:

-it plural marker plural
Vowel Head
delete final consonant, then delete the i of -it after two vowels
tuktu tuktuit caribou (pl)
ᑐᒃᑐ ᑐᒃᑐᐃᑦ  
inuk inuit people
ᐃᓄᒃ ᐃᓄᐃᑦ  
qimmiq qimmiit dogs
ᕿᒻᒥᖅ ᕿᒻᒦᑦ  
umiaq umiat boats
ᐅᒥᐊᖅ ᐅᒥᐊᑦ  

A final affix to slip in here would be the dual marker for nouns, which has its own morphological charm.

-*k dual marker dual
Vowel Head
delete final consonant, then ...
*Indicates "double the vowel, unless there already are two."
inuk inuuk two people
ᐃᓄᒃ ᐃᓅᒃ  
tupiq tupiik two tents
ᑐᐱᖅ ᑐᐲᒃ  
umiaq umiak two boats
ᐅᒥᐊᖅ ᐅᒥᐊᒃ  

That finishes the section on Morphophonology in general. Now we should take a look at some of the systematic differences among dialects.


This sketch map gives a general idea of the position and relationship of the dialects. Note that Aivilik is very close to North Baffin, and South Baffin to both East Baffin and Nunavik.


Now we'll look at some of the differences.

Vocabulary: Obviously we're not going to take the time to list the different vocabulary items. Such differences are almost all arbitrary.

Grammar: The differences in grammar, i.e., in morphology and syntax, are minor, and mostly arbitrary.

Phonology: Oh yes, indeed.

Static Differences in Phonology

Nasalizing word final consonants

Many speakers, mostly older, more in the west than in the east, tend to nasalize final consonants, especially t.

basic form: inuit ᐃᓄᐃᑦ people
nasalized: inuin ᐃᓄᐃᓐ  

the phoneme /s/ and its western equivalent /h/

Kivalliq, Natsilingmiutut and Inuinnaqtun dialects substitute /h/ for /s/. Because we have the letter h available in a roman alphabet, it is used in the roman orthography. Since the substitution is automatic, it should not be necessary to create a different syllabic symbol, so to my mind, these are perfectly acceptable transcriptions, with the readers pronouncing the syllabics as their dialect prompts them to. (Some of my Western students disagree, they want a phonetic representation in syllabics as well as roman.)

Eastern: siniktuq ᓯᓂᒃᑐᖅ people
Western: hiniktuq ᓯᓂᒃᑐᖅ  

the allophone [b]

Aivilik, Kivalliq, Natsilingmiutut and Inuinnaqtun dialects have the allophone [b] before /l/. Once again, since we have the letter b available in a roman alphabet, it is used in roman. In syllabics we are a little more subtle. We use the p final , and assume automatic voicing. Thus:

Western: kublu ᑯᑉᓗ thumb
Eastern: kullu ᑯᓪᓗ  

the glottal stop [ʼ]

Several dialects have a glottal stop. However the original phoneme it replaces varies from dialect to dialect. Many writers simply use the original phoneme: others use the apostrophe in both roman and syllabics.

Nunavik (Hudson Bay):
original q tupirmi tupiʼmi in a tent
  ᑐᐱᕐᒥ ᑐᐱˈ  
Baker Lake (Kivalliq):
original q Qamanittuaq Qamaniʼtuaq Baker Lake
  ᖃᒪᓂᑦᑐᐊᖅ ᖃᒪᓂˈᑐᐊᖅ  
Natsilingmiutut seems to glottalize ŋ in verb endings. (I haven't finished studying this):
  naalangmat naalaʼmat because he listens
  ᓈᓚᖕᒪᑦ ᓈᓚˈᒪᑦ  
But ...
  niuvirvingmi niuvirvingmi in the store
  ᓂᐅᕕᕐᕕᖕᒥ ᓂᐅᕕᕐᕕᖕᒥ  

Now we should start to look at the processes that have the greatest effects on dialectal differences, and on the pattern of usage of finals.

Assimilation of Place and Flow

We have concentrated on Regressive Assimilation of Manner, as it is the key process in Inuktitut Morphophonology. However, there is also Assimilation of Place.

labial alveolar palatal velar uvular
p t   k q
  s ɬ      
v l j g r
m n   ŋ  

Assimilation of Place

Consider the English expression in + tolerant, which comes out as intolerant,. Note that the n and the t are both in the alveolar column. Now consider the same structure, but with a different base in + possible. The n is still the same alveolar, but the p is labial. If we apply regressive assimilation of place, we get a labial nasal, m, to give us impossible.

In South Baffin and in Nunavik, the sequence ts exists as in natsiq seal. But the sequences ps and ks are subject to regressive assimilation of place, as in:

original SB / Nunavik  
natsiq natsiq seal
ᓇᑦᓯᖅ ᓇᑦᓯᖅ  
takugapsi takugatsi because you (pl) see
ᑕᑯᒐᑉᓯ ᑕᑯᒐᑦᓯ  
iksivautaq itsivautaq chair
ᐃᒃᓯᕙᐅᑕᖅ ᐃᑦᓯᕙᐅᑕᖅ  

Assimilation of Flow

By Assimilation of Flow I mean assimilation from stop to fricative, or vice versa. This is not very common, and in the example I am about to give it has no effect on the writing system, but we should dispose of it, if only for the sake of symmetry. Here is the table for voiceless consonants in Inuinnaqtun, including the fricative allophones, in square brackets. Note that in the Inuinnaqtun dialect the phoneme h replaces the phoneme s, and also the phoneme ɬ.

  Inuinnaqtun Voiceless Consonants
  labial alveolar palatal velar uvular aspirated
Voiceless Stops p t   k q  
Fricatives [f]     [x] [χ] h

Here is one of the basic participles compared with its Inuinnaqtun equivalent. Note the Regressive Assimilation of Flow: stop to fricative: the voiceless velar stop /k/ becomes the voiceless velar fricative [x].

basic form Inuinnaqtun form    
  (as written) (as pronounced)  
pisukɬuni pihukhuni [pihuxhuni] she / walking

North Baffin has a reverse example equivalent: Progressive Assimilation of Flow: fricative to stop:

basic form North Baffin form  
natsiq nattiq seal

Now we're ready for Gemination


Gemination means complete Assimilation. In Gemination one consonant becomes identical to its neighbour. We're going to give you a table indicating the spread of Gemination across the Nunavut and Nunavik dialects, increasing from west to east to south. We'll use these symbols in the table:

a consonant cluster in which the first consonant is labial:
e.g., pk vv mn
a consonant cluster in which the first consonant is alveolar:
e.g., tp lv nm
a consonant cluster in which the first consonant is velar:
e.g., kt gj ŋn
a consonant cluster in which the first consonant is uvular:
e.g., qt rl rm

Note: If the second consonant is a voiceless fricative, then the patterns will be more complex than shown here.

Here is the table. √ means the cluster exists as shown. gem means the cluster is regressively geminated.

  West Aivilik North Baffin South Baffin / Nunavik
labC gem gem
alvC gem gem gem
velC gem

And here is a partial table for clusters involving fricatives.

basic form North Baffin South Baffin / Nunavik
ps ss ts
ts tt ts
ks ks ts

As we give examples of these forms you can begin to see the effect on the pattern of finals in the different dialects. Dialects with a high proportion of geminates have obviously a smaller proportion of mixed clusters.

  West Aivilik N. Baffin S. Baffin / Nunavik
labC takugapku takugapku takugakku takugakku
  ᑕᑯᒐᑉᑯ ᑕᑯᒐᑉᑯ ᑕᑯᒐᒃᑯ ᑕᑯᒐᒃᑯ
        because I see her
  ublumi ublumi ullumi ullumi
  ᐅᑉᓗᒥ ᐅᑉᓗᒥ ᐅᓪᓗᒥ ᐅᓪᓗᒥ
  uvamnut uvamnut uvannut uvannut
  ᐅᕙᒻᓄᑦ ᐅᕙᒻᓄᑦ ᐅᕙᓐᓄᑦ ᐅᕙᓐᓄᑦ
        to me
alvC tikitpa? tikippa? tikippa? tikippa?
  ᑎᑭᑦᐸ? ᑎᑭᑉᐸ? ᑎᑭᑉᐸ? ᑎᑭᑉᐸ?
        has she arrived?
  milvik mivvik mivvik mivvik
  ᒥᓪᕕᒃ ᒥᕝᕕᒃ ᒥᕝᕕᒃ ᒥᕝᕕᒃ
        landing strip
  tikinmat tikimmat tikimmat tikimmat
  ᑎᑭᓐᒪᑦ ᑎᑭᒻᒪᑦ ᑎᑭᒻᒪᑦ ᑎᑭᒻᒪᑦ
        because he's arrived
velC auktuq auktuq auktuq auttuq
  ᐊᐅᒃᑐᖅ ᐊᐅᒃᑐᖅ ᐊᐅᒃᑐᖅ ᐊᐅᑦᑐᖅ
        it has melted
  iglu iglu iglu illu
  ᐃᒡᓗ ᐃᒡᓗ ᐃᒡᓗ ᐃᓪᓗ
  inungnut inungnut inungnut inunnut
  ᐃᓄᖕᓄᑦ ᐃᓄᖕᓄᑦ ᐃᓄᖕᓄᑦ ᐃᓄᓐᓄᑦ
        to people
uvuC No need to fill this in. Only Labrador ever geminates uvulars.

Now you can see some of the effects on the syllabic system:

North Baffin is the only dialect that needs final s , and then only in geminates from original ps: takugapsi because you (pl) see becomes takugassi: ᑕᑯᒐᑉᓯᑕᑯᒐᔅᓯ.

South Baffin & Nunavik only use all the other final symbols (except q and r ) in doubles or geminates, or at the end of words. These two dialects also have a high proportion of ts ᑦᓯ ᑦᓱ ᑦᓴ.

In all dialects the only words ending in p are nouns with the possessive noun ending -up.

The western dialects and Aivilik are the only ones that use a medial p, either to indicate the voiceless phoneme p as in takugapsi because you (pl) see ᑕᑯᒐᑉᓯ, or, in syllabics, to indicate the automatically voiced allophone b as in kublu thumb ᑯᑉᓗ.

Two dialectal peculiarities, an admission of failure, and then we're done.


One of these peculiarities does not affect the syllabic system at all, but would affect any kind of spelling checker or word generator that might conceivably be developed later (much later?)

Nunavik's Law of Double Consonants (LDC)

Nunavik refuses to permit a sequence of two consonant clusters. Any time this is about to occur, the first consonant of the second cluster is automatically deleted. This is a completely phonological process, and pays no heed to the underlying morphology.

The following three words express the same basic idea in the same structure, animal + hunt + he. Note the difference in the surface forms between South Baffin and Nunavik. We'll use tuttu caribou, aiviq walrus, and natsiq seal.

South Baffin LDC Nunavik
tuttusiuqtuq does not apply tuttusiuqtuq
tuttusiuqtuq intervening s breaks sequence tuttusiuqtuq
natsiqsiuqtuq natsi*siuqtuq natsisiuqtuq
natsiqsiuqtuq   natsisiuqtuq
aiviqsiuqtuq aiviqsiu*tuq aiviqsiutuq
aiviqsiuqtuq   aiviqsiutuq

Effect of "ghostly consonants"

Even after a consonant has been deleted by the LDC its ghost continues to wheel its phonological barrow. Compare these forms with the verbs sallu- lie (tell untruth), which ends in a vowel, and aullaq- depart, which ends in the uvular consonant q.

sallujuq he lies aullatuq she departs
Consonant Alternator:
-juq after vowels -tuq after the ghost of a consonant


sallugumajuq he wants to lie aullarumajuq she wants to depart
Uvular Alternator (S.B. / Nunavik forms):
-guma- after vowels -ruma- after the ghost of a uvular consonant

The extra phoneme of Natsilingmiutut

Natsilingmiutut, with other western dialects, substitutes h for s. I have already mentioned that although some speakers want their own syllabic symbol for that sound, in strict linguistic theory it is unnecessary. The fortuitous existence of the letter h in roman is irrelevant.

pisuktuq ᐱᓱᒃᑐᖅ pihuktuq ᐱᓱᒃᑐᖅ she walks

Local opinion may overcome linguistic pedantry, but that is a minor matter.


Natsilingmiutut is the only Canadian dialect to retain the difference between the voiced palatal phoneme j and the retroflex phoneme ɟ. In all other dialects these two have merged to j. I don't know this dialect well, but I have not found a minimal pair, where the only difference in meaning comes from a contrast in these two sounds, (like pet and bet in English.) However the occurrence of j and ɟ is not predictable:

iji ᐃᔨ takuɟuq ᑕᑯᖪᖅ

I don't know how seriously this problem is going to be regarded, but it exists.


It's a long story, but I'll shorten it. Back in 1976, at the ICI standardization conference, because of my belief that it was a good idea to mirror the Assimilation of Manner in the orthography, it was decided to use q for the first consonant in voiceless clusters, and r for the first consonant in voiced and nasal clusters.

That was a mistake. That particular distinction does not come natural to Inuit writers, (possibly because of the non-phonemic status of [ɴ].) Public signs, newspaper articles, government publications, children's literature produced by the Department of Education, all are littered with qs where there should be rs, and rs where there should be qs.

Kativik did the right thing in switching to the use of rs medially, with qs left for word initial and word final. When things settle down, maybe Nunavut will make that change. It won't affect the keyboard or the fonts, but it will reduce spelling errors among the otherwise literate by about 30%.

The original contents of this site was developed by Benoît Farley at the National Research Council of Canada.