* Proto- Phonology

Doug Hitch Independent Scholar

Proto-Inuit and the daughter dialects have two non-nasal, non- lateral consonant series, /p t c k q/ and /v ʐ j ɣ ʁ/. The distinctive feature separating these sets has been generally regarded as continuance but voicing is a better candidate. PI *c has been seen as affricate [ts] or [tʃ] but a voiceless palatal stop value [c] better explains the diachronic and synchronic data. PI *ʐ and modern /ʐ/ have received a wide range of phonetic characterizations but a voiced retroflex sibilant value [ʐ] best accounts for the historical and descriptive evidence. Some historical documents and modern descriptions are reassessed. Some suggestions are made for the reconstruction of PI.

1 Introduction

The Inuit language family stretches from Big Diomede Island in the Bering Strait to the east coast of . Probably all of the many intervening dialects have been studied, and while the studies have varying degrees of depth, there is no doubt about the basic identity of the phonemic units for any dialect and there is much known about phonological and morphophonological patterning. With the numerous distinct dialects, and the existence of much historical documentation, the language area offers a rich field for historical linguistics and linguistic typology. In these pages much effort is made towards refining our phonetic understanding of the segments /c/ and /ʐ/ in both Proto-Inuit and the daughter dialects. This information has both diachronic and synchronic implications. It can simplify both descriptions and make them more natural. Also presented here is a fresh look at the distinctive feature specification of the consonants. By regarding voicing rather than continuance as the primary distinguishing feature among the consonants, other improvements may be made in the descriptions. After the introductory sections (1.1–1.3) there are sections devoted to individual voiceless (2–2.6) and voiced (3–3.6) consonants. These sections cover historical and synchronic information relating to the phonetics and phonology of each of these consonants. The phonemic nasals /m n ŋ/ are phonologically straightforward and are not discussed. There is a final short section (4) on vowel epenthesis in consonant clusters. Appendices A to H contain lists of examples illustrating the phonetics of some segments. Appendix I provides a list of comparative symbology, and Appendix J is a list of abbreviations and dialect names with geographic indications.

1.1 Method

At least three levels of transcription are needed when working with non-phonological sources: the original transcription in the original orthography or non-standard transcription; a phonetic transcription; and a phonemic transcription. Sometimes there is overlap among them, but often all three require

* In early June 2017 Sean Guistini from College made a special effort to find and send a copy of Dorais 2003 which is out of print (3rd rev ed. in press). In late July 2017 I had a useful discussion with Elan Dresher and Daniel Hall about continuance being non-phonemic. The first draft of this paper received useful comments in October 2017 from Michael Fortescue, Louis-Jacques Dorais, Michael Cook and two anonymous reviewers. In the same month André Bourcier assisted with some and French issues.

Toronto Working Papers in Linguistics (TWPL), Volume 39 © 2017 Doug Hitch DOUG HITCH different symbols. Most scholarship to date makes compromises with these levels. They are combined or modified for the sake of expediency. This is not a criticism. Much very good work has been achieved with compromise transcriptions1 and some compromises are made here in two cases (see below). But at times a rigorous division of levels is necessary for accurate descriptions. Here, source transcriptions are in italics, phonetic in square brackets [ ], and phonemic in slanted brackets / /. These rules do not apply in quotations from other writers where the original apparatus is retained.. In the Inuit language field, the array of symbols in use now and in the past can be daunting. There is no consensus on symbology. The practice here has been to reproduce exactly what the original authors wrote. This is fairly easy now with Unicode and done almost without exception.2 For the phonetic and phonemic transcriptions, IPA is employed. It is not perfect but with so many symbols in use by Inuit language researchers, a symbolic system which is commonly understood is useful if not almost essential. Appendix X+1 has a chart comparing symbols from the major systems relied on here. Where there is any doubt about a representation, there should always be an accompanying phonetic or phonemic rendering in IPA. In phonetic transcriptions here the vowel uvularization is not marked. It predictably occurs before uvular segments. This is a kind of compromise phonetic transcription as it does not write some phonetic information. In phonemic transcriptions of Utkuhiksalingmiutut the glottal stop symbol ʔ is used to show where the glottal stop allophone of /t/ occurs. This is also a compromise transcription as it shows non- phonemic information. Many abbreviations for dialect names are necessary in work like this. Where there is material reproduced from the Comparative Dictionary (CED; Fortescue et al., 2010), the CED dialect abbreviations are used. Also, when the CED more narrowly defines a source by putting it in square brackets, this too is reproduced. Almost all Proto-Eskimo (PE) or Proto-Inuit (PI) reconstructions are from the CED and not especially noted as such. In the few cases where a new, non-CED reconstruction is suggested, this should be clear from the text. Page numbers are not given to sources which are properly indexed. There is little point in giving page numbers to the CED which has two editions with differing pagination. PE and PI forms from there are found in alphabetical order. In contrast to , Greenland, and the western Canadian Arctic, the word “Eskimo” has been deprecated in the eastern Canadian Arctic for more than a generation. Unfortunately, it is long established in the scientific community as the name of a family with three branches, Inuit, Yupik and Sirenikski. It is also used in the name of the parent family, Eskimo-Aleut or Eskaleut. The scholars involved in this field have the highest respect and admiration for the people whose language, culture, and history they study. They mean no disrespect through the limited scientific use of the “E-word”. Perhaps even more unfortunate is that no-one has been able to suggest a manageable alternative. Proto-Inuit-Yupik-Sirenikski is perhaps too unwieldy. It is easy to use the word Inuit when referring to the Inuit. It is difficult to find a new word to describe the parent family. “Proto-Eskimo” is used here with reservation.

1.2 The previous phonemic charts

I have noticed in the literature just one chart specifically of reconstructed Proto-Inuit consonants, but there are also examples of Proto-Eskimo charts which the authors appear to intend as serving also for Proto-Inuit. Bobalijk (1996: 325) provides a PI chart reproduced here as Table 1.

1 For instance, the CED, which is fundamental to this study, has a section stretching over four pages detailing the well-thought-out compromises needed to provide representations of a wide range of dialects and languages in the dictionary entries (xiii–xvi). 2 The only quasi-exception is with the subscript circle used to show the voiceless counterparts of [y] and [ŋ] in Yupik and Sirenikski transcriptions. Here the circle is put above the symbols: ẙ, ŋ̊ . Technically this is not an issue with Unicode but with the text processing software.

2 PROTO-INUIT PHONOLOGY

Table 1: Bobalijk’s Proto-Inuit phonemic chart labial coronal (laterals) velar uvular stops p t k q continuants voiced v ɹ, j l ɣ R voiceless s ł nasals m n ŋ

The CED chart has no phonetic or phonological labels, but much can be inferred by relative placement of the phonemic symbols. It also includes various symbols in parentheses which do not reflect phonemic units in PE. Those extra symbols are omitted in the version of the CED chart given here as Table 2:

Table 2: CED Proto-Eskimo phonemic chart p t c k q v ð l y ɣ ʀ ł m n ŋ

Dorais (1993) offers another presentation of the Proto-Eskimo consonant phonemes, based on that of the CED. His chart, Table 3, has phonological labels. He has substituted symbols more recognizable to people familiar with Canadian Inuit orthographies. He uses Ummaqmiutun r̂ in place of CED ð, the ICI & in place of ł, and ICI j in place of y. Beside three of these symbols he includes Canadian Roman spellings in square brackets, that is, ɣ [g], R [r] and ŋ [ng]. He has also grouped seven symbols under an “apical” label:

Table 3: Dorais’s (1993) Proto-Inuit-Yupik phonemic chart bilabial apical velar uvular stop p t c k q fricative v j r̂ ɣ [g] R [r] lateral & l nasal m n ŋ [ng]

Dorais (2003) presents a somewhat different chart. It has no labels. Under t there are d̶ , ł, l, and under c there is j. Under k is added x and under q is added x̂ . In all of these charts the content of the (bi)labial, velar and uvular columns is essentially identical (Dorais 2003 has extra x and x̂ ). The chief difference in the layouts occurs with the remaining seven symbols. The view here is that the hesitation is due to an unclear understanding of the phonological status of the segments concerned. Refinement is needed in both manner and place specifications. The top rows in all charts contains segments, p t k q, that in the daughter dialects are usually described as voiceless stops. The second rows in the charts have segments (v, ð~r̂ , y~j, ɣ, ʀ~R) that are usually defined as either voiced continuants or voiced fricatives. The two rows are traditionally described as differing in two features, voicing and continuance. In terms of distinctive features, just one of these can be primary, or phonemic, while the other is secondary, or phonetic. That is, in a phonemic description of the segments, just one feature needs to be used to define the fundamental contrast between the two rows.

3 DOUG HITCH

Descriptions either do not specify which is the underlying contrastive feature or they choose continuance as that feature.3

1.3 Proposed phonemics

There is evidence that the contrastive feature is voicing. First, while the segments in the first rows may appear in all positions, the segments in the second rows regularly only occur in the interior of words, rarely initially, and never finally. For example, in Iñupiaq, “Occurring initially are [p], [t], [k], [q], [s], [m] and [n] with [y], [l], and [v] found only occasionally and not in all dialects” (Kaplan, 1981: 32). In the bases section of the North Alaskan Inuit Index in the CED (638–665) there is one entry with initial /l/, livilivilaaluk ‘semipalmated sandpiper’, one with initial /j,/ yuvguq ‘fish slime’ and none with initial /v/. Similarly, in Iñupiaq, “Possible word-final consonants are [t], [č], [k], [q], [m], [n], [ñ], and [ŋ]” (Kaplan, 1981: 33). Farther to the east, there is also final /p/ on the surface in the relative or genitive case ending, e.g., North Baffin igluup ‘of the house’, tupiup ‘of the tent’, inuup ‘of the person’ (Spalding, 1979: 59– 60). Second, in some dialects, in some consonant clusters, the “continuant” segments from the second rows may be phonetically stops, for instance, /v/ may be [b], /ɣ/ [g], and ʁ [ɢ] (see sections 3.1, 3.5, 3.6 below). This encourages the argument that the segments are not defined underlyingly for continuance. They become continuant intervocalically and are realized as stops in some clusters. Third, in some dialects, in some positions, the “stop” segments from the top rows may be phonetically continuants, for instance, /p/ may be [f] or [ɸ], /k/ [x] and /q/ [χ] (see sections 2.1, 2.5, 2.6 below). In those cases, the continuance is also environmentally conditioned and so not phonemic. A fourth bit of evidence involves the laterals. In many dialects there is a contrast between /l/ and /ɬ/. Compare Siglitun taliq ‘arm’, iłuaqtuq ‘is correct’, akłaq ‘grizzly bear’, aglu ‘seal’s breathing hole’. This contrast requires that voicing must be a distinctive feature in those dialects. There is no similar requirement for continuance. If continuance is non-phonemic, it would be technically correct to use stop symbols for the voiced consonants from the second row. That is, /v/ could be rendered as /b/, /ɣ/ as /g/, and /ʁ/ as /ɢ/. Underlying /b g ɢ/ are realized phonetically as [v ɣ ʁ] in most environments. In these pages, for the sake of clarity, I continue the tradition of using continuant symbols for phonemic representations of the voiced segments. Table 3 below is a feature-minimalist chart of the proto-Inuit phonemic units. Known allophony in modern and older recorded dialects is given beside each segment in square brackets. This allophony is discussed farther below in separate sections for each segment. Long sections are devoted to the voiceless palatal /c/, and to the highly interesting voiced apical /ʐ/ (CED PE ð, PI ž; Dorais r̂, d̶ ). There is also new commentary on the [ʔ] allophone of /t/ and the stop [ɟ] allophone of /j/.

Table 4: Feature-minimalist Proto-Inuit phonemic chart Labial Apical Lateral Palatal Velar Uvular voiceless p [p f ɸ] t [t ʔ] ɬ [ɬ] c [s h c ç] k [k x] q [q χ] voiced v [v b β w] ʐ [ʐ ʂ ɖ] l [l d] j [j ɟ] ɣ [ɣ ɡ] ʁ [ʁ ɢ ɴ] nasal m [m] n [n] ŋ [ŋ]

With regard to point of articulation, there are two arguments for placing the laterals on a separate plane. Firstly /ɬ/ cannot begin or end a word, unlike all other voiceless segments (cf. the statements above on Iñupiaq from Kaplan, 1981). It is in a different phonotactic category. Secondly, there are phonological interactions, both diachronic and synchronic, which the laterals exhibit with both the apicals and the palatals in various dialects, and which the apicals and palatals exhibit with each other. The lines in Figure

3 Of the works consulted here only Bobalijk 1996 specifies a contrastive feature. An example of a fairly recent work specifying continuance as the contrastive feature, in descriptions of a wide range of Yupik and , is Compton 2008.

4 PROTO-INUIT PHONOLOGY

1 connect segments involved in interactions. Each segment interacts with three others. All interactions are described in the sections on segments below.

Figure 1: Symmetrical phonological interactions t c ɬ ʐ j l

Placing the laterals on a secondary point of articulation plane allows the remaining points on the primary plane to be distinguished from each other by the same three features needed to distinguish the four proto-Inuit vowels from each other.

Table 5: Proto-Inuit place features high low grave lateral labial - - + (-) apical - - - (-) (lateral) (-) (-) (-) (+) palatal + - - (-) velar + - + (-) uvular - + - (-) /i/ + - - /u/ + - + /ə/ - - - /a/ - + -

2 Voiceless consonants

2.1 Labial /p/ [p f ɸ]

The consonant cluster /pc/ is often realized as [fs] or [ff]. In Alaskan Iñupiaq the first segment becomes a continuant. In Barrow the cluster is [fs] but is non-phonemically written vs as in tavsi ‘belt’. In Kobuk in this environment “speakers tend to pronounce a bilabial, rather than a labio-dental fricative,” i.e., [ɸ] and prefer to write tapsi (Kaplan, 1981: 25). The closely related Uummaqmiutun has [ff] for /pc/: piffi ‘dry fish’, qaffit ‘how many?’, aqiffaaraa ‘he kicked it again’ (Lowe, 1984: xxii). Similarly, Kangiryuarmiutun from Holman on Victoria Island () also has [ff] and it is written ff in the orthography: iffuktaa ‘shook it’, tamaffi ‘all of you’, piffi ‘dry fish’ (Lowe, 1983: xix). In relation to Utkuhiksalik pipsi ‘dried fish’ Dyck and Briggs (2005: 24) note, “In /ps/ clusters, the /p/ is pronounced either as [p], [ɸ] or [f], and the /s/, as [s] or [sʲ]”. Here follow some forms from the CED illustrating some of the proto-bases to words mentioned above. While the PE forms have *vc or mc, the PI must have had *pc.

(1) PE *əvcuɣ- ‘shake’, but looks like PI *əpcuk-, SPI ipsuk- ‘shake out’, NAI ifsuk- ‘vibrate’, WCI ipsuk-, ifsuk- ‘shake’. (2) PE *pimci ‘dried fish’, but looks like PI *pipci, SPI pipsiʀaq, NAI, WCI pipsi, Lab [Peck] pipsi. (3) PE *tavci ‘belt’, but looks like PI *tapci, SPI tapsi, NAI tafsi, Nu tapsi, Sig tapsi, Cop taffi. (4) PE *qavcit ‘how many’, but looks like PI *qapcit, SPI qapsit, NAI qafsit, Mal qapsit, WCI qapsit, Cop qaffit.

5 DOUG HITCH

Where the phonetic fricativization /p/ > [f] occurs, it is often parallel to /k/ > [x] and /q/ > [χ] (see sections 2.5, 2.6 below).

2.2 Apical /t/ [t ʔ]

The glottal stop occurs as an allophone of /t/ in the Central Arctic dialects Utkuhiksalik and Natsilik. I tentatively proposed this for Natsilik in 1994 after a short project with speaker Attima Hadlari and technical assistant Norman Keenainak (Hitch, 1994a, 1994b). Now more complete documentation is available for Utkuhiksalik.

2.2.1 Utkuhiksalik /t/ > [ʔ]

Cook (2010) has prepared a useful monograph on glottal stop in Utkuhiksalingmiutitut. It occurs before the three nasals, [ʔm], [ʔn], [ʔŋ], and before the labial, apical and lateral voiced consonants [ʔv], [ʔʐ], [ʔl] (2010:3). Cook follows Dyck and Briggs (2005) in calling the three voiced consonants /v ʐ l/ sonorants, thus putting them in the same natural class as nasals. The segments /v ʐ/ are not sonorant any more than /ɣ ʁ/ are. There is no need to have a feature of sonorance in this language. Some other principle is creating this pattern. The strongest evidence for including [ʔ] as an allophone of /t/ may lie with the synchronic [utiC~uʔC] alternations listed by Cook (2010: 6): piqutiminik~piquʔminik 'his own possessions', qavřutiřaa~qavřuʔřaa (ř=/ʐ/) 'she put out the lights on him'. This involves the applicative verbal or instrumental nominalizer postbase -uti. When the affix final weak i is optionally deleted before a following “continuant”, the -ut- becomes -uʔ-. The clusters listed are -ʔm-, -ʔř-, and -ʔv- From the earlier statement about sonorants we can infer that -ʔg- and -ʔr- do not occur. It would be useful to have a positive statement about how the language behaves with postbases beginning with -g- /ɣ/ or -r- /ʁ/ that follow -uti. Presumably in these cases the i may not be deleted. There is also synchronic /t/ > [ʔ] where a morpheme final /t/ meets a morpheme initial /n ʐ l/. Presumably the environment is before any nasal or /v ʐ l/ but I find just three examples, two from Cook (5-6) and one from Dyck & Briggs (7):

(5) igluʔnut ‘to your house’ PE iglu- + 2S abs. -n + All. P -nun; Utku iglu + -t + -nut. (6) niritiʔlunga ‘while I was eating...’ niri- ‘eat’ + -tit- 4 (PE *tət- ‘cause or let’ “used with the subord./contemp. mood” CED 472) + -lunga ‘while I’. (7) apuʔřaqtuq ‘it’s buried in deep snow’ (Dyck and Briggs 2004:26), cf. ECI aput ‘snow on ground’.

There is substantial historical and comparative evidence showing that [ʔ] in Utkuhiksalik clusters is an allophone of /t/. There appear to be two sources of glottal stop in these clusters. One is historical *t before nasal or *v, *ð/*ʐ, *l, *y such as PE *tutmaʀ- > Utku tuʔmaq- ‘tread on’ (examples in Appendix A). The other is historical voiceless consonant before an apical voiced consonant (*n, *y, *l, *ð/*ʐ) such as PE *əpnaʀ > Utku iʔnaq ‘cliff’ (examples in Appendix B). Cook hesitates to conclude that [ʔ] is an allophone of /t/ in Utkuhiksalik because of the peculiar affix -ʔa- (expressing conjecture). He describes it “as the only regular occurence of a glottal stop in intervocalic position” and gives two examples, qaiřuuʔařuq ‘I think she has come’ and iniqtaɬugiʔařaa ‘I think she finished it’ (Cook, 2010: 7). Michael Fortescue (p.c.) suggests this postbase may be an alternative reflex for PE *tya- ‘intend or be about to’ which also appears in Natsilik and Utkuhiksalik as -ʔya- ‘vague future’ (CED: 474; and below, Appendix A, PE *ty > ʔj). At any rate, this instance is highly marked, and a description such as “[ʔ] is an allophone of /t/ except in -ʔa-” is reasonable.

4 Analysis suggested by Michael Fortescue p.c.

6 PROTO-INUIT PHONOLOGY

2.2.2 Natsilik /t/ > [ʔ]

The historical origins of Natsilik [ʔ] parallel those for Utkuhiksalik. There is *t > [ʔ] before certain nasal and voiced segments as in PE *natmaɣ- > Natsilik naʔmaktuq ‘he carries (something) on his back’. There is much less phonetic information on this dialect. From Hitch 1994a and 1994b there are examples of historical *t > Natsilik [ʔ] before *m, *ŋ, *l, *y in Appendix C. As with the Utkuhiksalik spellings above, the orthographic renderings contain ʔ rather than the phonemic t. The same sources have two instances of historical *k > [ʔ], one before *n and the other before *l:

(8) iʔniq 'bullet primer' and iʔnialuk ’bullet shell casing’. See the entry for Utku iʔniq ‘fire, flame’ in Appendix B. PE *ək(ə)nəʀ ‘fire’. (9) kiʔliani ‘at/on the edge’. The -nl- in PE *kənlə ‘border or edge’ is supported by Yupik NSY, CSY kənla and Sirenikski kənlə(ŋ)a, but not by Yupik AAY, Nun kəŋla. The Inuit forms support PI *kəkli: NAI kɨlli(q), [Jen.] kiɣlik, WCI kiɣli(k), kiɣliniq, killiniq ‘limit’, ECI killi, [Schn] kiɣli, GRI kiłłik, old ortho. kigdlik. Also Kang. kiglingani ‘at the edge of’. Compare also PE *kətliʀ ‘one far out in front or at sea’ and Net ki’liq ‘one farthest from land’.

They also offer one case of synchronic /k/ > [ʔ], before n:

(10) qamutiʔniittuq ‘he is on your sled’. PI *qamutək ‘sled’ [dual], NAI qamutik ‘low sled’, Sig qamutik ‘sled’, B. Lake [Dor.] qamutiik, ECI qamutiik du., NG qamutik ‘the two sled runners’. (+ - ni ‘on’ + -t ‘your’ + -tuq ‘he is’)

2.2.3 Other instances of glottal stop

In several Inuit dialects there is a glottal stop which is not an allophone of /t/. Some dialects feature a phonetic glottal stop between the two segments in a cluster, i.e. /CC/ > [CʔC]. Many Alaskan Inupiaq speakers insert [ʔ] when the first consonant is a voiceless stop as in [ikʔnɨq] ikniq ‘fire’ (Dorais, 2003: 43). Dorais (2003: 62) reports Natsilingmiutut inserts glottal stops where the first consonant reflects a PE voiceless stop and gives the example qim’miq or qip’miq ‘dog’. The Paallirmiut and Ahiarmiut subdialects of Kivalliq similarly have qim’miq ‘dog’ or kuug’juaq ‘big river’ (Dorais, 2003: 79) The last example shows this to also be a synchronic development, kuuk ‘river’ + -juaq ‘big’. In Qairnirmiut a very small number of words features glottal stop, e.g., man’na ‘thank-you’, -n’naaq ‘small’ (as in anin’naaq ‘small, preferred brother’), or -t’tuaq ‘prominent’ (as in Qamanit’tuaq ‘prominent widening in a river; Baker Lake’) (Dorais, 2003: 79). In Itivimiut (Arctic Quebec, Hudson Bay coast) most speakers have a glottal stop in clusters where the second segment is voiced /v, ʐ, ɣ, ʁ/. The first segment is then sometimes dropped. These examples are in Dorais’s transcription (j̵ = [ʐ]): iv’it~i’vit ‘you’, qur’vik~qu’vik ‘chamber pot’, aj̵’j̵i~a’j̵i ‘picture’, qar’j̵uk~ qa’j̵uk ‘arrow’, ag’gait ~a’gait ‘hand’, tar’raq~ta’raq ‘shadow’ (Dorais, 2003: 115). Somewhat similarly, in Thule “a glottal stop occurs within vv, gg, rr, ll, rv and rl clusters” (Dorais, 2003: 141). In the Kobuk dialect of Malimiutun an intervocalic [k] /k/ is sometimes phonetically replaced by glottal stop as in nirikami~niri’ami ‘when he/she eats’ (Dorais, 2003: 43).

2.3 Lateral /ɬ/ [ɬ]

Dorais (2003: 34–35) juxtaposes the reflexes of PE *ɬ in seventeen dialects. For intervocalic position he uses ‘suitable’ which has the shape iłuaqtuq in ten dialects. In four dialects PE *ɬ has merged with the reflexes of PE *c, giving phonetic [s] or [h]: Cape Dorset, , SE Baffin (some speakers) isuaqtuq; Inuinnaqtun ihuaqtuq. In four dialects PE *ɬ has merged with /l/: SE Baffin (some speakers) iluaqtuq, Thule, W Greenlandic iluartuq, E Greenlandic ilivartiq. For clusters the pattern is

7 DOUG HITCH fundamentally the same, although obscured by other phonological developments. Dorais uses the word for ‘rope, strap’ which is akłunaaq in nine dialects. In Labrador ałłunaak and W Greenlandic ałłunaaq the PE *ɬ phonetically remains and the first segment has assimilated to it. In Thule aglunaaq the entire cluster has voiced and PE *ɬ has merged with /l/. In four of the five remaining dialects the reflex of PE *ɬ has merged with the reflex of PE *c: Inuinnaqtun akhunaaq; Cape Dorset, Nunavik atsunaaq; E Greenlandic atsinaaq. In SE Baffin attunaaq we see PE *ɬ reflected in modern t because the cluster has apparently evolved *kɬ > *ɬɬ > tt. (Dorais, 2003: 96). Later in the same work Dorais treats the reflexes of *ɬ on Baffin Island. In North Baffin /ɬ/ still exists, while in Southeast and Southwest Baffin the segment is realized as [l], [t], or [s]. For example: N iłuaqtuq~SE iluaqtuq/isuaqtuq~SW isuaqtuq ‘is alright’; N tikiłłuni~SE tikittuni~SW tikitsuni ‘while arriving, [s]he…’; N kangiqłuk~SE kangiqtuk~SW kangiqsuk ‘bay, inlet’ (Dorais, 2003:96). In Nunavik, /ɬ/ has merged with [s]; cf. Lab iłuittuk~Nunavik isuittuq ‘bad, uncomfortable’, Lab ałłunaak~Nunavik atsunaaq ‘thong, rope’ (Dorais, 2003: 114).

2.4 Palatal /c/ [s h c ç]

The CED uses the symbol c to represent a segment found in PE and some daughter languages. The authors describe the phonological nature of this segment as:

A palatal affricate, [č], with the following exceptions: an alveo-palatal affricate, [ts], in Central Alaskan Yupik when followed by ə, a dental affricate, [t̪s], in Sirenikski, and a palatalized dental stop [tʸ], when followed by another consonant in North Alaskan Inuit. (CED: xx, fn. 1)

This description is not clear to me.5 There appears to be a mismatch between the symbols č and ts and their definitions. Where č is used elsewhere in language work, as in Americanist tradition, it denotes an alveopalatal (also called palato-alveolar or postalveolar) affricate, IPA [tʃ], as in English chance or itch, not a palatal affricate. As far I know, ts is used elsewhere for an alveolar affricate IPA [ts], possibly for a dental affricate [t̪s̪ ] and never for an alveopalatal affricate [tʃ]. With Inuit phonology it is essential to rigorously distinguish between alveopalatal and palatal points of articulation. Scholars (including myself) have sometimes loosely described č as palatal.6 In addition, there is sometimes the use of ‘palatal’ loosely for ‘alveopalatal’. For instance, for Iñupiaq Kaplan (1981: 30) notes, “What we refer to below as ‘palatals’ are phonetically alveo-palatals”, and Lowe (1985a: 298), for whom Siglitun (1984: xx) “ch sounds as in English chance” includes ch as a “palatal” in a phonemic chart. But there are languages which contrast alveopalatal [tʃ] with palatal [c], like Hungarian; cf. csupa [tʃupɒ] ‘full’ with tyúk [cuːk] ‘hen’, and kincs [kintʃ] ‘treasure’ with ponty [ponc] ‘carp (fish)’. This can be a challenging distinction to master for English speakers. I suspect that some descriptions of Inuit dialects list [tʃ] as a phonetic segment where technically the segment is palatal [c]. The common spelling Netchilik for Natsilik reflects the interpretation of [c] as [tʃ] by English ears. I suspect that the CED may have selected c as the PE symbol because c is Americanist for IPA [ts], and the segment is alveolar [ts] sometimes in CAY, and is dental [t̪s̪ ] always in Sirenikski. With regard to the segment and its reflexes in Inuit, and Proto-Inuit, the choice of c is serendipitous (and possibly confusing) as the segment, in my view, is underlyingly in many dialects a palatal stop, IPA [c]. Today in Inuit dialects, in intervocalic and initial positions, the phonetic reflexes of PE *c seem to almost always be either [s] or [h] (or at least described as such). The reflexes of PE *kicaʀ ‘anchor’ (cf. CAY kicaq) in Inuit are NAI, WCI, ECI, GRI kisaq and Uum, Kang kihaq (SPI kizaq shows *s>z). The reflexes of PE

5 “As regards /c/ in the CED, that is indeed a little confusing and was meant purely as an orthographic convention to cover both the palatal stop in Yupik and various reflexes in Inuit” (M. Fortescue, p.c.). 6 The practice of describing č as a palatal affricate is widespread, and just a few randomly selected examples are listed here: for Californian languages, Golla (2011: xiii, 205); for Solomon Islands Pijin, Jourdan and Selbach (2008: 170); for Old Church Slavonic, Huntley (1993: 133).

8 PROTO-INUIT PHONOLOGY

*caviɣ ‘knife’ (cf. CAY cavik ‘knife, metal, iron’) in Inuit are: WCI savik ‘knife, iron’, NAI, ECI, GRI savik ‘knife’, Uum, Kang, Utku havik ‘knife’. The position here is that historical or synchronically underlying /c/ in initial and intervocalic positions becomes continuant [ç] and then either [s] or [h]. Fish River Qawiaraq in Alaska has a segment written ch where Proto-Inuit had */c/: chawik ‘knife’, uchuk ‘penis’, ichiq ‘smoke’ (other dialects have savik/havik, usuk/uhuk, isiq/ihiq; cf. Dorais, 2003: 44). This palatal (or alveo-palatal) nature is generally thought to reflect an evolution from an older *[s]. For instance, according to Fortescue (1983: 14b), “A peculiarity of the Qawiažaq dialect around Fish River is the replacement of /s/ everywhere by /č/”. And, according to Dorais (2003: 44), “In Fish River Qawiaraq (for most speakers), s is pronounced ch”. Instead of an evolution [c > s > tʃ] it would be simpler to envision [c > tʃ] (if not [c > c]). This is added credence by the evidence from Petitot and Stefansson that the phonetic change [c > s] in Siglitun is recent (see below).

2.4.1 Sixteenth and eighteenth century /c/

In 1576 Martin Frobisher’s assistant collected a list of 17 words while anchored in what is now Frobisher Bay. As Dorais (1993: 39) noted, the entry chewat ‘ear’ (cf. PE *ciɣun, NAI siun, ECI siuti, GRI siut) seems to preserve the older plosive value of the word-initial segment. For us it means that /c/ in initial position retained the value [c] and had not yet become [s] (or [ç]). From around two centuries later there is some sparse vocabulary recorded by a number of mostly French explorers, traders and missionaries from their contact with Inuit in Labrador. Dorais (1993: 42, 44) presents some examples of these from which nine may show [c] or [ç].7 These are given in Table 6. The top eight reflect PE *c.8 The ninth and last item has [c] arising from the palatalization of /t/ by strong i as, for instance, in Iñupiaq isiqtuq ‘enters’.

Table 6: /c/ in eighteenth century Labrador Date Original transcription PE base phonetic Dorais’s analysis 1730 kilocto ‘it rains’ *cila [cilaluktuq] silaluktuq ‘it rains’ 1745 nutchade ‘the hair’ *nuccat pl. [nucçat] nutchat ‘head hair’ 1745 tchiou ‘the ear’ *ciɣun [ciut] siut ‘ear’ — qui oucty — *ciɣu [ciutik] siutik ‘ears (du.)’ — kinicto — PI *cinək- [ciniktuq] siniktuq ‘sleeps’ 1771 pipshy ‘dried cod’ *pimci [pipçi] pipsi ‘dried fish’ — ouctouchic — *utɣuciɣ [utkuçik] utkusik ‘cooking pot’ — jacquoc — PI *icaqquq du. [içaqquk] isaqquk ‘wings (du.)’ 1730 iqui ocpo ‘come in!’ *itəʀ- [iciqpuq] itiqpuq ‘comes in’

While /c/ was a stop [c] in initial position, between vowels it was a continuant [ç] as the spellings ch in ‘cooking pot’ and j in ‘wings (du.)’ show. The exception is the last item where presumably the strong i palatalization rule follows the rule making /c/ between vowels continuant. The one spelling by an English speaker, pipshy, suggests a continuant pronunciation [ç] for /c/ as the second segment in a cluster. In all of these items except ‘hair’ the modern correspondent has [s]. In none of the thirty-six or so words cited by Dorais (1993: 42, 44) is there an s corresponding to modern [s].

7 Dorais describes the phonetics of the segment as “word initial c (ch/ts)” (1993:44) probably intending IPA alveopalatal affricate [tʃ] or alveolar affricate [ts]. 8 *nuccat is my reconstructed plural; CED singular *nuyaʀ ‘hair’. *icaqquq du. reflects my reconstructed PI base *icaʁuq, cf. NAI, WCI, ECI, GRI isaʀuq (CED PE *iyaquʀ ‘wing’).

9 DOUG HITCH

2.4.2 Petitot’s Tchiglit () Vocabulary

Émile Petitot was a brilliant linguist who, from 1862 to 1882 with some interruptions, resided in northwestern and extensively documented several Athabaskan languages. In 1869–1870 he lived with a people he called Tchiglit, and in 1876 he published a French-Siglit vocabulary with grammatical notes and an ethnography. Petitot did not provide a transparent spelling for /c/. In his chart “Alphabet et classification des lettres” there are three relevant entries. He offered a spelling for the palato-alveolar sibilant [ʃ]: “CH, se prononce comme dans charmant, chant” (Petitot 1876: xxxix). He gave a spelling for [s], noting that this sound was rare: “S, se prononce comme dans savoir, souci. Son emploi est très-rare” (Petitot 1876: xxxix). He also noted that “S” is not found at the beginning of words (Petitot 1876: xl). The third alphabet entry suggests an affricate which is neither alveolar [ts] nor palato-alveolar [tʃ]: “TS, a une prononciation mixte entre ts et tch” (Petitot 1876: xxxix). This phonological information tells us that /c/ was still pronounced [c] or perhaps [ç] in most cases, and possibly that the transition towards [s] intervocalically was just beginning. Although not discussed in his classification of letters, Petitot also made use of a symbol ç in the combination tç to write /c/. He transcribed /c/ in a few ways, mostly tç or tch but at least once ts, and at least once he wrote /tc/ (/cc/?) as tci. Appendix D provides examples of all of these spellings together with presumed phonetic and phonemic forms, proto-forms, NAI cognates and modern Siglitun forms.

2.4.3 Stefansson’s trade jargon

Vilhjalmur Stefansson reported on “The Eskimo Trade Jargon of Herschel Island” in 1909. Most of the words derive from Inuit dialects from Point Barrow to the Mackenzie but there are also some from English and Kanaka (Hawai’ian). In most entries he included forms from “Mackenzie Eskimo” or Siglitun for comparison, and some entries have Iñupiaq from Point Barrow. He spent the winter of 1906– 1907 among the Siglitun of the Mackenzie Delta and learned or recorded something of their language. His stay was 36 years after that of Petitot and it is possible that the language had evolved since then. For instance, his notes may show the beginning of the change of initial [ç] to [s]. Eight of his renderings show s where modern dialects have [s]; compare Mac. Esk. sĭr-kĭn-nĭrk ‘sun’ with Sig siqiniq, and Mac. Esk. sī′-la ‘the outdoors’ with Sig sila 9 ‘weather’. Sixteen, or twice as many, of his renderings show Mackenzie or Point Barrow c where modern Sigilitun or North Slope have s. He appears to use different spellings in the same word. He wrote ‘knife’ sa′-vik with s, and then in the next entry he wrote ca-vī-kō′- yak ‘rice’, which he suggested is “perhaps from ca′-vĭ-ĭt, the scrapings (like sawdust) from wood or ivory when scraped with a knife”. 10 This inconsistency may reflect [ç~s] hesitation among speakers, or Stefansson’s inability to distinguish the phonetics. About his transcription system he stated, “c is used as the equivalent of (English) sh” (Stefansson, 1909: 222). Technically, English has alveo-palatal [ʃ] but he was presumably hearing palatal [ç] in Siglitun of that date. Some of the entries below could conceivably reflect [ʂ] for etymological *ʐ after a voiceless consonant. But there is no compelling evidence for the survival of retroflexion. His Mac. Esk. kē′rūk ‘wood’ (Stefansson, 1909: 226) seems to suggest that

9 In Writing on Ice, The Ethnographic Notebooks of Vilhalmur Stefansson, the editor, Gísli Pálsson, includes a short glossary showing “most of the Inuit and non-English terms used in the diaries” (Pálsson, 2001: 325–329). Caution must be exercised with this list. It gives no indication of context, so dialect is unclear. As well, the transcription system(s?) does not match any other I know of. The entry, “tjuna pishunktu—what do you want?” must be the same as Jargon cuna picuktu? ‘what do you want?’ from Stefansson (1909) with extraneous n. It is not clear to me if the initial tj- and medial -sh- represent Stefansson’s early tentative spellings. If they are, they suggest he was hearing an initial stop or affricate, [c] or [tʃ], and a medial sibilant [ç] or [ʃ]. Tjiglik ‘Sigliq’ also has tj- (cf. Petitot Tchigleϱk; 1876: x). The entry “tsila—outside” (Sig sila, Pet tçilla ‘temps’) with unique initial ts- also implies an affricate, in contrast to the Mac. Esk. sī′-la of the jargon list. 10 Under copeau de varlope ‘plane shavings’ Petitot gave tçavit which may be the same word. He also gave copeau très-menus servant de bourre ‘very fine shavings serving as stuffing’ as tçavilit, also a derivative of /cavik/ ‘knife’.

10 PROTO-INUIT PHONOLOGY

Siglitun may have still distinguished /ʐ/ from /j/, cf. Sig qiyuk, Uum qir̂ uk ‘wood’. But note that Petitot earlier had y in kϱéyuk ‘bois’.11 Examples of Stefansson’s /c/ spellings are given in Appendix E together with proposed phonetic and phonemic renderings, proto-forms, modern Siglitun spellings and the corresponding entry form Petitot where available.

2.4.4 Utkuhiksalik /c/

Dyck and Briggs (2005: 313) report that PE *c developed into “[hʲ] between vowels”. Their phonetic rendering, [hʲ], is likely the same phonetic segment as IPA voiceless palatal fricative [ç], which is the continuant correspondant to IPA voiceless palatal stop [c]. The examples given could be regarded as having phonemic /c/ which becomes phonetically continuant [ç] between vowels. For instance, ahi ‘other of a different kind (including ‘another person’)’, PI *aci ‘somewhere or something vague or remote’ may be /aci/ and [açi]. The root in Utku qahungařuq (with [hʲ]) ‘flabby, floppy, slack, not taut’ (ibid.), from PE *qacu- ‘become loose or slack’, may be /qacu-/ and [qaçu-]. And ihuq (with [hʲ]) ‘be murky’ (ibid.) from PE *əcuʀ- ‘be murky’ may similarly be /icuq/ and [içuq]. They report that, “Some words are pronounced exclusively with [hʲ] …, while other words have [h] or [hʲ] in free variation (ibid.). Presumably, for some speakers, intervocalic [ç] is simplifying to [h]. Phonetically revealing is the relationship between single /c/ and double /cc/ in Utku. Dyck and Briggs (2005) provide a number of words which may have underlying /cc/. Utku atsaga ‘my father’s sister’ reflects PE *accaɣ ‘paternal aunt’, SPI assak, NAI atcak, Net atsak, Cop attak, ECI, GRI atsa(k). Utku qitsuktauguma ‘if I am scratched’ reflects PE *qəc(c)uɣ- ‘scratch or dig claws into’, Qaw qissuk- ‘scratch’, NAI qitsuk-, Cop qitsuk-, ECI qitsu(k)- ‘scratch, claw (several times)’, GRI kitsuɣ- ‘scratch, dig nails into (several places)’. For ‘lake’, Dyck and Briggs (2005: 314) list singular tahiq with single /c/ and dual tatsik, plural tatsit, relative tatsip with geminate /cc/. The spellings reflect phonetic [taçik, tacçik, tacçit, tacçip]. Similarly, singular pihiq, plural pitsit ‘song, hymn’ (ibid.) reflect [piçiq, picçit]. The cluster [cç] is probably what I heard Natsilik speaker Attima Hadlari say in 1994 in words he was writing in syllabics with , such as [nacçiq] natsiq ‘ringed seal’. At the time, with my eastern arctic orientation, the phonetics puzzled me. But after working with Hungarian, which distingushes /c/ from /tʃ/ (Americanist /č/), I now wish to correct that initial description. I portrayed the cluster as [tç] (Hitch, 1994a: 10) but [cç] may be more accurate. Two other Natsilik words with /cc/ noticed then are katsungařuq ‘slow, hesitant, without enthusiasm or zeal’ (Hitch 1994a:10), PI *katcuq- ‘be settled’, SPI kassuq- ‘be finished’, finish’, NAI katcuq- ‘be pacified, settle’, Sig katsuq-, ‘be satisfied, full’, ECI katsuŋa- ‘be slow, without zeal’, GRI katsuʀ- ‘be calm, at peace’, and katsuaq ‘biceps’ (ibid.), PE kay(y)u(C)aʀ, WCI, ECI, GRI katsuaq ‘biceps’. In their earlier monograph, Dyck and Briggs (2004: 15) describe Utkuhiksalik “/ts/” as “heavily assibilated” and phonetically “[ts ~ tš]”. Presumably they were trying to write [cç]. One example is qala[ts]araittuq, qala[tš]araittuq ‘it cooks quickly’ (ibid.) with PE *yyaɣ-12 ‘reach state of’ (CED: 483; not PE *ya, tya be liable or apt to’ as Dyck and Briggs). Another example is Utkuhiksalik itsaq ‘tent skin’

11 In Pálsson (2001) there is an effort to write [ʂ], but it is possible that the [ʂ] forms are from Iñupiaq dialects, not Siglitun. The [ʂ] spellings are: “innukcuit—deer stockade” for [inukʂuit] ‘cairns’ PI *inukšuk ‘cairn’; “kaksrauk— black throated and red-throated loon” for [qaqʂauq] ‘red-throated loon’ PE *qaqða(C)uʀ ‘loon’; “nerkiksran parkittutin—you have found something meant for you to eat” containing [niqikʂan] ‘your thing to eat’; “oktjuk— whale meat” and “oxsrogon—with animal fat” both seem to contain [uqʂuk] ‘blubber’. The use of sr for [ʂ] is found in Iñupiaq orthography as is the use of r for [ʐ] in “pubraktuak—to swim” for [puubʐaqtuaq], cf. NAI puuvžaq-, WCI puuvyaq- [Pryde], and in “ugrug—bearded, square flipper seal” for [ugʐuk], PE *uɣðuɣ, NAI uɣžuk ‘’. The form pubraktuak ‘he swims’ has the third singular -tuaq particular in a present sense to Siglitun. Petitot has ugiuk ‘phoque veau marin’ which looks necessarily to show [j] in [ugjuk]. He does not appear to have a stem like *puubyaq- or *puubdǰaq-. Uum is puuvr̂ aqtuq. 12 PE *yyaɣ- might be Proto-Inuit *ccak-, cf. NAI, WCI, ECI (#)tsak- ‘become’ [with emot. roots], GRI (#)tsaɣ-. PE *yy may consistently be PI *cc (see below).

11 DOUG HITCH

(Dyck & Briggs 2004: 16) to PI *itcaq ‘tent skin’, NAI itca ‘skin for tent, ECI itsaq ‘roof of tent or house’, etc. An interesting example involves utjuk ‘vulva’. Dyck and Briggs (2005: 315) suggest this is an instance of PE intervocalic *c > Utku /tj/, but the word is actually an ancient dual, as noted in the CED. Under PE *ucuɣ ‘sexual organ’ the CED notes “NAI usuk ‘penis’, (du.) utcuk ‘vagina’”. All of the CED Inuit entries have a single consonant for ‘penis’ and double for ‘vagina’. I suspect the NAI utcuk spelling may contain [cç]. As all of the forms for ‘vagina’ listed in the CED contain only voiceless consonants, the in Utku utjuk looks suspicious. The statement that /tj/ “is pronounced somewhat like [tž]” (Dyck & Briggs 2005: 315) may reflect transcriptional difficulty with a [cç] cluster. Conor Cook, working on the same materials a few years later, while discussing the unusual cluster [ʔj] in Utkuhiksalik states, “j does not normally occur in clusters at all in this dialect” (Cook 2010: 7). The underlying forms in Utku are likely /ucuk/ ‘penis’ and geminated dual /uccuk/ ‘vulva’, phonetically [uçuk] and [ucçuk]. Utku natsiq ‘seal’ and natsiaq ‘baby seal’ (Dyck & Briggs, 2005: 316) should both also contain [cç] like the Natsilik. The CED gives PE nayyiʀ ‘ringed seal’ with -yy-, but as all Inuit forms have a voiceless cluster, PI looks like *nacciq, cf. SPI Qaw nassiq, NAI natciq [Nu natciʀžuaq ‘sealskin coat’], WCI, ECI natsiq, GRI natsiq, Petitot natçeϱk ‘phoque barbu’. In a discussion of the Utku postbase psʸaaq- ‘more’ (from PE *mcaɣ- ‘finally or more?’), Dyck and Briggs (2004: 12, fn. 10) list the cognate Net phaaq- ‘again; more’. In a footnote to this cognate they write: “A Natsilik speaker, Janet McGrath (p.c.) reports that the < h > in this Nattilingmiutut postbase is a sound intermediate between [h] and [s], i.e. a palatal-like sound”. In the later version the authors give the post-base as psaaq- with the same Natsilik information (Dyck & Briggs, 2004:12). In relation to pipsi ‘dried fish’ Dyck and Briggs (2005: 24) note, “In /ps/ clusters, the /p/ is pronounced either as [p], [ɸ] or [f], and the /s/, as [s] or [sʲ]. The [s] is not fully palatalized to [š] in /ps/ clusters”. /ps/ likely contains [pç]. There is evidence for the palatality of /c/ in the phonetics of the reflexes of PE or PI *kc in Utkuhiksalik. Dyck and Briggs (2005: 328) state that, “while /kh/ clusters are often pronounced [xx], they sometimes have an ‘s-like’ timbre, which we transcribe as a fronted velar fricative [x̯ x̯ ɕ]”. Their examples are PI *makcaq- > Utku makhaq- [max̯ x̯ ɕaq-] ‘sing a lullaby’, PI *akcut ‘hard or with effort’ > Utku akhut- [ax̯ x̯ ɕut-] ‘try harder, do with effort’, PE *aʀcar- ‘grab or otherwise act excessively’13 > Utku akhaaq- [ax̯ x̯ ɕaaq-] ‘grab’. The phonetic development of /kc/ is [kc > kç > çç]. The symbol x̯ is Americanist for IPA ç.

2.4.5 PI *cc

It is possible that all of the PE words containing *yy have *cc in PI. The CED gives PE *əyyi(ʀ) ‘yolk’ with *-yy- but the Proto-Inuit here seems to contain *-cc-. All of the Inuit reflexes show two voiceless segments: SPI iksiq ‘yolk’, Nu itciaq ‘membrane inside eggshell or around internal organs’, Car ixxi ‘yolk’, Holman I iktiq, NBI iksi, ECI itsi, NG itsiq, GRI itsik ‘white of an egg’, Up itsiq. The CED states, “the westerly Inu forms with ks are unexplained” (132a). The Inuit forms may be understood if the proto cluster was *cc. As intervocalic [c] is becoming a sibilant in the daughter dialects, the cluster undergoes differing reinterpretations. The first segment, the palatal stop /c/ [c], is reinterpreted as either an apical /t/ or velar /k/. That is, the cluster */cc/ [cc], either evolves through [cç] > [tç] > [ts], or through [cç] > [kç] > [ks]. The eastern -ts- forms may also reflect usual velar assimilation (cf. Baffin iglu, Quebec illu). Another example is PE *iyyuʀ- ‘poke head out for a look’ which may be PI *iccuq, cf. SPI issuq ‘go out to check weather’, NAI itcuq- ‘go out after waking’, ECI itsu(q)- ‘go out to take a look’, GRI itsuʀ- ‘look outside for something’. The CED gives the related PI *itcuaq ‘peek in or out’ which might

13 Instead of listing PE *aʀcaʀ- as the proto-form for Utku akhaaq- ‘grab’, it would be better to use PI *aqcaaq- ‘remove by force from (s.o.)’, cf. NAI aqsaaq- ‘grab away from’, WCI aqsaq- ‘take away from’. This example suggests the reflexes of PI *qc and *kc have merged in Utkuhiksalik. More examples are desirable.

12 PROTO-INUIT PHONOLOGY better be listed as *iccuaq, cf. SPI issuaq-, NAI itcuaq-, WCI itsuaq-, ECI itsua(q)-, GRI itsuaʀ-. For PE *nayyiʀ ‘ringed seal’ a PI *nacciq was suggested above. There are also PE words with *cc. E.g.:- PE *acciʀ- ‘name’, (AAY, CSY aciʀ-), SPI assiq-, NAI atciq-, WCI atsiq-, ECI atsi(q)-, GRI atsiʀ-; PE nuccuɣ- ‘pull or tug’, NAI nutcuk- ‘pull out’, WCI nutsuk- ‘pull on’, ECI nutsu(k)- ‘draw toward oneself’, GRI nutsuɣ- ‘pull toward self’ (CED). Inuit dialects with more conservative morphology feature a type of plural and dual formation in which a single medial consonant is doubled, i.e., /VCV/ > /VCCV/. The consonant is also the onset of the last syllable in the word. With most consonants the relationship between the single and double consonant is transparently one of simply doubling. Some Iñupiaq singular~dual pairs: ‘tent’ tupiq~tuppak, ‘walrus’ aiviq~aivvak, ‘captain’ umialik~umiallak, ‘canoe’ qayaq~qayyak, ‘eye’ iri~irrak, ‘boot’ kamik~kammak, ‘person’ iñuk~iññuk (MacLean, 1986: 76–77). But where the consonant to be doubled is [s] the relationship is not transparent. In Siglitun orthography these words have s in singular and tch in dual and plural. Some singular~plural pairs: ‘shoulder blade’ kiasik~kiatchit, ‘sleeve’ asiq~atchit, ‘ hood’ nasaq~natchat’, ‘fish scale’ kapisiq~kapitchit, ‘neck’ qunisiq~qungitchit, ‘lake’ tasiq~tatchit, ‘anchor’ kisaq~kitchat, ‘word’ uqausiq~uqautchit, ‘navel’ qalasiq~qalatchit. Diachronically the relationship is transparent, */c/ > */cc/. But if we posit an underlying /c/ in the synchronic language, the relationship may be simply described also as /c/ > /cc/, e.g., /kicaq/~/kiccat/ ‘anchor’. After doubling, rules apply giving the surface form, i.e., /VcV/ > [VsV], and /VccV/ > [VtʃːV] (the latter if tch truly writes an alveopalatal geminate and not [cc]).

2.4.6 Strong i palatalization

In North Slope and Malimiutun alveolar consonants /l ɬ n/ are palatalized after strong i in all positions. Dorais (2003: 46–47) writes the results as, for instance, ilʸu ‘the inside of something’, sikłʸaq ‘pickaxe’, iñuk ‘human being’. At the same time, /t/ is palatalized to ch when final as in iqaluich ‘fish (pl.)’, or to s intervocalically as in isiqtuq ‘enters’. It appears that /t/ palatalizes to /c/, which is then, in intervocalic position, fed into the rule /c/ > [ç] > [s]. In Uummaqmiutun this latter rule is /c/ > [ç] > [h] as in ihiqtuq ‘enters’ (ibid.). Also in these dialects the orthographic cluster tch created through the palatalization of geminate tt after strong i as in nirinngitchuq ‘does not eat’ (ibid.) represents geminate /cc/, in a way parallel to the palatalization of ll after strong i to lʸlʸ as in ilʸlʸatir̂ uq ‘joins a group’ (ibid.).

2.4.7 Other information on /c/

Palatal pronunciations of the reflex of PI *c are found in Greenland. In Thule, “Particularly characteristic phonetically is the /h/ (varying from [h] to [ʃ] through palatal [ç] intervocalically and initially)” (Fortescue, 1983: 8b). And, “When geminate, the pronunciation [çç] or [ʃʃ] is most common” (ibid. fn. 35). In East Greenlandic the segment corresponding to initial “/s/ of W Greenlandic becomes a palatalized [ᵗs] (almost [tʃ])” (Fortescue, 1983: 8b). This description appears to be of [c]. Fortescue 1983 uses the symbol ‘c’ to denote this sound though not intending the IPA usage. Also in E Greenlandic “/s/ following an /r/ is always /c/” (ibid.: fn. 30). Instead of [c] evolving from an older [s], the possibility should be considered that this dialect, on the extreme eastern edge of Inuit territory, has preserved [c] from the proto-language. Apparently, some elderly North Baffin speakers still have [cc] where younger speakers now have [tt]: “they pronounce tj (or, perhaps, tch, as in natjiq or natchiq ‘seal’)” (Dorais, 2003: 97 p.c. from Alexina Kublu). In North Labrador [ts] is one of only two phonetic clusters with dissimilar first and second segments (utsuk ‘blubber’). The second is [kq], the usual pronunciation of /qq/ (akqutik ‘road’). All others are geminates (ibid.: 113). The preservation of [ts] may show that the cluster was or is viewed as a geminate already. PI *c in some environments in some dialects merges with PI *t, and vice versa. Aivilik shows *c > *t: “As a general rule, cluster ts does not exist. As in North Baffin, it is generally replaced by tt” (ibid.:

13 DOUG HITCH

79). For example, Kiv natsiq~Aiv nattiq ‘seal’, Kiv uvatsiaru~Aiv uvattiaru ‘in a moment’ (ibid.). Some S Baffin dialects show *t > *c: “In Iqaluit and Kimmirut, initial t tends to become s when the next syllable starts with a s (cf. sasiq ‘lake’, suqsuuk ‘house porch’, rather than Southwest Baffin tasiq and tuqsuuk)” (Dorais, 2003: 97).

2.5 Velar /k/ [k x]

The voiceless velar consonant may be a fricative in some dialects when it occurs in a cluster before the reflexes of PI *ʐ, *ɬ or *c. In Uummaqmiutun /kʐ/ is pronounced [xʂ] as in ikpakr̂ aq [ikpaxʂaq] ‘yesterday’ (cf. Lowe, 1984: xxii). In Kangiryuarmiutun, “the combination kh results … in a double velar fricative” (Lowe, 1983: xx). There is [xx] for kh in hinikhaq ‘fur trim on a parka’, anaakhiq ‘whitefish’, hakhagiaq ‘redpoll’ (ibid.). To this may be added the Copper [Pryde] form maxxak- ‘sing a baby to sleep’ from the CED. Historically Kang or Cop [xx] in these words may come from *kɬ, cf. NAI aanaakłiq ‘Arctic cisco’. Or it may derive from *kc, cf.: PI *cakcakiq ‘redpoll’, ECI satsaɣiaq ‘little bird; and PI *makcaq- ‘sing a baby to sleep’, NAI maksaq-, Sig maksaaq-, Cop [Pryde] maxxak-, old GRI [Fab.] maksar-. In Thule, corresponding to [ks] in many dialects, there is [xx] written gh, e.g., ikpaghaq [ikpaxxaq], cf. Aivilik ikpaksaq ‘yesterday’ (Dorais, 2003: 137). Another description is, “The combination of /g/ plus /h/ (orthographic ‘gh’) is a geminate velar fricative [xx]” (Fortescue, 1983: 8b).

2.6 Uvular /q/ [q χ]

Phonemic /q/ may be phonetically continuant [χ], either in all positions, or in clusters reflecting *qc or *qʐ. In Labrador and Arctic Quebec many speakers realize /q/ as [χ]. “This explains why the Nunavik schools teach children to write taarpat, ursuq and qurturaq, rather than standard taaqpat, uqsuq and quqturaq” (Dorais, 2003: 114). In Uummaqmiutun /qʐ/ is pronounced [χʂ], as in qaqr̂ auq [qaχʂauq] ‘red-throated loon’ (cf. Lowe, 1984: xxii). In Kangiryuarmiutun, “the combination … qh results in a double uvular fricative” (Lowe, 1983: xx). There is [χχ] for qh in miqhuqtuq ‘sewing’, uaqhiyuq ‘is washing something’, qaqhauq ‘red- throated loon’. In Thule these clusters are also [χχ] and written rh, e.g., urhuq [uχχuq], cf. Aivilik uqsuq ‘blubber’ (Dorais, 2003: 137). Another description of the Thule clusters is that, “/r/ plus /h/ (orthographic ‘rh’) [is] a geminate uvular fricative [χχ]” (Fortescue, 1983: 8b).

3 Voiced consonants

3.1 Labial /v/ [v b β w]

In the eighteenth century spellings from words heard in Labrador by French travellers the /v/ is sometimes written with b or p. In the items offered by Dorais (1993: 42–43), /v/ occurs just five times, written always either b or p, not v, which is notable since French /v/ is written v.

Table 7: /v/ in eighteenth century Labrador Date Original transcription Phonetic Dorais’s analysis 1717 ibiéné ‘the breast’ [ibiaŋiq]] iviangiq ‘the breast’ 1717 calquipia ‘the lips [kakkibiaq] kakkiviaq ‘the upper lip’ 1730 oubignarou ‘a shirt’ [ubiɲiruq] uviniruq ‘shirt’ — tibougalo — [kibᵘɣaluk] kivgaluk ‘muskrat’ — tibailloc — [tibajuq] tivajuq ‘dances’

14 PROTO-INUIT PHONOLOGY

Although this list is short, it is clear that sometimes, if not always, /v/ was phonetically bilabial. In Table 6 the segment is phonetically transcribed as a stop [b] but it is possible that it was a continuant [β] (see below for voiced bilabial continuants elsewhere in Inuit). As the first element in a cluster, /v/ is [b] in some dialects. Kangiryuarmiutun has [b] in the cluster [bl], e.g., orthographically ublumi ‘today’, abluqtuq ‘takes a step’, ubluriaq ‘star’ (Lowe, 1983: xix). Siglitun has [b] in both [bl] and [bj]. There is [bl] in kublu ‘thumb’, ublumi ‘today’, publa ‘bubble’, tablu ‘chin’. There is [bj] in isibyuktuaq ‘whispers’, qibyauyaq ‘string’, kubyaq ‘fishnet’ (Lowe, 2001: xvii). A possible third instance of Siglitun [b] may be in [bb] from PE *vv. Petitot’s tibukiyaϱneϱk ‘baver’ (‘slobber, drool’) has orthographic b seemingly for [bb] (/vv/). It reflects PE *təvvuɣ- ‘spit’, cf. NAI tivvuk- ‘spit out’, GRI tiffuk ‘splash, squirt’. But Petitot has other words with etymological *vv which are written with v: Pet avaϱkϱ ‘moitié’ PI *avvaq ‘half’, WCI avvaq; Pet kiviktoaϱk ‘soulever’ PI *kəvvak- ‘lift gradually’, Sig kivvak- ‘lift up’. Words with etymological *vv are uncommon. The other cases noticed of Petitot writing intervocalic b are purely orthographic. He seems to write VbiV for [VbjV], e.g., itçibiuktoaϱk ‘chuchoter’ with /icibjuk-/ for PE *əcəmðuɣ- ‘whisper’, NAI isivžuk-, WCI isivyuk- (more examples below).

3.1.1 Utkuhiksalik [bʐ], /pʂ/, and /pc/

Utkuhiksalingmiutut has the cluster [bʐ]. Dyck and Briggs (2004) have chosen to write this cluster př. They list three examples: Utku apřaq ‘mattress; caribou hide mattress’, PE *avðaʀ ‘hindrance or protection’, NAI avžat ‘bedding’, GRI aššaq ‘defense, shield, curtain’; Utku ipřit ‘you’, PE *əlpət or əłvət ‘you’, Net ilvit, ižvit, GRI old ortho. ivdlit ‘you’; Utku tapřa ‘here’ right now’, PE *taðva ‘there (you are), Net tavža ‘there (you are), NAI tavža, tažva ‘that’s it, enough’ (Dyck & Briggs 2004: 13, with extra dialect examples from the CED). About the cluster written př Dyck and Briggs (ibid.) state, “/př/ is pronounced [b̥ ᵒř]; the first segment is a partly- to fully-voiced labial stop that has a lenis release when before /ř/”.14 This is an extremely useful phonetic comment that is not found in the 2005 version of the paper. It is clear that the cluster phonetically is neither [vʐ] or [pʐ] but is [bʐ]. Phonemically the cluster is /vʐ/. Dyck and Briggs have in contrast chosen to associate phonetic [b] instead with the phoneme /p/, writing the cluster phonemically as /př/ and orthographically as př. Implied in this choice is the idea that continuance is the distinctive phonemic feature rather than voicing. By using orthographic p and phonemic /p/ for the stop [b], there is in theory a risk of not distinguishing [bʐ] (/vʐ/) from [pʂ] (/pʐ/). Both would be written př. Instead, Dyck and Briggs have chosen to write [pʂ] as ps. While this avoids a conflict between [bʐ] and [pʂ], it creates a conflict between [pʂ] and [pç], confusing the retroflex and palatal sibilants. They do somewhat recognize the phonetic presence of both. They describe the retroflex sibilant as “rhotacized”, even putting the word in italics:

Isolated *labial-ð clusters, notably some *vð clusters, became a /ps/ cluster with a rhotacized /s/ cluster in Utkuhiksalingmiutut. In such clusters, the /s/ is consistently pronounced with a distinct rhotic quality as [psʴ] or [pʃʴ]. (Dyck & Briggs 2004: 29 fn. 21)

In the later version they repeat this but now use a subscript circle to show voicelessness: “The /ps/ cluster … consistently has a distinctly rhotic quality, and is pronounced as [pšʴ̥] or [šʴ̥]” (Dyck & Briggs 2005: 318–319). Their one example is “Utku taapsuma [taaps⁽ʴ̥⁾uma] ‘of that one’s batch’”, cf. NAI taavžuma, taafšuma, taaptuma (CED s.v. demonstrative PE *uv-, fn. to prefixed adverb tavžani), WCI taaffuma, Sig taavyuma (ibid. fn. to prefixed abs. sing. pronoun), GRI taššuma (ibid. fn. to prefixed abs. sing. pronoun). They describe the palatal sibilant [ç] in a number of ways, none of which is retroflex:

14 Cook (2010: 2 fn.4) writes this cluster [bɹ] which may mean that Utkuhiksalik scholarship now regards [b] as fully voiced. Cf. “/ɹv/ (including from underlying /lv/) normally undergoes metathesis, becoming [bɹ]”.

15 DOUG HITCH

PE clusters containing a labial (*m, *p, *v) plus *c (PI *s) typically became /ps/ in Utkuhiksalik (14). For example, PE *pimci dried fish became /pipsi/ (14.c). In /ps/ clusters, the /p/ is pronounced as [ɸ] or [f], and the /s/, as [s], [š], [sʸ], or [fʸ]. (2004: 11)

Their examples: Utku psʸaaq- ‘more’, PE *mcaɣ- ‘finally or more?’, Nu psaaq- ‘again’, NAI fsaaq- ‘again, more’; Utku apsʸak-tuq ‘a sound that’s not very audible’, PI *apcak- ‘make loud noise (by pounding)’; Utku pipsi-t ‘dried fish’, PE *pimci ‘dried fish’, WCI pipsi ‘dried fish’; Utku pui-psʸu-laaq- tuq ‘crackles (ice in warm water’, PE mcuɣ(-) ‘(a) little’, cf., NAI pifsukaq- ‘barely escape danger, barely miss hitting’, Cop [Pryde] kaffuk ‘a little while’ (2004: 12). They carry this ambiguous phonemic analysis over into the 2005 version. In a chart the labial+*ð cluster is defined as “/ps/ (rhotacized)” while the labial + *c cluster is “/ps/”. Instead, by recognizing that [ʂ] and [ç] reflect a phonemic contrast, the description may be simplified and nonambiguous phonemic and orthographic representations may be used. Table 8 displays various symbologies related to these Utkuhiksalik clusters. Note the ambiguity of ps in the Dyck & Briggs orthography. The suggested phonemic orthography is likely too abstract. The suggested practical orthography writes allophones known from English. A drawback is that there is no b symbol in syllabics, so vř might be preferred to maintain correspondence between roman and syllabics.

Table 8: Utkuhiksalik cluster symbology phonemic /pc/ /pʐ/ /vʐ/ phonetic [ɸç] [ɸʂ] [bʐ] Dyck & Briggs orthography ps ps př phonemic orthography pc př vř practical orthography ps př bř

Dyck & Briggs (2005: 319) propose that PE *vð became Utku /vj/ in kuvjat ‘net’. However, the cognates suggest that the Utku form would be *kuvřat [*kubʐat]: PE kuvðaʀ ‘net’, SPI, NAI kuvžaq ‘net’, WCI kuvyaq ‘net’. Stefansson’s (1909: 226) jargon entry kūb′-ra, kūb'-dja ‘a net – for fish or seal’ suggests Inuit sources with [kubʐaq, kubdʒaq] but these phonetic clusters seem to no longer exist in the west, cf. Uum kuvr̂ aq, Sig kubyaq ‘fishing net’. Petitot kϱubiaϱk ‘filet’ appears to have [bj]. Possibly an Utkuhiksalik alternate with [bʐ] still exists.

3.1.2 Other information on /v/

Petitot, who distinguishes b, v, w, has w sometimes for /v/: nuwiyoaϱk ‘enfiler’ reflecting PE *nuvə- ‘thread’, NAI, WCI, ECI nuvi- ‘thread’; uwaña, uvaña ‘moi’ for /uvaŋa/ ‘I’. Possibly intervocalic /v/ was sometimes [β]. In Bering Strait and Qawiaraq dialects, when following an unstressed syllable, /v/ is pronounced [w] (Dorais, 2003: 46); for example, SPI kiɣžawik ‘peregrine falcon’. In Kivalliq and Aivilik, the cluster /vl/ is pronounced [βl]. Dorais (2003: 78) describes the phone as “β sounding like Spanish b, i.e. halfway between English b and v”. The cluster is usually written bl in roman transcription, e.g., tablu ‘chin’, qablu ‘eyebrow’. Under certain conditions /v/ is bilabial in E Greenlandic and Thule (Fortescue, 1983: 8b and fn. 28). I am not sure what is meant by, “the special ‘double’ labio-velar articulation of /vv/ in Lichtenau” (ibid. fn. 28). Perhaps [ww] is intended. In Labrador /vv/ is pronounced [ff] and written ff: saffik ‘chest’ (N Baffin sagvik), affik ‘whale’ (N Baffin arvik) (Dorais, 2003: 113). Some older speakers in North Labrador “say qablunaak, rather than qallunaak (‘white person’), or ibjuk, rather than ijjuk (‘soil’)” (Dorais, 2003: 115).

16 PROTO-INUIT PHONOLOGY

3.2 Apical /ʐ/ [ʐ ʂ ɖ]

This segment probably offers the most challenges in description and analysis.15 It is the voiced correspondent of /t/. Both are apical, but the /t/ is phonetically alveolar while the /ʐ/ is phonetically retroflex. It is not necessary for segments which share phonemic place of articulation to share an identical phonetic place of articulation. In some descriptions of English phonology, the segments /s/ and /θ/ share a place of articulation and are distinguished by a feature of stridency, with /s/ being strident and /θ/ non- strident. All analyses of Inuit dialects assign /p/ and /v/ the same place of articulation even though /p/ is phonetically bilabial and /v/ is labiodental. Possibly the retroflexion of /ʐ/ reflects a need to increase the phonetic distance from /l/. /ʐ/ is in a synchronic voiced~voiceless relationship with /t/ in some dialects. In N Slope there is /t > ʐ/ before voiced consonants as in /naŋit-/ ‘be sick’ + /-vik/ ‘place’ > /naŋiʐvik/ naŋirvik ‘hospital’, /aqpat-/ ‘run’ + /-vik/ > /aqpaʐvik/ aqparvik ‘running track’, /aʁnat/ ‘women’ + /-ɣuuq/ ‘it is said that’ > /aʁnaʐɣuuq/ aġnarguuq ‘it is said that women …’ (MacLean, 1986: 16, Glossary of Stems). There is the same change between vowels. Compare the -tuq~-ruq variation in aqpat- ‘run’ aqpattuq ‘he runs’ and niġi- ‘eat’ niġiruq /niɣiʐuq/ ‘he eats’ (Kaplan, 1981: 173, 175). English borrowings into Iñupiaq show /ʐ/ for source [d]: para /paʐa/ ‘butter’, marasiq /maʐaciq/ ‘medicine’ (Kaplan, 1981: 209). In East Greenlandic *ʐ merges with /l/, cf. ili ‘eye’ (see below). In Inuinnaqtun (CED Copper), etymological *ʐ has merged with /l/ before another voiced segment (Dorais, 2003: 60), e.g.: algak (Cop alɣak) ‘hand’, PE *aðɣak, NAI ažɣak, Net ažɣak, GRI aššak; talva ‘over there!’, PE *taðva ‘there (you are)’, NAI tažva, NAI PH talva ‘that’s it, enough’, GRI tašša there (you are), that’s enough’, Utku [tabʐa] (see above); kilgavik ‘falcon’, PE *kəð(ə)kaviɣ ‘falcon or hawk’ (PI *kəʐɣavik?), NAI kižɣavik ‘goshawk, falcon’, GRI kiššavik ‘hawk’. The segment /ʐ/ does not occur initially or finally. Between vowels it is a continuant. In clusters the phonology is complex. After voiceless segments /ʐ/ devoices to [ʂ], likely already in Proto-Inuit. The cluster [pʂ] was discussed above, and the clusters [kʂ] and [qʂ] are treated in separate sections below.16 The phonetics of /ʐ/ as the first or second element in a voiced cluster are discussed in this section. In most Canadian dialects *ʐ and *j merge in /j/ when intervocalic or in voiced clusters. To illustrate the intervocalic pattern, Dorais (2003: 34) juxtaposes the reflexes of Proto-Eskimo *naja ‘sister’ and *əδə ‘eye’ in seventeen dialects. In most Canadian dialects these are /naja/ and /iji/. In Nunavik Itivimiut *ʐ and *j also merge but in /ʐ/: /naʐa/ and /iʐi/ (see below). In Alaskan dialects, Natsilingmiutut and Utkuhiksalingmiutut the segments retain their Proto-Inuit values: /naja/ and /iʐi/. In Greenland the segments do not merge. In East Greenlandic *ʐ merges with *l: /naja/ and /ili/. In , *ʐ first devoiced to /ʂ/ as shown by the older spelling issi for /iʂi/ ‘eye’, but is now /s/ for most speakers and spelled isi. In Thule ‘eye’ is ihi. For the pattern of PE *δ as the first element in voiced clusters, Dorais (2003: 36–37) has collected reflexes of PE *aðɣaɣ ‘hand’. The segment is still voiced and retroflexed in Alaska and Natsilingmiutut “aj̴gak” /aʐgak/. (It is also still retroflex in Utkuhiksalik ařgak (Dyck & Briggs, 2005: 316.) Elsewhere retroflexion is lost except in older W Greenlandic which has [aʂʂak] (CED GRI aššak; Dorais gives younger assak). Siglitun, Kivalliq and some Natsilingmiutut speakers have adjgak /ajɣak/ showing merger with /j/. Inuinnaqtun algak shows merger with /l/. Elsewhere in Canada there is complete assimilation to aggak. Thule aghak [axxak] shows further devoicing and E Greenlandic attak reflects complex phonological developments including loss of retroflexion.

15 Cf. “The symbol ‘š’ (‘sr’ in the Iñupiaq orthography) … is needed for the alveolo-palatal sound (approaching English ‘sh’, but retroflex like its voiced counterpart /ž/)” (Fortescue, 1983: 13a). 16 For PE *tð in a stem I have noticed only PE *qatðaʀ ‘bass or booming voice’, Nu qatžaq ‘echo’, GRI qaššaʀiɣ- ‘have a good deep voice’. Presumably the Nu form would contain [tʂ]. It is plausible that Uummaqmiutun avatr̂ aq ‘rear flipper’ would also feature [tʂ]. I find no cognates for this word.

17 DOUG HITCH

As the second segment in a voiced cluster, *ʐ remains continuant in most places. The CED offers PI *naɣžuk (i.e. /naɣʐuk/) ‘antler’ with retroflex continuant shown in SPI, NAI naɣžuq, GRI naššuk, and palatal continuant in WCI naɣyuk, ECI nayyuk. Petitot (1876: vii) wrote ‘antler’ in several ways: nagiuk ‘corne’, nakdǰiulik ‘cornu’, nagyuk, nagǰuk ‘antenne’, and nagdǰiuk ‘corne’ which implies that at that time the second segment varied between continuant and non-continuant. Presumably he was sometimes hearing the stop allophone [ɟ] of /j/. Cf. modern Siglit nagyuk ‘antler’. Another example is PE *qaʀðuʀ ‘arrow’, SPI, NAI qaʀžuq ‘arrow, bullet’, WCI, ECI qaʀyuk, GRI qaʀšuq. Compare also Natsilik ᖃᕐᖪᖅ qarřuq [qaɢᵃʐoq] ‘bullet’ (Hitch 1994a:8). Petitot kϱaϱioϱk ‘flèche’ appears to have [ʁj] like modern Siglitun qaryuq. There may be some evidence for a voiced stop allophone [ɖ] for *ʐ. I did not pay attention to the possible existence of [ɖ] in my 1994 Natsilik work. But it now appears we may have collected evidence. We recorded ᑕᑦᕙᐃᓐᓇᖅ tatvainnaq ‘now, at this moment, immediately’ (Hitch, 1994b: 8). This reflects PE *taðva ‘there (you are)’. Our Natsilik spelling ᑕᑦᕙ- tatva- with a stop before v may reflect [taɖva-] /taʐva/. Synchronically the cluster is voiced, but it is not clear if the first segment is retroflex [ɖ] or has merged with the alveolar [d] allophone of /l/ (see below). Our transcription was non-phonemic, following the syllabics practice of using the t final ᑦ for preconsonantal [d], for example, S Baffin ᐃᑦᓗ [idlu] /illu/ ‘house’. Utku iřřittuq ‘his eye is injured’ (Dyck & Briggs, 2005: 313), may feature [ɖʐ]. Under PE *əðə ‘eye’, the CED lists PI *əžžit- ‘get s.th. in one’s eye’, SPI, NAI ižžit-, Net itži ‘dust in the eye’ [Dor.], B. Lake iyyia- ‘’get dust in one’s eye’, ECI iyyi(t)- and iyyi ‘dust in the eye’, GRI iššit- ‘get s.th. in eye’ and iššia- ‘get dust in eye’. The Natsilik [Net] form itži in the CED from Dorais may show [iɖʐi] since /tʐ/ would be [iʔʐi]. Similarly, Utku uřřiqtuq ‘he feels anxious, worried’ (ibid.) may show [ɖʐ]. It reflects PI *užžiq- ‘show what to do (by signs?)’ (CED s.v. PE *uðəɣ- ‘try’); cf. NAI užžiqsuq- ‘give directions’, WCI uyyiqtuq- ‘watch over (so that s.th. is not done to excess)’, ECI uyyituq ‘watch over (to protect)’, GRI uššiʀ- ‘try, tempt, point at’. It is possible that the Utkuhiksalik pronunciations of ařgak ‘hand’ and aqiřgiq ‘ptarmigan’ (Dyck & Briggs, 2005: 316) might contain [ɖɡ] for some speakers. Dyck and Briggs (2005: 316, n. 24) offer the intriguing footnote: “Utkuhiksalik /řɡ/ is sometimes still pronounced as [ðɡ] (e.g., [haunaaðɣit] ‘eggshells.’)”. This is the only place I have noticed a [ð] phone described for any Inuit language. The devoicing of *ʐ after voiceless segments is well known from Iñupiaq where the orthographies use sr to write the voiceless retroflex [ʂ] allophone (possibly phonemic in Kobuk; cf. Kaplan, 1981: 26). It was not clearly known for Natsilingmiutut until the 1990s.17 Appendix F lists forms with [kʂ] and [qʂ] reported in Hitch, 1994a:5–618 together with proto-forms and cognates. Dyck and Briggs (2005: 326) are aware of the Natsilik devoicing from the 1994 project. But for Utkuhiksalik they state that the reflex of PE *ð after /k/ or /q/ is /s/, “pronounced as [s], [š], or [sʲ]” (ibid.). They earlier transcribed the Utkuhiksalik descendant of PE *kðaʀ ‘(s.th. for the) future’ as -ḵsaq; ḵšʸaq ‘potential; material for’ (ibid.:21) If the sibilant transcriptions are correct, then this implies a loss of retroflexion and a merger of /c/ and /ʐ/ after /k/ and /q/. This merger would contrast with the preservation of the /c~ʐ/ contrast after /p/ (see above, 3.1.1). Dyck and Briggs also note that, “In /ks/ and /qs/ clusters, the plosives /k/ and /q/ are realized as [x] and [χ]” (ibid.).

17 There are transcriptions in Knud Rasmusson’s work on the Netsilik which we can now identify as showing [ʂ]. One example is qaɔrʃuAq ‘the big brow’ (Rasmusson, 1931: 221) which reflects [quaqʂuaq]; cf. WCI, ECI quaq ‘forehead’ + Nats. -řuaq ‘big’. Another is unalErʃuAq ‘the great Indian’ which is [unaliqʂuaq], reflecting PE *unaliʀ ‘person from another tribe or from the coast?’, WCI Car unaliq ‘Cree Indian’ also with Nats. -řuaq ‘big’. 18 Possibly qarřukřaq ‘fish hook’ in Hitch (1994a) is an error for qarřuqsaq(?) so is not included in Appendix F. Compare PI *qaʀžuqcaq ‘jigging hook’, NAI qaʀžuqsaq ‘skin-covered hook left in water’, Cop qaʀʀyuXXaq ‘special hook for jigging for salmon [Ras. also for Net as ‘fish hook’], Lab qaʀʀyu(k)saq ‘bait’, GRI qaʀsuʀsaq ‘hook (esp. for fish)’.

18 PROTO-INUIT PHONOLOGY

In contrast, after the reflexes of the etymological voiced velar and uvular segments they observe that the retroflexion is retained. That is, PE *ð > Utku ř after PE *ɣ and *ʀ. However, they do not regard the resulting clusters as phonemically voiced. Instead, they have chosen to represent them as phonemic /kř/ and /qř/, and orthographic kř and qř. They offered slightly different statements on these clusters in 2004 and 2005:

“As for pronunciation, in synchronic clusters with a plosive plus /ř/, the plosive is generally voiced; as well, there is a lenis release between the plosive and the /ř/, so that /kř/ is pronounced as [gᵒř], and /qř/ as [Gᵒř]” (Dyck & Briggs, 2004: 23).

“/kř/ is pronounced as [gř] and /qř/ is pronounced as [qᵒř]; the first member of the [gř] cluster is lenis and at least partly voiced; the /qř/ [qᵒř] cluster has a lenis release between the plosive and the /ř/” (Dyck & Briggs, 2005: 327)

From these descriptions, it would be reasonable to represent these clusters simply as phonetic [ɡʐ] and [ɢʐ] and phonemic /ɣʐ/ and /ʁʐ/. For [uɡʐuk] ‘bearded seal’ they write ukřuk (Dyck & Briggs, 2004: 22), cf., PE *uɣðuɣ, NAI uɣžuk, WCI uɣyuk GRI old ortho. ugssuk. For [aɢʐaq] /aʁʐaq/ ‘ash, gunpowder’ they write aqřaq (Dyck & Briggs, 2005: 327), cf. PE *aʀða ‘ash’, NAI aʀža ‘ash, gunpowder’, Uum ar̂ ra, Sig arya ‘gunpowder’, GRI aʀšat ‘ash’. In those dialects where /ʐ/ has merged with /j/, the result is mostly /j/ (see above). The exception is Itivimiut (Arctic Quebec, Hudson Bay coast) where the result is /ʐ/ (examples below). The reflexes of PE *nayaɣ ‘man’s younger sister’ have /j/ in all dialects except for Itivimiut. The reflexes of PE *əðə ‘eye’ retain /ʐ/ in Alaska, Uummaqmiutun, Natsilingmiutut, and Itivimiut. The retroflexion was retained in West Greenlandic until recently, where the word was issi [iʂi]. In Thule, /ʐ/ has merged with /h/, ihi, and in East Greenlandic it has merged with /l/ between vowels, ili. But in the rest of Canada /ʐ/ has merged with /j/, iji (cf. Dorais, 2003: 34). Some examples of the preservation of /ʐ/ in Itivimiut, contrasted with Tarramiut (Arctic Quebec northwest coast) /j/ (in Dorais’s transcription with j̵ for /ʐ/): Tarr najaarjuk~Itiv naj̵aarj̵uk ‘small sister’; Tarr nauja~Itiv nauj̵a ‘seagull’; Tarr ikajuqpuq~Itiv ikaj̵uqpuq ‘helps’; Tarr iji~Itiv ij̵i ‘eye’ (Dorais, 2003: 114). One or two generations ago, most Central West Greenlandic speakers contrasted two voiceless sibilants, [s] written s and a second one written ss. The second one is described as alveo-palatal [ʃ] by Fortescue (1983: 5b). It reflects PI *ʐ and PE *ð. In the CED it is written š, e.g., GRI iši ‘eye’, PE *əðə, NAI iži, ECI iyi; GRI uššuk ‘bearded seal’, PE *uɣðuɣ, NAI uɣžuq, WCI uɣyuk; GRI ippaššaq ‘yesterday’, PI *ikpakʐaq (< PI *ikpaq ‘yesterday’ + PE *kðaʀ) ‘yesterday’, NAI ikpakšaq. The modern W Greenlandic spellings of these words are isi, ussuk and ippassaq. The segment *ʐ was likely already voiceless after voiceless consonants in Proto-Inuit. At the earliest stage of devoicing of intervocalic *ʐ and of voiced clusters with *ʐ in W Greenlandic, the segment was possibly retroflex before shifting to alveopalatal.

3.3 Lateral /l/ [l d]

The allophone [d] is found for /l/ in some dialects as the first member of clusters. The cluster /ll/ is phonetically [dl] in some dialects. Stefansson (1909: 224) reported Mackenzie Eskimo aud-la′k-tu-ak ‘he goes, travels’ which is Siglitun aullaqtuaq (Lowe). Similarly, his Mac. Esk. kō-ōd′-lĭk ‘a lamp’ (ibid.: 226) is Sig ‘lamp’, Uum qulliq ‘traditional stone lamp’. Petitot had kotluk with tl. Where North Baffin and west has /iɣlu/ [iɡlu] or [iɣlu] ‘house’, South Baffin and the rest of Canada has /illu/ [idlu] or [illu]. In older W Greenlandic /l/ could have a voiced stop [d] or affricate [dl] pronunciation. Compare older mardluk ‘two’ and tatdlimat ‘five’ (Barnum, 1901: xix) with modern GRI maʀłuk and tałłimat.

19 DOUG HITCH

In Thule there is [iglu] but younger speakers have “[idlu] with what sounds like a long tapped /l/” (Fortescue, 1983: 8b and fn. 34). Also, “/bl/ is found but more often than not sounds like [ʷdl] as in qaʷdlunaaq” (ibid.: fn. 34). The Natsilik word for ‘sledge runner’ we transcribed ᐱᑦᕌᖅ pitraaq (Hitch, 1994a: 8). This is PE *piʀlaɣ ‘sled runner’, Nu piʀlʸa<ž>aq ‘long sled slat’, GRI piʀłaaq ‘sled runner’. With cluster metathesis some cognates are WCI pilraak (dual) ‘sled runners’, Car pelrat ‘sledge shoes’ [Ras.], pilraaq ‘sledge shoe’ [B.-S.], ECI piʀʀaaq ‘metal strip for sled runner’. Our pitraaq was probably [pidʁaaq] for /pilʁaaq/. Our spelling was likely influenced by the syllabic system which does not have a dedicated symbol for [d]. Phonemically, the syllabics would be ᐱᓪᕌᖅ pilraaq. In Thule, “both single and grouped l are either pronounced as a ‘flap-l’ or as d (single or geminate only)” (Dorais, 2003:138). Some comparisons with Aivilik in Dorais’s transcription: Thule ilua/idua~Aiv ilua ‘its inside’; Thule /udu~Aiv ulu ‘woman’s knife’; Thule ullu/uddu~Aiv ublu ‘day’; Thule tullia/tuddia~Aiv tullia ‘its next’ (ibid.). PE * t becomes /l/ intervocalically in East Greenlandic (CED: xvi).

3.4 Palatal /j/ [j ɟ ʝ]

The segment /j/ may be a palatal stop [ɟ] or fricative [ʝ] in clusters in some dialects. In Uummaqmiutun double /cc/ and double /jj/ have parallel phonetics. According to Lowe (1984: xxii), who is writing for a general audience, “Consonants ch and dj are found either double or preceded by consonant t. In both cases, they are written tch and tdj respectively”. This seems to mean that tch can stand for /tc/ or /cc/ and tdj for /tj/ or /jj/. It is possible to phonetically distinguish [tc] from [cc] but I am not sure any language would do so. It would be easier to distinguish [tj] from [jj] but since all clusters appear to have uniform voicing in this dialect, a *[tj] does not exist. Phonetically, the clusters may be closest to [cc] or [cç] and, in parallel fashion, to [ɟɟ] or [ɟʝ]. Lowe gives three Uummaqmiutun voiceless cluster examples (11–13) and two voiced (14–15):

(11) aatchauqtuq ‘is yawning’. The closest cognate to the stem is NAI aatcauq- ‘yawn’. (12) natchiq ‘seal’. This is also spelled natchiq in Iñupiaq. In Natsilik the phonetic shape is [nacçik]. The CED gives PE *nayyiʀ ‘ringed seal’ but PI appears to be *nacciq, Qaw nassiq, NAI natciq, WCI, ECI, GRI natsiq. (13) tikitchuq ‘arrived’. In most dialects the stem is tikit- and the inflected form tikittuq. Uummaqmiutun, as a dialect of Iñupiaq also features strong i palatalization. In this case, [tt] becomes [cc]. In Barrow the form is also written tikitchuq19. (14) qatdjat ‘canoes’. This is a plural with a geminated single medial consonant, i.e., singular /qajaq/, plural /qajjat/. The form is spelled qayyat in Alaska (cf. Kaplan, 1981: 234) which presumably means the cluster there is pronounced, at least by some speakers, as [jj]. In Uummaqmiutun /jj/ is [ɟɟ]. (15) putdjukkaa ‘pinched him’. The base putdjuk- corresponds to NAI puyyuk- ‘pinch’ (PE *pumyuɣ-). As above, Alaskan [jj] is Uummarqmiut [ɟɟ].

Nineteenth century Siglitun had [ɟɟ] for /yy/. Petitot wrote pudǰuaϱtoaϱk ‘pincer’ for the verb ‘pinch’ and pudǰiuk (s.v. pincé) presumably for modern putdjuuk ‘grip of thumb and index’ (Lowe, 2001: 99). In these words he used dǰ and dǰi for [ɟɟ] /jj/. He described the digraph dǰ in the introduction to his grammatical sketch: “DJ̌ , a une prononciation mixte entre DJ et DZ. Prononcez l’une ou l’autre de ces consonnes doubles, les dents serrées” (ibid.: xxxix). Although this description lacks modern phonetic precision, it seems clear that the sound is different from palato-alveolar dj [dʒ] and alveolar dz [dz].

19 About the last two examples in Iñupiaq Kaplan (1981: 90) wrote, “Similarly, č may result from palatalization (tikitchuq ‘arrives from tikiṭ- ‘arrive’ + -tuq 3s) or may not (natchiq ‘seal’). Where no alternation is present, as in natchiq, č must be underlying”.

20 PROTO-INUIT PHONOLOGY

The example of ‘pinch’ is the only one I can find in Petitot with historical /j/ in a cluster (PE *pumyuɣ-). PI */ʐ/ and */j/ have merged in Siglitun /j/. All other examples I have found in Petitot of clusters with /j/ reflect PE *ð or PI *ž (/ʐ/):

(16) kϱoϱk-dǰiuk ‘cygne’ [quɡɟuq]. PE *quɣðuɣ ‘swan’, NAI quɣžuq, WCI quɣyuq, ECI quyyuq, GRI quššuq [old ortho. qugssuk]. (17) nakdǰiuk ‘corne’ [naɡɟuk]. PI *naɣžuk ‘antler’, NAI naɣžuk, WCI naɣyuk, ECI nayyuk. (18) ogǰiuk ‘phoque (veau marin)’ [uɡɟuk]. PE *uɣðuɣ ‘bearded seal’, NAI uɣžuk, WCI uɣyuk, ECI uyyuk, GRI uššuk [old ortho. ugssuk].

All of these items now have [j] where Petitot wrote dǰ or ǰ: Siglitun ugyuk ‘bearded seal’, nagyuk ‘antler’, qugyuk ‘whistling swan’. Siglitun kubyaq ‘fishing net’ may also have earlier featured [bɟ] as shown by Stefanssons’s (1909: 226) Mac. Esk. kūb-djak ‘a net’. However, Petitot’s kϱubiaϱk ‘filet’ has bi which looks like a spelling for [bj]. In fact, it appears Petitot consistently writes bi for [bj] /vj/. I have not noticed any spellings like *bdǰ, *bǰ, or *bǰi while bi is common. Some other examples:

(19) ibiaϱiaϱk ‘portage’ /ivjaʁaq/. PE *ətəvyaʀaʀ, NAI itivyaaq. Siglitun ibyariaq. (20) itçibiuktoaϱk ‘chuchoter’ /icivjuk-/. PE *əcəmðuɣ- ‘whisper’, NAI isivžuk-, WCI isivyuk-. Siglitun isibyuktuq ‘whisper’. (21) kϱébiaϱk ‘natte’ /qivjaq/. PE *qipðaʀ ‘braided sinew’, SPI qipšaq, NAI qivžaq, Nu qipžaq ‘twine’, WCI qivyat ‘caribou sinew thread, tresses in the Indian style’, ECI qiyyaq ‘twist of thread or tobacco’, GRI qiššaq ‘thread’. Siglitun qibyauyaq ‘string’. (22) tçibiaϱk ‘hanche’ /civjaq/. PE *cipyaʀ ‘hip’, SPI sipyaq, Mal sipyaaq, NAI, WCI, ECI sivviaq, GRI siffiaq, Utku hiʔviaq. Cf. also curious Kang hilviaq ‘hip bone’.

The somewhat surprising conclusion from this is that Siglitun at that period mostly retained [ɡɟ] and that *[bɟ] had already transitioned to [bj]. I have not collected possible instances of [ɢɟ] but aϱia ‘cendre’ (Siglitun arya ‘gunpowder’, aryaq ‘ash’) does not look like it contains a stop. Dyck and Briggs (2005: 317) describe only one possible case of Utku /jj/ in the morpheme -jjak- which they relate to PE *-tyaɣ- ‘little or thing resembling something’, and for which they describe “a long [žž]-like sound”. This might be [ɟʝ] in parallel to [cç]. The few words I have noticed with PE *yy are discussed above (2.4.5 PI *cc). They all appear to have *cc in PI, like PE *nayyiʀ ‘ringed seal’ with likely PI *nacciq: Qaw nassiq, NAI natciq, WCI, ECI, GRI natsik. I notice one word with PE *ty which may be relevant. PE *katyaɣ ‘place where two things come together’, SPI kassaaq ‘fork in river’, NAI kayyaaq, Mal katyaaq. Petitot has katçaϱk ‘confluent’ (‘confluence’; with apparent [cc]) but kadǰiaϱék ‘bifurcation’ (‘fork’; with apparent [ɟʝ]). We appear to have two words here, one with voiceless cluster, and one with voiced. The voiced palatal stop phone [ɟ] appears to be present in Aivilik: “As second element in a cluster, phoneme j is generally realized y in Kivalliq (ugyuk, ‘bearded seal’), but dj in Aivilik (ugdjuk)” (Dorais, 2003: 79). Because of a general process of devoicing in Labrador, “jj is pronounced tj” (Dorais, 2003: 114). Presumably this means that ujjuk ‘bearded seal’ is pronounced [uccuk]. This [cc] cluster may play a role in the preservation of the cluster [ts] (cf. utsuk ‘blubber’) in N Labrador where otherwise all historical clusters have become geminates (ibid.: 113).

3.5 Velar /ɣ/ [ɣ ɡ]

Because all orthographies use g to write both allophones [ɣ] and [ɡ] of /ɣ/, we need to rely on descriptions to tell us where the segment may be a stop. Such detail in a description is rare. Fortunately,

21 DOUG HITCH there are now audio files available online for several dialects in which [ɡ] can be heard in some clusters. Some examples are given Appendix G.

3.6 Uvular /ʁ/ [ʁ ɢ ɴ]

No orthography has separate letters for the [ɢ] and [ʁ] allophones of /ʁ/ and descriptions are rare that note this phonetic distinction. A further complication with the description of /ʁ/ is that it also has a nasal allophone [ɴ]. Appendix H has audio files illustrating these phonetics. There are examples of [ɢ] and [ɴ] ins some dialects in some clusters. There are also examples of the cluster /ʁŋ/ which has challenging phonetics but in the examples the /ʁ/ is clearly nasalized.

4 Vowel epenthesis in consonant clusters

When the first consonant is a voiced non-nasal, many Alaskan Inupiaq speakers often insert a short vowel as in [ɑɢᵃnɑq] arnaq ‘woman’, [ɨɢⁱnɨq] irniq ‘son’, and [ilⁱvic] ilvich ‘you’ (Dorais 2003:43). This process may have been present historically in Siglitun. Stefansson (1909: 224, 230) reported Mackenzie Eskimo īg-nĕ′-ra-vik ‘a place for fire’ and cı̆ -nı̆ g′-a-vĭk ‘sleeping place’, both with unetymological ‘a’. Compare modern Siglitun ingnirvik ‘stove’, and sinigvik ‘bedroom’ (Lowe). Presumably Stefansson heard [ignɨʁⁱvik] and [ciniɣⁱvik]. He also heard unetymological ‘u’ in Mac. Esk. ta′-dju′-va ‘there!’ (Stefansson 1909: 231) for Sig tadjva ‘there!’ which suggests [taɟᵃva]. There may also be traces of vowel insertion in Petitot. aϱvénèlœϱit for modern arvinilgit ‘six’ seems to show [arvinilᵊɣit]. málœϱok for modern malruk ‘two’ seems to show [malᵊʁuk]. Similarly there may be evidence from Frobisher’s list. arered ‘eye’ (Dorais, 1993: 37) appears to be the plural with medial gemination [əʐᵊʐit], cf. Iñupiaq dual irrak [iʐʐak], plural irit (MacLean, 1986: 251). An eighteenth century French orthography spelling from Labrador, acquillat ‘ashes’, modern aryat (Dorais, 1993: 43), may stand for [aɢᵊʐat]. Dorais prefers [aʁjat] with ll as in fille (ibid.). Another, tibougalo ‘muskrat’, modern kivgaluk (ibid.), may stand for [kibᵘɣaluk]. In Natsilingmiutut there may be an epenthetic vowel in clusters beginning with glottal stop (and ending in a nasal?). For example, qamutitniittuq ‘he is on your sled’ sounds like qamutiʔⁱniittuq, and qatnguqtaqtuq ‘it rocks, moves, shakes’ sounds like qaʔᵒnguqtaqtuq (Hitch, 1994: 8). An epenthetic vowel may be heard in the following Inuinnaqtun audio files:

(23) iglu ‘house’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/inuinnaqtun/531.mp3 (24) arnaq ‘woman’http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/inuinnaqtun/1235.mp3

5 Conclusion

All Inuit dialects appear to have a distinctive feature of voicing which separates the non-nasal consonants into two groups, phonemically voiceless and phonemically voiced. A feature of continuance is strictly phonetic. Continuance is predictable, although each dialect may have particular rules governing its appearance. Voicing rather than continuance was also distinctive in the proto-languages PI and PE. This analysis of the distinctive features in the modern and proto languages is contrary to the conventional view which contrasts stops and fricatives. Most modern dialects have an underlying voiceless palatal consonant /c/ rather than the generally assumed /s/. The palatality can still be heard in Natsilingmiutut and Utkuhiksalingmiutut in the realizations [c] and [ç]. There is also palatality in some cases in Thule and E Greenlandic. These phonetics may be more widespread. Some reports may not distinguish between palatal [c] and alveo- palatal [tʃ]. But it does seem that in some dialects the /c/ may be realized as alveo-palatal [tʃ], as is commonly reported. In Fish river Qawiaraq PI *c is consistently pronounced [tʃ]. Historical documents prove that some dialects historically had palatal pronunciations for /c/ where today they have alveolar [s]. It is clear that PI had palatal /c/ and not /ts/ or /tʃ/ as conventionally held.

22 PROTO-INUIT PHONOLOGY

The voiced segment corresponding to voiceless /t/ is in some dialects and PI best described phonetically as a retroflex voiced sibilant /ʐ/. The segment still exists in Iñupiaq, Natsilingmiutut and Utkuhiksalingmiutut. In Itivimiut PI *ʐ and *j have merged in /ʐ/. Older W Greenlandic distinguished two sibilants, [s] written s, and a second descended from PI *ʐ and written ss. At one time this second sibilant was likely retroflex [ʂ]. These observations simplify and improve the accuracy of the phonological analyses of the modern dialects, of PI, and of the evolution between daughters and parent. They may also help to improve our understanding of PE.

References

Barnum, F. (1901). Grammatical Fundamentals of the Innuit Language as Spoken by the Eskimos of the Western Coast of Alaska. Boston and London: Ginn & Company. Bobaljik, J. D. (1996). Assimilation in the Inuit Languages and the Place of the Uvular Nasal. International Journal of American Linguistics 62(4), 323–350. Compton, R. (2008). Contrast in Inuit Consonant Inventories (Generals paper in phonology). University of Toronto, Toronto. Retrieved from http://www.ensany.ir/storage/Files/20101125160650-13.pdf Cook, C. (2010). Unusual developments in Utkuhiksalingmiutitut consonant clusters: The troublesome case of the glottal stop. Presented at the 17th Inuit Studies Conference, Université du Québec en Abitibi-Témiscamingue. Retrieved from http://www.uqat.ca/isc-cei-2010/publications/Cook_CEI- ISC-2010.pdf Dorais, L.-J. (1993). From Magic Words to Word Processing. A History of the Inuit Language. Iqaluit: Arctic College - Nunatta Campus. Dorais, L.-J. (2003). Inuit Uqausiqatigiit. Inuit Langauges and Dialects (second, revised edition). Iqaluit: Arctic College. Dyck, C. J., & Briggs, J. L. (2004). Historical developments in Utkuhiksalik phonology. St. John’s Newfoundland. Retrieved from http://homes.chass.utoronto.ca/~inuit/UIDP/Historical%20Developments%20in%20Utku Dyck, C. J., & Briggs, J. L. (2005). Historical antecedents of /h/, /s/, /j/ and /ř/ in Utkuhiksalik (). Études Inuit Studies 29(1–2), 307–340. Fortescue, M. (1983). A comparative manual of affixes for the Inuit dialects of Greenland, Canada, and Alaska. Copenhagen: Nyt Nordisk Forlag. Fortescue, M., Jacobson, S., & Kaplan, L. (2010). Comparative Eskimo Dictionary, With Aleut Cognates (second edition). Fairbanks: University of Alaska Fairbanks. Hitch, D. (1994a). Natsilik Dictionary Pilot Project, General Report. unpublished manuscript, Yellowknife. Hitch, D. (1994b). Natsilingmiutut Dictionary Pilot Project. Tusagatsait 5, 4–6. Kaplan, L. D. (1981). Phonological Issues in North Alaskan Inupiaq. Fairbanks: Alaska Native Language Center, University of Alaska Fairbanks. Retrieved from https://ia601305.us.archive.org/5/items/ERIC_ED398760/ERIC_ED398760.pdf Lowe, R. (1983). Kangiryuarmit Uqauhingita Numiktittitdjutingit. Basic Kangiryuarmiut Eskimo Dictionary. Inuvik, : Committee for Original Peoples Entitlement. Lowe, R. (1984). Uumarmiut Uqalungiha Mumikhitchir̂ utingit. Basic Eskimo Dictionary. Inuvik, Northwest Territories: Committee for Original Peoples Entitlement. Lowe, R. (1985). Siglit Uqausiita Ilisarviksait. Basic Siglit Inuvialuit Eskimo Grammar. Inuvik, Northwest Territories: Committee for Original Peoples Entitlement. Lowe, R. (2001). Siglit Inuvialuit Uqautchiita, Nutaat Kipuktirutait Aglipkaqtat. Siglit Inuvialuit Eskimo Dictionary, Second Edition, Revised & Expanded. (Second revised and expanded edition). Québec: Éditions Nota bene.

23 DOUG HITCH

MacLean, E. A. (1986). North Slope Iñupiaq Grammar, First Year (Third edition, revised). Fairbanks: Alaska Native Language Center, University of Alaska Fairbanks. Pálsson, G. (2001). Writing on Ice, The Ethnographic Notebooks of Vilhalmur Stefansson. Hanover and London: University Press of New England. Petitot, É. (1876). Vocabulaire Français-Esquimau, Dialecte des Tchiglit des Bouches du Mackenzie et de l’Anderson, précédé d’une Monographie de cette Tribu et de Notes Grammaticales. Paris: Pinart. Retrieved from https://ia600206.us.archive.org/24/items/vocabulairefran00goog/vocabulairefran00goog.pdf Pirurvik Centre. (2017). Tusaalanga. Learn the Inuit Language. Glossary. Retrieved from www.tusaalanga.ca/glossary Rasmussen, K. (1931). The Netsilik Eskimos, Social Life and Spiritual Culture (Vol. 1–2). Copenhagen: Gyldendalske Boghandel, Nordisk Forlag. Stefánsson, V. (1909). The Eskimo Trade Jargon of Herschel Island. American Anthropologist 11(2), 217–232.

Appendix A: Historical *t > Utkuhiksalik ʔ before *m, *n, *ŋ, *v, *ð/*ʐ, *l, *y

All Utkuhiksalik forms discussed below are from Cook (2010) unless otherwise noted. All other dialect forms and the reconstructions are from CED under the PE base alphabetical listing except where otherwise noted.

PI *tm > ʔm tuʔmaqtaqqut ‘we trod on it’. PE * tutmaʀ- ‘tread on (repeatedly)’, AAY, CAY, NSY, CSY tutm̥ aʀ-, SPI tupmaq-, NAI tunmaq-, tuŋmaq-, PH tummaq-, Nu tutmaq-, Sig tunmaq-, ECI tumma(q)-, GRI tummaʀ-, GRI old ortho. tungmar-. PE *tm is continued in Inuit by Nu tutmaq-, in nasal-assimilated form by NAI, Sig tunmaq-, and as phonetic [ʔm] by Utku. It shows place assimilation in SPI tupmaq, and place and manner assimilation in PH tummaq-, ECI tumma(q)-, and GRI tummaʀ-. The ŋm shown by NAI tuŋmaq- and GRI old ortho. tungmar- presumably reflects the early neutralization of *nC and *ŋC as *ŋC (M. Fortescue p.c. 14.Oct.2017). Cf. also PE *tutməqaʀ ‘stair or step’, SPI tupmaq, Qaw tutmiʀaq, NAI tunmiʀaq, tuŋmiʀaq, ECI, GRI tummiʀaq, NG tuŋmiqqat ‘Orion’s Belt’. qaʔmaqtuqtuq ‘he calls to an animal’. PE *qalmaʀ- ‘call to dogs’, Cop [Pryde] qanmaq-, ECI qammatuq-, NBI qaŋmaq-, Lab [Peck] kangmar-, GRI qaʀmaʀ-, GRI [Fab.] kangmar-. Natsilik has the identical form qatmaqtuqtuq (=[ʔm]) ‘s/he calls the dogs’ (Hitch 1994a:7, 1994b:5). The CED PE form with -lm- is supported by CAY, CSY qalmaʀ- ‘summon a dog vocally’. But there is no trace of a lateral20 in Inuit and a PI form would be *qatmaq(tuq)-. There are no Alaskan Inuit cognates. The PI *tm is most directly supported by Cop [Pryde] qanmaq-, and Utku qaʔmaqtuq-. The forms with a velar nasal, NBI qaŋmaq-, Lab kangmar- [Peck], and GRI [Fab] kangmar-, reflect the early merger of *nC into *ŋC ([tm>nm>ŋm]; see above). GRI qaʀmaʀ- perhaps further shows [ŋm>ɴm] with the uvularity echoing the preceding [q] and following [ʁ]. kiʔmik ‘heel’. PE *kitmiɣ, AAY, CAY, NSY kitŋ̊ ik SPI, Mal kikmik, Qaw, Nu kipmik, NAI, Sig, ECI, NG, GRI kimmik, Cop, NBI kiŋmik, GRI old ortho. kingmik. Among the Inuit forms, the Utku has the most direct information on *t. For -ŋm- see above. Qaw and Nu kipmik show place assimilation [ʔm > pm]. Mal kikmik may show *ŋm>km. Probably PI *kitmik. qiʔmiq ‘dog’. PE *qikmiʀ, AAY AP, CSY qikm̥ iq, NSY qiqm̥ iraq. The Inuit forms suggest a PI

20 Possibly a similar development may be seen in the Inuit reflexes of PE *nałmi(nəʀ) ‘one’s own’ which suggest PI *natminəq: Mal natmiinʸ ‘close relative’, NAI nanminʸɨq* ‘peer, one’s own property, relatives’, Cop [Pryde] nanminiq ‘own’, WCI naŋminiq ‘one’s own relatives, possessions’, ECI, GRI namminiq. Also similiar is PE *məlŋuʀ ‘water beetle’ which is discussed below under miʔnguq ‘beetle’.

24 PROTO-INUIT PHONOLOGY

*qitmiq, following the pattern in the previous three examples. The alveolar is shown by Cop [Lowe, Pryde] qinmiq and Net, Car qi’miq. The *nC>ŋC neutralization is shown by Point Barrow qiŋmiq, WCI qiŋmiq, ECI qiŋmiq, and GRI qimmiq, GRI old ortho. qingmeq. Di qipmik, NAI qimmiq, Mal qipmiq, ECI qimmiq place or place and manner assimilation. PI *tn > ʔn aʔnuraaq 'clothing'. PI *atnuʀaaq ‘clothing’, Nu atnuraaq, NAI, WCI, ECI, GRI annuraaq. PE *tŋ > ʔŋ aʔniqtuq 'is hurt'. PE *atŋiʀ- ‘be in pain’ should perhaps be revised to *aqniʀ-21 and a PI *atniʀ-, reflecting *qn > *[ʔn] then reanalysis to *tn, seems plausible. The *tn in PI *atniʀ- ‘be in pain’, is clear in the daughter languages: Mal, Uum atniq- ‘hurt, get hurt’ (Imaq ŋ in atŋiq- ‘be hurt’ may show Yupik influence), NAI, WCI anniq-, GRI anniʀ- ‘have pain’. aŋuʔnuniq ‘temporary husband’ (not in Cook 2010 but presumed from CED Car, Net [B.-S.] forms). PE *aŋutŋunəʀ ‘woman’s male relative’. The tŋ is assured by Yupik AAY aŋutŋ̊ unəq ‘woman’s younger brother’, CAY, NAY aŋutŋ̊ unəq ‘brother’. The PI appears to be *aŋutnunɨq, with older tŋ becoming tn, presumably in echo of the n beginning the following syllable. *tn is clear in SPI aŋutnuniq ‘brother of woman’, Nu aŋutnunɨq ‘husband, male maternal parallel cousin’, and underlying in the [ʔn] reflected in Car, Net [B.-S.] aŋu’nuniq ‘temporary husband’. Other dialects show nn as in NAI aŋunnunɨq ‘husband, male maternal parallel cousin’, Igl [B.-S.] aŋunnuniq ‘temporary husband’, GRI aŋunnuniq ‘adulterer’. kiʔŋuřuq ‘it (boat, sled) capsizes or overturns’. PE *kitŋu ‘capsize’. PE *tŋ is reflected by AAY CAY kitŋ̊ u, SPI kitŋu-, NAI kinʸŋu-, Cop kinŋu- [Pryde]. The ʔŋ is seen in the Utku and in Net ki’ŋu [Pryde]. It is assimilated to kŋ in Nu kikŋu-, GRI kiŋŋu-. The pm in Qaw kipmu- may show ŋ > m before u. miʔnguq ‘beetle’. PE *məlŋuʀ ‘water beetle’, all Yupik (CAY NSY, CSY, Sir) have -lŋ- but possibly this is like PE *qalmaʀ- ‘call to dogs’ for which a PI *qatmaq(tuq)- was suggested above. A PI *mətŋuq is plausible from the Utku, Qaw minŋuq, NAI mɨnŋuq, Sig minŋuq. qaʔŋuqtuq 'moves from fixed position'. PE *qatŋuʀ- ‘pry (heavy object) loose?’ cf. Nu qatŋuq-, NAI qanŋuq- ‘move, budge a heavy object’, WCI qanŋuq- ‘unstick, detach’. PE *tv > ʔv qaʔvik ‘wolverine’. PE *qatvik or *qavciɣ, but PI must be *qatvik. The alveolar is shown by Cop qalvik, Net qa’vik [Pryde], and the Utku. There is place assimilation in Mal, Nu qapvik, NAI, Sig, Car, ECI qavvik, and GRI qappik. The SB, NG qaɣvik probably show manner assimilation *tv>*ʐv and then place resolution *ʐv>ɣv in dialects without /ʐ/. PI *tv > ʔv paʔviřuq ‘he is bothered’. PE *papði- ‘be bothersome’ is attested by Yupik CSY papši- ‘fidget’, AAY, CAY, papsi(ke)- ‘find bothersome’ and by Inuit Qaw papši- ‘bother’ and NAI papži-. In Utku we might at first blush expect to find *paʔři-. Instead, there appears to have been place feature metathesis

21 The CED proposes that PE *tŋ in *atŋiʀ- evolves to kŋ through assimilation, and to ŋq and qn either through influence from aŋqutə- ‘stumble’ or aqə- ‘kick (ball)’ or “due to the final uvular, undergoing metathesis in some languages”. The CED further suggests comparing PE *aŋutŋunəʀ ‘woman’s male relative’ for the Inuit developments. An alternate explanation may be simpler. The Yupik forms of PE *aŋutŋunəʀ, AAY, CAY, NSY aŋutŋ̊ unəq, show the regular development of *tŋ to tŋ̊ , and a proto-form PE *aqniʀ- permits a simpler, more motivated derivation of all forms. PE *qn in suggested *aqniʀ- is supported by Sir aqn̥ əʀ- ‘be hurt’, CSY aqn̥ iʀ- ‘be in pain’, and HBC aqn̥ iXtə-. Other Yupik forms show place assimilation, HBC aqŋ̊ iXtə- ‘hurt, get hurt’, or both assimilation and metathesis, AAY aŋqəXtə- ‘be in pain’, Nun aŋqiXtə-, NSY aŋqiʀ- ‘be ill’, where ŋ is presumably written for the uvular nasal [ɴ]. The development of the velar k in some CAY forms, akŋ̊ iXtə- ‘be in pain’, HBC akŋ̊ ia- ‘suffer pain’, akŋ̊ iq ‘nerve’ may have been influenced by the association of phonetic uvular [ɴ] with phonemic velar /ŋ/.

25 DOUG HITCH

/pʐ>tv/ resulting in an alternate or post PI stem *patvi- which gives Utku paʔvi-, ECI pavvinaq ‘be annoying’ (cf. next item). hiʔviaq 'ball of hip'. PE *cipyaʀ ‘hip’ with *py is supported by Sirenikski sipẙəX, Yupik NSY sipẙaq and Inuit SPI sipyaq, Mal sipyaaq. But Utku hiʔviaq and NAI, WCI, ECI sivviaq, and GRI siffiaq support a late or post PI *citviaq. The relationship between this and the PE form may involve a place feature metathesis /pj > cv/ and reanalysis > /tv/ similar to that in the previous item. PI *tʐ > ʔʐ (*tð > ʔř)22 aʔřuqtuq 'puts on outer clothes'. PI *atžuq- (i.e. *atʐuq-) ‘put on clothes’. Qaw aššuq- ‘keep putting on clothes’, WCI ayyuq- ‘disguise oneself in another’s clothes’, ECI ayyu(q)- ‘put on a coat, disguise oneself in another’s clothes’, GRI aššuʀ- ‘put on clothes—esp. fine Sunday ones’, EG [Tersis] atsuaq- ‘wear one’s best clothes’ (CED s.v. PE *atə- ‘put on’). aʔřia ‘its likeness’. Under PE *atə(δi)- ‘be the same’, the Inuit nominals listed in the CED permit reconstruction of PI *atʐi ‘likeness, copy, picture’: SPI ašši, Mal atži ‘copy, likeness’, NAI ažži, WCI ayyi, Aiv ayyi ‘likeness, picture, photograph’, GRI ašši(k), EG atsiq ‘picture, photograph’. CED also notes “Net a’ziɣiit ‘similar ones’ acc. Briggs — Dor. has atži ‘picture’”. iʔřiliqtuq ‘the weather is very cold’ (Dyck & Briggs 2005:322). PE *itδə ‘cold’ but as the Yupik and Sirenikski offer no clusters, we should only erect a PI *itʐɨ: SPI išši, Mal itžɨ, NAI ižži, Net [Dorais] itži, WCI, ECI iyyi, GRI išši. PE *tl > ʔl aʔla 'another of different kind'. PE *atla ‘other’. Mal, Uum atla, NAI, WCI alla, Net [Briggs] a’latqiit ‘dissimilar ones, ECI alla ‘non-Inuit First Nations person’, NG allaXXuaq ‘North American Inuk’; GRI ałła ‘other’, GRI [Fab] adla. PE *ty > ʔj -ʔja- postbase ‘vague future’. Examples: qauřimaʔjaŋŋitquurama ‘because I thought I wouldn’t know the words’, kiliʔjařruurama ‘I’m afraid I’ll cut myself”. PE *-tya- ‘intend or be about to’ (CED:474). This postbase is the only example of the cluster -ʔj- in the language, “indeed, j does not normally occur in clusters at all in this dialect” (Cook 2010:7).

Appendix B: Historical *p, *k, *q > Utkuhiksalik ʔ before *n, *ð/*ʐ, *l, *y

PE *pn > ʔn iʔnaq ‘cliff’. PE *əpnaʀ, Mal ɨpnaq, NAI ɨmnaq, Sig imnaq, ECI, GRI innaq. pre-Utku or PE *pð > ʔʐ kiʔřiřunga ‘I have cut my tendon’ (Cook 2010:6). A base kiʔři-, is not otherwise known. It may reflect contamination of PE kəpə- ‘cut’, cf. WCI, ECI kipi-, by PE *kəpδaʀ- ‘cut off (several or a lot)’, SPI kipšaq-, ‘cut hair’, NAI kivžaq-, WCI kivyaq- ‘cut with scissors’, and one might assume a pre-Utku *kipʐi- > *kitʐi-. tiʔřauliqtuq ‘is moving away, driven by breeze’. PE *təpδaʀ ‘flotsam or jetsam’, Mal tipžaq ’s.th. washed ashore, driftwood’, NAI tivžaq, WCI tivyat ‘driftwood’, ECI tiyyat ‘floating wreck’, GRI tiššaq ‘wreckage washed up on shore’.

22 The CED lists Net [Briggs] hi’žaq ‘den, burrow’ (s.v. CED PI *citə). At first blush this form looks relevant to this discussion. However, this is cognate to NAI sižžaq, NBI siɣyaq, siyyaq ‘fox’s lair’ for which a PI *ciɣʐaq with voiced ɣ before ʐ may be reconstructed. Possibly the Net form is *[çiɡᵃʐaq] with short epenthetic vowel between the voiced segments.

26 PROTO-INUIT PHONOLOGY

PE *pl > ʔl tiʔli 'food or other dirt around mouth'. PE *təpli ‘food around the mouth’, AAY, CAY təpłi- ‘have food around the mouth’, CSY təfłi ‘food around the mouth’, Mal tɨpliiyaun ‘napkin’, NAI tɨvli ‘food around the mouth’, WCI [Met.] tivlit, ECI tilliq, Lab [Peck] tivli, GRI tiłłi. haʔlu 'blindfold'. Possibly pre-Utkuhiksalik *caplu, related to PE *capə- ‘block’ cf. CAY capa ‘barrier, curtain’, but I find no cognates. PE *py > ʔj iʔjanŋuřuŋa ‘I’m suffocating’, iʔjanŋunaq ‘how stuffy it is in here’ (Dyck & Briggs 2005:321). PE *əpyalŋu- ‘feel suffocated’ (CED s.v. əpə- ‘suffocate’), SPI ipyaŋu-, Qaw ipsanŋu- ‘be short of breath’, NAI ivyanŋu- ‘be short of breath’, ECI iyyaŋu-, GRI issaŋu-, GRI old ortho. ivsangu-. The Inuit forms suggest a PI *ipjanŋu-. PE *kn > ʔn iʔniq 'fire, flame'. PE *ək(ə)nəʀ ‘fire’, SPI ikniq, Qaw itniq, Mal iknɨq, NAI iɣnɨk, Cop, Sig iŋnik, Net [Pryde] i’niq, ECI, GRI inniq. piʔna ‘up there’. PE *pik- ‘up above’, *pikna ‘that up-above (abs. sing.)’, SPI pikna, NAI piɣnʸa, WCI piŋna, ECI pinna, GRI piŋŋa (CED: 516). PE *qn > ʔn paʔnaiqtuq 'he prepares (clothes, possessions) for departure', PE *paqna- ‘prepare to leave’, SPI paqnak- ‘get ready’, Mal paqna- ‘get ready to go’, NAI paʀna-, Net [Ras.] paʀnak-, ECI paʀnak-. qiʔnaqtuq ‘it (the weather) is freezing (Cook 2010:6). The base qiqi- in qiqiřuq ‘it is frozen’ is from PE *qiqə-‘freeze’, NAI qiqɨ- ‘be frozen’, Aiv, NBI qiqi- ‘be frozen’. qiʔnaq- may be pre-Utku *qiqnaq- from PE *qiqə- + *-naʀ- ‘be such as to or -able’. CED also notes that the GRI +naʀ- also means “ ‘one -s’ (depersonalized subject)” (CED: 457) which may be related to the impersonal weather construction in Utku. (aʔniqtuq 'is hurt' might belong here if PE *atŋiʀ- ‘be in pain’ is revised to *aqniʀ-; see Appendix A under *tŋ > ʔŋ)

Appendix C: Historical *t > Natsilik ʔ before *m, *ŋ, *l, *y

PE *tm > ʔm naʔmaktuq ‘he carries (something) on his back’. PE *natmaɣ- ‘carry on back’, Qaw, Mal, Nu natmak- , NAI nanmak-, WCI naŋmak-, Net [Pryde] na’mak-, ECI namma(k)-, GRI nammaɣ-, old ortho. naŋmaɣ-. Also Uum natmak-. qaʔmaqtuqtuq ‘s/he calls the dogs’. See the discussion above on the Utkuhiksalik word of identical shape. PE tŋ > ʔŋ qaʔnguqtaqtuq ‘it rocks, moves, shakes’. PE *qatŋuʀ- ‘pry (heavy object) loose?’, SPI qatŋuq- ‘go away, remove, pry, lift up, take off (airplane)’, Nu qatnuq- ‘move, budge a heavy object’, WCI qanŋuq- ‘unstick, detach’ PE *tl > ʔl tuuʔlik ‘common loon’. PE *tu(C)utləɣ ‘loon or northern diver’, Nu, Mal tuutlɨk ‘yellow-billed loon’, NAI tuullɨk, Sig tuullik, Net [Ras.] to.tlik, Net [Pryde] tuu’lik, ECI tuulliq ‘common loon’, GRI tuułłik. PI *ty > ʔy puʔju ‘fingertip’. The *-my- in PE *pumyuɣ- ‘pinch thumb and forefinger together’ is supported by

27 DOUG HITCH

Sirenikski pumyuXqə(s)- and CSY pumsuɣ- ‘pinch’. All other Yupik have voiceless clusters: AAY puksuɣ-, AP, CAY pupsuɣ-, NSY pupẙuɣ- [ANLC]. The Inuit forms (except for Nu pupsuk- ‘pinch’ from Berg which is marked since Nu also has puyyuɣiaq ‘crab’) support a PI *putjuk- from *pupyuk- as suggested by the CED: Mal putyuk-, NAI puyyuk- ‘pinch’, WCI, ECI puyyuk ‘thumb and forefinger’. Also Uum /putjuk-/ (/pujjuk-/?) cf. putdjukkaa ‘pinched him’.

Appendix D: Petitot’s spellings for /c/

Below is a more or less random selection of words containing /c/ from Petitot’s list, presented in his spelling. These are given phonetic and phonemic interpretations. There follows a PE or PI proto-form and then a form from NAI. The last column contains modern Siglitun spellings from Lowe (1984).

Petitot Phonetic Phonemic PE/PI NAI Lowe 1984 ts tsaviϱatsiaϱk [cavik-] /cavik-/ PE *caviɣ savik ‘knife’ savik ‘couteau’ ‘knife’ tç tçik-tçik ‘marmotte’ [cikʂik] /cikʐik/ PE cikδiɣ sikšik ‘arctic siksik ‘ground ‘squirrel’ ground squirrel’ squirrel’ tçitamat ‘4’ [citamat] /citamat/ PE *citamat sisamat sitamat ‘four’ taptçiϱk ‘ceinture’ [tapciq] /tapciq/ PE tavci ‘belt’ tafsi, Nu tapsi tapsi piptçi ‘poisson sec’ [pipci] /pipci/ PE *pimci pipsi pipsi nakatçuk ‘vessie’ [nakacuk] /nakacuq/ PE *nakacuɣ nakasuk nakasuk ‘bladder’ ‘bladder’ kϱatçigéaϱk ‘phoque [qaciɣiaq] /qaciɣiaq/ PE *qaðiɣyaʀ qasiɣiaq (Uum qahigiaq marbré’ ‘spotted seal’ ‘black seal’) (but PI *qaciɣiaq?) kitçit ‘nombre (n. [kicit] /kicit/ PE *kəcit- kisit- kisitait ‘counted pl.)’ ‘count’ them’ pitçoyaϱtoaϱk [picu-] /picu-/ PE *piyuɣ- Nu pisuk-, Mal pisuk- ‘walk’ ‘marcher lentement, ‘walk’ pišuk- péniblement’ natçeϱk ‘phoque [nacciq] /nacciq/ PE *nayyiʀ natciq ‘seal’ natchiq ‘seal; barbu’ ‘ringed seal’ ringed seal’ (PI *nacciq?) ipiktçawn ‘pierre à [ipikcaun] /ipikcaun/ PE *ipəɣcaʀ- ipiksaq- ipiksaun aiguiser’ ‘sharpen’ ‘sharpener’ tçaϱvaϱk ‘courant’ [caʁvaq] /caʁvaq/ PE *caʀvaʀ saʀvaq sarvaq ‘current’ ‘current’ tçibiaϱk ‘hanche’ [cibⁱjaq] /civjaq/ PE *cipyaʀ Mal sipyaaq (ECI sivviaq) ‘hip’ tçaneϱktoaϱk [caniq-] /caniq-/ PE *caniʀ- Qaw saniq- saniq- ‘balayer’ ‘sweep’ tçauneϱk ‘os’ [cauniq] /cauniq/ PI *caunəq saunɨq sauniq ‘bone or pit’ tçiùn ‘oreille’ [ciun] /ciun/ PE *ciɣun ‘ear’ siun siun ‘ear’ tçiklaϱk ‘pioche’ [cikɬaq] /cikɬaq/ PE *ciklaʀ sikłʸaq sikłaq ‘pickaxe’

28 PROTO-INUIT PHONOLOGY

‘pickaxe’ tçuvaϱk ‘oeuf de [cuvak] /cuvak/ PE *ciłuvaɣ suvak suvak ‘fish egg’ poisson’ ‘fish egg’ tçigdǰaϱk ‘rivage’ [ciɡʐaq] or /ciɡʐaq/ PE *ciŋðaʀ siɣžaq [Jen.] (CED Sig [ciɡdʒaq] or /ciɣjaq/ ‘beach or siɣyaq) shore’ tçigôϱk ‘bec’ [ciɢɣuk] /ciɣɣuk/ PE *cuɣðuɣ or siɣɣuk sigguk ‘beak; *ciɣɣuɣ ‘beak muzzle’ or muzzle’ tçukkaϱk ‘poteau’ [cukaq] /cukaq/ PE *cukaʀ Qaw. sukaq sukaq ‘support’ ‘post or other ‘post supporting support’ fish drying rack’ tçokϱak ‘fanons de [cuqqaq] /cuqqaq/ PE *cuqqaʀ SPI suqqaq suqqaq ‘baleen’ baleine’ ‘baleen’ tch nutchuϱaga ‘tirer’ [nuccuk-] /nuccuk-/ PE *nuccuɣ- nutcuk- ‘pluck’ cf. nusugaa ‘pull or tug’ ‘snatched it’ koϱtchoϱk ‘résine’ [kuccuq] /kuccuq/ PE *kuccuʀ kutcuq kutchuq ‘resin or gum’ ‘chewing gum’ kotchoϱaϱk [kucuɣaq] /kucuɣaq/ PE *kucukaʀ SPI kuzuɣaq kusugaq ‘ice ‘stalagmite’ ‘icicle’ stalagmite’ itchuma ‘désir’ [icuma] /icuma/ PE *icuma isuma ‘mind’ isuma ‘thought’ tchiti ‘nid [citi] /citi/ PI *citə ‘den or Nu sinʸ siti ‘burrow’ d’hirondelles, de lair’ fourmis, etc.’ tchitkϱoϱk ‘genou’ [ciiʈquq] /ciiʐquq/ PE *ciɣəðquʀ siicquq siitquq ‘knee’ ‘knee’ tchiniktoaʀk ‘dormir’ [cinik-] /cinik-/ PI *cinək- sinʸik- sinik- ‘sleep’ tchoϱlon ‘narine, [cuɢlu] /cuʁlu/ PE *cuʀlu SPI suʀlut ‘nasal surluk ‘nasal naseau’ ‘nostril’ cavity’ cavity’ tci ig’utcièϱk ‘bourdon [iɣutcaq] /iɣutcaq/ PI *əɣutcaq or iɣutcaq ‘bumble igutchaq ‘bee’ des mousses’ *əɣutcak ‘bee’ bee’

Appendix E: Stefansson’s /c/ spellings

Table 17 has the spellings Stefansson called Mackenzie Eskimo (1909: 224–232), followed by assumed phonetic and phonemic renderings. The fourth column has the PE or PI form from the CED. The Siglitun column has items from Lowe (2001) unless otherwise noted (Uummaqmitun items are from Lowe 1984). The last column has items from Petitot (1876) where available.

“Mackenzie phonetic phonemic PE or PI Siglitun Petitot Eskimo” PE *c written c, tc ä-tcū' ‘I don’t [accu] /accu/ — (Uum atchuu!) — know’ ikpûkcak [ikpakcak] or /ikpakcak/ or PE *ikpaq ikpaksaq ‘day ikpèktçiaϱk

29 DOUG HITCH

‘yesterday’ [ikpakʂak] /ikpakʐak/ ‘yeserday or before ‘hier’ (Uum recently’ with yesterday’ with ikpakr̂ aq -kðaʀ ‘(s.th. -ksaq ‘material ‘yesterday’) for the) for; potential’ future)’ kap'-cĭt ‘how [qapcit] /qapcit/ PE *qavcit qapsit kraptçinik? many’ ‘how many’ (Uum qaffit) [kr for kϱ?] ‘combien?’ kĭ-tcĭ-mĭ [kiçimi] /kicimi/ PE *kəði- kisimi ‘only kitçimi ‘le ‘only, alone, ‘alone or only’ him’ seul’ nothing else’ ōk′-tcûk ‘fat, [uqcuq] /uqcuq/ PE uqðuʀ uqsuq oϱktchoϱk oil, blubber, ‘blubber or (Uum uqr̂ uq) ‘lard’ etc.’ seal oil’ ō-kûm-mai- [uqumaiccuaq] /uqumaiccuaq/ PE uqimaŋit- uqumaittuaq ‘is okϱumaītuaϱk tcū-ak ‘it is (strong i ‘be heavy’ heavy’ (Uum ‘lourd, de (adj. heavy’ palatalization?) uqumaitchuq ‘is v.)’ heavy’) pĭ-cuk-tū-ak [picuktuaq] /pituktuaq/ ? pisuktuaq pitoktoaϱk ‘he wants, (/t > c/ from ‘wants’ ‘vouloir’ (no desires’ strong i ?) strong i ?) pĭ-cūñ-nīt- [picuŋŋiccuŋa] /pituŋŋittuŋa/ PE -nʀit- ‘not’ -ŋŋit -ŋŋit- ‘not’ tcuña ‘I do not desire’ pī-tcûk [piiccuq] /piittuq/ PE piŋit- ‘not — NAI piic-, ‘be ‘nothing’ (strong i have absent, palatalizaion?) (anything) or missing’, WCI be absent’ piit- ‘not have any (left)’ cag-lū(k)-tu′- [caɡlu-] /caɣlu-/ PI *caɣlu- (NAI saɣlu- ) ECI sallu- yu-ak ‘he tells ‘tell a lie’ lies’ ca′-vĭ-ĭt ‘knife [caviit] /caviit/ PE *caviɣ saviit ‘knives’ tçavik scrapings’ ‘knife’ ‘couteau esquimau’ cĭ-nĭg′-a-vĭk [ciniɡᵃvik] /ciniɣvik/ PI *cinək- sinigvik tchiniktoaϱk ‘sleeping ‘sleep’ ‘bedroom’ ‘dormir’ place’ cĭ-nĭk′-tu-ak [ciniktuaq] /ciniktuaq/ " siniktuaq " ‘he sleeps’ ‘sleeps’ cū′-na ‘what, [cuna] /cuna/ PE *cu(na) suna ‘what?’ tchuna? which’ ‘what’ ‘que?’ ūñ-a-cĭk-tū-ak [uŋaciktuaq] /uŋaciktuaq/ PI *uŋacik- ungasiktuaq ‘is uñatçiktoϱ ‘it is far off’ ‘be far’ (CED far’ ‘très loin’ has *uŋasik-) a-tau′-tcīk ‘1’ [ataucik] /ataucik/ PE *ataʀuciʀ atausik ataotçiϱk ‘un’ ‘one’ pīñ-a-tcūt ‘3’ [piŋacut] /piŋacut/ PE *piŋayut pingasut piñatçut ‘three’ ‘trois’ PE *c written

30 PROTO-INUIT PHONOLOGY s sĭr-kĭn-nĭrk [çiqiniq] /ciqiniq/ PE *ciqinəʀ siqiniq tçikϱeyneϱk ‘sun’ ‘sun or ray of ‘soleil’ sunlight’ e-lĭp′-sī ‘you [ilipçi] /ilipci/ PE *əlpət ‘you ilipsi illipçi ‘vous’ (pl.)’ (sg.)’ sū-blū-a-raun′ [çubluaʁaun] /cuvluaʁaun/ ? cf. Tanik ‘white tchubluaϱaotit ‘white man’ person’ ‘les blancs’ kissimi ‘only, [kiçimi] /kicimi/ PE *kəði- kisimi ‘only kitçimi ‘le alone, nothing ‘alone or only’ him’ seul’ else’ sa′-vik ‘knife’ [çavik] /cavik/ PE caviɣ savik tçavik ‘knife’ ‘couteau esquimau’ sī′-la ‘the [çila] /cila/ PE *cila or sila tçilla ‘temps’ outdoors’ ciła ‘(spirit of) weather or outside world’ tū-sar′-yu-ak [tuça?juaq] /tuca?juaq/ PE *tucaʀ- tusaayuaq tutchâyok ‘he hears’ ‘hear or ‘hears’ (Uum ‘comprendre’ understand’ tuhaar̂ uq) sīs-sa-mat or [çitamat] /citamat/ PE *citamat sitamat tçitamat sī′-ta-mat ‘4’ ‘four’ (NAI sisamat)

Appendix F: Audio files showing [ɡ] in some clusters

PE *kð > Natsilik /kʐ/ [kʂ] inukřuk ‘inukshuk’. PI *inukšuk ‘cairn’, NAI inʸukšuk, inʸuksuk ‘pile of rocks used in corralling caribou’, WCI ‘cairn of rocks’, ECI inutsuk, GRI inussuk. It might be more accurate to give the PI form as *inukʐuk. qajukřaq ‘something that can be used to make soup’. PE *qayuʀ ‘meat broth’, NAI, WCI, ECI, GRI variously ‘tea, boiled blood, broth, meat soup, soup’ plus the postbase PE *kðaʀ ‘(s.th. for the) future’. itpakřaq (ikp-?) ‘yesterday’. PI *ikpaq ‘yesterday or recently’ + *kðaʀ ‘(s.th. for the) future’, Qaw, NAI ikpakšaq ‘yesterday’, WCI ikpaksaq, ikpaxxaq ‘yesterday’, NBI ikpaksaq ‘yesterday’, ECI ippasaq, GRI ippassaq. PI would be *ikpakʐaq. Petitot ikpèktçiaϱk ‘hier’. ikhivautaq akřalialik ‘wheelchair’. PE *akðaɣ- ‘roll or turn over’, SPI akšak- ‘roll’, NAI akšaliɣaq ‘wheel, tire’, WCI aksaluɣaq ‘wheel’, Cop axxaluaq ‘wheel’ [Pryde], ECI atsaluaq ‘wheel’, GRI aššakaašuq ‘wheel’. Natsilik Devoiced /ʐ/ after q miqřuqtaq ‘sewing’. PE *miŋqə- ‘sew’ but PI looks like *miqʐuq23, cf., Cop miXXuq-, Net meqSor- [Ras., B.-S.], ECI miʀsu(q)-, GRI miʀšuʀ-. qaqřauqpiarřuq ‘red-throated loon’. PE *qaqða(C)uʀ ‘loon’, SPI, NAI qaqšauq ‘red-throated loon’, WCI qaqsauq, qaXXauq, ECI qaʀsauq (for qaqsauq?), GRI qaʀšaaq, NG qaXXauq.

23 Next to the GRI entry CED has a note relating the GRI, Cop and ECI forms with the postbase qðuʀ- ‘repeated action’.

31 DOUG HITCH

Appendix G: Audio files showing [ɡ] in some clusters

Inuinnaqtun [ɡl] in /gl/ igliq ‘bed’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/inuinnaqtun/92.mp3 iglu ‘house’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/inuinnaqtun/531.mp3 Kivallirmiut [ɡl] in /gl/ igliq ‘bed’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/kivallirmiut/92.mp3 niglahuktuq ‘cold weather’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/kivallirmiut/239.mp3 Kivallirmiut [ɡv] in /gv/ imialugvik ‘bar’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/kivallirmiut/82.mp3 quviahugvik ‘Christmas’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/kivallirmiut/206.mp3 Uqqurmiut (Pangnirtung, SE Baffin) [ɡɡ] in /gg/ agga ‘no’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/uqqurmiut/738.mp3 aggaut ‘forearm; http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/uqqurmiut/396.mp3

Appendix H: Audio files showing [ɢ] and [ɴ] in some clusters

Uqqurmiut [ɢdʒ] in /ʁj/ Talurjuaq ‘Taloyoak’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/uqqurmiut/1039.mp3 umiarjuaq ‘ship’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/uqqurmiut/1039.mp3 Inuinnaqtun [ɢv] in /ʁv/ aaniarvik ‘hospital’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/inuinnaqtun/525.mp3 tukpirvik ‘tent site’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/inuinnaqtun/1053.mp3 Inuinnaqtun [ɢm] in /ʁm/ tayarmiaq ‘bracelet’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/inuinnaqtun/1247.mp3 imarmi ‘in water’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/inuinnaqtun/1149.mp3.

The uvular nasal allophone [ɴ] may perhaps most clearly be heard in the cluster /ʁm/ in those dialects where it is pronounced [ɴm].

Uqqurmiut [ɴm] tujurmijuq ‘stays somewhere’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/uqqurmiut/1003.mp3 nagguarmiq ‘ring’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/inuinnaqtun/1149.mp3 Kivallirmiut [ɴm] irmirvik ‘sink’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/kivallirmiut/944.mp3 turjurmivik ‘hotel’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/kivallirmiut/529.mp3 It is also heard in /ʁn/ [ɴn] in Uqqurmiut (Pangnirtung, SE Baffin Island) arnaq ‘woman’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/uqqurmiut/1235.mp3 irniq ‘son’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/kivallirmiut/985.mp3

32 PROTO-INUIT PHONOLOGY

The cluster /ʁŋ/ can sometimes have challenging phonetics. In the examples below the /ʁ/ is clearly nasalized.

Aggurmiut (Iglulik and Sanirajaq/Hall Beach, N Baffin) /ʁŋ/ irngutaq ‘grandchild’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/aggurmiut/463.mp3 upirngaaq ‘spring’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/aggurmiut/997.mp3 Uqqurmiut /ʁŋ/ irngutaq ‘grandchild’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/uqqurmiut/463.mp3 upirngaaq ‘spring’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/uqqurmiut/997.mp3 Kivallirmiut /ʁŋ/ irngutaq ‘grandchild’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/kivallirmiut/463.mp3 upirnngaaq ‘spring’ http://www.tusaalanga.ca/sites/tusaalanga.ca/files/dialect/kivallirmiut/997.mp3

Appendix I: Symbology

In the Inuit language field there is significant variation in the symbols used in phonological descriptions, and in practical orthographies. The method here is to reproduce the source symbology as accurately as possible, which can make arguments hard to follow. This chart intends to compare all discrepancies among these systems. Several minor systems are not included. Bobalijk 1996 uses ɹ for what is here regarded as the voiced retroflex sibilant which means that in these pages at least seven distinct symbols are used for this segment. The Dorais, Lowe and D&B (Dyck and Briggs) columns contain symbols from all works referred to.

IPA CED Dorais Iñupiaq ICI Lowe D&B voiced uvular continuant ʁ ʀ r, R ġ r r r, voiceless apical lateral ɬ ł ł, & ł ł, & ł ł voiced retroflex sibilant ʐ ð, ž ð, r̂, j̵ r — r̂ ř voiceless retroflex sibilant ʂ š sr voiced palatal approximant j y j y j y j voiceless palatal stop c č, c tch — tch voiceless postalveolar affricate tʃ č (tch) (tch) voiced palatal stop ɟ dj, tdj voiced postalveolar affricate dʒ (dj, tdj) palatal nasal ɲ ñ, nʸ ñ — velar nasal ŋ ŋ, ng ŋ, ng ng ng ng ŋ, ng glottal stop ʔ ’, ʔ ’, ʔ ’ — — ’, ʔ

In addition, for technical reasons, in these pages ŋ̊ stands for CED ŋ+ ̥ , and ẙ for CED y + ̥ .

33 DOUG HITCH

Appendix J: Abbreviations and Dialect Names

AAY Alaskan Yupik Aggurmiut N Baffin, Iglulik, Sanirajaq (Hall Beach) Aiv Aivilik, NE Hudson Bay B.-S. CED source for GRI and WCI: K. Birket-Smith 1928 Briggs CED source: Jean Briggs’ manuscript on Natsilik and Utkuhiksalik Car Caribou = Kivalliq, Baker Lake (Qamanittuaq), Arviat (Eskimo Point) CAY Central Alaskan Yupik CED Comparative Eskimo Dictionary; Fortescue et al. 2010 Cop Copper = Inuinnaqtun, western Canada, Cambridge Bay (Iqaluktuuttiaq), Qurluqtuq (Coppermine), Ulukhaqtuaq (Holman Island) CSY Central Siberian Yupik DI Little Diomede, island in the Bering Sea Dor CED source: Louis-Jacques Dorais, various works ECI Eastern Canadian Inuit, CED term comprising NBI, SBI, Aiv, Itiv, Tar, and Lab. EG East Greenlandic Fab CED source for old GRI: O. Fabricius 1804 GRI GRI old ortho. Spellings found in materials from before the Greenlandic orthographic reform of 1973 HBC Hooper Bay-Chevak, CAY dialect ICI Syllabics and Roman Inuktitut orthographies developed under the auspices of the Inuit Cultural Institute in 1976 Inuinnaqtun see Copper Itiv Itivimiut, Arctic Quebec, eastern Hudson Bay Kivalliq see Car Kivallirmiut see Car Kang Kangiryuarmiutun, from Holman on Victoria Island, subdialect of Inuinnaqtun (Copper) Lab Labrador Lowe CED source: Ronald Lowe, various works Mal Malimiut, lower NAI Met CED source: M. Métayer 1953 typescript on WCI NAI North Alaskan Inuit Natsilik see Net Natsilingmiutut see Net NBI, NB North Baffin Inuktitut Net Natsilingmiutut, Natsilik (Netchilik), Pelly Bay, Taloyoak (Talurjuaq, Spence Bay), Gjoa Haven NG North Greenlandic/Polar Eskimo, Thule NSY Naukan(ski) Yupik, Chukchi Peninsula, Siberia Nu , , NE Alaska, subdialect of NAI Nun Nunivak CAY, Mekoryuk (Mikuryaq) PE Proto-Eskimo (Proto-Inuit-Yupik-Sirenikski), reconstructed parent of the Inuit, Yupik and Sirenikski languages PH Point Hope NAI PI Proto-Inuit Pryde CED Source: lexicographic materials of Duncan Pryde [not available for the first edition]. Qaw Qawiaraq (SPI) SB South Baffin Sig Siglitun as quoted in the CED. Forms from Lowe 2001 are listed as Siglitun.

34 PROTO-INUIT PHONOLOGY

SPI Seward Peninsula Inuit Tarr Tarramiut. Arctic Quebec NE coast. Tersis CED source on E Greenlandic, N. Tersis 2008 Thule see NG Uum Uummaqmiutun, , Inuvik, NAI dialect Up Upernavik (GRI) Utku Utkuhiksalingmiutut, Utkuhiksalik, south of Natsilik, between Chantrey Inlet and Franklin Lake Uqqurmiut Pangnirtung, SE Baffin WCI Western Canadian Inuit, CED term comprising Sig, Cop, Net, Car

35