The phonemes of Bella Coola are as shown in Table 1 below. Per centum distribution of phonemes across the entire morpheme inventory 16 is indicated in parentheses. Note, in particular, the high numbers for /s, ł, i, u, a/ vs the very low frequency of /h/, and the prevalence of fricatives and sonorants over stops (plosives). Aspiration of /p/ and /t/, and affrication of all back plain stops, in non-prevocalic environments are treated in Nater (1984, p. 4). Note that /p’, t’, q’, etc./ = /p̓, t̓, q̓, etc./; /kʷ’, qʷ’/ = /k̓ʷ, q̓ʷ/; #’R = #ʔR (where # = continuant or morpheme boundary, R = sonorant); /ł/ = /ɬ/; /χ/ = /x̌/; /χʷ/ = /x̌ʷ/.

Mellesmoen’s addition of /ə/ (p. 228) is baseless, because schwa is not phonemically distinctive: it appears (1) predictably in sonorant clusters and XN̩X, (2) marginally in PNV, and (3) never in SNV 16. Thus, we note that /ə/ is in the boxed PNV portion of not just quite reduced (p. 233, fn. 7), but unsurprisingly absent: one hears (…c’ʔ͡me̝) (not *(…c’ɪ̆me̝)), where glottal release coincides with onset (lip closure) of (m). Figure 1

In order to give the reader an idea of the settings in which voiceless clusters exist, I list two-member clusters in charts based on 16. Coding and patterns are: (a) cells that would have accommodated avoided clusters are greyed out; (b) unpredictable absence is marked as ^____; (c) exceptional (hypothetically avoided) sequences are boxed; (d) clusters found only in words containing one or more vocalic sonorants are preceded by an asterisk; Table 2

(e) {homorganic front stops} and {plain front stop + ejective front stop} clusters are avoided:

(f) {plain back stop, /x/, /χʷ/ + ejective front stop} clusters, except {/q/ + ejective front stop} and boxed clusters, are avoided: Table 3

(g) {/p’/ + back obstruent} clusters except /p’x/ and /p’χʷ/ are avoided: Table 4

(h) back+back obstruent clusters except xkʷ’, χkʷ, kʷ’qʷ are homorganic, while {/xʷ/, /χʷ/ + back obstruent} and {/x/ + back obstruent except /kʷ’/} clusters are avoided: Table 5

Data presented in 3.1-3 appear in a practical orthography in Nater (1990), where ts = /c/, ts’ = /c’/, lh = /ł/, tl’ = /ƛ’/, c = /x/, cw = /xʷ/, x = /χ/, xw = /χʷ/, 7 = /’/. Throughout 3.1-3, tr. = transitive verb.

The data displayed further below contradict Mellesmoen’s claim that there is only one all-stop root, viz √tp ‘spotted’ (which she, relying on 2021 FirstVoices data rather than 17, cites as a free form). Mellesmoen briefly ponders affricates as potential nuclei (p. 230), but then imposes a restriction: “For a narrower scope …”. Regardless, she does not abide by this restriction, having already decided that tp is the one and only all-stop word (Table 1 and comments thereafter). It is Mellesmoen’s foregone conclusion that affricates (which she does not consider stops) are indeed potential nuclei, as implied by “Have a Fricative or an Affricate” in her Table 1. I quote:

From a total of 1506 words on FirstVoices, there were only a total of 51 words that fit the definition of an obstruent-only word. The break down is shown in Table 1.

(…) This leaves a single obstruent-only word that is comprised only of stops (tp ‘spotted’). (…) While virtually every word in Bella Coola has at least one (+continuant) segment, and fricatives [sic] may be naturally paired with fricatives, rather than stops, this involves larger questions pertaining to the nature of affricates that are beyond the scope of the present paper. For a narrower scope and due to a small number of obstruent-only words without fricatives, this paper focuses on fricatives as potential nuclei and leaves the investigation of affricates for future study. […] The distributional patterns described in Section 2.2 show that Bella Coola allows fricatives, but not stops, in a nuclear position.

15

^{1P = any stop but /c/, N = front sonorant, S = /c/ or fricative, V = vocalic R, X = obstruent or word boundary}.

Below, entries that contain none of the affricates /c/, /c’/, /ƛ’/ are rendered in boldface.

^{2Found in kƛ’i ‘short’, ’ Ayxc’mqa ‘a woman’s name’, skʷc ’ ‘wrinkled’, kʷƛ ’ (tr.) ‘put in place’.}

It is the large inventory of voiceless Bella Coola words and roots containing at least one stop and one fricative that seems to have inspired GENH in the first place. However, making a GENH-based decision on syllabicity is problematic where two consecutive fricatives appear within a word, as marked in boldface below.

The boldfaced entries beg the question of how Mellesmoen would decide which fricative(s) is/are nuclear, e.g.:

● Should psx (tr.) ‘train, educate’ be parsed as /ps̩x/ or /psx̩/?

Below I list words and roots that consist entirely of fricatives. Note recurrence of front+back and back+front sequences.

Again, Mellesmoen does not show us a procedure for determining nuclearity, and one might ask, for instance:

^{3The two exceptions are p’x ‘ crabapple ’ and p’χʷłt ‘bunchberry’. p’x has no known etymology, while p’χʷłt (dissimilated */ p’χʷ-łp /?) may be a blend of older * p’uqʷ (ay) ‘a shrub’ (Kuipers, 2002, p. 81) and Kwakwala √ p̓qʷ ‘to be flat on s.th.’ (Lincoln & Rath, 1980, p. 69) and/or Heiltsuk √ p̓h̓x̌ʷ ‘to change colours (in the fall)’ (Lincoln & Rath, 1980, p. 70).}

^{4Rows and colums that accomodate non-existent clusters only are excluded in Table 5. xkʷ ’, χkʷ , kʷ’qʷ are unique clusters found in resp. s’ixkʷ’xʷ ‘ bushtail rat’, ’ awanaaχkʷ ‘raft’ and Kʷ’qʷla ‘Clayton Falls’. Of these, s’ixkʷ’xʷ is likely based on Proto-Salish * k’ʷaxʷ ‘rat, mouse’ (Kuipers, 2002, p. 51), while the other two surely have North Wakashan origins: Haisla hawanàqi̓akʷ ‘loaded on top (said of the canoes covered with boards)’ (Lincoln & Rath, 1986, p. 121), resp. Heiltsuk √ k̓ʷh̓qʷ ‘?’ (in k̓ʷík̓ʷáqʷx̌ƛawa ‘lichen’ (Lincoln & Rath, 1980, p. 292).}

● Should sxłł ‘afterbirth’ be parsed as /s̩xłł/, /sx̩łł/, /sxł̩ł̩/ /, or /s̩xł̩ł̩/?

Long obstruent clusters and voiceless words also exist in two North Wakashan subdialects that are geographically close to Bella Coola (Figure 2): Heiltsuk (Bella Bella, Wáglísla) and Oowekyala (Rivers Inlet). Some Heiltsuk voiceless clusters and words 10 are:

(128) txtxní ‘owl’ (129) ttxstú ‘bulging eyes’

(130) txʷsx̌ƛáqà ‘to jump over s.th.’ (131) tpkʷ ‘flashlight’

(132)  qqs ‘eye’ (133) ƛxx̌s ‘cross-piece of a canoe’

While Heiltsuk is tonal, Oowekyala, like Bella Coola, lacks distinctive stress or tone 12. However, Oowekyala differs considerably from Bella Coola on the phonemic-phonotactic level. In re long-standing contact, 4 states that South Bentinck and Rivers Inlet were connected via a frequently used overland route, as reiterated by 8: Figure 2

(…) the Oowekeeno at the top of Owikeno Lake had regular contact with the Nuxalk of South Bentick Arm, and intermarriage between the two tribes was not infrequent (McIlwaith 1948). (…) The extremely rare consonant clustering property for which Nuxalk is notorious 161 is also found in Oowekyala (…)

8

Some voiceless words are copies of Oowekyala roots that are themselves voiceless: Table 6

Other voiceless words derive from **C₁VC₂ (> *C₁V̥C₂ > C₁C₂). In Table 7, Salish data are from 11 Kwakwala data from Lincoln and Rath (1980), Lillooet data from 21, Sechelt data from 3, Upper Chehalis data from 9, areal data (Salish + Tsimshianic (+ North Wakashan)) from 18. Kw = Kwakwala, Li = Lillooet, NW = North Wakashan, PS = proto-Salish, PCS = proto-Central Salish, PIS = proto-Interior Salish, Se = Sechelt, Ts = Tsimshianic, UCh = Upper Chehalis.

^{5North Wakashan has aspirated stops in addition to plain (often phonetically voiced) and ejective ones, as well as /λ/ and /ƛ/. See Howe (2000, pp. 9-15) for “degenerate” syllables and schwa epenthesis in Oowekyala}.