PROTO-CENTRAL SALISH PHONOLOGY AND SOUND CORRESPONDENCES Brent Galloway Sections presented as a late paper 17th International Conference on Saliah and Neighboring Languages August 9-11, 1982 Portland State University Portland, Oregon PROTO-CZNTRAL SALISH PHONOLOGY AND SOUND CORRESPONDENCES Brent D. Galloway 0. Introduction. This article began as a preliminary look at Nooksack and Proto-Central Salish sound correspondences, It was preliminary because I began only looking for cognates with the 189 Nooksack words given in Amos8 (1961). The purpose of the article was to find especially those correspondences which separate Nooksack from its neighbors, Halkomelea, Straits 3al- ish, aad Lushootseed (Puget Salish). This should provide a starting point for a aerie8 of in-depth studies I plu to do of the unpublished materials and tape8 of Nookaack gathered by Pamela Amosa, Laurenca Thornpaon and Barbara Efrat and their pre- Mr, Sindick Jimmy, the last fluent speakor of Nookaack, died July 12, 1977, He, like all the other speakers of whom we have tapes, and like many speakers of Nooksack for the past 1% years, was fluent also in the Ohilliwack dialect of Halkomelem and the Skagit dialect of Lushootseed. A8 Thompson (196) points out: The custom of importing wives from other communi- ties contributed to the early obsolescence of the dis- tinctive language [~ooksack]originally spoken there [the Nooksack apeeoh area]; by the time of white con- taot this process seems to have been at an advanced stage, so that now only a few persons recall the time when a number of old men in the villages spoke among themselves 'the old language,' while for general pur- poses other languages were in use--primarily Chilliwsck Halkomelea and Skagit, It is now difficult to deter- mine very surely what the original Nooksack language was like, beoause the materials that have been collec- ted all bear marks of borrowing from the neighboring languages, especially from Halkornelem. In order to sort out what linguistic data is Nooksack and what is not, several levels of apparent language variation will have to be separated: language-switching ( infomant error) , legitimate diffusion or borrowing (at various time depths), incomplete sound changes, features of language death (for exam- ple loss and inconsistencies), and normal variation within and between speakers, Figuring out the cognates and sound corres- pondences between Nooksack and its neighbors will allow U8 to isolate cases of language-switching. The other cases of lang- uage variation should yield to the application of internal re- construction and comparative reconstruction. It should also be kept in mind that as the descriptive analysis of the Nooksack materials becomes complete the sound correspondences here should be re-examined and revised where necessary, Also at that point many more cognates will be available, and they can be added to those provided by Kuipers (1970, 19811, Thompson, Thompson and Efrat (1974) , Haeberlin (1974) , and other published cognate sets. It was difficult to keep from adding cognate sets for which there was no Nooksack equivalent in Amos8 (1961). At last I gave in to the temptation when I saw that the correspondences could be made fairly complete with additions from languages poorly represented or diagnostic of different correspondences, The additional 100 cognate sets (190-289) are incomplete in them- selves; I have not searched all the sources as I did for 1-189 to find cognates in all the Central Salish languages, Thus while reconstructions could be made from 190-289 for the sound correspondences for which they are dia-ostic and for those which are always one-to-one, reconatructions of the other proto- phonemes in these cognates (especially vowels) are best post- poned till the cognate sets are as complete as for 1-189, Simi- larly no attempt is made to discuss reflexes from sets 190-289 in section 3, except for the correspondences these sets were chosen to fill in or illustrate. At first it seemed that the irregularities and exceptions were just a tedious headache, as I tried to understand and account for them, But it soon became apparent that they not only prove the rules but also are the evidence of some of the most interesting historioal processes in Central Salish phonolo- gical history, They are also sometimes the crucial evidence for explanation of the regular correspondences, After a while I came to seek them out, ney also show the network of borrow- ings with Central Salish; intrafamilial influence is one of the hardest elements to grapple with in Northwest Coast historical linguistics, Some irregularities show connections between sound correspondences (parallel shifts), Some show intermediate stages of development preserved, And many show the individual- ity of each language and dialect. Since Nooksack terms are central and present in every cog- nate set which follows (till I%), they are listed first; then cognates are presented from the languages proceeding north from Halkomelem to Comox, then from Nooksack proceeding generally south and west (Lushootseed, Twana, Northern Straits, Southern Straits), Thus some contiguoue sound shifts may be seen and some priority can also be given to Nooksaok's immediate neigh- bors, Transcriptions from the various sources have been stan- dardized, The languages (from north to south) from which data is presented have the following abbreviations and phonemic sys- tems (sources listing the phonemes are given here; comments on resonant-glottal stop clusters are given because some analyses show glottalized resonants as phonemes; no statements are imp- lied about occurrence of glottal stop before or after other con- sonants) : Mainland Comox (Cx) : Davis (1970) gives /p, ps , m, (te), te' (normalized from here on as <8*>), 8, t, t9,n, s, k, k', &,(I?, W 8, 8'9 3, 8, (k), (k'), g [kPl,y,kw, ktW, xu, W, 9, 9'9 5, P 9 P'", f, , h i, c, e, o, a/; ['I is predictable (first syll- able) 8 /te/ occurs in one morpheme only, a first person singular pronoun affix; rare phonemes are parenthesized, Bouchard (Iwla) has the same consonants but treats resonant + 3 clusters as glot- talized resonants; his vowel inventory is /i, i*,a, a*, u, u*, e/ with [ c] under /i/ and [o 1 under /u/ , Pentlatch (Pt) : Kinkade' s (1980) and Davis's (1981) reconstitu- tions of Boas's Pentlatch field notes (n ,do) include the follow- ing (in parentheses are likely additional phonemes): /p, (p*), m, te, 0, t, t*, n, c9, s,ks, a, 1, E, 89, J,i,~,k~,k*~,xu, W, P, q*, ?, (qw), q*w, T~,3, h, i, 9, u, a, '1; Ite/ again probably occurs in one morpheme only, a first person singular pronoun affix . Sechelt (Se) r Beaumont (l976), Timmers (19771, and Bouchard (l971b) show nearly the same inventory: /p, p' , m, t, t ' n, c, c*, s, 8, E*, , k %, 1, k, k', x, 7, kW, ktW, xW, W, 9, p', X, qw, q*w, ?w, 7, h, i, e, u, a, '1; Bouchard omits /d and he and Timmers add /m9, n9, I*, y9, w9/ where Beaumont has clus- ters /lo, ?y/ and perhaps others; Timmers also adds /'/; inci- dentally, the symbols c and c9 stand for d and d9 respectively throughout this article except in Halkomelem, q,v, Squamish (Sq) ; Kuipers l967b has /p, p*, m, t, t*, n, c, cs, s, E, , I, k', i, 1, , '9Y kW, k'w, fW, w, q, q9, 3, qw, VW, xw, ?, h, i, e, u, a, '/; he also gives a rare /u[c*- I*]which doesn't appear in the sets here; Kuipers writes clusters of + resonant and resonant + 3 with a11 resonants. Halkomelem (Hk) , Chilliwack dialect group (Ch) : Galloway (1977) gives /p, p*, m, 8' [te,l, e, t, t9, c, c*, 8, kt, L, 1, (k), , 9, Y, kW, k*w, xW, w, 9, P*, X, qW, P*~,g, ?, h, i, €9 e , a, u, o, , ', '/ ; Ch throughout stands for all the upriver dialects (Tait , Pilalt, ~hilliwack/~ardis,and Chehalis) ; they agree lexically in all but s few cases and those are noted; /c/ and /c'/ have allophones [d, 81 and [d*, 8'1 respectively; /k/ and /k9/ occur only in borrowings and a few pet names; /'/ and /'/ are respectively high and mid pitch-stress or tone (low tone is unmarked); sporadic lusters of + resonant or resonant + occur only as a rare result of reduplication, unlike the case in downriver (Ma) and island (Cw) dialects; /k/ is usually [dl, Halkomelem, Musqueam dialect (~s)r Elmendori and Suttles (1960) list the same phonemes as I have for Ch except than /n/ is added, /kg/ is omitted, /*/ is marginal (probablr best parenthesized), and ['I is probably not phonemic; resonants can be preceded or followed by 3 and such clusters are frequent. Halkomelem. Cowichan dialect (Cw); Elmendorf and Suttlea (1960) list the same phonemes as I have for Chilliwack except that /n/ is added, /k*/-is omitted, /8/ replaces /g/,/*/ is marginall resonants can be preceded or followed by and frequently are4 Leslie (19798 9) lists /te/(appears in one morpheme only, a demonstrative article) and /m9, n9, Is, y*, w*/ but mentions that wglottalized resonants ... are posited not on phonetic, but on morphophonemic ground^.^ Nooksack (Nk)r Amoss (1961) gives /p, p9, m, t, t9,n, c, c9, s, 8, 119, 8, Y, kv, , 1, kWkvW, x", w, q, q*, 5, q", q*w, ?W, 3, h, i, a, a, o, '/; resonants can be preceded or followed by 3; a significant note is found in Amos8 (1961: 9): wThe free variation in tongue grooving occurred only in the idiolect of George Swanaset . George Swanaset frequently realized /a/ as [el, /c/ as [~s]and /E/ as [c] he consistently produced /8/ as [g1.