* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Why Tlingit is not an Athabaskan language: An introduction to Tlingit
Arabic grammar wikipedia , lookup
Latin syntax wikipedia , lookup
Malay grammar wikipedia , lookup
Kannada grammar wikipedia , lookup
Udmurt grammar wikipedia , lookup
Ukrainian grammar wikipedia , lookup
Esperanto grammar wikipedia , lookup
Portuguese grammar wikipedia , lookup
Navajo grammar wikipedia , lookup
Old Norse morphology wikipedia , lookup
Modern Greek grammar wikipedia , lookup
Ancient Greek grammar wikipedia , lookup
Swedish grammar wikipedia , lookup
Zulu grammar wikipedia , lookup
Spanish grammar wikipedia , lookup
Old English grammar wikipedia , lookup
Ojibwe grammar wikipedia , lookup
French grammar wikipedia , lookup
Sotho parts of speech wikipedia , lookup
Polish grammar wikipedia , lookup
Yiddish grammar wikipedia , lookup
Turkish grammar wikipedia , lookup
Old Irish grammar wikipedia , lookup
Romanian nouns wikipedia , lookup
Serbo-Croatian grammar wikipedia , lookup
Why Tlingit is not an Athabaskan language: An introduction to Tlingit for Athabaskanists James A. Crippen University of Hawaiʻi at Mānoa [email protected] April C . I In this paper I will present some comparisons and contrasts between Tlingit (Eng. /ˈklɪŋˌkɪt/, not */t(ə)lɪŋgɪt/) and the Athabaskan languages. is is not meant to be a comprehensive study, but merely a shallow survey of some of the more interesting features which can be compared between the languages. It is well established that Tlingit and the Athabaskan languages are part of the greater Na-Dene family which also includes Eyak (leer:). A current historical hypothesis calls for a larger family of Dene-Yeniseic which subsumes NaDene and the Yeniseic family of languages in central Siberia (vajda:), but I will not address this issue here. Due to its ancestral connection, Tlingit has a large number of features in common with the Athabaskan languages, but there are some significant differences between them which may come as surprising to the typical Athabaskan specialist. e relatively obscure documentation on Tlingit has contributed to Athabaskanists having limited knowledge of the language, despite its importance in the reconstruction of the family and as a source of comparative insights. For an Athabaskanist, looking at Tlingit is like viewing an Athabaskan language through a distorting funhouse mirror, where reflections are seen in strange, convoluted forms, but the overall structure is still recognizably familiar, as will be seen throughout this paper. . D Tlingit is sometimes considered to be a well documented language, and in some respects this is true. e first serious aempt at describing the language was due to Ivan veniaminov: a Russian Orthodox missionary who was later canonized as St. Innocent. krause: offered a small lexical inventory, but his ear was far worse than Veniaminov’s. kelly: wrote a lile known work for the US government that is fairly accurate in its transcription, but the grammatical description is naïve. swanton: is oen mentioned and is relatively well known because of inclusion in the Handbook of American Indian languages, but Swanton’s time with Tlingit speakers was limited, his transcriptions are marginal, and his grammatical description is not very useful. Swanton’s work was superseded by the much beer but lesser known work of boas: whose collaboration with native speaker Louis shotridge: is all too oen overlooked. Description of the language languished while the Na-Dene controversy began, and very lile was work was done up through the s (miller:; velten:; velten:). In the s Constance M. Naish and Gillian L. Story, two young women from the Summer Institute of Linguistics, visited Angoon and did extensive work on the language. ey produced a noun dictionary (naish:; naish:), two master’s theses (naish:; story:), a translation of the Gospel of John (anon:), and a verb dictionary (story:). e master’s theses and the verb dictionary are probably their most important contributions, although the theses are somewhat inaccessible due to their extensive use of tagmemics. e verb dictionary is a treasure trove of example sentences, and contains an extraordinarily extensive listing of verb themes, but is hampered by lacking some of the less obvious features of themes including the thematic conjugation prefixes and stem variation information. Jeff Leer has worked on Tlingit since the s, but has published only a small amount of his work. Particularly important documentary sources are his work with the last speakers of the Tongass dialect (williams:), his dissertation (leer:), texts from his favorite consultant Elizabeth Nyman (nyman:), and the extensive noun dictionary of Interior Tlingit that he assisted with (leer:). A major problem with his dissertation for those looking for description is that since the dissertation’s emphasis is on the semantics of the aspectual system rather than the language as a whole, many descriptive insights which he notes are are illustrated with only one form, or worse yet have no examples whatsoever. Most of his subsequent work has been in the historical realm, and as such the descriptive details that he provides are oen overshadowed by comparative and historical content. Nevertheless, leer: has still been the most prolific linguist associated with Tlingit, and commands a deep understanding of the language that is unequaled. One issue he raised about the language which should not be overlooked is the likelihood that modern Tlingit resulted from a merging of somewhat different but closely related languages (leer:), a claim which not only has profound diachonic implications but also suggests explanations for many peculiar synchronic phenomena. To date there is no published grammar of Tlingit, and this issue is a lacuna which I hope to fill in the not-too-distant future. Leer’s dissertation serves as a source of documentation about a variety of issues in the language, but as previously noted it is problematic in a number of ways. Story’s master’s thesis is also useful as a descriptive adjunct, but the heavy emphasis on tagmemics in Naish’s master’s thesis makes it less accessible. Boas’s work is useful, but he was limited by the general linguistic sophistication of his time and as such many phenomena are not noted or are misunderstood. In contrast with grammatical descriptions, text collections are abundant, particularly with the excellent work of the Dauenhauers who provide records of a variety of oratorical and narrative forms (dauenhauer:; dauenhauer:; dauenhauer:) . S e sound system of Tlingit is complex, with between and consonants depending on dialect and era. It has received serious phonetic study only recently (maddieson:). e consonants are given in table ⁇. ere is a three-way distinction between unaspirated, aspirated, and ejective obstruents, as well as a two-way distinction between plain and ejective fricatives (except */ʃʼ/). Aspiration is neutralized in the coda, although all orthographies show obstruents as aspirated, whereas in fact they are released but unaspirated. Labialization is phonemic in the posterior consonants, but is not reflected in the anterior consonants. e ejective fricatives /xʼ/, /xʼʷ/, /χʼ/, and /χʼʷ/ are unique to Tlingit, having never been described in any other human language (Maddieson, p.c. ). ere are no voiced phonemes other than /j/, †/ɰ/, /w/, /n/, and %/m/. e velar approximant †/ɰ/ is extinct among living speakers and it is now realized as either /j/ or /w/ depending on the labialization environment, but was recorded as late as the s; it is still used in linguistic descriptions (wrien 〈ÿ〉) where it explains alternation between /j/ and /w/. ere are no labial consonants other than /w/ and %/m/, the laer of which is restricted to two Interior Tlingit communities in British Columbia and the Yukon where it is a morphophonologically predictable variant of Coastal Tlingit /w/ that may have developed from Athabaskan contact. e two consonants ?/ʔʷ/ and ?/hʷ/ are marginal, with the former occuring only occasionally as in Ḵakʼw.weidí /qʰakʼʷʔʷeːtí/ “Basket Bay clan”, and with ?/hʷ/ appearing in certain idiolectal (possibly communilectal) variants of the second person plural independent pronoun, e.g. yeehwáan /jiːhʷáːn/, a word which has extensive variability even within a single dialect. e Tlingit vowel system is quite simple, with four vowels /i/, /e/, /a/, and /u/. Both length and tone are distinguished in vowels, but the exact system varies among the dialects. e majority Northern dialect has high and low tones with a binary length distinction, a system which is represented in table ⁇. e Southern dialect, also known as Sanya-Henya, is described by leer: as having a tone register system with upper high/mid and lower mid/low tones, as well as a falling tone that occurs at register shi boundaries. Unfortunately I lack recordings or exposure to this dialect, and leer: offers very lile in the way of detail about this system. leer: (williams:) also describes the Tongass dialect as having a four-way system of vowel qualities, with plain, gloalized, murmured, and long forms for each vowel. e Tongass system seems to reflect the ancestral source of the tonal systems, and corresponds closely with vowels reconstructed in Proto-Athabaskan-Eyak. Nasal vowels do not occur phonemically in Tlingit except for the two Interior dialects that feature %/m/, where again the nasal vowels are probably due to Athabaskan contact. Compared to the Athabaskan family as a whole, Tlingit has a somewhat larger consonant inventory. No Athabaskan languages have ejective fricatives, and many lack uvular consonants. Although Athabaskan languages tend to have very few labial consonants, Tlingit essentially lacks them entirely if the labial-velar approximant /w/ is excluded. e total lack of voiced fricatives is also unusual in comparison to Athabaskan languages. Also interesting is the lack of voiced */l/ in Tlingit, although it is universal in Athabaskan languages. Despite these differences, the consonant system of Tlingit is still familiar. Indeed, Tlingit is much closer to Proto-Athabaskan than many Athabaskan languages, though it lacks the retroflex series, has velars rather than palatals, and lacks voiced fricatives. As for vowels, Tlingit’s lack of nasalization is interesting but not unprecedented among Athabaskan languages. Aside from nasalization its vowel system is strikingly similar to that of Navajo, although where Navajo has /o/ for a back round vowel, Tlingit instead has /u/. e tone system of Northern Tlingit parallels the many Athabaskan languages which have marked high tone, and the Tongass system is reminiscent of the vowels found in Eyak. unaspirated stop t k aspirated stop tʰ kʰ ejective stop t’ k’ % nasal stop (m) n fricative s ʃ ɬ x ejective fricative s’ ɬ’ x’ unasp. affricate ts tʃ tɬ aspirated affricate tsʰ tʃʰ tɬʰ ejective affricate ts’ tʃ’ tɬ’ † approximant j (ɰ) kʷ q qʷ ʔ kʰʷ qʰ qʰʷ k’ʷ q’ q’ʷ labial-gloal gloal labial-uvular uvular velar labial-velar posterior lateral bilabial alveolar postalveolar palatal anterior ? (ʔʷ) xʷ χ χʷ h ?(hʷ) x’ʷ χ’ χ’ʷ w Table : Tlingit consonants in IPA. % = dialectal; ? = marginal; † = extinct. íː éː á low i e ba ú central front í é long ba high central front short úː áː u a iː eː uː aː Table : Northern Tlingit vowels in IPA. .. O Tlingit writing is distinctly different from the relatively homogenous orthographies that are in use among Athabaskan languages. Tlingit suffers from an embarrasment of riches in terms of writing systems: there are several orthographies in current publication, and no one system is used in all communities. e two primary orthographies are what I call the Revised Popular orthography and the Canadian orthography. In addition there are a few publications which use variations on the original Naish-Story orthography, in particular the first noun dictionary (naish:), Dikée Anḵáwoo doo Yéet dàt John- kawshixidee Yoox̱’utúnk: e gospel of John in the Tlingit language (anon:), and the verb dictionary (story:). e Revised Popular orthography uses an underscore diacritic to represent uvular consonants 〈ḵ〉, 〈x̱〉, 〈g̱〉, high tone is marked with an acute accent 〈á〉, ejectives are marked with an apostrophe 〈t’〉, 〈s’〉, and gloal stops are represented with a period 〈.〉 which is also used to indicate separation of non-digraphs. e apostrophe is never used to indicate the gloal stop, unlike in Athabaskan orthographies. Because there is no voiced lateral articulation in Tlingit, the symbol 〈l〉 is unambiguously used to represent voiceless laterals. e long vowels are represented with English-inspired digraphs, so /eː/ is 〈ei〉, /uː/ is 〈oo〉, and /iː/ is 〈ee〉, but /aː/ is 〈aa〉; high tone is only marked on the first leer of a digraphic vowel: 〈óo〉. e Revised Popular orthography is the most common in publications, found in all of the Dauenhauers’ work (dauenhauer:; dauenhauer:; dauenhauer:), the second noun dictionary (naish:), and most academic publications in both linguistics and anthropology (cable:a; cable:; kan:; goldsmidt:; grinev:). Because the uvular underscore diacritic in 〈g̱〉 is inconvenient on computers and is typographically unpleasant, there are a variety of other forms in use such as 〈ǥ〉, 〈ɢ̱〉, and 〈ḡ〉; I use the laer in this paper. e Canadian orthography uses digraphs 〈kh〉, 〈xh〉, 〈gh〉 to represent uvulars, extending the use of 〈h〉 from the digraphs 〈ch〉 and 〈sh〉. e voiceless lateral fricatives are represented with the familiar barred ell 〈ł〉, although in affricate digraphs the unbarred ell is sometimes found instead: 〈tl〉. e ejectives and gloal stop are represented the same as in the Revised Popular system. Short low tone vowels are unmarked 〈a〉, a grave accent is used for long low tone vowels 〈à〉, an acute accent for high tone short vowels 〈á〉, and a circumflex accent for high tone long vowels 〈â〉. e most significant publications which use this orthography are the Interior Tlingit noun dictionary (leer:) and Gágiwduł.àt:¹ Brought forth to reconfirm (nyman:). At present this system is only used in Teslin, Carcross, and Whitehorse, with Atlin officials having recently decided to switch back to the Revised Popular system for their elementary school curricula. Although not in use for published works, there is another system which I call the E-mail orthography that uses the uvular digraphs from the Canadian orthography and the vowels from the Revised Popular orthography. is compromise system developed anonymously in response to restrictions on the coding of electronic text before the widespread adoption of Unicode. It is frequently seen on ephemeral printed materials, presentation slides, and occasionally even on regalia like drums and clothing. Proponents of the two major orthographies sometimes claim that it makes words “too long”, but its lack of official recognition from any authority makes it a popular alternative in the fractious world of Tlingit language politics. . D According to leer: there are few dialects in Tlingit despite a large geographic distribution, with high intelligibility between extant dialects. Based on my experience, it is entirely reasonable to claim that all Tlingit dialects are mutually intelligible, although speakers are acutely aware of differences from their own. leer: names the major dialects Northern Tlingit, Transitional Southern Tlingit, Sanya-Henya Tlingit, and Tongass Tlingit. I have renamed these for a few different reasons. First, I retain the term “Northern”, because the geographic division between the Northern varieties and the non-Northern varieties is roughly along Sumner Strait, with the varieties in the southeast being markedly different from those in the northwest. Second, I call the dialect spoken in Wrangell and Kake “Transitional” rather than “Transitional Southern”, with the reasoning that this dialect shares features of both the Northern dialect and the non-Northern dialects. My “Southern” dialect corresponds to Leerʼs Sanya-Henya, however I have renamed this because the term “Sanya-Henya” is limiting. e Saanyaa ḵwáan² and the Heinyaa ḵwáan are definitely speakers of this dialect, but the Kooyu ḵwáan may have spoken this dialect before being mostly wiped out by smallpox, and the Tʼaḵjikʼaan ḵwáan may have been speakers of this dialect before their . gági=-ÿu-du-l-ʔaat into.open=--.-[+, l, −]-go. . A ḵwáan is a sociogeographic grouping which cuts across clan lines, and is composed of nearby villages with close ties. forced merger with the Heinyaa ḵwáan in the late th century. us the term “Southern” is more inclusive for these other ḵwáan. I retain Leer’s term “Tongass” because it is appropriate as only the Taantʼa ḵwáan (resident at the village called Tongass in English) apparently spoke the dialect and the last two speakers selfidentified as Tongass people. e Tlingit dialects are diagrammed in figure ⁇. Note that I have not fully enumerated the Southern dialects for space reasons and because they are insufficiently documented. ere are no speakers of Tongass Tlingit alive today, probably less than ten Southern speakers mostly from Klawock,³ and only a small handful of Transitional speakers of whom most are from Kake. Northern speakers thus form the majority of present Tingit speakers, and hence their speech is the major focus of all documentation and revitalization efforts today. I focus on the Northern dialect in this paper. In terms of documentation, there is only a small amount published about the Tongass Tlingit dialect (williams:), and vanishingly lile about the Transitional or Southern dialects. Within each ḵwáan, territory was divided by matrilineal clans which themselves were organized into two moieties. Dialectal substructure, largely in the form of communilects, seems to have developed because of the regional organization of permanent selements, and communilects are more or less variable today depending on the intensiveness of exogamy in the last few generations and on immigration from now abandoned selements. I propose a likely substructure of the Northern Tlingit dialect in figure ⁇. e distinctiveness of different communilects within each dialect group is greater in the Interior dialects than it is among the Coastal dialects, and this distinction is found most strongly in certain regular phonological phenomena as well as borrowings from neighboring Athabaskan languages. On the other hand, the distinction between different communilects on the coast is mostly represented by semantic differences in temporal terminology (e.g. names of months and seasons), use of various contractions, and forms of certain highly variable function words. . W Free phrase order is uncommon among Athabaskan languages. thompson: says that Hupa and Koyukon have somewhat free phrase order, but most languages in the family have fixed SOV order outside of dislocation, focus, or subjectobject inversion. Going further afield, Michael Krauss (p.c. ) says that Eyak . I asked people to conduct a crude community census in early which turned up eight speakers, seven of whom were resident in or originally from Klawock. Tlingit Tongass Southern Transitional Saanyaa Heinyaa Northern Coastal Gulf Interior Figure : Traditional dialects of Tlingit, from south (le) to north (right). Northern Archipelago Outer Inner Gulf Interior Dry Bay Yakutat Atlin Teslin Tagish Sitka Angoon Juneau Hoonah Chilkat Figure : Northern subdialects and communilects. is clearly SOV except for dislocation – characterized by “comma” effects – and focus phenomena. Gary Holton (p.c. ) confirms this, claiming Eyak is beer described as V-final because “it is rare to have more than one direct argument expressed as a full NP or independent pronoun” – just as with Tlingit – and that other arrangements are used “for particular effect”. We can conclude that Eyak lacks free phrase order like most of the Athabaskan languages. In contrast with the Athabaskan languages and Eyak, free phrase order in Tlingit is possible in a variety of sentences, although it is oen difficult to make out due to the lack of words other than the verb. In his analysis of covert Ascrambling, cable: offered a paradigm of sentences with all possible phrase orders. In (⁇) the SOV form is fully glossed, and then in (⁇) I repeat cable:’s exhaustive variations. () O-S-[−, s]-teen “S see O” wé shaawátch xóots awsiteen wé shaawát-ch xóots a-ÿu--si-teen . woman- brown.bear ---[−, s, +]-see “the woman saw the bear” () a. b. c. d. e. f. (cable:) SOV: wé shaawátch xóots awsiteen OSV: xóots wé shaawátch wusiteen SVO: wé shaawátch wusiteen xóots OVS: xóots awsiteen wé shaawátch VSO: awsiteen wé shaawátch xóots VOS: awsiteen xóots wé shaawátch e SOV and SVO orders are unsurprisingly the most common in Tlingit. Nevertheless, all of the above examples are identical in meaning, although as Cable notes, “there are of course discourse-pragmatic effects associated with particular orders”. e astute reader might notice that examples (⁇) and (⁇) have a slightly different form of the verb. is is due to alternation in the third person object prefix which is explained in section ⁇. . T As with the Athabaskan languages, the verb in Tlingit has received the majority of every researcher’s aention. Nonetheless, the noun deserves its own share of research, and has some surprises in store for anyone who might care to poke at it. .. P Tlingit nouns can be divided into two major classes: possessable and unpossessable. Possessable nouns can have possessors, and unpossessable nouns cannot. Nearly all nouns in the language are possessable, with unpossessable nouns restricted to proper names. e possessable nouns can be further divided into alienable and inalienable subclasses. As with most systems of alienability, many nouns can be converted from one class into the other through regular morphological processes. As Ken Regh puts it, alienability “is a property of the relationship between noun and possessor, and is not a property of nouns themselves” (p.c. ). However, it is relatively clear from morphology that certain nouns have default membership in one or the other class, and are marked when converted to the other class. Inalienable nouns are generally easy to characterize. Nearly all body parts are represented by inalienable nouns, as are most kinship terms. ere is a large class of relational nouns which are also inalienable, and which represent spatial, temporal, or structural relationships between their possessors and other referents. e generic form of an inalienable noun includes a default possessor which may be any one of the indefinite nonhuman possessive pronoun a, the indefinite human possessive pronoun ḵaa, or the third person possessive pronoun du. Examples are ḵaa waaḵ “person’s eye(s)”, du jín “his/her hand(s)”, a kú “its (fish) tail”. Lexicographically, inalienable nouns are usually listed with a default possessor in parentheses, e.g. (a) déin “(its) vicinity”. A few exceptions to the rule of inalienable body parts are dook “skin”, sʼaaḵ “bone”, and dleey “flesh, meat”, all of which can be preserved and used aer death (though usually not that of humans!). Many characteristics of individuals are also represented by inalienable nouns, and can be profitably lumped together with the body parts. Alienable nouns comprise two major subclasses, one being the alienated body parts and the other being the “everything else” category. e “everything else” category is self-explanatory, but the alienated body parts deserve some discussion. Possession is marked with a possessed suffix -ÿí⁴ on the possessum as well as with a prenominal possessive pronoun, a system similar to the Athabaskan model. Alienable nouns are marked with the possessed suffix when they are the possessum in a possessive construction. In contrast, inalienable nouns require . e possessed suffix can take any of the forms -yí, -yi, -í, -i, -wú, -wu, -ú, or -u depending on labialization, tone sandhi, and final consonants. It is an excellent example of why the underlying velar approximant is still analytically relevant despite not being a real phoneme among living speakers. possessive pronouns, but are not marked with the possessed suffix. However, they can be alienated by the addition of a possessed suffix, which changes their relationship with the possessor. e resultant relationship is generally one of separation from the possessor, although this can vary in subtle ways. Probably the best way to understand this phenomenon is through examples. () a. du shá . head: “his/her head” b. xóots shá brown.bear head: “brown bear’s head” sháyi c. xóots xóots shá-ÿí brown.bear head- “brown bear head” In examples (⁇) and (⁇) the noun (a) shá “(its) head” is inalienable, and is in a normal relationship in that it is connected via the neck to the body. In contrast, in example (⁇) the addition of possessive marking has changed the relationship between head and body; it is implied that the head is separated from the body and is an independent object. English can approximate this distinction between the two relationships using the exact opposite possessive marking, as can be seen in the translations. .. N, , Tlingit has two numeral systems, an older quinquevigesimal system and a newer quinquedecimal system. Both systems have numbers from six to ten constructed from the numbers from one to five. us tléixʼ “one”, déix̱ “two”, násʼk “three”, dax.oon “four”, and keijín “five” have echoes in tleidooshú “six”, dax̱adooshú “seven”, nasʼgadooshú “eight”, gooshúḵ “nine”, and jinkaat “ten”. Both systems construct the teens in the same way, namely as jinkaat ḵa tléixʼ “ten and one”. e difference arises at twenty, which is deix̱ jinkaat “two ten” in the quinquedecimal system and tleiḵáa “one.man” in the quinquevigesimal system. One hundred hándit and one thousand taawsan are both loanwords from either English or Chinook Jargon or both. ere are special numerals for counting certain things and indicating distributives, all of which are formed from the basic numerals. For example, one person is tléináx̱, two people is dáx̱(i)náx̱, three people is násʼgináx̱, and so forth. Distributives are similar, with tlékʼḡaa “one by one”, dáx̱ḡaa “two by two”, násʼgiḡaa “three by three”, etc. Distributive numerals for people combine the two suffixal elements tlékʼḡaanáx̱ “one person at a time”, dáx̱ḡaanáx̱ “two people at a time”, etc. Numbers of repetions are counted with a different sort of suffix: tleidahéen “once”, dax̱dahéen “twice”, nasʼgidahéen “thrice”. Collectives of people or things can be counted with a different suffix, e.g. tleiyeeká “one group”, dax̱yeeká “two groups”, nasʼgiyeeká “three groups”. Diminutives are marked with a special suffix -kʼ, which has an allomorph -kʼw following labialized consonants and rounded vowels. It precedes the possessed suffix, causing the laer to be realized as a vowel. It has the usual meaning of diminutives, but sometimes may become somewhat lexicalized. () ax̱ hítkʼi káxʼ x̱waanoogú yé ax̱ hít-kʼ-ÿí ká-xʼ ÿu-x̱a-ÿa-nook-ÿí yé . house-- - -.-[−, , +]-sit- thus “I sit here in my lile house” (F. DeWi) A periphrastic diminutive can be formed with the adjective yádi which is derived from the alienable noun yát “child” plus the possessive suffix -ÿí, thus literally meaning “child o” something. An example of its use is shanaxwáayi yádi “hatchet” which literally means “child of an axe”. Plurality is a particularly interesting nominal phenomenon, with several different methods of marking but none which are obligatory or even very common. As will be seen in section ⁇, plurality is marked in the verb using plural pronominals, plural verb roots, a generic plural marker, and two distributives. On nouns it is perhaps most commonly marked with a suffix -xʼ (also -xʼw in labialized environments) which is homophonous with the primary form of the locative suffix. Like the diminutive, the plural suffix occurs before the possessed suffix. I have not encountered the diminutive and plural together, although I have not aempted to elicit it. () du yátxʼi du yát-xʼ-ÿí . child-- “his children” ere are three other morphemes which mark collectives in various situations, generally used only with terms referring to humans. e collective sáni is found with terms for relatives and requires a plural suffix, for example du yádxʼi sáni “her babies” (naish:). e collective yán is uncommon, but appears in a set phrase ax̱ káani yán “my brothers-in-law” which is used frequently in ceremonial oratory. e collective hás is found in postnominal position which disambiguates it from the third person plural hás. () .. a. hasdu káak hás jeeyáx̱ hasdu káak hás jín-yáx̱ . mat.uncle hand- “like their maternal uncles’ hands” b. hás, a yáx̱ áwé, yéi has awsinei hás a yáx̱ áwé yéi=has-a-ÿu--si-nei it . thus=----[−, s, +]-do “they, like that, they did it so” (F. DeWi) (F. DeWi) C Tlingit has a relatively large inventory of case suffixes and relational postpositions. e ergative - marks noun phrases which are subjects of transitive verbs. It also does double duty as an instrumental marker in some sentences, but the exact situations in which it arises are not described. Finally it also seems to serve a special function where it answers the question “why”. As is fairly common among languages with ergative marking, Tlingit lacks an absolutive marker. ere are eight oblique cases in Tlingit, namely the locative -xʼ “at, in, by, on”, the allative -déi “to, toward, until, in the manner o”, perlative -náx̱ “along, by, via, during”, ablative -dáx̱ “from, out o”, adessive -ḡaa “around, about, by, for”, punctual -t “at a point, to a point, about a point”, pertingent -x̱ “at, against, in the form of, concerning”, and instrumental -teen “along with, using, as soon as”. e oblique cases are distinguished from postpositions by their phonological behavior. Four cases consist solely of consonants, and must be considered suffixes. ere are three cases which are syllabic but undergo tone sandhi, and since tone sandhi does not occur across word boundaries they must be suffixes. Finally the last one is considered to be a suffix purely by custom, and it is uncertain whether it may in fact be an independent postposition. e locative case and the instrumental case show significant allomorphy. e locative has four forms, -xʼ, -ː, -i, and -, with the variants occuring in phrases that are bound to the verb. e vowel lengthening allomorph occurs aer open syllables with high tone, thus simply lengthening the final vowel. It also occurs uniquely with the first person singular independent pronoun x̱át where the combination x̱át-ː is x̱áa. e other allomorphs have similar phonological restrictions. e instrumental case varies primarily between -teen and -(ee)n, with the laer occuring in conditions which are similar to those of the locative. One consultant has suggested to me that the two forms of the instrumental are spliing, with the -(ee)n form being used more with humans and hence more comitative and the -teen form being more frequent with inanimates; this is a problem that needs corpus research. e relational postpositions are described by leer: as being “opaque combinations of relational noun plus postposition”. Whatever their internal structure, they are phonologically independent words which indicate specific relationships with their host nouns. e class comprises the similative yáx̱ “like, as, similar to, equivalent with”, abessive ḡóot “without, lacking”, elative náḵ “away from”, benefactive yís “for, benefiting”, superlative yáanáx̱ ~ yáanax̱ “more than”, and the sublative ḵín “less than”. e similative and benefactive seem to be undergoing a change towards becoming more like a case suffix, as there are a number of basic words where phonological changes make the postposition part of the word. is phenomenon may be more common in the non-Northern dialects, but there is only sparse evidence as of yet. . T As a cousin of the Athabaskan languages, Tlingit is traditionally described using a slot-and-filler template for the verb. I follow this tradition here although the template should not be taken as theoretically justified.⁵ e most recent published description of the verb template is from cable:a e template used here is extended from his description, and is given in table ⁇. A nearly complete enumeration of the affixes can be seen in table ⁇. e Tlingit verb is of roughly the same complexity as the Athabaskan verb in terms of number of slots, long distance dependencies, and distribution of functions. As with all Athabaskan languages, the lexical entry is a theme which includes not only the root but also the classifier and various other affixes. Tlingit does not have a clearly defined division between disjunct and conjunct zones, but there are some phonological indications that the lemost prefix slots are more independent than those closer to the root. For example, some preverbs and incorporates have lexical high tone whereas the prefixes close to the verb never . For more on the template morphology controversy in Athabaskan languages, see rice:b and mcdonough:b among others. disjunct conjunct Cable & Crippen Leer (1991) + + + + + + + preverbs reciprocal & distributive plural number object areal alienable incorporated nouns inalienable incorporated nouns + proclitic adjunct phrases + b number prefixes a ” + b incorporated obj. pronominals + + + + + + + + + + vertical surface yahorizontal surface kaself-benefactive outer aspect/conjugation irrealis inner aspect/conjugation perfective inner distributive subject classifier root b ” a ” + e schetic prefixes d ” c ” b ” a ” + distributive prefix + subject pronominals + classifier − derivation − − a b − − − suffix − duration − − − − a incorporated alienable nouns + c incorporated inalienable nouns stem variation modes modes postverbal auxiliaries derivational suffixes inner durative suffixes outer durative suffixes inner mode suffixes outer mode suffixes epimode & clause type suffixes Table : Tlingit verb template. Prefixes are positive, suffixes are negative. Suffix classifications for the Cable & Crippen column are tentative. e disjunct and conjunct zones are for organizational purposes and are not necessarily justified. Position Affixes + () ḡunayéi ~ ḡunéi=, áa=, shóo=, héeni=, gági=, éeḡi=, daaḡi= () ḵut=, yux̱=, yaax̱=, héenx̱=, ux̱=, ḵwáaḵx̱=, yedx̱=, ÿaanax̱ ~ ÿaʻnax̱T = () yan*=, neil*=, haa*=, yóo*=, ḵux̱*=, kux*= () kei=, yei=, yeeḵ ~ yeiḵ ~ iʻḵT =, daaḵ=, daak= () yéi ~ yeʻT = () yaa= () yaa=, yoo= + dax̱-, woosh+ has ~ s+ x̱at ~ ax̱-, haa-, i-, yee-, ash-, a ~ -, ḵu ~ ḵaa-, at-, aa-, sh ~ + ḵu+ yaan-, shakux-, yata-, x̱ʼasakw-, ḡax̱-, xee ~ xei-, ḵee ~ ḵei-, yee ~ yei-, kanik-, yaḵa-, saa-, aan-, naa-, sha.ax̱w-, yakw-, hin+ ji-, x̱ʼa ~ ḵʼa-, tu-, sha-, shu-, lu-, se ~ sa-, x̱a-, gu-, ta-, daa-, x̱oo-, x̱an-, x̱ʼaa-, tʼéi-, t’aa-, yik-, yee-, ḵi-, gin-, x̱i-, sʼaan-, lidíx̱ʼ-, waḵ-, sʼaḵ-, x̱ʼus-, sʼee-, duk-, laka-, tlʼaḵ-, keey-, tóoxʼ-, x̱ʼatu-, tuḵx̱ʼe ~ tuḵʼe-, daa.it-, tax̱ʼ+ ÿa+ ka+ ga+ ga-, ḡa+ u+ -, na-, ḡa+ ÿu+ daḡa ~ dax̱+ x̱a-, tu-, ee-, yi-, -, du-, du+ [d, s, i ]: d ∈ {+, −}; s ∈ {, s, l, sh}; i ∈ {+, −} root − -án, -shán, -ch, -áḵw, -aa, -x̱aa, -ÿí, -ee, -k(w), -(chʼ)álʼ, -ḵ, -nas, -násʼ, -kátʼ − () -ʻ, -k, -x̱, -ch () -t, -xʼ, -tʼ, -sʼ, -lʼ − -ː, -ʻ, -ʼ, -ÿ, -n − -ch, -nee ~ ee, -ín ~ ún − -ee, -een, -(ee)ḵ, -ÿi − =nooch ~ neechI ~ nukchY , =noojeen ~ ?neejeenS , =nóok(w) ~ néekwS , =núgwni ~ nígwniS , =ḡanú gun ~ ḡanígúnI ~ ḡaníkwS Table : Tlingit verb affixes. E preverbs can take certain locative suffixes. Subscripts mark dialect-specific forms (T: Tongass, S: Southern, I: Interior, Y: Yakutat). have tone. A loose boundary might be posited between the incorporated nouns and the vertical surface ÿa- and horizontal surface ka- prefixes, since the laer two may lose vowels but the incorporates in most cases do not. For organizational reasons I will adopt this division here, but it must be stressed that this is essentially stipulative and not firmly established. Indeed, the boundary could justifiably be placed between the self-benefactive and the horizontal surface prefix, or even between the incorporates and the objects. One interesting thing about the Tlingit verb is that close examination reveals that Tlingit has a split between intransitive types. is phenomenon is termed “fixed-S” by dixon: “active/agentive” by mithun: and “active” by ogrady: Among all verb themes there is a fixed division between those that are unergative [+S, −O] and those that are unaccusative [−S, +O]. is is indicated in the themes by presence of only an object slot or only a subject slot. .. C e classifier is a portmanteau morpheme [±, , ±]→(d)C(V)- composed of three functional elements, the D component, the S component, and the I component. It is distinctly more complex than the classifier in the Athabaskan family. e various forms of the classifier are given in table ⁇. Phonologically the D component represents the presence or absence of the consonant d, the S (“series”) component represents one of the consonants s, sh, l, or the lack of a consonant as , and the I component represents the vowel i if positive or a if negative. Functionally the D component represents middle voice,⁶ the S component represents valency and/or noun class,⁷ and the I component disagrees with the irrealis prefix. e S component is lexically specified in the majority of verb themes although there is some regular alternation. For example sh generally represents unpleasant or deprecatory forms of or s themes, and pairs of statives and s causatives are common. e D component appears as + in nearly all semantically middle voice verbs even if this is not otherwise indicated with pronominals; it is also lexically specified in some themes, e.g. O-S-[+, ]-naa “S drink O”, Osha-ka-[+, l]-kooʼ “O have curly hair”, O-S-[+, s]-ḡaax̱ “S ask for O with no intention of returning it”. When compared to the Athabaskan classifier, the Tlingit classifier is more . e D component is + in reflexives, reciprocals, self-benefactives, revertive motion with ḵux̱=, locomotives (“move while doing”), dissimulatives (“pretend to do”), lusives (“play at doing”), and thematically. See naish: krauss: and leer: . is noun classification function is the source of the term “classifier”. Although this function is largely absent within the Athabaskan family, see leer: −D +D S↓ −I +I −I +I s l sh ÿa- da- disasis- dzilalil- dlisha- shi- sh- ji- Table : e Tlingit classifier (+). complex and perhaps more productive. e Athabaskan D component maps directly to the Tlingit, a feature reconstructed in Proto-Na-Dene (leer:). e S component in Tlingit is similar to the ł component of the Athabaskan classifier, but instead of being a binary opposition of - vs. ł-, it is a quaternary system {, s, l, sh}. Finally, the I component of the Tlingit classifier appears to be unique, since the Athabaskan classifier is not described as having any vowel alternation paerns, nor does it seem to indicate any realis/irrealis distinctions. Overall, however, the functions of the classifier are quite similar on both sides, with the primary function being voice and valency indication, and with a variety of idiosyncratic and lexically specified uses. .. C As noted earlier there is no solid definition of conjunct versus disjunct in Tlingit, but I have stipulated for the purposes of this paper that this division exists between the incorporated nouns and the vertical surface ÿa- and horizontal surface ka- prefixes. e functions of the so-called conjunct prefixes in Tlingit are clearly similar to those in the Athabaskan conjunct zone. In particular the aspect/conjugation prefixes are strongly reminiscent of the Athabaskan ones, although they are not necessarily identical historically. e innermost prefixes of the conjunct zone are the subject pronominal prefixes. Tlingit marks first and second person singular (x̱a-, ee-) and plural (tu-, yi-), as well as third person (-), third person obviate (du-), and indefinite human (du-). is is in marked contrast with the Atahbaskan family, where third person subjects appear near the object pronominals, and only the first and second person subjects are found next to the classifier. e distributive prefix daḡa- ~ dax̱is immediately le of the subjects, and in some ways can be considered part of the subject pronominals. It has an allomorph dax̱- which appears in the disjunct zone just rightward of the preverbs, in the same slot as the reciprocal woosh-. e Tlingit aspect/conjugation prefixes ga-, -, na-, ḡa-, and ÿu- are reminiscent of the situation and viewpoint aspect prefixes of the Athabaskan verb, but their functions are noticeably different. e perfective ÿu- is the sole form of perfective marking in Tlingit, in contrast to the lexically specified four way system {, n, s, ɣ} of perfectives in Proto-Athabaskan; the same is true for the imperfectives. Due to this lack of variation, the perfective is only an aspectual prefix, and has no conjugational function. is can be seen in the following examples, where arbitrary verbs all show the same perfective marking. () a. O-ÿa-ka-S-[−, ]-dlaakw “S scratch face of O” dóosh ash yakaawadlákw dóosh ash-ÿa-ka-ÿu--ÿa-dlaakw cat .-----[−, , −]-scratch “the cat scratched up his face” (story:) b. a-S-[+, l]-seen “S hide self out of sight” awdlisín nadáakw tayeexʼ nadáakw ta-yee-xʼ a-ÿu--dli-seen table underside-beneath- ---[+, l, −]-hide “he hid beneath the table” (story:) In contrast with the tendency in Athabaskan to have only one aspect/conjugation prefix in a verb, in Tlingit the combination of aspect/conjugation prefixes is relatively common. e most common example is ga-u-ḡa-, a string which indicates the future, with the two aspect/conjugation prefixes ga- and ḡa- as well as the irrealis u-. In this string the meaning of the two aspect/conjugation prefixes is essentially opaque. I oen gloss the sequence in combination with the subject prefix as a single portmanteau morpheme for future tense to simplify analysis, as in the following examples. () a. O-sha-S-[−, l]-dzoo “S hit O on head with missile” gánch ashagux̱ladzóo gán-ch a-sha-gux̱-la-dzóo firewood- -head-.-[−, l, −]-hit.with.missile “he will hit it on the head with firewood” b. O-S-[−, s]-teen “S see O” yéi ikḵwasateen yéi=i-kḵwa-sa-teen thus=.-..-[−, s, −]-see “I will see you” (story:) e conjugational use of the aspect/conjugation prefixes appears for example in the imperative, where one of the the non-perfective prefixes is lexically specified for a given theme, as shown in the following examples. (Note that the verb root nook is irregular in losing its final consonant, and that . is null in the imperative.) () a. S-[−, sh]-taa “S sleep (angry speaker)” nashtá! na--sh-taa -.-[−, sh, −]-sleep “go to sleep, dammit!” b. S-[−, ]-nook “sg. S sit down” ḡanú! ḡa---nook ̱-.-[−, , −]-sit. “sit down!” c. O-S-[−, ]-x̱aa “S eat O” x̱á! ---x̱aa -.-[−, , −]-eat “eat!” e self-benefactive prefix ga- is a rare, lile used prefix in the conjunct zone. It requires + in the classifier, and indicates an action done for one’s own benefit. It also apparently occurs in the comparative according to leer: but he does not provide any examples of this. () O-S-[−, s]-ʔee “S cook O” a. x̱wasi.ée -ÿu-x̱a-si-ʔee --.-[−, s, +]-cook “I cooked it” b. gax̱wdzi.ée -ga-ÿu-x̱a-dzi-ʔee-? ---.-[+, s, +]-cook “I cooked it for mysel” (story:) .. D e prefixes in Tlingit which might be considered analogous to the disjunct prefixes in Athabaskan are roughly those leward of the horizontal surface ka- and vertical surface ÿa-. is includes the incorporates, the areal, the objects, the plural, the reciprocal, the distributive, and the preverbs. e division between conjunct and disjunct is somewhat arbitrary, but it is not insignificant that the rightmost prefixes which can bear lexical tone are among the inalienable incorporated nouns. e incorporates are split into two slots, although (leer:) does not give any examples where prefixes from both groups occur and I have yet to encounter any such forms. A possible distinction is that the inalienables can take object prefixes as possessors, but the alienable incorporates do not. e division between the two is instead functional, with the inner slot composed of nearly all inalienable nouns and the outer slot composed of alienable nouns. e only noninalienable nouns which appear amongst the inalienables are sʼaḵ- from sʼaaḵ “bone” (cf. a sʼaaḡí “its bone”) and duk- from dook “skin” (cf. a doogú “its skin”). ere are fewer alienable incorporates, and they generally derive from abstract concepts like yaan- “hunger”, kanik- “news”, and ḵee ~ ḵei- “dawn”, although there are a few concrete examples like yakw- “canoe” and hin- “fresh water”. e inalienable incorporates are more common than the alienable incorporates, and can be found in a number of very common verbs. A couple of examples of various incorporates are given below. () a. P-eet yaan=[−, ]-haa “P be hungry” du éet yaan uwaháa du-ʔée-t yaan=ÿu-ÿa-haa .-- hunger=-[−, , +]-fit “he is hungry” (story:) b. O-x̱ʼa-ÿa-[−, l]-sʼeilʼ “O tear loose from hook in mouth” daak ax̱ʼayawlisʼélʼ daak=a-x̱ʼa-ÿa-ÿu-li-sʼeilʼ seaward=-mouth---[−, l, +]-tear “it (salmon) tore loose from the hook” (story:) Unlike some Athabaskan languages (rice:), there are no incorporates which can be clearly distinguished as activity incorporates. e alienable ḡax̱- derived from ḡaax̱ “crying, weeping” might be construed as one, but it is not used for “do X while crying” but instead functions more like “do X which involves crying”, implying that the activity denoted by the verb is crying and thus crying is not a cooccuring activity. e following example shows ḡax̱- in use. () ḡax̱=S[]-[−, s]-tee “pl. S cry” kei ḡax̱ gax̱yisatée kei=ḡax̱-gax̱yi-sa-tee up=cry-..-[−, s, −]-be “you (pl.) will cry” (story:) Leward of the incorporates are the objects, which are more numerous than the subjects and have distinctly different forms. ey are first person singular x̱at ~ ax̱-, first person plural haa-, second person singular i-, second person plural yee-, third person proximate ash-, third person a ~ -, indefinite human ḵu ~ ḵaa, indefinite nonhuman at-, partitive aa-, and reflexive sh ~ -. e third person object prefix varies between a- and - in a way that is vaguely reminiscent but not the same as the yi- vs. bi- alternation in Athabaskan languages. In Tlingit the alternation is purely morphological, with - being the usual unmarked form and a- appearing only when the subject is third person - or is not present, except when a noun phrase marked with the ergative suffix - immediately precedes the third person object – requiring that no other prefixes le of the object are present – in which case the prefix switches back to -. () O-S-[−, ]-x̱aa “S eat O” a. x̱waax̱aa -ÿu-x̱a-ÿa-x̱aa --.-[−, , +]-eat “I ate it” b. aawax̱aa a-ÿu--ÿa-x̱aa ---[−, , +]-eat “he ate it” e following examples illustrate the effect of ergative-marked noun phrases which cause the third person object prefix to switch back to - even when there is a third person object. () O-S-[−, s]-teen “S see O” a. wé shaawátch xóots awsiteen wé shaawát-ch xóots a-ÿu--si-teen . woman- brown.bear ---[−, s, +]-see “the woman saw the bear” (cable:) b. xóots wé shaawátch wusiteen xóots wé shaawát-ch -ÿu--si-teen brown.bear . woman- ---[−, s, +]-see “the woman saw the bear” (cable:) Beyond the objects are what leer: terms the number prefixes, one slot containing the plural number s ~ has- and the next containing the reciprocal wooshand the distributive dax̱-. e plural prefix is somewhat common, indicating number of either the subject or object third person arguments, and can appear even when plurality is unambiguous from other sources like verb roots and plural nouns. It is clearly related to and perhaps derived from the plural particle hás and the plural prefix has- found in the independent third person plural possessive pronoun hasdu. Some speakers use s- very frequently, which seems to be a characteristic of Southern and Transitional dialects, however other speakers tend to use has- more oen instead, which seems to be characteristic of Northern dialects. () yándei yaa s nakwán yán-dei=yaa=s-na---kwán shore-=along=---[−, , −]-swim. “they are swimming ashore” (story:) e most leward morphemes in the verb are the preverbs, which in function should be immediately familiar for Athabaskanists, though the forms are mostly quite different. (leer:) divided the preverbs (his “proclitic adjunct phrases”) into six or seven subgroups which I have retained. e majority of Tlingit preverbs refer to direction although there are a few manner preverbs as well. e most abstract preverb is yéi= “thus”, the others have more obvious meanings. A number of preverbs can take certain locative case suffixes: punctual -t, pertingent -x̱, and allative -déi. .. S Tlingit has a much more elaborate suffix system than what is found among the Athabaskan languages. e aspectual suffixes described for Athabaskan have a parallel described by leer: in what he calls the “inner mode suffixes”, and which I call simply “stem variation”. Most of these are notional suffixes which represent the regular apophonic changes of the vowel in the verb root, although the -n suffix does surface as a separate segment along with its associated vowel change which can be seen in the following example. () O-S-[−, ]-x̱aa “S eat O” a. at guḡax̱áa at-ga-u-ḡa---x̱aa-ʼ .---̱--[−, , −]-eat- “he will eat (something)” b. yándei yaa anax̱éin yán-dei=yaa=a-na---x̱aa-n -=along=---[−, , −]-eat- “he’s almost finished eating” (story:) Because I do not fully understand the stem variation process, and because the extant verb documentation provides only a few notes on what kinds of paerns exist in the lexicon, I generally avoid glossing stem variation and instead simply treat the apophony as an opaque process as one might for example with Koyukon. story: offers more examples with an analysis based strictly on the Northern dialect, but does not describe the process in terms of a suffix. e derivational suffixes are infrequent, and of varying productivity. ey attach to certain roots and derive a new verb with a related meaning. e original root undergoes apophony when aached with a derivational suffix and its resulting form is fixed, i.e. root+derivational forms do not have stem variation. e resulting verbs are generally paradigmatically restricted, with only certain conjugations available. e most frequent and semantically transparent derivational suffix is the amissive -x̱aa which indicates missing a target. It can only combine with verbs whose meaning includes aiming at targets, and story: give seven verbs which include the suffix. e example below compares the ordinary verb for “shoot” with its amissive derivation “shoot and miss”. () a. O-S-[−, s]-ʔoon “S shoot O” dzískw x̱waa.ún dzískw -ÿu-x̱a-ÿa-ʔoon moose --.-[−, , +]-shoot “I shot a moose” (story:) b. O-ya-S-[−, s]-ʔún-x̱aa “S shoot and miss O” ayawsi.únx̱aa a-ya-ÿu--si-ʔún-x̱aa ----[−, s, +]-shoot- “he shot and missed it” (story:) Other derivational suffixes include the restorative -án, simulative -aa, intensive -(sh ~ ʼ)án, deprivative -( ~ y ~ ʼ)áḵw, denominal -ÿí, liabilitive -ee, coplural -ḵ, and several with unknown meanings: -k(w), -(ʼ)álʼ, -ʼ, -nas, -násʼ, and -kátʼ. e durational suffixes indicate the duration of an event that is described by the verb. Only some of these suffixes are fully productive, others are apparently restricted to certain verbs. e most common and fully productive are the frequentive -, repetitive -k(w), and habitual -x̱(w). e plural object suffix -xʼ is infrequent but seems to be somewhat productive, the other plural object suffix -tʼ is rare and unproductive. e serial -sʼ indicates actions repeatedly in a sequence. ere are also the suffixes -t and -lʼ which are extremely rare with unknown functions. Examples of a few of the durational suffixes are given below. () a. S-[+, ]-tees “S shuffle” nadatísch át á-t na--da-tees-ch it- --[+, , −]-shuffle- “they shuffle around noisily on it” b. O-ya-S-[−, s]-ḵaa “S say O” yéi yoo ayasiḵéik yéi=yoo=a-ya---si-ḵaa-k thus==----[−, s, +]-say- “he says it (so) again and again” c. O-S-[+, s]-xʼaa “S twist O” sʼú áwé dusxʼéix̱ sʼú áwé --du-s-xʼaa-x̱ root . --.-[+, s, −]-twist- “they always twist roots” d. awlixʼéisʼ a-ÿu--li-xʼaa-sʼ ---[−, l, +]- “he twisted it over and over” (story:) (naish:) (story:) (story:) e. O-ka-S-[−, l]-xaakw “S grind O” aklaxákwlʼ a-ka---la-xaakw-lʼ ----[−, l, −]-grind- “he’s grinding it” (naish:) e mode suffixes cover the contingent -(i/u)n, conditional -ni, decessive -een, optative -(ee)ḵ, aributive -ÿi, and subordinative -ÿí. ey are relatively frequent and are fully productive paradigmatically. e examples below illustrate a few of these suffixes. () a. O-ka-S-[−, ]-heix̱ “S plant O” akahéix̱een a-ka---heix̱-een ---[−, , −]-plant- “he used to plant” b. S-[−, ]-goot “sg. S go” gwál haax̱ ugoodeeḵ gwál haa-x̱=u---goot-eeḵ hither-=--[−, , −]-go.- “I wish he’d come here” c. haat ḡagúdin haat=ḡa---goot-in hither=̱--[−, , −]-go.- “whenever he comes here” d. át igútni á-t i--goot-ni it- .-[−, , −]-go.- “if/when you get there” (naish:) (naish:) (naish:) (naish:) e enclitic auxiliaries are meaningless hosts for several of the verb suffixes. eir use is somewhat idiosyncratic: some speakers almost never use them, some speakers use them extensively, and most are somewhere in the middle. ey arise when suffixes and root shape produce a complex consonant cluster that is phonotactically undesirable, in which case suffixes are shied to the auxiliary. eir shapes are conventionalized but vary between the dialects in parallel with variation in other specific verb roots. ere are auxiliaries for the frequentative, frequentative decessive, repetitive, conditional, and contingent. Use of an auxiliary precludes stem variation, although the reason for this is not clear. e origin of the auxiliary host is obscure, but it appears to be from a verb which might have been O-S-[−, ]-neekw meaning loosely “S feel O”, or perhaps P-t S-[−, ]-neekw “S feel like doing P” (leer:). () a. sh kanx̱alneekch sh-ka-na-x̱a-l-neek-ch .---.-[−, l, −]-tell- “I always tell the story” b. sh kanx̱alneek nooch sh-ka-na-x̱a-l-neek=nooch .---.-[−, l, −]-tell=. . R . A , , + + person allative, “to, toward” alternative, “back and forth” areal, “space, environment, weather, extent” aributive auxiliary meaningless base for pronominal aachment classifier: voice, valency, realis, noun class middle voice or thematic {, s, l, sh}: valency, noun class, or thematic realis conditional, “if, when” contingent, “if ever, whenever” decessive demonstrative diminuitive distal, far from speaker and listener distributive ergative case, subject of transitive verb focus frequentative (leer:) ̱ future tense ga-conjugation/aspect ḡa-conjugation/aspect habitual horizontal surface, “top, flat, surface” inalienable noun indefinite human, “somebody” indefinite nonhuman, “something” instrumental case, “using, with” irrealis locative, “at, in, by, on” mesiodistal, nearer to listener mesioproximal, nearer to speaker na-conjugation/aspect negative object obviate discourse argument optative perfective perlative case, “through, along, across” pertingent case, “at, in contact with, form of, member o” plural punctual case, “at/to a point” progressive proximal, closest to speaker proximate discourse argument possessive pronoun or possessed suffix reciprocal repetitive reversive, “back, backwards” reflexive subject self-benefactive serial simulative, “like, as, similar to, equivalent with” singular unspecified suffix terminative, “cease, end, coming to a point” stem variation vertical surface, “side, face, flat” -conjugation/aspect