Central Alaskan Yupʼik

Yup'ik contrasts four vowel qualities: /a i u ə/. The reduced vowel /ə/ always manifests phonetically short in duration, but the other three vowel qualities may occur phonetically short or long: [a aː i iː u uː]. Phonetically long vowels come about when a full vowel (/a i u/) is lengthened by stress (see below), or when two single vowels are brought together across a morpheme boundary. The effect is that while phonetic vowel length may yield a surface contrast between words, phonetic length is predictable and thus not phonemically contrastive.[8]

The vowel qualities [e o] are allophones of /i u/, and are found preceding uvular consonants (such as [q] or [ʁ]) and preceding the low vowel [a].[8]

Yup'ik does not contrast voicing in stops, but has a wide range of fricatives that contrast in voicing. The phoneme /l/ is not phonetically a fricative, but behaves as one phonologically in Yup'ik (in particular with regard to voicing alternations, where it alternates with [ɬ]; see below). Contrasts between /s/ and /z/ and between /f/ and /v/ are rare, and the greater part of the voicing contrasts among fricatives is between the laterals /l/ and /ɬ/, the velars /x/ and /ɣ/, and the uvulars /χ/ and /ʁ/. For some speakers, there is also a voicing contrast among the nasal consonants, which is typologically somewhat rare. Any consonant may occur as a geminate word-medially, and consonant length is contrastive.[8]

The table above includes the allophones [χʷ], [ts], and [w]. The voiceless labialized uvular fricative [χʷ] occurs only in some speech variants and does not contrast with its voiced counterpart /ʁʷ/. The voiceless alveolar affricate [ts] is an allophone of /tʃ/ before the schwa vowel. The voiced labiovelar approximant [w] is an allophone of /v/ that typically occurs between two full vowels, excepting when it occurs adjacent to an inflectional suffix. For example, /tʃali-vig-∅/ "work-place-ABS" is pronounced [tʃaliːwik] (orthographically, calivik), since /v/ occurs between two full vowels and it not adjacent to the inflectional suffix. With /tʃav-utə/ "oar" by contrast, since /-utə/ is an inflectional suffix, /v/ does not undergo the allophonic alternation: [tʃavun] (cavun).[8]

In Norton Sound, as well as some villages on the lower Yukon, /j/ tends to be pronounced as [z] when following a consonant, and geminate /jː/ as [zː]. For example, the word angyaq "boat" of General Central Yup'ik (GCY) is angsaq [aŋzaq] Norton Sound.[12][8]

Conversely, in the Hooper Bay-Chevak (HBC) dialect, there is no /z/ phoneme, and /j/ is used in its place, such that GCY qasgiq [qazɣeq] is pronounced qaygiq [qajɣeq]. HBC does not have the [w] allophone of /v/, such that /v/ is pronounced [v] in all contexts,[12] and there are no labialized uvular fricatives.[8]

In the Nunivak dialect, one finds /aː/ in place of GCY /ai/, such that GCY cukaitut "they are slow" is pronounced cukaatut, there is no word-final fortition of /x/ and /χ/ (see below), and word-initial /xʷ/ is pronounced [kʷ].[12]

There are a variety of voicing assimilation processes (specifically, devoicing) that apply mostly predictably to continuant consonants (fricatives and nasals); these processes are not represented in the orthography.[12]

• The voiced fricatives /v, z, l, ɣ, ʁ/ undergo voicing assimilation when adjacent to the voiceless stops /p, t, k, q/; this occurs progressively (left-to-right) and regressively (right-to-left). Thus ekvik is pronounced [əkfik], and qilugtuq /qiluɣ-tu-q/ is pronounced [qiluxtoq] (compare qilugaa /qiluɣ-a-a/ [qiluːɣaː]).

• Progressive voicing assimilation occurs from fricatives to fricatives: inarrvik /inaχ-vik/ is pronounced [inaχfik].
• Progressive voicing assimilation occurs from stops to nasals: ciut-ngu-uq "it is an ear" is pronounced [tʃiutŋ̊uːq].
• Progressive voicing optionally occurs from voiceless fricatives to nasals: errneq is pronounced [əχn̥əq] or [əχnəq].[8]

Occasionally these assimilation processes do not apply, and in the orthography an apostrophe is written in the middle of the consonant cluster to indicate this: at'nguq is pronounced [atŋoq], not [atŋ̊oq].[12]

Fricatives are devoiced word-initially and word-finally.[8]

Another common phonological alternation of Yup'ik is word-final fortition. Among consonants, only the stops /t k q/, the nasals /m n ŋ/, and the fricative /χ/ may occur word-finally. Any other fricative (and in many cases also /χ/) will become a plosive when it occurs at the end of a word. For example, qayar-pak "big kayak" is pronounced [qajaχpak], while "kayak" alone is [qajaq]; the velar fricative becomes a stop word-finally. Moreover, the [k] of -pak is only a stop by virtue of it being word-final: if another suffix is added, as in qayar-pag-tun "like a big kayak" a fricative is found in place of that stop: [qajaχpaxtun].[8]

The voiced velar consonants /ɣ ŋ/ are elided between single vowels, if the first is a full vowel: /tuma-ŋi/ is pronounced tumai [tumːai] (with geminate [mː] resulting from automatic gemination; see below).[8]

Yup'ik has an iambic stress system. Starting from the leftmost syllable in a word and moving rightward, syllables usually are grouped into units (termed "feet") containing two syllables each, and the second syllable of each foot is stressed. (However, feet in Yup'ik may also consist of a single syllable, which is almost always closed and must bear stress.) For example, in the word pissuqatalliniluni "apparently about to hunt", every second syllable (save the last) is stressed. The most prominent of these (i.e., the syllable that has primary stress) is the rightmost of the stressed syllables.[8]

The iambic stress system of Yup'ik results in predicable iambic lengthening, a processes that serves to increase the weight of the prominent syllable in a foot.[17] When lengthening cannot apply, a variety of processes involving either elision or gemination apply to create a well-formed prosodic word.[12][8]

Iambic lengthening is the process by which the second syllable in an iambic foot is made more prominent by lengthening the duration of the vowel in that syllable.[17] In Yup'ik, a bisyllabic foot whose syllables each contain one phonologically single vowel will be pronounced with a long vowel in the second syllable. Thus pissuqatalliniluni /pisuqataɬiniluni/ "apparently about to hunt" is pronounced [(pi.'suː)(qa.'taː)(ɬi.'niː)lu.ni]. Following standard linguistic convention, parentheses here demarcate feet, periods represent the remaining syllable boundaries, and apostrophes occur before syllables that bear stress. In this word the second, fourth, and sixth syllables are pronounced with long vowels as a result of iambic lengthening.[17][18][19] Iambic lengthening does not apply to final syllables in a word.[12][8]

Because the vowel /ə/ cannot occur long in Yup'ik, when a syllable whose nucleus is /ə/ is in line to receive stress, iambic lengthening cannot apply. Instead, one of two things may happen. In Norton Sound dialects, the consonant following /ə/ will geminate if that consonant is not part of a cluster. This also occurs outside of Norton Sound if the consonants before and after /ə/ are phonetically similar. For example, /tuməmi/ "on the footprint" is not pronounced *[(tu.'məː)mi], which would be expected by iambic lengthening, but rather is pronounced [(tu.'məm)mi], with gemination of the second /m/ to increase the weight of the second syllable.[17][18][19]

There are a variety of prosodic factors that cause stress to retract (move backward) to a syllable where it would not otherwise be expected, given the usual iambic stress pattern. (These processes do not apply, however, in the Norton Sound dialects.[8]) The processes by which stress retracts under prosodically-conditioned factors are said to feature regression of stress in Miyaoka's (2012) grammar. When regression occurs, the syllable to which stress regresses constitutes a monosyllabic foot.[8]

The first of these processes is related to the inability of /ə/ to occur long. Outside of Norton Sound, if the consonants before and after /ə/ are phonetically dissimilar, /ə/ will elide, and stress will retract to a syllable whose nucleus is the vowel before the elided /ə/. For example, /nəqə-ni/ "his own fish" is not pronounced *[(nə.'qəː)ni], which would be expected by iambic lengthening, but rather is pronounced neq'ni [('nəq)ni], which features the elision of /ə/ and a monosyllabic foot.[8]

Second, if the first syllable of a word is closed (ends in a consonant), this syllable constitutes a monosyllabic foot and receives stress. Iambic footing continues left-to-right from the right edge of that foot. For example, nerciqsugnarquq "(s)he probably will eat" has the stress pattern [('nəχ)(tʃiq.'sux)naχ.qoq], with stress on the first and third syllables.[12][8]

Another third prosodic factor that influences regressive is hiatus: the occurrence of adjacent vowels. Yup'ik disallows hiatus at the boundaries between feet: any two consecutive vowels must be grouped within the same foot. If two vowels are adjacent, and the first of these would be at the right edge of a foot (and thus stressed) given the usual iambic footing, the stress retracts to a preceding syllable. Without regressive accent, Yupiaq /jupiaq/ would be pronounced *[(ju.'piː)aq], but because of the ban on hiatus at foot boundaries, stress retracts to the initial syllable, and consonant gemination occurs to increase the weight of that initial syllable, resulting in [('jup)pi.aq].[8] This process is termed automatic gemination in Jacobson's (1995) grammar.

Yup'ik also disallows iambic feet that consist of a closed syllable followed by an open one, i.e. feet of the form CVC.'CV(ː), where C and V stand for "consonant" and "vowel" respectively. To avoid this type of foot, stress retracts: cangatenrituten /tʃaŋatənʁitutən/ has the stress pattern [(tʃa.'ŋaː)('tən)(ʁi.'tuː)tən] to avoid the iambic foot *(tən.'ʁiː) that would otherwise be expected.[8]

In descriptive work on Yup'ik, there are four regions within nouns and verbs that are commonly identified. The first of these is often called the stem (equivalent to the notion of a root), which carries the core meaning of the word. Following the stem come zero or more postbases,[12] which are derivational modifiers that change the category of the word or augment its meaning. (Yup'ik does not have adjectives; nominal roots and postbases are used instead.) The third section is called an ending, which carries the inflectional categories of case (on nouns), grammatical mood (on verbs), person, and number.[20] Finally, optional enclitics may be added, which usually indicate "the speaker's attitude towards what he is saying such as questioning, hoping, reporting, etc."[20] Orthographically, enclitics are separated from the rest of the word with a hyphen.[12][8] However, since hyphens are already used in glosses to separate morphemes, there is potential for confusion as to whether a morpheme is a suffix or an enclitic, so in glosses the equals sign is used instead.

Because post-bases are derivational morphemes, and thus can change the part of speech of a word, many verbs are built from noun stems, and vice versa. For example, neqe-ngqer-tua "I have fish" is a verb, despite the fact that neqe- "fish" is a noun; the postbase -ngqerr "have" makes the resulting word a verb. These changes in grammatical category can apply iteratively, such that over the course of word formation, a word may become a noun, then a verb, then back to a noun, and so on.[8][21]

The conjugation of Yup'ik verbs involves obligatory marking of grammatical mood and agreement.

Yup'ik has a great number of grammatical moods. The moods can be categorized according to whether the clause in which they are found is independent or subordinate. There are four so-called independent moods: the indicative, optative, interrogative, and participial; these typically are found on the main verbs of independent clauses. Yupʼik also has ten connective moods, which occur on the verbs of adverbial clauses; the connective moods are the Yup'ik equivalent of many subordinating conjunctions of English, and are often translated as 'because', 'although', 'if' and 'while'.[22] The form of a given mood may depend on the transitivity of the verb (e.g., the intransitive form of the participial mood suffix is usually -lriar, but when this mood is suffixed to a transitive verb, its form is -ke instead),[8] on the person of the grammatical subject (e.g., the optative mood is marked with -li only if the subject is third person), or on the phonological or morphological environment.

In addition to the connective moods listed above, there are five so-called "quasi-connective" moods. Though these are adverbial adjuncts to main clauses and thus are similar in function to the connective moods, they inflect like nominals (they inflect with case, not agreement).[8]

Yup'ik has a rich system of agreement on verbs. Up to two nominal arguments may be cross-referenced (intransitive verbs agree with their sole argument, and transitive verbs agree with both arguments). Three numbers (singular, dual, and plural) are distinguished, as well as at least three persons (first, second, and third). The third person is unmarked when cross-referencing subjects,[8] and the verbs of dependent clauses may have two types of third person forms depending on whether some argument is co-refers with the subject of the verb in the independent clause (see "Co-reference across clauses" below).[12][8] To the extent that subject and object agreement markers are not fusional, subject agreement linearly precedes object agreement.

Depending on the grammatical mood of the verb and which grammatical persons are being cross-referenced, agreement may display either an ergative pattern (where the sole argument of an intransitive verb is cross-referenced with the same morpheme that it would be if it were the object of a transitive verb) or an accusative pattern (where the sole argument of an intransitive verb is cross-referenced with the same morpheme that it would be if it were the subject of a transitive verb).[8]

Agreement markers vary in form depending on the grammatical mood of the verb. The two examples below illustrate this. In (1), the 1SG>3SG agreement marker is -qa because the verb is in the indicative mood, while in (2) the agreement marker is -ku due to verb being in the optative mood.

The participial and indicative share a set of agreement markers, and all the connective moods likewise share a common set (which is shared also with some possessed nouns).[8]

The form of 3rd-person agreement in dependent clauses may vary depending on whether that 3rd-person argument is the same referent as, or a different referent than, a 3rd-person subject of the independent clause. In some descriptive work on the language, when the subject of the independent clause is co-referential with the relevant argument in the dependent clause, the agreement in the dependent clause is said to reflect a "fourth"[12] or a "reflexive third"[8] person. Jacobson (1995) uses the following contrast to illustrate:

The intransitive agreement in the dependent clause ermig-pailg-an in (3) is -an, indicating that the argument of the dependent clause is a different referent than the subject of the independent clause nerellruuq, while in (4) the agreement -mi indicates that the arguments of each clause are co-referential. Some grammatical moods do not have associated agreement markers that contrast these two types of third person.[8]

Some researchers have argued that the contrast in (3-4) exemplifies a type of switch-reference,[23][24] though McKenzie (2015) claims Yup'ik does not have the true characteristics of switch-reference, and that the Yup'ik system is better understood in terms of obviation or long-distance anaphora.[25]

Yup'ik nouns inflect for number, case, and show agreement with the person and number of a possessor if present.[8]

The morphosyntactic alignment of Yupʼik is ergative-absolutive,[8] meaning that subjects of intransitive verbs bear the same grammatical case (the absolutive) as the objects of transitive verbs, while the subjects of transitive verbs have a different case (the ergative). For example, the sentence Angyaq tak'uq ("The boat is long") features an intransitive verb, and the subject (angyaq, "the boat") is in the absolutive case. By comparison, in the sentence Angyaq kiputaa ("He buys the boat"), the verb is transitive, and it is now the object (angyaq, "the boat") that bears the absolutive.[26] This contrasts with nominative-accusative languages like English, where the subjects of intransitives and transitives are identical in form ("He slept", "He ate the bread"), while the objects of transitives have a different case ("The moose saw him").

In addition to the absolutive and ergative structural cases (the latter of which is syncretic with the genitive; collectively the ergative and genitive are usually called the relative case[12][8]), there are at least five other cases that are mostly-nonstructural: ablative-modalis (a historical syncretism of ablative and instrumental cases), allative, locative, perlative, and equalis.[8]

The forms of these grammatical cases are variable, depending on the grammatical person and number of the head noun as well as the person and number of its possessor (if there is one).

Possessed nouns, like all other nouns, inflect for number and case, but also show person and number agreement with their possessor. For example, consider a few forms of saskaq "cup". The two leftmost nouns below are unpossessed, but the third is marked for a first person singular possessor -ka (pronounced in this case as -qa after assimilating to a uvular place of articulation). The final example marks plural number for both the noun itself and its possessor.

Possessors are often optional, but when present are marked with relative case:[8]

Yup'ik has considerably more freedom of word order than English does. In English, the word order of subjects and objects with respect to a verb reflects the thematic roles of the subject and object. For example, the English sentence The dog bit the preacher means something different than The preacher bit the dog does; this is because in English, the noun that comes before the verb must be the agent (the biter), while the noun following the verb must be the theme (the individual or thing that is bitten).

In Yupʼik, word order is freer because the rich inflectional system often serves to unambiguously identify thematic relations without recourse to word order. The Yup'ik sentences Qimugtem keggellrua agayulirta (dog.ERG bit preacher.ABS) and Agayulirta keggellrua qimugtem (preacher.ABS bit dog.ERG) both mean "the dog bit the preacher", for instance: the word order varies between these sentences, but the fact that qimugtem ("dog") is marked with ergative case (-m) is sufficient to identify it as the thematic agent. Thus, to say "the preacher bit the dog" in Yup'ik, one would need change which noun gets ergative case and which gets absolutive: qimugta keggellrua agayulirtem (dog.ABS bit preacher.ERG).[27]

Despite the greater freedom of word order, there seems to be a general preference for Subject-Object-Verb (SOV) order (though VSO is also common, and pragmatic factors also play a significant role).[8] This can be observed in circumstances where the inflectional system will not unambiguously determine which noun is the agent and which is the theme. This obtains, for instance, when both arguments of an indicative transitive verb are third person plural and unpossessed: elitnauristet mikelnguut assikait can, in principle, mean either "the teachers like the children" or "the children like the teachers", since the case marking on elitnauristet "teachers" and mikelnguut "children" does not distinguish ergative from absolutive case (-t marks unpossessed ergative plurals as well as unpossessed absolutive plurals). In cases like this, the SOV preference comes into play, and the sentence is most readily interpreted as "the teachers like the children".

Yup'ik has a rich system of spatial deixis. That is, many of the spatial properties of things and events are linguistically encoded in great detail; this holds true for demonstrative pronouns (like English "this one", "that one") as well as spatial adverbs ("here", "there").

There are twelve categories that define the orientation of a thing or event with respect to the environment. The environment in this sense includes topographical features (e.g., there is a contrast between upriver and downriver), the participants in the speech event (e.g., there is a contrast between proximity to the speaker and proximity to the hearer), and the linguistic context (one of these twelve categories is used for anaphora). This twelve-way contrast is cross-cut by a trinomial contrast in horizontal extension/motion: this determines whether the referent is extended (horizontally long or moving) or non-extended, and if non-extended, whether distal (typically far away, indistinct, and invisible) or proximal (typically nearby, distinct, and visible).

To illustrate, the spatial demonstrative roots of Yup'ik (which are then inflected for case and number) are presented in the following table from Miyaoka (2012).

Note that Classes I and II lack distal forms due to an inherently non-distal meaning (these forms only locate things that are near to the speaker/hearer). Class III is purely anaphoric, and thus only has a distal form.[8]