Алломорфия в суффиксации якутского языка Текст научной статьи по специальности «Языкознание и литературоведение»

UDC 81

Duck-Soo Kang, K. Egorova, Sang-Cheol Ahn

Allomorphy in Sakha Case Suffixation

The Sakha(Yakut) language has a variety of case suffixes, including nominative, comitative, accusative, partitive, dative, ablative, instrumental, and comparative. Moreover, depending on the environments, there are numerous allomorphic distributions in case suffixation processes. In this paper, we attempt to provide a uniform representation for each case suffixation and explain why such allomorphic distributions occur. To this end, we employ the basic concepts of feature underspecification for the alternating vowels and consonants which surface as full segments depending on the given contexts, i. e., vowel harmony or the featural properties of the preceding root-final segments. Moreover, employing the framework of Optimality Theory, we show how those consonantal as well as vocalic alternations can be accounted for by several simple constraints, such as *VV, Vowel Harmony, *OnsetCC, Share, etc. In dative suffixation, furthermore, we observe that stem-final vowels may affect the segmental specification of the suffix-initial consonants. For this vowel-consonant interaction, we invoke the vocalic feature [open] which spreads to the suffix-initial consonant, raising the consonantal aperture. The paper is an explanation for why choose one or the other allomorphs. The work will serve as a reference material for researchers, will help focus on the allomorphs Yakut language. This optimally-theoretical approach could be applied to other studies in the field of phonology and morphology of the Yakut.

Keywords: Yakut, case, case suffixation, underspecification, Optimality Theory, vowel harmony, consonant alternation, open vowel, consonantal variation, syllable structure.

Дуксу Канг, К. Г. Егорова, Сан Чоль Ан

АЛЛОМОРФИЯ В СУФФИКСАЦИИ ЯКУТСКОГО ЯЗЫКА

Якутский язык имеет множество суффиксов таких падежей, как именительный, комитативный (совместный), винительный, партитивный, дательный, аблатив (исходный), инструментальный, сравнительный. Кроме того, в зависимости от окружения имеются многочисленные алломорфные распределения в процессе суффиксации. В данной статье мы попытаемся равномерно объяснить, каким образом в каждом конкретном случае суффиксации происходят алломорфные дистрибуции. Для этого мы используем основные понятия неполной идентификации для чередующихся гласных и согласных, в которых полноправные сегменты зависят от заданных контекстов, т. е. гармонии гласных или свойственной характеристики предыдущих корневых финальных сегментов. Кроме того, используя рамки теории оптимальности, мы покажем, как согласные и чередования гласных можно объяснить с помощью нескольких простых ограничений, таких как *VV, *VH (гармонии гласных), *OnsetCC (стечение согласных) и т. д. В суффиксации дательного падежа заметим, что гласные

КАНГ Дуксу - профессор Хангукского университета иностранных языков, Республика Корея.

E-mail: kangds@hufs.ac.kr

KANG Duck-Soo - Professor of Hankuk University of Foreign Langauges, Korea.

ЕГОРОВА Кюннэй Григорьевна - ст. преподаватель кафедры восточных языков и страноведения ИЗФИР СВФУ им. М. К. Аммосова.

E-mail: kiounnei1@yandex.ru

EGOROVA Kyunney Grigor'evna - Senior Lecturer of the Department of Oriental Languages of the Institute of Modern Languages and International Studies, M. K. Ammosov North-Eastern Federal University.

АН Сан Чоль - проф. кафедры восточных языков и страноведения ИЗФИР СВФУ им. М. К. Аммосова.

E-mail: scahn.mail@gmail.com

AHN Sang-Cheol - Professor of the Department of Oriental Languages of the Institute of Modern Languages and International Studies, M. K. Ammosov North-Eastern Federal University.

основы слова могут повлиять на сегментную спецификацию начальных согласных суффикса. Мы обращаемся к свойствам гласных [открытости], которые распространяются на суффикс начального согласного. В работе представлено объяснение тому, почему выбирается тот или иной алломорф. Работа послужит опорным материалом для исследователей, поможет ориентироваться в алломорфах якутского языка. Данный оптимально-теоретический подход может быть применен для других исследований в области якутской фонологии и морфологии.

Ключевые слова: якутский язык, падежи, суффиксация, неполная идентификация, теория оптимальности, гармония гласных, чередование согласных, открытые гласные, разновидность согласных, структура слога.

Introduction

Sakha (Yakut) is a northern Turkic language which is spoken by 400.000 people mainly in the Sakha Republic, Russian Federation. Like all other Turkic languages, Sakha is an agglutinative language and employs vowel harmony. Shown below are the phonemic inventories of consonants and vowels in this language (Djachkovskij 2000: 62-63) (We can consider /x/ and /у/ as velar fricatives. Based on Djachkovskij (2000: 63), however, <x> is categorized as the uvular stop /q/ (Kang 2011) in this paper. Also, /r/ is described as a trill in Djachkovskij (2000: 63)).

According to the literature (Korkina et al. 1982: 130, Sleptsov1990), Sakha language has eight cases for noun stems: basic (nominative), accusative, dative, ablative, instrumental, comitative,

Table 1

Sakha consonantal inventory

Bilabial Dental Alveolar Palatal Velar Uvular Glottal

Plosive voiceless p t с k q

voiced b d 3 g

Fricative voiceless s

voiced у h*

Nasal m n V Ч

Approximant plain l j

nasalized J

Trill r

Note: *A fricative [h] has been regarded as an positional allophone of the consonant /s/, which occurs intervocalicaUy (as in a derived verb ahaa- 'to eat' from a noun as 'food'). According to Korkina et al. (1982: 59-60), however, it acquired a phonemic status which can be observed in such words as hajdaa- 'to drive cow' and ahayas 'open'.

Table 2

Sakha vowel phonemes (Kruger 1962, Djachkovskij 2000: 21, Sasa 2007: 158-159)

Front Back

unrounded rounded unrounded rounded

Close short i ü i u

long i: ü: i: u:

Open short e ö a o

long e: ö: a: o:

Diphthong ie üö ia uo

Note: The phonetic symbols [u], [o] and [i] indicate the IPA [y], [0], and [ra] respectively. - 81

partitive, and comparative. Moreover, the case forms are further divided into dual types: simple and possessive types. Simple types have a nominal stem with case suffixes, while possessive types take possessive suffixes before the case suffixes.

(1) Simple vs. possessive

Comitative

Accusative

Dative

Ablative

Instrumental

Partitive

Comparative

aya-liin 'with father' aya-ni 'father' aya-ya 'to father' aya-ttan 'from father' aya-nan 'by father' uu-ta 'some water' aya-taayar 'than father'

aya-bi-niin 'with my father' aya-ti-n 'my father' aya-ti-gar 'to my father' aya-bi-ttan 'from my father' aya-bi-nan 'by my father' uu-ttan 'some of my water' aya-bi-naayar 'than my father'

Having both simple and possessive types, the Sakha noun case system is quite complicated. In fact, even the simple type of suffixation is quite complicated as shown in the academic grammar of the Sakha language (Korkina et al. 1982: 129-147). For example, the comitative case of the simple type has 8 suffix variants, of which the representative form is considered to be /-liin/. Most earlier studies on this issue (Korkina et al. 1982, Sleptsov 1990), however, have just listed all the variants, without explaining when and how those variants occur. This paper, therefore, will show the underlying representation of each noun suffix and explain how the allomorphic variants occur. To this end, we employ the concept of feature underspecification for the alternating vowels and consonants which surface as full segments depending on the context, i. e., vowel harmony or the featural properties of the preceding root-final segments. We also employ the basic framework of Optimality Theory (McCarthy & Prince 1995, McCarthy 2008) as our tool of analysis.

Optimality Theory (OT henceforth, McCarthy and Prince 1993, McCarthy 2008) is a model of constraints and constraint interactions, whereas the standard generative theory is a model of rules and derivations. There are two major components in OT: GEN(erator) which maps the input onto an infinite number of output candidates, and EvAL(uator) which evaluates the output candidates by a set of ranked constraints and selects the optimal output among the possible candidates. Therefore, Gen produces a set of logically possible candidates of the input and Eval selects the optimal analysis of the input.

(2) The input-output mechanism in OT Gen: Input — {Candidate 1, Candidate 2, Candidate 3, .. .Candidate n} Eval: {Candidate 1, Candidate 2, Candidate 3, ... Candidate n} — (Optimal) Output

The main analytical proposal of OT is that constraints are ranked in a hierarchy of relevance. Therefore, the selected output is the most optimal candidate without fatal violation, i. e., it violates the lower-ranked constraints minimally, whereas suboptimal candidates violate the top or upper-ranked constraints. For example, the following tableau shows the selection of the optimal candidate, based on the dominance relationship among the four constraints. (Constraint violations are represented with * symbols (fatal violation with! symbol) and the optimal candidate is marked by the index The solid line is used for the dominance relation between constraints, while the dotted line for non-dominance.) The tableau shows that Candidate 3 is selected as optimal since its violation is minimal, i.e., violation of the two lower-ranked constraints.

(3)

Input Constraint 1 Constraint 2 Constraint 3 Constraint 4

a. Candidate 1 *!

b. Candidate 2 *!

^c. Candidate 3 * *

Constraint ranking: Constraint 1, Constraint 2 >> Constraint 3 >> Constraint 4

There are two types of constraints in Optimality Theory, i. e., the failthfulness and markedness constraints, which are intrinsically conflicting. The following list shows the basic faithfulness constraints.

(4) Max Input segments have output correspondents. (No deletion). Dep Output segments have input correspondents. (No epenthesis). lDENT[aF] Correspondent segments in input and output have identical values for [aF] .

Max prevents deletion, while Dep prohibits epenthesis. The constraint Ident[oF] requires that corresponding segments in the input (i. e., underlying representation) and the output (i. e., surface representation) must have the same feature representation. As the faithfulness constraints interact with the markedness constraints, we can employ the following markedness constraints in our analysis.

(5) *VV (No Hiatus): A two consecutive vowel sequence may not be permitted. VH (Vowel Harmony): A suffix vowel and the preceding root vowel should agree in [back] and [round] properties.

The markedness constraint *VV prohibiting a consecutive vowel sequence is a very common constraint found in many languages. The vowel harmony (VH) is also very common, especially in Turkic languages in which the backness and roundness of the suffix vowel is determined by the preceding root vowel, due to the vowel harmony.

Noun case suffixes

Nominative

The most basic form is the nominative which lacks any segmental property. Thus, a noun with a nominative case is used without any additional case form.

(6) Kun kieheren barda. 'Day began to darken.'

Bu kinige miene. 'This book (is) mine.'

(7) Ubaj-im uccutal. 'My brother (is) a teacher.'

In the example (6), kun 'day' and kinige 'book' are subjects taking a zero suffix. On the other hand, ubaj-im 'my brother' in (7) shows a nominative case with possessive suffix {-im}, indicating that the first person singular, 'my', does not carry a case segment.

Accusative

There are two types of allomorphs in the accusative suffixation: a nasal + vowel or a single vowel in which the quality of the vowel is determined by the vowel harmony with the root vowel.

taba-ni 'reindeer'

at-i 'horse'

aan-i 'door'

ehee-ni 'grandpa'

et-i 'meat'

erej-i 'pain'

oyo-nu 'child'

ot-u 'grass'

boro-nu 'wolf'

boy-u 'trash (box)'

In the data shown above, we observe the alternation pattern which disallows a nasal consonant, if a root ends with a consonant. Therefore, we get -V after a consonant, while -nV after a vowel. This indicates that the suffix-initial nasal shows up breaking up a possible V+V sequence, while it drops out if it is preceded by a root-final consonant. Moreover, the backness and roundness of the suffix vowel is determined by the preceding root vowel, due to the vowel harmony.

(9) -ni/ -ni/ -nu/ -nu — suffixed to a vowel-final root -i/ -i / -u / -u — suffixed to a consonant-final root

Considering the allomorphic distribution, we postulate the input form as /-nI/ in which the high vowel /I/ is underspecified for [round] and [back]. That is, depending on the backness and roundness of the root vowel, the underspecified suffix vowel can be realized as one of the four high (i. e., closed) vowels, [i, i, u, u]. Moreover, the surface realization is determined by the syllable

structure of the preceding root. That is, if the root has a coda consonant, the suffix-initial nasal is not required. If the root ends with a vowel, however, the nasal surfaces to break up the vowel hiatus.

(10) V (C) + nj (I = High (i. e., closed) vowel) [bk'rnd]

As the alternation of the suffix segments can be determined by several constraints, we employ the two markedness constraints, *VV (No vowel hiatus) and VH (vowel harmony). The following tableau shows how these markedness constraints interact with the faithfulness constraint.

ehe: -nl *VV VH Max

a. ehe : -i *! *

b. ehe -nu *!

c. ehe -ni *!

^d. ehe:-ni

Here the two constraints *VV and VH are not violable, whereas the lower-ranked Max and Ident[back] can be violated (Max is the lowest in ranking as it is the only violable constraint in this tableau). We also note that all the candidates including the optimal one violates the Ident[back] since they need the vowel specification. If the root ends with a consonant, however, we need to invoke a new constraint *CN which disallows a suffix-initial nasal after an obstruent. The following tableau shows that this constraint is also inviolable.

(12) *CN: A suffix-initial nasal consonant is not allowed after a root-final obstruent.

ot-nl *CN *VV VH Max

a. ot-ni *! *

b. ot-nu *!

c. ot-i *! *

^d. ot-u *

As shown above, the constraint *CN, along with *VV and VH, plays a crucial role in the selection of the optimal output of accusative suffixation.

Instrumental

The crucial role of *CN can be found not only in the accusative suffixation, but also in the instrumental suffixation. The following list shows the allomorphic variants of the instrumental suffix.

(14) saa-nan 'with gun'

at-inan 'with a horse'

suge-nen 'with an ax' et-inen 'with meat'

solko-non 'with silk' ot-unan 'with grass'

kohoruu-nen 'with carriage cattle' boy-unen 'with trash(box) '

The pattern of allomorphic distribution can be summed up as follows.

(15) -nan/-non/-nen/-non — suffixed to a vowel final root

-inan/-inon/-inen/-inon/-unan/-unen — suffixed to a consonant final root

The list in (15) shows that the appearance of the [+high] suffix-initial vowel is determined by the shape of the stem-final segment, as the avoidance of a vowel hiatus is apparently the major factor.

Below we can show a constraint-based account of the instrumental suffixation in a similar fashion as the constraint *CN plays a crucial role along with the two other constraints *VV and VH.

(16) ot-unan

'with grass'

ot-InAn *CN *VV VH Max

a. ot-inan *! (i)

b. ot-nan *! *

c. ot-unen *! (e)

^d. ot-unan

(17) solko-non

'with silk'

solko-InAn *CN *VV VH Max

a. solko-inan *! * (i)

b. solko-unon *!

c. solko-nen *! (e) *

^d. solko-non *

We observe how the underspecified vowels are fully specified by the constraint interaction. (The underspecified vowel I has only the [+high] specification, while the other underspeicified vowel A has [-high] having the a~o~e~o alternations). We can also see how the suffix-initial vowel surfaces or disappears, depending on the occurring environments. In addition, just as in accusative suffixation, *CN, VV and VH play crucial roles.

Ablative

Unlike the earlier two suffixation patterns, the ablative suffixation shows the t~tt alternation in addition to the vowel alternation affected by vowel harmony.

(18) taba-ttan 'from reindeer'

mas-tan

ehee-tten

qaar-tan

oyo-tton

qaar-tan

boy-ton

'from a tree' 'from grandpa' 'from snow' 'from a child' ' because of snow' 'because of trash'

(19) -ttan/-tton/-tten/-tton ^ after a vowel final root -tan/-ton/-ten/-ton ^ after a consonant final root

As the suffix allomorphy shows the t ~ tt alternation, we postulate the input form as shown below, in which the quality of the suffix vowel is determined by the vowel harmony, i. e., [back, round] property of the root vowel.

(20) V (C) + t t A n (A = Nonhigh (i. e., open) vowel)

[bk,md]

According to Baertsch & Davis (2003), Yakut does not allow onset clusters. If the stem ends with a consonant, therefore, the input /tt/ may not be allowed as an onset cluster on the surface. That is, so only one -t shows up if the root ends with a consonant due to the following markedness constraint. (We omit the inactive constraints *CN and *VV for a simpler description).

(21) *CC Consonant clusters may not be allowed

(22) mas-tan

'from a tree'

mas-ttAn *CC VH Max

a. mas-ttan *!

b. mas-tten *! *

c. mas-ten *! *

^d. mas-tan *

If the root ends with a vowel, however, the input /tt/ is split as a coda and an onset on the surface. (The period indicates a syllable boundary).

(23) ehe: -tten 'from grandpa'

Ehe: -ttAn *CC VH Max

a. ehe -t. tan *!

b. ehe -tan *! *

c. ehe -ten *

^d. ehe:-t.ten

Comitative

While the accusative and instrumental suffixes have the full segmental specification for their initial consonant, i. e., /n/ and /tt/, the comitative suffix allomorphy shows that the feature specification of the suffix-initial consonant is determined by that of the preceding root-final consonant. That is, showing the t~d~n~l alternation, the suffix-initial consonant is realized as [t] after a voiceless consonant, [d] after a non-lateral voiced consonant, [n] after a nasal, and [l] elsewhere, i. e., after a vowel (According to Korkina et al. (1982: 131), the comitative case has a suffix {-liin} and its variants).

(24) taba-liin 'with reindeer' sahil-liin 'with fox' kiil-liin 'with animal' kiis-tiin 'with girl' at-tiin 'with horse' qaar-diin 'with snow' tellej-diin 'with mushroom' sirej-diin 'with face' ij-diin 'with moon' aan-niin 'with door' kihin-niin 'with winter' tiig-niin 'with squirrel'

(25) -tiin/-tiin/-tuun/-tüün — after a voiceless consonant

-diin/-diin/-duun/-düün — after a voiced consonant

-niin/-niin/-nuun/-nüün — after a nasal

-liin/-liin/-luun/-lüün — after a vowel or /l/

While Korkina et al. (1982: 131-132) proposed the representative form of the comitative suffix as /-liin/, we formulate the input form of the comitative suffix as -llln, representing the underspec-ified high vowel as /I/ (Suffix vowels vary in accordance with their stem vowels, so {-liin} cannot represent the whole set of suffix variants of the comitative case. As vowel qualities are determined by vowel harmony, we presume the underspecified vowel /I/). The variants [l, t, d, n] share coronal "noncontinuancy", but [l] differs from the others in laterality, voicelessness and nasality. (Also, the capital /I/ indicates a high vowel for which other features [back, round] are determined by vowel

"armo"y). (26) V C - l I I n

V

[+voice] — d [-voice] — t [+nasal] — n

Based on the description shown above, we postulate three "Share" constraints: Share[voice], Share[lateral], and Share[nasal]. Note that the data in (25) indicate that the root-final /r, j/ share [-lateral, +voice] with the surfacing suffix-initial -d.

(27) Sh [ ' ] The [±voice] property of the root-final consonant ( ) [ ] should be shared with the suffix-initial consonant.

Sh [1 t 1] The [±lateral] property of the root-final consonant L J should be shared with the suffix-initial consonant.

„, r ,, The [±nasal] property of the root-final consonant [ ] should be shared with the suffix-initial consonant.

The following tableau shows how we get the optimal output with the relevant constraints. (We omit the inactive *OnsetCC and Max for convenience).

(28) kiis-tiin 'with girl'

kiis-lIIn Share[voice] VH Ident[lateral]

a. kiis-liin *! **

b. kiis-tiin **! *

c. kiis-tuun **! *

^d. kiis-tiin *

qaar-diin_'with snow'

qaar-lIIn Share [lateral] Share [voice] Share [nas] VH Ident [lateral]

a. qaar-liin *! **

b. qaar-tiin *! *

c. qaar-niin *! *

^d. qaar diin *

These tableaux show that the Share constraints are ranked higher than the Ident constraints in the selection of the optimal outputs. Partitive

The allormorphic distribution in partitive suffixation is equivalent to the comitative suffixation as shown in the following examples.

taba-ta 'a few reindeer'

sittik-ta 'some pillow'

mejii-te 'a little brain'

et-te 'some meat'

mas-ta 'some tree'

uuneeji-te 'a little plants'

balik-t 'some fish'

qaar-da 'a little snow'

tellej-de 'some mushroom'

sir-de 'some land'

qajihar-da 'a little ski'

suol-la 'some trace'

aan-na 'some door'

salgin-na 'some air'

kepseen-ne 'some story'

However, the partitive allomorphic distribution is different in that the input suffix-initial consonant in the comitative suffixation is specified as -l due to its occurrence after a vowel. In partitive suffixation, however, -l occurs only after a root-final /l/, while -t occurs in the intervocalic position.

(30) -ta/-to/-te/-to ^ after a vowel or a voiceless consonant -da/-do/-de/-do ^ after a voiced segments /r, j/ -la/-lo/-le/-lo ^ after the lateral /l/ -na/-no/-ne/-no ^ after a nasal

Considering the distribution pattern, we postulate -tA as the input form of the partitive suffix, so that /t/ is realized as -d, -n, and -l by spreading the [voice], [nasal], or [lateral] feature of the preceding root-final consonant, while -t occurs elsewhere, i. e., intervocalically. Therefore, the initial suffix-initial consonants of the partitive and comitative suffixes are realized in a different way in the intervocalic environment, whereas both of them show up as -n after a nasal and -d after a voiced consonant.

(31) V C - t A

C-'

[voice] [nasal] [lateral]

As for the difference from the comitative suffixation, the following tableau shows that the Share constraints, i. e., Share[voice], Share[nasal], and Share[lateral], play crucial roles. (We omit VH and Ident[back] for convenience).

(32) qaar-da 'a little snow'

qaar-tA Share [voice] Share [nas] Share [lat] Ident [voice] Ident [nas] Ident [lat]

a. qaar-ta *!

b. qaar-na *! *

c. qaar-la *! *

^d. qaar-da *

Comparative

The Share constraints shown above play crucial roles again in the analysis of the comparative suffixation. The comparative suffixation allomorphy can be observed in the following data.

(33)

taba-taayar

kiil- laayar

qaar-daayar

aan-naayar

oq-tooyor

tuu-teeyer

salgin-naayar

mas-taayar

qallaan-naayar

sap-taa jar

'than reindeer' 'than animal' 'than snow' 'than door' 'than a bow' 'than fluff' 'than air'

'than wood'/'than tree' 'than the sky' 'than yarn'

As shown above, the suffix of the comparative begins with one of the dental consonants /t, d, l, n/. Here we observe that -t occurs not only after a voiceless consonant but also after a vowel. We therefore postulate the input form of the comparative suffix as -tAAyar. The initial consonant /t/ alternates in accordance with the final element of a stem. In the tableau shown below, several Share constraints play the major roles again, i. e., Share[voice], Share[nasal], and Share[lateral]. (We omit VH and Ident[back] for convenience).

(33) qaar-daayar 'a little snow'

qaar-tAAyAr Share Share Share Ident Ident Ident

[voice] [nas] [lat] [voice] [nas] [lat]

a. qaar-taayar *!

b. qaar-naayar *! *

c. qaar-laayar *! *

^d. qaar-daayar *

Vowel effects on consonant realization: dative suffixation

The segmental underspecification is not limited to those cases shown so far, as we observe the consonantal variation in dative suffixation. As shown in the following list, there are a number of allomorphic consonantal as well as vocalic variants in dative suffixation.

(34) a. /g/ after /l, r/ and a high vowel /i, u, i, u/: -ga/-go/-ge/-go oyolor-go 'to children'

uol-ga 'to boy'

atastahii-ga 'to/for exchange'

silgi-ga 'to horse'

qonuu-ga_ 'to field'

oopuu-ga 'for play'

bilii-ge 'to knowledge'

uoruu-ge 'to joy/ joyfulness'

sir-ge 'to earth'

b. /к/ after an obstruent: -ka/-ko/-ke/-ko

kiis-ka 'to girl'

muus-ka 'to ice'

qonuk-ka 'to days'

uok-ka 'to fire'

kolqos-ka 'to kolkhoz'

c. /у/ after /a, e, o, o/: -ya/-yo/-ye/-yo

taba-ya 'to reindeer'

oskuola-ya 'to school'

ule-ye 'to new job'

oyo- yo 'to child'

d. /q/ after /q/ only: -qa/-qo/-qe/-qo

inaq-qa 'to cow'

uquoq-qa 'to bone'

sorudaq-qa 'to task'

tiqiraq-qa 'to nails'

oloq-qo 'to life'

e. /q/ after a nasal: ■ - qa/- qo/- qe/- qo

aay-ya* 'to door'

kiim-qa 'to spark'

ayilaahiq-qa 'to breathlessness'

sariq-qa 'to shoulder'

Note: *the stem-final nasal consonant /n/ (i. e., /aan/) is velarized by being assimilated to the suffix-initial velar nasal.

As the -gV variants in (34a) show the widest distribution, we can describe the dative suffixation in the following manner. First, the -kV realization in (34b) is a consequence of the [-voice] spreading (In kolqos-ka'to kolkhoz', we might say that the syllable-final voiced obstruent gets devoiced before suffixation, so [s] in kolqos takes -ka in kolqos-ka 'to kolkhoz'. The final consonant /s/ in Kolqos is merely an orthographic representation as the word was adapted as kolqos in Sakha). Then, the examples in (34c, d) can be described in a similar way, considering that the uvular /q/ is [-high], while all the other consonant variants /g, k, y, q/ are [+high] velars.

(35) a. VH

A--.

V C - g V ^ kV

[-voice]

b. VH

V C - g V ^ qV

i-'

[-voice]

[-high]

c. VH

V C - g V ^ gV ^ qV

r K

[+nasal]

The example in (35c), however, is much more complicated than other cases since -yV suffixation

occurs only if the root ends in an open vowel /a, e, o, 0/. That is, the consonantal variation is affected by a preceding vowel, rather than a consonant. Unlike the common consonantal assimilation, it is quite unusual to observe a case where a vowel affects a consonantal change. In this case, therefore, we conjecture that the "openness" of the preceding vowels impose more sonority to the suffix consonant, which ends up with [y],

(36) Vocalic effect on consonantal variation in yV suffixation

VH

r^. x

V - g V ^ yV

rV n

[+open]

The vocalic feature [open] (i. e., based on the categorization in Table 2) is used to categorize the non-high vowels which are distinct from closed high vowels /i, i, u, u/. So, we interpret that the openness of the stem-final vowel spreads to the suffix-initial consonant to increase the consonantal aperture. As a consequence, the stop /g/ becomes the more sonorous fricative [y].

Based on these descriptions, we can now show how the optimal outputs are selected in the following tableaux. (We omit the inactive constraints *VV, VH, *CN, and Max for convenience).

(37) oloq-qo_'to life'_

oloq-gA Share[voice, high] Ident[voice] Ident-C[hi]

a. oloq-go **!

b. oloq-ko *! *

c. oloq-yo **!

^d. oloq-qo * *

(38) sarig-ga 'to shoulder'

sarig-gA Share[nas] Share[voice, high] Ident[nas]

a. sarig-ga *!

b. sarig-ka *! *

c. sarig-ya *!

^d. sarig-ga *

(39) taba-ya 'to reindeer'

taba-gA Share-VC[open] Share[voice, high] Ident[open]

a. taba-ga *!

b. taba-ka *!

c. taba-ga *!

^"d. taba-ya *

In these analyses, it is shown that the suffix allomorphy is determined not only by the consonantal characteristics of the root final consonant but also by the [open] feature of the root final vowel as shown in (36) and (39).

Concluding remarks

So far we have discussed how the various patterns of Sakha case allomorphy can be accounted for. To this end, we employed the basic framework of Optimality Theory as our tool of analysis. During the discussion, we have postulated several crucial constraints responsible for the selection of the optimal outputs, such as *VV, VH, *OnsetCC, etc. These markedness constraints are higher ranked than the faithfulness constrains like Ident and Max. Here we also showed the important role of *CN which takes a crucial role in accusative, instrumental, and ablative suffixations.

As for the input specification, we have employed featural underspecification for segmental variations, so that the missing or underspeicified features can be filled in by vowel harmony or consonantal assimilation. As for the vocalic underspecification, we need to consider the dominant backness and rounding harmony in the Sakha language. The segemental specification can be determined mostly by the consonantal assimilation, i. e., spreading certain features of the stem-final consonant to the following underspecified suffix-initial consonant. Therefore, we invoked a

series of several Share constraints which are also inviolable just like the constraints for the vocalic alternations, i. e., *VV and VH.

In the discussion of the dative suffixation, we have observed a rare case in which stem-final vowels may affect the segmental specification of the suffix-initial consonants. This sort of vowel effect on consonantal specification is uncommon, so we invoked the vocalic feature [open] which spreads to the suffix-initial consonant, raising the consonantal aperture.

R e f e r e n c e s

1. Afanasjev P. S., Xaritonov L. N. 1968. Russko-Jakutskij slovarj. Moskva: Sovetskaja Enciklopedija.

2. Baertsch, Karen, Stuart Davis. 2003. The split margin approach to syllable structure. ZAS Papers in Linguistics 32.2003. Pp. 1-14.

3. Djachkovskij N. D. 2000. Sakha Bilingngi Tyla: Fonetika. Yakutsk: YAGU.

4. Eginova S. D. 2007. Sakha Tyla: tyly yrytyy khalyyba. Yakutsk: Yakut State University.

5. Kang Duck-Soo. 2011. Yakut Eo (Yakut Language). Seoul: Hankuk University of Foreign Studies Press.

6. Kolodeznikov S. K., Kolodeznikova V. S. 2010. Sakha Tyla. Samouchitel' jakutskogo jazyka. Yakutsk: Yakutia.

7. Korkina E. I. et al. 1982. Grammatika Sovremennogo Jakutskogo Literatunogo Jazyka. Moskva: Nauka.

8. Kruger John. 1962. Yakut Manual (Uralic and Altaic Series). Bloomington, IN: Indiana University Publication.

9. McCarthy, John and Alan Prince 1995. Faithfulness and reduplicative identity. University of Massachusetts Occasional Papers in Linguistics 18: Papers in Optimality Theory. Pp. 249-384.

10. McCarthy, John. 2008. Doing Optimality Theory. Malden, MA: Blackwell.

11. Neustroev N. N., Vasiliev E. K., Semenova S. S. 2008. Sakha Tyla: Morfologija. 6th grade textbook. Yakutsk: Bichik.

12. Novrogodov Innokentij N. 2010. Yakut-Evenki linguistic mutual-relations-Languages of people of Russian Federation (Yakut language). A dissertation summary of Doctor of Philology. Kazan, 2009. (https:// dvs.rsl.ru/SWU/Vrr/SelectedDocs?docid=%2Frsl01003000000% 2Frsl01003489000%2Frsl01003489787%2 Frsl01003489787.pdf).

13. Sasa Tomasa. 2009. Treatment of Vowel Harmony in Optimality Theory. PhD dissertation, University of Iowa.

14. Sleptsov P. A. 1972. Jakutsko-Russkij Slovarj. Moskva: Sovetskja Enciklopedija, 1972.

15. Sleptsov P. A. 1990. Izuchajushchim Jakutskij Jazyk (Let's learn Yakut language). Yakutsk: Yakutskoe knizhnoe izdatel'stvo.