1.1 Geography

The Tokunoshima dialect (hereafter, Tokunoshima or TKN) is spoken in Tokunoshima Island. The three maps on figure 2.1 show the Japanese Archipelago, the islands of the Northern Ryukyus, and Tokunoshima Island, in order from left to right.

Tokunoshima is divided into three towns: Amagi Town in the northwest, Tokunoshima Town in the east, and Isen Town in the south. Out of dozens of villages (TKN: sima, or sjuuraku) on the island, the present chapter focuses on the dialect of three villages in the southern part of Isen Town unless otherwise stated: Higashihama (also known as Higashiomonawa in Japanese and Agarebaa in Tokunoshima), Kenbuku (TKN: Kɨnbuu), and Uemonawa (TKN: Unnoo).

1.2 Genetic Affiliation and Typological Classification

The Amami language genealogically belongs to the Northern Ryukyuan branch of the Japonic family (Pellard 2015). The second map on figure 2.1 shows the islands where the Northern Ryukyuan languages (Amami, Kunigami and Okinawa) are spoken.

As far as the author recognizes, there is no comparative linguistic study on the genetic relationship among areal variations of Amami. Uemura (1972) classified the Northern Ryukyuan languages into four groups based on their typological features: Kikaijima, Oshima-Tokunoshima, Okinoerabu-Northern Okinawa, and Southern Okinawa. According to Uemura (1972), Tokunoshima falls into the Oshima-Tokunoshima group with Amami Oshima, Kakeromajima, Ukejima, and Yorojima. Nakamoto (1990) regarded the dialects of islands located from Amami Oshima in the north to Yoronjima in the south as ‘Amami dialect’, and classified them into five groups: Northern Amami Oshima, Southern Amami Oshima, Tokunoshima, Northern Kikaijima, and Southern Kikaijima-Okinoerabujima-Yoronjima.

Tokunoshima has dozens of villages, each of which has a unique dialect. As far as the author recognizes, there is no comparative linguistic study on the genetic relationship among areal variations of Tokunoshima. Sakimura (1983) divided the dialects of Tokunoshima into three groups (Northwest, East, and South) based on their typological features, and classified Isen into the Southern group. Hirayama et al. (1966) divided the Tokunoshima dialects into two groups (North and South) based on their typological features and classified Isen into the Southern group.

View Full Size

Figure 2.1

Maps of Japanese Archipelago

• Download Figure
• Download figure as PowerPoint slide

Isen has a number of features which are substantially different from other Amami dialects or even from other areal varieties in the island in terms of vocabulary, phonology, and morphosyntax (see Hirayama et al. 1966).

1.3 Data

This chapter is based on the data set collected through field works (2017–2020) and mail surveys (2020–2022) by the present author. Table 2.1 below gives the information about the consultants.

2.1 Phonemes

There are three subtypes of phonemes: vowel, glide, and consonant.

Tokunoshima has seven vowels, /a, i, u, e, o, ɨ, ɛ/. Table 2.2 summarizes the inventory of vowels.

(3) shows words that are distinguished by the vowels.

(3) a. /mii/ ‘fruit’, /mɨɨ/ ‘hole’, /mɛɛ/ ‘fron’, /maa/ ‘time’, /muu/ ‘algae’

b. /too/ ‘octopus’, /tuu/ ‘ten’

c. /jan/ ‘potato’, /jen/ ‘relationship’

Back vowels, /u/ and /o/, are also [+round]. Back rounded vowels are articulated with rounded protruded lips.

Scholars argue that /ɨ/ in Kametsu Tokunoshima is an advanced central-close vowel (Hattori 1959: 284, Hirayama 1966: 31, Nakamoto 1976: 112). Kato (2021) argues that /ɨ/ in Kametsu Tokunoshima is distinguished from /i/ by F3, and not by F1 or F2, which suggests /i/ and /ɨ/ are distinguished by some articulatory feature which is not frontness.

The height and the frontness of /ɛ/ is controversial. Hirayama et al. (1966: 31) argues that /ɛ/ is an advanced central-open vowel, and Shibata (1981: 32) maintains it is a central-open mid vowel, based on sound impression respectively. Phonetic analysis, however, suggests /ɛ/ is a front-mid vowel, which is barely distinguished from /e/ by F1, F2, and F3 (Kato 2021), although speakers recognize /ɛ/ and /e/ are different sounds.

There are four glides: /j, jˀ, w, wˀ/. Although their sound qualities are similar to that of vowels, they do not constitute morae.

Table 2.3 shows the consonant inventory.

Table 2.4 summarizes phonetic realizations of alveolar plosives and affricates.

2.2 Syllable Structure and Phonotactics

The maximum possible syllable structure is (C1)(G)V1(V2)(C2), where C stands for a consonant, G for a glide, and V for a vowel. Parentheses indicate that the slots are optionally filled. Laryngealized segments can C1 or G only when word-initial. Non-laryngealized consonants can fill C1 at any part in the word. C2 can be filled by /n/ at any part in the word and by /p, t, k, s/ word-medially as the first half of a geminate formed at the boundary of two syllables. CV is the most frequent syllable. Super-heavy syllables occur very infrequently, but they are not completely absent. For example, there are two tokens of boon in the appendix text.

2.3 Mora

V1, V2 and C2 in a syllable each constitute a mora. Isen utilizes morae as phonological units. As noted in § 2.4, the mora is the basic unit of accentuation.

Also, the constraint on the length of a word is sensitive to the number of morae; that is, a word must be bimoraic or longer. Monomoraic words should be lengthened to bimoraic in their surface form (e.g., kɨ ‘tree’ → /kɨɨ/).

2.4 Accent

Isen is a pitch accent language where H and L are distinguished. Nouns have a three-way distinction; that is, each noun lexically has one of the three tonal patterns (Hirayama et al. 1966). (5) shows the ternary accentual distinction among bimoraic nouns.

(5) a. A vs. B: /kóó/ HH ‘river’, /kòó/ LH ‘skin’

b. A vs. C: /hásí/ HH ‘bridge’, /hásì/ HL ‘chopsticks’

c. B vs. C: /kàmɨ́/ LH ‘turtle’, /kámɨ̀/ HL ‘pot’

We have not yet been able to generalize the mechanism behind the surface tonal realizations that would produce them.

Verbs have a two-way accentual distinction: one class of verbs has a rising pitch contour and the other lacks this rise. Table 2.5 shows the two-way distinction.

2.5 Intonation

Isen shows at least two intonation patterns: falling and rising. Interrogative sentences usually have a rising intonation. The collection and classification of intonation patterns in Isen requires further attention.

3.1 Morphological Units

This subsection introduces three morphological units: the word, clitic, and affix. The word is the smallest syntactically and phonologically independent unit. The clitic is a morphosyntactically independent but phonologically dependent unit. The clitic almost always occurs with its host. One morpheme is sometimes phonologically dependent and sometimes not depending on its morphosyntactic environment. Affixes are morphologically dependent and always attach to a stem.

3.2 Word Classes

Five word classes are identified: nouns (§ 4), verbals (§ 5), particles, adnominals, and adverbs. Nouns are words that head a nominal phrase. A noun functions as an argument of predicates or as a predicate itself. Verbals head a predicate phrase and inflect for tense, mood, and voice. Verbals include verbs, whose roots are verb roots, and adjectives, whose roots are PC roots. Particles attach to clauses or phrases, and encode various grammatical functions such as the case of nouns, modality, or grammatical relations between clauses. Adnominals modify NPs without any case marking. Adverbs are words that modify other words or phrases except nouns.

3.3 Root Classes

Three major root classes are identified: nominal roots, verb roots, and PC roots. Nominal roots can appear as a word without suffixation (address nouns and some pronouns require number marking. See § 4). Verb roots have obligatory inflection and optional derivation (see § 5.1). PC roots need to be verbalized as a rule (see § 5.5.1 for exceptional sequential and adnominal forms).

4.1 Pronouns

Pronouns are organized in terms of person (first-person and second-person) and number (singular, dual, and plural). Tokunoshima is not sensitive to clusivity, unlike some Ryukyuan dialects. Demonstratives are used to address a third-person object (See § 6.1). In addition to person and number, second-person pronouns are sensitive to politeness. Also, first-person singular pronouns are divided into three forms (unmarked fusion, unmarked non-fusion, and bare) depending on their morphosyntactic behavior as noted below. Table 2.6 summarizes the pronominal system.

The bare form, which is unique to first-person singular, shows a strong selectional restriction in that it precedes only ⸗ga (nom1). The distinction between fusional¹ and non-fusional pronouns is morphosyntactic; the former stand as the topical subject of a clause or the modifier of a noun without a case particle as in waa kuma⸗nu tjootjoo (‘I am the mayor of the town’). However, the latter require a particle when they appears in a sentence. In addition to this, non-fusional forms are used in nominal predicate phrases.

4.2 Lexical Nouns

A lexical noun can be extended by suffixation or compounding. The maximum structure of a noun is as follows: (prefixal numeral -) stem core (- aug/dim) (- pl). -ganasi is a dedicated augmentative suffix used to refer to personified natural entities and older relatives (e.g., tida-ganasi ‘the sun’, wuba-ganasi ‘aunt’). Also, kinship terms and names of professions are used as augmentative when they are attached to personal names (e.g., ziru-aka, ‘sister Ziru’). A diminutive suffix, -gwa, is typically used for small or beloved objects. There are two types of plural suffixes: -taa and -nkja. The former is associated with human nouns and demonstratives whose referents are humans. Although dual number is marked by pronominal inflection, there is no dual suffix for lexical nouns. Dual number is encoded by phrasal expression as in taroo⸗tu ziroo tˀaari (pn⸗com pn two.people) ‘Taroo and Ziroo, the two of them’.

Lexical nouns are divided into two sub-types: address nouns and non-address nouns. Address nouns are used as terms of address. For example, ama ‘mother’ in (6) is used as a term of address.

(6)

amama	kiiga
ama⸗ma	k-i⸗ga
mother⸗add	come-inf⸗q

‘Mom, will you come with me?’

Proper names, elder kinship terms, and some profession names (e.g., sinsii ‘teacher’) fall into this category. Number marking is obligatory for address nouns. If an address noun is marked by nothing, it is interpreted as singular noun.

4.3 Numerals and Adnominals

5.1 Verbs

The internal structure of the verb is as follows: optional derivational suffixes follow the stem core to form a stem, and one and only one inflectional suffix follows the stem. The stem core is a term adopted by Shimoji (2018: 192–193), to denote the smallest unit that constitutes a stem. A stem core can be a single verb root, but it can also be a compound stem, or a PC root and a derivational suffix, as in the following example, where pairs of square brackets indicate the domain of the stem core: [ut-i+hugas]-i (hit-inf+dig-inf), ‘to drill a hole’, [naga-mɨr]-i, ‘to lengthen’.

5.2 Inflectional Morphology

A verb root requires mandatorily one and only one inflectional suffix. Inflectional suffixes are classified into five types depending on the syntactic property of their inflected form: finite, converb, sequential, participle, and infinitive. Table 2.8 shows the list of inflectional suffixes, types, their functions, and alternation stems they attach to. In Table 2.8, koow- ‘to eat’ exemplifies surface forms. Table 2.9 summarizes the syntactic behavior of inflectional types. In the ‘Type’ column, F, S, C, P, I stand for Finite, Sequential, Converb, Participle, and Infinitive respectively.

For a detailed descriptions of syntax of each verb type, see § 9 and § 10.

5.3 Derivational Morphology

A verb can bear optional derivational suffixes between the stem core and the inflectional suffix. Table 2.10 schematizes the general internal structure of a verb. For a detailed explanation of functions of suffixes, see § 9.

Negative past -adar- does not co-occur with progressive -tur- or past -tar-, and past -tar- does not co-occur with negative -ar-. Also, non-past -jur- do not co-occur with progressive, negative, past, or negative past. The participle inflectional suffix -n cannot immediately follow the stem core, and obligatorily requires at least one of the suffixes in the ‘tense, aspect, polarity’ slot. -ar- that is glossed as pass in the table also functions as the potential marker. -jur- is selected in non-past, affirmative, and non-progressive environments.⁷

The inflectional and derivational affixes introduced so far that begin with /t/ (past -tar-, progressive -tur-, sequential -ti, anterior -taari, and juxtaposition -taatu) and infinitive -i are subject to the morphophonological rules noted below depending on the base-final consonant (e.g., jum-tur-n (read-prog-ptcp) is realized as judun, whereby /m-t/ becomes /d/). The morphophonological rules of /t/-initial suffixes are shown in (7), and the morphophonological rule of the infinitive suffix is shown in (8).

(7) a. Suffix-initial /t/ becomes /d/ after /b, m/ and deletes /b, m/.

b. Suffix-initial /t/ becomes /tj/ after and /k, s, t, j/ deletes /k, s, t, j/.

c. Suffix-initial /t/ becomes /zj/ after /g, n/ and deletes /g, n/.

d. Delete /r, w/ before suffix-initial /t/.

(8) /t/ becomes /tj/ before infinitive /-i/.

5.4 Existential, Resultative and Copula

5.5 Adjectives

Adjectives are words that describe the nature, color, state, etc., and mainly function as predicates in clauses. Words suffixed with -na, however, are not considered to constitute an independent class since the number of words is exceedingly small and it is unclear whether suffixation of -na is productive or not.¹² They are referred to as ‘adjective-like words’ in § 5.5.2. Hereafter, the term ‘adjective’ refers only to those derived from PC roots.

6.1 Demonstratives

The demonstrative system of Tokunoshima is ternary, and each series begins with a unique segment: proximals with /k/, medials with /u/, and distals with /a/. Table 2.11 shows the list of demonstratives.

There are two usages of demonstratives: reference to a concrete object (deixis) and reference to an object already mentioned in the discourse (anaphora). In deictic usage, either a proximal, medial, or distal demonstrative is chosen depending on the psychological and physical distance among the referent, the speaker, and the hearer.

The juxtaposition of a distal and a proximal demonstrative gives a phrase that means ‘various/every/a lot of things, people, and so forth’ as in (12).

(12)

arɨ	kurɨ	jun
arɨ	kurɨ	iw-jur-n
that	this	say-npst-ptcp

‘to say a lot of things’

arɨ kurɨ, ama kuma ‘everywhere’, and assi kassi ‘in various ways’ are attested so far.

6.1.1 Demonstrative Pronoun

Tokunoshima has no third-person pronoun. Instead, demonstrative pronouns (arɨ, kurɨ, and urɨ) and demonstrative adnominals (an, kun, and un) are used to refer to third-person nouns. They can refer to all nouns regardless of their animacy, but different plural suffixes are chosen depending on the animacy of the referent: -taa for personal nouns (e.g., kurɨ-taa as reference to people), and -nkja for non-human nouns (e.g., kurɨ-nkja as reference to cars).

Different situations to investigate the usages of demonstrative pronouns are illustrated by pictures in figure 2.2d, where the person with an open mouth asks the other to look at the referent object.¹⁶

View Full Size

Figure 2.2

Different situations of deictic reference

• Download Figure
• Download figure as PowerPoint slide

Proximal kurɨ is used only for 2.2 and 2.2a, where the referents are closer to the speakers than than addressees. Mesial urɨ is used for 2.2b, where the referent is closer to the hearer. Finally, arɨ is used for 2.2c, where the referent is far from both the speaker and the addressee. urɨ is mainly used for indefinite deictic referents as in (28) of the appendix. It is uncertain whether urɨ is substitutable with the proximal. arɨ functions as a definite deictic reference as in (13).

(13)

kiinu	kootan	arɨ	mutji	tjii
kiinu	koow-tar-n	arɨ	mut-ti	tjii
yesterday	buy-pst-ptcp	that	bring-seq	come.imp

‘Bring those goods (someone) bought yesterday to me!’

6.2 Interrogatives and Indefinites

There are seven interrogatives: personal nominal taru/tan; non-personal nominal nuu; locative nominal daa; selectional nominal dɨn; temporal adverbial/nominal icɨ; numeric adverbial/nominal ikucɨ; and verb of manner ikjasi.

Interrogative nouns function as nouns, in that they can be followed by case particles or be arguments of verbs. However, they cannot be modified by any other words. tan, taru, and nuu can be reduplicated (tandaru, tarudaru, and nuunuu) to function as plural interrogatives (e.g., tarudaru⸗nu kii⸗ga ‘Who and who will come?’). icɨ and ikucɨ function both as nouns and adverbs; it is possible for them to be followed by a case particle, or be arguments of verbs. The interrogative verb ikjasi modifies nouns in its participle form (e.g., ikjasjun mun kootan⸗ga (what.kind thing ate⸗q) ‘What kind of things did you eat?’), or modifies a predicate in its sequential form (e.g., ikjasi izjan⸗ga (how went⸗q) ‘How did you get there?’). Interrogative words are usually located in the same place as the corresponding words in declarative sentences.

7.1 The Head

Various nouns can be the head of a nominal phrase. Lexical nouns can be the head without any modifier. Some nouns, which we will call formal nouns, necessarily require a modifier. Formal nouns have undergone grammaticalization. For example, duki, whose lexical source is tuki ‘time’, always requires a participle verb before it and functions as the head of a phrase which indicates ‘the time when’.

Tokunoshima also displays headless noun phrases, in which participle words stand as modifiers without head nouns as in (14).

(14)

mɛɛ	katjan	mutji	tjii
mɛɛ	kak-tar-n	mut-ti	tjii
before	write-pst-ptcp	have-seq	come.imp

‘Bring to me what (I) wrote before!’

mɛɛ katjan functions as a nominal phrase without a head noun, ‘what (I) wrote before’.

7.2 The Modifier

There are four types of words that modify nouns: the participle form of a verb, the adnominal form of an adjective, demonstrative adnominals, and nouns with the genitive particle. They precede and modify nouns (e.g., akk-jur-n tju (walk-npst-ptcp man) ‘walking man’, or jaa⸗nu mɛɛ (house⸗gen front) ‘garden’). Exceptionally, some pronouns directly modify nouns without a particle (see § 9.3).

7.3 Case and Other Role Marking

Table 2.12 is a list of case particles and their usages. The relevant noun phrases are enclosed in square brackets in the ‘example’ and ‘translation’ columns.

⸗ga and ⸗nu/⸗no¹⁷ alternate depending on the animacy of its host. Also, the nominative and genitive cases show different patterns of alternation. Table 2.13 shows patterns of alternation, where G stands for ⸗ga, and N stands for ⸗nu/⸗no.

Case particles lack their own accentual information and their tonal realizations depend on the tonal information of the succeeding word. However, case particles sometimes have their own accent (e.g., ⸗juri⸗ma is always realized as LLL regardless of its host). Phonological dependencies of case particles need further examination.

8.1 Verbal Predication

8.2 Nominal Predication

A noun or the infinitive form of a verb may compose a complex predicate with a copular verb (cf. (10)). In a non-past affirmative main predicate, a noun can be the predicate without a copular verb. In other environments, copula or existential verbs appear. See § 5.4.2 for details of V2.

9.1 Sentence Type (Declarative, Interrogative, Imperative, Etc.)

Sentence types can be divided into two main categories: sentences that narrate and sentences that demand something from the other person. The former is a declarative sentences. The latter includes interrogative sentences,¹⁸ imperative sentences, and hortative sentences.

The basic word order in Isen is subject (topic)–indirect object–direct object–predicate. The basic word order basically follows this. In natural discourse, sentences with a word order that differs from the basic word order can be observed as shown in (17), where the subject wan⸗ja follows the predicate ikjun wakɨjoo.

(17)

hosjuuzjugjootji	jitjijaa	ikjun	wakɨjoo	benkjoo	siigatji	jitji	wan.ja
⟨hosjuuzjugjoo⟩⸗tji	jiw-ti⸗jaa	ik-jur-n	wakɨ⸗joo	⟨benkjoo⟩	s-i⸗ga⸗tji	jiw-ti	wan⸗ja
remedial.class⸗quot	say-seq⸗cfp	go-npst-ptcp	fn⸗cfp	study	do-inf⸗pur⸗quot	say-seq	1sg⸗top

‘It was called a remedial class. Saying “in order to study”, I used to go to remedial class.’

9.2 Alignment

Tokunoshima has marked nominative alignment, whereby subjects of transitive and intransitive sentences are almost always marked by the nominative marker, and objects are never marked. As noted in § 7.3, the nominative markers ⸗ga and ⸗nu are chosen depending on the subject’s animacy. Subjects may also be unmarked, although it is rare. Attested unmarked subjects occur in sentences that express weather, existence, or the emergence of things. Since unmarked subjects are observed more often in natural discourse than in elicitation, this complicates investigation under controlled conditions, and requires more data. In addition, subjects of predicates with low transitivity, such as perceptual verbs, potential verbs, and existential verbs, can be marked by the dative case (19).

(19)

wannin.ja	saaran
wan⸗nin⸗ja	sa-ar-ar-n
1sg⸗dat⸗top	do-pot-neg-ptcp

‘I can’t do that.’

9.3 Possession

There are two possessive constructions: phrasal possession (corresponding to ‘my house’ in English) and possessive predicates (corresponding to ‘I have this.’ in English).

Phrasal possessions are encoded through linking a possessor and a possessee by a genitive marker, or by direct modification. Nouns other than personal pronouns modify another noun with the genitive case marker (e.g., maju⸗nu mɨɨ (cat⸗nom eye) ‘cat’s eyes’).

Pronouns show different patterns of possessive expression in that some are marked by the genitive case marker and some modify a noun directly in a possessive phrase. Out of the 16 pronominal forms in Table 2.6, wan, the first-person singular non-fusional, does not occur in possessive phrases. Other non-fusionals and their plural forms, uri, ura, uri-taa, and ura-taa, are marked by ⸗ga to modify nouns (e.g., ura⸗ga jaa ‘your house’). Fusionals and plurals that end with -kja modify nouns directly (e.g., wakkja azja ‘our father’). No possessive structure is attested so far for dual pronouns.

Second, there are two possessive predicates: the dative possessive and the lexical possessive. Locative possessive phrases are composed of a possessee subject, a dative possessor, and an existential verb (20a), and lexical possessives of a possessor subject, a possessee object, and a lexical possessive verb (20b).

(20)

a.

wannin.ja	kanɨnu	nen
wan⸗nin⸗ja	kanɨ⸗nu	neer-n
1sg⸗dat⸗top	money⸗nom	neg.ext-ptcp

‘I have no money.’

b.

an	tjuja	teegee	kanɨ	mutjun
an	tju⸗ja	teegee	kanɨ	mut-tur-n
that	person⸗top	much	money	have-prog-ptcp

‘S/he has much money.’

9.4 Valency Changing Operations

Subjects are marked by a nominative marker unless they are marked by the additive ⸗ma or topic ⸗ja. Objects are unmarked in active affirmative sentences. This subsection discusses operations that introduce additional arguments and/or change their case marking.

9.5 Polarity

The affirmative is the unmarked polarity, and negative is overtly marked. Particular negation (Some S is not P) is encoded by predication. Universal negation (No S is P) is encoded by a combination of an interrogative word and a negative predicate. There are three ways to encode negation in a predicate: suffixation of the negative suffix -ar- onto to the main verb, suffixation of the negative suffix onto the light verb, or the negative existential verb. When the verb is not followed by any particle, the negative -ar- derives the negative form (e.g., kak-ar-n (write-neg-ptcp)). Table 2.14 summarizes strategies for negation of verbs, adjectives, and nominal predicates. Each class is represented by kak- ‘write’, aa ‘red’, and wan⸗ja sinsii ‘I am a teacher’ respectively. Translations for universal negations are as follows: ‘(I) don’t write anything’, ‘Nothing is red’, and ‘No one is a teacher’. ‘ptcl’ in the table stands for any one of the limit particles and information particles (such as ⸗ma, ⸗du, ⸗ja and so on). Ones enclosed in parentheses may or may not require a particle. Ones NOT indicated in parentheses always require a particle. Thus, while aa-ku neer-n is grammatical, kaki s-ar-n is ungrammatical. ⸗ma in universal negations cannot be substituted by any other particle.

9.6 TAM

9.7 Information Structure and Its Formal Encoding

Topicalized nouns are marked by the topic particle ⸗ja. Nominative particles do not co-occur with the topic marker. Focalized elements are sometimes marked by the focus particle ⸗du, which can also co-occur with nominative particles. ⸗du appears within focalized sentences in which the speaker recalls the contrast compared to the focalized element.

(22)

mukasija	jangadu	hanmɛɛ	jatan
mukasi⸗ja	jan⸗ga⸗du	hanmɛɛ	jar-tar-n
old.time⸗top	potato⸗nom⸗foc	food	cop-pst-ptcp

‘Potatoes were the only food at that time.’

In (22), the speaker recalls other foods like rice or bread, and excludes them as candidates for what he was eating at that time.

10.1 Clause Combining Strategies

10.2 Quotatives

A quotative clause is a clause formed with the quotative marker ⸗tji. Any word, phrase or sentence followed by the quotative marker can form a quotative clause. A quotative clause functions as the complement of speech or cognitive verbs.

(23)

arigatooja	simagutjisi	oboradaanitji	jun
⟨arigatoo⟩⸗ja	simagutji⸗si	oboradaani⸗tji	iw-jur-n
thank.you⸗top	local.dialect⸗ins	thank.you⸗quot	say-npst-ptcp

‘We say Oboradaani in the local dialect for “Thank you”.’

10.3 Insubordination

A subordinate clause may appear in a sentence without a main clause as in (24).

(24)

ugwasi	natikajaa
ugwasi	nar-ti⸗ka⸗ja
so	become-seq⸗cond⸗cfp

‘I wish that were true (lit. If it became …).’

The sequential form of verbs (-ti) functions as a predicate of the main clause. This usage of the sequential form may have originally been an insubordination of a dependent clause whose predicate is a sequential verb.