Reported speech in Bezhta

Толық мәтін

ABBREVIATIONS

i, ii, iii, iv, v — genders (noun classes) i–v

abl — ablative

abs — absolutive

add — additive enclitic

antip — antipassive

antr — anterior converb

apud — apudessive, location ‘near’

attr — attributive

caus — causative

comit — comitative

cond — conditional converb

cop — copula

cvb — converb

dir — directive

encl — enclitic

erg — ergative

foc — focus marker

gen — genitive

hpl — human plural

imm — immediate converb

imp — imperative

in — inessive, location ‘in, inside’

inch — inchoative

inf — infinitive

ins — instrumental

inter — interessive, location ‘in (a dense mass)’

interr — interrogative

irr — irrealis

lat — lative

msd — masdar

neg — negation

nhpl — non-human plural

obl — oblique

opt — optative

pl — plural

poss — possessive, location ‘at’

proh — prohibitive

prs — present tense

pst — past tense

ptcp — participle

q — question marker

quot — quotative

refl — reflexive

rem.pst — remote past

rep — reportative enclitic

sg — singular

sim — simultaneous converb

sub — subessive, location ‘under’

sup — superessive, location ‘on, above’

transl — translative

until — temporal converb

1. Introduction

It has been observed that the well-known dichotomy between direct and indirect speech, which is common for Standard Average European (SAE) languages, is inapplicable in many languages [Aikhenvald 2008; Evans 2013; Nikitina, Bugaeva 2021]. One such language is Bezhta, a language belonging to the Tsezic group of the Nakh-Daghestanian (also called East Caucasian) family. There has been relatively little research conducted on reported speech in Nakh-Daghestanian languages. Recent studies have focused on languages such as Tsez [Polinsky 2015], Archi [Daniel 2015], Sanzhi Dargwa [Forker 2019], Aqusha Dargwa [Ganenkov 2023], and Tabasaran [Bogomolova 2023]. Comrie et al. [2010] present a general overview of reported speech in the Tsezic languages. This study is the first in-depth investigation of reported speech in Bezhta. The paper contributes to the general discussion on reported speech constructions using data from a less-described minority Nakh-Daghestanian language.

When describing reported speech in Bezhta, I use the established terminology found in Aikhenvald [2008], Culy [1994], etc. In example (1a), Murad is the original speaker, i.e. the author of the quote (Murad is a male name and serves as the main character in my elicited examples), and the boy is a participant of the quote. The clause containing a speech verb is called the framing clause, and the embedded clause consisting of the quote is the quotation clause, or the reported utterance. Sentence (1b) is an example of self-quotation, where the personal pronoun in the framing clause is both the current speaker, i.e. the person who produces the speech report, and the original speaker at the same time.

(1) a. Murad said, ‘The boy likes khinkal.’

1. I said, ‘I like khinkal.’

Bezhta reported speech constructions can involve ordinary quotation (with speech verbs), self-quotation (i.e. original speaker = current speaker), and quotation with an unknown source of evidence (i.e. hearsay evidential, see Section 3.2). The reported speech construction in Bezhta consists of the framing clause based on a speech verb and a quotation clause, often followed by the quotative enclitic ƛo, which is attached directly to the quotation. Other functions of this enclitics are discussed in Section 4. The most common speech verbs are niso- ‘say, tell’, nisol- ‘ask’, -iqʼläše- ‘talk’, iƛe- ‘call, shout’, -ek’- ‘tell’, ɬazi -oː- ‘inform, warn’, čoq- ‘inform’, lalaƛo- ‘shout’, χabar m-eː- ‘tell a story’, gerda- ‘swear, scold’, morƛo- ‘scold’, qäƛe- ‘shout’, jäže niƛ- / ƛ’odo -oχ- ‘promise’, harzi -oː- ‘ask, beg’, maɬzi -o:- ‘teach’, p’ärƛö- ‘chat’, hardizi -aq- ‘ask, beg’, χantoq- ‘ask, plead’, eʁƛ’a -eⁿƛ’e- ‘ask’, amro -oː- ‘order’.

Reported speech constructions in Bezhta can involve not only speech verbs like niso- ‘say’, but also propositional attitude verbs like urʁezi -aq- ‘think’, propositional attitude verbs like gic’- ‘seem’, direct perception verbs like tuq- ‘hear’, achievement verbs like jak’ƛ’a -oⁿq’- ‘remember’, and some others. Thus, reporting includes not only speech reports, but also reporting one’s own or someone else’s thoughts and ideas. Reported thought constructions are treated together with reported speech constructions here.

The data in the paper are predominantly drawn from naturalistic folkloric text collections [Khalilov 2020], which include fairy tales, fiction, anecdotes and the like, and few elicited examples are used. The data represent the Bezhta proper dialect (i.e. the dialect of the village of Bezhta itself). The paper is organized as follows. Section 2 describes reporting strategies in Bezhta. Section 3 deals with deictic and anaphoric elements used in reported speech, including indexical shift and the behavior of reflexives. Section 4 treats the synchronic functions and the grammaticalization source of the quotative enclitic. Section 5, which describes the morphosyntactic status of the reported speech construction, sums up the paper.

2. Reporting strategies

Like some other complement-taking predicates in Bezhta, utterance predicates are able to use more than one complementation type. The distribution of the complementation types associated with verbs of speech does not depend on coreferentiality, but rather on whether the perspective of the original or the current speaker is taken, cf. Section 3.

The complement clauses of verbs of speech can be either finite or non-finite. Finite strategies include reported speech presented either by simply juxtaposing the framing clause and the utterance being reported, i.e. without any quotative enclitic, or by attaching the quotative enclitic to the final element in the quotation clause, the latter being the most common strategy. Another possibility for conveying reported speech is to use the quotative directly followed by the perfective converb of ‘say’. Non-finite reporting is based on nominalization, which is a less common way to encode reported speech.

2.1. Reporting by juxtaposition (without a quotative)

The reported speech construction can be introduced without any marking, i.e. without the quotative marker. The speech reported in this way can consist of a simple statement in the indicative mood (2), but also a request (3), question, command, promise, threat, etc.

(2) bečedab χalq’-ila niso-na gej wahli gisa Ø-eh-da,

rich people-erg say-cvb cop that(i) out i-let.out-cond

wahdo-la-s teli okko nic-ca.

that(iii)-obl-gen1 much money(iii) give-prs

‘The rich people said, “If you let him out, we will pay a lot of money.”’

(3) do holcoqa harzi j-oː-jo łina wodo niƛ-a dil

1sg.erg 3sg.poss beg iv-do-pst five day give-imp 1sg.lat

bado mäče m-iqo-cal.

another(iii) place(iii) iii-get-until

‘I begged him, “Give me five more days until I find a new place.”’

In the reported speech construction without the quotative enclitic, the order of the framing clause and the quote is free.

(4) mesed naː-ƛʼa-d j-iqʼläše-š mi, wahgi j-ekʼ-ca hoɬo.

honey where-sup-q ii-talk-prs 2sg.abs this ii-tell-prs 3sg.erg

‘“Honey, where are you speaking from?” she says (this).’

2.2. Reporting with the quotative enclitic

The most common type of reported speech constructions consists of the framing clause and the quotation clause, followed by the quotative enclitic ƛo. Reported speech is mainly introduced by verbs of speech, which include niso- ‘say, tell’, nisol- ‘ask’ (5), -iqʼläše- ‘talk’, -ek’- ‘tell’, ɬazi -oː- ‘inform’ (6), and lalaƛo- ‘shout’, etc.

(5) sud wahallijo kuwaƛʼa caːⁿ gej꞊ƛo nisol-ca aboqa öždi.

why such canyon.sup name cop꞊quot ask-prs father.poss son.erg

‘“Why does the canyon have such a name?” the son asked his father.’

(6) hugiʔƛʼa꞊na ɬazi j-oː-s, haχar bejten-l-il hadurɬi j-ow-a꞊ƛo

there꞊add inform iv-do-prs well wedding-obl-lat preparation(iv) iv-do-imp꞊quot

niso-na.

say-cvb

‘Then they inform them, “Well, get ready for the wedding!”’

The framing clause containing the speech verb may follow (5), precede (6) or interrupt the reported speech (7).

(7) naː-d gej꞊ƛo mähmä-hažijaw, niso-s munahal čurijaw,

where-q cop꞊quot Magomed-gadzhi say-prs sin.pl forgivable

saʔdulla abo, naː-d gej꞊ƛo.

Saadulla father.erg where-q cop꞊quot

‘“Where are you, Magomedgadzhi?” says father Saadulla (God forgive his sins), “Where are you?”’

Reporting is possible with propositional attitude verbs such as ‘think’ and ‘seem’. The quotative enclitic, which mainly occurs in reported speech with verbs of speech and manipulative verbs, is also used to mark complement clauses of propositional attitude verbs (e.g. urʁizi -aq- ‘think’, pikro bo:- ‘think’, gic’- ‘seem’), the direct perception verb tuq- ‘hear’ (8), and the achievement verb j-akʼƛʼa -oⁿqʼo- ‘remember’ (9). Sentence (8) is an example of a hearsay evidential.

(8) tuq-na zuqʼo-jo dil, itʼinaː qʼaridɬi j-oⁿqʼo-da j-uqʼo

hear-cvb be-pst 1sg.lat small.pl difficulty(iv) iv-come-cond iv-big

j-iqʼe-š, j-uqʼo j-oⁿqʼo-da itʼinaː gäččö-zü j-ega-s꞊ƛo.

iv-know-prs iv-big iv-come-cond small.pl cop.neg.cvb-foc iv-see-prs꞊quot

‘I have heard that if small problems come, they feel like big problems, but if big problems come, small problems do not matter.’

(9) dil j-akʼƛʼa j-oⁿqʼo-na qʼurban untila-qa Ø-uʁo-na꞊ƛo gic-ca.

1sg.lat iv-heart.sup iv-come-pst Kurban(i) disease.obl-poss i-die-pst꞊quot seem-cvb

‘I remembered that Kurban died from a disease.’

The quotation clause formed with propositional attitude verbs such as k’oc’ola- ‘think’, pikro boː- / pikro baq- ‘think’ is often accompanied by gic’- ‘seem’:

(10) kʼocʼola-na, sud hoƛoʔ do Ø-ec-al꞊ƛo gicʼ-na Ø-eⁿƛʼe-jo

think-cvb why here 1sg.abs i-stay-inf꞊quot seem-cvb i-go-pst

šahar-li-ɬ-da:.

town-obl-inter-dir

‘Thinking, “Why I should stay here?”, he went into town.’

(11) pikro gej hollos hoⁿs äjšät-ä-ʔ gisdaː hide-l kal.kal

thought cop 3pl.gen1 one Ayshat-obl-in outside refl.pl-lat obstacles(iii)

b-oh-cas sukʼo gäʔä꞊ƛö gicʼ-ijo.

iii-do-prs.ptcp person cop.neg.prs꞊quot seem-pst

‘They thought that except for Ayshat, there was no one who could hinder them.’

The framing verb together with the quotation clause can make up one part of a more complex sentence. In example (12), the framing clause is itself embedded, and thus the speech verb appears in the contextual converb form nisona; in (13) the speech verb appears in the present tense form nisos in the matrix clause.

(12) hini-zu zukʼ-ca qʼäm holco, äbdäl do꞊ƛo niso-na.

refl.erg-foc hit-prs head 3sg.erg stupid 1sg.abs꞊quot say-cvb

‘He himself hits his head, saying, “I am stupid.”’

(13) hinqa baɬaj koːʔ j-oⁿqʼo-ɬ, holco niso-s zuhrat-i ile

refl.poss sword(iv) hand.in iv-get-antr 3sg.erg say-prs Zuhra-erg 1pl.abs

b-ečä<bä>nne-jo꞊ƛo.

hpl-leave<pl>-pst꞊quot

‘When he gets the sword in his hands, he says, “Zuhrat has left us.”’

The reported utterance can be introduced without any speech verb, as it can easily be retrieved from the context. Reporting without speech verbs is often found in dialogues and conversations.

(14) Ø-oⁿqʼo-jo abdurahmaⁿ waccal.

i-come-pst Abduraxman cousin(i)

‘Cousin brother Abduraxman came.’

sijo-d j-aq-ijo jaccal꞊ƛo?

what-q iv-happen-pst cousin(ii)꞊quot

‘What’s happened, cousin sister?’ [he said]

2.3. Reporting with a speech verb and the perfective converb of ‘say’

Another possibility is for the reported speech with the quotative ƛo to be directly followed by the perfective converb of niso- ‘say’, in addition to the verb of speech which occurs in the framing clause. The converb nisona does not add any semantic or syntactic value; it merely serves to indicate that the text is a quotation. It must always immediately follow the quotative particle. Such reported speech constructions are frequently used in narratives. In examples (15)–(17) the quotation clause is followed by the quotative enclitic and the perfective converb nisona.

(15) haχar sijo-d dul j-aq-ijo꞊ƛo niso-na nisol-ca ijo-qa.

well what-q 2sg.lat iv-happen-pst꞊quot say-cvb ask-prs mother-poss

‘“So what happened to you?” they ask their mother.’

(16) hogco-qa lalada-s꞊ƛo, haχar j-oⁿqʼo-s were j-iƛʼ-a꞊ƛo

3sg.obl-poss shout-prs꞊rep hey nhpl-come-prs beware nhpl-kill-imp꞊quot

niso-na.

say-cvb

‘(They) shout at him “Hey, they are coming, beware, kill them!”’

(17) wahla: ƛobaʁak-ca meχ, iƛe-la-baː-s naː-d j-aq-ijo꞊ƛo

so become.noon-prs time call-antip-pl-prs where-q ii-happen-pst꞊quot

niso-na.

say-cvb

‘So it was getting dark, and they were calling out “Where have you ended up?”’

2.4. Nominalization

Another way to present reported speech is to use non-finite complementation with utterance predicates, as well as with propositional attitude, direct perception, and achievement verbs. Nominalization in Bezhta, traditionally known as masdar form with the suffix -ɬi, occurs with various types of complement-taking predicates including utterance predicates. Morphologically the masdar is a noun, but syntactically it is a verb as it retains all arguments in their regular cases. Here are some illustrative examples of the masdar strategy with predicates niso- ‘say’ (18), χabar m-eː- ‘tell (a story)’ (19), and -ega- ‘see, dream’ (20).

(18) hogƛo zamal-laː-ʔ äƛäʔ ädäm-läː-l nis-al nucʼ-aʔa-s zuqʼoː,

3sg.obl time-pl.obl-in village.in man-pl.obl-lat say-inf must-neg-prs be.pst

sijo j-aq-ijo-ɬi꞊na; hide-l-zu j-iqʼe-š zuqʼoː.

what iv-happen-pst.ptcp-msd꞊add refl.pl.obl-lat-foc iv-know-prs be.pst

‘In the village, at that time, there was no need to inform others about what had occurred; they were aware of it themselves.’

(19) bičʼ-li-ƛʼa Ø-oⁿqʼo-ɬ, m-eː-š χabar niʁa Ø-uʁojo-ɬi꞊na.

consciousness-obl-sup i-come-antr iii-send-prs story(iii) how i-die.pst.ptcp-msd꞊add

‘When he regains consciousness, he tells how he died.’

(20) miƛaɬ-zu꞊na Ø-ega-s badlola raƛʼ-li-ƛʼa-laƛʼaː

sleep.inter-foc꞊add i-see-prs other.obl earth-obl-sup-transl

sapar-li-ʔ Ø-eⁿƛʼe-cas-ɬi.

journey-obl-in i-go-prs.ptcp-msd

‘I dream that I travel to different lands.’

3. Reporting with deictic and anaphoric elements

All four reported speech strategies readily employ deictic and anaphoric elements in the reported clause. These elements can receive different interpretations, i.e. they can be considered from the perspective of either the original or the current speaker, or represent mixed speech.

Based on Evans [2013: 70], direct and indirect speech are distinguished as follows. Canonical direct speech reproduces the original speaker’s words and thus includes all linguistic peculiarities of the original, presenting all deictically sensitive expressions from the perspective of the original speaker. Canonical indirect speech assimilates the material presented by the original speaker to the reporting speaker’s perspective, abstracts away from all linguistic particularities of the original and replaces them with a stylistic representation consistent with the rest of the reporting speaker’s stylistic choices; in canonical indirect speech, all deictically sensitive expressions are presented from the perspective of the reporting speaker.

However, languages with indexical shift, like Bezhta, do not always meet the criteria for direct or indirect speech in full. In the following sections, I will lay out the principles of reporting with indexical shift in Bezhta and discuss the notion of logophoricity.

3.1. Indexical shift

Deictic and anaphoric elements used in reported speech constructions in Bezhta demonstrate different interpretations. When reporting involves embedded first and second person pronouns, we observe indexical shift in Bezhta. The phenomenon of indexical shift involves a situation where interpretation of embedded indexicals, i.e. first and second person pronouns, depends on an utterance event. The embedded indexicals can be interpreted either from the perspective of the original speaker (shifted reading) or that of the current speaker (unshifted reading). Central theoretical studies on this phenomenon are Schlenker [2003], Anand and Nevins [2004], and Deal [2020]. Currently, typological work on indexical shift is growing. Languages attested as showing indexical shift include some other Nakh-Daghestanian languages of the Caucasus: see the recent work of Bogomolova [2023] on Tabasaran, Ganenkov [2023] on Akusha Dargwa, and Polinsky [2015] on Tsez, a closely related sister language of Bezhta. The latter work on Tsez served as a model for this study, which is the first investigation of indexical shift in Bezhta. This study does not propose a formal analysis but is predominantly a descriptive treatment of the main properties of indexical shift found in Bezhta.

Consider the Bezhta example (21) below, with the first person pronoun in the reported clause. This sentence is ambiguous, as the embedded first person pronoun has two possible interpretations. The default reading is the ‘shifted reading’ whereby the first person pronoun refers to the original speaker Murad. This interpretation is always primary if no overriding context is given. If some such context is provided, the first person pronoun can refer instead to the current speaker.

(21) Murad-i niso-s do tabał-na gej꞊ƛo.

Murad-erg say-prs 1sg.abs be.tired-cvb cop꞊quot

1. ‘Murad says, “I am tired.”’
2. ‘Murad says that I am tired.’

Indexical shift is possible with second person pronouns:

(22) Murad-i nisol-ca aboqa dul naːna siⁿ b-ega-ʔeš-di꞊ƛo.

Murad-erg ask-prs father.poss 2sg.lat anywhere bear(iii) iii-see-neg.pst-q꞊quot

1. ‘Murad asked the father, “Have you seen a bear anywhere?”’
2. ‘Murad asked the father_i if he_i had seen a bear anywhere.’

The interpretation of non-declarative aspects of speech such as imperatives, prohibitives, optatives, vocatives, questions, and exclamations/interjections can be shifted or unshifted. For instance, unlike in most Indo-European languages, where imperative forms are replaced by infinitives or subjunctives in the context of indirect speech, imperatives in the Tsezic languages are always reported without any formal change. In sentence (23) the embedded goal can be either the original or the current speaker, i.e. the meaning of dil ‘for me’ is either shifted or unshifted, as signalled in translation.

(23) Murad-i niso-s b-ow-a dil χok’a꞊ƛo.

Murad-erg say-prs iii-do-imp 1sg.lat khinkal(iii)꞊quot

‘Murad_i says to prepare khinkal for me_i/j.’

Bezhta exhibits the same characteristics as are commonly observed in languages with indexical shift [Deal 2020]. One of these properties is sensitivity to the structure of the embedded clause. In Bezhta as in, e.g., Tsez [Polinsky 2015], indexical shift is possible only in complements that are finite. Indexical shift occurs with three types of reported speech construction: reported speech formed without the quotative, reported speech formed with the quotative and reported speech formed with the quotative and the converb ‘say’. The description of indexical shift in this paper is based on standard reported speech constructions formed using the quotative. In reported speech expressed by means of nominalization (cf. Section 2.4) indexical shift is not possible. Thus, structures with nominalization clearly correspond to indirect speech, which involves reporting from the current speaker’s perspective. Sentences in (24) illustrate non-finite and finite reported clauses with the verb of speech χabar m-eː- ‘to tell (a story)’.¹

(24) a. Murad-i χabar m-eː-š nito do bazaj-ja-ʔ Ø-eⁿƛ’e-jo-ɬi.

Murad-erg story(iii) iii-send-prs when 1sg.abs market-obl-in i-go-pst.ptcp-msd

‘Murad_i tells the story about the time I_j/*i went to the market.’

1. Murad-i χabar m-eː-š nito do bazaj-ja-ʔ Ø-eⁿƛ’e-jo꞊ƛo.

Murad-erg story(iii) iii-send-prs when 1sg.abs market-obl-in i-go-pst꞊quot

‘Murad_i tells the story about the time I_j/i went to the market.’

The obvious general difference between finite and non-finite verb forms in Bezhta is that a finite verb heads a main clause and a non-finite verb heads a subordinate clause. However, there are structures where a verb form that typically heads a dependent clause can instead occur as the head of a main clause. This is an instance of a phenomenon that has been referred to as insubordination [Comrie et al. 2016]. In (25) the reported clause contains the irrealis subordinating conjunction, which normally occurs in the protasis of conditional clauses; here it heads a main clause. The given example has two possible interpretations, shifted and unshifted. Therefore, indexical shift is possible not only in finite complementation but in insubordination.

(25) öždi niso-s do enekzi Ø-aq-eʔeš-q’odaː꞊ƛo.

boy.erg say-prs 1sg.abs listen i-become-neg.pst-irr꞊quot

‘The boy_i says, “If only I_i/j had not listened!”’

Another phenomena of indexical shift is the Shift Together constraint, which says that either all embedded indexicals shift or none of them does [Deal 2020: 18]. Consider example (26), which is drawn from a fairy tale about the three sons of the khan. Here the current speaker (the narrator) tells how the youngest son, who is the original speaker, falls into a pit and finds himself in a strange place. In (26) there is no discrepancy between spatial and temporal deixis, i.e. the temporal adverb hoƛoʔ ‘here’ and the two first person pronouns dil and do are both shifted and interpreted from the perspective of the original speaker.

(26) wah, dil sijo-d hoƛoʔ j-aq-ijo꞊ƛo niso-na gej it’ina-so-la

oh 1sg.lat what-q here iv-happen-pst꞊quot say-cvb cop small.obl-attr-obl

is-t’i, naː-d do Ø-oⁿq’o-jo꞊ƛo niso-na gej.

brother-erg where-q 1sg.abs i-come-pst꞊quot say-cvb cop

‘“Oh, what has happened to me here?” said the younger brother, “Where have I come?”’

Indexical shift becomes obligatory in some contexts and under certain circumstances. In most narratives, such as fairy tales, anecdotes and the like, the indexicals always reflect the perspective of the original speaker, since in such stories the narrator does not usually talk about himself. Sentence (27) is taken from the same story about the khan’s three sons. It illustrates the reporting of wh-questions and imperatives, and the retention of the first person pronoun referring to the original speaker, the youngest brother.

(27) it’ina is-t’i niso-na gej naː-d miže woddiʔ m-e<ja>ƛ’e-š,

small.obl brother-erg say-cvb cop where-q 2pl.abs day.in hpl-go<pl>-prs

nis-a di:-qa꞊na꞊ƛo.

say-imp 1sg.obl-poss꞊add꞊quot

‘The youngest brother asked, “Where do you go every day? Tell me as wellǃ”’

The interpretation of embedded indexicals can also depend on the presence of agreement. Let us consider the following constructed example. Native speaker consultants immediately reject example (28a), as they consider it ungrammatical on the grounds that the embedded predicate should show feminine agreement because the embedded first person pronoun has a female referent (i.e. kibba ‘the girl’), and they correct the example using a predicate featuring the gender II (feminine) prefix j- (28b). However, as soon as an appropriate context is given, informants recognize that example (28a) is in fact grammatical. If the current speaker in (28a) is male, only the unshifted reading is possible, whereas example (28b) allows both readings.

(28) a. kibba niso-s do Ø-ek-na gej꞊ƛo.

girl.erg say-prs 1sg.abs i-fall-cvb cop꞊quot

‘The girl_i says I (male)_j/*i have fallen down.’

1. kibba niso-s do j-ek-na gej꞊ƛo.

girl.erg say-prs 1sg.abs ii-fall-cvb cop꞊quot

‘The girl_i says I(female)_i/j have fallen down.’

So far, the data presented have demonstrated the use of embedded indexicals within one clause. Example (29) below demonstrates indexical shift across several complements. It illustrates the use of the first person singular, the first person plural, the second person pronouns and the demonstrative, which serves as a third person pronoun in Bezhta.

(29) [Aishat talks about what Saida said to her yesterday:]

Saida huł j-ek-ca gih mi꞊na iloʁoj q’ej b-ec-al

Saida yesterday ii-say-prs come(ii).imp 2sg.abs꞊add 1pl.comit down hpl-sit-inf

kafeliʔ m-eƛ’e-na b-iqlä<bä>š-äl꞊ƛo, ijo꞊na žo꞊na do꞊na,

café.in hpl-go-cvb hpl-talk<pl>-inf꞊quot mother꞊add thing꞊add 1sg.abs꞊add

gih, holo꞊na j-oⁿqo-s.

come(ii).imp 3sg.abs꞊add ii-come-prs

‘Yesterday Saida_i said, “You_j also come with us_i+j+k, (we) will go and sit at the café to talk, (our) mother_k and others_z and I_j/*i, (you) come, and she_x will also come.”’

This example comes from a dialogue between two female cousins, and features mixed reporting. The first part of the quotation [gih mina iloʁoj q’ej becal kafeliʔ meƛ’ena biqläbäšäl-ƛo] is reported from the perspective of the original speaker, Saida, who is inviting her sister Aishat for a get-together at the café; here the second person pronoun refers to the current speaker, while the first person plural pronoun refers to the current and original speakers together. This embedded utterance is closed off by the quotative ƛo. The next part [ijona žona dona] is reported from the perspective of the current speaker, Aishat, as the embedded first person singular pronoun refers to her. The next part [gih], which is an imperative form, is reproduced from the point of view of Saida, featuring direct reporting. The last part [holona joⁿqos] can be considered to be reporting from the perspective of either the current speaker or the original speaker, as the feminine third person pronoun refers to a new referent (their new sister-in-law). Third person pronouns or demonstratives do not refer to either speaker, but to a third person, i.e. to a new referent. Unlike the Shift Together constraint, the indexical shift does not occur at the level of the entire discourse, but is clause-bound [Polinsky 2015: 22]. This sentence is an illustrative example of what is known as biperspectival [Evans 2013: 67], referring to reported speech in which the perspective of the original speaker alternates with that of the current speaker.

Indexical interpretation is also available in reporting with wh-words. In (30a) the (in situ) wh-word łoʁoj corresponds to the comitative phrase in the declarative embedded clause. Here we observe two readings, shifted and unshifted. In (30b) the wh-word is moved to the front of the matrix clause. The preference is for the unshifted reading, although the shifted interpretation is also permitted.

(30) a. Ruslan-i niso-s do ło-ʁoj Ø-iqläše-š꞊ƛo?

Ruslan-erg say-prs 1sg.erg who.obl-comit i-talk-prs꞊quot

‘Who did Ruslan_i say I_i/j was talking to?’

1. ło-ʁoj Ruslan-i niso-s do Ø-iqläše-š꞊ƛo?

who.obl-comit Ruslan-erg say-prs 1sg.erg i-talk-prs꞊quot

‘Who did Ruslan_i say I_j/?i was talking to?’

The use of reflexive pronouns in reported speech blocks the indexical interpretation, just like in Tsez [Polinsky 2015: 28]. Consider the following examples with the first person reflexive (31a) and the third person reflexive (31b).

(31) a. kibba niso-s do-zu j-ow-al ken꞊ƛo.

girl.erg say-prs 1sg.abs-foc iv-do-inf food(iv)꞊quot

‘The girl_i said, “I_i/*j (myself) will make the food.”’

1. kibba niso-s hini j-ow-al ken꞊ƛo.

girl.erg say-prs refl.erg iv-do-inf food(iv)꞊quot

‘The girl_i said, “I_i/*j (myself) will make the food.”’

The reflexive pronoun used in reported speech refers back to the speaker of the original speech situation only, thus conveying the same reference relations as a first person pronoun. Thus, the simple reflexive pronoun (the 1^st person and 3^rd person reflexives) and the 1^st person pronoun are used as alternatives encoding the original speaker. In Section 3.2, the behavior of reflexive pronouns in reported speech is discussed.

3.2. Anaphoric elements in reported speech

Bezhta has simple, reduplicated and intensified reflexive pronouns,² which differ in their anaphoric properties. The simple reflexive (and the intensified reflexive) can be bound in local and non-local domains, i.e. used in long-distance reflexivization, whereas the complex reflexives are always locally bound, i.e. bound in simple clauses [Comrie et al. 2014]. The discussion below is based solely on the simple reflexive and the intensified reflexive pronouns. In complex clauses, e.g. in complement clauses, the simple reflexive pronouns show coreference with a main clause referent. In examples (32) representing the complement clause, the embedded third person reflexive pronoun hinis ‘refl.gen1’ shows long-distance reflexivization, i.e. it is bound across the complement clause boundary referring to the main clause subject Adil.

(32) adil Ø-uqʼadoj Ø-eⁿƛʼe-š hinis is zoʁijas nis-al.

Adil(i) i-big.apud i-go-prs refl.gen1 brother find.pst.ptcp.gen1 say-inf

‘Adil_i went to the head_j (of the village) to tell (him) about his_i/*j brother’s finding.’

Just like in complex clauses, in reported speech construction the simple third person reflexive pronoun used in the quotation clause of reported speech constructions refers to a main clause referent, i.e. encodes the original speaker. The use of simple reflexive pronouns in logophoric function is attested in many Daghestanian languages, cf. [Testelets, Toldova 1998; Daniel 2015; Polinsky 2015; Forker 2019], etc. In the typological literature the use of reflexive pronouns in reported speech in logophoric function is considered as a diagnostic of semi-direct speech, a third type of reported speech positioned between direct and indirect speech. Semi-direct speech, which is found only with third person speech subjects (‘s/he said that…’), involves incomplete person shift and the mixture of properties of direct and indirect speech [Aikhenvald 2008].

The simple reflexive pronoun encodes the original speaker and never the addressee, whereas the intensified reflexive pronoun can encode participants both inside and outside clause boundaries. The following elicited examples illustrate the use of simple and intensified reflexives respectively.

(33) a. Murad-i_i harzi j-oː-jo Muhamad-qa_j hinƛ’a_i/*j jak’o j-oq-ala꞊ƛo

Murad-erg ask iv-do-pst Muhamad-poss refl.sup heart(iv) iv-come-opt꞊quot

niso-na.

say-cvb

‘Murad_i asked Muhamad_j to trust him_i/*j.’

1. Murad-i_i harzi j-oː-jo Muhamad-qa_j hinƛ’a-zu_j/i jak’o j-oq-ala꞊ƛo

Murad-erg ask iv-do-pst Muhamad-poss refl.sup-foc heart(iv) iv-come-opt꞊quot

niso-na.

say-cvb

‘Murad_i asked Muhamad_j to trust him_i/himself_j.’

In reported speech personal reflexive pronouns function similarly to the third person pronouns, i.e. encode the original speaker. In the following example the personal reflexive dozu is coreferential to the main clause referent p’awleli.

(34) Do dil j-üⁿq-äl žo j-iⁿzcalaʔ, do-zu čaqallil m-üq-äl

1sg.erg 1sg.lat iv-eat-inf thing(iv) iv-search.sim 1sg.abs-foc jackal.lat iii-eat-inf

m-iqo-jo꞊ƛo niso-na pikro b-oh-na gej p’awle-li.

iii-get-pst꞊quot say-cvb thought(iii) iii-do-cvb cop cat-erg

‘The cat_i said, “While I_i was looking for something to eat, I myself_i became a meal for the jackal.”’

Sentence (35) exemplifies the use of deictic and anaphoric pronouns: the simple reflexive pronoun žü and the first person pronoun dijela both encode the original speaker abo ‘father’. In instances when the indexicals and reflexives co-occur, this reading is the only possibility.

(35) abo niso-s öždäː-qä, žü Ø-uʁo-na müʁättä,

father.erg say-prs son.pl.obl-poss.ess refl i-die-cvb after

dijela šünäƛ’ä b-eče꞊ƛo niso-na.

1sg.gen2 grave.sup.ess hpl-stay.imp꞊quot say-cvb

‘The father_i said to his sons that after he_i dies, you each stay near my_i grave.’

Example (36) illustrates the use of the simple reflexive, the first person pronoun, and multiple uses of the quotative particle, as well as the use of the deictic adverb. The deictic element hoƛoʔ ‘here’ in this example is interpreted from the original speaker’s perspective and refers to the location in which the original utterance took place, viz. Turkey; the first person pronoun dije is also used from the perspective of the original speaker. The simple reflexive pronoun žü encodes the original speaker, who is referred to in the matrix clause by the third person pronoun holco.

(36) holco nisos, dije ijo-abo zu<wa>qʼo-jo꞊ƛo daʁistanɬilaʔas žü

3sg.erg say.prs 1sg.gen1 mother-father be<pl>-pst꞊quot Daghestan.in.abl refl

Ø-aq-na꞊ƛo hoƛoʔ, ijo-abo꞊na b-u<wa>ʁo-na Ø-uqʼɬi-na

i-happen-cvb꞊quot here.in.ess mother-father꞊add hpl-die<pl>-cvb i-grow-cvb

gej꞊ƛo.

cop꞊quot

‘He says, “My parents were from Daghestan, and I was born here (in Turkey), and when my parents died, I grew up alone.”’

Thus, the simple reflexive pronoun is used in reported speech to refer back to the speaker of the original speech situation only, conveying the same reference relations as a first person pronoun. That is, the simple reflexive pronoun and the 1^st person pronoun are used as alternatives encoding the original speaker.

3.3. Deictic elements in reported speech

In Bezhta reported speech there is usually no spatial and temporal shift of the kind required in canonical direct and indirect speech constructions in English and elsewhere; instead, the interpretation of adjuncts in reported speech depends on the interpretation of indexical pronouns. Consider the place adverbs hoƛoʔ (proximal) and hugidaː (distal), which are derived from the corresponding proximal and distal demonstrative pronouns [Comrie et al. 2015b: 229]. Bezhta demonstrative pronouns show a three-way spatial distinction: close to the speaker, close to the hearer, and far from the speaker and the hearer. The proximal adverb hoƛoʔ is derived from the demonstrative hoƛo (close to the speaker), and the distal adverb hugidaː is derived from hugi (far from the speaker and the hearer). The reference point of these place adverbs is still ambiguous in (37a), as both adverbs can take the position of either the current or the original speaker. The ambiguity can be resolved, for example, when the long-distance reflexive is used in the reported clause. In this case, the only possible reading of the place adverbs is with indexical shift (37b).

(37) a. abo niso-s do zuq’o-jo hoƛoʔ / hugidaː.

father.erg say-prs 1sg.abs be-pst here there

‘The father_i says, “I_i/j was here/there.”’

1. abo niso-s žü zuq’o-jo hoƛoʔ / hugida:.

father.erg say-prs refl be-pst here there

‘The father_i says that he_i/*j was here/there.’

4. Quotative, reportative and other functions of the enclitic ƛo

The enclitic ƛo has several functions. Its main function is to mark quotation in the context of speech verbs, propositional attitude verbs (e.g. urʁizi -aq- ‘think’, pikro boː- ‘think’), a few verbs of knowledge and acquisition of knowledge (e.g. -iq’- ‘learn, know’), and fearing verbs (hic’o -ek- ‘fear’).

Another of its functions is reportative. It is used to mark hearsay evidential value in narration, something often used in storytelling. Interestingly, in the West Tsezic languages the reportative and quotative enclitics are distinct (cf. Hinuq reportative eƛ, quotative ƛen; Khwarshi reportative ƛo, quotative ƛun/ƛin; Tsez reportative ƛax, quotative ƛin), whereas in the East Tsezic the same enclitic is used in both quotative and reportative functions (Bezhta ƛo, Hunzib ƛe) [Khalilova 2011].

The system of evidentiality in the Tsezic languages shows a contrast between witnessed and unwitnessed events, which is expressed only in the past tense [Comrie, Polinsky 2007; Forker 2018; Khalilova 2011]. In the West Tsezic languages there is an equipollent evidential contrast between past witnessed and past unwitnessed forms. In the East Tsezic languages including Bezhta, meanwhile, the contrast is based on the distinction between the preterite (also called aorist), a morphologically simple past tense, and the perfect, which is a compound tense based on the perfective converb and the present tense copula. The witnessed/unwitnessed contrast is based on the distinction between visual and non-visual sources of information. In addition to these basic functions, the opposition of witnessed and unwitnessed forms has extended uses, first mentioned for Tsez in [Comrie, Polinsky 2007: 344]: “⟨...⟩ the contrast is used to report direct/indirect access to emotions, sensations, or beliefs. In this usage, past witnessed is normally associated with the first person, because it is the speaker who has direct access to his/her own feelings or knowledge. Past unwitnessed is used otherwise. This usage ⟨...⟩ is functionally similar to that of the historic present”. It is found in narratives, where the past witnessed regularly refers to events which the narrator did not in fact witness. In the following example (38), the enclitic ƛo functioning as a hearsay evidential is added to the finite verb in the past witnessed tense. The use of hearsay with a direct evidential is another way to express unwitnessed evidential value in Bezhta [Ibid.].

(38) a. aqo j-üⁿxöł-ijo꞊ƛo kotakalda biχzi j-aq-al.

wife(ii) ii-begin-pst꞊rep very be.anxious ii-become-inf

‘The wife started to become very anxious.’

1. sud꞊ƛo niso-da äli-s-so nako-wa j-ä<ä>ƛʼel-ca

why꞊quot say-cond Ali-gen1-attr melody-pl nhpl-hit<pl>-prs.ptcp

zu<wa>qʼo-na gej.

be<pl>-cvb cop

‘If you ask why, (it’s because) these were melodies played by Ali.’

(39) holło gähijo jal j-eže-jo꞊ƛo.

3sg.erg be.pst.ptcp hay(iv) iv-take-pst꞊quot

‘She took all the hay.’ [they say]

Since the quotative and the reportative (also called hearsay evidential) coincide in Bezhta, it is sometimes hard to tell whether we are dealing with the use of multiple quotative particles, or of one quotative and one reportative enclitic. In example (40), the enclitic ƛo occurs twice, in the middle and at the end of the quote; the latter definitely demonstrates the quotative usage, as it marks the end of the quote, but the former may indicate hearsay evidential value.

(40) holco niso-s qʼac’cʼo b-ow-al꞊ƛo ömrö, žensalaʁa müʁättäː

3sg.erg say-prs together iii-do-inf꞊quot/rep life(iii) today.cmpr back

mi dije Ø-uqʼo is gej꞊ƛo.

2sg.abs 1sg.gen1 i-big brother(i) cop꞊quot

‘He says, “Let’s live together, you are like my older brother nowadays.”’

In example (41) the enclitic ƛo, in its quotative function, is followed by the speech verb niso- ‘say’ in the embedded clause, while in the main clause the finite verb is marked with the enclitic ƛo registering the hearsay function.

(41) were tʼutʼ-aːqʼa꞊ƛo niso-na iƛe-na Ø-eⁿƛʼe-jo꞊ƛo hugi hoglodoj.

hey shoot-proh꞊quot say-cvb call-cvb i-go-pst꞊rep 3sg.abs 3pl.apud

‘Shouting, “Wait, do not shoot”, he went towards them.’

The enclitic ƛo can even attach to the speech verb niso- ‘say’, indicating a hearsay usage.

(42) niso-l-lo꞊ƛo hoglo is-naː-qa haːⁿ sijo-d j-aq-ijo,

say-caus-pst꞊rep 3pl.obl brother-obl.pl-poss hey what-q iv-happen-pst

sud j-aq-ijo꞊ƛo niso-na?

why iv-happen-pst꞊quot say-cvb

‘They asked those brothers, “What happened? Why did it happen?”’

The same element is obligatory in naming constructions, e.g. his name is…, it is called…, to name, and similar contexts:

(43) holloƛʼa äčʼäräl꞊ƛo niso-na caːⁿ gul-ca.

3pl.sup Adjara.people꞊quot say-cvb name put-prs

‘They are called the Adjara people.’

(44) holcos caːⁿ zuqʼo-jo Mähämäd꞊ƛo niso-na.

3sg.gen1 name be-pst Muhamad꞊quot say-cvb

‘His name was Muhammad.’

Enclitic ƛo is obligatorily used to introduce onomatopoeia.³

(45) χemuχa: b-oχ-eʔe-š zaƛi b-ek-il-lo χöχöcöj

root.pl iii-hold-neg-pst.ptcp wind.erg iii-fall-caus-pst.ptcp tree(iii).cmpr

dapʼ꞊ƛo qʼeː Ø-ek-ca tʼahir.

dap꞊quot down i-fall-prs tagir

‘Like a tree without roots, which the wind made fall, Tagir fell with a crash.’

Bezhta uses secondary onomatopoeias, which are words derived from primary onomatopoeic elements referring to sounds. For example, Bezhta possesses a class of onomatopoeic verbs, e.g., ʁäʔƛö- ‘caw’, hicƛo- ‘sneeze’, öhƛö- ‘cough’, hik’ƛo- ‘hiccup’, hahƛo- ‘yawn’, which have a single S_A argument in the Ergative and have arisen diachronically through the incorporation of an onomatopoeic element into the verb iƛe- ‘call’ [Comrie at el. 2015a: 542]. Such verbs are based on onomatopoeic expressions covering natural sounds, i.e. those produced by humans and animals, whereas onomatopoeic expressions covering sounds made by artifacts and objects (as in the examples above) do not form unergative verbs. This is easily explained, as the speech verb iƛe- ‘call’ allows only animate agents.

Identifying the grammaticalization source of the Bezhta quotative enclitic is not straightforward. All Tsezic languages have a cognate quotative particle, which comes from the grammaticalization of the perfective converb of the verb ‘say’. In the closely related Tsezic language Khwarshi, the quotative enclitic ƛun/ƛin clearly derives from grammaticalization of the perfective converb of the verb iƛ- ‘say’, iƛin/iƛun ‘having said’ [Khalilova 2011; 2021]. The Bezhta quotative ƛo might have originated from the verb iƛe- ‘call’, though the perfective form of ‘call’ iƛena and the aorist form iƛejo do not look to be good candidates for reduction. In Hunzib, Bezhta’s sister language, the quotative ƛe is presumably a shortened form of the stem of the verb iƛe-, which in Hunzib means ‘call’ just as in Bezhta, but whose cognate in some other Tsezic languages means ‘say’. Additionally, Hunzib has an alternative quotative ƛen, whose grammaticalization source is unclear: as van den Berg [1995: 134] notes, “it is difficult to say whether ƛe.n is made up of the quotation particle and the coordination particle (and), a petrified verbal form with a past gerund ending, or the quotation particle and the incorporated first consonant of nɨsən, with which ƛe is usually found”. The grammaticalization source of the reportative (hearsay) enclitic in West Tsezic is still to be investigated.

Another enclitic used in reporting is the interrogative enclitic ƛona, which is employed in embedded clauses. It is morphologically complex, as it consists of the quotative enclitic ƛo and the additive enclitic na. The enclitic ƛona has additional (epistemic) modal meanings of doubt, possibility, or hesitation, and is often used in the interrogative sense. The interrogative ƛona with verbs of speech indicates that the reported clause is alleged, i.e. the reported statement is false. The embedded interrogative with verbs of speech can have only the unshifted interpretation, i.e. the embedded indexicals can only refer to the current speaker.

(46) Murad-i niso-s do armija-li-ʔ Ø-eⁿƛ’e-na꞊ƛona.

Murad-erg say-prs 1sg.abs army-obl-in i-go-pst꞊interr

‘Murad_i says as if I_j/*i went to the army.’

The embedded interrogative ƛona also occurs with the other verbs, e.g. verbs of thought: pikro b-aq- ‘think’, urʁizi -aq- ‘think’, kʼocʼola- ‘think’.

(47) hugi öžö urʁizi Ø-aq-ca, naː-d že holo b-aq-ijo,

that boy(i) think i-become-prs where-q now that iii-happen-pst

m-eƛ’e-jo hogo ƛaplaʁe꞊ƛona.

iii-go-pst that frog(iii)꞊interr

‘The boy was wondering where now the frog had gone.’

(48) hollo-s pikro zuqʼo-na, žüwöːl nitona-zu wahliʔ b-aːq-ijaː

3pl.obl-gen1 thought be-cvb refl.pl always-foc there hpl-become.pl-pst.ptcp.pl

gej꞊ƛona.

cop꞊interr

‘They thought they were always there.’

5. Conclusion

Reported speech constructions are traditionally described as part of complementation. Complementation in Bezhta — like other kinds of subordination, including adverbial and relative constructions — involves the use of non-finite verb forms in the embedded clause [Comrie et al. 2016]. Unlike standard complementation, Bezhta reported speech constructions involve finite embedded clauses. Reported speech complementation shows a weak degree of syntactic subordination: for example, it can be based on mere juxtaposition of the main and the embedded reported clause, or the reported speech can be used in an independent declarative clause, i.e. without the matrix speech predicate.

The reported speech construction is thus definitely a distinct type of complex structure, as it shows some similarities with complementation but also differs from it in certain respects. Like some complement-taking verbs, verbs of speech use several complementation strategies. For verbs of speech, these strategies are finite complementation (Sections 2.1–2.3), non-finite complementation based on nominalization (Section 2.4) and insubordinate complementation (Section 3.1). The behavior of the reflexive pronouns in the reported speech construction is the same as in other complement clauses. In complex clauses and in reported speech constructions, simple reflexive pronouns are used to refer to a main clause referent. The simple 3^rd person reflexive pronouns, which are used in logophoric function, encode the original speaker (Section 3.2).

The most distinctive feature of the reported speech construction is the use of finite predicates in embedded complement clauses. As mentioned above, reported speech is introduced not only by utterance predicates but also by some other predicates, including propositional attitude, direct perception and achievement verbs. Clauses embedded under these complement-taking predicates are finite as in reported speech. In the reported speech construction, the quotative enclitic regularly follows the embedded clause. However, the quotative enclitic does not function as a complementizer, since embedded finite clauses also occur without a quotation enclitic, cf. Section 2.1. The quotative enclitic is not a marker of direct speech either, but a marker of quotation in general.

The most common way of expressing reported speech in Bezhta is by means of the construction where the framing clause involves a verb of speech (utterance verb) and the reported utterance is followed by the quotative particle ƛo. It is also common for the reported speech marked with the quotative ƛo to be followed by the perfective converb of niso- ‘say’ along with a verb of speech in the framing clause. Reported speech may occur without a framing clause. Reporting in Bezhta can be introduced by predicates other than speech verbs. There are no restrictions on the position of the framing clause, i.e. it may precede, follow or interrupt the quote. Vocatives, interjections, questions, and non-indicative forms like the imperative, prohibitive, and optative are reported as in ordinary main clauses.

Bezhta reported speech allows indexical shift, i.e. any embedded indexicals can reflect the point of view of either the original speaker or the current speaker (Section 3.1). Indexical shift is common only in finite complement clauses. In non-finite complementation, only the unshifted reading is possible. The shifted interpretation becomes obligatory with embedded reflexive pronouns. The simple 3^rd person reflexive pronouns function as logophoric pronouns in reported speech, encoding the original speaker. That is, the logophoric pronouns in reported speech behave just like first person pronouns as they both refer to the original speaker.

The quotative marker in Bezhta has multiple functions: it is used to mark quotation in reported speech constructions, but also hearsay evidential value. In the closely related West Tsezic languages, the quotative enclitic is a grammaticalized form of the perfective converb of the speech verb ‘say’ [Khalilova 2009: 238]. However, the origin of the Bezhta quotative ƛo remains unclear.

¹ In the collection of Bezhta texts, no instances are found of embedded first-person pronouns within the nominalized complements of verbs of speech. However, elicited examples such as (24a) are perfectly fine.

² The 3^rd person simple reflexive pronoun is žü ‘refl’ in the singular and žüwö:l in the plural. The oblique forms are (suppletive) hini in the singular and hide in the plural. There are two types of reduplicated reflexive pronoun. One is based on the reduplication of an oblique form of the simple reflexive: the first part is an invariable form in the Genitive, hinis (sg) / hides (pl), and the second part is oblique hini/hide in the appropriate cases. The other reduplicated reflexive pronoun consists of an oblique form of the simple reflexive in the appropriate case followed by the simple reflexive in the Absolutive form, which varies for number only. The intensified reflexive is based on the simple reflexive in its oblique form hini/hide combined with the focus particle -zu. The simple first and second person reflexives are formed by attaching the focus particle -zu to the personal pronouns.

³ Onomatopoeic expressions in Bezhta cover two main semantic fields: natural sounds and artifacts. Natural sounds are divided into those made by 1) humans: hik’ ‘hiccups’, q’ur ‘rumbling in the stomach’, χur ‘running nose’, etc.’; and 2) animals: wic’ ‘squeak of a mouse’, χurr ‘purr of a cat’, wič ‘twittering of small birds’, etc. Artifacts are sounds of striking and knocking, e.g. 1) sounds produced by motion: dap’ ‘sound of falling, thud’, q’ip’-q’op’ ‘stomping’, qoc- ‘sound of heels, trot’, and 2) others: č’iq’-č’iq’ ‘sound of cutting with scissors’, piq’, poq’ ‘sound of boiling corn porridge’, č’iq’ ‘click’, diq, däq ‘knock’, q’ip, q’öp’ ‘knock’, etc.

Авторлар туралы

Zaira Khalilova

Хат алмасуға жауапты Автор.
Email: zaira.khalilova@gmail.com
Ресей, Moscow

Әдебиет тізімі