(36) w?r ?Мssen-ж? [Ш-os?-\dd wжl?м]

Neg know.PerfN-lSgS [3MaSgS-come.PerfP-\Centrip or]

`I don't know whether he came.' [Heath 2011: 660]

In Udihe (Northern > Tungusic), a cliticalized disjunctive particle is used the same way.

According to the current data, disjunctive markers may be used as subordinators in embedded question construction only for polar interrogatives. However, unlike other subordinators, there is no complex sentence polysemy is attested for disjunctive ones. Also, disjunctive constructions are rarely allowable in independent questions. The only exception is disjunctive suffix -As(i), which is attested in both root (37) and indirect (38) questions.

(37) Iwana sin-du kusige-we bu-ge

Ivan you-DAT knife-ACC give-PERF

e-si-ni bu-o:s?

NEG-PAST-3SG give-DIS

`Did Ivan give you the knife or not?' [Nikolaeva & Tolskaya 2011: 812]

(38) Nua-ni xauntasi-e-ni Sonia skola-du

he-3SG ask-PAST-3SG Sonia school-DAT

bi:-we-n-de e-i-we-n-de.

be.PRP-ACC-3SG-DIS NEG-PRP-ACC-3SG-DIS

`He asked whether Sonia was at school or not.' [Nikolaeva & Tolskaya 2011: 50]

That facts let me suppose that disjunctive markers are specific for indirect question constructions in case they are used in a complex sentence as a subordinator. That feature differs disjunctive markers from other strategies, that may be either not specific for indirect questions or have synchronous polysemy in polypredicative constructions.

4.5 “Null” constructions

“Null” strategy is an indirect question construction that does not involve any specific interrogative markers. In such root questions, interrogative semantics is defined contextually or intonationally, e.g. in Russian (Slavic > Indo-European) independent polar question:

(39) Ты хочешь что-то сказать?

As for embedded questions, a direct quotation may be distinguished from indirect question construction via the pronominal deixis shift, e.g. cf. English (Germanic > Indo-European) direct quotation in (40) and embedded question in (41). Consider that in indirect question pronoun you turns into I:

(40) He asks me: “What do you want?”

(41) He asks me what I want.

Additionally, I prefer to divide “null” construction into two subtypes: “true null” constructions and wh-word questions. In “true null” ones interrogative semantics of a complement clause may not be explicitly expressed in any way syntactically except for deixis. Sometimes it can be accompanied with specific word order. Such construction is possible in Tundra Nenets (Samoyedic > Uralic) embedded polar questions (42):

(42) tыt?°=w°h n?i=w°h m?n?° yexara?-d°m

come.FUT-DUB NEG-DUB I ignore-1SG

`I don't know if he is coming or not.' [Nikolaeva 2014: 308]

In wh-word constructions, an interrogative word is the only question marker except for deixis shifts or word order. Consider an example from Neverver (Malayo-Polynesian > Austronesian), where wh-word abi is the only question marker (43):

(43) I-sus-ikh adr i-ver ar amt-uv

3REAL:SG-ask-APPL 3NSG 3REAL:SG-say 3NSG 3IRR:PL-go

abi.

where

`He asked them where they were going.' [Barbour 2012: 388]

Trivially, wh-word construction is possible only for constituent embedded questions, while “true null” strategy is allowable in alternative and polar ones. “Null” constructions are rather spread: “true null” is less frequent (2 out of 78 constructions), wh-word one is more (17 out of 78), that meets expectations.

Also, it is possible to formulate some typological generalizations:

- if a language allows using of “null” embedded polar question construction, then wh-word construction is also possible for constituent questions. E.g. Tundra Nenets works that way. The opposite conclusion is not true, e.g. Saamбka encodes embedded constituent questions with wh-word construction (44), but in indirect polar ones, only subordinator йe is used (45). There are 15 languages (consequently, the majority of languages that allow “null” constructions) that work the same way.

(44) Mi sбbi amb?м da i.

1S know who be 2S

`I know who you are.' [McWhorter 2012: 176]

(45) M?м sбbi йe jу sб kбi mi amг?jг?

1S.NEG know if 2S.FUT can call 1S tomorrow

`I don't know if you can call me tomorrow.' [McWhorter 2012: 75]

- if a language allows “null” construction in embedded question, it always allows “null” construction in a corresponding type of independent question. E.g. Hebrew allows wh-word construction in both embedded (46) and independent (47) questions. It can be noticed, that complement clause of the indirect question completely copies root question construction:

(46) sha'бlti matay ze met

I-asked when it died

`I asked when it died.' [Glinert 2004: 276]

(47) matay ze met?

when it died

`When did it died?' [Glinert 2004: 276]

4.6 Specific verbal constructions

Specific verbal constructions are rather spread in a given data: there are 9 entries out of 78. However, grammar authors usually describe special verbal constructions as an alternative question, although point out that such constructions invoke polar semantics. This means that, unlike true alternative questions, they can be answered by “yes” or “no”. Also, it sometimes occurs that in a given language a polar question may be expressed only with an alternative construction. However, in the current research, I considered such cases as specific verbal construction.

In such constructions, various means of question particles are used. Consider the following table:



Table 6. Question markers types in special verbal constructions

Question marker type	Constructions
“null”	1
question particles	2
subordinators	3
quotatives	0
disjunctive partiсles	3

Table 6 shows that special verbal constructions are possible within any strategy (I prefer not to take quotative one into account as data is insufficient) with comparably equal frequency.

“Null” strategy (so-called disjunctive-negative constructions) is almost not attested in the current data. It can be explained with the fact, that “null” strategy, as it was mentioned above according to the current data, is not frequent for embedded polar questions at all out if specific verbal constructions.

The following special verbal construction patterns are attested:

- V not V;

Such pattern is defined as disjunctive-negative construction. It does not involve any question markers of verb form, but head verb phrase is duplicated without any changes, consider the following example from Mandarin Chinese (Sinitic > Sino-Tibetan) (48):

(48) tв wиn wo tв mвmв mбng bщ mбng?

he ask I his mother busy not busy

`He asks me, whether his mother is busy'.

- V-Q negV-Q;

The pattern involves verb-phrase duplication in the negative form with the copying of the features of the main verb. For example, such strategy is considered is the most frequent mean of indirect polar question marking in Chuvash (Oghur > Turkic) [Egorova 2020]. There either question clitic =i or =?i may be used in a special verbal construction:

(49) ma?? kil-ed=i kil-mest=i te-ze

Masha come-npst[3sg]=q1 come-neg.npst[3sg]=q1 QUOT-cv_sim

va?? it-r?-л.

Vasya asked-pst-3sg

`Vasya asked, whether Masha would come.' [Egorova 2020]

The pattern is the most frequent in the given data: 6 out of 9 instances are constracted like this.

- V-Q neg.AUX-Q;

In such cases, instead of verb phrase duplication, a negative auxiliary is used. The auxiliary copies the features of main verbal phrase in a clause. So, if a verb contains a question marker, it is also copied. Cf. an instance from Tundra Nenets (Samoyedic > Uralic) (50). Here is a “null” strategy used, so there no specific question markers, however, all other features (form and person inflexion) of a verbal form are repeated in the negative auxiliary:

(50) ti-m xada-wer° n?i-wer°

reindeer-ACC kill-INFR-2SG > SG.OBJ NEG-INFR-2SG > SG.OBJ

yexara?-d°m

ignore-1SG

`I don't know whether you killed a reindeer or not.' [Nikolaeva 2014: 308]

However, it is also possible with other strategies. E.g. consider another Tundra Nenets example (51), whether a conditional verbal form is used as an indirect question subordinator. It is copied to negative auxiliary as well:

(51) m?n?° yexara?-d°m tыt°-b?-ta=w°h

I ignore-1SG come.FUT-COND-3SG-DUB

(n?i-b°-ta=w°h)

NEG-COND-3SG-DUB

`I don't know if he is coming or not.' [Nikolaeva 2014: 308]

- V-Q PRO.V-Q

Also, a case of proverb usage was attested, cf. an example (52) from Udighe (Northern > Tungusic):

(52) Nua-ni xauntasi-e-ni Sonia skola-du

he-3SG ask-PAST-3SG Sonia school-DAT

bi:-we-n-de e-i-we-n-de.

be.PRP-ACC-3SG-DIS NEG-PRP-ACC-3SG-DIS

`He asked whether Sonia was at school or not.' [Nikolaeva & Tolskaya 2011: 50]

The patterns mentioned above may not be exhaustive, so further typological investigations are needed to define the whole list of possible options.

There are some typological conclusions, that may be extracted from the current data:

- specific verbal construction is possible only for polar questions;

- specific verbal constructions are more frequent in embedded question constructions rather than in direct ones. There are 5 cases out of 9, when the construction may be used in root questions as well;

- if a language allows special verbal construction in independent question, then it also possible in indirect ones. For instance, consider the following examples from Udihe, where a particle -nu may be used in both root (53a) and embedded (53b) questions. The examples are translated by the author as alternative questions:

(53a) Xeleba bie-nu anci-nu ?

bread be.PRES.HAB-CONT no-CONT

`Is there bread or not?' [Nikolaeva & Tolskaya 2011: 812]

(53b) Memi ise:-ni tege bie-nu anci-nu.

REF.ACC see.PAST-3SG gown be.PRES.HAB-DIS no-DIS

`She looked at herself to see whether or not she was wearing a gown.' [Nikolaeva & Tolskaya 2011: 817]

- oppositely, if a language allows special verbal construction independent question, then it is not necessarily possible in root ones;

- special verbal construction is rarely the only means of expressing embedded polar questions (e.g., it is so in Chinese). More often it is either combined with other strategies (e.g. Udihe) or a duplicate negative/verbal phrase may be omitted, so that will be regular embedded polar construction with a given strategy. For instance, Tundra Nenets works the latter way, consider the example (51) above, where negative auxiliary may be deleted.

4.7 Word order in embedded questions

The word order in embedded questions seems to be a stand-alone issue as there are various parameters that should be considered:

- whether the word order in embedded questions differs from the independent question;

- are there any parameters that may affect the position of the embedded clause regarding the main clause;

- are there any parameters that may affect word order and which way (e.g. focus);

In the current research, there is not enough data for typological investigation as the information on word order in indirect question constructions are not covered. Usually, there is no possibility to make any autonomous conclusions in view of the low number of diverse examples. So, I prefer to leave the issue of the word order to further research from various language sources.

However, there are some preliminary observations that need checking on the wider sample:

- languages comparably rare have a specific word order for indirect question construction, i.e. most often it is the same as in independent questions. In case question word order does not differ from independent declaratives, a specific role of intonation is pointed out.

- however, the opposite cases are attested, cf. English (Germanic > Indo-European) word order in independent (55) and indirect (56) questions. In root questions, auxiliary fronting is required, but on embedded it is not the same. The word order in embedded question resembles the one in independent declarative (54):

(54) You love me.

(55) Do you love me?

(56) I don't know whether you love me.

However, the assumption, that word order in indirect questions in English corresponds to the one in independent declaratives is wrong. Consider the following examples of direct (58) and indirect (59) constituent interrogatives:

(57) You are a linguist.

(58) Who are you?

(59) I know who you are.

Surely, in wh-question, the word order is not similar to neither independent declarative (57) nor root question. Unlike root interrogative, embedded question lacks auxiliary fronting, but in contrary to independent declarative, wh-word fronting remains. So, English indirect question word order is specific for indirect question constructions.

- situational changes of word order do not relate to features of indirect questions. E.g., word order may be changed in view of information structure see section 4.8 for more details).

4.8 The information structure of embedded questions

In the current data, there is no evidence, that specific means of focus or topic marking for indirect question constructions do exist. Usually, the means of expression of information structure in indirect question are shares with root interrogatives. However, the data is insufficient for definitive conclusions as information on focus in embedded interrogatives usually not covered in grammars, or examples are not provided.

All in all, specific information structure may be expressed indirect polar question constructions in the following ways:

- intonation;

The parameter is left out of the scope of the current research as there is no possibility to measure intonational contour based on corpora and grammatical descriptions. However, it is sometimes mentioned in grammars as the only means of information structure marking. That is usually true for “null”-strategies. E.g., consider English (Germanic > Indo-European) examples with wh-word strategy, where focalized constituents are stressed intonationally:

(60) I don't know, whether you love me [so I would like to ask you about it].

(61) I don't know, whether YOU love me [but my mother surely does].

- word order change;

Commonly, the informational structure is expressed with a specific word order: in case expression is unmarked, the word order is standart, otherwise, focus- or topic-fronting is used.

For instance, in Mosetйn (Moseten-Chonan), focalized constituents are moved to clause-initial position [Sakel 2011: 433]. In Basque (Isolate) embedded questions, foci occur strictly in preverbal position: in constituent ones, the wh-word and the verb are preferably moved to the clause-initial position, while in polar questions are mostly signalled by intonation pattern only and verb fronting can optionally be used [Hualde 2011]. On the contrary, in Lango (Nilotic > Nilo-Saharan), clause-initial position is interpreted as a topic [Noonan 2011: 172].

- change of question marker position;

Change of question marker position is considered in case a focus particle is used in embedded interrogative constructions. For instance, in Ket (Northern Yeniseian > Denй-Yeniseian), focus particle ы is used for direct and indirect polar question marking. In a neutral case, it is posed right before the verbal phrase. e.g.:

(62) bы mana ??tn ы

3SG she.said 1PL QUEST

{i}8-a?a7-k5-s4-aq0-an-1

18-to.forest7-TH5-NPST4-go.MOM0-AN.PL-1

`She asked if we are going to the forest.' [Nefedov 2015: 62]

Otherwise, it is used right before the constituent the speaker wants to question (strangely, in the translation given in the following example does not supports the author's claim, but still):

(63) вd ы вm-an bоn si7-ba6-t5-o4-n2-oq0

1SG QUEST mother-CAR MIR R7-1SG6-TH5-PST4-PST2-become.PST0

`Was I (really) born without a mother?' [Nefedov 2015: 62]

- attachment of focus/emphatic particles to focalized constituents;

Also special focus particles may be used in embedded interrogatives along with question markers. E.g., in Jamsay (Dogon > Niger-Congo), focus clitic =?n is attached to a focalized constituent, e.g.:

(64) [[... мnи kу jщn?м-?? fъ] k?М:-r?м]

[[... person.L NonhO know.Impf-Ppl.Sg all] be.Nonh-Neg]

wб [?Мn?м=?n wу w?М:-? mб? kв:n]

say [Logo=Foc NonhO be.Hum-Ppl.Sg Q also]

“...there will be nobody who knows,” he said, “whether (=that) it was I [focus] who killed you-Sg…” [Heath 2008: 476]

The means of informational structure expressing in polar questions may be mixed. For instance, in Russian (Slavic > Indo-European), focus particle li, used in indirect polar questions, works the same way as in Ket: if en expression is unmarked, li is attached to the verbal phrase (65), otherwise it attaches to focalized constituent (66). However, unlike Ket, in Russian focussed constituent is obligatory clefted to the clause-initial position. So, in unmarked case, the verbal constituent is fronted:

(65) Я не знаю, пришла ли мама [когда узнаю -- скажу].

(66) Я не знаю, мама ли пришла [может быть, это папа пришел].

As for information structure in constituent question, there is no evidence, that focalization may affect on indirect question construction in any way.

4.9 Main verb semantic type

According to the current data, the main verb semantic type is relevant for the choice of the strategy of indirect question marking. Two semantic groups were taken into account: speech verbs (e.g. to say, to ask) and mental verbs (e.g. to know, to hear, to see). There are 6 constructions out of 78, where a type of construction is chosen regarding the main clause verb meaning. However, I suppose, that the influence is more frequent than it was attested as for some languages the data is not completely full, because grammars do not cover information in various semantic types.

In the current data, there are two types of diversification of strategies were attested:

- indirect question constructions with speech verbs are expressed as a direct quotation, while mental verbs require special question marking. For example, in Tamashek (Berber > Afro-Asiatic), speech verbs require a direct quotation. Cf. example (67), where root polar question marker ?мk highlites direct quotation usage:

(67) ??мs i-nna-\?-s ?мk t-?ss?мn-жd

only 3MaSgS-say.PerfP-\Dat-3Ss yes/no? 2S-know.Reslt-2SgS

s ъlli t-н-dж? t-?d?n-жd

that goats Fe-Dem.Pl-Anaph 2S-tend.Reslt-2SgS

t-м-nin.

Fe-Dem.Pl-1 SgPoss

`Well', he said (=asked), `do you-Sg know that these goats that you have been tending, they are mine?' [Heath 2011: 713]

However, indirect question constructions with mental verbs require a disjunctive strategy with particle wжlб:

(68) w?r ?Мssen-ж? [Ш-os?-\dd wжl?м]

Neg know.PerfN-lSgS [3MaSgS-come.PerfP-\Centrip or]

`I don't know whether (or not) he came.' [Heath 2011: 660]

Moreover, there are three indirect question strategies are attested in Tamashek, and none of them, according to available language data, may be used with speech verbs.

- more frequently, mental verbs require specific complementation strategy, which is impossible for speech verbs. In such cases, indirect questions with speech verbs are expressed via a different construction.

For instance, in Bardi (Nyulnyulan), gerund marker ma- is used as a basic complementation strategy for verbs of emotion and perception [Bowern 2013: 671]. Consider the following examples of ma- gerund usage in regular complementation (70) and in indirect constituent question with the main verb `know' (69):

(69) Boonyja-nim aamba oorany arr irr-moonggoon ginyinggi Rooji

all-ERG men women NEG 3A-know 3MIN R.

liyan-nga i-na-ma-na Marroolal m-onja-n-ngan.

want 3-TR-put-REM.PST M. GER-spear-CONT-ALL

`All the people didn't know that Rooji wanted to spear Marroolal.' [Bowern 2013: 527]

(70) Ni-moonggoon=irr ma-nya-n-ngan aarli.

3-know=3A.DO GER-catch-CONT-ALL fish

`They know how to catch fish.' [Bowern 2013: 672]

Ma- complementizer is one of three possible indirect question constructions in Bardi. It is available for both polar and constituent indirect questions. However, two other ones do not have restrictions for the semantic type of the main clause verbs. For instance, consider the example with “null” wh-word construction in constituent indirect question:

(71) Nganjal irrgoordoo lalin i-na-ng-gala-n-an

INTERROG how year 3-TR-PST-wander-CONT-REM.PST

Iwany-i.

Sunday.Island-TOP.LOC

`We don't know how many years he stayed on Sunday Island.' [Bowern 2013: 242]

Nonetheless, it is important to notice, that such constructions as ma-V in Bardi need separate consideration as they are not equal to regular indirect question markers. Spot the difference between examples (72a) and (72b):

(72b) Я знаю, как готовить овсяную кашу.

(72b) Я знаю, как она готовит овсяную кашу.

In constructions like (72a) special future-oriented verbal forms are used (e.g. infinitive, irrealis, potentialis), that are impossible for indirect questions like (72b). I assume that in further research such constructions should me considered as a special type not equal to regular indirect question constructions.

- however, there is single evidence that speech-verbs specific embedded question construction without direct quotation also exist. In Basque (Isolate), a construction ba al V-(e)n (ba is a conditional particle, al is a question particle, V-(e)n is a relativizer), used in Guipuzcoan dialect only, is possible only with verbs meaning `ask' [de Rijk 2018: 442]:

(73) Erregeak zerbait egiten ba al zekien

king.ERG something do.lMPF ba Q knew.COMP

galdetu zion.

ask AUX

`The king asked him whether he knew [how] to do anything.' [Hualde 2011: 483]

4.10 Synchronous polysemy

Indirect question constructions' counterparts rather frequently have more than one, function. The functions are partly discussed in previous sections 4.2-4.4, so, it was established, that subordinators, quotatives and disjunctive particles are possible counterparts of indirect question construction. In section 4.2 frequency of various subordinators (complementizers, relativizers and adverbializers) also was covered. In addition, the types of adverbial complements were taken into account.

However, indirect question markers' polysemy is not restricted to the types mentioned above. In table 7 below there is a full overview of polysemy attested in the current data:



Table 7. Syncronous polysemy of indirect question markers

Subordinator type	Constructions
complementizer (except for quotatives)	15
relativizer	7
adverbializer	12
quotative marker	2
disjunction marker	6
focus marker	8
additive marker	3
evidential marker	1
transitvizer	1

The table shows, that apart from already discussed types, focus, additive and conjunction markers are rather frequent as indirect question markers. Also, there are cases where embedded question markers are used to encode evidentiality and as transitivizers.

Focus markers' usage is quite expectable in polar questions. In 4 out of 7 attested cases focus marker is possible as a polar question strategy. The position of question/focus particle affects the informational structure thus expected answer (see section 4.8) for more details.

However, focus markers are also attested in indirect constituent questions. E.g., in Humburi Senni (Songhay > Nilo-Saharan), constituent question is formed through focalization of question word in both direct (74a) and indirect (74b) question:

(73a) [meмy seм] ?gaм=?м=?м=?м haмruм

[who? Dat] Focus=2SgS=Tr=3SgO say

`To whom [focus] did you-Sg say it?' [Heath 2014]

(74b) iМ suМ beмy [meмy gaм kaМ]

1SgS ImpfNeg know [who.Q Focus come]

`I don't know who came.' [Heath 2014: 367]

That is a rather rare type because usually, languages use the other strategies of wh-word focalization, e.g. intonation or word order change: wh-words may be either fronted or clefted to the clause-final position regarding the focus marking principle in a language.

One of the features of focus particles in question constructions is that they are universal for both direct and indirect questions, so, there is no embedded interrogative specific focus particles.

Thus, exhaustive focus particles (with the meaning similar to the English cleft) have a double function: if they are used to encode the focused element in questions, they function both as a focus marker and as an interrogative particle. This double functioning singles out exhaustive focus markers between other types of focus particles (scalar focus, additive etc.). E.g., it is known for Lezgian (Lezgic > Northeast Caucasian) question suffix -ni: it occurs in both direct and indirect question constructions (75) when suffixed to a finite verb, however, as an additive or a focus clitic, it may be attached to anything but finite verbs (76).

(75) gomer haqiqat.d-a xa-ji-di ja-ni tuљ-ni

[Homer reality-INESS be-AOP-SBST.SG COP-Q COP:NEG-Q]

hele sada-waj-ni tamam-diz subutar-iz xa-nwa-и

yet one-ADEL-even [perfect-ADV prove-INF] can-PRF-NEG

`No one has been able yet to prove conclusively whether Homer existed in reality or not.' [Haspelmath 1993: 426]

In the example below, -ni is used as a focus particle. To make focus meaning more explicit, kwaz (also a focus particle) is used after the ficalized constituent in addition to -ni:

(76) иi Qabustan=ba.di-kaj sew-er-iz-ni kwaz

we:GEN Qabustan=ba-SBEL bear-PL-DAT-also even

kiи'e-da!

afraid-FUT

`Even bears are afraid of our Qabustan-ba!' [Haspelmath 1993: 237]

In Kayardild (Tanghik > Macro-Pama-Nyungan) complementizing case -ntha used in indirect question constructions (77) may also be interpreted as an evidential marker (78). The distribution, as in Lezgian, is clear: if a complementizing case is used in the embedded clause, it bears subordinating semantics. In case it is used in independent clause, it is conceived as evidential marker. Note, that complementizing case marks each word in the clause:

(77) balmbi-wu nyingka kurri-ju, [ngaaka-ntha dangka-ntha

morrow-MPROP 2sgNOM see-POT who-COBL person-COBL

ngijin-jinaa-ntha wangal-wu-jarra-nth]COBL

lsg-MABL-COBL boomerang-V.GIFT-PST-COBL

`Tomorrow you will see who gave me the boomerang.' [Evans 2011: 518]

(78) [Dan-kurrka ri-in-kurrka thardawankawuru-ntha

here-LOC:COBL east-FROM-LOC:COBL aeroplane-COBL

burri-jurrk! ]

appear-IMMED:COBL

`(I can hear) the aeroplane coming in just now, here from the east!' [Evans 2011: 6]

4.11 Rare features

In the current section, I discuss a number of infrequent types attested in the sample. However, supposition about its rarity cannot be admitted in such a narrow data, so that is the issue for further research.

Special wh-word series

In most languages constituent interrogatives share one set of wh-words for both direct and and constituent questions. However, there are 3 constructions for 2 languages, where specific series of indefinite pronouns is attested in indirect question constructions.

For example, in Udihe (Northern > Tungusic), there are 3 various indirect constituent constructions:

- wh-word;

That is a regular wh-word strategy, that it rather spread in the current sample (see section 4.5 for more details). As ordinary “null” strategies, wh-word construction in Udihe is possible in both direct and indirect questions. Consider the following examples:

(79) Kuliga-tigi ni mafa-la-za?

snake-LAT who man-V-SUBJ

`Who would marry a snake?' [Nikolaeva & Tolskaya 2011: 814]

(79) Bi ei-mi sa: ono bi-zeye-i.

me NEG-1SG know how be-FUT-lSG

`I don't know how I will live.' [Nikolaeva & Tolskaya 2011: 814]

- wh-word=dA FinS;

In the strategy, an indefinite clitic =dA is attached to the question word, so that non-specific indefinite pronoun is derived. By other means, it is share features of regular wh-word strategy, cf. (80). However, indefinite pronouns are impossible in direct constituent question (81).

(80) J'e-le сansule-i?

what-LOC study-2SG

'Where do you study?' [Nikolaeva & Tolskaya 2011: 801]

(81) J'e-uxi-de susa:-ni ni-de ei sa:.

what-LAT-IND escape.PAST-3S who-IND NEG know

`Where she escaped to, nobody knows.' [Nikolaeva & Tolskaya 2011: 814]

- wh-word=dA PTCP;

The strategy differs from the previous one only with non-finite complement clause with accusative participles (82). Notably, in case of non-finite embedded clause, regular wh-word series is impossible.

(82) Si min-tigi noxoli gakpa-ja bi

you me-LAT first shoot-IMP.2SG me

ise-lege-i si ono-do gakpa-i-wa-i.

see-PURP-lSG you how-IND shoot-PRP-ACC-2SG

`Shoot at me first, so I can see how you shoot.' [Nikolaeva & Tolskaya 2011: 814]

In [Nikolaeva & Tolskaya 2011] there is no data on whether the constructions above are distributed semantically. It is also unknown, whether finite indefinite pronoun strategy may substitute regular wh-word strategy or not.

However, for Tamashek (Berber > Afro-Asiatic) such suppositions are provided. In Tamashek, like in Udihe, both wh-word and indefinite question word series are attested. As in Udige, indefinite question word strategy is impossible for root question constructions.

However, author notes that “the indefinite element functions as the head noun of a definite relative clause. The construction is therefore translatable as e.g. `(I don't know) the thing that he ate', `(I don't know) the place to which he went', and so forth, except that some of the indefinite nouns are reduced in form”. [Heath 2011: 660] Thus, syntactically indefinite question word strategy is a relativization, so therefore I take into account as a subordinator strategy with a relativizer.

In the following examples, regular question word ?nd?kй `where' (83) is replaced with indefinite analog и (84):

(83) ?nd?kй-t t-?nh?мy-жd

where?-3MaSgO 2S-see.Reslt-2SgS

`Where did you-Sg see him?' [Heath 2011: 654]

(84) w?r ?Мssen-ж? [и t-?kka]

Neg know.PerfN-lSgS [where 3FeSgS-go.to.PerfP]

`I don't know where she went.' [Heath 2011: 660]

Possibly, the instances of wh-word=dA PTCP strategy in Udihe may be interpreted as definite relative clauses well. The structure of relativization in Udihe resembles indirect question construction (85). However, in contrary to interrogative strategy, accusative marking of a participle is not obligatory: case marking depends on argument structure if a main verb.

(85) Ni: su:le-i-we-ni-de ja-u-mi.

[man be.ill-PRP-ACC-3SG-FOC] medicine-CAUS-lSG

`I cure the people who are ill.' [Nikolaeva & Tolskaya 2011: 673]

The same is true for Humburi Senni language (Songhay > Nilo-Saharan) (86), but in the current data it is unknown, whether boМr `person' is an indefinite pronoun or not.

(86) iМ suМ beмy [boМr kaм kaМ]

1SgS ImpfNeg know [person Rel come]

`I don't know who came.' [lit. `I don't know the person who came.'] [Heath 2014: 367]

All in all, there are a list of issues on special question word series that I leave for further research:

- if there any other pronoun series, except for indefinite one, are possible in indirect constituent questions?

- how finite embedding with indefinite pronoun (like in Udihe) may be interpreted?

- how frequent relativization usage with indefinite pronoun in indirect question construction is?

- whether such constructions are related to special semantics of `know'? The verb `know' may have two interpretations: the knowledge about some information (“I know who came”) and being familiar with someone (“I know the man who came”), that may require different constructions. It is unknown, whether such difference persist in mentioned languages.

Narrowing of functions

According to data analysis discussed above, a considerable tendency may be found: if indirect question marker is possible in direct interrogatives, then, possible question types correspond each other. However, Lezgian (Lezgic > Northeast Caucasian) and Russian (Slavic > Indo-European) languages is an exception for the conclusion.

In Lezgian, interrogative marker -ni may be used in polar and alternative independent questions. It attaches to the finite verb, e.g.:

(87) Betxoven.a-n muzyka wa-z k'an-da-ni?

Beethoven-GEN music you-DAT like-FUT-Q

`Do you like Beethoven's music?' [Haspelmath 1993: 417]

In indirect question constructions, conditional marker -t'a may be used for polar (88) and constituent (89) interrogatives:

(88) za sadra, kkal.i xa-nwa-t'a akwa-n

I:ERG PT [cow(ERG) bear-PRF-CND] see-HORT

`Let me see whether the cow has calved.' [Haspelmath 1993: 425]

(89) Wun inal kkwe-kaj extilat fi-zwa-t'a

you:ABS [here what-SBEL talk go-IMPF-CND]

gawur.d-a hat-na-и-ni?

understanding-INESS get-AOR-NEC-Q

`Haven't you understood what we are talking about here? (lit. what the conversation is going about?)' [Haspelmath 1993: 425]

-ni suffix may used in indirect question as well, however author points out, that only alternative indirect interrogatives are possible with -ni. The suffix -ni attached to both verbs:

(90) gomer haqiqat.d-a xa-ji-di ja-ni tuљ-ni

[Homer reality-INESS be-AOP-SBST.SG COP-Q COP:NEG-Q]

hele sada-waj-ni tamam-diz subutar-iz xa-nwa-и

yet one-ADEL-even [perfect-ADV prove-INF] can-PRF-NEG

`No one has been able yet to prove conclusively whether Homer existed in reality or not.' [Haspelmath 1993: 426]

(91) ada-z im axwar ja-ni, xabar ja-ni

he-DAT [this:ABS dream COP-Q news COP-Q]

иi-zwa-и-ir

know-IMPF-NEC-PST

`He did not know whether this was dream or reality.' [Haspelmath 1993: 426]

Russian exhibits a similar behaviour regarding indirect alternatives questions. A “null” strategy is used in direct polar (92) and alternative (93) questions (interrogative semantics is marked intonationally), however, like in Lezgian, only alternative indirect questions are available for the construction (94).

(92) Ты придешь?

(93) Ты будешь чай или кофе?

(94) Кто-то спросил Ї мужчины это или женщины, и если это одни мужчины, то почему?

Such constructions reveal a list of unusual features:

- there is no other instances of functional narrowing from direct to indirect question constructions in a given data;

- dedicated markers for indirect alternative questions are extremely rare in the current sample;

All in all, I can see the following directions for investigation of the phenomenon:

- accumulation of alternative question data for indirect questions as the current data is insufficient to judge, whether alternative question-specific markers are rare or not.

- checking the tendency of available question types similarity for direct and indirect constructions and define types of cases that may violate it.

5. Summary

In the current research various means of indirect question marking are observed. According to the data, the following groups of marking are defined:

- direct speech marking;

The type is not covered in the current research, however, is a rather spread strategy of indirect question marking. In case of direct speech construction, an indirect question is expressed as a quotation without deictic shift. Such constructions may involve root question markers and quotative particles.

- “null” marking;

“Null” strategy implies absence of any specific question, subordination, etc. markers in a construction. Indirect question constructions of “null” strategy embedded question clause is juxtaposed with a main clause. However, deixis shift of pronouns differs “null” strategy from direct quotation: it is necessary in indirect question constructions, while direct quotation entails exact repetition of a given expression.

- question particle strategy;

Question particles may be expressed as affix, clitic or autonomous word. They tend to be universal for both direct and indirect interrogatives, however, embedded question specific particles are rather spread, too. Question particles may combine more than one functions and be, e.g., also focus markers.

- subordinator strategy;

Subordination markers are rather frequently used in indirect question constructions. Three types of subordinators are attested in the constructions: complementizers, relativizers and adverbializers. Complementizers are the most frequent means of indirect question marking as many languages interpret embedded questions as complement clauses. Complementizers may be either indirect question specific or not. Also, indirect question constructions frequently share marking means with adverbial conditional or concessive clauses. Other types of adverbial subordinators (e.g. the ones for purpose clause) are also attested, but rarely. In addition, in some cases indirect question clause may be constructed as relativization. Subordinators sometimes may be used in root question, but that implies special semantics (e.g. rhetorical).

- other polysemy markers;

There is quite a wide range of polysemy options, that are attested in the current data. For example, disjunctive particles are rather frequently used as question particles. There is an instance of quotative marker usage in indirect question constructions.

Rather frequently indirect polar question constructions invoke duplication of a verb in a negative form or negative auxiliary or a pro-verb with question markers attached to both verbal forms. Syntactically, they resemble alternative questions. In the current research this strategy is termed as specific verbal construction. However, such constructions need clarification of definition and classification in further investigation.

The following implications follows from the observed data:

1. If a marker is available in both direct and indirect question, then possible question types are corresponding (there are only 2 exceptions for this). E.g. in case a construction marks indirect polar question, then in marks root polar question.

2. Question particles tend to mark both root and indirect questions, while subordinators tend to be indirect question specific. “Null” strategy is always possible in both root and indirect interrogatives.

3. If a language allows “null” strategy for indirect polar question, than it is allowed for indirect constituent question, too.

4. Languages tend to not encode different question types with different indirect question constructions. More than half of attested markers are not specific for question types.

5. However, some types of construction are always specified for certain question types: e.g. special verbal constructions and disjunctive markers may be used only in polar questions.

6. Specific word order is rather rare for indirect question constructions. Usually word order in embedded question corresponds to root question one.

7. Main verb semantic type affects the choice of indirect question construction considerably.

8. On the contrary, frequently informational structure of embedded question does not influence the choice of strategy. However, it may affect word order or intonation.

The list of implications is surely not exhaustive, however, the further investigation is needed for data clarification and widening of the sample.



Conclusions

To sum up, the current research is dedicated to the typology of indirect question constructions. It is a pilot study aimed to test the methodology of investigation and make preliminary typological implications.

Despite active interest in subordinate constructions (such as complement, relative and adverbial clauses) in the recent years, the indirect question domain remains unexplored in macrolinguistic prospect and is limited to specific languages' observations. However, the construction has a variety of special features resulting from the combination of both subordinate clauses and independent question properties. As a result, the means of expressing indirect questions in the languages of the world vary to a great extent.

In this research, I applied typological methods to collect data on a sample of 30 languages. Grammars and field questionnaires served as main data sources, thus the sample is compiled on the basis of the sufficiency of syntax description. Also, the languages using direct quotation as the only indirect question strategies were skipped as well. The data collection resulted in a typological database, where single indirect question construction is a basic database unit. Normally, a language has more than one construction of such type. Each unit is analyzed with the use of comparative parameters, that allow comparing structurally different languages. The parameters cover different types of questions (both dependent and independent), synchronic polysemy and additional data on diachrony, word order, etc.

Observation of the language sample resulted in 78 indirect question constructions. They were grouped into strategies according to the semantics and syntactic status of the constructions' counterparts: question particles, subordinators, quotative, disjunctive markers and “null” constructions. However, the data analysis revealed that the list of features is not exhaustive. Also, specific verbal constructions are taken into account.

Though grammar-based approach did not result in full and sufficient data on the constructions, several typological implications are found: a number of global implications for the indirect question syntax and construction types were established.



References

1. Aikhenvald 2008 -- Aikhenvald, Alexandra. (2008). Semi-direct speech: Manambu and beyond. Language Sciences. 30. 383-422.

2. Akmajian 1997 -- Akmajian, A. (1977). The complement structure of perception verbs in an autonomous syntax framework. In Formal Syntax, P.W. Culicover, T. Wason & A. Akmajian (eds), 427-460. New York NY: Academic press.

3. Bakker 2010 -- Bakker D. Language sampling. (2010). The Oxford handbook of linguistic typology. Oxford: Oxford University Press.

4. Bakker & Miestamo -- Bakker, D., & Miestamo, M. (2016). Sampling for variety. Linguistic Typology, 20(2).

5. Barbour 2012 -- Barbour, J. (2012). A Grammar of Neverver. Berlin, Boston: De Gruyter Mouton.

6. Bowern 2013 -- Bowern, C. (2013). A grammar of Bardi. Berlin: De Gruyter Mouton.

7. Coupe 2008 -- Coupe, A.R. (2008). A Grammar of Mongsen Ao. Berlin, Boston: De Gruyter Mouton.

8. Carlson 2011 -- Carlson, R. (2011). A Grammar of Supyire. Berlin, Boston: De Gruyter Mouton.

9. Cristofaro 2003 -- Cristofaro S. (2003). Subordination. Oxford: Oxford University Press.

10. Deal 2017 -- Deal A. R. (2017). Shifty asymmetries: universals and variation in shifty indexicality.

11. Egorova 2020 -- Egorova A.D. (2020). Field data.

12. Evans 2011 -- Evans, N.D. (2011). A grammar of Kayardild: With historical-comparative notes on Tangkic. Berlin, Boston: De Gruyter Mouton.

13. Fedden 2011 -- Fedden, Sebastian Olcher. (2011) A Grammar of Mian (Mouton Grammar Library 55). Berlin: Mouton de Gruyter.

14. Fox & Thompson 1990 -- Fox, B., & Thompson, S. (1990). A Discourse Explanation of the Grammar of Relative Clauses in English Conversation. Language, 66(2), 297-316.

15. Glinert 2004 -- Glinert L. (2004). The Grammar of Modern Hebrew. Cambridge: Cambridge University Press.

16. Haspelmath 1993 -- Haspelmath, Martin. (1993). A Grammar of Lezgian. Berlin: Mouton de Gruyter.

17. Haspelmath 2016 -- Haspelmath M. (2016). The serial verb construction: comparative concept and cross-linguistic generalizations. Language and Linguistics 17: 291-319.

18. Heath 2008 -- Heath, J. (2008). A Grammar of Jamsay. Berlin, Boston: De Gruyter Mouton.

19. Heath 2011 -- Heath, J. (2011). A Grammar of Tamashek (Tuareg of Mali). Berlin, Boston: De Gruyter Mouton.

20. Heath 2014 -- Heath, J., & Hantgan. A. (2014). Grammar of Humburi Senni. Berlin, Boston: De Gruyter Mouton.

21. Hцlzl 2016 -- Hцlzl, A. (2016). Towards a new typology of questions. Bavarian Working Papers in Linguistics 5: 17-28.

22. Honeyman 2016 -- Honeyman, Thomas. (2016). A grammar of Momu, a language of Papua New Guinea. Berlin, Boston: De Gruyter Mouton.

23. Hualde 2011 -- Hualde, J., & Ortiz de Urbina, J. (Eds.). (2011). A Grammar of Basque.

24. Karttunen 1977 -- Karttunen, Lauri. (1977). Syntax and Semantics of Questions. Linguistics and Philosophy.

25. Keenan 1985 -- Keenan, E. (1985). Relative clauses. In Language Typology and Syntactic Description: Complex Constructions, T. Shopen (ed.). Cambridge: Cambridge University Press.

26. Kiefer F. (1980). Yes-No Questions as Wh-Questions. In Searle J.R., Kiefer F., Bierwisch M. (eds) Speech Act Theory and Pragmatics. Texts and Studies in Linguistics and Philosophy, vol 10. Springer, Dordrecht.

27. Kina 2018 -- Dobrushina N., Daniel M., Maisak T. et al. (2017-2020). Kina Rutul Field Data.

28. Kцnig & Siemund 2007 -- Kцnig E., Siemund P. (2007). Speech act distinctions in grammar. In Shopen T. (ed.). Language Typology and Syntactic Description. Vol. I. Cambridge: Cambridge University Press.

29. Koul 2009 -- Koul, O. N. (2009). Modern Hindi grammar. Delhi: Indian Institute of Language Studies.

30. Lehmann 1998 -- Lehmann, C. (1988). Towards a typology of clause linkage. In Clause Combining in Grammar and Discourse [Typological Studies in Language 18], J. Haiman & S. Thompson (eds), 181-225.Amsterdam: John Benjamins.

31. Krifka 2013 -- Krifka M. (2013). Question. In Maienborn, C., Heusinger, K., & Portner, P. (eds.). Semantics. Volume 3. Berlin, Boston: De Gruyter Mouton.

32. Maier 2014 -- Maier E. (2014). Mixed quotation: the grammar of apparently transparent opacity. Semantics and pragmatics 7, pp. 1-67.

33. Miyaoka 2015 -- Miyaoka, O. (2015) A grammar of Central Alaskan Yupik (CAY). Berlin, Boston: De Gruyter Mouton.

34. McWhorter, J., & Good, J. (2012). A Grammar of Saramaccan Creole. Berlin, Boston: De Gruyter Mouton.

35. Nefedov 2015 -- Nefedov A. Clause linkage in Ket. -- Utrecht: LOT, 2015.

36. Nikitina 2012 -- Nikitina T. (2012). Personal deixis and reported discourse: towards a typology of person alignment. Linguistic Typology 16:233-263.

37. Nikolaeva 2011 -- Nikolaeva, I., & Tolskaya, M. (2011). A Grammar of Udihe. Berlin, Boston: De Gruyter Mouton.

38. Nikolaeva 2014 -- Nikolaeva, I. (2014). A Grammar of Tundra Nenets. Berlin, Boston: De Gruyter.

39. Noonan 2011 -- Noonan, M. (2011). A Grammar of Lango. Berlin, Boston: De Gruyter Mouton.

40. de Rijk 2018 -- Rudolf P.G. de Rijk. (2018). Standard Basque, A Progressive Grammar. Volume 1. Cambridge, MA: MIT Press.

41. Sakel 2011 -- Sakel, J. (2011). A Grammar of Mosetйn. Berlin, Boston: De Gruyter Mouton.

42. Siemund 2001 -- Siemund, Peter. (2001). Interrogative Constructions. In Martin Haspelmath, Ekkehard Kцnig, Wulf Oesterreicher, and Wolfgang Raible (eds.), Language Typology and Language Universals. Berlin: de Gruyter, 1010-28.

43. Testelets 2001 -- Testelets Ya. (2001). Vvedenie v obschij sintaksis. [Intoduction to general syntax.] Moscow.: RGGU.

44. Thompson & Longacre 1985 -- Thompson S.A., Longacre R.E. (1985). Adverbial clauses. // T. Shopen (ed.). Language typology and syntactic description. Vol. 2: Complex constructions. Cambridge, England: Cambridge University Press.

45. Uegaki 2019 -- Uegaki, W. (2019). The semantics of question-embedding predicates. Lang Linguist Compass.

Размещено на Allbest.ru