A Brief Description on Impersonal Constructions in Uyghur

Impersonal constructions have been a regular topic of investigation in various languages which belong to different language families across the world. Discussions over Impersonal constructions would constitute main contribution to the theoretical study. Although impersonal constructions are the main characteristics of Uyghur, but it’s been hardly noticed by theoretical linguists. With this research, I would like to put forward the idea that Uyghur also have a wide variety of impersonal constructions, the analysis of which would bring an interesting contribution to the typology of impersonality. Scholars conceive impersonal constructions in different terms, some apply morphological methods, and others adopt syntactic approaches. Whichever methodology they apply, it is unarguable that impersonal constructions are agentless by nature, in which the sentences may not have an overt subject. They are many varieties of types, such as agentless gerunds, agentless passives, existential sentences etc. Since Uyghur is a pro-drop language, omitting the pronominal subject also helps to form impersonal constructions. This paper provides a brief description on the types of impersonal constructions, explicates several ways of forming agentless sentences, and introduces main types of impersonal sentences in Uyghur.


Introduction
Impersonal constructions have been a regular topic of investigation in Indo-European studies, and similar constructions have been described in languages spoken in various areas of the world and belonging to different language families. It had been widely discussed in Anna Siewierska [1,2], Andrej Malchukov and Anna Siewierska [3], Andrej Malchukov and Akio Ogawa [4] etc.
Modern Uyghur is a direct descendant of old Uighur, which is spoken in western China. Even though Impersonal constructions are one of the major sentence types, we can rarely see exclusive description on them in Uyghur grammars. Some important works such as Hazirqi Zaman Uyghur Tili 'Modern Uyghur Grammar' [5,6] didn't even mention it. Since most Uyghur Grammarians mainly focused on morphology, detailed information on sentence structures are scarce. In spite of that, there are two well-discussed papers on this topic in Uyghur with Chinese translations. They explained the impersonal constructions from different perspectives. General description on impersonal constructions was published by several linguists. See Xaliq NIYAZ [7], Zäynäp NIYAZ [8], Niyaz TURDI [9], Cheng at al. [10], Hämit TÖMÜR [11] and Litip TOHTI [12]. Xaliq NIYAZ [7] mainly discussed sentence types in modern Uyghur and classified them according to their properties. He called one of these sentence types as Igisiz Jümlä 'subjectless sentence'. Here, he mainly implied the grammatical subject. Hämit TÖMÜR [11] and Cheng et al. [10] provide general information on Uyghur grammar, only scattered discussions on impersonal constructions can be found in these works. Litip TOHTI [12] thoroughly discussed the syntactic structure of Altaic languages in the frame of Generative Syntax. Page 139-140 includes some discussions on impersonal constructions, where he called them 'agentless sentence. Most of these works cover general discussion without details.

General Description
Impersonal construction mainly points to agentless construction in Uyghur grammar. The notion 'impersonal' in Uyghur grammar books is disparate because some scholars conceive it in morphological terms, while others adopt syntactic approaches. Most Uyghur grammarians adopt syntactic approach. Nevertheless, they have slightly different understandings on the nature of impersonal constructions.
The syntactic characterization of impersonality involves subjecthood. Impersonal constructions are seen to either lack a grammatical subject altogether or alternatively feature only a pleonastic (semantically empty) subject, be it overt one or covert on [1] When the gerund is in the dative case and combined with the construction toγra käl-'have to', it produces subjectless sentence which indicate the need for the action to be carried out.

Bu
iš-ni bügün tügit-iš-kä toγra kel-idu. this work-ACC today finish-VN-DAT have to-NPAST 'This work must be completed today.' When linked with the verb bol-'to be', it indicates the possibility that the action expressed by the adverbial will be carried out. Such sentences are without subject.
It is possible to finish this job within two days.'

Main Types of Impersonal Sentences
There are several types of impersonal sentences: agentless sentences, incomplete sentences, existential/ dependent construction, modal sentences etc.
1) agentless sentences The agent of these sentences are hardly identifiable, it only includes a core sentence which equals to predicate. e.g: One must not retreat in front of the difficulty.' Ay-ni etäk bilän yep-ip bol-ma-s. moon-ACC elbow with cover-CONV be-NEG-AOR 'It is impossible to cover the moon with the elbow. (One couldn't hide the crystal-clear truth.)' As Janet R. Aiken [13] pointed out, constructions lacking subject or verb or both are of a great variety of types, from the imperatives such as COME HERE and the omitted first person types WENT DOWN TOWN TODAY …. where it is difficult or impossible to construct a full sentence convincingly.
Such imperative sentence can also be impersonal in Uyghur. It is difficult to construct the agent of the sentence, since it does not agree with the agent-predicate agreement principle. E.g.: Silär-niŋ yardim-iŋlar-γa köp rähmät. You.pl-GEN help-POSS.2pl-DAT many thank 'Many thanks to your help' Heyt-iŋiz-γa mubaräk bol-sun. Eid-POSS.2sg-DAT happy to be-IMP.3sg 'Happy ramazan festival!' 2) Incomplete sentence These sentences are incomplete by nature, so it can be said non-sentence [13] since it is completely impracticable to supply the missing elements to make complete sentences.
This type exists in many languages, since short sentences agree with the economy principle. Main part of these sentences is indicated by nominative phrase, or it can be a short clause. e.g.: Bazar-da adäm köp-tur. street-LOC people many-COP 'There are lot of people on the street.' Existence or appearing is the main information in these sentences. Therefore the place of existence must be spoken out first, otherwise it will become non-existential sentence. 4) Modal sentence English modal verbs correspond to Uyghur modal adjectives. In these types, agent part is gerunds, predicate part is modal adjectives such as mümkin 'possible', keräk~lazim 'must, should, have to', šärt 'should'.e.g.: Öy-dä adäm yoq bol-iš-i mümkin. house-LOC man no be-VN-POSS.3sg possible 'May be there is no one in the house.' In this type of sentences, the modal adjectives and gerunds constitute a strict agent-predicate construction, which loosens the relationship between possessor and dependant. As a result, the genitive case drops. a.
Silär-niŋ ätä kel-iš-iŋlar-// keräk You.pl-GEN tomorrow come-VB-2pl.POSS must b. Silär-// ätä kelišiŋlar keräk 'You must come tomorrow.' 5) impersonal passives Langacker and Munro [14] argue persuasively that passive constructions are basically agentless, and that agentive phrases are derived from external sources. In this view, corresponding passive and active sentences are related semantically, but do not have a common conceptual (i.e. underlying) structure. They give evidence from a number of Uto-Aztecan languages and from Mojave, a Yuman language, to show that passives are basically impersonal constructions, derived 'from structures in which a clause with unspecified subject is embedded as subject complement to the predicate BE.
Explicit agents do occur with impersonal passives. Furthermore, when agents are not explicitly expressed, they are predictable from the context in a number of cases. Generally, agentless passives are derived in all cases by a transformation of indefinite agent deletion.
Passive constructions in Uyghur are produced by attaching the suffix -n (-n/-in) or -l (-l/-i-/-ul/-ül). Passive voice indicates that the grammatical subject of the sentence is actually the logical object of the original action. iii. It is necessary to especially accentuate the logical object.
Düšmän-lär yoqit-il-di. enemy-PL exterminate-PASS-PST 'The enemies were exterminated.' iv. In some situations it is necessary to point out the logical subject at the same time as accentuating the logical object. In such cases, if the logical object is a person, people in general, or some organization, the noun which indicates that logical subject is combined the the proposition täripidin 'by' to form an adverbial modifier.
Paša išan täripidin orunla-n-γan naxša alqiš-qa eriš-ti. Pasha Ishan by play-PASS-PARTCPL song applause-DAT obtain-PAST 'The song performed by Pasha Ishan was applauded.' v. If the logical subject was something else, the role of the logical subject is indicated in different ways.

Some Arguments over the Nature of Impersonal Construction
Most scholars agree on the nature of impersonal constructions. However, two authors put forward slightly different opinions.
According to the characteristics of some agentless constructions whether it can be converted into agentive ones, Niyaz TURDI [9] asserted that impersonal constructions can be divided into two types: absolute impersonal construction and relative impersonal construction. 1) Relative impersonal constructions a) The predicate part of relative impersonal construction is the combination of non-personal gerunds and modal adjectives, such as lazim 'should', keräk 'must' etc. By adding person marker to the gerund, the sentence can be converted into covert subject sentence. E.g.: Qiyinčiliq ald-i-da täwrä-n-mäs-lik keräk (impersonal) difficulty front-POSS.3sg -LOC shake-PASS-NEG-NOML must 'One must be unshakeable in front of difficulty.' Qiyinčiliq ald-i-da täwrä-n-mäs-lik-imiz keräk.
(personal) difficulty front-POSS.3sg-LOC shake-PASS-NEG-NOML-POSS.1pl must 'We must be unshakeable in front of difficulty.' The personal marker -imiz which added to the gerund is the key factor that converts impersonal construction into personal contruction.
b) If the first part of a sentence, whose predicates is bol-'to be', toγra käl-'have to', is gerund, it will form an impersonal construction. If the gerund part has person marker, it forms a covert personal construction. E.g.: U-niŋ öz-i bilän kör-üš-üš-kä toγra kel-idu. 'We have to meet his own self.' c) When the head of the sentence is dative or genitive noun phrase, and after that appears the combination of gerund and toγra käl-'have to', if the case marker drops off, the gerund also loses its possessive marker. As a result, the sentence becomes an agentive construction.e.g.
Mämät-kä qayt-ip ket-iš-kä toγra käl-di. Hämmi-miz-niŋ kel-iš-imiz zörür. All-POSS.1pl-GEN come-VN-POSS.1pl necessary 'It is necessary for all of us to come.' She argued that they are all agentive sentences. In these sentences, the -š gerunds as mundaq rohiŋizdin öginiš, hämmimizniŋ kelišimiz play the role of the subject, and keräk, lazim, zörür can play the role of predicates. She also argued that these sentences [the combination of gerunds+mümkin] as below were also mistakenly taken as subjectless: Bu kitab-din paydilin-iš mümkin. this book-ABL use-VN possible 'It is possible to make use of this book.' Biz-niŋ u yär-gä bar-mas-liq-imiz mümkin. we-GEN that place-DAT go-NEG-NOML-POSS.1pl possible 'Tt is likely that we go there.' In these sentences, bu kitabdin, bizning u yärgä barmasliqimiz are nominative gerunds which play the role of agents. And mümkin is a predicate whose copila -dur is omitted.
Thirdly, these sentences also belong to such mistakes:

Conclusion
To sum up, Zäynäp NIYAZ [8] and Niyaz TURDI [9] put forward different opinions on the nature of impersonal constructions and narrowed the scope. They thought that such sentences as hämmimizniŋ kelišimiz zörür were in fact subjective sentences, and treated hämmimizniŋ kelišimiz as a subject. But from the point of subject predicate agreement rule in Uyghur, in which the predicate must agree with the subject in person, for instance in män käldim 'I have come', the person marker -m in predicate agrees with the subject män, we have to deny their suggestions. In män käldim 'I have come', by omitting the subject män, a syntactic pleonasm will be formed. E.g.: män käldim 'I has come' käldim 'I has come' In this case, the pronoun män 'I' is grammatically optional, both sentences mean 'I has come'. But in the case of Zäynäp NIYAZ [8] and Niyaz TURDI [9], the covert pronoun biz 'we' doesn't agree with the predicate zörür in person.
Another sentence type, whose agent is difficult to define, is pro-drop ambiguity sentence. E.g.:

Tülkä
käl-gän täräp-kä qari-di fox come-PARTCPL direction-DAT look-PAST 'The fox looks at the direction from where it (the fox itself) comes. ' 'The fox looks at the direction from where he (human) comes. ' 'He (human) looks at the direction from where the fox comes.' The position of covert subject is the main reason of ambiguity in this sentence. It was extensively discussed by Muzappar ABDURUSUL [14]. Here I shall limit my discussion to conventional impersonal constructions.
To sum up, impersonal constructions are one of the main sentence types of Uyghur language. It not only displays universal features, but also shows language-specific characteristics. In this preliminary study, the author has discussed the types of impersonal constructions and their formation in Uyghur. Thus, this paper will pave a way for further studies of impersonal constructions and provides important languages facts for future cross-linguistic study.