Research on grammar of social cognition and human reference in Khalkha-Mongolian

Размещено на http://www.allbest.ru/

Research on grammar of social cognition and human reference in Khalkha-Mongolian

Dolgor Guntsetseg

Post-Doc Researcher,

Institute of Finnougric and Uralic Studies

Ludwig Maximilian University

Abstract

This paper presents empirical research on the coding of social cognition in Khalkha-Mongolian grammar. It is conducted in the framework of a larger international project that has developed an innovative field elicitation tool that yields natural spontaneous and interactive speech data as well as analysis methods for these data. Its goal is to find out which social and cultural parameters influence the grammar of different languages and how the mental processes of individuals are reflected in it. This paper discusses the collection of data and the preliminary results concerning human reference strategies and presents some initial insights regarding linguistic aspects of social cognition in Khalkha-Mongolian. Keywords: Khalkha-Mongolian, social cognition, field elicitation tool, corpus data, human reference, kinship terms, `kintax', possessive marking, reference and topic continuity.

Аннотация

ИССЛЕДОВАНИЕ ГРАММАТИКИ СОЦИАЛЬНОГО ПОЗНАНИЯ И РЕФЕРЕНЦИИ ЧЕЛОВЕКА В ХАЛХА-МОНГОЛЬСКОМ ЯЗЫКЕ

Долгор Гунцэцэг

Ключевые слова: халха-монгольский язык; социальное познание; инструмент полевого извлечения; данные корпуса; человеческое упоминание; термины родства; «синтаксис»; притяжательные конструкции; преемственность ссылок и тем.

Introduction

The aim of this paper is to introduce the research project “Grammar of Social Cognition in Khalkha-Mongolian”, which is supported by the German Research Foundation (DFG, project number 417675039, 2019-2022). The concept of the project shares the framework of a larger international project, “Language and Social Cognition”, headed by Prof. Nicholas Evans (Australian National University, the ARC Centre of Excellence for the Dynamics of Language), which aims to produce a detailed cross-linguistic study of the ways that social cognition can be coded in a language. Within this project, an international team of researchers is working on a large corpus called "Social Cognition Parallax Interview Corpus (SCOPIC)" that contains a broad sample of 30 languages from around the world and consists of "richly annotated data focusing on functional categories relevant to social cognition, the social and psychological facts that place people and others within an interconnected social context and allow people to interact with one another" [2, p. 1]. For each language, a translated and annotated corpus will be created; the tasks of the team in this regard are to generate broadly parallel, comparable discourse, including both narrative and dialogic elements, while leaving speakers free to produce spontaneous material. This is done by using a narrative problem-solving task that encourages naturalistic speech, and whose design allows people to linguistically encode whatever social-cognition relevant categories they recognize and choose.

The Khalkha-Mongolian data is being collected and analyzed by Dr. Dolgor Guntsetseg and Prof. Dr. Elena Skribnik. Cross-linguistic studies emerge from discussions in annual SCOPIC annotation workshops and in joint publications.

The structure of the paper is as follows. Section 2 introduces the term `social cognition' as a factor relevant for the structuring of linguistic expressions and is therefore necessary to consider in grammatical descriptions and able to provide some cross-linguistic insights. Since social cognition is a wide field of research, the research focus of the current project will be briefly introduced, too. Section 3 provides a brief description of the research design and elicitation methods of the study, both with regard to our field research and to annotation tools for comprehensive data analysis. Section 4 discusses preliminary data analysis concerning human reference in Khalkha-Mongolian followed by a brief summary in Section 5.

1. Social Cognition in grammar

In recent decades, it has been widely recognized that languages are not just fixed formal systems, but rather that they are dynamic systems that interact with cognitive processes such as perception, information processing and modelling the world picture of individuals or social and cultural groups based on their experiences, i.e. social cognition. Following Frith and Frith [5, p.724], social cognition is considered to be “the sum of those processes that allow individuals of the same species (conspecifics) to interact with one another”. Interaction between humans is primarily language-mediated communication supported with non-verbal signals such as facial expressions, gesturing and body posture. Accordingly, it is to be expected that the social knowledge of a speaker's community is encoded in many parts of a language's expressive resources including morphosyntax, lexis, prosody and gesture. Thus, language research has started focusing on social and cultural aspects, conducting empirical studies with usage-based methods of fieldwork, watching speakers in their natural social communication surroundings. Recent studies on linguistic expressions of social cognition [4, ch. 4; 10; 2] address the question of how the mental processes and social awareness are coded in a language during communication. Moreover, they consider the following two aspects of human interaction that are represented by linguistic expressions referring to social cognition:

1) Social facts: during an effective and successful communication within a social group, it is vital to code "social facts referring to kinship relationships, status and ownership" [10, p. 136].

2) Psychological facts: interactive communication between humans is full of information exchange about "individual's own desires, feelings and thinkings or estimating of these for others" [2, p. 3].

The research topics of the current project are selected from both groups: human reference, reported speech and thought, `private predicates', benefactive events and stance (including evidentiality, mirativity and epistemicity). In the present paper, some preliminary results of our analysis of human reference strategies in Khalkha- Mongolian will be discussed (see Section 4).

Such an approach, applied to Khalkha-Mongolian, will help to gain new knowledge, show already known phenomena in a new light, and present a description of a dynamic system in its use in communication, cutting through different levels of language, from morphosyntax to lexicon.

2. Research design and methods

The research fundament of this project is based on the Family Problems Picture Task [10], an interactive narrative problem-solving task developed by an Australian research group as a field elicitation tool for recording language data rich in social cognitional content and resulting in stimulus-based, semi-directed spontaneous speech data. The stimulus materials of the task contain 16 pictures (based on original drawings by Alice Carroll, see Figure 1) that build a coherent story and depict people and socially pregnant and emotionally charged events.

Figure 1. Homecoming

According to the instructions in San Roque et al. [10], the task has a predetermined structure consisting of the following three phases. In Phase 1, a pair of native speakers sees the pictures one at a time, in a predetermined but non-logical order, and describes each picture andwhat is happening in it. The participants are not told that the pictures set up a story but are asked to work together through the whole task, so that there arises a dialogic interaction between them. In Phase 2, the participants have to put the pictures into a logical order, i.e. into a coherent story. Again, collaborative work is needed that leads to more active discussions concerning the content of the pictures and to an intensive dialogue with directives, questions and attentionmoves. In Phase 3, participants are asked to retell the resulting story a) in the third-person perspective, i.e. `from outside', and b) in the first-person perspective, i.e. from the point of view of any chosen protagonist in the story. As an audience for this retelling, new participants can be involved.

This task enables naturalistic interaction between the participants solving a narrative problem and makes them employ their social experience to recognize or interpret the scenes in the pictures in terms of what their language can code. The third phase of the task induces different packaging for the same events as between third- person and first-person narratives and gets participants to return several times, in a natural way, to the characterization of the same events, giving them the opportunity to exhibit alternative ways of depicting the same thing.

From a linguistic point of view, the task enables the researcher to elicit language data containing a) narration describing social relationships, perceptions, speaking, thinking and the feelings of the protagonists in the pictures; and b) expressions of the participants' own processes of observation, inference, evaluation and estimation [10].

The field work in Mongolia for data elicitation was conducted in two steps. In 2017, two sessions were conducted with Khalkha-Mongolian native speakers living in urban regions. In 2019, native speakers living in province regions were involved: three sessions were held in Govi-Altai in the South-West, four in Dornogovi in the South-East, and three in Khentii in the North-East of Mongolia.

Figure 2. First data elicitation session 2017

Altogether, 12 sessions were conducted with 24 participants (not counting audience participants), ages 20-65 and of different genders. The professions of the participants were diverse: cattle breeders (nomads), workmen, civil servants, a museum guide, etc. To create a relaxed and open-minded atmosphere during the elicitation sessions, a pair of participants who knew each other well is generallyselected, e.g. couples, colleagues, friends, neighbors.

Each session including all three phases was recorded in video and audio (approximately nine hours of audio-visual materials). The video recording employed two cameras: one from the front facing the participants to capture non-verbal features like gestures and facial expressions; and one from a bird's eye view in order to see where a speaker was pointing on the picture during the picture description task.

The participants coped well with the first two phases of the task: description of the pictures and putting them into logical order. But they either completely refused to retell the story from 1^st person perspective or accepted doing so with discomfort, viewing itas a bad omen (because of the depicted events of drinking and domestic violence). Instead,participants wanted to tell a similar story that happened either to them or in their own social surroundings.

The linguistic annotation software ELAN, ELAN (EUDICO Linguistic Annotator) is an open-source tool for time-aligned linguistic annotation developed at the Max-Planck-Institute for Psycholinguistics, The Language Archive, Nijmegen, The Netherlands (cf. Sloetjes & Wittenburg 2008). a multimedia annotator tool, was used for the data analysis. For each session, one ELAN file is created that integrates the two video files (.mp4) and one audio file (.wav). Multiple annotation tiers are built in, linked to these associated media segments.

The first basic set includes four tiers. In the first tier, the spoken language materials are written down in Cyrillic script, following the literary language in orthography. For some phonetically specific features of the spoken language, an additional tier can optionally be added. The second basic tier presents the Latin transliteration, set out on the “one letter to one letter” principle, i.e. mapping exact orthographic rules of the written language in Mongolia. The next tier is for glossing following the Leipzig glossing rules; URL: https://www.eva.mpg.de/lingua/resources/glossing-rules.php finally the fourth tier contains a translation into research language English.

The main analysis part in the ELAN files consists of annotation tiers for tag- ging/coding the linguistic structures relevant for the research topics. The tiers are structured hierarchically but differentiated by individual speakers. Each tier includes a dependent notes tier for special remarks as well as coding of mimicry, gestures and body posture. Currently, annotation for human reference and reported speech and thought is in the works. Simultaneously, the annotated data are being analyzed both statistically and linguistically to detect social-cognitive aspects that influence grammatical structure. The plan for work in the near future covers the annotation and analysis of the data forthe research topics of private predicates, benefactive events and stance.

In the end, each ELAN file will contain a package of analysis of each data elicitation session with interlinked and comprehensive annotations that will enable the researcher to analyze the data from a holistic point of view and to discern the interplay of multiple aspects on a larger scale.

2.Human reference in Khalkha-Mongolian

This section discusses some preliminary results concerning the coding of human reference in Khalkha-Mongolian. The cross-linguistic analysis in SCOPIC workshops (Barth et al. submitted) has offered at least two interesting insights. First, languages/societies differ in strategies of human reference in discourse: some societies prefer using kinship terms from the start (`this one must be a grandfather; his son there...'); some societies prefer general terms with descriptions (`the younger man coming in'); some societies operate mostly with terms of professions/social roles (`this one must be a student coming home...'). Khalkha-Mongolian data clearly shows that it belongs to the first group: in almost all sessions, the participants start to describe the protagonists in the picture using kinship terms, thus presenting family ties as central in their understanding of society structure. Second, if speakers formulate reference through kinship expressions, it often means the presence of kinshipsensitive categories in a language's grammar (`kintax'). Conversely, a speaker of a language with at least one grammatical category sensitive to kinship uses a higher proportion of kinship formulations in human reference [1]. For human reference, Khalkha-Mongolian uses different types of linguistic devices at different levels:

- at the lexical level, there is a well-developed lexicon of kinship terms, social roles, and generic and descriptive reference;

- at the morphosyntactic level, there exists a set of descriptive syntactic devices, both simple, like `the one with the stick', and of different grades of complexity, e.g. relator constructions with possessive markers, participial relative clauses (commonly with the nominalizer xun `person'), etc.

- at the discourse-pragmatic level, pronouns and demonstratives are used for both deictic and anaphoric reference (Khalkha-Mongolian has no 3^rd person singular pronouns, so in this function demonstratives are used); the reference continuity in discourse frequently uses zero anaphora (pro-drop) and possessive marking.

The preliminary statistical analysis of one session shows the following results. Out of 338 instances of overt human reference in this data, the most common type consists of different terms denoting kinship (altogether 153 tokens, or 45%). Generic nouns with different attributes -- demonstratives, descriptive adjectives, participial relative clauses -- are the second most common type (altogether 107 tokens, 37%). Numerals (42 tokens) are used even more frequently than words for social roles (10 lexemes, 36 tokens).

Kinship terms differentiate between paternal and maternal relatives of the parents' generation. Lexically specified gender distinction is relevant for kinwho are older than EGO, e.g. Khalkha-Mongolian distinguishes between genders of elder siblings but not of younger siblings:

Parents are expressed by dyads `father mother', formal eceg ex and more intimate aav eez, the only kinship term with a formality opposition. Kinship terms are used almost exclusively with possessive marking, which can be either personal or reflexive, depending on the syntactic role of the term in the clause. In the case of personal possessive clitics, the overt expression of possessor in the same clause is not obligatory, their reference point can move with the discourse and has to be reconstructed by the hearers. Reflexive-possessive markers are obligatory. They mark all non-subject NPs within a clause that stand in a relation to the subject of the same clause, irrespective of whether it is overt or not.

For some words, the use of possessive marking differentiates between generic and kinship meaning:

(2) хуу xuu `boy' vs. хуунь xuu n' `(his/her) son'

охин oxin `girl' vs. охиннь oxin n' `(his/her) daughter'

нехер noxor `fellow' vs. нехернь noxor n' `(her) husband'

The few tokens without possessive marking demonstrate a non-referential use, be it predicative or attributive:

(3) Хоёр хуухэдтэй айл байна.

Xojor xuuxed-tej ajl baj-na

two child-COM family be-PRS

`This is a family with two children (a two-child-family)'.

Expressions for social roles in the present data include mostly professions, many of them formed with the agentive suffix -(g)c: tarialanc `peasant', gudamzny xudaldagc `street vendor', cagdaa `policeman',prokuror `attorney', xorigdol `prisoner', sorongijn xjanagc `prison guard'. Possessive marking here is rare, used only with two lexemes containing inherent relation in cases when this relation involves the main protagonist of the picture (`his guard', `his victim').

Descriptor devices can be single words or syntactic constructions. All descriptor lexemes in the present data have ane valuative meaning: xasir `someone who is experienced in a hard way', xoorxij `poor one', etgeed `person/organization (in legal terms)' or `fellow (with a negative evaluation)'. The absolute majority of descriptors are combinations of generic nouns with attributes -- adjectives or participial relative clauses; the most frequent generic noun xun `person' can already be considered a nominalizer alongside with yum `thing' for its inanimate counterpart. The differentiation of participles into past, present, habitual, etc. allows one to distinguish temporary and stable characteristics, cf.:

(4) with the habitual participle in -dAg:

a. уудаг нехдууд uu-dag noxd-uud `(regularly) drinking buddies',

b. тэр хавьд амьдардаг хумуус ter xav'd am'dar-dag xumuus `people who live in that surrounding';

(5) with the past participle in -sAn

a. холоос ирсэн хун xoloos ir-sen xun `the person who came from afar',

b. денгеж орсон хун dongoz or-son xun `the person who just entered (into the prison)';

(6) with the present participle in -(g)AA

ирж байгаа хун нь ir-z baj-gaa xun n' `the person who is now arriving (to them)'.

Possessive marking with this group is rare; the only example with 3sg/pl marker n' is given in (6), where the relation is established between the already discussed family group and the person arriving; the reading is `arriving to them' (see Figure 1).

Descriptions are usually based on the narrative content (zoduulsan etgeed `the beaten person') including the location of the referent in the individual picture (ard talynx n' xun `the person behind him'). There are practically no elaborated descriptions from the point of view of speakers (e.g. `the person you point at'), except for some evaluations (jamar sonin xun `what a strange person' -- strange for the speaker).

However, Khalkha-Mongolian has some demonstratives oriented on speech act participants, i.e. nogoodox `the one we already know', naadax `the one you mean'. Furthermore, Khalkha-Mongolian exhibits an anamnestic (recognitional) demonstrative nogoo (cf. Guntsetseg 2016: 38), which, among its many functions, is a crucial reference-tracking lexical device expressing different cognitional nuances in a certain social context. Occurring alone in (7) modifying a noun, it neutrally refers to an antecedent in the former discourse that could take place as much as even a week ago. However, in combination with cin ', an originally 2^nd person possessive marker that has e.g. additional focus marking function, it can express emphatic focus or mirativ- ity depending on the context (8):

(7) Негее хун энэ байна.

Nogoo xun ene baj-na.

DEM.ANAM person this be-PRS

`This is the person we talked about before (yesterday/last week/some day)'

(8) Негеедех чинь энэ байна.

Nogoodox cin ' ene baj-na.

DEM.ANAM 2SG/FOC this be-PRS

`(Look!) The person we talked before (yesterday/last week/some day) is here.' It is important to mention that these different types of human reference expressions participate actively in information structuring: the choice from the topic continuity scale (cf. Givon 1983:18) is mostly between (elaborated) noun phrases and zero anaphora, rarely pronouns (not counting possessive marking). For the latter, the data has shown that Khalkha-Mongolian persistently uses a pro-drop topic strategy [9] like other discourse-oriented pro-drop languages such as Japanese and Chinese [9; 8].

`(a) It is a family with two children. (b) The (lit. its) elder son has studied abroad, completed his study and comes home on his holidays, so it is. (c) Then his father and mother are there. (d) There is his younger brother. (e) Now his younger brother, what is he holding, what would you say?'

In this discussion of the first picture, “Homecoming”, both speakers agree that the picture shows a family with two children (repeated twice by speaker A, once by speaker B as acceptance of the first speaker's reasoning). Thus `family' was made topical for the following segment of discourse; the following 3poss marker indicates the relation between the marked NP (tom xuu n' `the big/elder son') and the topical NP, and thus the membership of the referent in the `family' group. In the next sentence, a topic change takes place: from now on `the big (elder) son' is the topic, so that in aav eez xojor n ' the 3poss marker refers not to the `family' anymore, but to the `elder son'. The 3poss in baga duu n' `small younger sibling' also refers to the `elder son' and not to the `family': in the latter case, it should have been baga xuu n' `the small/younger son (lit. boy) of the family'.

This shows that possessive marking in Khalkha-Mongolian, both reflexive and possessive, maintains identifiability of referents by establishing a net of interrelations between participants of the speech act and/or referents in the narrative.

Both personal possessive markers and possessive pronouns in Khalkha-Mon- golian show some interesting aspects of social cognition in their use. For instance, plural first person possessor marking (manaj/maan' `our') is often preferred over singular marking (minij/min ' `my'), even if the possessor is literally just one person, e.g. manaj exner `my wife'. This “plurality” can be explained by the way the family is perceived in the Mongolian culture: “I” am not alone, with respect to my family, and there is always a “we” implicated in any kinship relationship (cf. Barth et al. (submitted)). Additionally, speakers can use the words ax and egc to address or refer to a stranger who seemingly looks older than the speaker and with duu to a younger person, the latter often in combination with minij `my ', signaling empathy, protec- tivity and/or teaching attitude depending on the social context [3, p. 419-425].

As for the `kintax', Khalkha-Mongolian kinship terms, compared to common nouns, differ in their morphosyntactic behavior, e.g. in their combinatory patterns with possessive markers or in accusative and reflexive inflection [7, p. 58]:

(11) аавыгаа aav-yg-aa / ааваа aav-aa `father'

(12) гэрээ ger-ee / *гэрийгээ *ger-ijg-ee `yurt/home'

To sum up, this preliminary analysis of expressions of human reference in Khalkha-Mongolian offers some initial insights into the social-cognitive aspects that play a role for the grammatical structure of this language. The next step will be to proceed to analysis to detect a holistic social-cognitional pattern.

Summary

This paper has presented aresearch project that analyzes Khalkha-Mongolian linguistic resources from a new perspective, namely focusing on actual communication and its social-cognitive and cultural aspects, as a part of a larger international typological project. Not only has the new research focus on social cognition in the grammar of languages been introduced, as discussed in Barth and Evans [2], but an innovative data elicitation tool was also used [10].

The current state of the research work concerning the Khalkha-Mongolian data includes the creation of the basis for the data corpus (collection of the fieldwork material, its transcription, transliteration, glossing and English translation in ELAN) and analysis of one of the research topics -- human reference. At the next stage, the analysis of benefactive events, `private predicates' and stance is planned.

Acknowledgments: I am grateful to Elena Skribnik for helping me conduct the empirical field work and to analyze the data, and also for her valuable suggestions and comments on this paper. I also thank the audience of the 4th International Conference on Mongolic Linguistics for their useful comments.

References

1. Barth, D., Evans, N., San Roque, L., Mansfield, J., Gipper, S., Hodge, G., Rumsey, A., Arka, I.W., Bergqvist, H., Dickson, G., Dohler, C., Eka Pratiwi, I. D. P., Forker, D., Gast, V., Guntsetseg, D., Kashima, E., Kelly, B., Kimoto, Y., Knuchel, D., Kogura, N., Kurabe, K., Narrog H., Schalley, A. C., Schnell, S., Senge, C., Skribnik, E. and van Putten, S. (submitted). Frequency and grammar: The case of kinship.

2. Barth, D. and Evans, N. (2017): sCoPIC Design and Overview. In: D. Barth and N. Evans (Eds.), LD&C Special Publication No. 12: The Social Cognition Parallax Interview Corpus (SCOPIC): A Cross-linguistic Resource, 1-23.

3. Brosig, B. (2019): Terms of address and self-reference in Ulaanbaatar Mongolian. In: Bettina Kluge & Maria Irene Moyna (eds.) It's not all about 'you' -- New perspectives on address research. Topics in address research 1. Amsterdam: Benjamins: 415-433.

4. Evans, N. (2010): Dying words: Endangered languages and what they have to tell us. Wiley-Blackwell, Chichester UK.

5. Frith, C. D. and Frith, U. (2007): Social cognition in humans. Current Biology 17. R724-R732.

6. Givon, T. (1983): Topic continuity in discourse. A quantitative cross-language study. John Benjamins.

7. Guntsetseg, D. (2016): Differential case marking in Mongolian. Tunguso- Sibirica, Band 39. Wiesbaden: Harrasowitz Verlag.

8. Hasegawa, N. (1985): On the so-called “zero pronouns” in Japanese. The Linguistic Review 4, 289-341.

9. Huang, J. C.-T. (1984): On the distribution and reference of empty pronouns. Linguistic Inquiry 15 (4), 531-574.

10. San Roque L., Gawne, L., Hoenigman, D., Miller, J.C., Rumsey, A., Spronck, S., Carroll, A. and Evans, N. (2012): Getting the story straight: Language fieldwork using a narrative problem-solving task. Language documentation and conservation 6, 135-174.

11. Sloetjes, H. and Wittenburg. P. 2008: Annotation by category -- ELAN and ISO DCR. 6th International Conference on Language Resources and Evaluation (LREC 2008).

Размещено на Allbest.ru