Variation in three-participant constructions in Eastern Mansi.
In this article I will outline three-participant constructions in Eastern Mansi, concentrating on the variation in constructions with an animate Recipient. In other words, I will consider only the so-called caused possession- constructions, and the caused motion -constructions (for the definitions see Rappa-port Hovav, Levin 2008) will be excluded. I will show the frequencies of different constructions and the factors affecting the variation.
In the following sections I will consider two different ditransitive constructions in active voice and one in passive. The active constructions are called PO/SO construction (Primary Object--Secondary Object; see Dryer 1986) and DO/IO construction (Direct Object--Indirect Object; see Dryer 1986). Also the passive will be presented with and without an Agent. I will illustrate the most typical situations where each of these three constructions occurs, and present some conclusions concerning the correlation between pragmatic and syntactic functions: according to my data, each syntactic role (subject, object, oblique) is usually occupied by a certain information structural function (focus, primary topic, secondary topic).
Some details of the topic have been mentioned in earlier studies, but the variation between these three constructions has not been thoroughly examined in any linguistic study so far. One of the aims of my study is to fill in the lack of empirical data on the theme and create a comprehensive outline on the variation in three- participant constructions and the factors affecting it.
In Section 2 I will briefly introduce the Mansi languages and their most important typological features from the point of view of the three-participant constructions. My research data is presented in Section 3. In Section 4 I will outline the theoretical background to my study, and in Section 5 the most important previous studies on the theme. The results of my study are presented in Section 6, and the conclusions are gathered in Section 7.
2. Eastern Mansi language
Mansi (Vogul) is a Uralic language spoken in Western Siberia in the Autonomous district of Chanty-Mansijsk of the Russian Federation. It is a language distantly related to Finnish and more closely related to Hungarian: its closest relative is Khanty (Ostyak). Mansi and Khanty together build the Ob-Ugric branch of the Uralic language family. According to the Atlas of the World's Languages in Danger by Unesco, Mansi is a critically endangered language: only 24% of the 11,400 Mansi speak Mansi nowadays.
Eastern Mansi is one of the four main dialectical groups of Mansi, and it was spoken close to the Konda river. Nowadays all the Eastern dialects have vanished; Eastern Mansi is not spoken anymore. Eastern Mansi used to have four sub-dialects: Upper Konda, Middle Konda, Lower Konda and Jukonda. My data represents only Middle Konda which was spoken in the middle section of the Konda river.
The word dialect is often used in this connection, even though the differences between the main groups are quite remarkable, and the four main groups can be regarded as different languages. The dialects are not mutually intelligible. Grammars have been produced for both the Northern Sosva dialect and the Eastern Konda dialects. For that reason I will talk about the Eastern Mansi language and the Mansi languages in the following sections.
From the typological point of view, Eastern Mansi can be characterized as an agglutinative language. All the Uralic languages are rich in affixes: words are inflected and derived mainly with suffixes, not prefixes. In Eastern Mansi there are also postpositions, but they can be inflected with e. g. possessive suffixes as well. Verbs and nouns have three number classes: singular, dual and plural.
There are two verb inflectional categories: the subjective conjugation and objective conjugation. The subjective conjugation expresses only the number (Sg/Du/Pl) and person (1./2./3.) of the subject:
(1) suj-no koal-aam, suj-sot oos-om
moor-LAT step-1SG moor-lord be-1SG
'I step on the moor, I am the lord of the moor'
The objective conjugation expresses also the number of the object (Sg/Du/Pl):
(2) sos-mo uus koalt-os-to
moose-ACC again frighten-PST-SG<3SG
'He frightened the moose again'
Objective conjugation appears in all the Ugric languages and also in some other Uralic languages. Eastern Mansi can be classified as a DOM-language (DOM = Differential Object Marking, see Bossong 1985; Aissen 2003): only some of the direct objects are marked. There are several morphological devices used for marking the topical direct objects: both verb agreement and noun marking are involved in the coding of objects. However, Eastern Mansi does not represent the most typical type of DOM-languages presented by Aissen (2003), because animacy does not affect the variation. In my data, the animate and inanimate direct objects are marked equally: replacing an animate direct object with an inanimate one does not change the surface structure of the sentence. According to my data and earlier studies (see Skribnik 2001; Nikolaeva 1999), direct object marking is based on information structure: topical objects are marked, focal ones are not. My data shows that the topical direct objects are primarily verb marked: case-marking has rather a complementing function.
In all the Mansi languages passive covers a very active-like function: passive is not only for expressing impersonal action, but it is the inverse main category for the active voice (see Kulonen 2007 : 165). The most frequent passive construction is the so called personal passive, where the Agent of the event is also identified. In a two-participant passive clause Patient occupies the place of subject slot, and the predicate verb is inflected according to the number and person of the subject (Patient). A passive clause also often contains a lative-marked agent.
It is typical for all the Mansi languages that the most topical arguments are expressed explicitly only by a cross-reference to the verb (zero-anaphora). In (3) both the subject and the direct object are explicitly referred to only with a verb ending:
'(He) ate (it)'
Sentences without any nominal or pronominal subject or object constituent are fairly frequent: 85 % of all the sentences in my data lack a nominal subject constituent, and 35 % of the transitive active sentences lack a nominal object constituent.
3. Research data
My research material is collected from the folklore collection gathered by Artturi Kannisto in the beginning of 1900s and published some decades later (see Kannisto 1951; 1955; 1956; 1958; 1963), which can be considered as a vast and variable database representing different text genres. The collection consists of six parts including different text genres such as mythological texts, war songs, bear songs, tales and fate songs. The collection includes text patterns from all the dialectical groups of Mansi.
My data used for this article is gathered mainly from parts I and III of the collection. I have looked over all the texts (altogether 170 pages) written in the Middle Konda dialect in parts I and III and gathered all the sentences that indicate transitivity in terms of semantics. There are also some single examples from part II of the collection. I follow the definitions of transitivity by Hopper and Thompson (1980) and Kittila (2002): transitivity refers to the notion that action carries over from agent to patient. The whole corpus includes more than 1000 clauses, of which approximately 200 are ditransitive. Both active and passive clauses are included in the data.
The data shown in this article is converted to the phonetic transcription created specially by Kulonen for the needs of Eastern Mansi. This transcription was first presented in her Eastern Mansi grammar (see Kulonen 2007: 9-24). The original texts are written by Kannisto (see Section 3) with a Finno-Ugric Transcription (FUT).
4. Theoretical background
My approach to the topic is based on information structural analysis, especially the definitions and terminology stated by Lambrecht. Information structure is the formal expression of the pragmatic structuring of a proposition in a discourse (see e. g. Lambrecht 1994 : 5). Lambrecht emphasizes that with information structure there is talk about the "structuring of propositions into portions which a speaker assumes an addressee already knows" (Lambrecht 1994 : 7). In other words, with the needs of information structural analysis we can come up with how the sentences are structured basing on what the speaker assumes the addressee already to know as opposed to what they do not. There are two kinds of relation between denotata and proposition: the topic relation and the focus relation (Lambrecht 1994 : 335). Topic is often described with the word aboutness. Lambrecht (1994 : 188) has defined the topic as "the thing which the proposition expressed by the relation of sentence is about". Consequently, topic is an element that is already known: the speaker expects it to be familiar to the addressee. Focus is the exponent of new, non-recoverable information: something that the speaker does not expect the addressee to know already. According to Lambrecht (1994 : 207), focus is "the unpredictable or pragmatically non-recoverable element in the utterance".
In cases where we have several topics in one sentence, we can differentiate between primary and secondary topics. Nikolaeva (2001) introduces secondary topics in Khanty, the closest relative language of Mansi. She presents the notions of primary and secondary topics as follows: "The primary and secondary topic have essentially the same properties but two different degrees: the primary topic is more important, continuous and recurrent than the secondary topic" (Nikolaeva 2001 : 11).
There are certain difficulties concerning data gathered from a historic corpus and providing information structural analysis without any live discussion corpus. Petrova and Solf (2009) have discussed these questions and dispensed with the idea of assigning the main categories of topic and focus directly (Petrova, Solf 2009 : 144). However, despite Petrova and Solf's statements I have ended up following Nikolaeva especially and her terminology.
The aim of my study is to find out the factors affecting the variation between certain three-participant constructions. For that purpose using the main categories--primary topic, secondary topic and focus--gives sufficiently precise results.
5. Previous studies on three-participant constructions in Eastern Mansi
The variation between three-participant constructions has not been thoroughly examined, but the topic has been mentioned in connection to other questions in some previous studies. The following remarks by Skribnik and Kulonen support the results shown on the basis of my data.
5.1 Skribnik 2001
Skribnik touches on the topic of three-participant constructions in her article on the effect of information structure on object marking. According to Skribnik, the usage of the objective conjugation, the passive and the dative shift (for dative shift, see Section 6.1) in Northern Mansi has a pragmatic motivation and is a part of information structure marking (Skribnik 2001 : 222). All these three devices are closely connected to the variation in three-participant constructions.
Skribnik concludes the results of her study as follows: "The subject and the direct object in Northern Mansi are grammaticalized pragmatic roles: the subject is the primary clausal topic--Topic-1 [---] and the direct object is the secondary clausal topic--Topic-2, irrespective of their semantic roles. If any argument lower than Patient is a Topic-2, it is promoted to the direct object position" (Skribnik 2001 : 236). In this article I will show how these conclusions Concerning Northern Mansi by Skribnik can also be applied to the system of three-participant constructions in Eastern Mansi.
5.2. Kulonen 1989; 2007
Kulonen (1989) also mentions the question of three-participant constructions in connection to the passive voice. Kulonen states that in the Ob-Ugrian languages, one central function of the passive is the topicalization of another actant than the Agent: the given constituent is promoted to the subject position (Kulonen 1989 : 288). Applied to the three-participant constructions this means that whenever the Patient or the Recipient is the primary topic, it is promoted to Subject, and the whole clause is turned into the passive voice. The Agent is in the lative case.
(4) jag-otaan ontor-jaak-ong jalpong toagl-ol maj-w-os father-LAT3SG stomach-skin-ADJ sacred cloth-INSTR give--PASS--PST 'He was given a cloth of abdominal skin by his father'
Also in her Eastern Mansi grammar Kulonen emphasizes that in Eastern Mansi the passive is in more common use than in many other languages: passive is the inverse main category for the active voice, and also the Recipient or Beneficiary of a three-participant clause can be promoted to the subject of a passive clause (Kulonen 2007: 165). Later on, regarding active clauses and dative shift, she states that an also R-argument often takes the place of the primary argument (Kulonen 2007 : 54-55).
In the following sections I will introduce how the variation in three-participant constructions is based on the facts stated above. All the variation between three different constructions follows from the need of promoting the most topical element to the subject position.
6. Occurrence of the three-participant constructions in my data
6.1. PO/SO construction
PO/SO construction has also been called dative shift (see e. g. Kulonen 1989); both a direct object and an indirect object can be identified. The dative-constituent is promoted to direct object, and the Patient is marked with the oblique case. Applied to Eastern Mansi it means an accusative-inflected Recipient and an instrumental- inflected Patient:
(5) am naa-n tat-os-lom nee-l
1sg 2sg-sg2sg bring-PST-SG<1SG woman-INSTR
'I brought you a wife'
PO/SO construction has the highest frequency in my data with 45% of all the examples representing PO/SO. I could say that PO/SO is the primary active construction used in Eastern Mansi.
It is typical for an active clause that the Agent is the most topical element of the clause and occupies the place of subject. According to my data, PO/SO construction is used whenever the Recipient represents secondary topic, and due to the topicality of the direct object it is always supported by the objective conjugation. The Recipient is often referred to using zero-anaphora. The Patient is always a focus and marked with the instrumental. The principles are realized in examples (6) and (7):
(6) Soat lont woant-otaam joank-om-wooj-ol
seven goose flock-ACCSG3sG ice-PTPC-fat-INSTR
'He feeds his flock of seven geese with icy fat, with frozen fat.'
(7) aj-n-ol woar-os-to, tee-n-ol woar-os-to
drink-AKT-INSTR make-PST-SG<3sG eat-AKT-INSTR make-PST-SG<3sG
'She made him something to eat and drink'
To sum up, a in typical clause representing PO/SO construction the semantic, pragmatic and syntactic functions and the case marking correlate with each other as follows:
Table 1 Semantic Pragmatic Syntactic function: function: function: Case marking: Agent Primary topic Subject Nominative (0) Patient Secondary topic Direct Object Accusative Recipient Focus Oblique Instrumental
6.2. DO/IO construction
19% of my data represent DO/IO construction. DO/IO construction is also known as oblique strategy (Margetts, Austin 2007 : 402--403). In this kind of sentence the Patient occupies the place of the direct object and the Recipient is in the oblique case. In Eastern Mansi this means that the Patient is in the accusative case or unmarked (nominative), and the Recipient in the lative. The predicate verb is inflected in subjective or objective conjugation, depending on the nature of the direct object. Whenever the Patient is a focus (appearing in the nominative case), it is supported by subjective conjugation.
(8) om kurom lyox ak nag-noan tat-s-om
1SG three message only 2SG-LAT bring-PST-1SG
'I brought three messages just for you'. (Someone is knocking at the door and asks the housekeeper to let him in. The housekeeper tells him to go away, but he tries to say that he has brought a message for the housekeeper)
Also in the DO/IO construction the subject is always the primary topic. The prototypical situation of DO/IO construction is that of where the Patient represents secondary topic and occupies the place of direct object, whilst the Recipient is a focus and occupies the place of oblique. Also this construction is often (but not always) supported by objective conjugation.
(9) moot soon-toagol keelop-mo wo-s-to,
other bowl-full blood-ACC take-PST-SG<3SG
koop-posom-ot puw-otaan tow-mo-s-to
boat-stern-LOC son-LATSG3SG PREF-give-PST-SG<3SG
'He took the other bowl full of blood and gave it to his son at the stern of the boat'
(10) ton kuuly-toot-poal-mo eeko
that smock-sleeve-half-ACC woman
wisy-kom-no kuuly-tagl-ii junt-os-to
young-man-LAT smock-full-TRANSL sew-PST-SG<3SG
'The woman resewed the one sleeve of the smock into a full smock for her son'
However, with this table we can demonstrate only the very prototypical situation: my data also contains examples that do not fit the proto
Again we can see certain correspondences between the semantic, pragmatic and syntactic functions, as shown in table 2:
Table 2 Semantic Pragmatic Syntactic function: function: function: Case marking: Agent Primary topic Subject Nominative (0) Patient Secondary topic Direct Object Accusative Nominative (0) Recipient Focus Oblique Lative
However, with this table we can demonstrate only the very prototypical situation: my data also contains examples that do not fit the proto-type. Examples (11)-(12) are supported by subjective conjugation, and the Patient is a focus:
(11) pas-ong-kom jag-aanoll-no soat sor
light-ADj-man father-SG3PL-LAT seven kind
'They shouted [sacrificed] seven kinds of blood sacrifices to the bright man, their father'
In (11) both the Recipient and the Patient are foci. The agent is the only topical element in the clause and naturally occupies the place of subject. In the story people are busy with other things until they start sacrificing to their gods; the whole phenomenon of sacrificing is a new element brought to the discussion. Neither the god (bright man), nor the seven sacrifices has been mentioned before.
(12) om-noan pol eep-ong oano, eep-ong toas oat wott-aat
1sg-LAT PARTIC steam-ADJ bowl steam-ADJ plate neg put-3PL
'They do not give (in sacrifice) any steaming bowl, nor any steaming plate for me'
Also in (12) both the Recipient and Patient are focal: even though the Recipient is expressed with a first person personal pronoun, it is an element just brought to the discussion. The division of the pragmatic functions does not support the prototype shown is table 2, and that is also why the syntactic structure differs from the prototypical situation.
36% on my data represent the passive. In my data a case in which the Patient of a three-participant sentence would be promoted to subject does not appear; it is obvious that whenever the Patient is the primary topic, it is turned to the subject of a PO/SO construction (see Section 6.1). In all the three-participant passive constructions the Recipient is promoted to subject, and the Patient is marked with an oblique case. The predicate verb agrees with the person and number of the Recipient.
(13) koatom pol was eel-oolii meen-k moo-l law-w-aamon
sure PARTIC SUPERL before 2DU-STRESS land-NSTR tell-PASS-1DU
'Certainly we'll first be provided with land'
According to my data, the passive construction is used whenever the Recipient is the primary topic and needs to be promoted to subject position. The Patient is always in instrumental case. The possible Agent is in the lative case. In 75% of the cases in my data a passive clause includes an Agent:
(14) nee-taan jar-ol maj-w-os
woman-LATSG3SG scraper-INSTR give-PASS-PST
'He was given a scraper by his wife'
In 25% of the examples the Agent is not identified:
(15) alon-seetop-manyto, sagrop, panlog-oostyor, jiiw-oosymosy, soat lont
silver-rope-ball axe hemp-whip wood-key seven goose
woant joank-om-wooj-ol, pooly-om-woojol toxt-w-ot
flock ice-ADJ-fat-INSTR freeze-PTCP-fat-iNSTR feed-PASS-3PL
'The silver rope ball, the axe, the hemp whip, the wooden key and the flock of seven geese are fed with icy fat, with frozen fat'
To sum up, there is certain correlation between pragmatic and syntactic functions. The syntactic functions form a hierarchy, and the more topical the constituent is, the higher position it gets in the syntax hierarchy. Skribnik (2001) has stated that in Northern Mansi the most topical element always occupies the place of subject. Further, Nikolaeva (1999) has stated that in Khanty focus is encoded as an oblique, while the secondary topic is encoded as the direct object. We can see exactly the same features in Eastern Mansi: the current clause structure is always dependent on which element is the most topical one (primary topic) and receives the place of subject. Further, the second most topical (secondary topic) is always the direct object, and focus is expressed with an oblique case. The alternation between active and passive voice is purely motivated by the nature of the primary topic; primary topic occupies the place of subject independent of its semantic function. Consequently, we can describe the pragmatic variation as follows (table 3):
Table 3 PO/SO (45%) DO/IO (19%) Passive (36%) Agent Primary topic Primary topic Secondary topic SUBJECT SUBJECT OBLIQUE Patient Focus Secondary topic Focus OBLIQUE OBJECT OBLIQUE Recipient Secondary topic Focus Primary topic OBJECT OBLIQUE SUBJECT
In the table we can find the prototypical situations, and as can be seen in the data presented, there are also examples that do not fit the prototypes. With table 3 I have described the most typical cases in general; a three- participant clause is expected to contain one primary topic, one secondary topic and one focus. The other models are excluded from the table. Thus, the existence of such sentences that contain two secondary topics or two foci is natural. They are not exceptions, just less prototypical and less frequent cases.
However, the same principle is applicable to all of my data; the more topical the constituent is, the more central a role it gets in the sentence structure. The basis of the system can be described with simple correlations between the semantic pragmatic and syntactic functions. All the data do not fit the prototypes described in the tables, but they do fit the principle concerning the correspondence between the topicality level of the constituent and its role in the sentence.
Finally, the results of my study support the earlier statements concerning the topic. As both Kulonen and Skribnik have presented, topicality is the most remarkable factor affecting the variation in three-participant constructions. Within this study I have provided a more comprehensive description of the whole phenomenon, including also the areas examined in the previous studies.
Debreceni Egyetem. University of Debrecen
acc--accusative, adj--adjective derivater suffix, du--dual, akt--aktionsart, instr--instrumental, lat--lative, neg--negation particle, partic--particle, pass--passive, pl--plural, pref--prefix, pst--past tense, ptcp-participle, sg--singular, stress--stressed form of a personal pronoun, superl--superlative
The objective conjugation suffixes are glossed as follows: [number of object] < [person of subject][number of subject], e. g. SG<1SG, DU<3SG.
The possessive lative and possessive accusative forms are glossed as follows: [case][number of possessed][person of possessor][number of possessor], e. g. LATSG1SG.
Aissen, J. 2003, Differential Object Marking: Iconicity vs. Economy.--Natural Languages & Linguistic Theory 21, 435-483.
Bossong, G. 1985, Differentielle Objektmarkierung in den neuiranischen Sprachen. Empirische Universalienforschung, Tubingen.
Dryer, M. S. 1986, Primary Objects, Secondary Objects, and Antidative.--Language 62, 808-845.
Hopper, P. j., Thompson, S. A. 1980, Transitivity in Grammar and Discourse.--Language 56, 251-299.
Kannisto, A. 1951, Wogulische Volksdichtung I. Texte mythischen Inhalts.
Bearbeitet und herausgegeben von Matti Liimola, Helsinki (MSFOu 101).
--1955, Wogulische Volksdichtung II. Kriegs- und Heldensagen. Bearbeitet und herausgegeben von Matti Liimola, Helsinki (MSFOu 109).
--1956, Wogulische Volksdichtung III. Marchen. Bearbeitet und herausgegeben von Matti Liimola, Helsinki (MSFOu 111).
--1958, Wogulische Volksdichtung IV. Barenlieder. Bearbeitet und herausgegeben von Matti Liimola, Helsinki (MSFOu 114).
--1959, Wogulische Volksdichtung V. Auffuhrungen beim Barenfest. Bearbeitet und herausgegeben von Matti Liimola, Helsinki (MSFOu 116).
--1963, Wogulische Volksdichtung VI. Schicksalslieder, Klagelieder, Kinderreime, Ratsel, Verschiedenes. Bearbeitet und herausgegeben von Matti Liimola,
Helsinki (MSFOu 134). Kittila, S. 2002, Transitivity. Towards a Comprehensive Typology, Turku (PhD thesis. University of Turku).
Kulonen, U.-M. 1989, The Passive in Ob-Ugrian, Helsinki (MSFOu 203).
--2007, Itamansin kielioppi ja teksteja, Helsinki (Apuneuvoja suomalais-ugrilaisten kielten opintoja varten 15).
Lambrecht, K. 1994, Information Structure and Sentence Form, Cambridge.
Margetts, A., Austin, P. K. 2007, Three-Participant Events in the Languages of the World: towards a Cross-Linguistic Typology.--Linguistics 45, 393-451.
Nikolaeva, I. 1999, Object Agreement, Grammatical Relations, and Information Structure.--Studies in Language 23, 341-86.
Petrova, S., Solf, M. 2009, On the Methods of Information-Structural Analysis in Historical Texts. A Case Study on Old German.--Information Structure and Language Change. New Approaches to Word Order Variation in Germanic, Berlin (Trends in Linguistics. Studies & Monographs 203), 121-160.
Rappaport Hovav, M., Levin, B. 2008, Reflexions on Manner/Result Complementarity.--Syntax, Lexical Semantics, and Event Structure. Part I, Oxford (Studies in Theoretical Linguistics), 21-58.
Skribnik, E. 2001, Pragmatic Structuring in Northern Mansi.--CIFU IX. Pars VI, 222-239.
SUSANNA VIRTANEN (Debrecen)
|Printer friendly Cite/link Email Feedback|
|Date:||Jun 1, 2012|
|Previous Article:||Salis-livisches Worterbuch.|
|Next Article:||Ago Kunnap 70.|