Subject-Object versus Topic Agreement in Erzya (myv)

The Erzya language is known for its complex person-agreement in both the function of predication and that found in the discourse participant/non-participant identification strategies of possession. Predication person agreement, which I will occasionally be refered to as Trajector person agreement, on the one hand is not limited to finite verbal conjugation alone, rather it is manifest in other word classes, Nouns, Adjectives, Pronouns, Non-finites, etc. Possessor person agreement, or what I will often refer to as Landmark person agreement, on the other hand is not confined to noun anchoring, rather it anchors on other parts of speech, as well, Pronouns, Adpositions, Non-finites, etc. Hence person agreement in Erzya can be expected to be manifest in the text, whether that be marking for predication, discourse related identification strategies or combinations of both.

The relatively recent typological study Intransitive Predication by Stassen (1997) is a very good stepping stone for beginning the fundamental contemplation of Erzya Predication Person Agreement, namely the scope is narrow enough to draw ones attention to Erzya diversity. If we contemplate Erzya from this point of departure where (1) the domain is established as the subset of Intransitive Predications with one definite argument (Bisang, W. 2002: 11; Stassen 1997: 11), and (2) a four-way subcategorization made with (2a) Locational, (2b) Event and Action/State, (2c) Class-Membership, and (2d) Property-Concept Predication (Bisang, W. 2002: 11; Stassen 1997: 15-21), many specifics of Erzya will come to the fore.

How does one go about locating a definite argument for the Intransitive Predications. Both Proper and Common Nouns can take Determinate or Possessor agreement markers, absence of these in the former is typical of a definite referent, whereas their absence in the latter means at least non-deictic, non-associative and in the modern Erzya language indefinite. In other words, Proper Nouns might easily be found, but common ones must always be filtered for definiteness.

How should a loose category of Location be applied to a language where spatial predication is manifest in three subsubcategories, namely (a) source, (b) single-reference-point locative and (c) multiple-reference-point locative. Source predication, although more widely abound in its sister language Moksha, is also attested in Erzya, where the Elative and Ablative cases are its morphological markers. The concept of Source Predication would also entail [+time stable] status which would in turn move it closer to Class-predication types than the Locative-Spatial Predications cf. Bisang quoting in (Bisang, W. 2002: 12; Stassen 1997). The locative-spatial predications of (b) and (c) can be readily split with the assumption that (c) the multiple-reference-point locative, corresponging to Prolative case marking, can only be used with plural trajectors, whereas (b) the single-reference-point locative, corresponding to Inessive and Locative case marking, is associated freely with trajectors of either number. It would be interesting to see whether there is a difference between the specific single-reference-point locative and the less specific multiple-reference-point locative; do these align separately on the time-stability scale.

In Event, Action/State predication, Erzya requires a division between finite and non-finite verbs, whereas finite verb predication follows the pattern of accusative languages and the non-finite that of ergative languages (cf. Anderson, S. On the Notion of Subject in Ergative Languages. In Subject and Topic, edited by Charles N. Li. 1976). Contemplation of verbal predication therefore leads us inadvertently in the direction of Single versus Multiple argument relations being realized in predication structures -- something applicable to relative nouns, spatio-temporal adverbs, and adpositions.

Class-Membership and Property-Concept Predications are, at times, difficult to distinguish, i.e. Erzya does not have morphological comparison, which might reveal a group of relative adjectives. It does, however, employ the Indeterminate Genitive with nouns to derive what I will call here an Associative-Domain modifier, whereas the same Genitive on an adjective would render an indefinite, non-specific [+human] referent. This, of course, would only be applicable to adjectives used to modify [+human] heads. Evidence for adjectives used in this way for [-human] heads has yet to be found. Lexemes indicating ethnic background versus class-membership are difficult to distinguish, which is typical of Erzya where toponyms are readily used in the plural to indicate the residents of a given place. Analogically singular usage, especially with determinate or possessor agreement marking, is just as prevailent and frequent as it might be in the English words Seattlite and Kentite corresponding to the city of Kent and the near-by city of Seattle.

What then does the concept of Intransitive Predication have to offer an analysis of Erzya. First it brings to the fore a need to differentiate single versus multi-argument relations, and by so doing to delineate a semantic alignment for the different parts of speech on the basis of the following criteria: (a) can they act as anchors of person agreement, (b) is there a scale Landmark versus Trajector agreement predominance.

For Purposes of contemplating person agreement in Erzya let us assume a set of semantic relations with correlates or anchors in the finite set of Erzya lexemes and their concatenatible combinations for which either the Landmark, the Trajector or both correspond to real-world referent.

In the following text Implicit Landmark means that no syntactic or morphological Landmark is present. Whereas Tacit Landmark, although represented by morphological marking could just as well be indicated syntactically. The nature of this paper, however, focuses on person agreement in the morphological realm.

It should be noted that kindred terms such as 'sister' are definite by nature, and therefore their inadequate identification renders a constituent considered odd or unacceptible by the speakers of a given language. There might even be a continuum of felicity frequency within a language, which renders the question in (1a) more acceptible than that in (1b).

(1a) Who here is a mother?

(1b) Who here is a sister?
A person answering should (a) be a female, and (b) have one or more siblings.

In Finnish the analogical questions stated in (2a) and (2b) show that (2b) is considered at least problematic. Moreover (2b) sparks a discussion of the actual intent of the question, see (3a) vs (3b) and (3c).

(2a)
Kuka täällä on äiti?

(2b)
Kuka täällä on sisko?

(3a)
Kenellä on sisko?
'Who has a sister?'

(3b)
Kuka on jonkun sisko?
'Who is somebody's sister?'

(3c)
Kenellä on sisarus?
'Who has a sibling?'

The problem faced in (3) is how to coerce a naturally definite noun into indicating an indefinite non-definite entity. Obviously, (3a) loses the gist of the intent by retaining the indefiniteness of a concievable referent in the structure of an introductory predication, and shifting the reference to a different entity. In (3b) retention of indefiniteness is achieved through use of an indeterminate possessive modifier. Finally (3c) offers an entailment relation, which by content is answered by the wrong people, as well.

Thus in the 'older sister' analysis we must make the initial assumption that Implicit Landmark is pragmatically low in frequency.

The so called long-form present participle is applied in derivation to render deverbal Actor nouns. Since the semantics of the verb entail [+human] characteristics for Agent and sometimes Patient-Theme, it can be assumed that this verb is suited to an Ergative alignment parameter involving Landmark-Possession and Trajector-Predication.

THIS FAR

Finally, within the realm of non-finite structures Proper nouns can be coerces to operate as anchors for Class-Membership predication. can be both definite and indefinite, whereas proper nouns can also take Determinate or Possessor agreement markers. the author deals exclussively with the notion of In addition to what is traditionally considered subject and object agreement in the finite verbs, there are zero-copula functions involving non-finite verbs, and the notions of with subject and object paradigms involving the first, second and third persons. Documentation of finite and non-finite/non-verbal predication in Erzya dates back to the very first grammar of the Erzya language in 1838 by Gabelentz based on the Erzya translation of the Gospel in 1821.

More specifically, finite conjugation applies subject marking for transitive and intransitive verbs alike, with an additional option of marking the definite objects of transitive verbs. According to Moravcsik (cf. Givón,T. 1976: 151-2; Moravcsik, E. 1974), subject and definite object marking could be the two most prevalent types of agreement marking, whereas referential salience of [+human] characteristics in proper names is a matter observable in the object NP correlate and not the person-agreement marking of the finite verb itself. (cf. Givón,T. 1976: 151-2; Moravcsik, E. 1974).

Givón's thesis (Givón, T. 1976: 151) is that grammatical agreement stems from topic agreement, which has later been reanylized as subject or object agreement. In Erzya it would appear that pure subject agreement marking does not entail topic agreement, whereas the fusional type object agreement markers do restrictively align the object with topic agreement. Hence subjects in Erzya can be both definite and indefinite, whereas only definite objects can be correlates of object agreement.

In the syntactic structure of Erzya, a pro-drop language with multiple argument agreement, definite objects are not always attestable. In fact, the subject and object NPs can be regarded as correlates of the argument marking strategy on the finite verb, and therefore their physically presence in a contextual predication is not obligatory. If the person agreement marking on the finite verb is sufficient in some contexts, but not laboratory sentences, how should the syntactic subject and object be characterized.

The NP subject correlate of subject agreement in Erzya should be dealt with on a basis of clarification of the referent, namely on the cline common noun / proper noun -- pronoun -- zero the pronoun would appear to be no more referential than the zero.

In a discussion of SOV versus SVO with regards to Erzya, it would be far more pertinent to discuss matters of topic, i.e. topicality of subject, object or predication. In the following quotation it will be observed that a topic object can normally precede the finite verb.

A closer scrutiny of context, however, might reveal that pronoun use is closely related to the concept of anaphoric proniminalization, i.e. there is an introductory discourse structure providing a focused element directly before the pronoun is used.

When contemplating the syntactic subject and object of an Erzya predication, it is necessary that we draw a line between the finite verb, the subject and the object. In a sense these subject and object NPs then provide referential clarification for the argument slots manifest in the valency of the individual verb.

constituent. Since the verb in Erzya represents the core of an event relation and the subject and object are seen as elements providing referential clarification for argument slots, it is necessary that this treatise of person agreement consider the referential essence of subjecthood and objecthood.

In Erzya, the syntactic subject, the constituent triggering person agreement in the finite verb, can be divided into two groups, which are morphologically and semantically consistent. On the one hand there are the "-voms/-vems" passive verbs, which seem to reduce volitional properties of a verb and simultaneously emphasize the telic-aspectual information. Hence some intransitive verbs, such as 'to.fall', 'to.die', 'to.drown/to.sink', are not given to "-voms/-vems" passivization -- there is already no volition to be lost. Transitive verbs, however, show a promotion of the patient-theme to syntactic subject and the demotion of volitional agent to oblique dative. By subtraction, we are rendered the group of prototypical subjects, which is what I intend to speak of here.

The syntactic subject, the correlate of the subject-agreement marker on the verb in Erzya, is in the Nominative. It can be singular or plural; determinate, indeterminate or marked as a possessum of any of the three persons singular or plural.

In the Agent-Patient markers there is ambiguity present whenever (A) the object referent is first or second person, and (b) either the Agent or the Patient or both is non-singular.

1. Object referent is first or second person
2. Agent or Patient or both is/are not in the singular

In Topic, Pronoun, and Grammatical Agreement (Givón, T. 1976: 151-2) the author discusses the importance of distinguishing between subject-verb agreement and topic-verb agreement. Hence there should be an acknowledgement of the different grammatical processes involved in pronominalization versus person agreement on the verb itself.

In the confines of a sentence, it should be noted that a mere statement to the effect of Erzya's being a pro-drop language is no sufficient. First of all, Erzya has simple Subject predication marking for not only finite verbs, but other constituents, as well, i.e. non-verbal expressions of qualities, class membership, location and source can also be marked for Subject agreement. Secondly, when dealing with Subject-Predicate correlation, individual treatment should be given to the assessment of the Subject on one hand and the Predicate on the other. Hence the syntactic entities of the Subject will be divided three ways:

1. NP
2. Pronoun
3. the empty set

The Syntactic empty set, however, can correlate with (a) an implicit referent, previously introduced, or (b) a 0-refent, e.g. expressions of weather.

Person agreement marking on the finite verb appears to be subject agreement in intransitive predications, whereas the situation becomes more complicated in fusion-type Subject-Object marking of transitive predications and predications involving non-verbal constituents. In Subject-Object marking of transitive predications becomes ambiguous whenever the following two statements hold for the predication arguments (a) the Object is 1st or 2nd person and (b) at least one of the arguments Subject/Object lacks the specific reference of Singular.

, it should be stated that a subject finite-verb relation ought to be described in at least three different situations the Subject correlates to an endophoric constituent: (a) an NP, (b) a pronoun, or (c) {the empty set}. There is the third-person dimension in which the subject NP clearly indicates a referent, i.e. the subject is a proper of common noun.

Since evidence can be given from Erzya in support of Subject-agreement marking on the verb, whereas regardless of whether the Subject is topic or not the finite is always manifest in person agreement for the Subject.

Object agreement proves more difficult, i.e. in Erzya Object agreement implies definiteness of the object, but not all definite-determinate marked objects have corresponding object-agreement on the finite verb.

1. Definiteness of Object
2. Generic Object
3. Perfect aspect of Event Predication
4. State Predications (sodams 'to.know', vetʂkems 'to.love', ???)

Keresztes (1999: 11-12) summarizes the requirements of the Mordvin Object conjugation as (a) definiteness of the object, (b) and aspect of the verb, and that there are also differences between the Moksha and Erzya usages (cf. Alhoniemi, 1996). On the other hand mention of Hunfalvy, possibly, finding fault with the use of subjective forms where an anology of Hungarian objective forms would have been expected shows that some of the information has been neglected. Namely, there are instances in the Erzya Gospell text where 6Obj3 marking is used where the object ought to be 1st or 2nd.

Отвѣчасть тянзе: киякъ езизь сиведь минекъ, кортли сынянстъ: моледе тынь гакъ виноградонь садсъ монь, и мязе еряви, максанъ тынянкъ.

Abramov_K_1964:
Пахом Васильевич кандсь начко саконть , Дракин -- кукшононть .
Атясь мольсь мельгаст ды потясь мадезь цилименть .
Кие учось пиземе , -- витсь прянть Дракин .
Кеветее иесэ тейтерькась уш лови эсь прянть покшокс , ялатеке а чарькоди , мекс авардп авазо .
Аштесь толтомо , таргась ды ледстнесь ютазенть .
Эзь марявгак , кода чиизе кемгавксово вайгельбенть .
Мезе сынь теить ведьгемень ломанть губантнэнь каршо ?
Пивтиль стака арсематнень потсонзо , киненьгак сонзэ эзь чамдо .
-- Икеле неть цётамотнень тензэ цётамальть, аволь ней, зярдо тевесь теезь .
Сон авань седейсэ марясь ды чарькодсь , коли цёрась кадовсь совхозс меельсе чинтькак , кадовсь кинь-бути кисэ. Те кие-бутись , нама , кода сон арсесь , ульнесь Галя .