The Contemporary Use of DOM in South-Western Macedonian Dialects

Размещено на http://www.allbest.ru

Размещено на http://www.allbest.ru

The Contemporary Use of DOM in South-Western Macedonian Dialects

E. Buzarovska



Buzarovska Eleni - PhD in Philology; Professor, Faculty of Philology “Blaze Koneski”, Ss Cyril end Methodius University in Skopje, Republic of Macedonia

Ss Cyril end Methodius University in Skopje

The paper discusses the preposition-based pattern of differential object marking in the south-western Macedonian dialects of Ohrid and Struga. The author hypothesizes that the explanation of this synchronic phenomenon should be based on pragmatic-semantic criteria. The results from the analysis indicate that the preposition na marks a direct object only if its definite and animate referent is ascribed an important communicative status in spoken discourse.

Key words: direct object, animacy, discourse prominence, Balkan languages.



Бужаровска Элени - доктор филологических наук; профессор кафедры английского языка филологического факультета имени Блаже Конеского, Университет Святых Кирилла и Мефодия в Скопье, Республика Македония

Университет Святых Кирилла и Мефодия в Скопье

Дифференцированное маркирование объекта в юго-западных македонских диалектах Охрида и Струги

В статье обсуждается дифференцированное маркирование объекта при помощи предложной конструкции в юго-западных македонских диалектах Охрида и Струги. Автор выдвигает предположение, что объяснение этого синхронного явления должно основываться на прагматическо-семантических параметрах. Проведенное исследование показывает, что предлог на маркирует прямой объект, если его определенному одушевленному референту приписывается важный коммуникативный статус в разговорном дискурсе.

Ключевые слова: прямой объект, одушевленность, коммуникативный статус аргумента, балканские языки.



Introduction

This paper examines a peculiar dialectal feature of some peripheral Macedonian dialects absent in other Balkan Slavic languages: a direct object construction involving the preposition na as in (1). In other dialects and in standard Macedonian, the sentence in (1) is without na (Go gledam Marko).

The use of na with a direct object (1) instantiates a typological feature known as differential object marking (henceforth DOM). DOM is a rather common cross-linguistic phenomenon in over 300 languages [Bossong, 1984; de Swart, 2007], among which are Spanish, Turkish, Persian, Syrian, Hindi, Hebrew, Mongol, Tagalog, etc. Romance languages employ a prepositionbased strategy: a in Spanish [cf. von Heusinger, Kaiser, 2003] and pe in Romanian [Chiriacescu, von Heusinger, 2010]. The pe-marking has been borrowed by the east Bulgarian dialects in Romania [Mladenov, 1993]. A similar prepositional strategy with definite direct objects is used in southern and south-western Macedonian dialects [Koneski, 1986].1 The presence of this typological feature in these dialects was described in Balkan scholarship by Cyxun (1981), Koneski (1986), Topolinjska (1995), Asenova and Aleksova (2008) among others, and was also mentioned in dialectological studies on a particular local dialect from these two areas [Tosev,1970; Shklifov, 1973; Peev, 1979, 1987; Vidoeski, 1984, 1998; Karanfilovski, 1992; Labroska, 2003].

However, this phenomenon has been mainly studied in the context of language contact without taking into account the research on DOM and the respective literature; thus the first use of this term in Balkan Slavic scholarship is found in Adamou (2006). Traditionally, it was described under the term “accusative иа-construction” [Topolinjska, 1995, p. 93] in Balkan Slavic or “l'objet direct prepositionnel” [Niculescu, 1959] Definite direct objects in standard Macedonian and in central and western dialects require the obligatory use of pronominal clitics, but they may be optional in eastern and southern dialects. I am grateful to the anonymous reviewer for pointing out that this term appeared earlier in the Romance tradition in this source. and “nota accusativi personalis” [Asenova, Aleksova, 2008] in Balkan Romance. A number of linguists underline the novelty of this construction: Koneski (1986) refers to it as an “interesting innovation”, Pencheva [Pencheva, 1998, ф p. 301] as a radical innovation within Slavic languages.

Among the reasons why this innovative construction belongs to typologically relevant properties of Slavic languages is the fact that it makes use of the grammaticalized preposition na,1 which has several functions in caseless Balkan Slavic languages (Bulgarian and Macedonian): it codes the internal possessor (kolata na Marko `Marko's car') within a noun phrase and functions as a dative marker marking the recipient in transfer constructions (mu zboruvam naMarko `I am talking to Marko'). But in some peripheral Macedonian dialects the dative na has been copied in the accusative relation, marking the second human participant in a transitive clause. This led to the neutralizing of the formal opposition between direct and indirect construction, which is yet another manifestation of the instability of dative forms in Balkan languages [Topolinjska, Buzarovska, 2011; Asenova, Aleksova, 2008].

The goal of this paper is to provide an explanation for the contemporary use of DOM in south-western Macedonian dialects of the Ohrid-Struga region (Republic of Macedonia). These dialects are particularly interesting for Balkan linguistics because “the most Balkanized dialects of the five language groups are to be found in Central Balkans, around the lakes Ohrid and Prespa” [Lindstedt, 2014]. Both Macedonian The primary sense of the preposition na `on' is spatial (na masata `on the table'), but it has numerous extensions in temporal and metaphorical domains (na pladne `at noon', na zdrayje `to health'). DOM is not used by all speakers of this dialect: it is not found in ethnic Macedonian villages near Ohrid (e.g.Konjsko, Trpejca, etc). (2a) and Aromanian Except for the Aromanian dialect in the village of Krania (Thessaly, Greece) described in Sobolev (2008) and several villages in Epirus (Greece) [Asenova, Aleksova, 2008] object marking is not registered in other Aromanian dialects (Petar Atanasov, p. c.). speakers (2b) in these two towns (and some surrounding villages) on the shore of Ohrid lake may mark animate and definite objects [Markovik, 2007].

a. Go vidov па Такі.

b. L' vizdui pi Такі.

cl.3msg.acc see.PST.lsG dom Taki

`(I) saw dom Taki.'

The presence of DOM is also attested in the neighboring Debar region in the speech of the Macedonian-speaking Moslem population (3).

ke a darzat na nevestata

FUT CL.3fSG.ACC hold.PRS.3PL DOM bride.F:DEF

`(they) will keep the bride'. Their language is also characterized by some archaic features [Vidoeski, 1998, p. 225] such as the optional use of synthetic dative forms of masculine proper names (mu rece Stojanu `s/he told Stoyan'), and the synthetic genitive-accusative forms of these names (go vide Stojana `s/he saw Stoyan').

In view of the fact that the ««-marking in Ohrid and Struga dialects is restricted to definite and predominantly human objects it is logical to define it as semantically-driven DOM caused by a deviation from the transitive prototype [Comrie, 1989]. On the prototype approach, the subject position is reserved for an animate (human) agent, while the object is an indefinite and affected inanimate patient [Comrie, 1989, p. 128]. A patient animate object diverges from the prototype of a transitive clause [Comrie, 1979, 1989; Aissen, 2003] because its semantic features are typical of subjects. To avoid a possible confusion with the subject, some languages case-mark atypical objects differently. DOM on this approach exemplifies a “grammaticalized way of drawing the listener's attention to the non-canonical mapping of argument positions of a transitive construction” [Lyutikova, Ronko, Zimmerling, 2016, p. 119].

However, not all definite and human objects are ««-marked in Ohrid and Struga dialects. The variations in marking cannot be fully accounted for solely by semantic factors and require additional inclusion of pragmatic factors related to information structure. We believe that DOM in these dialects is triggered by the interplay of pragmatic and semantic factors: the pragmatic criterion of discourse prominence applies if semantic preconditions related to animacy and definiteness are satisfied. For this reason, the theoretical basis of our analysis presented in the next section encompasses the approaches to DOM involving both semantic and pragmatic properties of marked objects. It is argued that DOM is triggered by information structure role of definite animate direct objects. The speakers of these dialects use DOM to assign a prominent discourse status to the second definite (human) participant in a transitive situation.

The main hypothesis of the paper - that the contemporary use of DOM in these dialects is regulated by the interaction of semantic and pragmatic principles - is supported by empirical data obtained by the questionnaire involving a sample of 150 speakers from Ohrid and Struga. The data are interpreted with reference to the animacy and definiteness hierarchies [Croft, 2003] and to information structure [Dalrymple, Nikolaeva, 2011, p. 14], which relates casemarking to topicality. The analysis of questionnaire results aims to determine whether the above semantic and pragmatic factors potentially condition DOM in these dialects. It is assumed that DOM is applied only to topical objects with certain semantic features, objects that are I pragmatically prominent.

The paper is organized in six sections. After providing the theoretical basis of our investigation in the next section, section 3 gives a brief overview of DOM in Macedonian dialects. Section 4 discusses the possible reasons for the emergence of DOM from a diachronic and synchronic perspective.

The central part of the paper, section 5, presents the empirical study on the use of /-/«-objects in south-western dialects. The main points of the analysis are underscored in the conclusion.



Theoretical considerations

Semantic explanations of DOM are based on the notion of an argument-triggered DOM in which “the properties of the differentially marked argument alone are responsible for a particular marking” [Witzlack-Makarevich, Serzant, 2017]. These authors distinguish between inherent and non-inherent DOM-triggering properties. The inherent lexical semantic features are intergrated into several implicational hierarchies or scales: animacy, person, uniqueness, number. They predict that objects higher on these hierarchies tend to be case-marked. The animacy hierarchy [Silverstein, 1976; Dixon, 1979; Croft, 2003], Croft's hierarchy, based on the hierarchies in Silverstein (1976) and Dixon (1979) also includes definiteness. which follows the ontological scale (humans > animals > inanimates), intersects with the other hierarchies: person (first and second person > third person), uniqueness (proper > common noun), and number (singular > plural > dual). We show that ««-marking generally follows the animacy and the above semantic hierarchies, but may be overridden by the non-inherent features of the second argument.

These features include the referential and discourse properties of the second argument. Referential features, captured in the Definiteness hierarchy [Croft, 2003, p. 132]: definite > specific (indefinite) > non-specific (indefinite), rely on identifiability of the referent in discourse by both interlocutors (definiteness) or by the speaker (specificity). As already mentioned, ««-marking applies only to definite individuated participants, although cross-linguistically DOM languages may differ in “extensions of object marking along the animacy/definiteness hierarchy” [Malchukov, 2008, p. 206]. Moreover, all these properties “interact in an intricate fashion” [Witzlack-Makarevich, Serzant, 2017].

The discourse status of the object referent stems from its information structure role in the sentence and a wider discourse environment. Dalrymple and Nikolaeva [Dalrymple, Nikolaeva, 2011, p. 13] propose a topicality theory of DOM “which emphasises the role of information structure in the marking patterns of objects”. Topical objects are characterized by pragmatic prominence [Aissen, 2003; Leonetti, 2004; De Swart, 2007]. Dalrymple and Nikolaeva, in the spirit of Lambrecht (1994), define topicality “as a pragmatic relation that holds between a referent and the proposition expressed by an utterance” [Dalrymple, Nikolaeva, 2011, p. 14]. Thus, it is not an inherent property of the referent, but discourse-dependent because it “depends on the speaker's construal of the situation within the given communicative context” [Dalrymple, Nikolaeva, 2011, p. 52]. In such situations, the pragmatically presupposed referent of the topical object is placed in the centre of attention and interest in the conversation. They argue that in some languages the use of DOM marking is conditioned on the topical information structure role of non-prototypical objects. The unmarked topiccomment information structure of a transitive clause assumes subject/topic and object/focus alignment: prototypical objects, which introduce a new participant in the discourse, are in focus. However, an animate and definite object, whose referent may be activated from the previous discourse or from common ground shared by the interlocutors, functions as a secondary topic. Hence, objects have two prototypical information structure functions: focus and secondary topic [Dalrymple, Nikolaeva, 2011, p. 173]. discourse grammatic transitive

The above definition of topicality echoes the one provided by Givon (2001), though the connection between topicality and reference is more clear-cut in the latter. According to Givon [Givon, 2001, p. 227], topicality subsumes two pragmatic aspects of reference: referential accessibility (anaphoric) and thematic importance (cataphoric). The former involves the speaker's judgement about how accessible a referent is to the addressee, while the latter covers “referent-coding devices” which signal “how important the referent is going to be in the subsequent discourse, so that the thematic structure of yet unfolding discourse may be organized around the important referent.” [Givon, 2001, p. 228].Given that most topics are definite and animate, topicality correlates with semantic prominence, which involves the inherent semantic features of this referent.1 Thus defined, topicality seems to overlap with de Swart's [de Swart, 2007, p. 138] understanding of pragmatic or discourse prominence in terms of centrality of a referent in the discourse. It can be concluded that ««-marked objects, being topical, possess a certain degree of pragmatic prominence2 absent in unmarked nontopical objects. The term prominence seems to be more adequate for our analysis, than topicality because the latter traditionally refers to unfocused elements.

The distribution of DOM across south-west dialects

As mentioned above, the ««-accusative pattern in Macedonian dialects is presumed to have been modeled after the dative ««-pattern where na typically marks a human definite recipient. Since humanness and g definiteness imply prominence, ««-marking by analogy may have spread onto unprototypical patients (human and definite). Further entrenchment of this pattern allowed marking on those definite animate and inanimate participants that were perceived prominent. In Macedonian dialects, a na-marked object refers to a human participant who is uniquely identifiable in the discourse by both interlocutors. More rarely this class may include definite animals and inanimate entities. The speaker chooses to assign an important discourse status to a definite animate object when it serves a particular communicative need: mainly to draw listener's attention to this participant and to signal its importance. As a result, the novel иа-marking has contributed both to the transparency and informativeness of the message, especially in multilingual environments. It should be pointed out that the south-west region has a very complex ethnic and religious make up: the Macedonian-speaking Christian population has lived side by side with Aromanian and Turkish speakers in the towns, which are surrounded by Macedonian-speaking Moslem and Christian villages. The neighboring Debar area, where Albanian-Macedonian bilingualism is common, is populated by ethnic Albanians, Macedonians, Turks and Macedonian-speaking Moslems [Vidoeski, 1998, p. 214]. The latter, known as Torbes, live in almost all villages in two districts: Zupa (roughly between Struga and Debar) and Gora (a stretch along the Albanian border).

According to the dialectal sources, today DOM exists in the following regions in the Republic of Macedonia: the Ohrid-Struga region, the Debar region, the Dojran and Gevgelija regions and marginally in the town of Strumica.1 It was present in some dialects in Greece (Lagadin / Langada area, Kostur/Kastoria, Enidze Vardar/Yanitsa, Kukus/Kilkis), Strumica is a small town in the east of Republic of Macedonia. This feature is absent in the Strumica dialect apart from the vernacular of some speakers who mark topical animate objects (testimony of Zule Stojanova). They claim that their predecessors were refugees from Kukus/ Kilkis during the Balkan Wars a hundred years ago. The two toponyms represent the Macedonian and the Greek variants, respectively. but the number of these minority dialect speakers is declining and more recent evidence is lacking.

DOM seems to be a relatively recent phenomenon in Ohrid-Struga region. Aa-objects are not attested in the vernacular language of Daniel's Tetraglosson at the very end of the 18th century [Nichev, 1997] (in view of the fact that the translation from Greek was written in the Ohrid dialect); neither is it present in the local folk poetry collected by the Miladinov brothers in the mid 19th century. Tosev (1970) notes that its use was not registered in 19th century Aromanian texts from Struga but thinks that the Aromanian refugees from Moschopolis/Moscopole (Albania) may have carried this feature into these urban areas. After several pogroms in the second half of 18th century by Ali Pasa troops, Moscopo- le (Voskopoje, Albania), mostly inhabited by Aromanians (also fluent in Greek), was finally destroyed in 1916.

Explanation of DOM

Following Koneski (1986), the use of DOM in this region is traditionally ascribed to Aromanian influence: it is claimed that the accusative pattern with a иа-marked object represents a syntactic loan from the local Aromanian dialect. This position is countered by Asenova and Aleksova (2008), who are inclined to see an independent development of this construction.1 Similarly to the Aromanian account, the use of DOM in the southernmost Macedonian dialects, which have been in close contact with Greek, is tentatively attributed to Greek influence [Topolinjska, 1995]. Following this line of reasoning [Buzarovska, 2001], points to the existence of a common syntactic pattern for accusative and dative relation in Northern Greek dialects. Both the accusative and the dative patterns employ the accusative singular form of the definite article ton (m), tin (f) and to (n) Asenova (p.c.) thinks that the syncretism of dative and accusative case marking in the Balkan dialects is related to the rise of DOM. More on this topic in Topolinska, Buzarovska (2011). For instance, with the past form verbs idha `see' and ipa `say' in the sentences Idha ton Kosta `I saw Kosta' and Ipa ton Kosta `I said to Kosta'. thereby leveling the structural distinction between them. The filled syntagmatic slot (by a casemarked article) in the accusative pattern may have been reanalyzed as a case marker in contact situations and may have influenced the insertion of na into the South Slavic accusative pattern. However, a detailed analysis of the two oldest sources containing this feature undermines the Greek contact hypothesis [Buzarovska, forthcoming], but does not rule out replication of the dialectal Greek accusative-dative pattern. Both texts, the Kulakia Gospel [Mazon, Vaillant, 1938] and the collection of fairy tales by Verkovik [Penusliski, 1985], were created in the mid 19th century in the “innovation zone” of Bal- kanisms, as Cyxun (1981) calls it. The Kulakia Gospel was written in Chalastra village (former Kulakia), north-west of Thessaloniki, while the Verkovik's collection of folk tales originates from a wider area to the north of Lagadin (mostly from the small town of Yanitsa/Enidze Vardar). This area comprised the Low Vardar dialects in the Lagadin region, north and north-east of Thessaloniki (Greece).

An internal origin hypothesis seems viable to explain the rise of this feature in other peripheral dialects. It is highly possible that similar sociolinguistic conditions in this second, south-western “innovation zone” resulted in the rise of DOM. The иа-accusative construction may have appeared in this multilingual area independently. There is no ample evidence that it was transferred from the southernmost innovation zone of the Lagadin region. In the first innovation zone (Lagadin), the past language situation, reflected in the Kulakia Gospel, betrays strong traces of bilingual use and heavy transfer from Greek. The inherited Slavic nominal case system at that time underwent radical changes: the clitics (short pronouns indicating oblique case and agreement categories) either dropped or lost gender and number distinctions. Clitics became homophonous because the same ones rendered both direct and indirect dependency between the verb and its definite object, or the same accusative relation was marked by several 3rd person clitics (the dative masculine mu, the accusative feminine a, dative feminine i, plural gi). Free variation of adverbal proclitics referring to a definite referent with the same verb was common (a vidoa na Afendot / i vidoa na Stopanot `(they) cl1 saw dom the Lord'; gu pitaa na negu / mu pitaa na negu `(they) cl asked dom him'); variation existed between dative and accusative clitics in direct dependency even in the same sentence (go fatia aramiite і mu sublikoa na negu `the robbers cl caught and cl stripped dom him'). In addition to the overuse of clitics (Isus... na Nazaret gi a najdemi `Jesus, in Nazareth (we) cl cl found), we find clitics (the feminine singular accusative a) marking an intransitive relation (on a /legal idno kaik `he cl came onto a small boat'), or a clitic marking an indefinite object (a vide idno covek `he cl saw a man').

All these changes resulted in the lack of discrimination between the agent and the patient roles. In the process, two communication-driven tendencies were violated: the tendency for explicit identification of referents via cliticscl - stands for clitic. Indicating agreement, they served as “markers of referential categories” [Topolinska 1995].and the principle of transparent marking of grammatical relations [Lindstedt, 2000, 2017]. Lindstedt points out the following: “What unites most grammatical Balkanisms structurally is a tendency towards explicit marking of grammatical relations with particles, prepositions and other uninflected function words that are identifiable across languages” [Lindstedt, 2000, p. 234]. In Lindstedt (2017), the author argues that with analytic argument marking (with prepositions and clitic doubling) “an even higher grade of intertranslatability among languages could be attained than with case endings.” It is argued here that the insertion of na before the direct object was originally motivated by the discriminating function, the necessity to make the patient argument distinct from the agent by a more transparent means. Moreover, this was done in a most economic way: presumably by replicating the existing dative pattern which led to decategorization of the dative marker na. It may be classified as construction-oriented pattern-replication [Matras, Sakel, 2007, p. 843].

Such semantically-conditioned иа-marking is most evident in the oldest source, the Kulakia Gospel, created in the first innovation zone. The important property of the Gospel text is that the marking of human objects is regular: almost all direct objects with definite human referents are preceded by na (277 marked vs 10 unmarked objects). Moreover, in several cases the marking spread onto material entities, illustrated in (4) and (5).

(4) Utidi da kupa na zemnjata.

gO.PST.3sG SUBJ buy.PRS.3sG DOM land.F:DEF

`(he) went to buy some land.'

I oni utidia і

and they go.psT.3pL and

`And they went and closed the

zaklucile na grobut.

lock.PRF.3PL DOM tomb.M:DEF

tomb.'

However, to the north of Kulakia village, DOM is defined by information structural properties of semantically prominent objects. In the second source from this period, Verkovik's folk tales [Penusliski 1985], the distribution of DOM is irregular. Only one third of all direct objects referring to definite human referents carry the marker na: 83 marked vs 33 unmarked. Similarly to the Gospel text, there are several marked non-human definite objects: animals (29instances),materialandabstractentities(30instances).Thediscourseanalysis of these texts shows that the marked objects are prominent in contrast to the unmarked nontopical objects. For instance, the marked objects in (6) and (7) serve as secondary topics in the narrative. The иа-marking signals a pragmatic relation of prominence: the second participant (the ox, her pleas) is viewed important for the development of the story. They refer to previously introduced participants placed in the center as instrumental in the unfolding of the plot.

Table 1

(6) Ajde da se	pitame	na	toj	vol!
HORTAjde is a common Balkan hortative particle. SUB REFL `Let's ask this ox!'	ask.PRs.1PL	DOM	this	ox

I bog poslusa na nejnite

molbi.

pleas

oldest sources

and Lord listen.psT.3sG dom her.F:DEF

`And the Lord listened to her prayers'.

The difference in the distribution of DOM in the two suggests that in the epicenter of the first innovation zone of Balkanisms, DOM arose due to semantic reasons: the case marker na differentiates the patient from the agent. According to Siewierska and Bakker (2008), case exponents perform a discriminatory and/or an indexing function. Although closely intertwined, the former implies definiteness and animacy, while the latter relates casemarking to the degree of affectedness of the second participant by the activity of the agent. Animacy, in turn, indirectly correlates with affectedness: animate entities, being more prominent in discourse, are perceived to be more affected in comparison to inanimates [Nsss, 2004; Malchukov, 2008]. Under the second, indexing role of case markers animacy effects are “epiphenomenal, generated through correlations between animacy features and role-related characteristics of arguments” [Malchukov, 2008, p. 211].

The regularity of na-marking in the Kulakia Gospel testifies to the prevalence of the discriminating function in the epicenter of the “innovation zone” of Balkanisms at that period (more than 150 years ago). However, DOM is applied optionally in Verkovik's tales. It seems that the discriminating function must have initially caused the use of DOM but weakened during the diffusion of this feature resulting in the subsequent foregrounding of the indexing function.

DOM in Ohrid and Struga dialects today. The study

The main hypothesis in this investigation is that the contemporary use of DOM in the south western dialects of Ohrid and Struga is pragmatically motivated, but semantically conditioned. To prove this, we conducted an acceptability judgement test involving speakers from these towns. The sample consisted of 150 young participants: 66 subjects (44%) were from Ohrid and 84 (56%) from Struga. All of them were students: the majority, 114 (76%) attended elementary schools and 36 (24%) secondary schools. Only 23 (15.33%) subjects had contacts with Aromanian. However, this parameter was not considered because a direct link with the equivalent Aromanian Ohrid pattern cannot be proved.

The instrument used in the investigation was a questionnaire with 20 sentences containing a иа-marked direct object. Though topicality is a discourse-dependent property of referents, rather than clause-level one [Givon, 2001, p. 254], the questionnaire sentences were constructed in such a manner that their propositional content fitted the addressee's state of knowledge. Familiar everyday situations with discourse-linked and contextually salient topical participants (rendered with suitable vocabulary) facilitated information exchange and allowed assessment of pragmatic prominence of the object referent. It was assumed that these sentences could be uttered in isolation in different situations as short speech acts (two were connected in a pair). Three types of sentences with different information structure were designed (all 20 sentences are given in Table 1). The first group of sentences may be labeled “hot news”: the speaker informs the addresses about an important event in which the identity of a highly individuated object referent is easily established (sentences 2, 5, 12, 13, 14, 18, 19, 20); many have a prosodic | clue (exclamation mark) to convey the emotional attitude towards the object referent. The second group includes assertions stating a relation between the speaker and a familiar participant, but whether this participant is construed prominent was left to decide on the addressee (sentences 1, 3, 6, 7, 8, 10, 11). In the third group of sentences the object referent is under contrastive stress (sentences 4, 9, 15, 16, 17). It should be noted that pragmatic prominence can also be assigned to emphasized object referents “selected from a set of alternative candidates” [Dalrymple, Nikolaeva, 2011, p. 125].

All 150 participants were asked to evaluate the acceptability of each sentence using a 5-point scale ranging from 1 (certainly unacceptable) to 5 (certainly acceptable). The order of the sentences was random. The answers were fed into a database, which served for further quantification of the results. Below we present and discuss the results of the analysis.

Although our database enables an analysis using a number of different statistical tools, we used the two measures of central tendency and variability (mean and standard deviation), as sufficient to prove the main hypothesis. Table 1 shows the results for mean and standard deviation values for each of the 20 sentences. The sentences are ordered according to the mean value of acceptability of the иа-object in a given sentence, sorted out from the highest to the lowest. As it can be seen from Table 1, the mean of the sample is 2.88, which is close to the mean of normal distribution (3.00). The range of the means (min = 1.89, max = 3.60) confirms slight tendency towards unacceptability of иа-objects. The high standard deviation of the sample (SD 1.44) indicates high variability in individual responses, which is also confirmed by the range of the SD values (min = 1.18, max = 1.6).

Acceptability scale: mean and standard deviation for each sentence

Table 2

	Sentences	Mean	SD
1	Go vidov na Martin. `(I) saw dom Martin.'	3.60	1.53
2	Znaes, Verce go saka na Vlatko! `(you) know, Verce loves dom Vlatko!'	3.50	1.45
3	Ja prasav na nastavnickata. `(I) asked dom the teacher.'	3.45	1.39
4	Na Ivan go prasavme, ama ne znaese. `dom Ivan (we) asked, but he did not know.	3.39	1.35
5	Na kuceto go udri kola! `A car hit dom the dog!'	3.29	1.47
	Sentences	Mean	SD
6	Gi prasav na Maja і Petre. `(I) asked dom Maya and Petre.'	3.15	1.50
7	Gi vidov na decata, igraa vo dvorot. `(I) saw dom the children, they were playing in the yard.'	3.13	1.55
8	Go udri na kuceto. `(he) hit dom the dog.'	3.10	1.49
9	Go udri na kuceto, ne na maceto. `(he) hit dom the dog, not the cat.'	3.09	1.52
10	Go vidov na Petre koga ja pominuvase ulicata. `(I) saw dom Petre crossing the street.'	3.06	1.55
11	Ama pak Vlatko ne ja saka na nea. `But Vlatko does not love dom her.'	2.98	1.54
12	Nastavnicke, Igor go udri na Goran so tupanica! `Teacher, Igor hit dom Goran with a fist!	2.93	1.60
13	Zosto go butna na Sime? `Why did (you) push dom Sime?'	2.85	1.45
14	Ja butna na Sandra і і gi skrsi na naocarite. `(he) pushed dom Sandra and broke her glasses.'	2.76	1.41
15	Go vidov na nego, bese sam. `(I) saw dom him, he was alone.'	2.49	1.43
16	Na niv gi prasav, ne na tebe. `dom them (I) asked, not dom you.'	2.37	1.35
17	Me butna na mene koga trcase. `(he) pushed dom me while running.'	2.37	1.29
18	Ne go fakaj na telefonot, ke mi go skrsis! `Don't touch dom the phone, you'll break it!'	2.15	1.47
19	Me butna і ja skinav na tetratkata! `(he) pushed me and I dom tore my notebook!'	2.00	1.18
20	Vidi, Maja mi gi skrsi na naocarite! `Look, Maya broke dom my glasses!'	1.81	1.24
Total	2.88	1.44

Factors influencing the use of DOM

The analysis confirmed that the distribution of ««-marking is governed by the interplay of inherent and non-inherent properties. Aissen (2002) refers to DOM triggered by definiteness and animacy as two dimensional. Given the role of pragmatic factors DOM in Ohrid-Struga dialects may be considered multidimensional. The semantic factors limit the range of object referents to definite individuals, while pragmatics is responsible for their topicality. It is evident that the pragmatic criterion of discourse prominence prevails over the inherent property of humanness allowing marking on animals and things. However, affectedness does not seem to play a significant role in assigning an important discourse status to the second argument in these dialects: it is often used with state verbs see and love.

Semantic factors: animacy and number

The importance of definiteness and animacy criteria as determining factors was confirmed in the analysis. The respondents followed the animacy hierarchy in the use of DOM: person, animal, thing. The sentences below are ordered according to the obtained acceptability values. Most acceptable is the sentence in (8), whose object Martin represents an activated participant known to both interlocutors. The sentence in (9) with the animal object (the dog) is ranked eighth, while the least accepted is the sentence in (10), with a material object (the glasses).

Go vidov na Martin. (3.60)

`(I) saw dom Martin.'

Go udri na kuceto. (3.10)

`(he) hit dom the dog.'

Vidi, Maja mi gi skrsi na naocarite! (1.81)

`Look, Maya broke dom my glasses!'

The adherence to the animacy scale is evident in the distribution of the five most acceptable sentences: the objects in the first four sentences have individuated referents: Martin (11), Vlatko (12), Ivan (13), the teacher (14), and in the fifth one it is a familiar dog (13).

Znaes, Verce go saka na Vlatko! (3.50)

`(you) know, Verce loves dom Vlatko.'

Japrasav na nastavnickata. (3.45)

`(I) asked dom the teacher.'

Na Ivan go prasavme, ama ne znaese. (3.39)

`dom Ivan (we) asked, but he did not know.'

Na kuceto go udri kola! (3.29)

`A car hit dom the dog!'

The three sentences with the lowest degree of acceptability (ranked 18th, 19th and 20th) have definite inanimate objects: the telephone (15), the notebook (16), the glasses (17).

Ne go fakaj na telefonot, ke mi gi skrsis! (2.15)

`Don't touch dom the phone, you'll break it!'

Me butna і ja skinav na tetratkata! (2.00)

`(he) pushed me and (I) dom tore my notebook!'

Vidi, Maja mi gi skrsi na naocarite! (1.81)

`Look, Maya broke dom my glasses!'

However, the fact that inanimate objects were accepted at all undermines the primacy of the animacy criterion and suggests that other factors are at play. The object the dog in sentence (14), is a salient participant because an animal, familiar to both interlocutors, was affected in a traffic accident, so the sentence was ranked high (as fifth) on the scale.

Not all objects referring to humans exhibit prominence. Generally, pronominal objects anaphorically referring to discourse-old animate topics were not perceived salient enough by respondents. This explains why the two sentences containing anaphoric pronouns и/v `them' and mene `me' occupy the sixteenth and seventeenth position in Table 1 (see examples 18 and 21). The low prominence of pronominal objects becomes evident when comparing the ranking of sentences (18) and (13). Though they share the same predicative relation (prasa `ask') and information structure, (13) was ranked much higher - in the fourth position, because the topicalized object is not formalized as a pronoun (18), but as a proper noun (13). Interestingly, contrastive contexts seem to increase object's prominence even in cases of an animal referent; thus (19) was ranked ninth as opposed to the sixteenth position of (18).

Na niv gi prasav, ne na tebe. (2.37)

`dom them I asked, not dom you.'

Go udri na kuceto, ne na maceto. (3.39)

`(he) hit dom the dog, not the cat.'

A similar explanation can be offered regarding the acceptability of sentences in (20) and (21). In (20), ranked fourteenth, the agent pushed Sandra,

a topical referent whose salience is confirmed in the subsequent clause (which informs what happened to her), whereas a similar sentence (21) with an anaphoric object drops to the low seventeenth position on the acceptability scale.

Ja butna na Sandra і і gi skrsi na naocarite. (2.76)

`(he) pushed dom Sandra and broke her glasses.'

Me butna na mene koga trcase. (2.17)

`(he) pushed dom me while running.'

The prominent status of the second participant depends on how relevant s/he is perceived for the ensuing flow of events. Individuated singularized referents were considered more relevant by the respondents.

Thus the sentence in (22) with the object nastavnickata `the teacher' is ranked third. Compare a similar sentence in (23) with plural object referents (Maya and Petre), ranked sixth.

This suggests that DOM marking is sensitive to number, another relevant scalar dimension for DOM, also interacting with the animacy hierarchy [Witzlack-Makarevich, Serzant, forthcoming].1

Ja prasav na nastavnickata.

(3.45)

(3.15)

`(I) asked dom the teacher.'

Gi prasav na Maja і Petre.

`(I) asked dom Maya and Petre.'

Pragmatic factors: discourse prominence

This criterion applies when the object referent is construed pragmatically salient in the sentence and important for the subsequent discourse. The иа-marking directs attention to the second participant, who is known to both interlocutors irrespective of its referential accessability. In the comparative analysis of differently ranked sentences we tested several parameters that may influence discourse prominence: information structure role, pronominal realization, individuation and empathy. Below each of them is addressed in this order.

To explain the role of the information structure of the sentence in which the object occurs, we apply Lambrecht's classification of focus structures I [Lambrecht, 1994, p. 138]. In unmarked information structuring, secondary topics are either in sentence focus or predicate focus. Sentence-focus structures occur in sentences that report a new, often unexpected event, while predicate-focus structures are found in topic-comment sentences. Although predicate phrases carry new information, within the predicate domain the object is topical. In marked configurations, discourse prominent objects are contrastive topics indicated by higher pitch accent and/or topicalization.

The comparison between the information structure type of the sentences on Table 1 and their acceptability ranking does not show any correlation. For instance, sentence-focus structures (sentences 2, 5, 12, 13, 14, 18, 19, 20) rank both high and low. The same is observed in the second informationstructure type, whereas contrastive topics (sentences 4, 9, 15, 16, 17) show a certain correlation with prominence.

What “creates” prominence is speaker's emotional attitude and other factors. This is evident in sentences in (24) and (25), which form an adjacency pair and have a similar propositional content. In (24), ranked very high (second), the object Vlatko is a secondary topic in sentence-focus structure. The surprising news about the well known participants is preannounced by `you know' and conveyed via emphatic intonation. On the other hand, the иа-marked pronominal object nea `her' in (25), which anaphorically refers to the already introduced participant, is ranked eleventh. Sentence (25) is an example of a predicate-focus structure in which the topical anaphoric object is within the focused verb phrase. It was less accepted than the sentencefocus structure in (24). The low acceptability of (25) may be attributed to the pronominal realization of the object referent and lack of speaker involvement.

Znaes, Verce go saka na Vlatko! (3.50)

`(you) know, Verce loves dom Vlatko!'

Ama pak Vlatko ne ja saka na nea. (2.98)

`But Vlatko does not love dom her.'

Defocused objects have a lower prominence status. Defocusing here signifies the speaker's diffused rather than directed or focused attention towards the participant. In the most accepted sentence (26), the object Martin is in the focus of the speaker-perceiver, while Petre in (27) is less accepted (ranked tenth) because the speaker's focus is divided between Petre and the “background” activity he performs (crossing the street). Ranked seventeenth, the sentence in (28) with an anaphoric object (mene `me') was even less accepted because the participant (ego) is defocused by an adverbial temporal clause (while running).

Go vidov na Martin. (3.60)

`(I) saw dom Martin.'

Go vidov na Petre koga japominuvase ulicata. (3.06)

`(I) saw dom Petre crossing the street.'

Me butna па mene koga trcase. (2.37)

`(he) pushed dom me while running.'

Animal referents are also subject to discourse configuration effects. Malchukov notes that the speaker's empathy is related to animacy and discourse prominence [Malchukov, 2008, p. 204]. The different acceptability of sentences in (29) and (30) with the same object referent (the dog) may indicate that a topicalized participant is more salient than an ontologically identical object referent in a canonical linearization. Consequently, the marked accented dog was accepted more readily (ranked fifth) than its non- empathic counterpart (ranked eighth), although the difference in acceptability between these sentences is rather small (only 0.19).

Na kuceto go udri kola! (3.29)

`A car hit dom the dog!'

Go udri na kuceto. (3.10)

`(he) hit dom the dog.'

As mentioned earlier, inanimate participants in direct object positions received the lowest acceptability rates. Yet, even inanimates display different prominence, which probably depends on the degree of perceived affectedness of the inanimate object and the illocutionary force of the utterance. Thus, the sentence in (31) has somewhat higher acceptance rate than the one in (32) probably because the notebook is already affected, (i.e., destroyed) in contrast to the telephone. On the other hand, the number of subjects who gave the highest value to the sentence with the telephone is considerably higher (22 subjects)1 than the sentence with the notebook (7 subjects) because a telephone is a much more valued possession than a notebook.

Me butna і ja skinav na tetratkata! (2.00)

`(he) pushed me and (I) dom tore my notebook!'

Ne go fakaj na telefonot, ke mi go skrsis! (1.81)

`Don't touch dom the phone, you'll break it!'

The presented findings of the analysis are supported by the evidence I presented in [Tosev, 1970, p. 99] about the use of DOM in Struga almost

о 50 years ago. He lists 33 examples with DOM collected from the assignments

з written by Struga schoolchildren. In all the examples the direct objects

code individuated human referents that presumably were seen as important discourse participants. In (33) the direct object is topicalized and the generic agent is removed from the scene, while in (34), the object referent, introduced in the preceding sentence, undergoes an unexpected turn of events.

Na Dura ne go pustile da odit.

dom Gjura not cl.3msg.acc let.PRF.3pL sub go.PRs.3sG

`Gjura was not allowed to leave.'

Aleksa otisol da krade,

Aleksa go.PRF.3sG sub steal.PRs.3sG

no togas na Aleksa go fatija.

but then dom Aleksa cl.3msg.acc catch.PST.3PL

`Aleksa went to steal something, but then Aleksa was caught.'



Conclusion

The paper aims to account for the contemporary use of DOM in the Ohrid and Struga dialects on the basis of an empirical study. The results of the study clearly indicate that DOM in this historically attested “innovation zone” is triggered by information structure needs to mark a prominent discourse participant in the object position. This entails that the distribution of DOM in south-western dialect is regulated by pragmatic principles applied to semantically restricted set of referents.

Not excluding the role of contact, the paper suggests an internal origin of this preposition-based pattern. Originally meant to semantically disambiguate the obscured case relations involving a second definite human participant in multilingual communities, the casemarker na evolved into a discourse prominence marker, a “referent-coding device” that serves the communicative needs of the dialect speakers. This supports the position expressed by Topolinska (1995) that the pragmatic and semantic motivation for language changes in Balkan languages was based on the communicative need to make messages maximally transparent to speakers of other languages. By the same token, Friedman (2000) argues that the need for successful communication led to the creation of Balkanisms. One of the first scholars to acknowledge the role of pragmatics in shaping Balkan grammars is [Civian, 1979, p. 284]. She points out that pragmatics, not grammar, ensures successful communication “в условиях разноязычия” (i.e., multilingual g environment) among the speakers of the Balkan Sprachbund. Her view that: | “и в своем формально-грамматическом единстве и в своем разноообра- о зии система БЯС Балканский языковой союз. ориентирована на осуществление коммуникации $ в условиях максимума помех в канале связи (разноязычие, потенциальная необходимость перехода с языка на язык)...” [Civian, 1979, р. 279] predicts that such a non-Slavic phenomenon as a preposition-based DOM in peripheral Slavic dialects can be viewed as a communication strategy.



References

1. Adamou, 2006 - Adamou E. Le nashta. Description d'un parler slave de Grece en voie de disparition. Munchen, 2006.

2. Aissen, 2003 - Aissen J. Differential object marking: Iconicity vs economy. Natural language and linguistic theory. 2003. Vol. 21. Pp. 435-483.

3. Asenova, Aleksova, 2008 - Asenova P., Aleksova V. L'aspect balkanique de la nota accusativi personalis. Zeitschrift fur Balkanologie. 2008. Vol. 44. № 1. Pp. 1-22.

4. Bossong, 1984 - Bossong G. Animacy and Markedness in Universal Grammar. Glossologia. 1984. Vol. 39. № 2/3. Pp. 7-20.

5. Buzarovska, forthcoming - Buzarovska E. The contact hypothesis revised: DOM in Macedonian dialects. Journal of Language Contact. Forthcoming.

6. Buzarovska, 2001 - Buzarovska E. Accusative and Dative Clitics in Southern Macedonian and Northern Greek Dialects. Balkanistica. 2001. Vol. 14. Pp. 1-17.

7. Chiriacescu, von Heusinger, 2010 - Chiriacescu S., Heusinger K., von. Discourse Prominence and Ре-marking in Romanian. International Review of Pragmatics. 2010. Vol. 2. Pp. 298-332.

8. Civian, 1979 - Цывьян Т.В. Синтаксическая структура балканского языкового союза. М., 1979. [Civian T.V. Sintaksicheskaya struktura balkanskogo yazykovogo soyuza [The Syntactic Structure of the Balkan Sprachbund]. Moscow, 1979.]

9. Comrie, 1975 - Comrie B. Definite and animate direct objects: A natural class. Linguistica Silesiana.1975. Vol. 3. Pp. 13-21.

10. Comrie, 1989 - Comrie B. Language Universals and Linguistic Typology. Chicago, 1989.

11. Croft, 2003 - Croft W. Typology and universals. 2nd ed. Cambridge, 2003.

12. Cyxun, 1981 - Цыхун Г.А. Типологические проблемы балканославянского языкового ареала. Мн., 1981. [Tipologicheskie problemy balkarnslavyanskogo yazykovogo areala [Typological Problems of the Balkan Sprachbund].Minsk, 1981.]

13. Dalrymple, Nikolaeva, 2011- Dalrymple M., Nikolaeva I. Objects and information structure. Cambridge, 2011.

14. Dixon, 1979 - Dixon R.M.W. Ergativity. Language. 1979. Vol. 55. Pp. 59-138.

15. Friedman, 2000 - Friedman V.A. Pragmatics and Contact in Macedonia: Convergence and Differentiation in the Balkan Sprachbund. Juznoslovenski filolog. Vol. 56. № 3/4. Pp. 1343-1351.

16. Givon, 2001- Givon T. Syntax: An Introduction. Vol. 2. Rev. ed. Amsterdam, 2001.

17. von Heusinger, Kaiser, 2003 - von Heusinger K., Kaiser G.A. The Interaction of Animacy, Definiteness and Specificity in Spanish. Proceedings of the Workshop “Semantic and Syntactic Aspects of Specificity in Romance Languages”. von Heusinger K., Kaiser G.A. (eds.). Konstanz, 2003. Pp. 41-65.

18. de Hoop, Malchukov, 2008 - de Hoop H., Malchukov A.L. Case-Marking Strategies. Linguistic Inquiry. 2008. Vol. 39. № 4. Pp. 565-587.

19. Karanfilovski, 1992 - Каранфиловски М. Еницевардарскиот говор. Скоще, 1992. [Karanfilovski М. Enidzevardarskiot govor [The Dialect of Enidze-Vardar]. Skopje, 1992.]

20. Koneski, 1986 - Конески Б. Македонскиот XIX век, jазични и книжевно- историски паралели. Скоще, 1986. [Koneski B. Makedonskiot XIX vek, jazicni i istoriski paraleli [Macedonian 19th Century, Literary and Historic comparisons]. Skopje, 1986.]

21. Labroska, 2003 - Лаброска В. Говорот на селото Кула - Серско. Скоще, 2003. [Labroska V. Govorot na seloto Kula - Sersko [The dialect of Kula village in the region of Ser]. Skopje, 2003.]

22. Lambrecht, 1994 - Lambrecht K. Information Structure and Sentence Form: Topic, Focus, and the Mental Representations of Discourse Referents. Cambridge, 1994.

23. Leonetti, 2004 - Leonetti M. Specificity and object marking: the case of Spanish. Catalan Journal of Linguistics. 2004. № 3. Pp. 75-114.

24. Lindstedt, forthcoming - Lindstedt J. Diachronic regularities explaining the tendency towards explicit analytic marking in Balkan syntax. Balkan Syntax and Universal Principles of Grammar. B. Joseph, I. Krapova (eds.). Berlin/Boston. Forthcoming.

25. Lindstedt, 2000 - Lindstedt J. Linguistic Balkanization: Contact-induced change by mutual reinforcement. Studies in Slavic and General Linguistics. Vol. 28 (Languages in Contact). Gilbers D., Nerbonne J., Schaeken J. (eds.). Amsterdam/ Atlanta, 2000. Pp. 231-246.