Grammatical Convergence in Gostivar Turkish
© Şirin Tufan
This study focuses on a number of grammatical structures of Gostivar Turkish (GT) which are seen to have been syntactically re-arranged due to contact with the surrounding languages. The argumentation in this study is that in GT, being a Turkic language, the organisation of knowledge into propositions is triggered by contact with the surrounding Indo-European languages, c.q. Macedonian and Albanian. Language contact among these typologically different languages leads to a spread of linguistic features in order to harmonise differences in terms of typology, more in particular: constituent order, finiteness/non-finiteness etc. The order of the constituents in GT is viewed as an outcome of convergence. Following Matras (1998, 2003/4), it is proposed that convergence in the Balkan context is a cross-linguistic fusion of techniques for arranging and organising propositions, leading to language-internal grammaticalization processes. Hence, this study gives a description of techniques and employed strategies in mapping knowledge in GT.
The following strategies are discussed:
- ordering of constituents (nominal predicate and existential sentences);
- ordering of constituents + status/function of constituent (copular sentences; possessive phrasal constructions; complex sentences: modal clauses, relative clauses, adverbial clauses; interrogative constructions).
ST = Standard Turkish
MAC = Macedonian
ALB = Albanian
Ordering of constituents
Ordering of constituents in nominal predicate will be discussed. Consider the following examples:
(1) | a. | GT | O | ol-di | damat | |
he | be-PST | groom | ||||
b. | MAC | Toj | stana | mladoženec | ||
he | became | groom | ||||
c. | ALB | Ai | u bë | dhandër | ||
he | became | groom | ||||
d. | ST | O | damat | ol-du | ||
he | groom | be-PST | ||||
'He became a groom' |
(1a), (1b), (1c) and (1d) are instances of nominal predicate expressing change/ promotion of state or role. In (1a) the verb is in sentence non-final position as same as in (1b) and (1c), as to (1d) it is in sentence final position. Hence, in (1a) GT displays different word order than that of ST in (1d). Note that, Turkish, being classified as SOV language, following Greenberg (1963) allows us to predict certain properties of the language (e.g., adverb-verb, object-verb, adj-noun orders), but also following Erguvanli (1984) there would be verb-object orders which are pragmatically conditioned. Hence, a sentence like "O ol-du damat", having the verb in non-final position would be pragmatically marked structure in Turkish. However, as suggested by Heine & Kuteva (2005:50), under the influence of Macedonian and perhaps other Balkanic verb-medial (SVO) languages, speakers of West Rumelian Turkish have developed one of their pragmatically minor use patterns into an unmarked pattern in an attempt to establish syntactic equivalence with the languages of their Balkanic neighbours, in this case Macedonian and Albanian.
Ordering of constituents + status / function of constituents
In the aforementioned examples, we looked only at the ordering of constituents as a strategy in utterance organization in GT. However, not only the organization of the order of the constituents, but also the status/ function of constituents plays a role in the organization of the utterance structure in GT. Consider the following examples:
Nominal predicate:
Adjectival predicate:
(3) | a. | GT | O | dur | güzel | |
she | 3SG | beautiful | ||||
b. | MAC | Taa | e | ubav-a | ||
she | 3SG | beautiful-F | ||||
c. | ALB | Ajo | asht e | bukur | ||
she | 3SG | beautiful | ||||
d. | ST | (O) | güzel-dir | |||
she | beautiful-3SG | |||||
'She is beautiful' |
Locative predicate:
The above presented examples (2 a,b,c,d), (3 a,b,c,d) and (4a,b,c,d) are all copular sentences. (2a), (2b), (2c) and (2d) are copular sentences with nominal complements. If we, firstly, compare (2a) and (2d) we see that in (2a) the copula followed by the attached subject agreement suffix precedes the nominal predicate and preserves its unbound character, while in (2d) the subject agreement suffix functioning as a copula is a clitic attached to the nominal complement. There is one to one mapping between copula and subject clitic in ST (e.g. kız-sın). Hence, in terms of the status the copula in GT resists the grammaticalization that ST has undergone. Also, the examples of adjectival and locative predicate (3a) and (4a) reveal the same phenomena. Ordering of the constituents remains as a strategy that the speakers of GT apply in order to syncretize the utterance structure (Matras 2003/04) in GT. Furthermore, by looking at (2a) and comparing it to the relevant examples we see that there is merger at the level of clause (where the position of a constituent functions as a pivot constituent), but not at the level of phrase. Consider the relation between sen i-sın in (2a), ti si in(2b) and ti ye in (2c) with that of küçük bir kış-çe in (2a), edn-o mal-o devoj-če in (2b), nji vajzë e vogël in (2c) and küçük bir kız-sın in (2d). This shows that convergence in GT is selective.
Marking of possession
In this section I would deal with ordering of constituents in possessive constructions and marking of possession in GT. Here I will treat examples of possessive phrasal constructions. Thus, an order in possessive noun phrase in GT is that of possessee - possessor, that is GT behaves head initial. Hence, the dependent genitive follows the governing noun, as illustrated in (5a) below, an order opposite to that of ST in (5d), but similar to the ones in Macedonian (5b) and Albanian (5c). Thus, GT harmonizes its order with that of the surrounding languages. However, as to the nature of expressing possession in GT we see that the possessor has a genitive case attached, while the possessee has the possessive agreement suffix attached, agreeing with the possessor in person and number, same as in ST. In Macedonian and Albanian possession is marked by a preposition. Thus, the languages in contact have different means of expressing syntactic functions. Therefore, in GT the following convergence strategy seems to be employed: convergence in terms of word order, redundancy of analytically expressed na, since possession is already syntactically inherited.
Ordering of Clauses
In this section we look at strategies employed in complex constructions. First, modal sentences will be elaborated, followed by analysis of relative clauses.
Modal Clauses
In here we look at the usage of subjunctive. Consider the following examples:
In ST the dependent clause is non-finite and the main verb is in sentence final position following the dependent clause. Look at (6d) above. However, as illustrated in (6a), in GT the main verb precedes the dependent clause which is finite. Note that in Macedonian (6b) and Albanian (6c) there is the same ordering and status of constituents as in (6a). Thus, in order to establish syncretism with the surrounding languages GT undergoes language internal change triggered by contact. Firstly, there is change in the order of the pivot constituent and also replacement of non-finite structure to finite one. How is it achieved? The function of the optative (oyna-ya-yım) in Turkish is extended to the function to that as subjunctive. Thus, GT undergoes grammaticalization. Mood is expressed syntactically with its own devices. Note that in Macedonian and Albanian mood is expressed analytically. In Macedonian there is a modal particle da which is traditionally the maker of the subjunctive (Friedman 2002:33) and in Albanian the subjunctive is marked with the particle të.
So, in modal clauses of GT the dependent clause follows the main verb, the non-finite structure infinitive is replaced by finite structure optative which leads to loss of infinitive, mood is syntactically expressed.
Relative Clauses
Relative clauses in Turkish in relation to the position of the clause with respect to the head are predominantly prenominal. The relativized NP is expressed by a participle, there is no particle as a relativizer, consider example (7d) below. On the other hand, relative clauses in Macedonian and Albanian are postnominal, there is a finite verb form as a relativizer. Look at (7b) and (7c). GT, as illustrated in (7a) shows postnominal behaviour of the clause with respect to the head, and a relativizer in finite form.
Ordering of constituents as a strategy shows that there is an organization of the complex clause in two levels. Firstly, harmonizing towards the order of the constituents in the main clause. Emphasis on the verb of the main clause in order to establish SVO order. Secondly, harmonizing in terms of the order and status of the restricting clause. Thus, the restricting clause follows the head and the relativizer sets off the restricting clause as a relative clause and the status of the clause is that of finite. In order to harmonize the ordering of the constituents the gap of a relativizer is fullfilled through the grammaticalization of the interrogative ne into a relativizer. Heine & Kuteva (2005) refer to it as replica grammaticalization.
Yes/No Questions
Finally, we look at interrogative constructions. Consider the following yes/no questions:
In yes/no questions GT shows traces of convergence, not only in terms of pattern but also in terms of matter (adaptation of actual material).The word order of yes/ no questions in GT is that of: QP V IO, the same order as that of Macedonian and Albanian, while in Turkish would be that of: IO V QP. GT in order to harmonize its utterance structure, following the word order of Macedonian and Albanian, pursues a transfer of matter (the question particle 'a' is borrowed from Albanian). Transfer of matter is functionally motivated in GT. In Turkish, the question particle mi is inherently enclitic and enclitics generally appear in second place in their clause. In the sense of Haines & Kuteva's (2005) hierarchy of grammaticalization, the unidirectinality principle is justified in this example, where a grammatical bound morpheme cannot change to a free morpheme. Therefore, transfer of matter is attained.
The above presented phenomena are examples of convergence on syntactic level. However, language contact phenomena in GT is also evident at both, phonological and morphological levels.