Birsel Karakoç

Turkish has a close connection between verb nominalisation and clausal embedding;
complementizing nominalisers constitute a more or less closed class. Herkenrath & Karakoç
(2016) present the suffixal combinations -mIşlIK and -mAzlIK as questionable candidates for
complementizership in modern Turkish. These forms, which are rarely used (some two
instances each per hundred pages in our data), have not yet received much attention in the
literature (however Banguoğlu 1995 [1959], Lewis 1967, van Schaaik 2000: 116, Haig 2003),
and no attention at all in discussions of clausal complementation.
Inside these forms, the respective first components, -mIş and -mAz, express verbal
inflectional categories (aspect and, in -mAz, also negation), whereas the second component,
-lIK, is a highly productive nominal derivational (Borsley & Kornfilt 2000 for the idea of verbalnominally
‘mixed extended projections’, Haig 2003 on theoretical challenges presented by an
‘inflection-derivational borderline’ area). We interpret some of the -mIşlIK and -mAzlIK
constructions as clausal. Taking a different direction than Lehmann’s (1988)
‘desententialisation’, Matthiesen & Thompson’s (1988) discourse-rhetorical motivation of
hypotaxis, or Thompson’s (forthcoming) emergence of clauses as units from social actions,
our constructions seem to ‘clausalise’ out of nominals as some speakers/writers begin to
exploit their clausal potential.
The morphosyntactic criteria that operationalise our ‘nouniness/clausiness scale’ include, on
the clausal side: the realisation of verbal categories such as negation and diathesis, the
degree to which case-governed arguments occur (especially direct objects), the use of
adverbials; on the nominal side: the occurrence of nominal attributes, determiners, and
quantifiers. At the superordinate level, the occurrence of verba sentiendi and dicendi plays a
role in favour of clausiness.
Our data confront us with borderline phenomena at two levels: (1) between categories and
(2) within one given category. In the first case, a form may be introduced by a determiner, but
still assign accusative case. In the second case, which is in the focus of the present paper,
findings are dubious at the category-internal level, specifically with respect to the status of
construction-internal genitive NPs: genitive NPs can be either genitive attributes in nominal
possessive constructions or genitive subjects in clausal constructions. We find, in our data,
clear cases on both sides as well as some unclear cases.
Attempts to solve this problem on syntactic grounds alone run into problems of circularity.
The issue has to be tackled from a discourse-based position (e.g. Schroeder 1999). We
preliminarily identify the distinction between genitive attributes and genitive subjects as a
discourse-semantic one, more precisely as one of agency and illocutionary force (Nichols
1998). In order to clarify the morphosyntactic status of these NPs, our paper thus
investigates their discourse-semantic qualities.
We base our investigation on a corpus of contemporary literary Turkish texts, currently some
4,800 pages, forming a concordance of some 202 items so far, allowing us to move between
a systematising and an explorative approach. Our paper will suggest some categorial
distinctions that can help to draw a line.

This paper attempts to portray the status nascendi of a new evidential system in Turkish as spoken in Germany. Grammatical encoding of evidentiality is obligatory in Turkish, in indicating the indirectness of a source of information (inference, hearsay, perception) and in realising second-hand narratives (Johanson 2003, Aikhenvald 2004 for a typological contextualisation). The data (Rehbein, Herkenrath & Karakoç 2009) include child-adult narrative interactions in which children retell episodes of family history and other stories.

Previous studies found that bilingual children use Turkish evidential forms (-mIş and -(y)mIş) less than monolinguals and do not realise any -mIş-based narratives (e.g. Rehbein & Karakoç 2004, Karakoç 2007). Karakoç & Herkenrath (2016) take a discourse-interactional perspective, asking questions as to the use of evidential versus alternative forms and the achievement of pragmatic effects of evidentiality.

Our data show how bilingual children, in communicating with monolingually socialised adults, treat evidentiality as an optional category. From the perspective of a monolingual listener, this leads to ungrammaticality and/or poses challenges to understanding, especially with respect to the (lack of) distance between the child and the event. We hypothesise that optional encoding of evidentiality is a feature of bilingual Turkish, i.e. a new way of dealing with information source and text type conventions. In other words: the grammaticalised status of evidentiality seems to become shaky in bilingual contexts. Communication between the new generation of bilingual speakers and their monolingually socialised elders thus raises needs for clarification and/or adaptations at the receptive level. The present study looks at a range of phenomena in old-versus-new-speaker communication, with the following goals in mind:
− to describe the functional-semantic difference between the old and the new system,
− to reconstruct what happens in terms of interaction, listening, and understanding,
− to get a clearer picture of the status of conventionalisation.

The aim of our paper is to discuss the syntactic and semantic functions of some complementisers in Turkish with some digressions into other Turkic languages. We define ‘complementation’ as the embedding of a clause, i.e. a functional and syntactic unit realising a proposition and a predication and projecting an inner argument structure, into the argument structure of a matrix clause. The analysis is mainly based on spoken Turkish data from the ENDFAS/SKOBI corpus (Rehbein 2009, Rehbein et al. 2009); for comparison, spoken and written Noghay data (Karakoç 2005) as well as some data from other Turkic languages will be referred to.

Typologically, the most central feature of complementation, and more generally of subordination, in Turkic languages are non-finite, head-final (‘left-branching’) structures involving bound morphemes (Johanson 1975a, b), e.g. simple participles and verbal nouns -DIK (1), -(y)AcAk (2), -mA (3) and -(y)Iş (4), as well as some complex participles (5) in Turkish.

We argue that an analysis of complementation in Turkish has to centrally include these non-finite, i.e. suffixal forms. Intriguingly, these elements are not opaque complementisers, but consist of functionally and etymologically analysable forms that may also occur in other, not necessarily subordinating, functions.

There also exist strategies for embedding finite clauses. Among these, we exclude the particle ki, from Persian, which has sometimes been analysed as a complementiser (e.g. Kerslake 2007), on the grounds given in Johanson (1975a) and also Rehbein (2005, 2006). Another strategy, based on syntactic adjacency alone (6), can be discussed in relation to omission phenomena in other languages (Herkenrath & Karakoç 2007). Furthermore, subordinating particles based on converbial forms of the verbum dicendi de- ‘say’ are employed, for example diye in Turkish (7) or dep as in Noghay.

The paper aims at a typological and functional-pragmatic categorisation of Turkish complementisers in terms of the communicative potential of their smalles morpho-syntactic units (theorised as ‘linguistic procedures’, see Ehlich 1986/1996, Redder & Rehbein 1999, Rehbein 2002 and others).
(1) The complementisers under (1) to (4) have a morphosyntactic status somewhere in between nominal and verbal (Borsley & Kornfilt 1998). On their nominal outside, they occur in combinations of suffixes involving deictic and phoric possessives as well as case; on their verbal inside, they project a full argument structure. While some of them, such as -DIK and -(y)AcAK display more verbal features such as (at least partial) aspectual information, and can be shown to also occur in finite forms, others, such as the infinitival -mA and particularly -(y)Iş have more nominal features and their suffixes may even be more productive as derivational nominal suffixes.
(2) We would like to argue that the aspectual properties of -DIK and -(y)AcAK are not in contradiction to their complementising functions. The question rather concerns the functional contribution of the apsectual information to the complementising effect.
(3) The complementising suffixes -mA, -DIK and -(y)AcAK, in combination with possessive and case suffixes as well as postpositions and other lexical material, also occur as subordinators in adjunct clauses; moreover, -DIK, in combination with possessives, is employed to form attributive clauses and nominal actor clauses (next to -(y)An, see Johanson 1990: 199 for the term ‘nominal actor clauses’).
(4) How do the individual formal-communicative components of these suffixes and their combinations contribute to the complementising effect?
(5) How do (1) to (4) semantically differ from each other, e.g. in terms of modality (‘factive’ versus ‘actional’, Borsley & Kornfilt 1998, cf. Csató 1999), aspect, and epistemic categories (Van Lier & Boye 2009, Csató 2010 for a classification of matrix verbs in these respects)?
(6) -DIK (1) is an aspecto-temporally neutral complementiser and it takes complex forms such as in (5) to realise the full aspectual range. In contrast, the finite forms in (6) and (7) allow for the full aspectual marking. How do the non-finite strategies (1) to (4) on the one hand and the finite strategies in (6) and (7) on the other differ in terms of the aspectual and epistemic range of meanings that they can express?
(7) In a generative framework, several studies have addressed the question of whether or not it is justified to analyse the Turkish subordinating morpholoical elements as being part of the COMP system (see Borsley & Kornfilt 1998, Kural 1992, 1998 etc.).

This paper investigates employments of converbial clauses in monolingual and bilingual children’s Turkish. As a quantitative pre-study has shown, converbs are employed by very young children; on the other hand, in school-age children growing up in a monolingual school system in Germany, the development of more complex converb constructions tends to slow down. Moreover, other grammatical areas in our corpus (e.g. deictics, aspects) have shown functional reinterpretations involving changes of grammatical categories in the sense of ‘transpositions’ of linguistic procedures, as defined within the framework of functional-pragmatics.

The research question pertains to qualitative differences between the two groups: is it possible to establish a distinctly bilingual employment of converb clauses?