Abstract (english) | In the literature about the Croatian language, grammars devote very little space to the apposition. Their study revealed a number of inconsistencies in the approach and methodology of its description. In addition, work on syntactically marked corpora has shown that the formalisms developed for individual treebanks do not correspond to the interpretation of linguistic phenomena in grammars for the Croatian language. The examples given in Croatian grammars and the offered solutions do not satisfy the need for annotating real sentences from real texts in the corpora. Our initial research showed that one of such problematic places in Croatian grammar is the linguistic category of apposition, which is described inconsistently and inadequately, and in syntactically annotated corpora is annotated with rules developed for other languages. If the descriptions in Croatian grammars are compared with the descriptions in other Slavic languages, especially Czech, the inconsistency in the description of this syntactic category is even stronger. This can be seen in the description of the apposition primarily in the fact that most authors approached the topic simultaneously from different methodological points of view, i.e. formal-descriptive, using a transformational-generative model while at the same time including the semantic criterion. By combining the models of description, the interpretation of the apposition in the Croatian language sometimes seems wrong, and certainly incomplete. Czech syntaxes, on the other hand, have a rich tradition of normative handbooks and their study shows that grammatical methodologies have developed gradually and that new ones have consistently relied on the theoretical solutions of the previous ones. Such, a more systematic, approach to the explanation of grammatical problems is clearly seen in the example of the description of the apposition in Czech grammars. A review of previous research for the Czech language justifies the main motivation and main goal of this research, which rely in part on choosing the Czech model for describing the apposition in the Croatian language. The main goal of this thesis was to redefine the term of apposition in the Croatian language based on the evaluation of previous descriptions and data obtained from syntactically annotated corpora. We started from the assumption that the previous descriptions of the apposition in Croatian are very different from the descriptions in Czech due to unclear theoretical and methodological approaches, and not because it is a different phenomenon in the two languages. The intention was to take the poorly defined category of apposition as a sentence member in Croatian and better describe it as a sentence relationship modelled on descriptions for the Czech language. The main motive for the creation of this thesis is the fact that in the Croatian language there is no corpus-based research of the apposition that would clearly define the grammatical criteria for its determination. Apposition is described in grammars as part of the sentence structure without explaining how it relates to other parts of the sentence structure. The evaluation of previous descriptions of the opposition for Croatian has shown that the mentioned criteria cannot be used to describe all linguistic phenomena, that they are not based on the whole spectrum of real examples realized in real texts, which leads to problems in syntactic corpus annotation. The evaluation of Czech grammar manuals has shown that the Czech model can be used to describe linguistic phenomena and linguistic relations that are not currently included in descriptions for Croatian, but often occur in real text. The thesis is divided into three main sections. 1. An overview of theoretical approaches to apposition: In this section, all available descriptions of the apposition for the Croatian and Czech languages are presented in detail. Available descriptions include descriptions in grammars, other language handbooks, and the few published professional and scientific articles. In previous descriptions of the Croatian and Czech languages, the apposition is viewed in two ways – as a sentence element that is part of a noun phrase or as a sentence relationship of two or more syntagms. In Croatian grammars, apposition is usually understood as an appositive syntagm in the strict sense, i.e. a noun syntagm consisting of two juxtaposed nouns. In addition to grammatical criteria, in determining the apposition, authors dealing with the Croatian language also introduce a semantic criterion, i.e. discuss whether the apposition is an additional provision to a noun that defines and denotes it more clearly, whether it names the same concept in other words and has it got a wider or narrower meaning than the governing noun (Pavešić i Vince 1971, Mrazović i Vukadinović 1990, Težak i Babić 2005, Znika 2008). In addition, some authors (Kuna 2002) classify appositions as attributes. Some authors consider the juxtaposed syntagms that could be in the appositional relationship as a subsequent addition and subject of the description of elementary orthography (Marković 2008), while others extend the definition of apposition to the sentence relationship of the two syntagms, emphasizing that the noun itself cannot function as an apposition and such function can only be performed by a noun phrase whose notion completely or partially coincides with the notion of the noun to which it refers (Kovačević 1966, Pavešić and Vince 1971). Most interpretations of the apposition in Czech syntaxes (Havránek and Jedlička 1960, Šmilauer 1966, Grepl and Karlík 1989) rely on the structural-functional principles and theory of V. Mathesius, who defines apposition as a relation of broad identity and has similarities with other syntagmatic relations – predication, determination and coordination (Mathesius 1947, Cesar 2013). In Czech grammars, apposition is regularly interpreted as a sentence, appositive relationship that encompasses the relationship between non-restrictive complements and govern syntagms, not only of the noun type. This is actually a relationship between two syntagms, and not relationships within a single syntagm. Noun syntagms that are indicated as appositions in Croatian literature, are presented in the Czech language exclusively as noun attributes, where the predicate relationship determines which noun is the main one and which is its attribute. The review points out that the descriptions of the apposition in the two Slavic languages differ significantly, although they refer to the same linguistic phenomenon. Descriptions for the Croatian language do not meet the real needs in computational linguistics, they cannot explain the realizations of appositional structures that appear in real sentences, which is why we consequently started a corpus research of apposition. 2. Models of corpus representations of the apposition In this section, theoretically grounded and practically implemented models are presented, with the help of which the apposition as a grammatical category is annotated in syntactically annotated corpora, i.e. treebanks. The sources of corpus research were the two largest available syntactically annotated corpora – the Croatian Dependency Treebank (HOBS) (Tadić 2007) and the Universal Dependencies Corpus for the Croatian Language (Agić and Ljubešić 2015). The first part of the section explains the basic settings of selected available treebanks for both languages, the basic principles of annotation the dependent relationships between parts of a sentence and the bases of the label that these parts carry. The second part of the section describes in detail how the apposition was approached as a linguistic phenomenon in these treebanks, how it is annotated and in what relation it is to other parts of the sentence. 3. Redefining the apposition: In the last section the main goal of this thesis is achieved, new criteria for determining the apposition are adopted and it is redefined with regard to the data obtained in the corpus survey. The research was conducted on a subcorpus of sentences made up of sentences from HOBS and UD-HR corpus in which at least one sentence node is annotated as apposition, in order to obtain a relevant data source for redefining apposition. First, the structure of data obtained from syntactically annotated corpora is briefly presented. Apposition is defined as the sentence relationship between two syntagms that together form an appositional structure. Elements of the appositional structure are interconnected by connecting links, which can be either a punctuation mark or one of the linking expressions. Elements of the appositional structure are in an equal determinative relation to the superordinate element. Two related syntagms are part of the same sentence element, i.e. they perform the same function in a sentence, for example that of a subject or object. There are no clear grammatical criteria for determining which part of the appositional structure is superordinate and which is subordinate, especially since both are in the same relationship of subordination to the superordinate element. However, since we have determined that the second syntagm from the appositional relationship in some way additionally determines the first one semantically, it is concluded that the first part is the main, and the second part is considered an apposed part of the appositional structure. In the appositional structure, there are rules for formally expressing the appositional relationship. These formal ways refer to the forms in which the apposed member may appear. With the corpus we confirmed and exemplified five different forms in which the appositional structure appears. We sorted all the sentences from the corpus according to the connecting link by which the parts of the appositional structure are connected, and we detected all the ways in which the two syntagms can be connected by an appositional relationship. A total of 18 connecting expressions and 4 connecting marks were confirmed. Many authors emphasize reversibility as a property of apposition. It is not possible in principle, precisely because of the semantic significance of the second, apposed part of the appositional structure. Parts of the appositional structure name the same extralinguistic phenomenon, but emphasize different parts of that phenomenon. Thanks to that fact, but also to the fact that connecting expressions in themselves carry some meaning, we can talk about semantic types of appositional relationships. In our corpus, we have identified ten semantic types of appositional relationships, some of which also have subtypes. These are: specifying, identifying persons, identifying other named entities, naming, shortening, enumerating, exemplifying, highlighting, correcting, summarizing/classifying. Several meanings can be pronounced simultaneously by an appositional relationship, such as naming and enumeration, which we have called hybrid semantic types of appositional relationships. Not all semantic types are equally represented. In addition to the above, the thesis additionally explains the noun attribute. This stems from the study of syntagms in which two nominative nouns coincide. Nouns in genitive, dative and instrumental which are located next to the head of the syntagm are traditionally considered a noun attribute. Examples in the corpus have shown that in noun phrases in which two nouns coincide, the first of them is always the superordinate part, the syntactic head, and that the right noun in such a syntagm can be considered a subordinate part of the syntagm. If the subordinate part in genitive, dative and instrumental is viewed as a noun attribute, we believe that such an addition to the noun in nominative can also be interpreted as a noun attribute, which is usually not in agreement because in most instances it will not change the case with the main noun of the syntagm. Such an attribute changes by case only if it is a general noun that is not part of a petrified expression. The research also showed that treebanks built according to specific linguistic theoretical assumptions are a suitable resource for researching grammatical phenomena such as the apposition due to a good theoretical foundation and consistency. |