Ovaj je rad posvećen prijedložnoj dopuni u španjolskome, koju se opisuje u odnosu s predikatom i ostalim glagolskim komplementima te se određuju i analiziraju njezini ekvivalenti u hrvatskome. Prijedložna je dopuna argument glagolu, po čemu nalikuje izravnomu objektu, no oblika je prijedložnoga izraza, zbog čega se dugo svrstavala u skupinu priložnih oznaka. U literaturi se spominje više mogućih kriterija za njezinu identifikaciju i razlikovanje od ostalih komplemenata (supstitucija naglašenim zamjenicama uz čuvanje prijedloga, nemogućnost izostavljanja, nekompatibilnost s izravnim objektom uz isti glagol, prijedložna rekcija, gubitak značenja prijedloga itd.), od kojih svaki pokazuje određene poteškoće, a neki su i odbačeni, kako se i prikazuje u ovome radu. Ciljevi su rada odrediti ekvivalente španjolske prijedložne dopune u hrvatskome te stvoriti preduvjete za razvoj sustava za prepoznavanje tih elemenata u obama jezicima. Istraživanje je provedeno rabeći metodologiju kontrastivne analize i korpusni pristup, po uzoru na kontrastivne projekte Rudolfa Filipovića na Zavodu za lingvistiku Filozofskoga fakulteta Sveučilišta u Zagrebu. Sastavljen je usporedni španjolsko-hrvatski korpus s 2 milijuna pojavnica, koji se sastoji od španjolskih romana i njihovih objavljenih prijevoda na hrvatski. Kako bi se došlo do primjera prijedložne dopune, u korpusu su pretražene konstrukcije glagol + prijedlog, kod kojih prijedlog može uvoditi prijedložnu dopunu, a u analizu su uključene sve konstrukcije glagol + prijedložna dopuna s apsolutnom čestotom 15 i više, što čini ukupno 136 konstrukcija u 10.030 primjera. Skupine za analizu i usporedbu s hrvatskim određene su na temelju (ne)povratnosti konstrukcije, vrste konstrukcije s prijedložnom dopunom te prisutnosti drugih komplemenata u primjerima iz korpusa. Analiza je pokazala kako prijedložna dopuna u španjolskome ne odgovara samo jednoj sintaktičkoj funkciji u hrvatskome, nego su njezini ekvivalenti u hrvatskome različite dopune (neizravni i izravni objekt, adverbijalna dopuna, predikatna dopuna, subjekt) te u malom broju primjera adverbijalni dodatci. Semantički gledano, također je vidljiva heterogenost ovih elemenata te se osim najčešće semantičke uloge, trpitelja radnje, u hrvatskim primjerima ostvaruje i niz drugih uloga. Kod usporedbe prijevoda prijedložne dopune i drugih komplemenata istoga oblika, pokazalo se kako je hrvatski u većini slučajeva pridonio njihovu razlikovanju (prijevod drugim glagolima i/ili drukčijim konstrukcijama).
The main topic of this dissertation is the prepositional argument in Spanish (spa. complemento de régimen preposicional), a verbal complement which shares its form (prepositional phrase) with the adverbial adjunct or argument (spa. complemento circunstancial), but it’s similar in meaning to the direct object. This is why it was necessary to first describe verbal arguments and adjuncts in Spanish, with a special focus on direct and indirect object and the adverbial, to later compare their characteristics with those of the prepositional argument. The contrastive analysis with Croatian is conducted in order to try and use the inflexional richness of the Croatian language, where nouns can be marked for seven different cases, to better understand and distinguish between verbal arguments and adjuncts in Spanish, especially those formally identical. There are two main objectives: 1) to determine the Croatian equivalents of the prepositional argument in Spanish, assuming those will mainly be arguments of different kinds and 2) to lay the theoretical foundation for developing a system for detecting these structures in both languages, building on the assumption that the application of Croatian as a metalanguage will be of use. The dissertation is divided into three parts: theory, methodology and analysis. In the theoretical part we introduce and discuss the topics of transitivity, valency and rection, to then examine how verbal arguments and adjuncts are treated in Spanish and Croatian grammar. We especially focus on the distinction between these two groups of verbal modifiers in both languages, a frequent topic throughout functional grammar, valency theory and dependency grammar. The last chapter in the theoretical part of this work is dedicated to the prepositional argument in Spanish, from the seminal work of Alarcos Llorach (1966) where it was postulated, through later contributions (Bosque, 1983; Rojo, 1990; Martínez García, 1986; Cano Aguilar, 1987; Gutiérrez Araus, 1987; Alarcos Llorach, 1986, 1990, 1994/1998; García-Miguel, 1995b) which expand and refine the theory, focusing on its different aspects. Here we have decided to approach the topic concentrating on the proposed criteria for the identification of the prepositional argument in opposition to other verbal modifiers, namely, is it compatible with the direct object, whether it’s marginal or not towards the predicate and whether it can be omitted, how can it be substituted and what is the nature of its preposition. The methodology for this study is based on the one developed for the numerous contrastive projects carried out by Rudolf Filipović at the Institute of Linguistics, Faculty of Humanities and Social Sciences, University of Zagreb from the late 1960s. This includes the contrastive analysis of two languages using the translation method to identify the elements which have the same meaning and then describing those elements syntactically and semantically. For this purpose, the data was extracted from a parallel corpus, and since a suitable one didn’t exist, we constructed it. This parallel Spanish-Croatian corpus contains 2 million words (2 Mw) and is composed of eleven Spanish novels and their published translations to Croatian. The corpus was lemmatized and morphosyntactically (MSD) tagged and has been made freely available to be used in future language research. Based on the previous work about the prepositional argument (Martínez García, 1986; Cano Aguilar, 1987 and 1999; García-Miguel, 1995b; Candalija Reina, 2008) and the valency database of Spanish verbs – ADESSE (García-Miguel, González Domínguez i Vaamonde, 2010), we compiled a list of 560 verbs with one to five specific prepositions that can introduce the prepositional argument, allowing there to be up to five other elements between the verb and the preposition, which could be to the left or to the right of the verb. All the results were manually checked and we included in the analysis constructions for which the same preposition with the same verb introduced the prepositional argument in at least 15 cases. This way we got a list of 136 constructions in 10,030 examples in total. The results were divided into seven groups, depending on the verb meaning in ADESSE, whether the construction was reflexive or not, what the type of the construction was (whether the prepositional argument was, with the exception of the subject, the only argument – biactantial construction, or there were others, for example, the direct object – triactantial construction) and whether there were other constructions yielded by the aforementioned corpus search. This functionally and semantically complex element in Spanish doesn’t have an equivalent in one syntactical function or semantical role in Croatian. It was translated mostly by verbal arguments (all forms of indirect and direct object, adverbial argument, part of nominal or verbal predicate, subject) and in rare cases even by adverbial adjuncts. Semantically speaking, the most frequent semantical role these elements expressed in Croatian was the patient, but many others were found (based on the Croatian Valency Lexicon of Verbs – CROVALLEX: Mikelić Preradović, 2014). In the analysis of the corpus data, it was found that the translation of the reflexive and nonreflexive construction with the prepositional argument didn’t show any differences, which is in accordance with the conclusion that the relation between the verb and this argument is the same in both constructions (Cano Aguilar, 1987 and 1999). Comparing the translations of the prepositional argument in biactantial and triactantial construction with the same verb, there were no differences and most of the Croatian verbs used were also trivalent, with the third argument expressed or not, according to the type of the construction. Furthermore, comparing the translation of the prepositional argument and other verbal modifiers with the same verb, some were different (e.g. adverbs for adverbial modifiers, verb with a different preposition and/or case for direct and indirect object) and some were the same (e.g. when direct object and prepositional argument have the same semantic role, when the two arguments should be distinguished according to the animacy of their referent). Looking into the translations of the examples of a different verbal meaning than the one carrying the prepositional argument, mostly differences were found, but the similarities between the verbal meanings and the analysis criteria for this distinction in ADESSE were also a factor. Lastly, for some exceptionally polysemous verbs, that can be part of many different constructions and also form different expressions or phrases (e.g. llevar, hacer), additional restrictions should be introduced, in order to minimize the manual work on the data.