Nominalizaciones deverbales: denotación y estructura argumental

Spanish deverbal nominalizations are linguistic constructions characterized by presenting properties of common nouns but also by inheriting the argument structure of the verbs from which they derive. This duality aroused considerable interest in deverbal nominalizations in Linguistics. On the one hand, they can denote both the state or the result of the action expressed by the corresponding base verb as well as the same action or event expressed by the base verb, in the latter being paraphrases of sentence clauses. On the other hand, nominalizations are nouns with argument taking capacity, that is, they select arguments. In this sense, it is relevant to observe the patterns of the syntactic-semantic realization of the nominalizations arguments, since they represent an alternative way of expressing the same semantic content of a sentence. Besides the intrinsic linguistic value of studying these constructions in Spanish, having tools and resources dealing with deverbal nominalizations is essential in Natural Language Processing (NLP), both in terms of denotation and argument structure. Tasks such as coreference resolution or paraphrase detection may benefit from a tool or resource that addresses the denotation type of nominalizations. Applications such as information extraction or question answering, and semantic role labelling systems may also benefit from tools and resources that represent the argument structure of deverbal nominalizations. This thesis aims to study deverbal nominalizations both from Linguistics and NLP approaches. The thesis is divided into four parts, which reflect these two perspectives. The first part of this work gives background information on deverbal nominalizations. It defines the object of study, presents the methodology used and provides an extensive review of the literature, including both theoretical and computational works on deverbal nominalizations. The second part focuses on the argument structure of deverbal nominalizations. We present our corpus-based linguistic study of the syntactic-semantic realization of arguments. From this study, we extracted a series of hypotheses about which constituents are arguments of nominalizations and which are not, and what