Psychological predicates (PPs), referring to psychological, mental, or emotional states, pose challenges for both grammatical theory and natural language processing (NLP) tasks. While previous research has mostly addressed PPs expressed by a single word (typically verbs, but also nouns and adjectives), idioms such as ahogarse en un vaso de agua (“to worry excessively about something unimportant”) have received little attention. This study, situated within computational linguistics, focuses on verbal idiomatic psychological predicates (VIPP) in Rioplatense Spanish. We argue that the syntax of idioms occupies an intermediate position between phrase-level and sentence-level syntax, which makes them particularly challenging to model computationally. To explore this, we use NooJ, a linguistic development environment that enables the construction of linguistic resources that can be used by automatic text alorithms. Our methodology combines two components: (i) the formal description of a representative set of VIPP, and (ii) the implementation of syntactic grammars designed to recognize these idioms and their participants. The resulting algorithm was tested on a collection of sentences extracted from publicly available corpora. The evaluation demonstrates that the system successfully identifies VIPP along with their associated arguments, showing the feasibility of capturing this type of predicate in a computational framework. These findings contribute both to the grammatical characterization of idioms and to their automatic treatment in NLP applications.