This contribution aims at analyzing the Evaluative Binominal Construction (EBC) (e.g., EN a hell of a day), represented as the constructional schema [DET1 N1 PREP DET2 N2], which connects an N1 expressing an evaluative quality and an N2 indicating the entity to which the quality is attributed. The EBC has already been studied in several languages such as English (Aarts 1994, 1998), Dutch (Verhagen 2005, Van Goethem 2023) and French (Larrivée 1994), but cross-linguistic studies (such as Piunno 2022 for Romance languages) are scarce and often not corpus-based (e.g., Foolen 2004). The main objective of our study is to fill this gap by providing a corpus-based cross-linguistic constructionist analysis of the EBC. Our study includes four languages, two Germanic (English and Dutch) and two Romance languages (French and Italian). Our data are taken from the TenTen corpus family on SketchEngine. Firstly, only the top 20 EBC patterns per language are examined for a frequency analysis and classification in terms of their evaluative semantics and competing constructions. Secondly, the top 5 EBCs of each language is analyzed in terms of formal properties, productivity, and semantic specialization, based on corpus samples of 100 relevant occurrences per pattern (i.e., datasets of 500 occurrences per language). Preliminary results reveal on the formal level a preference for an indefinite article in DET1 position in Germanic and for a demonstrative in Romance. Furthermore, the Romance EBC often uses nominalized adjectives in the N1 slot (FR ce drôle de N; IT quell’idiota di N), contrary to the Germanic EBC. The constructional semantics is almost exclusively positive in Dutch (ameliorative/augmentative connotations), both positive and negative in English, and mostly pejorative in Romance. Also, the semantics of the N1 differs between Romance and Germanic languages, referring more frequently to a quality in French and Italian (FR cette saloperie d’histoire ‘this fucking story’), and more frequently to an object or a ...