Contrastive phraseology: old and new challenges

Abstract This article argues that contrastive phraseology is entering a new stage of theoretical and methodological development. Drawing on corpus- and usage-based approaches, it proposes a broader conception of contrastive phraseology that extends beyond the comparison of traditional idioms to encompass the full spectrum of lexicogrammatical and phraseological phenomena, including valency patterns, collocations and their frequent combinations, formulas, formulaic sequences, and constructional idioms. The discussion demonstrates that meaningful cross-linguistic comparison requires integrated descriptions combining semantic, syntactic, pragmatic, and discourse-functional information rather than relying on lexical equivalence alone. The article further argues that recent approaches defining phraseological units in terms of idiosyncrasy rather than idiomaticity provide a more comprehensive theoretical foundation for contrastive research. From this perspective, semantic, syntactic, and pragmatic irregularities represent different manifestations of language-specific conventionalisation and are therefore equally relevant objects of phraseological analysis. Such a framework enables phraseological units to be compared across languages in terms of their communicative functions despite considerable formal differences. Finally, the article considers the potential contribution of artificial intelligence to contrastive phraseology. While large language models can facilitate corpus exploration, hypothesis generation, and multilingual data analysis, their outputs require systematic linguistic evaluation. AI should therefore complement rather than replace philologically grounded, corpus-based research, with its greatest potential lying in the support of empirically informed analyses of phraseological variation across languages.