Slovenská jazykovedná spoločnosť

As Horsch and Boas have noted, “[o]ne of the main goals of constructional research is to find, document, and analyze all constructions of a language” (2025fc.), i.e., ‘mapping out’ the constructicon. Accordingly, projects have been initiated to ‘build’ constructicons for various languages, including English (Fillmore et al., 2012), Swedish (Lyngfelt et al., 2018), Brazilian Portuguese (Laviola et al., 2017), Russian (Janda et al., 2018), German (Ziem et al., 2019), Italian (Pannitto et al., 2024), Hungarian (Sass, 2024) and, most recently, Norwegian (Endresen and Mikkelsen, 2024). Inspired by these projects and acknowledging calls “to build constructicon resources for additional natural languages”, with the goal of “taking Construction Grammar to a new quantitative and qualitative level” (Janda et al., 2020, p. 161), in this contribution I discuss the need for a Slovak constructicon and consider concrete steps that will be required for building one. Based on other constructicon projects, especially the Russian one (Janda et al., 2020), we suggest a five-year time frame and three major phases: (1) Initial inventory (~2 years), (2) corpus-based expansion (~1 year), after which a “critical mass of over one thousand” constructions (Janda et al., 2020, p. 167) should be reached, and (3) system-based expansion (~2 years). We recommend focusing on partially schematic constructions such as the comparative correlative (cf. Horsch, 2019, 2021, 2023) at first, because these are “recurrent linguistic patterns that ‘fall between the cracks’ of dictionaries and grammars” (Janda et al. 2020: 164). Furthermore, we discuss whom the Slovak constructicon should be aimed at (linguists, learners, NLP researchers) and how this will influence presentation, e.g. the user interface. We also touch upon semantic and syntactic annotation (semantic tags adapted from the Russian constructicon and Universal Dependencies/Leipzig glossing), and how the relationship between constructions may be visualized using networks. Finally, we address data sources for identifying constructions: Crowd-sourcing, textbooks, research papers and corpus data (as suggested by Janda et al., 2020). [menej]