Sõnastikust konstruktikoniks? Taustu, eeskujusid ja väljakutseid

Abstract

A new field has emerged alongside constructional linguistic theories and electronic lexicography, known as constructicography (Lyngfelt 2018: 11). Its goal is to compile a constructicon, a resource that organizes and presents linguistic constructions much like dictionaries present words. This concept stems from the theoretical framework of Construction Grammar (see Fillmore 2006; Goldberg 2003; Hoffmann, Trousdale 2013). The main purpose of this study is to outline the necessary theoretical concepts, map out existing solutions, and identify the main challenges in creating different types of constructicons. This aims to inform decisions when beginning to create a potential Estonian language constructicon. First, we discuss the main theoretical considerations and concepts essential for constructicon building. The basic assumption of this theory is that there is a continuum-like transition from grammar to lexicon, rather than a sharp boundary between these two basic linguistic phenomena. According to this view, linguistic units are constructions – pairings of form and meaning with a variable number of lexical components and various levels of schematicity. We further examine the theoretical and methodological principles behind current constructiographic practices, using examples from eight initiatives across seven languages: English, German, Japanese, Russian, Hungarian, Swedish, and Brazilian Portuguese (Perek, Patten 2019; Ziem et al. 2019; Janda et al. 2020; Lyngfelt et al. 2018; Ohara 2018; Fillmore et al. 2012; Torrent et al. 2014; Sass 2023). Meta-analysis reveals that most constructicographic databases are linked to FrameNet resources created for respective languages. The target audiences for these resources vary from language experts to second language learners and language technology applications. The coverage of constructions ranges from 73 to approximately 13,000 entries. The constructicons include semi-schematic constructions, abstract valency patterns, argument structure constructions, and even idiomatic constructions located at the “lexical end” of the grammar–lexicon continuum. As constructions form networks characterized by inheritance relations, describing these relations is a crucial part of constructicon building. Challenges discussed in the paper include defining a construction – whether narrowly or broadly; deciding if a constructicon should encompass general phrase structure constructions or be limited to idiosyncratic constructions; determining which constructions should be included in the resource and how to arrive at a nomenclature; handling constructions at different levels of abstraction; ensuring that the meta-language of descriptions is accessible to learners, native speakers, linguists, and NLP applications alike; and providing effective search functionalities in the dictionary/database for users. Based on the analysis results, we highlight questions that need to be addressed if a constructicon is to be created as part of a relational lexicographical database for Estonian, specifically within the database of the EKI Combined Dictionary compiled within DWS Ekilex (Tavast et al. 2018).

Keywords

• linguistics
• lexicography
• construction-based approach
• constructicon
• constructicography