期刊
LINGUISTICS VANGUARD
卷 -, 期 -, 页码 -出版社
WALTER DE GRUYTER GMBH
DOI: 10.1515/lingvan-2022-0122
关键词
linguistic constructions; construction extraction; natural language processing; computer-assisted language learning
This article introduces an open-source tool, Constraction, for the automatic extraction and interactive exploration of linguistic constructions from textual corpora. Constraction features a generic algorithm and a browser-based interface, allowing customizable layers of linguistic annotation and visual representation of extracted patterns. Case studies demonstrate the utility of Constraction in language research and pedagogy.
A central task in empirical and quantitative language studies is the extraction of linguistic constructions important to linguistic theory and application. The great number and variety of such constructions increasingly necessitates computer-assisted extraction, which often proves challenging as it entails a simultaneous analysis of multiple layers of linguistic information latent in large-scale corpora. To address this, we present Constraction, an open-source tool for the automatic extraction and interactive exploration of linguistic constructions from arbitrary textual corpora. Constraction features a generic algorithm that integrates customizable layers of linguistic annotation (e.g., lexical, syntactic, and semantic) to identify constructional patterns of varying sizes and abstraction levels. Its browser-based interface allows users to configure various extraction parameters and enables visual, interactive exploration of the extracted patterns. We demonstrate the utility of Constraction through case studies and discuss its potential applications in language research and pedagogy.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据