29 September 2026 to 3 October 2026
OeAW Main Seat
Europe/Vienna timezone

Trend Word Detection for DWDS, a Lexical Information System for Contemporary German

1 Oct 2026, 15:00
30m
01 | Sitzungssaal (01 | Sitzungssaal, OeAW Main Seat, 1st floor)

01 | Sitzungssaal

01 | Sitzungssaal, OeAW Main Seat, 1st floor

Austrian Academy of Sciences Dr. Ignaz Seipel-Platz 2 1010 Vienna

Speakers

Alexander Geyken (Berlin-Brandenburg Academy of Sciences and Humanities) Gregor Middell (Berlin-Brandenburg Academy of Sciences and Humanities)

Description

This paper addresses the challenge of prioritizing dictionary revisions in corpus-based lexicography, specifically for the Digital Dictionary of German (DWDS), which relies heavily on legacy content. Given the massive data volume in modern corpora, manual selection of relevant words for updating is no longer feasible. We investigate two methods for identifying “trend words”—terms showing an overproportionally high present-day frequency relative to past intervals. The first method applies established linear regression to large monitor corpora, highlighting limitations due to reliance on the chosen time period. To mitigate this, the second method employs a dispersion-oriented keyness metric on a specialized “discourse corpus” of current online text. This metric scores lemmas based on their spread across unique sources, effectively prioritizing words relevant to contemporary discourse. Evaluation of the ranked candidate list confirms that the high-score group contains a significantly higher density of relevant trend words compared to lower-ranked groups. This system allows lexicographers to systematically prioritize legacy entries for revision, such as those that are outdated or missing a current sense, thereby optimizing the lexicographical workflow and ensuring the dictionary reflects current language use.

Presentation materials

There are no materials yet.