A Methodology for Building a Diachronic Dataset of Semantic Shifts and its Application to QC-FR-Diac-V1.0, a Free Reference for French

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 2117–2125 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 2117 A Methodology for Building a Diachronic Dataset of Semantic Shifts and its Application to QC-FR-Diac-V1.0, a Free Reference for French David Kletz†, Philippe Langlais†, Franc¸ois Lareau‡, Patrick Drouin‡ Universit´e de Montr´eal †Department of Computer Science and Operational Research (DIRO) / RALI ‡Department of Linguistics and Translation {david.kletz, philippe.langlais, francois.lareau, patrick.drouin}@umontreal.ca Abstract Different algorithms have been proposed to detect semantic shifts (changes in a word meaning over time) in a diachronic corpus. Yet, and somehow surprisingly, no reference corpus has been designed so far to evaluate them, leaving researchers to fallback to troublesome evaluation strategies. In this work, we introduce a methodology for the construction of a reference dataset for the evaluation of semantic shift detection, that is, a list of words where we know for sure whether they present a word meaning change over a period of interest. We leverage a state-of-the-art word-sense disambiguation model to associate a date of first appearance to all the senses of a word. Significant changes in sense distributions as well as clear stability are detected and the resulting words are inspected by experts using a dedicated interface before populating a reference dataset. As a proof of concept, we apply this methodology to a corpus of newspapers from Quebec covering the whole 20th century. We manually verified a subset of candidates, leading to QC-FR-Diac-V1.0, a corpus of 151 words allowing one to evaluate the identification of semantic shifts in French between 1910 and 1990. Keywords: Semantic Shift Identification, Semantic Neology, Neology, Reference Corpus, Evaluation 1. Introduction The meaning of a word is subject to variation over time. Many phenomena can trigger these changes, such as the appearance of new technology or social changes. Thus, an existing word can be assigned a new mean- ing, have a meaning withdrawn or remain stable be- tween two given dates. The study of semantic shift is a field that focuses on these changes in meaning over time. Semantic shift, or semantic neology, is differ- ent from lexical neology. Detecting lexical neologisms across periods of a diachronic corpus boils down to de- tecting new, recurrent tokens in the latest period, which is a fairly straightforward process. With semantic ne- ology however, the same token appears with different meanings in different periods. Detecting these cases is more involved. In this paper, we use the term semantic shift to refer to different cases, as explained below. The first case is the appearance of a new word sense. This can be further divided into new polysemy and new homonymy, depending on how the new sense is related to an existing one. But the difference between the two is basically one of degree, and there are cases that sit on the fence. For example, consider the French noun souris (‘mouse’) between 1900 and today. The first sense for this word is defined as1 petit mammif`ere rongeur …

Content truncated to 500 words