GASC: Genre-Aware Semantic Change for Ancient Greek

Abstract

Word meaning changes over time, depending on linguistic and extra-linguistic factors. Associating a word's correct meaning in its historical context is a central challenge in diachronic research, and is relevant to a range of NLP tasks, including information retrieval and semantic search in historical texts. Bayesian models for semantic change have emerged as a powerful tool to address this challenge, providing explicit and interpretable representations of semantic change phenomena. However, while corpora typically come with rich metadata, existing models are limited by their inability to exploit contextual information (such as text genre) beyond the document timestamp. This is particularly critical in the case of ancient languages, where lack of data and long diachronic span make it harder to draw a clear distinction between polysemy (the fact that a word has several senses) and semantic change (the process of acquiring, losing, or changing senses), and current systems perform poorly on these languages. We develop GASC, a dynamic semantic change model that leverages categorical metadata about the texts' genre to boost inference and uncover the evolution of meanings in Ancient Greek corpora. In a new evaluation framework, our model achieves improved predictive performance compared to the state of the art.

A computational approach to lexical polysemy in Ancient Greek

“…Nonetheless, these studies usually focus on the change of the primary sense of a word, and do not capture the inter-sense dynamics of polysemy across time. To the best of our knowledge, aside from the model used in this paper, the only advanced technique that tackles such dynamics is SCAN (Frermann and Lapata, 2016) and its extension known as GASC (Perrone et al 2019).…”

“…The analyses illustrated so far show the complexities of semantic change phenomena in Ancient Greek. In the rest of this article, in order to answer our original research questions on large-scale analysis of polysemy, we discuss the challenges involved in creating the Genre-Aware Semantic Change (GASC; Perrone et al 2019), a computational model aimed to allow scholars to scale up their manual analyses and identify general patterns in the semantic evolution of the Ancient Greek lexicon. Perrone et al (2019) applied the SCAN model by Frermann and Lapata (2016), which extends a Bayesian topic model to the case of semantic change in English in the 16 th -20 th centuries.…”

“…In the rest of this article, in order to answer our original research questions on large-scale analysis of polysemy, we discuss the challenges involved in creating the Genre-Aware Semantic Change (GASC; Perrone et al 2019), a computational model aimed to allow scholars to scale up their manual analyses and identify general patterns in the semantic evolution of the Ancient Greek lexicon. Perrone et al (2019) applied the SCAN model by Frermann and Lapata (2016), which extends a Bayesian topic model to the case of semantic change in English in the 16 th -20 th centuries. This enabled the analysis of the model's estimated probability distribution of a word's meaning over time and raised interesting questions regarding the challenges of this computational research on Ancient Greek.…”

“…This enabled the analysis of the model's estimated probability distribution of a word's meaning over time and raised interesting questions regarding the challenges of this computational research on Ancient Greek. Secondly, Perrone et al (2019) combined the insight on the role played by genre in semantic change phenomena in historical languages within a new computational model. Originally, GASC infers word meanings not only from the context of words, but also from the genre of the texts, making it possible to estimate the probability distribution of a word's meaning over time and by genre.…”

“…This can be described as a shift in probabilities, and it is clear that language spread should be dealt with in probabilistic terms. (Jenset and McGillivray (2017:44)) Here we explore the challenges of quantitatively modelling semantic change and polysemy based on the model by Perrone et al (2019), and we discuss preliminary results of a statistical model for semantic change in Ancient Greek. The focus on a historical language with a long textual tradition allows us to address important methodological challenges.…”