Recent & Upcoming Talks

Because of great efforts by the corpus-linguistic community, a large number of historical texts have been digitized (and sometimes even …

Machine-based exploration of culturally relevant datasets (e.g. newspapers, periodicals, correspondence or annals) often involves …

Linguistic variation and change have long been studied as properties of community-level language, i.e. the “shared system of …

This demonstration will focus on MacBERTh, a BERT-based model pre-trained on Early Modern and Late Modern English (3.9B (tokenized) …

This presentation demonstrates how MacBERTh may help researchers to (i) access and (ii) analyse the functional-semantic information encoded in linguistic corpus data in a (semi-)automatic way. More specifically, after surveying the performance of the model on general downstream NLP tasks, we home in on a specific case study in morphosyntactic variation in English ing-forms.

In this talk, Fonteyn and Manjavacas Arevalo present MacBERTh – a transformer-based language model pre-trained on historical English – and exhaustively assess its benefits on a large set of relevant downstream tasks. The experiments highlight that, despite some differences across target time periods, pre-training on historical language from scratch outperforms models pre-trained on present-day language and later adapted to historical language.

This talk introduces MacBERTh, a transformer-based language model pre-trained on historical English. We exhaustively assess the benefits of this historically pre-trained language model on a large set of relevant downstream tasks.

This talk outlines a hands-off methodology that helps tackle some common challenges in the use of functional-semantic vector representations to study cases of constructional change. The specific case we address concerns the development of the phrasal expression to death from a literal, resultative phrase into an ‘amplifying’ expression.

In this talk, we introduce the ‘MacBERTh’ project.