What has LeBenchmark Learnt about French Syntax? - GETALP Access content directly
Conference Papers Year : 2024

What has LeBenchmark Learnt about French Syntax?

Abstract

The paper reports on a series of experiments aiming at probing LeBenchmark, a pretrained acoustic model trained on 7k hours of spoken French, for syntactic information. Pretrained acoustic models are increasingly used for downstream speech tasks such as automatic speech recognition, speech translation, spoken language understanding or speech parsing. They are trained on very low level information (the raw speech signal), and do not have explicit lexical knowledge. Despite that, they obtained reasonable results on tasks that requires higher level linguistic knowledge. As a result, an emerging question is whether these models encode syntactic information. We probe each representation layer of LeBenchmark for syntax, using the Orféo treebank, and observe that it has learnt some syntactic information. Our results show that syntactic information is more easily extractable from the middle layers of the network, after which a very sharp decrease is observed.
Fichier principal
Vignette du fichier
2024.lrec-main.1521.pdf (723.34 Ko) Télécharger le fichier
Origin Publisher files allowed on an open archive
licence

Dates and versions

hal-04592628 , version 1 (29-05-2024)

Licence

Identifiers

Cite

Zdravko Dugonjić, Adrien Pupier, Benjamin Lecouteux, Maximin Coavoux. What has LeBenchmark Learnt about French Syntax?. LREC-COLING 2024, May 2024, Torino, Italy. pp.17493-17499. ⟨hal-04592628⟩
10 View
4 Download

Altmetric

Share

Gmail Mastodon Facebook X LinkedIn More