When French becomes Canadian French: The curious case of localizing COVID-19 terms with Microsoft Translator

Lynne Bowker*, Frédéric Blain

*Corresponding author for this work

Research output: Contribution to journalArticleScientificpeer-review

5 Citations (Scopus)

Abstract

In late 2020, the free online translation tool Microsoft Translator began to offer the option of translating into “French (Canada)” as a target language, alongside the previously offered “French”. Using a list of ten COVID-19 terms previously identified by Bowker (2020) as having different equivalents in Canadian French and European French, we evaluate the ability of Microsoft Translator to localize these terms into the two varieties of French. The findings indicate that while this tool does a good job of localizing the terms into Canadian French, it also uses a high number of Canadian French terms when the target language is set to “French”. One potential reason for this may be that the corpus used to train the tool for “French” contains a disproportionate number of examples from Canadian sources, and so there may be a problem of bias where the tool is amplifying Canadian French in the machine translation output.

Original languageEnglish
Pages (from-to)1-37
Number of pages37
JournalJournal of Internationalization and Localization
Volume9
Issue number1
DOIs
Publication statusPublished - 13 Oct 2022
Externally publishedYes

Keywords

  • bias
  • Canadian French
  • COVID-19 terminology
  • French
  • language varieties
  • localization
  • neural machine translation

Fingerprint

Dive into the research topics of 'When French becomes Canadian French: The curious case of localizing COVID-19 terms with Microsoft Translator'. Together they form a unique fingerprint.

Cite this