Gendering subtitles? Evidence from a Netflix corpus

Abstract

We start from the observation that language pairs like Polish and English display an intriguing asymmetry: Polish has the structural resources to express grammatical gender in nouns while English predominantly leaves these under-determined. Therefore, a speaker of English can conventionally say “I am a nurse” irrespective of the speaker’s gender, while in Polish male/female/non-binary speakers would likely use variable forms. This asymmetry leads to nuanced translation choices which this paper aims to explore. Based on a newly compiled corpus of English-to-Polish Netflix subtitles – amounting to 3.5 million words – we examine how 113 different occupation-related feminatives are deployed by translators. We address three main research questions (RQ1-RQ3). RQ1 is about the frequency of the relevant feminatives in our corpus. RQ2 deals with how the feminatives are distributed across the 284 productions analysed in the dataset. Finally, RQ3 addresses the relationship between the frequency of feminatives in Netflix subtitles and the acceptability of these feminatives, as judged by speakers.

Keywords

• gender; feminativa; interlingual subtitling; corpus data; streaming platforms.