Neutral is not fair enough: testing the efficiency of different language gender-fair strategies

Abstract

In many languages with grammatical gender, the use of masculine forms as a generic reference has been associated with a bias favoring masculine-specific representations. This article examines the efficiency of gender-fair forms, specifically gender-unmarked forms (neutralization strategy, e.g., "l'enfant") and contracted double forms (re-feminization strategy, e.g., "un·e enfant"), in reducing gender biases in language. Extensive empirical research has shown that gender-fair forms have the potential to promote more gender-balanced representations. However, the relative efficiency of these strategies remains a subject of debate in the scientific literature. In order to explore these questions, two experiments were conducted in French. We analyzed the response times and percent correct scores using a sentence evaluation paradigm, where the participants had to decide whether a second sentence starting with a gendered personal pronoun ("il" or "elle") was a sensible continuation of the first sentence written in a gender-fair form. Experiment 1 confirmed that gender-unmarked forms are not fully effective in neutralizing the masculine bias. In Experiment 2, a comparison was made between gender-unmarked forms and contracted double forms, to assess their respective abilities to generate more balanced representations. The findings indicated that contracted double forms are more effective in promoting gender balance compared to gender-unmarked forms. This study contributes to the existing scientific literature by shedding light on the relative efficiency of neutralization and re-feminization strategies in reducing gender biases in language. These results have implications for informing efforts to promote more inclusive and unbiased language practices.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 21 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.