Predictability in French gender attribution: A corpus analysis

Abstract

This article presents a corpus analysis designed to determine the extent to which noun endings in French are reliable predictors of grammatical gender. A corpus of 9,961 nouns appearing in Le Robert Junior Illustré was analysed according to noun endings, which were operationalised as orthographic representations of rhymes, which consist of either a vowel sound (i.e., a nucleus) in the case of vocalic endings or a vowel-plus-consonant blend (i.e., a nucleus and a coda) in the case of consonantal endings. The analysis classified noun endings as reliably masculine, reliably feminine, or ambiguous, by considering as reliable predictors of grammatical gender any noun ending that predicts the gender of least 90 per cent of all nouns in the corpus with that ending. Results reveal that 81 per cent of all feminine nouns and 80 per cent of all masculine nouns in the corpus are rule governed, having endings that systematically predict their gender. These findings, at odds with traditional grammars, are discussed in terms of their pedagogical implications.