Perceptual Insensibility in a Second Language
Examining gender processing in early and late bilinguals
Posted October 11, 2013
Post written by François Grosjean.
Winston Churchill once said on French radio, with his characteristic English accent when he spoke French, "Despite working so hard and coming so far with the French to help them win their freedom, I have never mastered the gender of French nouns!". His problem was a classic one for those who learn French late: Is "bateau" (boat) masculine or feminine? It's masculine. How about "montagne" (mountain)? It's feminine.
Not only is gender difficult for late learners of French (as it can be for late learners of Spanish and Italian, among other languages), but gender agreement marking on other words that accompany the noun such as an article, an adjective, or a pronoun, can also be difficult. This explains why you may hear a non-native speaker of French say, "le petit montagne" (the small mountain) instead of "la petite montagne".
We have known for some time that native listeners of languages with gender make use of gender marking cues (such as the pronunciation of the "t" in "petite" but not in "petit") to speed up the recognition of the following noun (e.g. "montagne"). The question my colleague, Delphine Guillelmon, and I asked was whether bilinguals would show the same effect. And does it depend on when they acquired their gender marking language?
Even though late learners of a gender language make more gender errors than early learners when speaking, we expected that in perception both early and late bilinguals would be sensitive to gender marking to the same extent. After all, we reasoned, if a language offers you a gender cue to speed up your recognition of the following noun, why not use it?
We asked early and late English-French bilinguals to do a very simple task: they were to listen to short phrases such as "le joli bateau" (the nice boat) and to repeat the word after "joli" (in this case, "bateau"). We compared the time it took them to do so in a congruent condition (the article "le" has the same gender as the following noun, "bateau"), in a neutral condition as in "leur joli bateau" (their nice boat) where "leur" carries no gender information, and in an incongruent condition such as "la joli(e) bateau" where the gender marking "la" is incorrect.
We first tested the early bilingual group (they had started using their two languages, English and French, as early as 5 years of age, on average) and we found that they behaved like monolingual French speakers. They too had become sensitive to gender marking early in life and they used it to speed up processing in perception.
The crucial question now became: Would late bilinguals (English speakers who had started speaking French on a regular basis at age 25, some 24 years before we tested them) show the same effect as early bilinguals? If gender marking is indeed important during language processing, then they should have become sensitive to it. However, if there is a critical (or sensitive) period for taking into account gender marking, at least in perception, and if they acquired their gender-marking language after this period, then they should show little, if any, effect.
The results we obtained surprised us. Late bilinguals were not only totally insensitive to gender congruency ("le joli bateau") but also to gender incongruency (the ungrammatical "la joli(e) bateau"). It was as if they simply could not use the masculine "le" cue or the feminine "la" cue during the processing of those short phrases.
We investigated whether this was due to a slightly slower overall speed of response (it wasn't) or to their inability to use gender agreement when speaking French (in fact, they made very few gender errors in production). Their level of language proficiency was not at stake either–their oral comprehension of French was generally excellent after more than twenty years of daily use of the language.
It would appear therefore that certain processing mechanisms in a second language will never be acquired (or only partly acquired) after a specific point in time. Of course, late English-French bilinguals still recognize words perfectly but recognition is neither facilitated by a congruent gender marking nor impeded by an incongruent one.
We could not resist concluding our study by extending Sir Winston Churchill's statement in the following way, "I have never mastered the gender of French nouns.... be it in production OR perception".
Photo of a blonde woman from Shutterstock.
"Life as a bilingual" posts by content area.
François Grosjean's website.