Grammatical Gender Influences Dutch 5-year-olds’ Pronoun Interpretation in a Pointing Task
Poster presented at SRCD 2015 in Philadelphia, PA, USA
Decades of research point to a peculiar effect when children acquire object pronouns compared to reflexives: During production, they seem to master the correct use of "her" in comparison to "herself" in sentences such as "Goldilocks is touching her/herself" (reviewed e.g. by O'Grady, 2014). But when asked for the meaning of those sentences, children up to school-age seem to be unsure of how to interpret the pronoun. Interestingly, their responses to "herself" indicate no such uncertainty.
While many proposals attribute a general underlying mechanism that is acquired late and affects all pronouns, evidence is beginning to emerge that contests this view, as for example verb type influences children's behaviour (Matthews, Lieven, Theakston, Tomasello, 2009). At the moment, no theoretical account can fully capture the diverse results that are pointing to effects of word type, frequency, and even the task (e.g. Bergmann, Paulus, & Fikkert, 2012).
If usage-based accounts (e.g., Tomasello, 2003) indeed capture the development of pronoun understanding, it should also be affected by pronoun gender (given that the masculine pronoun is by far more frequent). To assess this claim, we tested whether pronoun gender affects Dutch 5-year-olds' performance. We asked 34 children to point at the correct one of two pictures, one depicting a transitive action between two toy animals (the agent touching the patient, object pronoun interpretation) and one depicting an action of the agent with the second toy animal being an unrelated distractor. Sentences containing either the pronoun or the reflexive were read intermixed to the children. Crucially, they were tested in two blocks (order randomized across participants), with one only containing 'male' toys (Mister Frog, Mister Bear) and the other only 'female' toys (Miss Cow, Miss Cat). Thus, the gender of the pronoun did not disambiguate between possible referents.
We replicate near ceiling performance for reflexives (86 % correct, above chance with p<.001), whereas pronouns are seemingly confused in their meaning (37% correct, below chance with p=.02), performance across the two gender conditions differs significantly (p<.001). When comparing the two conditions, an advantage for the masculine pronoun emerges, which is the default, high-frequency choice in spoken and written language. Children were still at chance level for sentences referring to the male pair (50% correct, at chance level, p=.94), whereas they performed significantly worse in the condition with female toys (24% correct, below chance level with p<.001).
Our results point to an effect of linguistic gender in addition to the task, as children of the same age reliably look at the correct referent during a passive listening eye tracking task. The advantage for the masculine pronoun supports a usage-based interpretation, as this pronoun is more frequent. The performance below chance level for the feminine pronoun can be attributed to a less robust representation that more easily is mapped onto the simpler, syntactically bound reflexive interpretation. Pronoun acquisition is thus not a uniform process across instances of the phenomenon, but seems influenced by many factors such as word frequency.