2018 Friday Poster 6527

Friday, November 2, 2018 | Poster Session I, Metcalf Small | 3pm

When predictions fail: Adults and children stop predicting upcoming syntactic categories in unreliable contexts
B. Axel, N. Havron, I. Dautriche, A. de Carvalho, A. Christophe

To comprehend language, listeners must deal with a considerable amount of linguistic variability at all levels (speaker, situation, noise, Kamide, 2012). To do so, adult listeners quickly adjust their linguistic expectations—based on previous experience— to the statistics of a novel environment (Fine et al., 2013). It remains unclear however, when this ability develops. One hypothesis is that the architecture underlying linguistic adaptation remains the same throughout development and may be the key component through which language learning unfolds (Ellis, 2006). Alternatively, adaptation might be a feature that is only available in the mature language processing system as it may require robust linguistic knowledge and general cognitive skills, yet to be acquired by children. To investigate this question beyond adaptation occurring at phonemic perception (Yurovsky, Case, Frank, 2016), we focused on syntactic adaptation. We tested whether French 4-to-5-year-olds (n=48) and adults (n=48) stop relying on function words to predict the syntactic category of upcoming words (verb or noun) when exposed to a speaker who uses function words unreliably.

In a visual-world procedure, participants were randomly assigned to one of two conditions: In the Reliable Condition they heard a reliable speaker correctly using determiners before nouns (e.g., une banane, “a banana”) and pronouns before verbs (e.g., Elle mange, “She eats”); in the Unreliable Condition they heard a speaker incorrectly using determiners before verbs (e.g., *An eat) and pronouns before nouns (e.g.,

*She banana, see fig.1). All participants had the same test-trials (half informative and half non-informative). In informative test-trials, the two pictures depicted words of different syntactic categories (e.g., picture 1: a costume; picture 2: a boy descending stairs), so that the function word (e.g., il … ‘he …’) could be used to anticipate the target word (i.e., a pronoun predicts a verb and therefore the picture representing an action). In non-informative test-trials, the two pictures represented words of the same category (two actions or two objects), so that the function word could not help to select between pictures. To test for the effect of adaptation, we compared the gaze of participants in informative vs non-informative test-trials, depending on their condition.

Both children and adults in the reliable condition anticipated target words during the informative test-trials compared to non-informative test-trials (see fig.2). Moreover, participants in the unreliable condition stopped predicting upcoming syntactic categories, showing no anticipation effect. The interaction between condition and trial-types was marginally significant for children and significant for adults. The weak interaction might be coming from a weak anticipation effect in the reliable group. For example, when the target word is déguisement (“costume”) and the competitor image depicts découpe (‘cut’), with scissors cutting paper, children might have been slow to anticipate, thinking about “paper” as a possible noun candidate. We are currently running a second experiment in which images are named before the session, in the hope of increasing anticipation (as is usually done in adult studies visual-world studies). The current results already suggest that predictions are updated rapidly and continuously, even in children.

References

Ellis, N. C. (2006). Language acquisition as rational contingency learning. Applied linguistics, 27(1), 1-24.

Fine, A. B., Jaeger, T. F., Farmer, T. a, & Qian, T. (2013). Rapid Expectation Adaptation during Syntactic Comprehension. PloS One, 8(10), e77661. https://doi.org/10.1371/journal.pone.0077661

Kamide, Y. (2012). Learning individual talkers’ structural preferences. Cognition, 124(1), 66-71.

Yurovsky, D., Case, S., & Frank, M. C. (2016). Preschoolers Flexibly Adapt to Linguistic Input in a Noisy Channel. Psychological Science, 1 –9. https://doi.org/10.1177/0956797616668557