Poster #54 - Comprehension of Counterfactual Constructions in Children Aged 2 and 3

• In Event: Poster Session 10

Abstract

With counterfactual expressions like “If pigs had wings, they could fly” or “I wish pigs could fly” we communicate situations that are not actual and imagine what the world would look like if they were. While children start producing counterfactual constructions as early as age 2 (Bowerman, 1986; Tulling & Cournane, 2019), many argue that basic counterfactual comprehension develops around age 4 (e.g., Riggs et al., 1998; Robinson & Beck, 2000; Nyhout & Ganea, 2019). In this study, we investigate whether this apparent reversed production-comprehension asymmetry is rooted in linguistic and task complexity. By conducting a referent selection task that probes counterfactual comprehension indirectly, we aim to find out whether 2- and 3-year-olds understand simple counterfactual expressions.

Thirty 2- and 3-year-old children will be recorded online via Lookit as they complete a preferential looking task (Figure 1) that exploits toddlers’ established ability to update mental representations (Ganea et al., 2007). A narrator will introduce twins: Peter, who loves being clean, and Lily who loves being messy, and the twins’ two identical toys. With one of them, Panoo, we continue to play hide-and-seek, while the other one (‘Panoo’s friend’) remains in the background. After playing, both toys are concealed inside separate boxes and the boxes are shuffled. Then, one of the siblings either cleans or paints one of the boxes and spills inside the box. The other sibling, in response, comments on Panoo’s state with a target utterance (Table 1), and whether the spill affected him or not, e.g., “I wish Panoo was still clean. I wish Panoo was clean”. Then the child sees the two toys, one of them wet or dirty, the other one clean and dry. The child is prompted to identify Panoo by touch or pointing. The experimental conditions vary by ‘Construction’ (Factual, Negated or Counterfactual) and ‘Panoo’s state’ (changed or no change) (3x2 design). Each child will see 3 items of each condition (18 total, randomized) in three blocks, counterbalanced across participants.

We will code children’s looking behavior offline using the recordings, analyzing which side of the screen they looked at in the 3 second time window following the target utterance, in addition to their overt response (point/touch). Then, we will conduct general linear regression analyses to see if looking time and accuracy can be predicted by our experimental conditions and the child’s age. The factual condition serves as a control condition, and we expect children to display high accuracy on these items. We propose three hypotheses for the performance on the conditions of interest (negated, counterfactual):

(1) adult-like for both conditions, suggesting they comprehend both.
(2) poor for both conditions, which could indicate they struggle with non-actual language in general.
(3) adult-like with negation, but not counterfactuals, suggesting they struggle with repressing the actual world, rather than negating an event.

Additionally, we expect that the implicit measure of looking preference might reveal more competence in younger children than the overt behavioral response, providing us with the best possible chance to uncover potential counterfactual comprehension abilities.

Authors

• Maxime Tulling, Université de Montréal (UdeM)
  Presenting Author

• Maya Orey, New York University

• Ailís Cournane, New York University