Children’s overuse of possibility modals in counterfactual production

• In Event: How do children reason about hypothetical possibilities? Evidence from language production

Abstract

Possibility reasoning is central to counterfactual (CF) thought and language, and CF conditionals typically include a modal verb (MV) (e.g., could, would, (1)). CF conditionals differ from non-CF conditionals partly due to the closest possible world constraint (CPWC); CFs select only the closest worlds from the possible worlds generated from the antecedent (Lewis 1973). Experimental studies suggest that children under 6 do not reliably follow the CPWC and may rely on alternative reasoning methods (e.g., Nyhout & Ganea 2019).
We explore modal language production in preschool children. We measure how they express possibility vs. necessity (modal force) in CF conditionals and consider the implications of modal language use for possibility reasoning in CF situations. We find that while 4-year-olds’ modal force use matches adult patterns in non-CF contexts, they use a significantly greater proportion of possibility modals in CF contexts than adults. We argue that this pattern is consistent with children underapplying the CPWC (i.e., considering too many CF possibilities relative to adults).

Methods. We extracted all utterances containing “if” produced by 4-year-olds and adult input speakers from English corpora in CHILDES (23 corpora). We excluded utterances that were not conditionals, or did not have a consequent. We coded conditionals as counterfactual (CF) when the antecedent was false in the real world (given surrounding context), and non-CF otherwise. From 2376 child “if”- utterances, we identified 155 CFs. From 1600 adult “if”- utterances we identified 118 CFs. We additionally selected a random subset of non-CF conditionals (194 child utterances, 238 adult utterances). We coded the force of the MV in the consequent of CF and non-CF conditionals (possibility vs. necessity - Table 2).

Results. 4-year-olds’ CFs are significantly more likely to involve possibility modals compared to adults (χ = 5.63, df = 1, p < 0.05, shaded in Table 1). There is no significant difference for non-CF conditionals (χ = 2.95, df = 1, p = 0.09). Adults prefer necessity for MV force between CF and non-CF conditionals (p < 0.05), while children do not (p = 0.46).

Discussion. Why might children be overusing possibility modals in CF conditionals? By 4, children can maintain parallel open modal possibilities and command the semantics of possibility modals (e.g., Leahy & Carey 2020, Dieuleveut et al 2022). We suggest children’s overuse of possibility modals in CF consequents reflects their possibility reasoning - children tend to maintain more open possibilities, because they apply the CPWC less strictly than adults. This is consistent with experimental results which show that children lack a mature CPWC, potentially rely on simpler forms of reasoning (e.g., McCormack et al 2018, Nyhout & Ganea 2019), and are flexible in changing CF facts unmentioned in the antecedent, unlike adults (Grosu & Cournane 2021). Alternatively, children may not yet appreciate the force of necessity modals and avoid using them. However, this does not explain why only CF conditionals have non-adult proportions of MVs. MV usage therefore reveals that relative to adults, children are potentially not ruling out enough possible worlds.

Authors

• Ioana Grosu, New York University
  Presenting Author

• Ailís Cournane, New York University