Poster #17 - The relationship between children’s metacognitive judgments of knowledge and verbal disfluency

• In Event: Poster Session 12

Abstract

When we [uhh] have everyday conversations, our speech is [um] littered with [like] spontaneous pauses and interjections known as “verbal disfluencies”. In adults, verbal disfluencies are associated with a speaker’s certainty or knowledge level in both speech production and speech comprehension. That is, adults rate both their own and others’ confidence lower when they produce more verbal disfluency (Brennan & Williams, 1995; Smith & Clark, 1993). Surprisingly few studies have explored whether and when children’s verbal disfluency correlates with their own internal sense of confidence (Hübscher et al., 2019; Krahmer & Swerts, 2005; Visser et al., 2014). Given that young children struggle with explicit ratings of their own confidence, and show a bias toward overconfidence in a range of tasks (e.g., Finn & Metcalfe, 2014; Lipko et al., 2009; Schneider, 1998), these implicit cues may provide researchers a window into children’s metacognitive awareness, while removing the task demands of explicit reports. This study examines the association between verbal disfluency and confidence in 5–8-year-olds’ (N=60) naturally produced speech. Children answered fact-based questions about animals (e.g. “what are baby cats called?”) and performed numerical comparisons (i.e., they are asked “which side has more dots?” in non-symbolic dot displays). Then, they rated their confidence about these answers in a forced-choice metacognitive judgment paradigm: wherein they are asked which of two answers they feel more certain about. This forced-choice task removes the ability to respond in a biased manner, as children cannot respond “very sure” on every trial (Baer et al., 2021). We examine the association between verbal disfluency and the accuracy of children’s responses, as well as these explicit ratings of metacognitive confidence, showing that even our youngest children reliably produce more verbal disfluencies when they answer incorrectly (M=.36 filler disfluencies for incorrect answers; M=.17 filler disfluencies for correct answers; z=-9.56, p<.001, OR=0.54), and when they feel less confident (M=.32 filler disfluencies not-chosen trials in the forced-choice metacognitive judgment; M=.19 filler disfluencies for chosen trials; t(3614.2)=6.89, p<.001, d=0.22). Non-verbal disfluencies (e.g. “um”, “uh”) decreased with age (r(58)= -.25, p= .05), while lexical disfluencies like hedges (e.g. “I think”, “maybe”) increased (r(58)= .23, p= .08), suggesting that older children more frequently hedge their answers explicitly. Moreover, children’s verbal disfluencies predicted the accuracy of their response over and above their explicit ratings of confidence, suggesting that future work should consider examining verbal disfluency as a measure of children’s metacognition.

Authors

• Eloise West, University of British Columbia (UBC)
  Presenting Author

• Darko Odic, University of British Columbia (UBC)