Does Speech Rate Influence Children’s Conversational Responses? A comparison of Speech Rate Manipulation Techniques

Conversation is central for social interaction, but places substantial cognitive demands on speech processing and response planning, particularly when speech is fast. These demands may be especially challenging for vulnerable individuals, such as children whose language and cognitive abilities are still developing, and for neurodiverse individuals with communication difficulties. Despite this, there is little direct evidence on how speech rate affects children’s conversational responses.

Across two pre-registered studies, we examined whether a conversation partner’s speech rate influences children’s ability to acknowledge and/or elaborate on a topic (“topic-supporting” responses). Using a novel semi-naturalistic paradigm, children conversed with a researcher via Zoom, while pre-recorded utterances (“probes”) spoken in the researcher’s voice were interspersed at topic-appropriate moments. For example, the researcher may lead the conversation towards days out, at which point they would play the probe, “We like to go to the zoo.”

In Study 1, probes were digitally manipulated to be either Fast (80% of original speech rate; 4.83 syllables per second) or Slow (140%; 2.68 syllables per second). Participants were autistic and well-matched neurotypical children aged 6½–12 years (N = 100). Contrary to our prediction, children showed a trend toward providing more topic-supporting responses in the Fast (M = 76%) than the Slow condition (M = 72%; p = .07). However, the Slow condition involved greater digital manipulation, which may have reduced engagement. Additionally, the Fast condition probes overlapped with typical adult–adult speech rates.

Study 2 will address these limitations by using naturally produced speech. Fast probes reflect the upper end of the researcher’s natural speaking rate (6.18 syllables per second); Slow probes (3.42 syllables per second) are matched for prosody. To increase variability in responses, participants are younger children aged 6–7 years (target N = 36 neurotypicals). If the findings from Study 1 reflected reduced engagement with unnatural Slow speech, we predict more topic-supporting responses in the Slow condition. Alternatively, if faster speech is more engaging, children may produce more topic-supporting responses in the Fast condition. Together, these findings will inform the development of a mechanistic account of children’s conversational topic maintenance.