Skip to main content

Predicting the Timing of Talking

How do speaker and listener factors boost effective conversational timing for communicative goals?

Project focus

This project investigates:

  • how listeners dynamically interpret timing information in speech
  • how this ability may change due to stroke-acquired aphasia
  • what this tells us about the cognitive and neuropsychological basis of the timing of conversational interactions

How timing prediction works

In this project, we explore how listeners predict the timing of speech. These predictions help us to identify key features of spoken interactions, such as the starts of words and the ends of conversational turns. Crucially, listeners seem to use the rate of speech they have already heard to anticipate how long upcoming sounds should be. They then use deviations from those expectations to interpret spoken language more effectively.

What this research will do

The research will test competing neuropsychological theories about timing prediction, for example through testing whether artificially regular timing or the irregular flow of natural speech better supports listeners’ prediction.

To investigate this, we use a novel listening task in which participants detect short nonsense sequences of sounds embedded in longer utterances. This allows us to examine how timing prediction is shaped by factors such as:

  • speech regularity
  • background noise
  • meaningful content
  • familiarity with voices and accents
  • visual cues from speakers

By comparing people with typical language abilities and people with aphasia after stroke, the project aims to:

  • improve understanding of speech perception difficulties 
  • inform the design of speech technologies and communication support systems

Find out more

Predicting the timing of talking: How do speaker and listener factors boost effective conversational timing for communicative goals?

Principal Investigator Laurence White, Co-Investigator Julie Morris.

Acknowledgements

Funded by the Economic and Social Research Council (ESRC).