Unmasking the truth about clear conversation
Posted on 12 December 2008
Dr Dominic Watt and Dr Carmen Llamas, working with Philip Harrison of forensic speech and audio analysis specialists JP French Associates, are examining whether garments such as surgical masks, balaclavas or the niqab, the veil worn by some Muslim women, distort the speech sounds made by people wearing them. Their interest was prompted by the furore that followed comments by Jack Straw, then Leader of the Commons and now Justice Secretary, in 2006 revealing that he asked women wearing veils who visited his surgery to lift them in order to make conversation easier.
While there is a substantial body of experimental evidence regarding the link between facial movements and speech comprehension, little has been published about the impact face coverings have on how people hear what the wearer is saying.
The ramifications of any effect on communication at a day-to-day level are clear. A doctor wearing a surgical mask, for example, must be able to instruct other members of medical staff clearly in an emergency situation. However, there are also broader implications. The way an earwitness report of a crime scene is interpreted by the police or the courts may need to change if there is evidence that a face covering can degrade the speech signal.
In a preliminary study, the group conducted two separate tests. In the first, a panel of listeners was asked to listen through headphones to the speech of talkers they watched on a computer screen, and to identify target words. The talkers wore a variety of face coverings as well as reading the sentences with their faces uncovered, while keeping their facial expressions as neutral as possible. The listeners also heard the soundtracks of the video recordings, but with the picture removed so as to isolate any beneficial effect of being able to see the talker.
The second test examined the sound absorption qualities of each fabric. Samples of a wide range of materials which might be used to cover the mouth in medically or forensically relevant circumstances (for example a cotton handkerchief, a woollen scarf and even a nylon stocking like those stereotypically worn as a disguise by armed robbers) were placed between a loudspeaker and a microphone in a sound-treated room. Test signals were then played so that the absorption spectrum of each fabric could be measured.
Initial findings showed that the listeners experienced surprisingly little difficulty in correctly perceiving target words, whatever the talkers were wearing over their faces. Given that the results of the second test showed negligible differences in sound absorption between the various fabrics, the group suggests that claims about impaired speech intelligibility made in the wake of Straw's remarks probably derive more from listeners being unable to pick up facial clues when faces are covered than from interference with speech production or hearing per se.
The research team are now planning to investigate these issues further by conducting a broader study involving a wider variety of test conditions and participants, including individuals who habitually wear face coverings for professional, religious or recreational reasons.
About the researchers
Dr Carmen Llamas
Dr Carmen Llamas is a Lecturer in the Department of Language and Linguistic Science
Contact
Email: cl558@york.ac.uk
Dr Dominic Watt
Dr Dominic Watt is a Lecturer in the Department of Language and Linguistic Science
Contact
Email: dw539@york.ac.uk
Philip Harrison
- Philip Harrison, JP French Associates, 86 The Mount, York pth@jpfrench.com
Further information
- Studying in the Department of Language and Linguistic Science