Research

The primary focus of the SVO Lab is investigating auditory-perceptual outcomes for speech and voice, with a particular emphasis on speech that is not well-described using measures of intelligibility. Outcome measures are used to characterize baseline performance and to document change over time. In some cases, a speaker’s intelligibility fails to reflect the amount of effort a listener has to invest in order to communicate with her. It may be less relevant in those cases to verify that a listener has understood the speaker than it is to identify changes in how hard the listener has to work. Measures of perceived listener effort (PLE) capture the listener’s perspective on communicative interactions with an individual with a speech or voice disorder. The SVO Lab collaborates with Dr. Tanya Eadie of the Vocal Function Lab at the University of Washington on studies of perceptual characteristics of disordered voice and speech.

One way to improve both speech intelligibility and PLE for most listeners is to address speech naturalness and prosody. Individuals who have had a total laryngectomy (removal of the larynx), those who speak English non-natively, and those with motor speech or voice disorders are of particular interest to the SVO Lab. These populations present listening challenges related to intonation and speech naturalness.

Electrolaryngeal speech is notably unnatural sounding and relatively monotonous because of its atypical sound source (the electrolarynx).¬†Alternative “voice” sources lack the quality and flexibility to produce the rich, complex, periodic tones associated with the human voice or to produce intonation patterns corresponding with a speaker’s linguistic intentions. The SVO Lab collaborates with Dr. James Heaton on improving prosody in electrolaryngeal speech at the Center for Laryngeal Surgery & Voice Rehabilitation at the Massachusetts General Hospital.

Non-native speakers (NNS) of English often have a different problem with intonation – that is, they are likely to accentuate syllables or words in a way that is distracting to a native listener, which increases the complexity of listening. Frequently, depending on many variables such as native language, NNSs have concomitant problems pronouncing speech sounds in a native-like way.

The SVO Lab uses quantitative and qualitative methods to investigate these and other variables related to PLE with the goals of creating valid, reliable measures that reveal the effects of 1) disordered speech and voice on listeners, and 2) the reflection of listener reactions back onto speakers.