Tool for Intelligibility Assessment
Overview
Reduced intelligibility is a core functional deficit of motor speech disorders and a key indicator of disorder severity. Clinicians and researchers must ensure that their approach to measuring intelligibility will yield an accurate estimation of a speaker’s actual intelligibility when communicating outside of the clinic or lab. These professionals are faced with several decisions about their approach—which estimation method to use, which listeners to recruit, and what speech sample to present.
The Tool for Intelligibility Assessment will guide clinicians and researchers in making decisions about their approach to measuring intelligibility to maximize the efficiency of the process without undermining the accuracy of the measure.
Application
This tool was built using data from our work on measuring the intelligibility in people with Parkinson’s disease (Dahl et al., 2024). We evaluated different estimation methods, listener types, and a standardized tool (Speech Intelligibility Test [SIT]; Yorkston et al., 1996), all of which are used in clinical practice and research. This tool may thus only apply to intelligibility assessments that incorporate a combination of these elements.
- Population: The speaker sample included people with Parkinson’s disease. But because intelligibility is an outcome independent of etiology, we believe this tool may be cautiously applied to other populations
- Methods:
- Orthographic transcription
Listeners typed what they heard the speaker say, and this transcription was compared to the speaker’s intended message. The number of words the listener correctly understood was divided by the total number of words spoken to calculate how intelligible the speaker was (e.g., 6 out of 10 words correct = 60% intelligible). - Visual analog scale
Listeners placed a mark on a 100-mm line, with one end corresponding to 0% intelligibility and the other end 100% intelligibility. The distance of the mark from the 0% end of the line was measured to calculate how intelligible the listener thought the speaker was (e.g., a mark at 60 mm = 60% intelligibility).
- Orthographic transcription
- Listeners:
- Speech-language pathologists (SLPs) with experience assessing intelligibility
- Inexperienced listeners
- Speech sample: SIT sentences
Instructions
- Select an assessment method
- Select a listener type
- Set an accuracy threshold. We recommend a minimum of 80% accuracy, consistent with the convention for minimum statistical power
- Find Results reveals how many listeners and SIT sentences you need to accurately measure intelligibility using the selected parameters
Citation
Dahl, K.L., Balz, M.A., Díaz Cádiz, M., & Stepp, C.E. (2024). How to efficiently measure the intelligibility of people with Parkinson’s disease. American Journal of Speech-Language Pathology, ePub ahead of print.