List Equivalency of PRESTO for the Evaluation of Speech Recognition

Abstract

BACKGROUND:

There is a pressing clinical need for the development of ecologically valid and robust assessment measures of speech recognition. Perceptually Robust English Sentence Test Open-set (PRESTO) is a new high-variability sentence recognition test that is sensitive to individual differences and was designed for use with several different clinical populations. PRESTO differs from other sentence recognition tests because the target sentences differ in talker, gender, and regional dialect. Increasing interest in using PRESTO as a clinical test of spoken word recognition dictates the need to establish equivalence across test lists. PURPOSE:

The purpose of this study was to establish list equivalency of PRESTO for clinical use. RESEARCH DESIGN:

PRESTO sentence lists were presented to three groups of normal-hearing listeners in noise (multitalker babble [MTB] at 0 dB signal-to-noise ratio) or under eight-channel cochlear implant simulation (CI-Sim). STUDY SAMPLE:

Ninety-one young native speakers of English who were undergraduate students from the Indiana University community participated in this study. DATA COLLECTION AND ANALYSIS:

Participants completed a sentence recognition task using different PRESTO sentence lists. They listened to sentences presented over headphones and typed in the words they heard on a computer. Keyword scoring was completed offline. Equivalency for sentence lists was determined based on the list intelligibility (mean keyword accuracy for each list compared with all other lists) and listener consistency (the relation between mean keyword accuracy on each list for each listener). RESULTS:

Based on measures of list equivalency and listener consistency, ten PRESTO lists were found to be equivalent in the MTB condition, nine lists were equivalent in the CI-Sim condition, and six PRESTO lists were equivalent in both conditions. CONCLUSIONS:

PRESTO is a valuable addition to the clinical toolbox for assessing sentence recognition across different populations. Because the test condition influenced the overall intelligibility of lists, researchers and clinicians should take the presentation conditions into consideration when selecting the best PRESTO lists for their research or clinical protocols.