Intelligibility testing: issues in closed versus open format scoring
J Vigouroux, Nick Miller · 2007
This study compared diagnostic intelligibility test (DIT) scores using open versus closed format scoring. Audio recordings of twenty-seven dysarthric speakers with Parkinson's disease and twenty-four control speakers, producing single words and sentences were played to sixty-one naive listeners, who either orthographically transcribed what they heard (open format), or selected the target from a choice containing the target and 11 foils (closed format). Listeners also rated speakers ' perceived naturalness from a sample of spontaneous speech. Naturalness scores were compared to the open and closed intelligibility scores. Closed format intelligibility scores were significantly higher than open format scores. The correlation between scores from respective formats was significant, but closer analyses revealed intra-individual speaker variability. Intelligibility scores from the open format DIT correlated more closely with judgements of speakers ' speech naturalness than did closed format scores. Findings suggest open and closed format DITs should be viewed as qualitatively different assessment methods, which are not necessarily clinically interchangeable. 1.