Acoustic variability of examiners’ productions in language assessment

Abstract

Speech-Language Pathologists and other professionals administer assessments to determine whether children have communication disorders. Although these assessments specify procedures to ensure comparable administration across examiners, there are still individual factors that are unaccounted for. A commonly used measure in assessment is sentence repetition, in which a child repeats a live-voice production of a sentence. Sentence repetition tasks necessarily reflect the unique prosody used by experimenters, and in particular their rate of speech and f0 patterns. This study investigates examiners’ rate and f0 variation as they administer a sentence repetition task in order to quantify variability within and between the examiners. To do this, we analyzed recordings of 25 different speakers administering the Sentence Repetition subtest of the Clinical Evaluation of Language Fundamentals (CELF) to children aged 3 through 10 years old. For each production of a sentence, we measured the mean pitch, pitch range, and speaking rate. The results indicate the extent to which each examiner varied these prosodic features based on the assessment item and the age of the child undergoing the assessment. This variability is compared across examiners to capture a possible point of inconsistency that may affect children’s performance on these assessment tasks.