Measuring consonant identification in nonsense syllables, words, and sentences

Abstract

Sensorineural hearing loss (SNHL) produces deficits in speech comprehension in noise that primarily are due to impairments in identifying consonants. Here, we describe the California Syllable Test (CaST) that quantifies the identification of common American English consonants. In experiment I, 16 young subjects with normal hearing identified 720 consonant-vowel-consonant (CVC) syllables in three test sessions. Consonants were identified slightly more accurately in words than nonsense syllables, and small interactions were found between the processing of initial and final consonants. Consonant-identification performance correlated strongly with sentence reception thresholds (SeRTs) measured with both the Hearing in Noise Test and QuickSIN (Etymotic Research; Elk Grove Village, Illinois). At SeRTs, subjects with normal hearing could identify 32.5% of consonants in isolated CVCs. In experiment II, a patient with moderate SNHL showed large elevations in consonant-identification thresholds and smaller elevations in SeRTs. At SeRT levels, the patient could identify only 12.5% of consonants in isolated CVCs, indicating that sentence comprehension relied disproportionately on vowel cues and semantic constraints. Consonant-profile analysis revealed disproportional impairments in identifying consonants dependent on high-frequency acoustic cues. Consonant confusion analysis revealed a reorganization of consonant per ception. The CaST is a promising tool for evaluating consonant-specific processing deficits in patients with hearing impairment.