Impact of Lexical Parameters and Audibility on the Recognition of the Freiburg Monosyllabic Speech Test

Correct word recognition is generally determined by audibility, but lexical parameters also play a role. The focus of this study was to examine both the impact of audibility and lexical parameters on speech recognition of test words of the clinical German Freiburg monosyllabic speech test, and subsequently on the perceptual imbalance of test lists observed in the literature.

For 160 participants with normal hearing that were divided into three groups with different simulated hearing thresholds, monaural speech recognition for the Freiburg monosyllabic speech test was obtained via headphones in quiet at different presentation levels. A software manipulated the original speech material to simulate two different hearing thresholds. All monosyllables were classified according to their frequency of occurrence in contemporary language and the number of lexical neighbors using the Cross-Linguistic Easy-Access Resource for Phonological and Orthographic Neighborhood Density database. Generalized linear mixed-effects regression models were used to evaluate the influences of audibility in terms of the Speech Intelligibility Index and lexical properties of the monosyllables in terms of word frequency (WF) and neighborhood density (ND) on the observed speech recognition per word and per test list, respectively.

Audibility and interactions of audibility with WF and ND correctly predicted identification of the individual monosyllables. Test list recognition was predicted by test list choice, audibility, and ND, as well as by interactions of WF and test list, audibility and ND, ND and test list, and audibility per test list.

Observed differences in speech recognition of the Freiburg monosyllabic speech test, which are well reported in the literature, depend not only on audibility but also on WF, neighborhood density, and test list choice and their interactions. The authors conclude that future creations of speech test material should take these lexical parameters into account.