Acoustic variability underlying pathological voice quality

Pathological voice quality poses unresolved challenges for voice science. The presence of physical pathology does not necessarily imply poor voice quality, nor does a large deviation from modal phonation necessarily imply pathology. Creak, breathiness, and chaotic vibrations occur in both normal and pathological voice, and most speakers can produce a wide range of qualities, including those often labeled “pathological.” Why do some voices signal a disorder, rather than being perceived as part of a quality continuum that also includes normal voices? To address this question, we derived a psychoacoustic voice space for speakers with diagnosed vocal pathologies and compared it to the corresponding space for speakers without a diagnosis. Acoustic variables were measured from recordings of sentences read by speakers with and without diagnosed pathology (e.g., mass lesions, paralyses, functional disorders, neurogenic disorders) and analyzed using principal component analysis. We hypothesized that vocal pathology disrupts the transmission of biologically important information related to a speaker’s state of arousal and reproductive fitness. This information is available from the voices of speakers without pathology, and its loss in cases of pathology hypothetically undermines communicative functions that may represent a primary evolutionary basis for phonation.