The quality of transmission of prosody-related acoustic cues through a cochlear implant (CI; an auditory prosthesis that can partially restore hearing in individuals with sensorineural hearing loss) and the linguistic experience of listeners impose challenges on the identification of native and non-native prosody in CI users and vocoder simulations.
Prosody forms an important part of spoken language. It can, for instance, be used to identify linguistically meaningful contrasts in speech or the emotional state of speakers. Prosodic patterns can be expressed using acoustic cues such as variation in fundamental frequency (f0), intensity, and duration, and listeners are able to draw on these cues in perception. Identifying prosodic patterns and interpreting their meaning can be challenging for cochlear implant (CI) users because the electric speech signal transmitted through a CI is highly lacking in fine spectro-temporal detail, hindering their ability to draw on prosody-related acoustic cues – particularly f0 – that are readily available to normal hearing (NH) individuals. An additional challenge can be expected for CI users perceiving prosody in a non-native language, as cues may be used differently or may be less efficiently processed in the non-native language.
We present the results of a meta-analysis and a perception experiment which show where the challenges in the identification of native and non-native prosody mainly lie for CI users and NH listeners of vocoder simulations. The meta-analysis revealed a robust impact of CIs and vocoder simulations on prosody identification in the native language, where electric hearing limits the identification of prosody, mainly due to the poor transmission of f0. Similarly, the perception experiment revealed that prosody perception in the non-native language was less accurate for vocoded stimuli. Moreover, the results showed that listeners transferred the cue-weighting strategies from their native language into the non-native language and that this was not influenced by the perceptibility of available cues in the spectro-temporally degraded signal. Taken together, these studies show that the quality of transmission of prosody-related acoustic cues and the linguistic experience of listeners influence the identification of native and non-native prosody in CI users and vocoder simulations.
