Perception of the prosodic features of newscaster speech

It is frequently posited that the speech of groups with particular communicative goals constitutes cohesive prosodic genres. Here we investigate American English newscaster speech to discover whether listeners can distinguish it from non-newscaster speech based on prosody alone, and, if so, which features they use to do so and which are ignored. In a perception experiment, participants classified low-pass filtered audio clips as “newscaster” or “everyday” speech. We find that listeners were able to identify newscasters at a rate better than chance. They correctly utilized newscasters’ use of lower pitch, more intonational breaks, and little non-modal phonation, but overlooked the range of pitch accents that characterize newscaster speech. Whether this results from a coherent Newscaster genre or an interacting collection of indexical traits, our results suggest that listeners’ competence about speech registers includes sub-phonemic information about intonation, impacting the kind of data linguists must attend to when designing intonational models.