Second International Conference on Spoken Language Processing (ICSLP'92)
Banff, Alberta, Canada
Accents mark focal, i.e. important information. Published production and perception data show that accents may differ in "accent strength". In order to find out whether this information might be used by listeners to determine which information is more focal, an experiment was set up to find out whether listeners can make judgments of relative accent strength equally well for adjacent accents and non-adjacent accents: if they can do so only for adjacent accents, the usefulness of variation in accent strength as a cue to variation in communicative weight would clearly be limited. Judgments about relative accent strength for adjacent and non-adjacent accents were compared both for a real sentence and for reiterant speech. It was found that the task was much easier for the real sentence than for reiterant speech, and that for the real sentence judgments of relative accent strength can be made equally well for adjacent and non-adjacent accents. It is concluded that variation in accent strength might be used to signal differences in communicative weight. In addition, it was concluded that listeners need linguistic structure to support the interpretation of prosodic information.
Bibliographic reference. Terken, Jacques / Hombergh, Karin van den (1992): "Judgments of relative prominence for adjacent and non-adjacent accents", In ICSLP-1992, 735-738.