Potent prosody: Comparing the effects of distal prosody, proximal prosody, and semantic context on word segmentation

Recent work shows that word segmentation is influenced by distal prosodic characteristics of the input several syllablesfrom the segmentation point (Dilley & McAuley, 2008). Here, participants heard eight- syllablesequences with a lexically ambiguous four- syllableending (e.g., crisis turnip vs. cry sister nip). The prosodic characteristics of the initial five syllableswere resynthesized in a manner predicted to favor parsing of the final syllablesas either a monosyllabic or a disyllabic word; the acoustic characteristics of the final three syllableswere held constant. Experiments 1a–c replicated earlier results showing that utterance-initial prosodyinfluences segmentation utterance-finally, even when lexical content is removed through low-pass filtering, and even when an on-line cross-modal paradigm is used. Experiments 2 and 3 pitted distal prosodyagainst, respectively, distal semantic context and prosodic attributes of the test words themselves. Although these factors jointly affected which words participants heard, distal prosodyremained an extremely robust segmentation cue. These findings suggest that distal prosodyis a powerful factor for consideration in models of word segmentation and lexical access.