Intonation and Syntax in Spoken Language Systems

Phrasal intonation is notorious for a tendency to perceptually segment the word-string of a spoken utterance into groups which may violate orthodox syntactic notions of constituency. For example, the normal prosody for the answer (b) to the following question (a) imposes the intonational constituency indicated by the brackets (stress, marked in this case by raised pitch, is indicated by capit als) : (1) a. I know that brassicas are a good source of minerals, but what are LEGumes a good source of? b. (LEGumes are a good source of) VITamins . Such a grouping cuts across the traditional syntactic structure of the sentence. The presence of two apparently uncoupled levels of structure in natural language grammar appears to complicate the path from speech to interpretation unreasonably, and to thereby threaten a number of computational applications in speech recognition and and speech synthesis.