Just in case explanations weren’t really clear…
(Lexical) prosody concerns the prominence of (parts of) words. For example the prominence of stressed vs unstressed syllables. (In our discussion, it was pointed out correctly that prosody also concerns other properties, eg (rise or fall of) tone in so-called tone languages as in Mandarin Chinese.) (Lexical) stress is seen as linguistic information because it can distinguish between (the meaning of) different words. For example, “forBEAR” and “FORbear” (or “conCRETE” vs “CONcrete”) share the same sequence of sounds (phonemes) but differ in meaning.
Stressed syllables are on average louder, longer, and higher in pitch. This is a perceptual description, ie it describes prominence in psychological terms. However, this information must have a physical correlate. (How could we transmit information otherwise?) So far we know the physical parameters amplitude, duration, and fundamental frequency (f0) corresponding to the above mentioned properties of prominence, respectively. Although there’s a correlation between perceptual properties and physical parameters, this is not an exact (one-to-one) relation. Human voice processing may not change with physical parameters but with prototypicality (natural-manipulated-computer generated); see the Lattner et al paper.
No comments:
Post a Comment