Prosody

Language and speech. Speech spontaneity, articulation, vocabulary, and prosody should be noted. Pedantic speech may be a symptom of Asperger s disorder or of a nonverbal learning disorder. Diminished speech may be the result of depression or autism. Excessive speech may be the result of anxiety, ADHD, or hypo-mania. Unusual speech may be an early sign of neu-rodevelopmental vulnerability to psychosis. Slurred speech may be evidence of a neurologic abnormality or a medication side effect. 

Hoekert M, Kahn RS, Pijnenborg M, Aleman A. 2007. Impaired recognition and expression of emotional prosody in schizophrenia Review and meta-analysis. Schizophr Res 96 135-145. 

One of the central debates in the field of prosodic linguistics is j ust how far this scale extends. Crudely speaking, the argiunent coneems whether prosody is mainly a universal, non-arbitrary... 

The second aspect of prosody is called augmentative prosody. The basic purpose of this is to augment the verbal component. Taking a simple example first, we see that many sentences are highly confusable with respect to syntactic structure. In sentences such as... 

Affective prosody is when a speaker uses prosody to express emotion, speech act or other information which is conveyed sentence by sentence... 

Augmentative prosody is used to disambiguate and reinforce the verbal component. 

Both t5T3es of prosody are optional, in that speakers can choose to use no prosody if they wish and just communicate via the verbal component. Writing in fact uses just this component. 

Simply put, a speaker has a open choice of how much prosody or how much verbal content to put in a message they will balance these against each other, to create a final signal that conveys the message they wish in a way that is balanced between efficiency and effectiveness. A writer on the other hand has only a very limited ability to use prosody and so has to rely on the verbal component more. Because of this, he is likely to beef-up the verbal component, so that it is less ambiguous and more explicit than the spoken equivalent. In addition, the writer will verbalise any emotion if that is required (that is, e q)licitly explain the emotion by using words). 

Prosodic information can not in general be decoded from the text, and so everything is spoken with neutral prosody. 

Prosody from the text Following Ifom the assumptions of the complete prosody model, we find that if every utterance requires a detailed prosodic specification, then this must somehow be generated. A common assumption is that the text does in fact contain enough information to determine prosody, and so many TTS systems have modules which try and predict prosodic representations directly from the text, often with the assumption that this is an analysis process with a right and wrong answer. 

In many cases, no prosody is required, but in others a more sophisticate approach is to attempt to generate prosody that is appropriate for the message. This can not however by decoded from the text as that information was never encoded there in the first place. 

Finally, Chapter 6 describes how to predict prosody information from an often impoverished text input. In many ways, this subject shares similarities with text analysis. There is an important difference however in that while we can view text analysis as a decoding problem with a clear right and wrong, prosody prediction has no strict right and wrong as we are attempting to determining prosody from an underspecified input. 

---