After reading text listen to sound contours.
A speaker can successfully use the same structure to get a range of different messages across to a hearer. Take a simple subject-negative predicate sentence: not can be understood to deny the predicate or the whole sentence; also, not may seem to zoom in on any smaller part of the subject or the predicate, resulting in a positive implication. These differences and others systematically come with prosodic form.
‘Predicate negation’ versus ‘sentence negation’ is not a matter of ‘scope of negation’: what happens is that not reinforces a difference that comes with the subject noun phrase whether the sentence is positive or negative. The speaker can use this noun phrase either to refer to something that is there, in that case, he drops his pitch to base level between the top and the end of its final syllable. A following negative predicate is understood as ‘predicate negation’. Or he can use the noun phrase not to refer, in that case, he keeps its end above base level. Combined with a negative predicate, the information is ‘sentence negation’.
The hypotheses predict understanding of ‘predicate negation’ or ‘sentence negation’ for all utterances of these negative sentences, only depending on this minimal pair in pitch and independent of distribution and realization of pitch accent and of overall intonation contours. Hypotheses and predictions are first discussed in Koene (1984); for the latest version of the semantic theory specifying what exactly these prosodic forms contribute, see Koene (2007). Notation of prosodic form is given on two levels:
- the theoretical level of pitch forms recognized by the hearer as informative; here, only presence of pitch accent and the minimal pair proposed need to be represented; they form (part of) the skeleton beneath the actual intonation contour. The hypotheses apply to Dutch as well as English,
- the overall stylized contour, drawn as I hear it, roughly based on the notation developed in ‘Dutch School’ research (see references in the books elsewhere on this homepage). These are Dutch contours.
Listen to een zuster van Jan is niet geïnteresseerd, translating a sister of John’s is not interested. Contours (a) thru (f) all, arbitrarily, have one pitch accent on zuster (sister) and one pitch accent on geïnteresseerd (interested); the crucial point is the final syllable of the noun phrase: in contours (a), (c) and (e), the contour meets the low declination line at least at one point between its top and its end (followed by a very late final rise in contour (a) and (c)), in contour (b), (d) and (f), the contour remains above base level at that point. Contour (g) thru (h) all have the same very common ‘hat’ contour as contour (b) has, but they differ in the location of the two pitch accents: in all cases, it is when the ‘hat’ is pushed to either side of the noun phrase-predicate border that the understanding reverses from ‘sentence negation’ to ‘predicate negation’.
Not zooming in always is a special case of either ‘predicate negation’, as in case (n), or ‘sentence negation’, as in cases (l) and (m); they fit in with the hypotheses. The characteristic ‘fall-rise’ intonation has to be broken down into separate informative forms: peaking pitch on the spot singled out, an overall absence of pitch accent with the possible exception of the spot singled out, in case the spot singled out is within the noun phrase, its final syllable kept above base level, and a final rise. These forms all make their own contribution to the understanding. The analysis is first made in Koene (1994); for the latest version of the semantic theory specifying exactly how they result in the positive implication, see Koene (2007).