38 A Grammar of Spoken English Discourse
There is widespread support in the literature for the existence of declination
both in English and in many other languages. The term ‘declination’ was
coined by Cohen and ‘t Hart (1967: 184) to describe the downward trend of
pitch observable across many utterances in Dutch. Cohen and ‘t Hart, and
others working within the IPO^17 tradition, regard declination as no more
than a phonetic (i.e. not communicatively signifi cant) phenomenon, e.g.
(‘t Hart 1998: 100) though they recognize that speakers can exploit declina-
tion for linguistic purposes by resetting a natural declination by producing
a high onset (‘t Hart and Collier 1990).
Gårding (1998: 121), in the case of Swedish, argues that the slope of
declination is dependent solely on the length of the sentence. I take this
to mean that shorter sentences have steeper slopes than longer ones
irrespective of individual speaker communicative purposes. The belief
that declination is a gradual, and to a large extent regular, tapering off of
pitch has led numerous scholars to produce mathematical models of the
pitch contours of utterances (e.g. Fujisaki 1983 for Japanese; Gårding 1983
for Swedish; Gårding 1987 for Chinese). Such views are dubbed ‘overlay’
models by Ladd (1996: 24) who notes a number of potential problems
with them including the following. First, none of the proposed models has
produced a quantitative defi nition of the components they presuppose, e.g.
there is as yet no quantitative characterization of the slope of declination.
Second, the lack of precise defi nition of a pitch contour as a mathematical
function renders them incapable of prediction (ibid. 27). Third, all overlay
models are grounded upon the unproven assumption that intonational
meaning can be related directly to acoustic correlates presented as a pitch
contour without any mediating phonological categories (ibid. 20–4).
A note of caution in placing sole reliance on acoustic measurements is also
required. Pitch is usually taken to refer to how F0 is perceived by hearers
but other factors such as vowel quality; the nature of the surrounding
consonants (i.e. voiced or voiceless); loudness and duration also infl uence
how people perceive pitch (Chun 2002: 5). Thus, changes in absolute F0
values may not always be heard as changes in pitch level.
In an investigation of the intonation of Greek, Botinis (1998: 294) argues
that while the fi nal juncture is lower than the initial one, pitch does not
decline gradually across an utterance. Instead, he proposes, that reported
instances of declination in a number of languages may be artefacts arising
out of the relatively simple utterances utilized in laboratory experiments.
Cruttenden (1997: 121) reports that acoustic measurements of pitch peaks
in conversational speech have found little evidence of declination. Tench
(1996: 28) notes that the recognition of phonological paragraphs is most