A Review of A Grammar of Speech 37
tone unit as the beginning of a new paratone and instead marks it as the
fi nal tone unit of the second paratone. The boundary between the second
and the third paratone is signalled by the combination of low pitch on
breakfast and the immediately following high pitch onset on cereal.
Working within Discourse Intonation Barr (1990: 11) and Pickering
(2004: 24) recognize what they refer to as the sequence chain which consists
of one or more pitch sequences bounded by an initial high-key onset
and completed by a low termination which is itself immediately followed
by a high key. The second phonological paratone identifi ed by Tench is
a sequence chain which contains two pitch sequences. It seems that para-
tones are more closely related to sequence chains than they are to pitch
sequences. However, Tench (1990: 277) states that pitch sequences usually
begin with high key and, thus, tend to confl ate with sequence chains.
Other scholars agree that high initial pitch signals the beginning of a new
paratone while low pitch, at least partly, signals the closure of the paratone.
Brown, Currie and Kenworthy (1980: 26) argue that the delimiting criteria
for paratones are pause followed by an initial high pitch reset, though they
also recognize that speakers signal the completion of an existing paratone
by ‘dropping low in their pitch range’ (ibid. 25). Thompson (2003: 9)
recognizes low termination followed by a high onset as the criterial features
for identifying phonological paragraphs. Cutler and Pearson (1986)
describe a carefully designed experiment where ten speakers read two
versions of fi ve dialogues which differed only in the order of the sentences.
In one version, a particular sentence was turn medial; in the other turn
fi nal. These sentences were then judged in isolation as either turn medial
or turn fi nal. They found that the intonation feature which correlated
most highly with the perception of fi nality was a tonic syllable pitched
signifi cantly lower than the previous syllable,^16 while turn medial sentences
correlated most closely with a tonic syllable pitched higher than the previous
syllable (ibid. 152).
Brown, Currie and Kenworthy (1980: 136) argue that speakers mark new
topics or subtopics by raising initial stressed peaks and that if speakers wish
to signal that their contribution is a continuation of an existing topic they
will produce an initial pitch which is low in their pitch range (also Brown
1990: 92, Couper-Kuhlen 1996: 398, Cruttenden 1997: 123, and Gussenhoven
2004: 71). Brown et al. note that their analysis ‘bears a close resemblance to
that of Brazil who has examined the role of “key” in discourse’. However,
they fi nd evidence for only two pitch levels: high and low (1980: 137): a view
implicitly supported by Gussenhoven (2004: 114–15) and by those working
within the autosegmental tradition.