A Review of A Grammar of Speech 35
It is not necessary in an analysis which focuses solely on the relationship
between pitch sequences to take notice of key/termination levels internal
to the pitch sequence (Brazil 1997: 123). All that needs to be said is that
the initial high-key selection labels the entire pitch sequence (or in this
example increment) as containing information which is contrary to
previously generated expectations while the fi rst low-termination selection
closes the pitch sequence.
To sum up, this section has argued that key and termination realize
communicative value in the domain of increments and that a fully
descriptive grammar must codify the communicative value realized by
key and termination in increments.
2.3.2 Pitch peaks and troughs
Before looking more closely at possible communicative purposes served
by key and termination in increments, it is useful to broaden the picture
and consider what other scholars have written about high and low pitch at
the beginning and the end of utterances. This is done in order to situate
Brazil’s work in the wider literature, and demonstrate that, whilst phrased
differently, Brazil’s work does not necessarily confl ict with the work of
others. However, Tench’s (1990: 274) acknowledgement that perhaps
Brazil’s major contribution to the study of intonation was the development
of key as an independent variable separate from tone should prove cau-
tionary. Many other scholars have not abstracted the communicative value
of relative pitch level from that of tone and so it will not be possible to
match other scholars’ defi nitions exactly with the categories of key and
termination.
Brazil, Coulthard and Johns (1980: 61) argue that the downward drift of
pitch across an utterance is exploited as an organizing position; speakers
mark the boundaries of pitch sequences by producing low termination.
This view is widely supported in the literature. For example, Rost (2002: 34)
states that chunks of speech, known as paratones, which are similar to pitch
sequences, correspond to global planning units of the speaker’s text. Tench
(1996: 24) summarizes the criteria for identifying phonological paragraphs
or paratones. Among the criteria he identifi es are: high pitch on the onset
found in the initial tone unit of the paratone; a gradual lowering of pitch
until the fi nal tone unit is reached; and the depth of fall in the fi nal tone
unit is the lowest in the paratone. The unit described by Tench is clearly
similar to Brazil’s pitch sequence, but as his discussion of an extract from
Brazil, Coulthard and Johns (1980: 145–7) makes clear it is not identical