204 A Grammar of Spoken English Discourse
Two language features not present in Brazil’s original data, ellipsis and
dysfl uency, were explored and possible codings proposed. Brazil’s require-
ment that the grammar contain initial N V elements was shown to be an
idealized abstraction which speakers in situated discourse may not follow.
Speakers are free to produce chains which contain the minimum number
of elements appropriate to their communicative needs. Coding the situa-
tionally or textually mandated elements which were realized by zero with
the Ø symbol rendered the workings of the chains more semantically trans-
parent and ensured that utterances which met communicative needs but
breached Brazil’s strict syntactic coding could be included within the
grammar.
Brazil (1995) included the ‘.. .’ coding to notate points in the discourse
where speakers abandoned increments and this coding was found to accur-
ately describe dysfl uency which resulted in abandoned increments. How-
ever, many instances of dysfl uency located in the data, resulted not in the
abandoning of increments but rather in the insertion of lexical elements of
the same class membership as the previous lexical element. Within the
chain the second lexical element did not result in the creation of a new
intermediate state. Instead it cancelled the expectations created by the pre-
vious element and then re-imposed a new expectation. In order to make
the workings of the chains more semantically transparent and highlight the
fact that the replaced elements failed to result in an expectation which led
to the creation of target state, they were bracketed.
Brazil argues that chains are composed of word-like elements which move
from an initial state through optional intermediate state(s) to a target state.
Such a view appears opposed to much recent linguistic theory which has
argued that language is at least partly formed out of prefabricated chunks.
Evidence from the literature was produced in Chapter 4 which suggested
that idioms; phrasal verbs; verbs in phase; the future use of the to be-ing pat-
tern and compound nouns are more transparently coded as chunks rather
than decomposed into orthographic words. Chapter 6 confi rmed that
tonality selections indicated that speakers treated such elements as chunks.
The coding of such elements as PHR-V, PHR and N rendered the workings
of the grammar more transparent, e.g. example (12) look at in Chapter 5
which in Dc’s reading was coded as a v’p sequence rather than as a v’phr.
8.2 Limitations in the Research/Unresolved Issues
All research is constrained by limited time and space, a lack of open-ended
resources, and by the data employed, and so a choice must be made about