![](bg2.png)
phase prior to transcription and coding itself. In
contrast to a monolithic approach to transcription,
previously described, the production of intentionally
incomplete transcription and coding deliverables is
referred to here as partial transcription. The
advantages of partial transcription include amongst
other things the ability to encourage empowerment
research (see Cameron et al, 1992) by enabling the
participants themselves to determine what gets
recorded. This facilitates trust and improves the
research relationship between analysts and members
of organisations. An obvious difficulty with partial
transcription is that, depending on the kind of
analysis being undertaken, omitting sections of a
transcript will disrupt a number of spoken language
resources- some of these may be crucial to the
analysis being conducted. Fortunately, functional
linguistic theories exist which can give considerable
insight into which specific language resources will
be affected. We use Systemic Functional Linguistics
(SFL) a semiotic model of language (Halliday,
1985) because it has a concept referred to as texture
that encompasses and defines all the text-forming
resources that may be used in a transcript or any
other text. For example, texture has also been
applied to hypertext development and modeling
(Clarke, 1997). Any texts including all transcripts
must possess texture in order to function as a
semantic unit, as well as being relevant or
appropriate to a given social setting or occasion.
Whether knowingly or not, speakers and writers use
their experience of texture resources when
constructing texts, while listeners and readers use
their experience of these resources when interpreting
texts. Texts are generally read from start to finish
and so many of these resources flow through a text
in chains. This is an attribute of language referred to
as sequential implicativeness (Schegloff and Sachs,
1973). For example a text might start with the
sentence “Rod is in the Red Theatre” and if the next
sentence was “He is giving a seminar” we might
reasonably conclude that the ‘He’ is Rod. This is an
effect of sequential implicativeness in the so-called
Reference System (see below).
There are several models of texture within SFL.
The texture model we use (Martins’ 1992, 381
adaptation of Halliday and Hasan’s 1976 model)
recognises three major groups of text-forming
resources- Intrasentential Resources, Intersentential
Resources, and Coherence. Within each major
group, there are a number of sub-categories of text-
forming resources each having an associated
analysis method and some also have graphical
methods:
– intra-sentential resources (Martin 1992, 381) or
structural resources (Halliday, 1985)- involve
systems of THEME and INFORMATION and
spoken language specific systems involved in
Conversation Structure. All texts consist of sets
of clauses each of which can be divided into a
theme and a rheme. Listeners or readers rely
upon thematic progression, the specific pattern
of themes, to predict how the text should unfold.
Texts must also provide and ‘manage’
information. Listeners or readers come to rely
upon patterns of information units to build and
accumulate new meanings from those that have
already been given. Conversation Structure
involves speech functions the characteristic set
of moves enacted by participants involving
initiations (offers, commands, statements,
questions) or responses, as well as sequences of
speech functions that form jointly negotiated
patterns called exchange structure.
– intersentential text-forming resources of
Cohesion- describe how clauses within any text
are interrelated giving the appearance of a unity
thereby assisting listeners and readers in
understanding the meanings being negotiated.
There are a number of types of cohesion,
including lexical cohesion which describes how
lexical items (words) and sequences of events
are used to consistently relate a text to a topic,
reference which describes how participants are
introduced and subsequently managed, ellipsis
which establishes reference relationships through
the omission of otherwise repetitive lexical
items, substitution which employs alternate lexis
for original lexical items, and conjunction which
refers to the logical relations between parts of a
text.
– text forming resources of Coherence- which
describes how clauses in texts relate to the
contexts in which they occur. All texts must be
relevant to the immediate situational context,
referred to as situational coherence, while also
conforming to an appropriate genre, referred to
as generic coherence.
It is relatively easy to understand in principle
what happens when we partially transcribe.
Effectively we run the risk of disrupting sequential
implicativeness of many of these text-forming
resources. Partial transcripts may loose coherence,
and will most certainly have disrupted thematic and
informational intra-sentential resources. Perhaps the
group of text forming resources most disrupted will
be cohesion as omitting clauses make it more
difficult for readers to understand the transcript as a
unity. We could easily produce an unintelligible
partial transcript if we removed too much of it.
While texture theory can tell us which language
resources will be affected when we adopt partial
transcription, it can only provide part of the picture.
The theory of texture cannot tell us how significant
EFFECTIVE XML REPRESENTATION FOR SPOKEN LANGUAGE IN ORGANISATIONS
487