Beruflich Dokumente
Kultur Dokumente
html
Email: rlawedav@ecel.uwa.edu.au
Introduction
This paper reports, in brief, some of the findings of my PhD research, in which I
looked at differences between coherent and non-coherent tertiary examination essays.
The aim of the study was to discover the properties of text which discriminated
between these groups, looking principally at cataphoric/anaphoric relationships. The
'new' approach to which the title of this paper refers, reflects the finding that
coherence in text was found to relate strongly to the fulfilment or non-fulfilment of
reader expectations raised by cataphoric language functions in the text. This approach
is clearly different from that in which coherence is considered to relate primarily to
anaphoric functions.
In this paper, I shall give a brief background to the study with regard to theory and
methodology, and then describe the cataphoric functions of language found to be
important in discriminating between coherent and non-coherent texts. Similarities and
differences between Native speakers (NS), Non-Native Speakers (NNS) and
International (INT) students will also be indicated.
Background
In this study, it was considered that coherence of text lies in the judgement of the
reader; it is coherent if the reader can readily comprehend its meaning. Like others
before me, I sought to define the forms and functions of language in the text which
could contribute to those judgements. Johns (1986) distinguished two approaches to
the study of coherence;
The theoretical bases for the study were those of cognitive Schema theory (Carrell,
1983), Structure Building theory (Gernsbacher, 1990) and Functional Grammar
(Givon, 1993). The concerns of Schema Theory relate to reader expectations of the
form and content of text. These schemata develop from previous knowledge and
experience, and tend to be shared by members of a particular discourse community. In
this framework, judgements of text coherence relate to the degree to which in-coming
information meets the expectations that are based in a pre-existing schema. This
involves 'top down' information processing, in which the reader seeks to confirm
predictions made on the basis of such schemata. The correlate to this is 'bottom up'
processing, which involves the chunk-by-chunk processing of language units in the
text, and the creation of text-specific expectations, as examined in Structure Building
theory (Gernsbacher, 1990).
Structure Building theory is concerned with the reader's task of building a coherent
message. It considers text-specific reader expectations based on particular language
properties of the text, and the ways in which such expectations are fulfilled, and the
consequences if they are not. Specifically, cataphoric language serves to raise
expectations by 'pointing ahead' in the text, and anaphoric items serve to fulfil
expectations. Items which function cataphorically occur at the front of any given unit
of the text: that is, in the prompt or title, the text opening, the beginning of the
development and the front of the paragraph. In those positions, text-specific
expectations may be raised for lexical repetition, semantic inference, a particular set
of organisers or a text pattern.
Functional Grammar (Givon, 1993) is concerned with the ways in which language
serves communicative functions. Within a functional framework, the grammar of a
language is regarded as a "set of strategies that one employs in order to produce
coherent communication" (Givon, 1993, p.1). This approach to language is sensitive
to cognitive considerations such as word order and information order, when studying
the functional implications of syntactic or grammatical choice. Word order and
information order have important implications for the cognitive considerations of
focal attention and memory, which, in turn, impact on the reader's ability to 'follow'
and comprehend a text.
Method
The findings presented here are based on the results of quantitative and qualitative
analyses of 164 examination essays written by 67 students of Dentistry at the
University of Western Australia, in response to five different essay prompts. Students
were Native-Speakers (NS), Non-Native Speaker (NNS) residents of Australia, and
International (INT) students. Each group of essays was rated for coherence by four
raters: two content specialists and two Applied Linguists. Raters were untrained, and
were asked to rate the essays on a scale of 1-5, judging whether the essay was a
"readable, well-structured, logically argued and coherent answer to the question".
Based on these judgements, 17 High-rated essays (H-R) and 22 Low-rated (L-R)
essays were identified for comparison.
Two types of analysis were conducted; quantitative and qualitative. The quantitative
analysis compared the frequency of pre-selected language items that might be
predicted to discriminate between the two groups. The qualitative analysis defined
differences that related to the fulfilment or non-fulfilment of cataphoric language
functions, as well as other obvious differences associated with reader judgements of
coherence.
Results
The quantitative analysis revealed only two differences: low-rated essays were
significantly shorter than high-rated essays, and low-rated essays had a higher number
of Discourse-New (Prince, 1992) sentence subjects. Items that did not differ between
the groups were the frequency of items of coreference, logical connectors (Halliday
and Hasan, 1976), units of local coherence, and sentence-initial Inter-Clausal
Connectors (Givon, 1993).
The findings with regard to 'top down' expectations of textual form were that an
'appropriate' genre took the form of a text opening, development and conclusion, in
which a global list of numbered, headed paragraphs was acceptable. In some high-
rated essays, the lack of a conclusion was accepted by readers where this co-occurred
with fulfilled metatextual deixis (eg In this essay I will...). Forms that were not
acceptable included multiple lists and list/sub-list organisation, including the use of
dot or dash points. The issue of text content was addressed in text-specific terms,
relating to lexical repetition of key items in the prompt.
The findings with regard to 'bottom up', text-specific expectations showed that high-
and low-/mid-rated texts could be clearly differentiated on the basis of
fulfilled/unfulfilled expectations based on cataphoric language functions at the front
of textual units. On the basis of these differences, seven text-specific expectations
could be defined. In each case, high-rated essays fulfilled these expectations, and
many low- and mid-rated essays did not.
The first mention of a noun phrase in the prompt or title is cataphorically important.
Exact lexical repetition of that noun phrase is expected, at least once in the body of
the text. Where exact lexical repetition did not occur, the text was not high-rated.
Example
How does saliva contribute to the maintenance of health of the teeth and oral soft
Prompt:
tissues?
The first mention of a noun phrase in the text opening is cataphorically important.
High-rated essays fulfilled cataphoric functions by providing lexical/derivational
repetition, in the development, of a noun phrase first mentioned in the opening. Many
low- and mid-rated texts had 'misleading' text openings in which noun phrases
mentioned in the opening were not referred to again.
It should be noted that, unlike expectations based in the prompt, repetition of words
first mentioned in the opening did not need to be exact, but could be found in
relationships such as maintenance/maintains, growth/grows, lubrication/lubricant. The
following examples illustrate this finding.
High-rated Essay
The first mention of a semantic field in the text opening or the prompt is
cataphorically important, and raises expectations of cohesion by semantic inference.
Both high-rated and low-rated texts fulfilled predictions of semantic inference; the
difference between the two groups lay in the salience of the linking item. In high-
rated texts, the inferred nominal links were placed at the front of the paragraph. In
low-rated texts, the inferentially linked noun phrase was in a position of lesser
salience, such as in the predicate of the paragraph-initial sentence.
The definition of 'organisers' is derived from the cognitive approach of Harold (1995).
Organisers occur in paragraph-initial positions and serve to both group and divide
text. In this study a distinction was made between dividing organisers which indicate
topic change, and unifying organisers, which maintain the existing topic. It was found
that dividing organisers occur in sets (temporal, enumerative, sequential,
predicted/predictable noun phrases, headings), with a minimum number of two in the
set. In a minimal set, the organisers spanned the whole text, that is, the first dividing
organiser was placed near the front of the development, and the second towards the
end of the text.
High-rated essay
eg. enumerative set:
Low-rated essays
(NS 40)
b) incomplete/inconsistent set.
c) numbers only
Metatextual deixis:
This question will be answered by considering the functions of
and thereby deciding how the removal of these functions will affect the oral cavity.
(NS 28)
Low-rated essays
a) confusing opening:
b) unpredictable opening:
Discuss the growth and development of the human mandible
Prompt:
from birth to adulthood
Mandible is the first bone that begins to ossify. The ossification is
Opening
different from the maxillae. (INT 37)
c) Long opening
o gamma globulins,
o albumin and
o blood clotting factors IX, VII, VIII
(NNS 66)
Any factor that will give damage to ' the dentine will result
Opening:
in pulpal response.
o In restorative dentistry, the procedure....
o In biomechanical perspective, damage to
the pulp....
o The mechanical, chemical and thermal
insult will stimulate C and SA fibre....
o In order to reduce the damage to the
dentino-pulpal complex is
Development:
1. Use a lot of water
2. Use light pressure
3. stop the procedure...
4. In deep cavity, calcium...
5. Increase blood flow...
(INT 3)
This feature of text marked occasions where the noun phrase at the front of the
paragraph-initial sentence was not predicted, that is, there was no cataphoric
expectation raised in the previous text. In high-rated essays, the paragraph-
initial noun phrase was always predicted in terms of either previous mention in
the text opening, or predictable in terms of previous expectation by semantic
inference. In low-rated essays, writers often opened a paragraph with a totally
unpredicted noun phrase.
Opening:
o A Biological consideration..............................
o B. Mechanical
consideration................................
o C Chemical consideration.......................
o D Technical consideration....................
(INT 2)
In addition to the above differences between coherent and non-coherent texts,
the results of the qualitative analysis showed that NS, NNS and INT writers
displayed similar problems in failing to fulfil general schema expectations and
text-based expectations raised by cataphoric language functions. It seemed that
they shared a common problem in failing to appreciate the important,
predictive functions of language placed at the front of textual units. Some
differences between the groups were apparent, but in the context of this study,
it is the similarities that are most interesting, for they suggest that all student
writers could benefit from tuition in this regard.
Differences were:
Conclusion
The findings need to be tested in other texts and in other situations, but it
seems clear that this is an approach to the study of coherence that could impact
on the teaching of writing, where students could be taught to appreciate the
cataphoric functions of language placed at the front of textual units. The
reader invests those units with a 'meaning' beyond that of knowledge telling;
they are promises of what is to come, and where those promises are not kept,
the reader appears to have difficulty in forming a coherent message from the
text.
References
Peter White
Centre for Language Teaching and Research
The University of Queensland, Qld 4072
Australia Tel: +61 7 3365 6893; Fax: +61 7 3365 7077