Málfríður - 15.05.1992, Blaðsíða 7
Table 1. Discourse Scores
Classes viewed separately Pre-test Post-test
mean s.d.* mean sd.*
Class 1 (n = 20) 16,13 3,87 16,05 4,86
Experimental Students (n=12) 16,71 3,39 18,33 4,01
Comparison Students (n = 8) 15,25 4,61 12,63 3,42
Class 2 (2n= 16) 11,72 2,57 15,75 4,18
Experimental Students (n = 7) 11,50 2,99 17,07 3,69
Comparison Students (n = 9) 11,89 2,37 14,72 4,46
Both Classes Combined Pre-test Post-test
mean s.d.* mean s.d.*
Experimental Group (n = 19) 14,79 4,08 17,87 3,84
Comparison Group (n = 17) 13,47 3,88 13,74 4,03
* s.d.: standard deviation.
in Class 2 — an average increase of
about 4 points, from 11,72 to 15,75
— was significantly greater than in
Class 1 where the average score
remained almost exactly the same,
at around 16,1 (p = ,001).
3.3.2 Essay length
The Experimental Group wrote an
average of 18 more words per essay
on the post-test than on the pre-test,
rising from 208 to 226 words, while
the Comparison Group wrote an
average of 23 fewer words on the
post-test, declining from 219 to 196
words. This difference between
groups is significant (p=,026). Ag-
ain, it is interesting to note that
mean essay length in Class 2 rose
from an average of 173 to 190
words, while that of Class 1 declined
significantly, from 246 to 230
(p = ,042).
3.3.3. Written accuracy
No significant changes in written
accuracy were found between the
groups or classes, nor even in the
36-student sample as a whole, be-
tween pre-and post-test.
In the pre-test the Experimental
Group made an average of 11,2
errors per 100 words, or 88,8% of
their text was correct (s.d. 4,7). This
figure improved to 89,5% (s.d. 4,5)
on the post-test. Similarily, the Com-
parison Group rose from 86,9% (s.d.
6,5) to only 88,0% (s.d. 3,7) between
pre- and post-test.
Class 1 improved from writing
89,6% of their text correctly (s.d.
4,3) to 90,0% (s.d. 3,9) and Class 2
rose from 85,7% (s.d. 4,2). These im-
provements are insignificant.
3.4. Summary of results
• The study’s hypothesis, that care-
fully directed comments are more
effective in improving advanced
ESL students’ written English dis-
course than no such feedback,
was supported by the study.
• Both discourse scores and essay
length increased significantly
more in Class 2 than in Class 1.
• Neither the teacher’s written-in
corrections, nor a combination of
such corrections with discourse-
oriented comments, resulted in
significantly different written ac-
curacy over the six-week period.
4. Discussion
4.1. The positive effects of
teachers’ written comments
This study indicates that teachers’
written comments can be effective in
improving the written discourse of
advanced adult ESL students. How-
ever, this statement cannot simply be
interpreted to mean that comments
work. Rather, the study indicates
that teachers’ written comments
may be effective when used as an ele-
ment of a larger programme where
classroom instruction has prepared
students for the comments and
students act upon the comments by
rewriting their draft essays, incorpo-
rating the instruction that was con-
tained the feedback.
It is probably not the case that just
any comments are effective. The
comments in this study possessed
the following features:
1. The comments were focused only
on disourse.
2. They were designed to be read
and comprehended. For exam-
ple, the following passages were
used as terminal comments:
a) This is a very good first draft.
Out of this should come a good
essay if you manage to break it
down into paragraphs, each dis-
cussing a separate aspect of the
topic and each joined to the next
in a logical way.
b) Your conclusion sounds rath-
er negative, whereas your essay
on the advantages and disadvant-
ages of dty life was both positive
and negative. You should change
your conclusion!
3. Wherever possible, comments
were focused on specific parts of
the student’s text. For example
the following are examples of
marginal comments:
a) It seems that this [here a
large bracket: indicates four
lines of text] requires a whole
new paragraph because you are
discussing something new.
b) Can’t this [here an arrow po-
ints to text ] be joined to the pre-
vious paragraph, which has the
same idea?
7