Applied Linguistics 19/4 491-514 © Oxford University Press 1998

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
Assessing the Impact of Peer Revision on
L2 Writing
OLGA S VILLAMIL and MARIA C M DE GUERRERO
English Department, Inter American University of Puerto Rico

This study sought to investigate the impact of peer revision on writers' final
drafts in two rhetorical modes, narration and persuasion, among 14 Spanish-
speaking ESL college students Two questions were addressed (1) How were
revisions made m peer sessions incorporated by wnters in their final versions7
(2) How were troublesources revised according to different language aspects
(content, orgamzation, vocabulary, grammar, and mechanics)7 An analysis of
audiotaped interactions, first drafts, and final drafts revealed that 74 per cent of
revisions made in the peer sessions were incorporated In addition, wnters
made many further and self revisions after the sessions These revisions suggest
a pattern of behavior conducive to self-regulation among wnters Results also
show that students focused equally on grammar and content when revising in
the narrative mode and predominantly on grammar in the persuasive mode
Organization was the least attended aspect in either mode Only 7 per cent of
false repairs were found overall The study suggests that peer assistance can help
L2 intermediate learners realize their potential for effective revision, to the
extent their linguistic abilities permit It is the authors' belief that peer revision
should be seen as an important complementary source of feedback in the ESL
classroom

ASSESSING THE IMPACT OF PEER REVISION ON L2

WRITING
The benefits of using peer feedback as an aid to revision in wnting in first (LI)
and second (L2) language classrooms have been amply discussed in the
literature However, even though peer revision enjoys solid theoretical and
empirical support, there are still questions about the learners' capacity to help
each other in solving linguistic problems in their texts Among practitioners,
there seems to be a lingering feeling that L2 students are not knowledgeable
enough to detect and correct errors in the target language Among
researchers, there are also questions about the relative effectiveness of
teacher, peer, and self feedback, as well as whether L2 learners can go
beyond the mere editing of surface problems into higher-level textual aspects
It seems, therefore, that more research on the effects of peer revision on L2
wnting is needed in order to justify its continued use in the classroom In this
paper we assess the impact of peer revision on L2 writing in two rhetorical
modes, focusing particularly on two questions how revisions are incorporated
in final versions of texts and how the different language aspeas of the texts
are revised
492 IMPACT OF PEER REVISION

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
Concerns about the efficacy of peer revision among L2 learners seem to
spring mainly from the fact that m L2 contexts the language of oral and
written communication is, at the same time, the language the students are
trying to leam (Nelson and Murphy 1993) Although some of the problems of
peer revision may be attnbuted to cultural and soaal differences among the
members of response groups (Carson and Nelson 1994, Nelson and Murphy
1992a), the most recurring concern among practitioners is the belief that
students are not capable of detecting and correcting errors in the L2 (Leki
1990, Nelson and Murphy 1992b, 1993) A serious question that has been
posed is whether peer revision yields the type of quality product that teacher-
guided revision might produce and whether it leads 'to enough quality
revisions to warrant the class tune that is required' (Berger 1990 29)
Several studies have investigated the relative effects of different types of
feedback on revision In Berger's (1990) study, for example, peer revision
appeared to have more favorable effects than self revision In Zhang's (1995)
survey study, which gathered students' perceptions through a questionnaire,
peer feedback was preferred over self-feedback Research comparing peer and
teacher feedback has yielded mixed findings Connor and Asenavage (1994)
found that only 5 per cent of revisions done were based on peer group
collaboration and that most of the revisions incorporated were prompted by
the teacher and tutors. Cheong's (1994) results showed that students used the
three kinds of feedback—self, peer, and teacher—but mostly adopted the
changes suggested by the teacher Other studies, however, have presented
opposing results In Chaudron's (1984) study, peer feedback had as much
impact as teacher feedback on revision in terms of improvement Both types
of responses may have helped students to revise their texts Likewise, Caulk
(1994) found that intermediate and advanced ESL students seemed to provide
as much feedback to each other as their instructor Teachers' comments were
general, aimed particularly at the whole piece, whereas peer comments were
more specific in focus Similar results were obtained in Hedgcock and
Lefkowitz's (1992) comparative study of teacher and peer feedback in two
intact FL groups the expenmental group—peer revision—performed on an
equal level as that of the control group—teacher feedback
In spite of the growing body of research on peer revision in L2, very few
studies have looked at how students' negotiations lead to the incorporation of
peer suggestions Nelson and Murphy (1993) found that wnters actually used
their peers' comments while revising although they were not consistent in the
incorporation of peer suggestions throughout the revision sessions Wnters
were more likely to incorporate their peer suggestions when they had
interacted with their peers in a cooperative manner than when they had
interacted in a defensive manner or had not interacted at all Mendonca and
Johnson (1994) also studied how student negotiations influenced revision in
the L2 Their findings showed that wnters were selective about which peer
revisions to adopt, incorporating peer comments in 53 per cent of the cases
A problem that has been pointed out both in LI and L2 peer revision is that
 OLGA S VILLAMIL and MARIA C M DE GUERRERO 493

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
students tend to concentrate more on surface* revisions rather than on
changes that affect meaning Beason (1993), for example, found that surface-
level revisions were the most commonly addressed by LI students as they
revised papers in four courses across the curriculum More recently, Yagelski
(1995) demonstrated that students in an LI course which emphasized the
process approach to wnting tended to focus on surface and stylistic concerns,
suggesting there is no relationship between the classroom context and the
nature and effectiveness of revision Leki has voiced a similar complaint in
relation to L2 students
Instead of engaging with the text they are reading and responding as
real readers, they are likely to respond to surface concerns of grammar,
mechanics, spelling, and vocabulary, taking refuge in the secunty of
details of presentation rather than grappling with more difficult
questions of meaning (1990 9)
Results from empincal studies in the L2 do not present a clear picture on this
issue Berger' s (1990) study, which examined the effects of peer and self
feedback among advanced ESL college learners on three categories—linguistic
structures, content, and form—showed that most revisions were made in the
area of linguistic structures in both groups, followed by revisions in content
Rothschild and Klmgenberger's (1990) study of peer and self evaluation
among high intermediate ESL college learners revealed that the experimental
group—trained in the use of an evaluation scale—made more comments on
content and organization and fewer comments on structure and mechanics
than the control group In Hedgcock and Lefkowitz's (1992) FL study, the
basic wnters who received teacher feedback attended more to grammatical
accuracy, whereas those who were engaged with peers made more content
and organizational changes In Mendonga' and Johnson's (1994) study of peer
review among advanced ESL college students, the participants focused both
on local issues, e g vocabulary, and global issues, e g organization, and rarely
made comments about grammar Finally, Connor and Asenavage's (1994)
case study of two 4-membered groups enrolled in a freshman English course
showed that the types of revisions differed in each group one of the groups
revised more surface errors whereas the other made more text-based
revisions Also in this study, revisions made from teacher comments were
mostly surface changes
An area that has been left unattended in the L2 literature is how peers
revise texts written in different patterns of development or modes of
discourse Traditionally, the LI literature has mentioned that some rhetorical
modes are more difficult than others in terms of the cognitive burden they
may impose on the writer Moffet (1983) contended that the narrative mode
requires the lowest level of abstraction because it depends the most on
chronological order Fitzsimons (1983) and Matsuhashi (1981) have also
posited that, as a recording task, the narrative mode requires less planning
than other modes Persuasive tasks, in turn, have been found to be more
494 IMPACT OF PEER REVISION

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
complex and cogmtively demanding than other forms of discourse in both the
LI (Durst et al 1990, McCann 1989) and the L2 (Ferns 1994, Intaraprawat and
Steffensen 1995)
Although the tasks in some of the L2 studies already mentioned involved
the revision of narrative or expository texts, and/or the transformation of a
text written in one mode to another, the effects of the rhetorical vanable on
L2 revision have not been considered In only one study (Hedgcock and
Lefkowitz 1992) were the FL writers asked to revise two different writing tasks
in the expository and persuasive modes The authors acknowledged that task
differences may have influenced results but were not able to point out how
the groups differed in their performance on the two writing assignments To
our knowledge, no other study to date on L2 peer revision has paid attention
to the rhetorical vanable, that is, how differently students approach the task
of revising m vanous modes
In addition to the above mentioned issues of peer revision, methodological
problems have been pointed out McGroarty and Zhu (1997) complain that
peer revision studies, both in the LI and the L2, are characterized by
inconsistent findings This drawback, the authors claim, is due to the lack of
multiple perspectives, such as tnangulation would permit, with which to
assess the technique Zhang (1995), on the other hand, worries that
effectiveness of the technique has been assumed in the literature without
sound empincal support 1
Despite the need to fill some existing gaps in the literature, it should be
acknowledged that considerable advances have been made in the study of the
social dynamics of interaction dunng peer review Following the line of LI
research (Freedman 1992, Gere and Abbot 1985), a number of L2 studies has
examined the language of wnting groups in an effort to explore what actually
happens when students gather to talk cntically about a piece of writing These
studies have looked at the social dimensions of peer talk during revision,
focusing particularly on how students' stance affects peer revision activities
(Lockhart and Ng 1995, Mangelsdorf and Schlumberger 1992), how social
dimensions generated in a group affect interaction (Nelson and Murphy
1992a), and how training influences peer talk dunng revision (Stanley 1992)
Working within a Vygotskian framework, we have investigated in ,two
previous studies the social-cognitive dimensions of peer review Specifically,
we have focused on how students regulate themselves and others dunng
interaction (Guerrero and Villamil 1994) and have documented the kind of
social-cognitive activities, mediating strategies, and aspects of social behavior
that charactenze peer review (Villamil and Guerrero 1996)
Theoretically, Vygotsky's work (1978, 1934(1986)) offers an ideal stance for
the study of peer review With its emphasis on the social roots of higher
psychological processes, the Vygotskian framework helps us understand how
social interaction and collaboration in the classroom can contnbute to the
development of wnting abilities Seen in this light, peer review emerges as a
favorable instructional environment for readers and writers to work within
 OLGA S VILLAMIL and MARIA C M DE GUERRERO 495

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
their respective 'zones of proximal development' (ZPDs) (Vygotsky 1978,
1934(1986}), a psychological space where latent abilities are vulnerable to
maturation with the appropriate aid of another person. Effective other-
regulation as provided by adults or more expert peers can lead to self-
regulation, or the capacity for independent problem solving {see Wertsch
1985 for an elaboration of Vygotsky's notion of regulation) Recently, Ohta
adapted the concept of the ZPD to L2 learning as follows
For SLA purposes, I would like to conceptualize the ZPD as the
difference between the L2 learner's developmental level as determined
by independent language use, and the higher level of potential
development as determined by how language is used in collaboration
with a more capable interlocutor (1995 96)
Whereas traditional interpretations of the ZPD clearly pose the presence of a
more expert partner (a teacher, a tutor, or a more advanced peer) and suggest
that assistance is unidirectional (from expert to novice), L2 researchers within
sociocultural theory (Brooks and Donato 1994, DiCamilla and Anton 1997,
Donato 1994, Ohta 1995) have been investigating the effects of mutual help
in novice-novice interactions The crucial question is whether L2 learners at
similar stages of development can help each other advance through their
respective ZPDs As Aljaafreh and Lantolf (1994) point out, effective feedback
in the ZPD should be contingent on, and tailored to, the learners' specific
needs and potential level of development Can novice writers offer this type of
finely calibrated assistance to other novice writers7 In our view, peer revision
offers a formidable opportunity to observe the effects of such intervention and
growth in the L2 writing classroom
This study addresses some of the concerns related to peer revision still present
in the literature while incorporating the insufficiently explored rhetorical
variable Basically, we want to find out how and to what extent peer assistance
influences and shapes peer revision Ultimately, our question is whether the
measurable effects of peer revision support its continued use in the classroom
In this study we thus investigate the impact of peer revision on writers' final
drafts across two rhetorical modes, narration and persuasion, among Spanish-
speaking ESL college students Two specific research questions are posed
1 How were revisions made in peer sessions incorporated by writers into
their final versions7
2 How were troublesources revised according to different language aspects
(content, organization, grammar, vocabulary, and mechanics)7

METHOD
Participants
Fourteen intermediate ESL college students from a large private university in
Puerto Rico, working in pairs, constituted the sample for this study The
496 IMPACT OF PEER REVISION

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
participants were drawn from a pool of 54 students we had previously selected
for two former studies we conducted on peer revision (Guerrero and Villamil
1994, Villamil and Guerrero 1996) The students had been selected on the
basis of the following cntena they were native.speakers of Spanish (Spanish
being the dominant language in Puerto Rico), had a score of 500-599 on the
College Board ESLAT,3 had taken two previous courses on the intermediate
level, and had not spent more than a year in an English-speaking country
outside Puerto Rico The students were enrolled in a course which
emphasized the development of writing skills This course was the last of
three 3-credit English courses required within the General Education
component of the university curriculum Previous to this course, the students
had been exposed to little formal instruction on writing or rhetonc at the
university, and it is very likely they had not practised much writing at the
secondary school level' As part of the course activities, the students
participated in two peer'revision sessions, based respectively on a narrative
and a persuasive text The seven pairs selected for this study met the criteria
stipulated and remained constant in their pairings throughout both 'peer
revision sessions We focused on the pairs which remained constant because
we wanted to observe how differently the same pair of students approached
the task of revising in two rhetoncal modes, narration and persuasion In no
way were these students selected because of the extent of their'revisions or
their attitude toward peer revision 4

Data collection procedures

The data were collected in a writing course taught by one of the researchers
In it, students were exposed to two different rhetorical modes narration and
persuasion 5 Four weeks were devoted to the teaching of each rhetoncal mode
so that students would understand the demands of each mode Readings in
each mode were discussed in class, and a journal was assigned In the journal,
the students had to write anecdotes related to the narrative topic and analyze
commercials on TV in order to comprehend the appeals commonly used in
persuasion During the same period of time, students received training in self
and peer revision Sample student essays were projected on a screen, and the
class revised along with the teacher using specially designed writing prompts
to guide revision in each mode (see prompts in Villamil and Guerrero 1996)
Students then practised revising sample student essays utilizing the wntmg
prompts and discussed their revisions with the class Dunng the training
period, doubts regarding the content and language of the compositions as well
as revision procedures were discussed and clarified At the end of each four-
week penod, the students produced a first draft which would be peer revised
in the following class 6
Both peer revision sessions were set up by painng qualifying participants at
random in advance Two external raters, trained by the researchers,
determined which of the two first drafts in each pair 'needed more revision' 7
 OLGA S VILLAM1L and MARIA C M DE GUERRERO 497

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
Thus, a 'writer' (the author of the composition which needed more revision)
and a 'reader' (the other member of the pair, whose task was to assist the
writer) were identified by the raters to form each pair Three of the seven
selected pairs had the same reader/wnter roles in both sessions, while the
other four switched roles 8
Dunng the revision sessions, the students worked in reader/wnter pairs but
were not explicitly informed of the roles we had assigned them for purposes of
the study Each pair was provided with a tape recorder and instructed to
record all their interaction as they revised the draft The pairs were also given
a blank revision sheet for any of the partners to take notes or write comments
Wnters were instructed to initiate the task by reading the draft aloud To
revise the composition, the students were asked to focus first on content and
organization and then on language use and mechanics, just as they had done
dunng training At the end of each revision session, the author of the
composition was asked to produce a final version and submit it within a week
Tapes were transcribed word for word in their entirety by graduate students
and venfied by the researchers

Data analysis
The data for analysis consisted of transcnpts of fourteen audio recordings
(from seven pairs on two revision sessions), first drafts, final versions, and
comments wntten on the revision sheets The analysis focused on 'trou-
blesources' (Nystrand 1986), that is, perceived or potential problems, errors,
or deficiencies in the text which were sources of revision To qualify as a
troublesource, the problematic section of the text had to be the focus of
awareness for the participants, as evidenced through their oral interaction,
their markings on the text, or their wntten comments In some cases, the
source of trouble or revision was only discussed through oral interaction In
other cases, it was only marked on the text and there was no concurrent
discussion In the majonty of the cases, however, troublesources were both
discussed and marked or commented on in wntten form
The data were submitted to an iterative process dunng which trouble-
sources were first tentatively determined and categones were generated
Through this process of examination and re-exammation, troublesources were
finally identified and categones refined First, we identified the m-session
revisions (troublesources dealt with during the peer revision sessions) The
next step was to compare the m-session revisions with the changes observed
in final drafts Thus, changes were categonzed as (a) incorporated (adopted in
peer session and incorporated without changes in the final version), (b) not
incorporated (adopted in session but not in the final version), (c) further
revised (adopted in session and further developed at home), and (d) self
revised (performed at home, not discussed in session) Some of these
categones are exemplified below
498 IMPACT OF PEER REVISION

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
Example of incorporated revision
Draft version Legalize them, it would be to invite everybody to use them
In-sesswn revision {audtotaped)
Reader Here, in this place, you should say 'to legalize them' and you
should eliminate this 'it' because 'legalized', then comma, and this 'it'
does not make sense
Writer Circle it then [Changes appear marked in text ]
Final version To legalize them would be to invite everybody to use them
[grammar revision well-formedness of sentence]

Example of further revision

Draft version The next I was sick
Version adopted in peer session The next day I was sick [vocabulary
revision word 'day' has-been added]
Further revision by writer in final version The next morning I was sick
[word 'morning' has been substituted for 'day']
Example of self revision
Draft version When we found out they weren't followin us my friend
Samuel turned back found them and help them with the car
Final version When we found out they weren't following [mechanics
revision spelling] us my fnend, Samuel told me that maybe something could
happen to their car [content revision detail added] So we turned back in
Samuel car to look for them [content revision clarification of idea] When we
found them I ask Willy what had happened with the car and he told me that the
car had heated [content revision elaboration of idea, details added] We gave
them a bottle full of water so they could continue In a few minutes we were on the
road [content revision details added]
To code the data we first worked jointly on two of the transcnpts to develop
consistency in assigning troublesources to categories Each of the researchers
coded six transcnpts independently, and we then exchanged transcripts to
corroborate identification of troublesources and assignment to categones
Discrepancies were discussed and resolved by mutual agreement
Troublesources were then categonzed according to five language aspects
content, organization, vocabulary, grammar, or mechanics Rather than
defining the five language aspects, we conceptualized each language
component as compnsed of the descnptors that appear in Appendix A We
included linguistic features that reflected the types of troublesources the
students had dealt with as well as features that are typically included in
holistic-analytical essay rating scales (Bernhardt 1988, Hedgcock and
Lefkowitz 1992, Hughey et al 1983) Appendix B provides examples of the
troublesources revised in each language aspect Length, one of the
components of content, was measured by performing a word count, including
the title, of each composition in its draft and final version 9
The analysis included detection of false repairs, also termed 'faulty
 OLGA S VILLAMH. and MARIA C M DE GUERRERO 499

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
corrections' {Rothschild and Klingenberg 1990) or 'incorrect solutions' (Swain
and Lapkin 1995) By false repair, we mean wrongly correcting (a) something
that was correct or <b) something that was incorrect
Examples of false repairs
(a) In that moment -» On that moment
(b) The person who drive this truck can't stop on time -* The person who
dnved this truck can't stopped on time

RESULTS
To answer our first research question (How were revisions made in peer
sessions incorporated by writers into their final versions7), troublesources
revised dunng the peer sessions were counted, yielding the following number
of m-session revisions 179 in the narrative mode and 102 in the persuasive
mode Results presented in Table 1 indicate that the majority of revisions
made in both peer sessions (74 per cent) were incorporated into the final
versions, a few were further revised (8 per cent), and the rest were not
incorporated (18 per cent) In addition"to those revisions carried over from the
peer sessions, final drafts also showed revisions made by the writers on their
own These self revisions accounted for 39 per cent of the total number of
revisions adopted in the final drafts Furthermore, when the length of first
drafts was taken into account, it was observed that the narrative mode
generated a slightly higher number of revisions (115) per 100 words than the
persuasive mode (10 4)
The analysis of m-session revisions provided insights into the nature of peer
assistance and its impact on final drafts For purposes of illustration, we offer a
sample of how a text was modified from a writer's draft version to its final
version through peer (reader/wnter) interaction
Draft version But what really impressed me was that to a side was a downward
hole
Peer interaction
R(eader) 'But what really impressed me was that ' what does it say
here^
W(nter) to a side, 'to a side of where we were was a hole that went down',
'downward', a slope [una pendiente]
R A slope? So, a hole that went downward 'but what really impressed
me was that to a side', next to us, [proximo a nosotrosp
W Yeah, 'next to us' sounds better
R 'what really impressed me was that it was a downward, next to us it was,
next to us' [habfa]
W That it doesn't sound correct
R No, you are conjugating the verb When you say the verb 'to be' in
the-past, I was, you were, she was, he was, it was [hubo, estuvo, habfa]
W Then it would be [habfa], next to us there was
500 IMPACT OF PEER REVISION

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
Table 1 How revisions were incorporated infinal*versions
Narrative , Persuasive Total

Made in peer sessions

Incorporated 136 (76%) - 71 (70%) 207 (74%)
Not incorporated 28 (16%) 23 (22%) 5 1 ' (18%)
Further revised 15 (8%) S (8%) 2 3 (8%)
Total 179 102 281
Revisions on final draft
Incorporated 136 (57%) . 71 (51%) 207 (55%)
Further revised 15 (6%) 8 (6%) , 23 (6%)
Self revisions 90 (37%) 59 (43%) 149 (39%)
' Total 241 1*38 379
Revisions per 100 wordsa
.11 5 10 4

"These figures were calculated by dividing the total number of revisions on final
drafts by the total number of words in first drafts in each task (2,094 words,"
narrative task, 1,327 words, persuasive task) and multiplying by 100

R 'there was [de 'haber', habia] a downward hole

Final version But what really impressed me was that next to us, there was a
downward hole
' ' ' ,• -
In the above example both reader and writer demonstrated areas ,of weakness
and expertise The reader was responsible for calling the writer's attention to
the potential troublesource (the sentence beginning 'But what really
impressed '), offering a better alternative ('next to us'), considering the
use of an expletive ('it'), and giving the wnter a mini lesson in verb
conjugation The writer, for his part, was able'to 1 clarify the meaning of
'downward' for the reader and to discriminate between acceptable and
unacceptable alternatives* ('"next to us" sounds better', 'that it doesn't sound
correct') while solving the problem of'the correct expletive to use with the
verb Both reader and writer made contingent use of their LI' (Spanish) to
make meaning In short, both peers offered and received feedback on revision
that ultimately resulted in changes that were adopted by the writer in his final
draft ' • •'
To answer our second research question, revisions were analyzed to
determine what language aspect {content, organization, grammar, vocabu-
lary, and mechanics) they belonged to Results in Table 2 show that grammar
revisions (31 per cent) predominated in the narrative mode closely followed
by content revisions (27 per cent) In the persuasive mode, grammar revisions
 OLGA S VILLAM1L and MARIA C M DE GUERRERO .501

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
(38 per cent) also occupied the first place and vocabulary revisions (22 per
cent) the second place Among self revisions, content ranked first in the
narrative mode, while grammar ranked first in the persuasive mode Overall,
grammar was the most revised aspect in both modes, while organization was
the least attended to
In the process of incorporating peer revisions and rewriting final drafts,
writers modified the number of words As can be seen in Table 3, results on
length indicate that a higher number of words were produced in the narrative
than in the persuasive mode in both draft and final versions A comparison of
length between draft and final version showed that, whereas some final
versions decreased in number of words, most increased There was an average
increase of 68 words in the narrative mode and 38 57 in the persuasive mode
The highest single increase was of 397 words {Dyad VII, narrative mode)
Troublesources in all areas were revised by deleting, adding to, substituting,
or in some way modifying the text in the original version Whereas most
revisions implied repairing or attending to potential deficiencies in the text,
not all revisions were for the sake of repair Some entailed substituting
appropnate and correct language with other-sometimes better-aitematives,
for example, 'This leads me to say' was changed to 'This leads me to believe'
and 'It was very cold' was changed to 'The night turned very cold ' These
revisions can be taken as stylistic changes or attempts to sharpen the meaning
of the text made while students were operating in a revision mode A few false
repairs did occur, however There was a total of 32 false repairs, 18 of which
were m-session revisions and 14 of which were self revisions Most false
repairs were in the areas of grammar (15) and mechanics (15), two were in
the area of vocabulary (all revisions in content and organization were
considered appropnate) Five of the m-session false repairs were not
incorporated into the fmal versions Overall, there were 27 false repairs in
the final drafts This constitutes 7 per cent of all revisions (379) performed on
final drafts It should be noted that other rhetorical and grammatical problems
or errors may have remained unattended in the final drafts Our focus in this
paper, however, was not to count errors but to analyze those revisions that
were discussed and/or performed by students

DISCUSSION
The impact of peer collaboration on revision was first measured by the extent
to which writers incorporated their peers' suggestions and worked on their
final drafts on their own Results show that peer assistance had a substantial
effect on revising, insofar as the majority of troublesources revised dunng
interaction were incorporated into final versions If wnters had not had the
benefit of a peer assistant, it is possible the final number of revisions in final
drafts would not have been as high At the same time, the fact that some m-
session revisions were not incorporated and that several self revisions were
added demonstrates that, ultimately, it was the wnters' prerogative to choose
 >
n
H
O
Table 2 Revisions according to language aspect in narrative and persuasive modes
B
Narrative mode Persuasive mode

Language aspect In-session Self-revised* Total In-session Self-revised Total

o
Content 37 (24%) 29 (32%) 66 (27%) 11 (14%) 12 (21%) 23 (17%) z
Organization 4 (4%) 3 (3%) 7 (3%) 3 (4%) 4 (7%) 7 (5%)
Grammar 52 (34%) 22 (24%) 74 (31%) 30 (38%) 22 (38%) 52 (38%)
Vocabulary 30 (20%) 27 (30%) 57 (24%) 19 (24%) 12 (21%) 31 (22%)
Mechanics 28 (18%) 9 (10%) 37 (15%) 16 (20%) 9 (16%) 25 (18%)
Total 151 90 241 79 59 138

Note In-session figures include those revisions made in the peer sessions that were incorporated or further revised in the final
version

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
 OLGA S VILLAMIL and MARIA C M DE GUERRERO 503

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
Table 3 Number of words in draft and final versions
Narrative Persuasive

Dyad Draft Final Change Draft Final Change

I 455 487 +32 194 211 +17

n 242 195 -47 206 209 +3
m 231 293 +62 145 171 +26
IV 289 321 +32 285 454 + 169
V 329 337 +8 182 214 +32
VI 278 270 -8 127 115 -12
vn 270 667 +397 188 223 +35
Total 2,094 2,570 +476 1,327 1,597 +270
Mean 299 14 367 14 +68 189 57 228 14 +38 57

from the suggestions and exercise authonal control Our findings corroborate
those of Mendonca and Johnson (1994) and Nelson and Murphy (1993),
which showed that selective decisions about adoption were finally made by
the writers Mendonca and Johnson's study actually yielded very similar
findings to ours In their study, 53 per cent of revisions in final drafts were
incorporated suggestions made by peers, 10 per cent were not incorporated,
and 37 per cent were self revised As Nelson and Murphy (1993) suggest, it
may be the nature of peer interactions which determines to what extent peer
suggestions are incorporated, interactions where partners work cooperatively
generating more peer-based revisions than interactions characterized by
disagreements and defensive attitudes
In our study, peer sessions were characterized by collaboration Both
readers and writers became active partners in the revision task as they
reciprocally extended support From a Vygotskian perspective, regulation
shifted hands throughout the revision sessions with both partners sharing
expertise on a contingent basis We might say that expertise resided in their
joint efforts, rather than in one individual alone Writers, specifically, were
able to move through their ZPDs beyond their current levels of development,
as reflected in their first drafts, to higher levels of achievement by
appropriating the changes that were jointly generated and incorporating
them into their final drafts Of course, the writers' incorporation of changes
into final drafts acts only as tentative evidence of actual growth Whether
development occurred or not is to be seen in independent writing
performance, some signs of which were provided by our writers' 'further'
and 'self revisions Testing independent performance after peer intervention
was beyond the scope of our study, but it is an important step in our research
504 IMPACT OF PEER REVISION

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
agenda (For two studies that investigate the connections between peer
interaction and independent performance, see Swain 1998, Swain and
Lapkin, in press )
The 'pres'ence of .'further revisions' and 'self revisions in final drafts
indicates that wnters, before turning in their final versions, worked
considerably on their own on the basis of the previous peer work What the
writers did after peer interaction may be taken as symptoms of self-regulatory
behavior In other words, the writers' reformulation of the changes made in
sessions and ail other revisions made on their own could be indicative of the
effect that other-regulation (peer assistance) had on self-regulation (inde-
pendent performance), as suggested above Through other-regulation, certain
linguistic or rhetorical processes which were in a state of development or
instability may have had an opportunity to mature and consolidate, and new
knowledge may have been generated It should be pointed out that self-
regulation is never permanent nor all-encompassing (DiCamilla and Anton
1997) but it enables the individual to function independently in certain tasks
Thus, as we expressed elsewhere (Guerrero and Villamil 1994, ViUamil and
Guerrero 1996), the value of joint activities earned out dunng peer revision
lies in their role as social precursors of individual processes that are necessary
for independent writing production ,
j
The effects of peer collaboration on revision were also measured in terms of
how students revised troublesources with respect to the five language aspects
Results show that.students focused almost equally on grammar and,content
when revising in the narrative mode and predominantly on grammar in the
persuasive mode Why would students make more revisions in grammar than
in any other language area despite explicit instructions to concentrate on
content and organization first and then on language use and mechanics7 One
explanation may be that ESL learners at the intermediate level are still quite
unstable in their command of language structure and, consequently, feel the
need to deal with aspects of form Another explanation may be that students
needed to, so to say, 'clean' the text first of linguistic anomalies that might
have obscured the meaning of the text A third explanation is that, regardless
of the course emphasis on higher level textual concerns, students simply
followed their habitual tendency to focus on grammar, as probably learned
throughout much of their previous language instruction In the students'
emphasis on grammar we can perceive the historical effects of traditional
form-oriented L2 training
In-the literature, there is some evidence of L2 students' concern t with
grammar when revising In Berger's (1990) study on the effects of peer and
self-feedback on revision, L2 students -focused- pnmanly on 'linguistic
structures'—a category which in that study encompassed grammatical aspects
as well as vocabulary, spelling, and punctuation 10 Concern with surface and
mechanical revisions is not restricted to the L2 Fitzgerald has pointed out that
LI students at various levels of competence focus on surface modifications,
'often revealing a view of revision as proofreading' (1987 492) Likewise,
 OLGA S VILLAMIL and MARIA C M DE GUERRERO 505

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
Bereiter and Scardamaha have expressed that student writers 'are famous for
their avoidance of revision and for confining it to a cosmetic level' (1987 22)
Interestingly, in Hall's (1990) study comparing the revisory behaviors of LI
and L2 students, L2 students earned out more grammatical corrections than
LI students
A further point needs to be made in this respect It is commonly assumed that
a limitation of peer revision in the L2 is precisely the focus on surface-level
aspects such as grammar and mechanics Yet, as Beason has clearly put it.
Revision research as a whole does not warrant the conclusion that
focusing on Local-meaning Revision or Surface-level Revision fails to
benefit the wnter or text The overall quality of a text might not be
significantly enhanced by Surface-level Revisions, but even a mended
comma splice contributes in its own way to the readability of a text and
the writer's linguistic development (1993 416)
Beason further believes that 'revisions operating below the global level—
despite the somewhat trivializing labels of surface-level or microstructure
revisions—are often cogmtively demanding as well' (1993 416) Although
our ultimate goal as teachers may be that students focus primarily on deep-
level areas such as content and organization, we agree with Beason that L2
students' preoccupation with surface aspects represents a cognitive and
linguistic effort that needs to be acknowledged
This being said, we should not underemphasize the fact that the students'
second most important concern was content- in the narrative mode and
vocabulary in the persuasive mode The data revealed'that revisions in the
five language aspects varied according to the text's rhetorical mode We
hypothesize, then, that mode could .have affected the focus of attention
during peer revision This varying focus of attention has already been pointed
out in the literature Yagelski has noted in a study on self revision conducted
with LI [Writers that 'there might be differences in revisions across essay
forms' (1995 219) Hedgcock and Lefkowitz (1992) have also acknowledged
that task differences—expository and persuasive—may have influenced the
performance of their L2 sample during peer revision
The effect of the rhetorical variable was not only seen in the learners'
treatment of the different language aspects but also in the number of words
and revisions generated in each mode The fact that the narrative mode
resulted in longer first drafts, greater gains in length after revision, and a
slightly higher number of revisions than the persuasive mode strengthens the
notion that persuasion is a more cogmtively demanding genre According to
reports of the National Assessment of Educational Progress (NAP), persuasive
wnting tasks present considerable difficulty to elementary and high school
students in the United States (cited in Oliver 1995) It has been found that
scores on national exams reflect 'the difficulty that students have when they
move from narration and summary to a more analytic, interpretive
framework' (Durst et at 1990 223) Ferns (1994) and Intaraprawat and
506 IMPACT OF PEER REVISION

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
Steffensen (1995) have contended that the difficulty imposed by the genre
seems to be greater to L2 than to LI writers Intaraprawat and Steffensen posit
that

When writers are not highly skilled, when they are wnting in an L2,
when they are under time constraint, presenting content and
effectively organizing an argumentative structure are the most import-
ant considerations Increasing the likelihood of actually persuading the
reader by referencing an authority is a level of rhetorical sophistication
that does not occur with high frequency in these papers {1995 268)

It is possible that the higher cognitive demands of the persuasive mode

together with lack of familiarity with the mode may have been responsible for
the lower number of words and modifications produced in ,it by our
students u
Why and how L2 students revise differently according to rhetorical modes
needs to be further explored In particular, it would be interesting to find out
if some topics indeed elicit different writing and/or revising skills and if some
wnting tasks impose more cognitive demands on L2 writers This information
will enable researchers and practitioners not only to broaden their scope of
the act of revision across rhetorical modes, but also to make informed
decisions on how to better prepare students to meet academic demands and
actively participate in the discourse of their professional field Our belief is
that, given the well-attested differences presented in the literature, the
rhetorical mode is an important variable and should be included in the
research design of future studies on wntten revision in L2
Why organization received the lowest percentage m the number of
revisions is also an area that needs to be examined A possible explanation
might be that students did not revise organization as much as content
because, as they were expanding and elaborating on content, they assumed
they were also reorganizing the matenal It is also possible that when it comes
to quantifying revisions, organizational changes, which alter the macro-
structure of the text, will always be less numerous than lexical or grammatical
ones, which are done at the level of microstructure (Whereas moving one
whole paragraph around would count as one revision,' within the paragraph
there may be multiple revisions in terms of grammar, content, vocabulary,
and mechanics) The L2 literature, unfortunately, sheds little light on this
aspect of revision, as there is, to our knowledge, no study we can compare our
results with on the issue of how students revise organization 12 In the LI
literature on revision, however, there is some evidence that students do not
prioritize organization while revising (Beason 1993, Bernhardt 1988) Bereiter
and Scardamaha (1987) offer a possible answer to this question They state
that novice writers are 'under immediate threat of information overload',
which will enable them to focus on some local problems and even on the
generation of content but not on higher textual features such as organ-
izational patterns (1987 357) Furthermore, students might not modify their
 OLGA S VILLAMIL and MARIA C M DE GUERRERO 507

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
rhetorical organization once they have sketched their first drafts because
doing so would imply too big a change for the overall text Given that peers do
not seem to focus much on organization, it may then be advisable for teachers
to spend more time on this aspect and to leave other features like content and
vocabulary to the learners 13
Results on length provide another indication of the impact of peer
collaboration on revision The increase in length in final versions demon-
strates that peer interactions had an effect on rethinking and restructuring the
onginal text Two cases were particularly outstanding (see Table 3), with
increases of 397 (dyad VII) and 169 (dyad IV) words in the narrative and
persuasive mode, respectively Decreases in length, which occurred in only 3
out of 14 cases, cannot be regarded, however, in a negative light Even though
it may be too obvious to mention, it is sometimes desirable to cut words out
rather than add in order to improve a text Actually, because final figures on
length tend to conceal the nature of changes performed on drafts, it should be
noted that all compositions underwent word additions and deletions as
students revised the text, a fact which again bespeaks the students'
discriminating powers when it comes to revising
In general, the students' revisory behaviors in our study were quite
complex and vaned As peers, students were not only capable of repairing
perceived sources of error in all of the language aspects but also of altenng
length and reformulating the text in ways that went beyond mere editing and
correction Although the students did incur false repairs in some—a problem
of peer revision cited in the literature (Rothschild and Khngenberg 1990) and
a vivid preoccupation among some practitioners—we would like to offer the
possibility of viewing false repairs as manifestations of language growth
Swain and Lapkin (1995), for example, have suggested that incorrect
solutions in students' output could be indicative of important language
learning processes False repairing, overcorrecting,'or leaving errors uncor-
rected are typical behaviors of students whose linguistic systems are in a state
of development As Lantolf and Aljaafreh (1996) point out, it is natural for L2
language development to be irregular, uneven, erratic, and regressive,
therefore, this behavior is to be expected as students are restructuring their
language systems How false repairs are viewed, as 'problems' in L2 peer
revision or as signals of language development, has strong implications for
pedagogy and research How are teachers or external raters, for example,
affected by the presence of false repairs when assessing L2 writing7 Given the
significance of these types of error, will teachers continue to devote the
necessary class time to the practice of peer revision? The issue ments careful
consideration

CONCLUDING REMARKS
As Gere and Stevens have suggested, 'the most telling measure of the effect of
writing group response lies in the changes it induces' (1985 99) Our analysis
508 IMPACT OF PEER REVISION

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
of the changes induced by peer assistance in L2 writers' drafts across two
rhetoncal modes has yielded, in summary form, the following findings (a) the
majonty of revisions made in collaboration were incorporated by the writers
into their final drafts, (b) writers made further revisions and self revisions on
the basis of previous peer collaboration, suggesting a pattern of behavior
conducive to self-regulation, (c) students focused mainly on grammar and
content when revising in the. narrative mode, and on grammar in* the
persuasive mode, (d) grammar was the most revised aspect whereas
organization was the least attended to, (e) most final drafts increased >m
length, with a higher increase in the narrative mode than in the persuasive
mode, and (f) 7 per cent of all revisions were false repairs <
These findings provide a measure of the impact that peer revision has on
ESL learners trying to improve a piece of writing - To judge the extent of this
impact, we need to take rntO'consideration two very important factors First,
L2 learners, because of'the.very nature of their as yet underdeveloped L2
system, can only revise to the extent of their abilities When learners are
operating within their ZPDs, there is always the possibility of moving beyond
their actual level of development but never beyond their maximum
immediate potential As Vygotsky stated, 'with assistance, every child .can
do more than he can by himself—though only- within the limits set by the
state of his development' (1934(1986) 187) An intermediate ESL student's
maximum potential is still not equal to the writing ability of an expert, native
or otherwise, no matter how much help they are given and'by whom --It
would thus be'unrealistic and inappropriate to expect revised drafts to be free
of errors Second, collaboration in the classroom in the form of peer revision is
not a substitute for teacher feedback We do not see peer feedback and teacher
feedback as being in competition, rather, we see them as complementary
forms of assistance in the writing classroom Perhaps the time has come to
stop asking ourselves, 'Which is better (or which is more effective), .peer
feedback or teacher feedback"?' We should begin to ask ourselves instead,
'What and how can peer revision contribute to the students' writing
development in a way that complements teacher feedback?',
From the perspective of our study, we can say that peer revision can
contnbute to writing-development in many important ways The practice our
students acquired in the process of peer revising could not have been acquired
elsewhere Moreover, the expenence of peer revision provided our students
with an unparalleled opportunity to discuss textual problems, internalize the
demands of two rhetoncal modes, develop self-regulatory behaviors, acquire a
sense of audience, and in general become sensitive to the social dimension of
writing Above all, through peer revision the students were involved in ,the
process of acquiring strategic competence in revising a text, a competence
which will prove invaluable in their future academic and professional life
(Revised version received February 1998)
 OLGA S VILLAMIL and MARIA C M DE GUERRERO 509

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
ACKNOWLEDGEMENTS
To conduct this research we were granted a reduced course load from the administration of Inter
American University of Puerto Rico, for which we are very grateful We also wish to thank the
invaluable contribution of editors and anonymous reviewers for the improvement of earlier drafts
of this paper

APPENDIX A
Descriptors of language aspects
Narrative task
CONTENT Elaboration of ideas, clanty of ideas and meaning, ideas related to one main
event, relevance of ideas to topic, details to recreate event, dialogue, message, setting,
title, length
ORGANIZATION Parts of composition (beginning, middle, end, introduction, body,
conclusion), ideas connected chronologically, placement of ideas within parts of
composition, logical sequence of events, transition words, paragraphing, overall
structure
GRAMMAR Movement from sentence to sentence, well-formedness of sentences,
subject-verb agreement, complete sentences, morphosyntacuc fluency, verb tenses and
verb forms, use of articles, modals, prepositions, pronouns, and conjunctions, word
forms (if word is used as noun, verb, adverb, adjective, etc ), number (plural/singular),
gender, word order, negation, 'there' structure
VOCABULARY Variety of diction, avoidance of repetition, appropriateness of diction,
effectiveness in expressing meaning, accuracy of word choice, using English terms
(appropriateness of translation, use of Spanish words), idiomatic usage
MECHANICS Punctuation, contractions, spelling, capitalization, handwriting, other
conventions (use of words for numbers, parentheses, symbols, etc ), indention of
paragraph

Persuasive task
CONTENT Background knowledge, opposing viewpoint considered, adequate/enough
support (facts, examples, evidence, details), topic stated, opinion stated, elaboration of
ideas, clanty of meaning and information, relevance of ideas, message, setting, title,
length
ORGANIZATION Parts of composition (beginning, middle, end, introduction, body,
conclusion), ideas connected logically, cohesiveness of ideas, placement of ideas within
parts of composition, transition words, paragraphing, overall structure
GRAMMAR Movement from sentence to sentence, well-formedness of sentences,
subject-verb agreement, complete sentences, morphosyntacuc fluency verb tenses and
verb forms, use of articles, modals, prepositions, pronouns, and conjunctions, word
forms (if word is used as noun, verb, adverb, adjective, etc ), number (plural/singular),
gender, word order, negation, 'there' structure
VOCABULARY Variety of diction, avoidance of repetition, appropnateness of diction,
effectiveness in expressing meaning, accuracy of word choice, using English terms
(appropnateness of translation, use of Spanish words), idiomatic usage
MECHANICS Punctuation, contractions, spelling, capitalization, handwriting, other
510 IMPACT OF PEER REVISION

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
conventions (use of words for numbers, parentheses, symbols, etc), mdention of
paragraph

APPENDIX B
Examples of troublesources revised in the different language
aspects
CONTENT
Development of idea
Draft version In conclusion I think the decision is of the person or of the family
because it's difficult but if the person are suffer very much is better the euthanasia
Final version In conclusion I think the decision of this is in the person himself if he is
conscious of what he is domg But if the person doesn't know what its happening I
think the decision is a family affair
Details added
Draft version For that day the candidates had to be physically and mentally prepared
for any questions that the jury could ask
Final version For that day the candidates had to be physically and mentally prepared
for any questions that the jury could ask For example, they ask me. What people did I
admire'' How could I help young people7 etc
ORGANIZATION
Paragraphing (breaking one paragraph m two)
Draft version In that moment I said 'I want to visit my uncle and I can't nde with
you When I was in the street (I go alone) In the street #1 from Caguas to Rio Piedras
one truck
1
Final version In that moment I said 'I want to visit my uncle so I can't nde with
you' Then I went to my house alone
When I was in the street #1 from Caguas to Trujillo Alto one truck
GRAMMAR
Sentence well-formedness
Draft version I was impressed at how can a house could be so far from the roads
Final version I was impressed at how a house could be so far away from the roads
Verb forms
Draft version I'm not agree
Final version I don't agree
VOCABULARY
Word choice, use of English term
Draft version we have a case in the tribunals for neghllense
Final version we have a trial in court for negligence
MECHANICS
Use of words for numbers
Draft version After 2 hours
Final version After two hours
Spelling
Draft version were sopost to be pick up
Final version were supposed to be pick up
 OLGA S VILLAMIL and MARIA C M DE GUERRERO 511

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
NOTES
1 A reviewer has pointed out that one of the letters, and resumes Oral communication
key methodological problems of peer in job interviews Analysis of complex
review research is the lack of sufficiently grammatical structures in context'
fine-grained instruments, sensitive to 6 The topics for compositions were 'Narrate
changes in students' revisions throughout an experience that taught you something
multiple drafts about yourself or an experience that made
2 The students ranged in age from 18 to 25 you reflect on life' and 'Convince/persuade
years {M = 19 64) and were m their 2nd or your audience about that controversial
3rd year at the university majoring in issue that you have been working on for
various fields of study, such as accounting, the last few weeks'
biology, and social work Eight students 7 As part of the training, three raters (ESL
were female and six were male instructors at the college level) analyzed
3 The ESLAT (English as a Second Language pairs of sample student essays written by
Achievement Test) is the College Board ESL other intermediate ESL writers to deter-
test version for Puerto Rican college can- mine which paper needed more revision m
didates This test ranges from 200 to 800 terms of content, organization, vocabulary,
points A student taking the ESLAT would grammar, and mechanics Descriptors in
score approximately 100 points less in the current holistic-analytical essay ratings
TOEFL (Alderman 1981) were used to explain each language
4 A further note on our selection of partici- aspect Raters then practised on their own
pants, which could be considered a limita- and discussed their ratings with the group
tion of this study, is that there was no After the training, two of the raters were
matched control group with which to asked to work on the 14 sample papers of
compare our results A matched control this study In case of discrepancies, the third
group would be one in which teacher, rater was used Interrater agreement was 86
tutor, or self-feedback, instead of peer per cent
feedback, were offered This type of control 8 An anonymous reviewer has pointed out
could have shown whether the same results that since all the pairs did not remain
could have occurred with some other kind constant in their reader/wnter roles, the
of feedback Some of the studies reviewed comparison between modes is not war-
in our introduction have this type of ranted We do not agree with this comment
experimental design In our case, due to because it views the wnting task as the sole
the limited number of students available at property of the 'wnter' We envision the
the necessary proficiency level, it was not task differently For us, revision becomes a
feasible to include this type of control in the shared task where both reader and wnter
research design of our original larger study construct meaning jointly Dunng inter-
In addition, it was not our intent in this action the reader takes on the writer's
study to compare different types of feed- role, and the wnter takes on the reader's
back but to probe deeply into the nature of role What we are thus comparing is the
peer response product of joint rather than individual
5 The teaching—and sequencing—of these production
two rhetorical modes narration and per- 9 Some non-parametnc statistics were run
suasion, was required by the course sylla- but were ultimately disregarded because
bus This course is described in the they were not found to enhance the
university catalogue as follows 'Refine- analysis of the data in any impressive
ment of the communication skills acquired way The main problem in applying statis-
in the previous course Emphasis on the tics to our study was the limited number of
writing of short essays based on readings, pairs of students (7) For the benefit of
using a guided wnting technique Also researchers who would wish to replicate
includes the wnting of reports, summaries, our study, a bnef report of our statistical
512 IMPACT OF PEER REVISION

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
efforts is offered here The Wilcoxon order for some of the students We acknow-
matched-pairs signed-ranks test was per- ledge this as a methodological limitation
formed to observe differences between the that has bearings on the interpretation of
number of words in draft and final versions the results However, on pedagogical
and between the narrative and persuasive grounds, scheduling the persuasive task
modes in the total number of revisions (per before the narrative one would have been
100 words) in final drafts A significant unwise for our intermediate ESL wnters
difference (if alpha level were set at 05) 12 A category somewhat resembling our
was found in number of words when 'organization' category is Berger's (1990)
narrative and persuasive drafts were com- form' category although not enough infor-
pared (p = 018) and between drafts^and mation is provided by Berger to conclude so
final versions in the persuasive mode {p = unequivocally In that study, 'form' ranked
042) but not in the narrative mode There last in number of revisions after 'linguistic
was a non-significant difference between structures' (first) and content (second)
the narrative and persuasive mode in Nelson and(Murphy (1992b), on the other
number of revisions hand, found that organization ranked first
10 Unfortunately, Berger (1990) does not among several textual deficiencies identi-
specify how many of the changes included fied by L2 learners in their drafts In this
under the label 'linguistic structures' would study, students had been instructed not to
fall under the rubric of grammar , correct grammar, punctuation, and spel-
11 Because the sequencing stipulated by the t ling ,The elimination of these categories
, course syllabus called for the narrative task makes companson with our study difficult
to be scheduled before the persuasive task, 13 We are indebted to two anonymous
it was not possible to neutralize the effects reviewers for their insights into the issue
of the rhetorical mode by reversing the of organization

REFERENCES
Alderman, D L 1981 Language Proficiency as a Carson, J G and G L Nelson 1994 'Writing
Moderator Vanable in Testing Academic Aptitude groups Cross-cultural issues ' Journal of
(TOEFL Research Report No 81-41) Second Language Writing 3/1 17-30
Princeton, NJ Educational Testing Service Caulk, N 1994 'Comparing teacher and
Aljaafreh, A and J P Lantolf 1994 'Negative student responses to written work' TESOL
feedback as regulation and second language Quarterly 28/1 181-8
learning in the zone of proximal develop- Chaudron, C 1984 'The effects of feedback on
ment ' Modern Language Journal78/4 465-83 students' composition revisions ' RELC Jour-
Beason, L 1993 'Feedback and revision in nal 15 1-16
wnting across the curriculum classes ' Research Cheong, L K 1994 Using annotation in a
in the Teaching of English 111 A 395-422 process approach to wnting in a Hong Kong
Bereiter, C and Scardamaha, M 1987 The classroom ' TESL Reporter 27/2 63-73
Psychology of Written Composition Hillsdalc, Connor, U and K Asenavage 1994 'Peer
NJ Lawrence Erlbaum Associates , response groups in ESL wnting classes How
Berger, V 1990 'The effects of peer and self- much impact on revision?' Journal of Second
feedback ' The CATESOL Journal November Language Wnting 3/3 257-76
21-35 DiCamilla, F J and M Anton 1997 'Repeti-
Bernhardt, S A 1988 'Text revisions by basic tion in the collaborative discourse of L2
wnters From impromptu first drafts to take- learners A Vygotskian perspective \ The
home revision' Research in the Teaching of Canadian Modern Language Review 53/4
English 22/3 266-80 609-33
Brooks, F B and R Donato 1994 'Vygots- Donato, R 1994 'Collective scaffolding in
ban approaches to understanding foreign second language learning' in J P Lantolf
language learner discourse during commun- and G Appel (eds ) Vygotskian Approaches to
icative tasks ' Hispama 77 262-74
 OLGA S V1LLAMIL and MARIA C M DE GUERRERO 513

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
Second language Research Norwood, NJ tional Journal of Educational Research 23/7
Ablex 51-64
Durst, R , C Lame, L M Schultz, and W Leki, I 1990 'Potential problems with peer
Vilter 1990 Appealing texts The persuas- responding in ESL classes' The CATESOL
ive wnting of high school students ' Written Journal November 5-19
Communication 7/2 232-55 Lockhart, C and P Ng 1995 'Analyzing talk
Ferns, D 1994 Rhetorical strategies m stu- in ESL peer response groups Stances func-
dent persuasive wnting Differences between tions, and contents ' Language Learning 45/4
native and non-native English speakers' 605-55
Research m the Teaching of English 28/1 45-McCann, T M 1989 'Student argumentative
65 wnting knowledge and ability at three grade
Fitzgerald, J 1987 Research on revision in levels Research in the Teaching of English 23/
wnting ' Review of Educational Research 57/4 1 62-76
481-506 McGroarty, M E and W Zhu 1997 Tnan-
Fitzsimons, J T 1983 'Task and audience gulation in classroom research A study of
Effects of planning processes and commun- peer revision ' Language Learning All I 1-43
icative effectiveness of competent wnters' Mangelsdorf, K and A Schlumberger 1992
Dissertation Abstract International 44 2072A 'ESL student response stances in a peer-
(University Microfilms No 83-26, 170) review task Journal of Second Language
Freedman, S W 1992 Outside-in and mside- Writing 1/3 235-54
out Peer response groups in two ninth- Matsuhashi, A 1981 'Pausing and planning
grade classes ' Research in the Teaching of The tempo of wntten discourse production
English 26/'1 71-106 Research in the Teaching of English 15 113-34
Gere, A R and R D Abbott 1985 'Talking Mendonca C O and K E Johnson 1994
about writing The language of wnting 'Peer review negotiations Revision activities
groups ' Research in the Teaching of English in ESL wnting instruction ' TESOL Quarterly
19/4 362-85 28/4 745-69
Gere, A R and R S Stevens 1985 'The Moffet, J 1983 Teaching the Universe of Dis-
language of wnting groups How oral course Boston Houghton Mifflm
response shapes revision' in S W Preedman Nelson, G L and J M Murphy 1992a 'An
(ed ) The Acquisition of Written Language L2 wnting group Task and social dimen-
Response and Revision Norwood NJ Ablex sions ' Journal of Second Language Writing 1/3
Guerrero, M C M de and O S Vlllamil 171-93
1994 'Social-cognitive dimensions of inter- Nelson, G L and J M Murphy 1992b
action in L2 peer revision ' The Modern 'Writing groups and the less proficient ESL
Language Journal 1%I4 484-96 student TESOL Journal 2/2 23-26
Hall, C 1990 'Managing the complexity of Nelson, G L and J M Murphy 1993 'Peer
revising across languages ' TESOL Quarterly response groups Do L2 wnters use peer
24/1 43-60 comments in revising their drafts? TESOL
Hedgcock, J and N Lefkowitz 1992 'Col- Quarterly 2771 135-41
laborative orat/aural revision in foreign Nystrand, M 1986 'Learning to wnte by
language wnting instruction ' Journal of talking about wnting A summary of
Second Language Writing IB 255-76 research on intensive peer review in expo-
Hughey, J B , D R Wormuth, V F Hartfiel, sitory wnting instruaion at the University of
and H L Jacobs 1983 Teaching ESL Wisconsin-Madison' in M Nystrand (ed )
Composition Principles and Techniques New The Structure of Wntten Communication
York Newbury House Orlando, FL Academic Press
Intaraprawat, P and M S Steffensen 1995 Ohta, A 1995 Learner-learner collaborative
The use of metadiscourse m good and poor interaction in the ZPD' Issues in Applied
ESL essays' Journal of Second Language Linguistics 6/2 93-121
Writing 4/3 253-72 Oliver, E I 1995 The wnting quality of
Lantolf, J P and A Aljaafreh |1996) L2 seventh ninth and eleventh graders and
learning in the zone of proximal develop- college freshmen Does rhetoncal specifica-
ment A revolutionary expenence Interna- tion in writing prompts make a difference9
514 IMPACT OF PEER REVISION

Downloaded from https://academic.oup.com/applij/article-abstract/19/4/491/265645 by Hankuk University of Foreign Studies user on 04 December 2019
Research in the Teaching of English 291'4 422-
Villamil, O S and M C M. de Guerrero
50 1996 'Peer revision in the L2 classroom
Rothschild, D andF Klingenberg 1990 'Self Social-cognitive activities, mediating strat-
and peer evaluation of writing in the egies, and aspects of social behavior ' Journal
interactive ESL classroom An exploratory of Second Language Writing 5/1 51-75
study ' TESL Canada Journal 8 52-65 Vygotsky, L S 1978 Mind m Society The
Stanley, J 1992 'Coaching student writers to Development of Higher Psychological Processes
be more effective peer evaluators Journal of Cambridge, MA Harvard University Press
Second Language Writing 1/3 217-33 Vygotsky, L S 1986 Thought and Language (A
Swain, M (1998) 'Focus on form through Kozuhn, trans ) Cambridge, MA MIT Press
conscious reflection' in C Doughty and J (Original work published 1934)
Williams (eds) Focus on Form in ClassroomWertsch, J V 1985 Vygotsky and the Social
Second Language Acquisition Cambridge, MA Formation of Mind Cambridge, MA Harvard
Cambridge University Press University Press
Swain, M and S Lapkin 1995 'Problems in Yagelski, R P i 1995 "The role of classroom
output and the cognitive processes they context in the revision strategies of student
generate A step towards second language writers ' Research in the Teaching of English 29/
learning ' Applied Linguistics 16/3 371-91 2 216-38 •
Swain, M and S Lapkin {in press) 'Inter- Zhang, S 1995 'Reexamining the affective
action and second language learning Two advantage of peer feedback in the ESL
adolescent French immersion students work- writing class' Journal of Second Language
ing together ' Modem Language Journal 82/3 Writing 4/3 209-22