System 28 (2000) 229±245

www.elsevier.com/locate/system

Linguistic characteristics of ESL writing in

task-based e-mail activities
Yili Li *
School of Continuing Education, Hong Kong Baptist University, Kowloon Tong, Hong Kong

Received 15 June 1999; received in revised form 18 October 1999; accepted 30 November 1999

Abstract
This study investigated the ecacy of integrating task-based e-mail activities into a process-
oriented ESL writing class. In particular, it examined the linguistic characteristics of 132
pieces of e-mail writing by ESL students in tasks that di€ered in terms of purpose, audience
interaction and task structure. The analysis focused on the linguistic features of the students'
e-mail writing at di€erent levels, i.e. syntactic complexity, lexical complexity and grammatical
accuracy. Computerized text analysis programs were used to ensure internal consistency of
the linguistic analysis. Statistical analysis of the results using the repeated measures analysis
of variance and post hoc contrast tests showed signi®cant syntactic, lexical and grammatical
di€erences in the students' e-mail writing of the di€erent tasks. Speci®cally, in e-mail tasks
involving audience interaction, students tended to produce syntactically and lexically more
complex texts, and in tasks which allowed students self-selection of topics and content, stu-
dents also tended to use more complex sentences and richer and more diverse vocabulary.
However, an interesting trade-o€ e€ect was observed between linguistic complexity and
grammatical accuracy in the students' e-mail writing, indicating the complexity of the second
language writing process. The study provides pedagogical implications for designing e€ective
e-mail tasks for enhancing second language writing development. # 2000 Elsevier Science
Ltd. All rights reserved.
Keywords: Computer-assisted language learning; Computer-mediated communication; English as a second
language; Electronic mail; Second language writing; Writing assessment

* Tel.: +852-2339-5682; fax: +852-2339-5444.

E-mail address: yili9098@hkbu.edu.hk (Y. Li).

0346-251X/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved.
PII: S0346-251X(00)00009-9
230 Y. Li / System 28 (2000) 229±245

1. Introduction

The advancement of modern technology in recent years has brought about inno-
vative use of computers in second language learning and teaching, especially in the
area of writing instruction (Pennington, 1996; Boswood, 1997; Levy, 1997; War-
schauer and Healey, 1998). Although the word processor is still the primary tool for
computer-aided writing, computer-mediated communication realized through dif-
ferent networking tools such as e-mail, synchronous computer conferencing and the
World Wide Web has been introduced into second/foreign language instruction.
Obviously, computer-mediated communication has changed the language learning
environment as well as the dynamics of the language classroom. Language learn-
ing that takes place with the aid of computer networks has led to more social,
interactive, collaborative, communicative and student-centered classrooms. Besides,
computer networking has expanded the opportunity for students to learn and use
the target language naturally and with communicative purposes (Pennington, 1996;
Warschauer, 1996a; Beauvois, 1998).
Computer-mediated communication demonstrates a number of advantages for
enhancing language learning. One distinctive aspect of computer-mediated com-
munication lies in the interactive learning environment it provides to promote
interactive language learning and the opportunity for authentic use of the target
language (Chun, 1994). Such interactive language learning experience is in con-
gruence with the discourse-based second language acquisition theories which
emphasize the role communicative interaction plays in second language learning
(Long and Porter, 1985; Pica, 1987; Pica et al., 1987). In addition, computer-mediated
communication encourages collaborative writing in the second language classroom,
which in turn enhances second language writing development through increasing
engagement, con®dence and responsibility on the part of the second language
learners (Strasma and Foster, 1992). Computer-mediated communication also
serves to foster student empowerment, i.e. students' increased control of the content
and process of their own learning (Warschauer et al., 1994). This sense of con-
trol and mastery, which is termed ``learner autonomy'' by second language educa-
tors, is viewed as especially important for language learning (Wenden and Rubin,
1987). Furthermore, computer-mediated communication attests to the importance
of social and a€ective factors in second language learning by promoting student
motivation and interest in functional use of the target language, and providing
students with a less threatening means to communicate (Barson et al., 1993; Beau-
vois, 1995, 1998; Beauvois and Eledge, 1996; Warschauer, 1996b; Skinner and
Austin, 1999).
Among the various forms of computer-mediated communication in language
instruction, e-mail has so far been the most popular (Levy, 1997). As a way of
sending a message from one computer to one or more computers around the world,
e-mail provides an easy and fast means of communication for language teachers and
students. It helps students to maintain constant contact with the teacher and receive
immediate feedback from the teacher outside the classroom (Wang, 1993). Writing
via e-mail also creates an authentic purpose and audience for writing, which are
 Y. Li / System 28 (2000) 229±245 231

elements often lacking in the writing assignments in traditional writing classes. In

addition, e-mail enables students to engage in on-line collaborative writing activities
such as planning, drafting, revising and editing, thus enhancing the social process of
writing. Since e-mail enriches the opportunities for learning and writing practice
beyond the time and space restraints of the traditional writing classroom, students
have more time to write at their own pace. Consequently, they can take time to
process linguistic input and produce more elaborated and complex language than
electronic discussions taking place during synchronous computer conferencing ses-
sions. The convenient access to the e-mail system also makes it easier for teachers to
integrate e-mail activities into a semester-long course and make such activities part
of the curriculum (Li, 1998; Liaw, 1998).
Taking advantage of the e-mail technology, language teachers around the world
have implemented various e-mail projects involving students of di€erent languages
and cultures (Vilmi, 1995; Warschauer, 1995a). There have also been reports of
innovative use of e-mail in a single class setting. For example, e-mail was used as a
writing tool for second language learners to engage in ``dialogue journals'' with their
teachers (Wang, 1993). In the forms of class mailing lists, bulletin boards, class
newsgroups, e-mail can also be used as a means for on-line out-of-class discussion
and collaborative writing (Warschauer, 1995a, b).
Despite the increasingly popular use of e-mail for language teaching and learning,
systematic investigation into the impact of this new medium of communication and
instruction on language learning and teaching is scarce. Among the few empirical
endeavors, Sotillo (1997) reported a study on using e-mail for native speakers to
provide corrective feedback to ESL students on their writing. The study found that
the ESL students bene®ted from di€erent types of corrective feedback o€ered by the
native speakers, and the successful learners were able to incorporate more than 70%
of the native speakers' corrective feedback into their revised written work. Leppanen
and Kalaja (1995) conducted a quasi-experimental study in which students' peer
feedback to their written work given via e-mail outside the class was compared with
a tutor's feedback given with pen on paper. The results showed that the students
gave each other a great deal of feedback of di€erent kinds and demonstrated a great
sense of responsibility. In a collaborative manner, the students generated more
comments on each other's writing than the tutor could provide. The authors exam-
ined the interaction patterns students demonstrated via e-mail writing and con-
cluded that the e-mail system is ``a ¯exible means of exchanging ideas and
commenting in the written form, and consequently, of transforming writing into a
social and interactive activity'' (Leppanen and Kalaja, 1995, p. 35). In an English-as-
a-foreign-language setting, Liaw (1998) investigated the ecacy of integrating e-mail
writing into EFL classrooms to facilitate authentic target language communication
among second language students. The study found that the e-mail activities promoted
students' motivation, computer skills, social interaction, reading skills, and com-
municative skills in the written form. Barson et al. (1993) reported another study on
using e-mail in the foreign language classroom with English-speaking students
enrolled in intermediate French classes at three universities. A functional analysis of
the e-mail messages generated by students in the process of accomplishing the
232 Y. Li / System 28 (2000) 229±245

semester-long task revealed that a wide range of speech acts were performed by the
students in a foreign language, indicating e€orts made in the process of negotiation
of meaning with the acquisition of certain discourse strategies. Kern (1996) described
a content-based e-mail project involving exchanges between a group of English-
speaking French students at an American university and a group of high school
students in France. Still in its on-going process when the study was reported, the
project showed promise of promoting language learning and cultural awareness
through international communication.

2. Purpose of the study

It is worth noting that the few reported studies on the use of e-mail for second/
foreign language learning have focused on the social, communicative and moti-
vating aspects of this new technology when applied to the language classroom.
These studies surveyed students' a€ective reactions to the e-mail activities
(Liaw, 1998), examined the kinds of feedback students gave to each other through
functional analysis of the messages (Sottilo, 1997) and measured the amount
of student participation by calculating the number of e-mail exchanges (Leppanen
and Kalaja, 1995). So far, almost no attempts have been made to examine the
quality and characteristics of students' e-mail writing. Nor have previous studies
attempted to analyze e-mail writing under di€erent task conditions. As such,
although the bene®cial aspects of e-mail for second/foreign language learning
are apparent, little has been known about the characteristics of students' e-mail
writing. Neither has there been information on whether there is variation in stu-
dents' writing when di€erent kinds of tasks are assigned. It is argued that in order
to better understand the impact of e-mail technology on second language learn-
ing so as to e€ectively integrate such technology into the language curriculum, it
is important to look into the characteristics and qualities of students' e-mail writ-
ing, and examine the relationship between the characteristics of students' e-mail
writing and speci®c features of the task design. To render empirical support to this
argument, the present study uses data collected from a semester-long study to
investigate the ecacy of integrating task-based e-mail activities into an ESL writ-
ing curriculum. In particular, the study focuses on examining the relationship
between the task characteristics of writing purpose, audience interaction, task
structure and the linguistic characteristics of ESL students' writing in task-based
e-mail activities. The following research questions are speci®cally addressed in
this study:

1. Are there any signi®cant di€erences between ESL students' writing in e-mail
tasks with di€erent rhetorical purposes?
2. Are there any signi®cant di€erences between ESL students' writing in e-mail
tasks with and without audience interaction?
3. Are there any signi®cant di€erences between ESL students' writing in struc-
tured and non-structured e-mail tasks?
 Y. Li / System 28 (2000) 229±245 233

3. Methodology

3.1. Subjects

Twenty-two ESL students (9 female and 13 male) enrolled in a freshman com-

position course at an American university participated in this study. The students
came from 11 ®rst language backgrounds, and were enrolled in di€erent academic
programs at the university. A pre-activity questionnaire was administered during
the ®rst week of the class to gather information about the students' backgrounds.
Results from the questionnaire showed that all 1of the students had experienced
formal instruction in English for an average of 6 2 years by the time they came to
study in the USA. Since the majority of the students came to the USA shortly
before the semester began, they had very limited experience in the USA and little
exposure to the academic environment at an American university at the outset of
the study. In a self-assessing manner, most of the students ranked their writing
ability in English as fair (63%) and their typing ability as good (68%). A few stu-
dents thought their writing ability in English was poor (23%) and a few thought the
same about their typing ability (18%). All of the students reported having used
word processing before, and most of them had experienced using e-mail (90%) and
the World Wide Web (77%). It should be noted that these students' previous
experience with e-mail writing had been restricted largely to informal writing for
personal communication purposes. When they were instructed to subscribe to the
class mailing list in the ®rst week of class, they admitted that it was their ®rst
experience of joining an electronic discussion list for academic purposes. Overall, it
seems that with just a few exceptions, this group of ESL students had a certain level
of computer skills and knowledge, so requiring them to write via e-mail did not
pose a big technical problem for most of them at the outset of the semester-long
project.

3.2. The writing course

A process-oriented approach to writing instruction was implemented in the com-

position course at the institution where the present study was conducted. The
objective of the composition course was to help ESL students develop academic
writing skills needed for success in their discipline studies at the university. In the
composition course, students were engaged in a variety of classroom activities rela-
ted to the process of composing, from planning, pre-writing, drafting to revising
(Zamel, 1983; Silva, 1993). The students wrote multiple drafts, received feedback on
their ®rst drafts and went through a revision process before they turned in their ®nal
papers. The course emphasized the social aspect of writing. The students were
encouraged to share their ideas and writings in di€erent ways, e.g. in-class discus-
sions, group workshop on drafts, peer review, journal exchanges, and oral pres-
entations. To facilitate such a sharing process, a class mailing list was set up in the
®rst week of class to enable students to post their writings and circulate their written
works among class members.
234 Y. Li / System 28 (2000) 229±245

3.3. The writing tasks

Four e-mail tasks were designed as part of the writing assignments students were
required to complete in the composition course. Such writing assignments
were intended to help students practise di€erent writing skills relevant to academic
essays. Each of the tasks was actually related to the purpose and skills required of a
formal academic essay the students wrote in the course. For the research objective,
the four tasks were carefully constructed so that they were di€erent in terms of
rhetorical purpose, audience interaction and task structure. The students were
required to use e-mail to complete the writing tasks outside class and posted them on
the class mailing list so they could read each other's writing on-line. The students
had a week to complete each task, and they were able to take time to do it within
that week at their own pace.

3.3.1. Writing tasks of di€erent rhetorical purposes

Each of the four e-mail tasks required students to write for a particular purpose,
each in a di€erent mode of expression and related to an academic essay students wrote
in the composition course. Task 1 asked students to tell a personal story about one of
their ®rst experiences in the USA. It was related to Essay 1, a personal narrative essay
in which students narrated a signi®cant personal experience and showed the reader
how that experience a€ected them. Task 2 required students to exchange cultural
information on a topic of their own. It was related to Essay 2, a cultural comparison/
contrast essay in which students compared or/and contrasted on a cultural topic
between their own country and that of another class member. Task 3 invited students
to perform role play of either the proponent or the opponent of a controversial issue.
It was related to Essay 3, a persuasive essay in which students took their own stand on
a particular issue and developed an argument of their own. Task 4 asked students to
write about their reactions to an article. It was related to the ®nal exam, an in-class
timed essay based on questions derived from an assigned article. Students were asked
to express their personal opinions on the author's viewpoints and react to the article.

3.3.2. Writing tasks with and without audience interaction

Taking into account the importance of audience interaction for writing, the
researcher aimed to study the relationship between audience interaction and
the characteristics of students' e-mail writing. For this purpose, the four e-mail tasks
described in the previous section were grouped into two categories according to the
presence and absence of interaction between the student writer and the audience,
who were the class members. Tasks 2 and 3 fell into the ®rst category, which
involved interactive feedback from a speci®c peer audience and exchanges between
the student writer and the peer audience. In this group of tasks, students exchanged
cultural information with their peers (Task 2), or engaged in on-line debates over a
controversial issue with each other (Task 3). While completing these tasks, the stu-
dents were engaged in active exchange of ideas and information with a speci®c
audience who had mutual interests in the topics of writing. Take Task 3 for an
example; in this task, students were asked to send an initial message to the class
 Y. Li / System 28 (2000) 229±245 235

mailing list stating an issue he/she had a strong opinion about. Then they were asked
to pair up with each other and perform the role of either the proponent or the
opponent of a controversial issue they were most interested in. The students ®rst put
forward their opinions on the issue, stating their arguments and citing examples and
evidence to support their arguments. Then they were engaged in on-line debates with
their partners trying to convince their partners of their stance on the issue. It was the
active interaction between the student writer and the audience that distinguished this
group of tasks from the other (i.e. Tasks 1 and 4), which did not involve interactive
feedback from the peer audience and there was a lack of exchanges between the
student writer and the peer audience. In this second group of tasks, the students
shared personal stories with peers (Task 1), or expressed their personal opinions in
response to an article (Task 4), but they did not engage in exchanges with the audi-
ence, who were primarily silent readers.

3.3.3. Writing tasks of di€erent task structure

In an attempt to examine the relationship between task structure and characteristics
of the students' e-mail writing, a distinction was made between structured and non-
structured tasks according to the degree of speci®cation in the task assignment.
Among the four e-mail tasks, Tasks 1 and 3 were assigned to the students in a struc-
tured manner. In these tasks, speci®c questions were given to the students who wrote
in response to such questions rather than on self-selected topics of their own. For
example, in Task 1, the students were asked to write about their ®rst experiences in the
USA. They were given speci®c questions such as: ``What happened? Where and when
did it happen? Who was involved? What surprised you? What caused that surprise?
How did you feel about this experience? What did you learn from this experience?'' In
response to the assignment for Task 1, students had to keep all these questions in
mind and supply details related to these questions in their narrative. Due to the
structured manner in which these tasks were assigned to the students, the content and
details of their writing were relatively restricted to what was speci®ed in the writing
prompts. Tasks 2 and 4 were assigned to the students in an unstructured manner.
General guidelines were provided to the students conveying the instructor's expecta-
tions, but the students had the choice of topic selection and freedom to express their
ideas on their chosen topic. For example, in Task 2, a student was paired up with
another student from a di€erent country. The two students had a chance to discuss
orally in class what topic they were interested in knowing about each other's culture.
Upon deciding on the topic, the students started the e-mail exchanges after class,
trying to ®nd out the similarities and di€erences between the two cultures. They sent
out the ®rst message stating their interests and what they wanted to know about the
other culture, asking speci®c questions to elicit detailed information on their chosen
topics, then they responded to each other's questions, supplying the needed informa-
tion, making comparison and contrasts. The topics the students chose in Task 2 var-
ied from college sports, educational systems, family structure to wedding practices.
Thus, the structured and the non-structured tasks were supposed to allow students
di€erent degrees of freedom in terms of topic selection and content of writing in
responding to the writing assignments they completed via e-mail.
236 Y. Li / System 28 (2000) 229±245

3.4. The writing samples

The students' writing in response to the four e-mail assignments was collected
electronically for data analysis. The entire data base consisted of a total of 132
pieces of writing collected over the course of a semester from four e-mail tasks,
among which two tasks required exchanges between the partners, thus making a
total of six writing samples from each student. Although a minimum length of 350
words for each assignment was required, the students tended to write more than
required. It turned out that the average length for each written text was about 450
words, and the entire data base amounted to a total of 59,372 running words.

3.5. Analytical procedures

In this study, students' e-mail writing was analyzed through objective measure-
ments performed by computerized text analysis programs. The analysis focused on
the linguistic features of the written texts at the level of syntactic complexity, lexical
complexity and grammatical accuracy, which operationally characterized the ESL
students' writing performance on the e-mail tasks. Two measures were used to assess
the written texts at each of these linguistic levels. Syntactic complexity was assessed
by: (1) average sentence length, which was obtained by calculating the average
number of words per sentence in a given text; and (2) the ratio of subordinated
structures, which was measured by calculating the ratio of the number of sub-
ordinated structures to the combination of subordinated structures and coordinated
structures in a piece of writing. Two measures were used to assess lexical complexity:
(1) lexical diversity, which was calculated by having the number of di€erent words
including both content and function words divided by the total number of words in
a piece of writing; and (2) lexical density, which was calculated by having the num-
ber of lexical items excluding function words divided by the total number of words
in a piece of writing.1 There were two ways to tap into grammatical accuracy: (1) the
ratio of number of grammatical errors to the total number of sentences in a piece of
writing; and (2) the ratio of types of grammatical errors to the total number of sen-
tences in a piece of writing.2

1
The formula for calculating lexical diversity and lexical density was adapted from Laufer and Nation
(1995) about measuring vocabulary size and use in second language written production. The calculation
method is as follows:
Lexical diversity ˆ number of different lexical and functional words …types†  100=total number of tokens;
Lexical density ˆ number of different lexical words  100=total number of tokens:

Since the computer program used for data analysis computed the type/token ratio based on consecutive n-
word chunks of texts (speci®ed as 100 in the present analysis), the texts of unequal length were controlled
for.
2
In the analysis of grammatical accuracy, the ratios were calculated in order to control for texts of
unequal length in the writing samples. The calculation method is as follows:
Number of grammatical errors ˆ number of grammatical errors=total number of sentences;
Types of grammatical errors ˆ number of types of grammatical erros=total number of sentences:
 Y. Li / System 28 (2000) 229±245 237

Two computerized text analysis programs were used to perform the linguistic
analysis, i.e. ``Wordsmith Tools'' (Scott, 1996) and the Grammatik program inte-
grated into Word Perfect 7.0. Wordsmith Tools is an integrated package of text
analysis programs designed to examine how words behave in texts. In the present
study, two major programs of Wordsmith Tools were used, i.e. ``Wordlist'' and
``Concord''. Wordlist was used for the analysis of syntactic complexity in terms of
average sentence length since the statistic function of this program provides ready
information on the average sentence length of a given text. The Concord program
was used to search for the occurrences of subordinated and coordinated structures
for the calculation of the ratio of subordination used in a given text as a secondary
measure of syntactic complexity. Wordlist was used for the lexical analysis in terms
of lexical diversity and lexical density because this program provides statistics of
type/token ratio.3
For the analysis of grammatical accuracy, ``Grammatik'' in WordPerfect 7.0 was
used. Since Grammatik ¯ags not just grammatical errors but also improper usage
related to style and mechanics, e.g. colloquialism, capitalisation, number style, etc.,
customisation of its checking style and rules was required before the program was
put into use. First, among the default options of checking styles, student composition
was chosen because the texts to be analyzed were student writing samples. Then the
rules for checking were scrutinised and those not relevant to grammar checking were
removed, leaving only rules that really apply to the grammatical categories of
the English language. This customised checking style was then speci®ed as the
working checker for the purpose of the present analysis.

4. Results and discussions

4.1. Purpose and characteristics of writing

To examine the relationship between the purpose of writing and the characteristics
of e-mail writing, students' written production on the e-mail tasks was analyzed,
using the repeated measures analysis of variance (ANOVA) procedure. The results,
as presented in Table 1, show signi®cant di€erences across e-mail tasks of di€erent
writing purposes at the probability level of P<0.01 in several variables, i.e. syntactic
complexity in terms of ratio of subordination, lexical complexity in terms of both
lexical diversity and lexical density and grammatical accuracy in terms of number
and types of grammatical errors. Such observed di€erences in the students' writing
across the four tasks con®rm the ®ndings of previous studies in both ®rst and second
language writing, which have demonstrated that variation in writing is related to

3
Wordlist contains a stop list function. When a list of words is created and used as a stop list, the
words contained in this list will not be counted when the program calculates the type/token ratio. Making
use of this program function, a list of English function words was identi®ed and entered as the stop list to
yield the index of lexical density.
238 Y. Li / System 28 (2000) 229±245

Table 1
Di€erences in the linguistic measurements of e-mail writing of tasks of di€erent purposesa

Variable Task 1 Task 2 Task 3 Task 4 F-test P value

Narrative Informative Persuasive Expressive

Sentence length 16.18 (6.39) 17.10 (3.15) 17.43 (3.73) 17.65 (4.63) 0.82 0.488
Ratio of subordination 0.606 (0.1015) 0.610 (0.096) 0.708 (0.081) 0.713 (0.137) 8.25 0.001*
Lexical diversity 76.33 (2.81) 74.28 (2.77) 71.13 (2.74) 77.34 (2.58) 8.25 0.001*
Lexical density 82.78 (3.86) 77.05 (6.15) 78.62 (4.36) 81.12 (4.79) 8.19 0.001*
Number of grammatical 0.381 (0.247) 0.489 (0.2196) 0.601 (0.327) 0.614 (0.376) 5.81 0.001*
errors
Types of grammatical 0.218 (0.1024) 0.305 (0.1100) 0.291 (0.010) 0.330 (0.168) 5.73 0.002*
errors
a
S.D. are in parentheses.
*Signi®cant at P<0.01.

varying rhetorical purposes which, in turn, elicit di€erent modes of writing (Prater
and Padia, 1983; Koda, 1993; Foster and Skehan, 1996).
To probe further into the linguistic di€erences in students' e-mail writing in tasks
of di€erent purposes, a series of post hoc pairwise contrast tests were performed
to compare the mean di€erences in the linguistic measures for each pair of tasks
with di€erent writing purposes in order to ®nd out which pairs of means are sig-
ni®cantly di€erent. The Tukey honestly signi®cant di€erence (HSD) test was chosen
to run the contrast analyses with the overall alpha level set at 0.05.
The signi®cant results of the post hoc analysis as summarized in Table 2 indicate
that the most di€erences were found between narrative writing (Task 1) and per-
suasive writing (Task 3), with signi®cant results at the probability level of P<0.05
observed in several variables, i.e. ratio of subordination, lexical diversity, lexical
density and number of grammatical errors. Such ®ndings seem to be consistent with
previous composition studies which have found expository and argumentative writ-
ing to be more demanding and dicult than narrative and descriptive writing (Per-
ron, 1977). Perhaps the most striking result comes from the measurement of
grammatical accuracy. The data suggests that the students' writing in the e-mail
tasks di€ered the most signi®cantly in terms of the number of grammatical errors

Table 2
Post hoc pairwise contrast tests with signi®cant results (t value)

Variables Task 1 Task 1 Task 1 Task 2 Task 2 Task 3

vs. vs. vs. vs. vs. vs.
Task 2 Task 3 Task 4 Task 3 Task 4 Task 4

Ratio of subordination 3.41* 3.58* 3.25*

Lexical diversity 7.60* 4.60* 4.43* 9.03*
Lexical density 5.55* 4.01* 3.97*
Number of grammatical errors ÿ3.64* ÿ7.30* ÿ7.63* ÿ3.64* 3.98*
Types of grammatical errors ÿ3.72*

*Signi®cant at P<0.05. Blanks indicate non-signi®cant results.

 Y. Li / System 28 (2000) 229±245 239

the writing contained. As indicated in Table 2, among the six pairwise comparisons,
®ve of them show signi®cant di€erences in the variable of number of grammatical
errors. It is interesting to note that while the students tended to use more complex
sentence structures in persuasive writing (ratio of subordination=0.71) than in nar-
rative writing (ratio of subordination=0.61), they were also inclined to make a
greater number of grammatical errors in persuasive writing (ratio of number of
grammatical errors to total number of sentences=0.60) than in narrative writing
(ratio of number of grammatical errors to total number of sentences=0.38). This
®nding points to the importance of the trade-o€ e€ects between complexity and
accuracy in measuring performance on second language learning (Foster and Ske-
han, 1996). Foster and Skehan (1996) maintained that complexity in second lan-
guage learners' use of language indicates a greater willingness to experiment and to
take risks, whereas accuracy re¯ects a focus on form to achieve freedom from errors.
They argue that complexity and accuracy are two competing goals for second lan-
guage learners. Operating under information-processing constraints, second
language learners have to allocate attention to particular goals at the expense of
other goals. In the present study, such a trade-o€ e€ect between complexity and
accuracy can be observed in the students' writing in the two most cognitively dis-
tinctive writing tasks, i.e. narrative and persuasive. It explains why the students
performed with a higher level of syntactic and lexical complexity in the persuasive
writing task than in the narrative writing task while they also tended to make more
grammatical errors in the narrative mode of writing.

4.2. Audience interaction and characteristics of writing

To examine the relationship between audience interaction and writing, the mean
measures of the linguistic features being examined were compared between the stu-
dents' writing in the e-mail tasks with and without audience interaction. As indi-
cated in Table 3, signi®cant di€erences at the probability level of P<0.05 were
found between the tasks with audience interaction and those without audience
interaction in the means of several measures: i.e. ratio of subordination, lexical

Table 3
Di€erences in the linguistic measurements of e-mail writing of tasks with and without audience
interactiona

Variables Tasks with Tasks without t-test P value

audience interaction audience interaction

Sentence length 17.27 (3.42) 16.92 (5.56) ÿ0.487 0.631

Ratio of subordination 0.659 (0.101) 0.660 (0.131) 0.993 0.009*
Lexical diversity 76.84 (2.72) 75.71 (3.08) 8.48 0.001*
Lexical density 81.93 (4.37) 77.84 (5.33) 5.64 0.001*
Number of grammatical errors 0.545 (0.28) 0.498 (0.336) ÿ2.35 0.029*
Types of grammatical errors 0.298 (0.104) 0.274 (0.149) ÿ1.14 0.269
a
S.D. are in parentheses.
*Signi®cant at P<0.05.
240 Y. Li / System 28 (2000) 229±245

diversity, lexical density and number of grammatical errors. Such results suggest
variation in students' writing in e-mail tasks with and without audience interaction
at di€erent linguistic levels. In addition, analysis of the descriptive statistics show
some consistent results: (1) there was a consistently higher level of lexical complexity
in terms of both lexical diversity and lexical density in students' writing in the tasks
with audience interaction (lexical diversity=76.84, lexical density=81.93) than in
the tasks without audience interaction (lexical diversity=75.71, lexical density
=77.84); (2) despite the lack of statistically signi®cant di€erences in the means, the
average sentence length in the texts produced in the tasks with audience interaction
(mean=17.27) was longer than in that of tasks without audience interac-
tion (mean=16.92); and (3) a higher frequency of occurrence of errors was found in
the tasks with audience interaction (ratio of number of grammatical errors to the
total number of sentences=0.545) than in those without audience interaction (ratio
of number of grammatical errors to the total number of sentences=0.498).
The ®ndings summarized above indicate a higher level of linguistic complexity
syntactically and lexically when the students were engaged in active interaction with
their peer audience in e-mail writing. Such ®ndings have gained support from pre-
vious research which found a positive e€ect of collaboration through peer interac-
tion on networked computers (Chun, 1994; Liaw, 1998). The ®ndings from this
study further indicate that interaction with peer audiences on networked computers
can help students produce more sophisticated written language. However, it is also
interesting to note that while the interaction with a peer audience seemed to lead to
more complex sentence structures and richer and more diverse vocabulary, there
was, however, a higher frequency of grammatical errors in the texts. The data
reveals that the students tended to make more grammatical errors when they
exchanged feedback and information, or argued with peers interactively than when
they narrated a personal story, or expressed their personal opinions without inter-
acting with peers. It seemed that interaction with a peer audience helped the students
to generate more complex and sophisticated use of written language but, at the same
time, the grammatical accuracy of their language su€ered.

4.3. Task structure and characteristics of writing

Another follow-up analysis was conducted to examine the relationship between

task structure and characteristics of writing by comparing the means of the linguistic
features being examined between the students' e-mail writing in structured and non-
structured tasks. Table 4 summarizes the results of this analysis. Signi®cant di€er-
ences at the probability level of P<0.05 were observed in the means of several
dependent variables used to analyze the students' e-mail writing: i.e. ratio of sub-
ordination, lexical diversity, lexical density, number of grammatical errors, types of
grammatical errors. In addition, analysis of the descriptive statistics show some con-
sistent results: (1) there was a higher level of lexical complexity in student writing in
the non-structured tasks (lexical diversity=76.73, lexical density=80.69) than in the
structured tasks (lexical diversity=75.81, lexical density=79.09); and (2) the students
tended to produce longer and more complex sentences in the non-structured tasks
 Y. Li / System 28 (2000) 229±245 241

Table 4
Di€erences in the linguistic measurements of e-mail writing of structured and non-structured tasksa

Variables Structured tasks Non-structured t-test P value

tasks

Sentence length 16.81 (3.21) 17.38 (3.93) ÿ0.808 0.428

Ratio of subordination 0.657 (0.105) 0.662 (0.128) 4.83 0.001*
Lexical diversity 75.81 (3.06) 76.73 (2.77) 4.29 0.001*
Lexical density 79.09 (5.82) 80.69 (4.57) 2.21 0.038*
Number of grammatical errors 0.491 (0.307) 0.557 (0.312) ÿ7.98 0.001*
Types of grammatical errors 0.255 (0.107) 0.317 (0.141) ÿ2.95 0.008*
a
S.D. are in parentheses.
*Signi®cant at P<0.05.

(sentence length=17.38, ratio of subordination=0.662) than in the structured

tasks (sentence length=16.81, ratio of subordination=0.657), though the sig-
ni®cance testing of the variable of average sentence length did not yield a signi®cant
result. Similar to the ®ndings discussed above, a trade-o€ e€ect between linguistic
complexity and grammatical accuracy was observed in students' writing between the
structured and the non-structured tasks. Such ®ndings suggest that when the writing
task allowed the students to make choices of topic selection and more freedom to
express their ideas on their chosen topic, they tended to use more complex sentences
and richer and more diverse vocabulary than when the task required them to write
on speci®c topics and respond to speci®c questions. In other words, the students
were able to produce syntactically and lexically more complex written language in a
task condition under which they had more freedom of decision making about what
they wanted to write than in a task condition under which they were restricted in
the content of their writing. This ®nding gains support from recent claims about the
empowerment of students learning with computer networks. In second language
literature, there have been strong claims about the importance of student autonomy
for the enhancement of second language learning. It is believed that when students
assume more responsibilities and more control of their own learning, they will be
able to learn faster and better (Wenden and Rubin, 1987; O'Malley and Chamot,
1990; Oxford, 1990). Such claims can be used to explain the ®nding of more
sophisticated use of language in the non-structured e-mail tasks in the present study.
In the non-structured e-mail tasks, the students were given more freedom to decide
what they wanted to write about, and there were fewer restrictions on the details and
amount of information they could include in their writing than in the structured
tasks. Thus, it is natural to observe that as the students gained a greater control of
their language use in the non-structured e-mail tasks, their writing displayed more
sophisticated and more diverse use of written language.
However, it is crucial to point out that there seemed to be always trade-o€ e€ects
between linguistic complexity and grammatical accuracy in the students' writing,
which could be consistently observed from the data of this study. Perhaps such
e€ects are more obvious when the students are writing via e-mail rather than writing
242 Y. Li / System 28 (2000) 229±245

in the traditional way because of the more instant nature of e-mail writing. Since no
previous studies have been conducted to compare students' writing in the networked
writing environment versus that in the traditional writing condition in terms of both
linguistic complexity and grammatical accuracy, there is a need for empirical sup-
port for the observation made in the present study through further investigation.

5. Conclusions and pedagogical implications

To conclude, the ®ndings of this study indicate that there were syntactic, lexical
and grammatical di€erences in ESL students' writing in e-mail tasks of di€erent
purposes, and there was also variation between e-mail tasks with and without audi-
ence interaction, and between structured and non-structured e-mail tasks. In par-
ticular, in e-mail tasks in which an interactive audience was present, the students
tended to produce texts that were linguistically more complex. In addition, the
students wrote with a higher level of syntactic and lexical complexity in the non-
structured e-mail tasks than in the structured ones, indicating more sophisticated use
of language when the student writers were given more freedom and control of the
learning activities.
The ®ndings bear important implications for designing e€ective task-based e-mail
activities for enhancing second language writing development. First of all, in order
to make use of e-mail to enhance second language writing instruction, it is import-
ant to design e€ective e-mail writing tasks that are not only interesting, but also
meaningful and relevant to the objectives and content of the writing course. In
integrating e-mail activities into second language writing, teachers should take full
advantage of the on-line communication channel provided by computer networks to
stimulate interaction among the students, foster communication and encourage col-
laborative writing. For example, e-mail assignments which require the exchange of
information, ideas or debates on controversial issues will generate a great deal on
written interaction among the students in the process of completing the assignments.
In addition, teachers involved in applying computer-mediated communication to
second language writing instruction should assume appropriate, multiple roles in
di€erent phases of the project from planning and implementation to evaluation.
Besides designing e€ective learning tasks and constructing appropriate writing
prompts, teachers should also be responsible for monitoring the students' perfor-
mance on the tasks and ensuring that the tasks are completed in ways such that their
objectives and goals are met. The obvious trade-o€ e€ects between linguistic com-
plexity and accuracy in students' e-mail writing is revealing. The fact that while
students produced linguistically more complex texts they also paid less attention to
grammatical accuracy re¯ects to a great extent the complexity of writing in a second
language. Such trade-o€ e€ects may be attributed to the information-processing
constraints of writing in a second language, which require learners to allocate
attention to particular goals at the expense of other goals. There is therefore the
need to balance the development of both linguistic complexity and accuracy in sec-
ond language writing.
 Y. Li / System 28 (2000) 229±245 243

Although the ®ndings of the study are largely positive, several limitations should
be noted here. First, the subjects of this study were 22 ESL students enrolled in a
freshman ESL writing class at an American university. Thus, the ®ndings are limited
to subjects with a pro®le similar to those participating in this project. In interpreting
the results, we should also bear in mind that the subjects were from 11 ®rst language
backgrounds, and their previous academic backgrounds might be varied, which
might have a€ected their writing output. In addition, the e-mail writing which con-
stituted the data for analysis in this study is an informal kind of writing which
served as pre-writing of the students' academic written work. Since no information
was gathered about students' ®nal written products, i.e. the formal academic essays
related to the e-mail writing tasks, caution must be taken to avoid making claims
about the relationship between such e-mail writing tasks and students' achievements
in traditional academic essay writing.
Since the analysis in the present study has focused on the linguistic features
of students' e-mail writing, future investigation should also take into account
discourse-level written features such as coherence, development of main ideas and
organization so as to have a more comprehensive understanding of the character-
istics of second language writing. Qualitative analysis of students' writing processes
and the strategies they use while composing via e-mail will enable us to understand
better in what ways computer technology can serve to enhance second language
writing development. Consideration of individual learner di€erences such as moti-
vation, attitude, gender, learning style and how such variables may a€ect the use of
e-mail strategies in the electronic composing process could lead to a future research
project that could also add substantially to the literature on computer-assisted lan-
guage learning.

Acknowledgements

I am grateful to Professor Robert Ariew for his helpful comments and constant
encouragement and support throughout this study. I also wish to thank the anon-
ymous reviewer for thoughtful comments and helpful suggestions on earlier versions
of this paper.

References

Barson, J., Frommer, J., Schwartz, M., 1993. Foreign language learning using e-mail in a task-oriented
perspective: interuniversity experiments in communication and collaboration. Journal of Education and
Technology 4, 565±584.
Beauvois, M.H., 1995. E-talk: attitudes and motivation in computer-assisted classroom discussion. Com-
puters and the Humanities 28, 177±190.
Beauvois, M.H., 1995. Conversation in slow motion: computer-mediated communication in the foreign
language classroom. The Canadian Modern Language Review 54, 189±237.
Beauvois, M.H., Eledge, J., 1996. Personality types and megabytes: student attitudes toward computer-
mediated communication in the language classroom. CALICO Journal 13, 27±45.
Boswood, T. (Ed.), 1997. New Ways of Using Computers in Language Teaching. TESOL, Alexandria, VA.
244 Y. Li / System 28 (2000) 229±245

Chun, D., 1994. Using computer networking to facilitate the acquisition of interactive competence.
System 22, 17±31.
Foster, P., Skehan, P., 1996. The in¯uence of planning and task type on second language performance.
Studies in Second Language Acquisition 18, 299±323.
Kern, R.G., 1996. Computer-mediated communication: using E-mail exchanges to explore personal his-
tories in two cultures. In: Warschauer, M. (Ed.), Telecommunication in Foreign Language Learning:
Proceedings of the Hawaii Symposium (SLTCC Technical Report #12). University of Hawaii Second
Language Teaching and Curriculum Center, Honolulu, HI, pp. 105±120.
Koda, K., 1993. Task-induced variability in FL composition: language speci®c perspectives. Foreign
Language Annals 26 (3), 332±346.
Laufer, B., Nation, P., 1995. Vocabulary size and use: lexical richness in L2 written production. Applied
Linguistics 16 (3), 307±322.
Leppanen, S., Kalaja, P., 1995. Experimenting with computer conferencing in English for academic pur-
poses. ELT Journal 49, 26±36.
Levy, M., 1997. Computer-Assisted Language Learning: Context and Conceptualization. Oxford Uni-
versity Press, New York.
Li, Y.L., 1998. Using Task-based E-mail Activities in Developing Academic Writing Skills in English as a
Second Language. Unpublished PhD thesis, The University of Arizona, AZ.
Liaw, M., 1998. Using electronic mail for English as a foreign language instruction. System 26, 335±352.
Long, M., Porter, P., 1985. Group work, interlanguage talk, and second language acquisition. TESOL
Quarterly 19, 207±228.
O'Malley, J., Chamot, A., 1990. Learning Strategies in Second Language Acquisition. Cambridge Uni-
versity Press, Cambridge.
Oxford, R., 1990. Language Learning Strategies: What Every Teacher Should Know. Newbury House,
Rowley, MA.
Pennington, M.C., 1996. The Power of CALL. Athelstan, Houston, TX.
Perron, J., 1977. Written Syntactic Complexity and the Modes of Discourse. Paper presented at the
Annual Meeting of the American Educational Research Association, New York.
Pica, T., 1987. Second-language acquisition, social interaction, and the classroom. Applied Linguistics 8,
3±21.
Pica, T., Young, R., Doughty, C., 1987. The impact of interaction on comprehension. TESOL Quarterly
21, 737±758.
Prater, D., Padia, W., 1983. E€ects of modes of discourse in writing performance in grades four and six.
Research in the Teaching of English 17, 127±134.
Scott, M., 1996. Wordsmith Tools. Oxford University Press, London.
Silva, T., 1993. Towards an understanding of the distinct nature of L2 writing: the ESL research and its
implications. TESOL Quarterly 27, 657±677.
Skinner, B., Austin, R., 1999. Computer conferencing: does it motivate EFL students? ELT Journal 53,
270±280.
Sotillo, S.M., 1997. E-mail Exchanges and Corrective Feedback Between Native Speakers and Non-native
Speakers of English. Paper presented at the Annual Conference of American Association for Applied
Linguistics, Orlando, FL.
Strasma, K., Foster, G., 1992. Collaboration within writing classes: an ethnographic point of view. The
Writing Instructor, Spring/summer, 111±127.
Vilmi, R., 1995. The individual writing exchange. In: Warschauer, M. (Ed.), Virtual Connections:
Online Activities and Projects for Networking Language Learners (SLTCC Technical Report #8).
University of Hawaii, Second Language Teaching and Curriculum Center, Honolulu, HI, pp.
208±210.
Wang, Y.M., 1993. E-mail Dialogue Journaling in an ESL Reading and Writing Classroom. Unpublished
PhD thesis, University of Oregon at Eugene, OR.
Warschauer, M., 1995a. Virtual Connections: Online Activities and Projects for Networking Language
Learners (SLTCC Technical Report #8). University of Hawaii, Second Language Teaching and Curri-
culum Center, Honolulu, HI.
 Y. Li / System 28 (2000) 229±245 245

Warschauer, M., 1995b. E-mail for English teaching. TESOL, Alexandria, VA.
Warschauer, M., 1996a. Telecommunication in Foreign Language Learning: Proceedings of the Hawaii
Symposium (SLTCC Technical Report #12). University of Hawaii Second Language Teaching and
Curriculum Center, Honolulu, HI.
Warschauer, M., 1996b. Motivational Aspects of Using Computers for Writing and Communication
(SLTCC Net Work #1). University of Hawaii, Second Language Teaching and Curriculum Center,
Honolulu, HI.
Warschauer, M., Healey, M., 1998. Computers and language learning: an overview. Foreign Language
Annals 31, 57±71.
Warschauer, M., Turbee, L., Roberts, B., 1994. Computer Learning Network and Student Empowerment.
(SLTCC Research Note #10). University of Hawaii, Second Language Teaching and Curriculum Cen-
ter, Honolulu, HI.
Wenden, A., Rubin, J. (Eds.), 1987. Learner Strategies in Language Learning. Prentice Hall, Englewood
Cli€s, NJ.
Zamel, V., 1983. The composing processes of advanced ESL students: six case studies. TESOL Quarterly
17, 165±187.