SYSTEM

System 33 (2005) 623636

www.elsevier.com/locate/system

Cohesive features in argumentative writing

produced by Chinese undergraduates
Meihua Liu

a,*

, George Braine

Department of Foreign Languages, Tsinghua University, Beijing 100084, PR China

Department of English, Fung King Hey Building, The Chinese University of Hong Kong, Shatin, N.T., Hong Kong
Received 9 January 2003; received in revised form 16 August 2004; accepted 7 February 2005

Abstract
Because cohesion is important both to the reader and the writer to create and comprehend a text,
teachers have placed much emphasis on text cohesion and coherence in their teaching and evaluation
of writing. Using Halliday and Hasans [Halliday, M.A.K., Hasan, R. (1976). Cohesion in English.
Longman, London] taxonomy of cohesive devices and their framework for analysis, this study investigated the use of cohesive devices in 50 argumentative compositions created by Chinese undergraduate
non-English majors. It was found that the students were able to use a variety of cohesive devices in their
writing, among which lexical devices formed the largest percentage of the total number of cohesive
devices, followed by references and conjunctives. The quality of writing was also revealed to signicantly
co-vary with the number of lexical devices and the total number of cohesive devices used. Apart from
that, certain problems were identied in the writing in terms of the use of reference and lexical devices.
2005 Elsevier Ltd. All rights reserved.
Keywords: Cohesive features; Argumentative writing; Chinese undergraduate non-English majors; Reference tie;
Conjunction tie; Lexical tie

1. Introduction
As one of the four basic language skills, writing is more complex in that it tests a persons ability to use a language and the ability to express ideas. As a result, a person needs
*

Corresponding author. Tel.: +86 10 82381986/13552111138.

E-mail addresses: ellenlmh@yahoo.com (M. Liu), georgebraine@cuhk.edu.hk (G. Braine).

0346-251X/$ - see front matter 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.system.2005.02.002

624

M. Liu, G. Braine / System 33 (2005) 623636

to write not only coherently but correctly, which requires much more time and skills. This
is especially so when writing in a second/foreign language. Although neglected for a long
time in Mainland China (Yang, 1994; Liu, 2000), writing in English as a foreign language
is becoming more and more important owing to the fact that even EFL learners have to
write papers and theses in English at secondary and tertiary levels, and send job application letters and applications to foreign universities, among other functions. As a result, in
China, eective writing appears to be particularly important in teaching and learning of
English as a foreign language. Cohesion and coherence, two important textual elements
(Halliday and Hasan, 1976; Halliday, 2000), have long been recognized as important features of good writing.
Research on cohesion and coherence in writing has been ourishing since the publication of Cohesion in English (Halliday and Hasan, 1976). Halliday and Hasan propose that
in any language, such grammatical and lexical devices as reference, ellipsis, substitution,
conjunction and lexical cohesion create texture the property of being a text. These devices form cohesive relations between sentences and elements in sentences, thus contributing to the coherence of the text.

2. Literature review
Halliday and Hasan (1976) have provided a framework for the study of cohesion and
coherence in ESL/EFL writing (Green et al., 2000; Jafarpur, 1991; Johns, 1980; Johnson,
1992; Yvette and Yip, 1992; Zhang, 2000). However, the ndings of these studies have
been somewhat contradictory. Some have found that there is no dierence in the deployment of cohesive devices in good and weak writing (Johnson, 1992; Zhang, 2000).
Others indicate that highly rated essays dier from low rated ones in the use of cohesive
devices (Jafarpur, 1991).
According to Johnson (1992), there was no dierence in the degree of cohesion between
good and weak compositions written in Malay by native speakers or in English by
native and Malay speakers. Zhangs (2000) study of cohesion in 107 expository compositions created by Chinese English majors generated similar ndings as Johnsons (1992). On
the other hand, some researchers nd that compositions scored holistically high contain
more cohesion than those scored low (Jafarpur, 1991). In addition, it is generally agreed
that highly rated essays contain more lexical collocations than do low rated essays (Johns,
1980; Zhang, 2000). The researchers also hold that lexical cohesion is the most common
category in both good and weak essays, followed by reference and conjunction (Johns,
1980; Zhang, 2000).
At the same time, some peculiar features have also been identied in the writing of ESL/
EFL learners (Khalil, 1989; Kuo, 1995). Khalils (1989) analysis showed that the Arab students overused reiteration of the same lexical item as a cohesive device, but underused
other lexical and grammatical cohesive devices. The case was the same with the writing
composed by Spanish-speaking students (Palmer, 1999) and by Chinese undergraduate
English majors (Zhang, 2000). In Wikborgs (1990) study, it was found that Swedish students often showed cohesion problems in their writing ranging from missing or misleading
sentence connection to malfunctioning cohesive devices to too great a distance between the
cohesive items in a cohesive chain. Consequently, the misuse of these cohesive devices affected or even broke the coherence of the text.

M. Liu, G. Braine / System 33 (2005) 623636

625

These empirical studies demonstrate that cohesion is an important element of any type
of writing and that L1 and L2 learners of English have considerable diculty in using
cohesive devices.
3. Rationale for the study
In the past decade or so, China has emerged as a powerhouse in English language
teaching because English is now a required subject from Grade 3 onwards and estimates are that more than 225 million students are enrolled in primary, secondary,
and university levels in China (Mu, 2004; Zhang and Luan, 2002). Further, in Chinese
classrooms, more attention is now being paid to writing (Li, 2000; Yu, 2000). Nevertheless, to our knowledge, only one study (Zhang, 2000) appears to have been conducted to investigate the cohesive features in writing produced by undergraduate
English majors in China.
Though all undergraduates at Chinese universities are required to take English courses,
English majors study English courses full-time while English is only one of the subjects
taken by non-English majors. Nevertheless, non-English majors form the overwhelmingly
dominant population of EFL learners in China. Further, English, in both written and oral
forms, plays a crucial role when these students compete for further education and employment. Thus, their use of English in writing and speech calls for more research.
Moreover, one of the authors functioned as a marker of the writing component in the
College English Test1 (CET) bands 4 and 6 and Tsinghua English Prociency Test I2, and
found that the majority of Chinese undergraduate non-English majors did not appear to
know the use of cohesive devices. They had problems such as a restricted choice of lexical
items, inadequate or wrong use of collocations, and the use of pronouns (Ma and Wen,
1999).
Adopting Halliday and Hasans (1976) cohesion theory as the framework, this study
seeks to identify the general cohesive features in argumentative writing composed
by Chinese undergraduate non-English majors. It aims to answer the following
questions:
(1) What are the cohesive features in the argumentative writing produced by Chinese
undergraduate non-English majors?
(2) What is the relationship between the number of cohesive devices used and the quality
of writing?
(3) Do Chinese students have any problems regarding the use of cohesion in their writing? What are they?

1
College English Test (CET) bands 4 and 6 are national English prociency tests targeted at undergraduate and
graduate non-English majors. The written test of the CET is composed of ve parts: listening, reading, vocabulary
and grammar, translation and writing. Certicates for the written and the speaking tests are awarded if the
candidate satises the requirements at dierent levels such as pass, good and excellent.
2
Tsinghua English Prociency Test I (TEPT1) aims at undergraduate non-English majors at Tsinghua
University in Beijing. Similar to but more demanding and dicult than the CET, TEPT 1 has two parts: the
written test and the speaking test. Failure in either test leads to the failure in TEPT1.

626

M. Liu, G. Braine / System 33 (2005) 623636

4. Research methodology
4.1. Participants
From 96 rst-year band 33 students (in three intact classes) enrolled in Basic Writing in
the second term of the academic year of 20012002 at Tsinghua University, Beijing, 50 students were randomly chosen as the subjects for the present study. Randomly put in the
classes taught by one of the authors, these students came from dierent departments ranging from Computer Science to Civil Engineering to Chinese. Since very few universities in
Mainland China oer writing courses to undergraduate non-English majors, these students at Tsinghua University were particularly appropriate for this study.
4.2. Research design
At the rst class meeting of Basic Writing, the marking criteria for the writing component of both the CET and the Tsinghua English Prociency Test2,4 were clearly explained
to the students, and requirements and aims were specied. Thus, the students had a clear
idea of how they should prepare for the writing class, what they should pay attention to in
writing, and what they were expected to achieve at the end of the term.
During the term, a process-oriented approach was adopted though only one draft of
each writing task was required for teacher evaluation. Meanwhile, explicit explanations
and hands-on practice were provided before each task: sample papers were analyzed, relevant reading and writing assignments were completed, and discussions of students own
writing were conducted. Cohesion and coherence were emphasized through explicit
instruction throughout the term both in required reading and in writing. Thus, for each
piece of writing, the students had a clear understanding of their strengths and shortcomings, where improvement was needed, and what was expected of them in the next writing
task.
Altogether, each of the students produced thirteen pieces of writing of dierent genres
during the 16-week term. During the nal class meeting, the students were given a test.
One of the tasks was to write a 150200-word argumentation on Whether Computer
Games Should or Should Not Be Forbidden in Universities within 30 min. Finally, 50
randomly chosen compositions from this test served as the data for the present study.
4.3. Data analysis
Cohesive devices used in each composition were identied and counted by the researchers (Cronbach alpha = 0.87). Then, frequency, mean and standard deviation of the cohesive devices in each category were computed using SPSS. Correlation was also examined to
determine the relationship between the frequency of the use of cohesive devices and the
quality of writing (see Section 5 below).
3
All the students have to take a placement test upon entering Tsinghua University and are placed in dierent
band groups from 1 to 3 (band 1 is the least procient and band 3 the highest) according to their performance in
the test. After a terms formal learning at the University, the students are automatically promoted into a higher
band group.
4
These criteria are quite similar.

M. Liu, G. Braine / System 33 (2005) 623636

627

Table 1
Mean, Standard deviation, etc., of the scores
Mean

Standard deviation (SD)

Standard error (SE)

Minimum

Maximum

Range

Median

20.6

2.36

0.33

14

24

10

20

5. Findings
5.1. Assessment results
Altogether, 50 compositions were collected and marked by two raters (a = 0.91 according to the marking criteria specied for the Tsinghua English Prociency Test I.5 The
scores given to each composition by the raters were averaged and the mean determined
as the nal score for that piece. The results are presented in Table 1.
As shown in Table 1, the mean score of the 50 compositions was 20.6 (out of a maximum score of 25). Thus, the compositions scored 23 or above were considered the best,
while those scored 16 or below the weakest. Since most of the scores fell into the range
from 18 to 23 (mean score SD), the students in this study could be considered similar
in writing ability.
5.2. Cohesive features found in the argumentative writing
Applying Halliday and Hasans (1976) cohesive framework as the basis for data analysis, the number of cohesive devices used in each composition were counted, followed by
the determination of the frequency, mean and standard deviation of the cohesive devices in
each category. Then, the relationship between the frequency of the use of cohesive devices
and the quality of writing was examined through the use of correlation and analysis of variance. However, in the actual analysis, two types of cohesion, substitution and ellipsis,
were not included because they are more characteristically found in dialogues (Halliday,
2000, p. 337) and they are seldom used in formal writing.
5.2.1. Cohesive devices
Table 2 presents the numbers and percentages of the dierent subcategories of grammatical and lexical cohesive devices identied in the compositions. This shows that the students in the present study employed a variety of cohesive devices with some types of
devices used more frequently than others.
Based on the percentage of each cohesive tie, it is evident that the lexical devices had the
highest percentage (55.6%), followed by the reference devices (29.8%) and the conjunction
5
Marking criteria for writing in Tsinghua English Prociency Test 1 (similar to that of CET): 2125: Main
ideas stated clearly and accurately; well organized and perfectly coherent; very eective choice of words; no
errors; full control of complex structure. 1620: Main ideas stated fairly and accurately; fairly well organized and
generally coherent; eective choice of words; almost no errors; good control of structure. 1115: Loosely
organized but main ideas clear; logical but incomplete sequencing; adequate choice of words but some misuse of
words; many errors. 610: Main ideas not clear or accurate; ideas disconnected; lacks logical sequencing; limited
range and confused use of words; a lot of errors; poor control of structure. 05: Main ideas not at all clear or
accurate; no organization; incoherent; very limited range and very poor knowledge of words; full of errors; no
control of structure.

628

M. Liu, G. Braine / System 33 (2005) 623636

Table 2
Cohesive devices used
Type of cohesive
devices

Reference
devices

Conjunction
devices

Lexical
devices

Total number of
cohesive devices

Frequency
Mean/composition
Standard deviation
Standard error
Range
Percentage based on total

1423
28.46
11.49
1.63
63
29.8%

698
13.96
5.41
0.77
30
14.6%

2658
53.16
11.75
1.66
56
55.6%

4779
95.58
23.12
3.27
112
100%

devices (14.6%). This nding is consistent with that of previous research studies (Yvette
and Yip, 1992; Zhang, 2000). On the other hand, the greater use of reference devices in
the argumentative writing found in this study might be due to the fact that the denite article the was also calculated and integrated into reference devices. On the contrary, the use
of the denite article the was excluded in some research studies (Zhang, 2000). The following presents the further elaboration of the use of dierent cohesive devices.
5.2.2. The use of cohesive devices in each category
5.2.2.1. The use of reference devices. As shown in Table 3, among the three sub-categories
of reference devices, pronouns (60%) had the highest percentage of use, followed by the
denite article (29.3%) and the comparatives (6.1%). Demonstratives (4.6%) had the least
percentage of use. These ndings are somewhat dierent from those of Zhang (2000),
where comparatives were the least used in expository compositions produced by Chinese
undergraduate English majors.
The predominant use of pronouns in the compositions were they, followed by it
them, these, we and I, which might be due to teacher inuence. The students were
encouraged to use third person pronouns in order to be more objective and authoritative
in argumentative writing. Among demonstratives, this and these occurred much more
frequently than that and those. This might manifest that the students were more comfortable at using items that refer to something near, as claimed by Zhang (2000). All these
can be demonstrated by the following extracts from students compositions.
Table 3
Reference devices used
Pronominal
devices

Frequency
Mean per composition
Standard deviation
Standard error
Range
Percentage

854
17.08
7.44
1.05
32
60%

Most frequently
used cohesive items

They, it,
them, their,
we, I

Demonstrative devices
and the denite article
Demonstratives

The denite
article

65
1.3
1.33
0.19
6
33.9%
4.6%
This, these,
that

417
8.34
6.53
0.92
32

Comparative
devices

Total number of
reference devices

87
1.74
1.71
0.24
9
6.1%

1423
28.46
11.49
1.63
63
100%

29.3%
More, more
and more,
the most, same

M. Liu, G. Braine / System 33 (2005) 623636

629

Example 1 (score: 21). Firstly, playing computer games costs them too much time, . . .
What they, college students, should do is to study hard for future but not waste time in
playing. They would not have a job when they leave if they dont study knowledge and
learn skills now.
Example 2 (score: 23). It wastes the students studying time; hurts their eyes and mind. It
even causes more and more serous society problem. The college students is the hope of the
country, so its necessary to forbid them to play computer games and give them a better
future.
Example 3 (score: 19). First, their study will become worse and worse if they play games.
. . . Sometimes students play computer games to release themselves. Its right. But the
games often let students indulge in it. They will forget any things except games when they
are playing.
A further analysis of the reference devices used in the compositions reveals that nearly
all of the reference devices used were anaphoric with only a few cataphoric devices (4),
which is also indicated by the above examples.
5.2.2.2. The use of conjunction devices. As shown in Table 4, among the ve sub-categories
of conjunction devices, additive devices (44.4%) accounted for the largest percentage of
use, followed by causal (19.5%), temporal (18.6%), and adversative (15.6%) devices. Continuative devices occurred only 13 times in the 50 compositions analyzed, probably because they are seldom used in formal written work.
As can be seen from Table 4, all the students appeared to be aware of the ve sub-categories of conjunction devices. Additive devices were the most frequently used because
they are helpful in connecting phrases, clauses and sentences in writing. The cohesive items
with the highest frequency were and, also and or, all of which belong to the group of
easy words and expressions Chinese students start to learn as soon as they have access to
English. Among adversative devices, but was used with the highest frequency (86), while

Table 4
Conjunction devices used

Frequency
Mean per
composition
Standard
deviation
Standard error
Range
Percentage
Most frequently
used cohesive
items

Additive
devices

Adversative
devices

Causal
devices

Temporal
devices

Continuative
devices

Total number
of conjunction
devices

310
6.2

109
2.18

136
2.72

130
2.6

13
0.26

698
13.96

3.4

1.3

2.54

2.15

0.49

5.41

0.48
17
44.4%
And, also,
or

0.19
6
15.6%
But

0.36
16
19.5%
So, because
of, for

0.3
12
18.6%
First, rstly,
second, secondly,
third, thirdly,
nally

0.07
2
1.9%
Of course,
surely

0.77
30
100%

630

M. Liu, G. Braine / System 33 (2005) 623636

on the contrary and instead occasionally occurred in the writing. This may imply that
the students were not competent enough to use other cohesive items to indicate transition
of meaning such as however, rather, and on the contrary. Among causal devices, so
was the most frequently used item, followed by because of and for. Other items such
as as a result and thus were only occasionally or rarely used. As to temporals, the students preferred using rst, rstly, second, secondly, third, thirdly and nally to
indicate the sequence of time or importance, etc. However, items as sum up, meanwhile,
and all in all were hardly found in the writing. The following are some of the examples
taken from the students compositions:
Example 4 (score: 21). I just give three reasons, but therere still many other reasons. So,
playing computer games should not be forbidden. It has many advantages, though it also
has disadvantages. So, I must say forbidding is forbidden.
Example 5 (score: 23). And for another reason, therere many other things except computer games that affect students study. . . ., and some one may spend too much energy
on playing music instruments or watching movies. In all, there are many other things that
can attract students attention, and they can be reasons to have effects on their study.
Example 6 (score: 16). First, college students is not the same as high school students. . . .
So they shouldnt spend all their time on studying. Second, after a whole weeks study, who
want to go on? Everybody need rest.
5.2.2.3. The use of lexical cohesion. As shown in Table 5, in the ve sub-categories of lexical
devices, repetition (83.8%) of the same word formed the highest percentage of use, followed by collocation (6.6%), synonym (6.5%) and antonym (2.4%). Superordinates and
general words were rarely used.
As further indicated in Table 5, the students tended to use repetition more often than
synonym, collocation and antonym. Further, the majority of the most frequently used
cohesive items were computer, game, student, college, and forbid. These were the
words given in the title and closely related to the theme of the topic.

Table 5
Lexical devices used

Frequency
Mean per
composition
Standard
deviation
Standard error
Range
Percentage
Most frequently
used cohesive
items

Repetition

Antonym

Synonym

Superordinate

Collocation Total number

of lexical
devices

2227
44.54

63
1.26

173
3.46

20
0.4

175
3.5

2658
53.16

9.77

1.5

2.3

0.95

2.56

11.75

1.38
47
83.8%
Computer,
game, student,
college, study

0.21
4
2.4%
Tiredrelaxed,
successfailure

0.33
12
6.5%
Forbidban

0.13
0.36
3
13
0.7%
6.6%
Collegeteacher
student,
familyparent

1.66
56
100%

M. Liu, G. Braine / System 33 (2005) 623636

631

5.3. The relationship between the number of cohesive devices and the quality of writing
The gures in the above Tables show that the students had a knowledge of cohesive devices and used a variety of them in their writing. Nevertheless, it is worth noting that all
three leading cohesive devices used (reference, conjunction and lexical cohesion) had a
considerable range, from 30 to 63. For example, although the denite article the was
extensively used in some compositions, it was seldom used in others. In order to investigate
the relationship between the number of cohesive devices and writing quality, correlation
was computed between the numerical composition scores and the frequency of devices (total devices per composition) in terms of cohesive categories (reference, conjunction and
lexical cohesion). The results are displayed in Table 6.
The correlations among composition scores and dierent cohesive variables demonstrated in Table 6 reveal that the composition scores signicantly co-varied with the total
number of cohesive devices (r = 0.315). Moreover, the composition scores were highly correlated to lexical devices among the three main categories of cohesive devices (r = 5.36).
Contrary to the ndings claimed by other researchers (Johnson, 1992; Zhang, 2000), this
nding is quite unexpected and might be partly due to the fact that compositions with high
scores tended to be longer and involved more lexical items.
5.4. Problems with cohesion
As shown in the preceding Tables, the students involved in this study were able to use a
variety of cohesive devices in their writing to make it cohesive and coherent. However, certain cohesion problems also existed in the argumentative writing created by these undergraduate non-English majors. This is similar to Wikborgs (1990) ndings that Swedish
students often had problems with cohesion. These problems were mainly concerned with
the use of reference and lexical cohesion.
5.4.1. Problems with reference devices
Although reference devices were the second most extensively used categories of cohesion in the students writing, the students appeared not to be at ease with their use. The
problems with the use of reference devices were mainly of three types: the shifted use of
pronouns; omission or misuse of the denite article; and underuse of comparatives and
overuse of the phrase more and more.
Table 6
Correlation between composition scores and cohesive devices

Composition scores
Reference devices
Conjunction devices
Lexical devices
Total number of
cohesive devices
*
**

p < 0.05.
p < 0.01.

Composition
scores

Reference
devices

Conjunction
devices

Lexical
devices

Total number
of cohesive devices

1
0.065
0.046
0.536**
0.315*

1
0.432*
0.497*
0.851**

1
0.372**
0.638**

1
0.842**

632

M. Liu, G. Braine / System 33 (2005) 623636

The students in this study, like those in Zhangs (2000), tended to shift pronouns within
or between clauses such as from the rst person to the second or from the singular form to
the plural. As a result, the referents and the referring items were made inconsistent, which
often times confused the reader and even caused problems in comprehension. Below are
some examples from the students writing.
Example 7 (score: 20). When a student came into college, they are no longer children,
they have grown up.
Example 8 (score: 19). College students must learn how to plan their time before he go to
the society.
Example 9 (score: 14). They need a way to relax themselves. Computer games can be very
good. It can also develop your skills and ability of thinking. So we shouldnt get it always
without thinking.
Another problem that the students in this study had in the use of reference devices
was the omission or misuse of the denite article the. The students sometimes confused the use of denite and indenite articles, but, more often, they omitted obligatory articles or inserted unnecessary ones in their writing. This misuse of the denite
article might be due to the transfer and interference of their mother tongue because
the Chinese language does not have any articles. The same problem was found in
Zhangs (2000) study. Below are some of the examples with the revised version in
parenthesis.
Example 10 (score: 21). With the broad use of computer, the computer games is
becoming very popular to college students (x).
Example 11 (score: 18). They look into it, and discover that nowadays many students
spend a lot of time on the computers instead of reading books in the library (x).
Example 12 (score: 18). As branch of software industry, computer games produce
some billions of dollars in US and even more in Japan (a, the).
In addition, the analysis of the compositions also reveals that the students were not
condent in using comparatives in their writing. The most striking feature was the underuse of comparatives and the overuse of the phrase more and more. Among the total
number of 4779 cohesive devices used in the 50 compositions, comparatives occurred only
87 times, accounting for 6.1% of the total number of sub-categories of reference devices.
This is a low percentage given the fact that the students had to employ comparison and
contrast to support their opinions that college students should or should not be forbidden to play computer games. Nevertheless, this might suggest that the students had difculty in using comparative expressions such as as . . . as, not so . . . as and more than.
On the other hand, the comparative expression more and more seemed to be the students favorite in their writing, which is probably attributed to language transfer in that
a similar expression yue lai yue is extensively used in Chinese. The following examples
can illustrate this.
Example 13 (score: 19). Most college students say that they feel heavier and heavier press
day by day.
Example 14 (score: 23). It even causes more and more serous social problem.

M. Liu, G. Braine / System 33 (2005) 623636

633

Example 15 (score: 22). We all know that, nowadays, college students have more and
more schoolwork to do.
5.4.2. Problems with lexical devices
Being the main carrier of message and the means of expression, lexical items are the
principal components of any composition. That may contribute to the fact that lexical devices are the most extensively used categories of cohesion in students writing. However,
because lexicon involves both meaning and usage, it becomes a much more complicated
and dicult task for foreign learners of English. The analysis of the compositions in this
study indicates that the students had two areas of diculty in using lexical devices: a restricted choice of lexical items and the wrong use of collocation. This is implied by the fact
that the majority (83.8%) of the lexical devices were the repetition of words. Further, the
repeated words were conned to a narrow range such as computer, game, student and
forbid. Other types of lexical devices such as synonym, antonym and superordinate were
hardly found in the writing. This restricted use of lexical items might be due to the students English prociency because they had little exposure to English outside the English
lessons oered by the University. An additional reason could be that they only had 90 min
per week for writing in English. The students knowledge of vocabulary justies the statement that vocabulary is the thorniest part of learning a foreign language (Hu et al., 1982;
Kang, 1995).
The second problem the students had in using lexical devices was the wrong use of collocations, as illustrated by the following examples with the revised version in parenthesis:
Example 16 (score: 20). First, it makes students interest in the computer (interested).
Example 17 (score: 18). It means not only the students but also the teachers should
face computer games in a proper attitude (take a proper attitude toward computer
games).
Example 18 (score: 17). Compare to its wasting time, it has much more advantages
(compared with those disadvantages such as wasting time, it has many more
advantages).
As shown by the examples, the wrong use of collocations involved phrases, verbs,
nouns and prepositions. According to Zhang (2000), this is due to the Chinese tradition
of teaching vocabulary out of context, isolated from semantic meaning.
6. Discussion and conclusions
The ndings reveal that lexical devices (55.6%) constituted the highest percentage of
the total number of cohesive devices used in the argumentative writing, followed by
reference devices (19.8%) and conjunction devices (14.6%). Because of calculation of
the use of the denite article the, reference devices accounted for a higher percentage
in the writing than in other research studies (Johns, 1980; Zhang, 2000).
The qualitative analysis of the use of cohesive devices in the compositions shows that,
among the three sub-categories of reference devices, pronominals (60%) were the most frequently used devices while demonstratives (4.6%) least used. Further, the students, especially
the less procient ones, were also found to have diculty in using reference devices in a con-

634

M. Liu, G. Braine / System 33 (2005) 623636

sistent and eective way. As Chang claimed (1987, cited in Zhang, 2000), Chinese students
have diculty in using articles consistently and correctly: they tend to omit necessary articles,
insert unnecessary ones, or confuse the use of denite and indenite articles in their writing.
With regard to the use of conjunctions, it seems that the students were capable of using a
variety of devices to bridge the previous sentence(s) and the following one(s) to make their
writing clearer and more logical. However, only those commonly used items as and, but,
or and so were the students favorites, whereas the items learned later such as furthermore,
on the contrary, moreover, in addition, on the whole, and nevertheless seldom occurred
in their writing.
Although lexical cohesion was the most extensively used category of cohesive devices, it
appeared to be an area that needs improvement for Chinese undergraduate non-English
majors. The students demonstrated a limited choice in the use of lexical items and the great
majority of the lexical devices were repetitiously used. The rare use of synonyms, antonyms, superordinates, and general words, a general trend that can also be found in the
writing of CET and Tsinghua English Prociency Test I, indicates that much needs to
be done in the teaching of vocabulary in China.
A further statistical analysis shows that the composition scores were highly correlated with
the number of lexical devices and the total number of cohesive devices used. This indicates
that there was a signicant relationship between the number of cohesive devices used and
the quality of the argumentative writing created by these undergraduate non-English majors.
On the whole, given the fact that the students admitted to Tsinghua University (top 1 or 2
university in China) are more procient and competent in using English than most university
students in the country and that the subjects involved were at an advanced level and received
training in English writing for one term, it could be hypothesized that Chinese undergraduate non-English majors in general tend to use more lexical devices in their writing, followed
by reference and conjunction devices. Concerning the use of each category of cohesive devices, they most probably have more diculty in making eective use of pronouns, articles,
comparatives and lexical items due to low English prociency and/or little or even no training in writing. As a consequence, much needs to be done in the teaching of writing to enhance
the students awareness of the importance and use of cohesive devices in their writing.
7. Pedagogical implications
As illustrated by the above ndings and conclusions, this study oers some implications
for the teaching of English writing for undergraduate non-English majors in Mainland
China in general.
First of all, as marking criteria set the standard for judging a piece of writing, writing
teachers and learners should be familiarized with the marking criteria for dierent writing
tests. It is the writing teachers responsibility to rst comprehend and then explain the
marking and assessing criteria in the class, and thereby enhance students awareness of
what contributes to the quality of writing (Densteadt, 1996). As a result, the students will
know what to emphasize when writing in English.
Moreover, as illustrated above, some of the students were found to have problems in
using cohesive devices eectively and accurately. Therefore, explicit instruction with
examples should be provided by the writing teachers in class instead of accumulated
awareness through learning (Al-Jarf, 2001; Reichelt, 2001), especially because Chinese
undergraduate non-English majors usually do not have much time to learn or much ac-

M. Liu, G. Braine / System 33 (2005) 623636

635

cess to English. In addition, focused activities should be developed in combination with

explicit instruction. The students can be required to write a paragraph for 35 min, using
dierent cohesive devices. Following that, peer review could be used to analyze the cohesive devices used in each others writing and comment on the eects of using those cohesive devices. After each task is nished, it is necessary for the writing teacher to choose a
sample composition for critique, stressing the importance of using cohesive devices eectively and appropriately and reminding the students to avoid overusing or underusing
cohesive devices.
Since the acquisition of English vocabulary by EFL learners is a thorny task because
there are subtle devices in each word as far as meaning and use are concerned, this kind
of teaching may not be eective with the learning of lexical devices, which concern far
more aspects of language such as semantics and pragmatics. In order to raise the students
syntactic and semantic awareness, they should be encouraged to read as extensively as possible. By doing this, they will not only extend their vocabulary but also better understand
the use and meaning of words in dierent contexts. Apart from reading, as suggested by
Liu (2000), it is necessary to train the students to paraphrase words or phrases by means of
synonyms, antonyms, hyponyms or examples through exercises. Thus, they would not
only recall as many words as possible but will be able to create more vivid and eective
compositions through the use of lexical devices.
Finally, because reading is closely linked to writing (see Carson, 1993; Kroll, 1993;
Leki, 1993), reading should be integrated into the teaching of writing. The students in
the present study were generally aware of the use of cohesive devices through training
by means of explicit instruction, reading and practice. This indicates that reading and
writing should be taught in combination. However, this is not the case in China: Zhang
(2000: 89) claimed that teachers of writing tend to neglect reading and vice versa.
Thus, reading teachers could integrate some writing into their teaching by analyzing
the use of language, cohesive devices and text patterns and assigning corresponding writing tasks. Reciprocally, writing teachers could incorporate reading into their teaching by
assigning reading materials. The focus should be on the features explained in the class
and the students need to be asked to produce a type of writing modeled on the text they
have read. In this manner, the students are expected to become more sensitive to the
characteristics of good English writing and more able to create writing of high quality
as far as cohesion is concerned.
In conclusion, these implications may apply beyond China to other EFL contexts as
well. However, methods of improving EFL/ESL learners writing and the eects of dierent teaching methods need further examination in actual contexts.
References
Al-Jarf, R.S., 2001. Processing of cohesive devices by EFL Arab college students. Foreign Language Annals 34,
141150.
Carson, J., 1993. Reciprocal themes in ESL reading and writing. In: Carson, J.G., Loki, I. (Eds.), Reading in the
Composition Classroom: Second Language Perspectives. Henley & Henley, Boston, MA, pp. 85104.
Densteadt, J., 1996. Balancing the teaching and assessment of writing. TESOL Quarterly 6, 3536.
Green, C.F., Christopher, E.R., Jacquelyn, L., 2000. The incidence and eects on coherence of marked themes in
interlanguage texts: a corpus-based enquiry. English for Specic Purposes 19, 99113.
Halliday, M.A.K., 2000. Introduction to Functional Grammar, second ed. Foreign Language Teaching and
Research Press, Beijing.

636

M. Liu, G. Braine / System 33 (2005) 623636

Halliday, M.A.K., Hasan, R., 1976. Cohesion in English. Longman, London.

Hu, Z., Brown, D.F., Brown, L.B., 1982. Some linguistic dierences in the written English of Chinese and
Australian students. Language Learning and Communication 1, 3949.
Jafarpur, A., 1991. Cohesiveness as a basis for evaluating compositions. System 19, 459465.
Johns, A.M., 1980. Cohesion in written business discourse: some contrasts. The ESP Journal 1, 3644.
Johnson, D.P., 1992. Cohesion and coherence in compositions in Malay and English. RELC Journal 23, 117.
Kang, S., 1995. The eects of a context-embedded approach to second-language vocabulary learning. System 23,
4356.
Khalil, A., 1989. A study of cohesion and coherence in Arab EFL college students writing. System 17, 359371.
Kroll, B., 1993. Reciprocal themes in ESL reading and writing. In: Carson, J.G., Leki, I. (Eds.), Reading in the
Composition Classroom: Second Language Perspectives. Heinle & Heinle, Boston, MA, pp. 6181.
Kuo, C., 1995. Cohesion and coherence in academic writing: from lexical choice to organization. RELC Journal
26, 4762.
Leki, I., 1993. Reciprocal themes in ESL reading and writing. In: Carson, J.G., Leki, I. (Eds.), Reading in the
Composition Classroom: Second Language Perspectives. Heinle & Heinle, Boston, pp. 931.
Li, J., 2000. Brainstorming: an eective warm-up activity in the English writing class. Teaching English in China
23, 4447.
Liu, D., 2000. Writing cohesion using content lexical devices in ESOL. Forum, 38. Retrieved on April 13 from:
<http://ericir.syr.edu/plweb-cgi/fastweb?search>.
Ma, G., Wen, Q., 1999. The study of the factors aecting college students writing. Foreign Language Teaching
and Research 4, 5561.
Mu, Y., 2004. Retrieved on June 16 from: <http://news.sina.com.cn/e/204-04-28/11253173485.shtml>.
Palmer, J.C., 1999. Coherence and cohesion in the English language classroom: the use of lexical reiteration and
pronominlisation. RELC Journal 30, 6185.
Reichelt, M., 2001. A critical review of foreign language writing research on pedagogical approaches. The
Modern Language Journal 85, 578593.
Wikborg, E., 1990. Types of coherence breaks in Swedish student writing: misleading paragraph division. In:
Connor, U., Johns, A.M. (Eds.), Coherence in Writing: Research and Pedagogical Perspectives. TESOL,
Alexandria, VA, pp. 131149.
Yang, A., 1994. Ways to improve university students writing ability in English. Journal of Guangdong Medical
College 12, 284286.
Yu, Y., 2000. Helping EFL students learn to write through feedback. Teaching English in China 23, 5559.
Yvette, F., Yip, L., 1992. A comparison of Internet conjunctive cohesion in the English essay writing of
Cantonese speakers and native speakers of English. RELC Journal 23, 1528.
Zhang, M., 2000. Cohesive features in exploratory writing of undergraduates in two Chinese universities. RELC
Journal 31, 6193.
Zhang, H., Luan, F., 2002. A survey on EFL teaching in dierent countries of the world. Retrieved on June 20
from: <http://www.cbe21.com/subject/english/html/050203/2002_11/20021105_1977.html>.