master's thesis
DESCRIPTION
An Investigation of Writing Errors Made by High Proficiency and Low Proficiency First Year Students at Sultan Qaboos University of OmanTRANSCRIPT
1
An Investigation of Writing Errors Made by High Proficiency
and Low Proficiency First Year Students at Sultan Qaboos
University of Oman
By
Hashil Al-Saadi
MA Thesis
The University of Bristol
2000
© Hashil Al-Saadi, 2000
2
Chapter one
Introduction
The study of learners‟ errors has been a primary focus of second language acquisition
research during the last few decades. The increasing interest in the analysis of learners‟
errors stems from the assumption that such analysis would provide teachers, as well as
second language acquisition researchers, with valuable information on the areas of difficulty
for second/foreign language learners when learning another language and on the major
sources of learners‟ errors. However, by the late 1960s, such an assumption became more
evident and the interest in studying learners‟ errors has attracted the attention of more
researchers especially after Corder‟s (1967) seminal article on the “significance of learners‟
errors”. In this article, Corder shifted the status of errors from undesirability to that of a
guide to the areas of difficulty that language learners have. He also proposed other possible
sources for errors besides the mother tongue interference such as over generalisation of the
Target Language (TL) rules and strategies of second language learning. This climate gave
rise to the Error Analysis (EA) movement which appeared in the late 1960s as a refreshing
alternative to the prevailing but more restrictive Contrastive Analysis (CA). Since then,
teachers and researchers have much of their time extracting errors from student compositions
and conversations, submitting them to close analysis, and using them for theory construction
and classroom practice.
To date, in Oman there have been no studies of second language learners that have yet been
completed in the field of EA; no studies have been devoted to the analysis and description of
the frequency and types of errors made by students neither at the pre-university nor the
3
university level. Therefore, this dissertation attempts to address this lack of research in the
area of EA in Oman by reporting on the findings of a small-scale study into the types and
frequency of errors made by high proficiency and low proficiency first year students at Sultan
Qaboos University (SQU) of Oman. I hope that the findings of the present EA study will
prove useful for practising teachers and contribute to the increasing research in the field of
second language acquisition with regard to second language learners‟ errors in general and
Arab learners‟ errors in particular. Also, it is hoped that the present research will pave the
way to more studies into the field of EA which address the areas of difficulty for Omani
students by a thorough investigation of the types, frequency and sources of students‟ errors at
various stages of proficiency.
1.1. Aims of the Study
The underlying objective of the present EA study is twofold. On the one hand, the study
aims to investigate the most prevalent error types that occur in the writing of first year
students at Sultan Qaboos University of Oman at two different levels of proficiency. On the
other, it aims to survey the methods used by writing teachers in responding to their students‟
errors in writing.
One of the fundamental hypotheses that the present research is designed to examine is the
one based on the conception that the types and frequency of errors will vary with the
students‟ level of proficiency (see hypothesis 1 below). In other words, the more proficient
the students are, the lesser errors they are likely to make and vice versa. Therefore, the
present study aims to test the above hypothesis via the qualitative and quantitative treatment
of error types extracted from the actual writing of 60 Arabic-speaking students at two
different levels of proficiency who represent the subjects of the present EA study
4
Also examined within this study is the assumption that all teachers will react similarly to the
same error types. Basically, this research is aimed at testing the hypothesis which claims that
the same error types will be noticed and marked as deviant by all teachers (see hypothesis 2
below). This aim is achieved by means of presenting the eight teachers who represent the
teacher sample with a text produced by one of the students from the low level of proficiency
and inviting them to mark it.
1.2. Study Questions
Generally speaking, the present study addresses the three following questions:
I. What are the most common errors made by high and low proficiency first year students
in writing?
II. What is the relationship between students‟ level of proficiency and the types and
frequency of errors they make?
III. How do teachers respond to their students‟ writing errors?
Basically, this research is aimed at investigating three major areas. These are: the nature,
types and frequency of errors made by students at low level of proficiency (pre-intermediate
band 2 students); the nature, types and frequency of errors made by students at a higher level
of proficiency (upper-intermediate band 4 students) and the approach available to teachers for
responding to their students‟ errors in writing.
Basically, the two underlying hypotheses that stimulated the present research are:
I. The type and frequency of errors will vary with the student level of proficiency.
II. All teachers will react similarly to the same error types.
5
The assumption is that the careful analysis of students‟ errors in each type will provide ample
information on the areas of difficulty that students at each level have. Such information
would have direct implications for teachers who would be in a better position in dealing with
their students‟ difficulties in various areas of the target language (TL). In the light of the
findings of such analysis, teachers will be able to design materials and adopt techniques that
account for the students‟ areas of difficulty and, hopefully, help them overcome these
problems.
It is also the assumption that the comparison of students‟ errors at both levels of proficiency
would shed light on the process of fossilisation (Selinker, 1972). This is basically a process
in second or foreign language learning which refers to certain erroneous features lexis and
grammar which tend to become fixed or fossilised in second language learning.
1.3. Nature of Investigation
It is worth noting at this point however that no attempt is made in the present EA study to
explain the students‟ errors, i.e., tracing back every error type to its potential source. The
discussion and the classification of the errors will be limited to the descriptive aspects of
errors on the assumption that “the accurate description of errors is a separate activity from
the task of inferring the sources of those errors” (Dulay, et al, 1982:145). In the light of the
above assumption, errors will be labelled and classified according to some observable surface
feature of the error itself (e.g., the linguistic level that the error affects), without reference to
its underlying cause or source.
1.4. Organisation
6
The five chapters that make up this dissertation are divided as follows. The present chapter
constitutes the initial chapter of the thesis in which the theoretical position of the researcher
is set out together with the study aims and objectives. The second chapter addresses the
recent discussion and arguments in the literature of EA regarding various aspects of
ESL/EFL learners‟ errors. In this chapter, the literature is carefully reviewed for the recent
definitions, classifications and implications of learners‟ errors and error types. The third
chapter sets out the overall design of the study. It specifies in absolute terms the research
method, study population and sample, research instruments and data analysis methods. The
major findings of the present EA study are detailed and discussed in the fourth chapter, while
the main conclusions, implications and recommendations reached in the light of the present
findings constitute the last chapter.
7
Chapter Two
Literature Review
2.0. Introduction:
One of the most important and significant shifts in the field of Applied Linguistics in recent
years has been from the view of teacher as controller of language learning towards a more
learner-centred view which stresses the learner‟s active role in managing his or her own way
towards the linguistic competence sufficient for their communicative needs. One major
result of such shift of attention has been the increasing concerns in the monitoring and
analysis of the learner‟s language. This interest in the learners‟ errors stems from the
significant and useful information and data that such error-based studies provide for a better
understanding and explanation of the process of second/foreign language acquisition, which,
in turn, have stimulated the major changes in teaching practices (Dulay et al 1982:138).
Therefore, there has been an ever-increasing research interest in the analysis of learner‟s
language in general and learner‟s errors in particular. This increasing recognition in the
importance of the learner‟s language has taken two phases: the first phase represents the
systematic analysis of and comparison between the learner‟s mother tongue and the target
language while the second concerns the analysis of and comparison between the learner‟s
second/foreign language and the target language. The former phase has been the focus of the
Contrastive Analysis (CA) movement which attracted great interest among linguists who
were involved in language teaching and production of language teaching materials in the
fifties and sixties (James 1998:4). The latter, however, has been the focus of the Error
Analysis (EA) movement, the new paradigm for the study of second/foreign language
learning and the movement which has been preferred over the traditional CA since it has
8
succeeded in providing valid explanations to some of the learners‟ difficulties in learning
second/foreign language and has proposed a more psychological approach to dealing with
these difficulties.
Unsurprisingly, the field of error analysis has attracted many researchers who have made
important contributions to our knowledge and have furthered our thinking about learners‟
errors and the strategies they employ in learning a second/foreign language (e.g. Corder,
1981; Richards, 1974; Dulay, Burt and Krashen, 1982; Selinker, 1972; Nemser, 1971; Ellis,
1986; James, 1998;).
As with regard to errors, the many studies that have been devoted to the description and
explanation of learners‟ errors have resulted in a wide range of discussions and so have given
rise to many controversial issues concerning different approaches to labelling and classifying
errors as well as other issues regarding how and when teachers should correct their students‟
errors.
The thrust of the present chapter, however, highlights the recent discussions and arguments in
the literature concerning the area of error analysis and error correction. I shall begin by
defining errors and highlighting the difference between errors and mistakes and briefly
review the historical development of the concept of „error‟. A discussion of the climate that
EA movement emerged in will follow. An attempt is made to cover all the different error
types and error classifications and taxonomies that have been proposed by researchers in the
domain. Insights gained from different studies in the area of error analysis into the process of
second/foreign language acquisition process will also be presented. Then, the different
studies that are aimed at studying a particular feature of language errors such as error
frequency and error gravity will also be reviewed. Special attention will be paid to those
9
studies that were aimed at studying the interlanguage system and the commonest type and
frequency of errors made by Arab EFL/ESL learners. I will conclude the chapter by
highlighting the recent arguments in the literature on how to detect and correct learners‟
errors and presenting the recent studies that have investigated issues like error correction and
teacher responses to student writing.
2.1. Defining Error:
It is fruitful to admit, in the first place, the fact that errors are inevitable part of any learning
process. That is, people often make errors when they are engaged in a learning process of
any type and in any discipline. In language learning situations, though, language teachers,
during their long careers, have come to realise that errors are a natural phenomenon in the
process of language learning and so they cannot be eradicated completely neither in the
students‟ speech nor in their writing. The writing errors are therefore the focus of the present
chapter which serves as a background to the whole study.
It seems of great importance at this stage, though, to define „language errors‟ before we
proceed on discussing various aspects of the term. In this regard, I shall present four
definitions of language errors followed by a discussion of different aspects of these
definitions. The Longman Dictionary of Language Teaching and Applied Linguistics
(1992:127) defines errors as follows:
“In the speech or writing of a second or foreign language learner, the use of a
linguistic item (e.g. a word, a grammatical item, a speech act, etc.) in a way which a
fluent or native speaker of the language regards as showing faulty or incomplete
learning.”
Another definition has been offered by Dulay et al (1982:138):
10
“Errors are the flawed side of learner speech or writing. They are those parts of
conversation or composition that deviate from some selected norm of mature
language performance.”
James (1998:64) endorses Lennon‟s (cited in ibid: 64) definition of errors who views errors
as
“linguistic form which, in the same context … would in all likelihood not be
produced by the learner‟s native speakers counterparts”.
Later on, based on the learner‟s „intentionality‟, James (ibid:78) proposes another definition
of error in which he defines error as:
“an instance of language that is unintentionally deviant and is not self-corrigible by
its author”.
By devoting a close look at the definitions presented so far, it is plainly obvious that all,
except the last one, contain some kind of reference points or criteria for deciding whether or
not a particular bit of language is an error. In the first and third definitions, the correctness of
learner‟s language is being measured against that of a fluent or native speaker of the target
language, while in the second, the criterion for correctness is some selected norms of mature
language performance. It is clear then that errors are being defined with reference to native
speakers of the target language based on the Chomskian line on this issue who claims that
native speakers do not and can not commit errors since they know their language perfectly
and comprehensively (James,1998:83). But James (ibid) suggests in this regard that we
should not compare the language of a real learner and an ideal native speaker with uniform
experience in an ideal speech community with no dialect diversity and no variations among
speakers. He insists that “the comparison should rather be between two real individuals or at
least two categories of persons” (64). It is because of such observation that James endorses
the third definition presented above.
11
Further more, the issue of what is to be accepted as a correct form of language and what is to
be regarded as a deviant or erroneous stimulates the following question: on what grounds can
we place our judgement about students‟ language correctness? In other words, what are the
measures of incorrectness and/or deviance? James (1998:64) suggests that learners‟
ignorance of the target language can be expressed in terms of four measures which serves as
reference points to our judgement: grammaticality, acceptability, correctness and
strangeness and infelicity. I am going to briefly highlight the first three measures only as
they most concern us here.
Grammaticality is synonymous with „well-formedness‟. It is the grammar that decides
whether something said or written by learners is grammatical. Grammaticality is viewed as
an objective measure of learners‟ correctness because the decision made is out of the orbit of
human whim (ibid:65). However, grammaticality can be questioned in terms of its reliability
as a reference point. That is, grammaticality works well in the clear-cut cases where there is
only one form which can be accepted. In some other contexts, however, two forms can be
accepted depending, in the first place, on that very context and secondly on the speaker‟s or
writer‟s intention.
The second measure of whether a piece learners‟ language is deviant is acceptability. An
acceptable piece of language is one that has been, or might be, produced by a native speaker
in some appropriate context and is accepted by other native speakers as belonging to the
language in question (James,1998:67). It can be deduced then that acceptability has to do
with context rather than rules but, however, we cannot think of a case where an utterance or
written discourse is ungrammatical but acceptable. In other words, as James puts it:
grammaticality is a prerequisite to acceptability.
12
The third measure of learners‟ language is correctness. A piece of learners‟ language can be
grammatical and acceptable but, in the prescriptive grammarians‟ point of view, is incorrect.
A closer to hand example of this case is when a learner says: “they went with Tom and me”.
This is a case where an „accepted‟ piece of language is edited by “they went with Tom and I”
by members of Prescriptive Grammar School.
Again, the first definition implies some sources of errors: faulty or incomplete learning. It is
implied here that learners make errors because they lack the knowledge necessary for a well-
formed language. As a matter of fact, many EA studies have related errors to different
sources (e.g., interlingual, intralingual, developmental, induced, avoidance, learning
strategies, communication strategies, etc.), but in some cases the boundaries between these
sources are not clear cut. In broader terms, as Johnson (1988:90) suggests, there are at least
two reasons why learners get things wrong. One is that learners either do not have the
appropriate knowledge of the TL, or has some false knowledge. The result is what Corder
(1967) terms as an “error”. The second reason why learners get things wrong is because they
may lack the ability to process the knowledge they already have in certain conditions and the
result would be what Corder (ibid.) refers to as a “mistake”. (A distinction between errors
and mistakes is made in section (2.2) below).
A close look at the last definition yields in two characteristics of errors: intentionality and
self-corrigibility. According to James (1998:77), an error arises only when there was no
intention to commit one and so we cannot talk about so-called “deliberate errors”. In fact, we
can reasonably determine, based on the learner‟s intention, whether a deviance in his or her
language is a mistake, slip or rather a genuine error. The other feature that the last
definition contains is that errors cannot be corrected by learners themselves although their
intention has been drawn to them because they lack the knowledge that would enable them
13
do so. This is, in fact, another difference between errors and mistakes which will be the
focus of the next section.
2.2. Errors Verses Mistakes:
A distinction is always made in the literature between errors and mistakes as the two
concepts imply different characteristics of language learner‟s underlying knowledge. The
error:mistake distinction was first introduced into the modern debate by Corder (1967) who
associates errors with failures in „competence‟ and mistakes with failures in „performance‟.
For Corder, errors are systematic and represent learners‟ knowledge of the target language up
to date and so they prove useful for both teachers and researchers. Mistakes, in contrast, are
not systematic and do not reveal so much of learners‟ competence in the target language
since they are a result of some performance failures when performing the target language.
Corder (1974:24-5) wrote:
“We must therefore make a distinction between those errors which are the product of
chance circumstances and those which reveal his [learner] underlying knowledge of
the language to date … It will be useful to therefore hereafter to refer to errors of
performance as mistakes, reserving the term error to the systematic errors of the
learner from which we are able to reconstruct his knowledge of the language up to
date.”
James (1998:77) adds another dimension to the error:mistake distinction. He stresses that
learner‟s intentionality is another important factor to determine whether learner‟s erroneous
output is an error or a mistake. He suggests that if the learner has the intention to produce the
very structure and cannot correct it, then the faulty output is considered as an error. If, by
contrast, the learner has no intention to produce what he/she has produced and have the
ability to correct it when his/her attention is drawn to it, then it is a mistake.
14
As with regards to intentionality and errors, Taylor (1986:154) also seems to fall in
agreement with James (ibid) but he distinguishes two sorts of intentions. He points out that
“the only way we can reasonably determine whether a mistake is a slip or a genuine error is
by reference to the writer‟s semantic and structural intentions.” Here Taylor clearly
distinguishes between what the learner wants to say (semantic intention) and how he
expresses it (structural intention).
Duskova (1983:219) seems to disagree with the claim that all mistakes are a product of
failures of performance and are self-corrigible by their authors as James (1998:78) and
Corder (1967) view them. He insists that these features do not apply to all „nonce‟ mistakes.
He also makes another observation regarding the established correspondence between errors,
as earlier defined as a result of failures of competence, and defect in knowledge. He states
that:
“Many of the recurrent errors of systemic character, which we might be inclined to
describe as errors in competence, reflect no real defect in knowledge, since most
learners know the pertinent rule and can readily apply it, but the mechanism of
application does not yet work automatically”(P.219).
We can conclude then that errors, as James (1998:79) puts it, are everything that mistakes are
not: they do reflect knowledge; they are not self-correctable; only learners of L2 make them;
and they are of great significance. The significance of errors will be our next section.
2.3. Significance of Errors:
The study of learners‟ errors has been a primary focus of L2 researchers during the last three
decades. Since the emergence of Corder‟s article on the significance of learners‟ errors and
his arguments for what errors can provide for both research and pedagogy, researchers and
teachers in various part of the world have spent countless hours extracting, counting, and
15
classifying errors from students‟ compositions and conversations using them as a base for
theory construction and classroom practice. If such extraordinary interest in learners‟ errors
reflects something, it would certainly reflect the importance of errors as an indicator of how
languages are learnt and at what rate.
According to Dulay at al (1982:138), the study of learners‟ errors serves two major purposes:
(a) they provide data from which inferences about the nature of the language learning process
can be made; and (b) they indicate to teachers and curriculum developers which part of the
target language students have most difficulty producing correctly. As a matter of fact, the
Error Analysis movement, as we shall see in (2.4) below, has yielded insights into the L2
acquisition process that have stimulated major changes in the teaching practices, (ibid).
As it has been clearly and evidently established in the previous section (see 2.2) that errors,
not mistakes, reveal much about the ongoing process of second/foreign language learning, it
is vital to emphasise the fact that teachers, researchers and error analysts who are interested
in the pedagogical implications of EA should concentrate more on learners‟ competence
errors than on errors of performance (mistakes according to Corder 1967). This is due to the
fact that it is the former type of errors which reveal the actual learners‟ difficulties. It is
fruitful to quote Corder (1974) who, with respects to the significant of learners‟ competence
errors argues that
“They are significant in three different ways. First to the teacher, in that they tell
him, if he undertakes a systematic analysis, how far towards the goal the learner has
progressed, and, consequently, what remains for him to learn. Secondly, they provide
to the researcher evidence of how language is learnt or acquired …Thirdly, (and in a
sense this is their most important aspect) they are indispensable for the learner
himself, because we can regard the making of errors as a device the learner uses in
order to learn” (P.25).
16
What this entails, then, that a careful analysis of errors is an indicator for both teachers and
learners of what has been learned and what still needs to be worked on and what not yet
being acquired. Additionally, they are a way learners have of testing their hypotheses about
the nature of the target language by comparing their output to the teacher‟s or other
competent users.
After the introduction of the concept „interlanguage‟ into the literature by Selinker (1972),
much interest has been devoted to the study of learners‟ errors as they permit the formulation
of rules for learners‟ interlingual systems and provide data on the nature and significance of
the of the obstacles that lie in the path towards discovery of the target language rules
(Richards 1974:IX).
It therefore follows that a careful analysis or learners‟ errors enriches our knowledge on the
nature, the process and the steps of second/foreign language acquisition and consequently on
the planning of courses incorporating the psychology of second/foreign language learning.
The experience and observation have shown that teachers do rely on their learners‟ errors to
evaluate the effectiveness of their teaching as well as to discover what types of obstacles their
learners seem to encounter and use this knowledge, that error analysis provides, to design
remedial work for their students.
However, if full advantage of error analysis is sought, our learners not only need to be
encouraged to observe their own errors, but to reflect on these errors and try to correct them.
In other words, our learners need to be encouraged to develop a „reflective approach‟ in
dealing with errors.
17
In the next section, I will explore the Error Analysis movement highlighting the climate that
helped it arise and its contributions to the understanding of the nature of learners‟ error and
their sources.
2.4. The emergence of Error Analysis movement and Interlanguage
studies
At least two major approaches to the study of second language learners‟ errors and error
explanation can be distinguished: Contrastive Analysis and Error Analysis. It is fruitful to
draw the attention at this point, however, that the thrust of the discussion here will be devoted
to the EA movement as it is the scope of the present section and frames the methodology of
the present study as a whole. The CA movement will be referred to as the background and
climate that the EA movement emerged in.
During the fifties and early sixties, the predominant approach to the study of learners‟ errors
was the CA paradigm. Researchers, linguists and teachers were attracted at that time by the
Behaviourists‟ as well as the structuralists‟ attitudes to errors who viewed errors as a
symptom of ineffective teaching or as evidence of failure. Errors were also viewed as mainly
due to the mother-tongue interference, which teachers, researchers and course designers have
failed to predict and account for. Errors could be eradicated completely by providing
learners with the correct forms only. The prevailing idea was that these errors were thought
of as a result of mainly the mother-tongue interference and a careful comparison between
learners‟ native language and the target language in question, would reveal the area of
differences between the two languages. All errors were believed to result from such
differences between the two languages. The absence of any theoretical framework for
explaining the role played by errors in the process of SLA led to no serious attempt at that
time to analyse learners‟ error and to account for other potential sources of errors besides
18
mother-tongue interference (Ellis, 1986:51). In such climate, the interest in CA had grown
and become the working approach to second/foreign language teaching.
By late 1960s, the need has emerged for more research on learners‟ errors. Such need was
stimulated by the fact that some errors were found unexplainable by CA. For example, some
errors were found to represent the TL rule complexity than interference from learners‟ MT.
Matter (1989:77) cites the following two examples of errors made by Arabic-speaking ESL
learners that are neither explainable by CA nor traceable to the learners‟ MT: “if we wanted
to visited the fort…” and “my father is work in a bank…”. Interestingly, such errors were
also found to be made by children as they are acquiring their first language, of course no
interference from other languages is possible in this case. It has become evident that mother-
tongue interference is not the only source of errors in SLA process and so there should be an
alternative approach which accounts for all types of learners‟ errors and explains their
potential sources. By the early 1970s, however, some misgivings about the reliability of CA
began to voice mainly because of its association with an outdated model of language
description (Structuralism) and a discredited learning theory (Behaviourism) (James, 1998:4).
In such climate the enthusiasm for CA has declined and more interest in analysing learners‟
errors has increased.
According to Ellis (1986:51), it was not until the late 1960s that there was a resurgence of
interest in EA. It was Corder (1967, 1971, 1978, etc.) who marked such resurgence by a
series of articles which helped to give EA its direction. According to Corder (1975:64), “EA
can be seen to serves two related but distinct functions: pedagogical, „applied‟, and the other
theoretical, leading to a better understanding of second language learning processes and
strategies”. Put differently, the former function is classroom-oriented, while the latter is
research-bound. More specifically, three principal goals can be identified for carrying out an
19
EA. Firstly, EA is carried out to identify strategies which learners use when learning a
second/foreign language. Secondly, it also carried out in an attempt to identify the causes of
learners‟ errors. Finally, errors are analysed to obtain information on common difficulties in
language learning, as an aid to teaching or in the preparation of teaching materials.
The procedure for EA is spelled out in Corder (1975). They are summarised by Ellis
(1986:51-2) as follows. (1) A corpus of learners‟ language is selected taken into account the
size of the sample, the medium to be sampled and the homogeneity of learners (their age, sex,
L1 background, etc.). (2) The errors in the corpus are identified. In this very step, errors,
which, by definition, are the result of lack of competence, are to be distinguished from lapses
and mistakes which are the result of some neurophysiological breakdown or processing
limitations (Corder, 1975:63). (3) The errors are classified. Then, (4) the errors are
explained. In this last stage, the errors are related to their possible psycholinguistic causes as
proposed by Selinker 1972 (see below). Finally, (5) the errors are evaluated. This stage
involves assessing the seriousness of each error in order to make decisions about teaching
priories.
As with regards to error causes, Selinker (1972:37) proposes five processes that could
account for learners‟ errors. These are: language transfer, if the errors are due to the L1
interference; transfer of training, if the erroneous items are a result of identifiable items in
training procedures; strategies of second language learning, if the identified errors are a
result of some kind of strategy applied by the learners when presented with learning materials
of any kind; strategies of second language communication, if the identified erroneous
items are due to learners‟ endeavour to communicate with native speakers of the target
language; and finally overgeneralisation of TL linguistic material, if they are traceable
back to a clear overgeneralisation of TL rules and semantic features.
20
Thus, it is the result of the significant shift that took place in the late1960s and early 1970s
that such different sources of errors have become evident. Another result of such shift has
been the change in researchers‟ and practising teachers‟ attitudes towards errors and their
sources which has resulted in intensive research projects and studies devoted to the study of
learners‟ interlanguage in an attempt to discover its features and test hypotheses about
second/foreign language acquisition. Early on, Corder (1967) introduced the concept
“transitional competence” into the EA literature followed by “idiosyncratic dialect” (ID) in
1971 marking the beginning of a new paradigm in studying learners‟ language. He argues
that all language learners, during their learning careers, develop a special sort of dialect
which is shared by the same learners having similar L1 and is different from dialects of
learners from different L1 background (Corder, 1971: 158). In the same year, however,
Nemser (1971:55) suggested that learners employ an approximative deviant linguistic system
in their attempt to utilise the target language. Such approximative systems vary in character
in accordance with proficiency level. They are also transient and represent the current stage
of the learners‟ stage of proficiency in the TL. Nemser‟s (1971) approximative system seems
a resemblance of Corder‟s (1971) idiosyncratic dialect. A year later, Selinker (1972:35)
introduced the term interlanguage to describe the learners‟ developing linguistic system
which results from learners‟ attempted production of the TL norm. Among the above three
terms, the term „interlanguage‟, needless to say, is most frequent and recurrent in the
literature to refer to the learners‟ “version” of the TL. It is clear, thus, that the emergence of
the EA movement has paved the way to research and invited researchers to better understand
learners‟ interlanguage and to account for potential learning difficulties.
21
By the late 1970s, researchers began to view learners‟ interlanguage as a continuum
stretching from MT to TL, but the case for such a continuum has been largely theoretical
(Ellis, 1986:54).
Generally speaking, there are two types of studies that have been devised to investigate
learners‟ interlanguage: longitudinal and cross-sectional. These studies are aimed at to
investigate learners‟ interlanguage, which is the major aim for EA. Much insight has been
gained from these studies into the nature and process of SLA as well as the potential error
causes (as oppose to the single or prime cause of error claimed by the CA). According to
Ellis (1986:52-3), EA studies provide two kinds of information about learners‟ interlanguage.
The first concerns the linguistic type of errors produced by L2 learners (a discussion and
classification of error types are detailed in section 2.5 below). The second type of
information, however, concerns the psycholinguistic types of errors. This second type of
errors refers to the strategies that learners employ not only to learning a SL but also to
overcome obstacles in the course of their learning such as avoidance strategies and
overgeneralisation.
To sum up, the EA movement has succeeded in shifting the ground of SLA research. It has
succeeded in providing convincing and evident explanation of the SL learners‟ errors. It is
fruitful to quote Ellis (1986:53) as with regards to EA contributions. He states that:
“The most significant contribution of Error Analysis, apart from the role it played in
the reassessment of the Contrastive Analysis Hypothesis, lies in its success in
evaluating the status of errors from undesirability to that of a guide to the inner
workings of the language learning process. As a result of interlanguage theory and
the evidence accumulated from Error Analysis, errors were no longer seen as
„unwanted forms‟”.
22
The next section will be on the different error types and subtypes as well as different
classifications of these errors that have been examined and pointed out by different
researchers in EA field.
2.5. Types and classifications of errors:
In trying to make sense of the literature on EA, I have found it of great value, at this stage, to
shed the light on different types of errors and various attempts to classify them. It is
interesting to note in this regard, though, that no real attempts had been made to categorise
nor to classify learners‟ errors during the 1950s and the early 1960s when the CA movement
was the predominant practice in the field of EFL/ESL since it was the belief that there was
only one type of error: those resulting from mother tongue interference. With the emergence
of the EA movement in the late 1960s, however, and with the increasing number of
interlanguage studies, the need to classify learners‟ errors according to their various sources
has brought itself to the surface. As I have noted earlier (see section 2.4 above), the
classification of errors has been a prime goal for EA after the identification of their sources.
Errors have been labelled and classified in various ways depending on (a) the type of study
conducted, (b) the aim and focus of study, (c) the approach followed by researcher, and (d)
according to the level of analysis.
The early attempts to classify errors tend to categorise errors based solely on their sources as
interlingual, intralingual or induced errors, as oppose to the more recent ones which account
for the linguistic analysis of the error itself. Early on, Schumann and Stenson (1974:4)
divided students‟ errors into three types: errors due to incomplete acquisition of the target
language grammar (which may in turn be due to either inter- or intra-lingual difficulties),
errors due to the exigencies of the teaching/learning situation, and errors due to normal
23
problems of language performance. As simple and inadequate this classification may appear,
it forms the bases for the forthcoming categorisations by other researchers who seems to
draw their classification on this early one, as we shall see.
Generally speaking, four types of error classifications can be identified in the literature of
EA. They are, as Dulay et al, (1982: 146) present them: (1) linguistic categories; (2) surface
strategy; (3) comparative analysis; and (4) communicative effect. In the following lines, I
shall consider each of them.
2.5.1. Linguistic Category Taxonomy:
Taxonomies based on the linguistic categories of errors concern with a description of the
language constituents that are affected by errors. Put differently, this taxonomy classifies
errors “according to the particular linguistic constituent that the error affect” (Dulay et al,
1982:146). It first indicates on what level (or constituent) the error is located: in phonology,
syntax, morphology, semantics, lexicon, or discourse. Then, within the level specified, it
indicates the category in which the error occurs. For instance, if an error is located in the
grammar level, the next step is to specify the category of grammar. That might include the
auxiliary system, passives, etc. As a refinement of Dulay‟s et al system, James (1998:105)
suggests that after having established the level in which the error occurs, the class, rank and
system that the error affect then need to be specified. For instance, an error may occur at the
grammar level, involving the class noun, the rank morpheme, and the system of countability.
According to James (ibid.), this framework, after the refinement, is useful and can handle the
errors of relatively advanced learners (P.105). I have to agree with James on his sub-
categories as they prove useful in accounting for various types of error which can be included
here. It is also of value to note here that curriculum developers have long used linguistic
24
category taxonomies to organise their language lessons in student textbooks. “Such materials
permit teachers and students to feel that they have covered certain aspects of the language in
their classes” (Dulay, et al, 1982:147).
2.5.2. Surface Strategy Taxonomy:
This is the second type of taxonomies proposed by Dulay et al (1982). It highlights the ways
the surface structures of the TL are altered. The types of error under this taxonomy include:
errors of omission, addition, misformation and misordering. Two convincing arguments
made by James (1998) as with regards to this taxonomy. In the first place, the label „surface
structure‟ seems unsatisfactory since the surface:deep distinction is immaterial and there
being no „deep‟ structure taxonomy. He suggests “Target Modification” as an alternative
label. Secondly, it is quite misleading to say that learners „alter‟ the correct form of TL so as
to produce an error. I am quite in agreement with James as it is clear that „alter‟ implies as if
the learners knew the correct form all along. This taxonomy is based on the ways surface
structures are alerted or „modified‟ from the TL structure. This modification takes one or
more of the following four processes:
2.5.2.1. Omission
These types of error are characterised by “the absence of an item that must appear in a well-
formed utterance [and/or written form]” (Dulay, et al, 1982:154). Observation shows that
some types of words and morphemes are omitted more than others. For example, studies
have found that learners in their early L2 acquisition stages tend to omit function words
25
(grammatical morphemes such as articles, auxiliaries, -s, etc.) more than content ones. More
advanced learners tend to be aware of their ignorance of the content words and so they resort
to „compensation strategies‟ to get their message across rather than omit one (James,
1998:107).
2.5.2.2. Addition
An addition error type occurs when learners include an item which must not appear in a well-
formed utterance and or written form. Dulay, et al (1982:156) suggest that addition errors
occur in the later stages of L2 acquisition and are the result of “the all-too-faithful use of
certain rules”. They also suggest subtypes of addition errors. First, regularisation, which
involves overlooking exceptions and spreading rules to domains where they do not apply, for
instance producing the regular (in the learner‟s perspective) buyed for the irregular bought.
James (1998:107) suggests that some learners also „irregularise‟ some rules when learners
wrongly assume that certain forms might be exceptions to regular ones and he gives dove (as
the past tense of dive) as an example, although dove is used as past tense of dive in American
English. The second subtype of addition is double marking defined as “failure to delete
certain items which are required in some linguistic constructions but not in others” (Dulay, et
al, 1982:156). Examples of this subtype would be, „He doesn‟t knows me‟ and „I didn‟t went
with them‟. The last subtype of addition is simple addition and it caters for all additions not
described as double marking or regularisation.
2.5.2.3. Misformation
26
This is the third category of Dulay‟s et al (1982) surface strategy taxonomy. They define it
as “the use of the wrong form of the morpheme or structure” (P.158). They also identify
three subtypes of misformation. The first subtype is regularisation errors, they occur when
a regular marker is used in place of an irregular one as in „runned‟ for ran and „gooses‟ for
geese. The second is archi-forms, they occur when learners select one member of a class of
forms to represent others in the class such as the use of „that‟ out of the set
this/that/those/these. James (1998:108) suggest the term misselection when refer to this type
of errors, which is a reasonable suggestion. The third subtype is alterating forms, this
occurs when learners freely alternate various members of a class with each other. For
example, a learner may use „I seen her yesterday‟ for „I saw…‟ or „I have already saw him‟
for „I have already seen him‟.
2.5.2.4. Misordering
Part of the linguistic competence, in addition to selecting the right forms to use in the right
context, is to arrange them in the right order (James, 1998:110). Misordering errors occur
when learners misarrange or misplace words in sentences. Examples of this type are: „He
always comes late home‟ and „What Daddy is doing?‟ According to Dulay‟s, et al
(1982:163) observation, learners often produce this type of errors when carrying out a word-
for-word translation of their native language surface structure.
2.5.2.5. Blends
This category has been added by James (1998) as a complementary category to Dulay‟s et al
(1982) surface strategy taxonomy. This category is “typical of situations where there is not
just one well-formed target, but two … and the learner is undecided about which of these two
targets he has in mind” (James, 1998:111). An error of this type occurs when there are two
27
possible semantically related structures and either of which could serve the speaker or
writer‟s purpose but he/she fails to make a clear choice. Instead, they combine a part of each
to produce a structure which characterises both (ibid.) as in „a typical Indian meal comprises
of‟‟. This error is a result of combing the following well-formed structures: „a typical Indian
meal comprises rice, …‟ and „a typical Indian meal is comprised of rice, ..‟
To summarise so far, Dulay, et al (1982) offer a useful descriptive taxonomy based on a
comparison of the forms learners use with forms used by native speakers in the same
situation. As with regards to the value of such taxonomy it is fruitful to quote Dulay, et al
(1982:150) as they defend their taxonomy. They wrote:
Analysing errors from a surface strategy perspective holds much promise for
researchers concerned with identifying cognitive processes that underlie the learner‟s
reconstruction of the new language. It also makes us aware that learners‟ errors are
based on some logic”.
Having presented Dulay‟s, et al (1982) linguistic and surface strategy taxonomies, I have
found it also of value at this stage to shed the light on another linguistic taxonomy proposed
by James (1998:129). In the following few lines, I shall briefly present the types and
subtypes of this taxonomy.
2.5.3 Linguistic Taxonomy Based on Levels of Error
James (1998) based his taxonomy on three principal levels of language: the levels of
substance, text and discourse. The first two are of interest here and I shall discuss them
presently. It seems of great importance, though, to draw the attention to the fact that only
writing errors will be considered in this taxonomy, speaking errors are out of the scope of the
present study.
28
2.5.3.1. Substance Errors
This category caters for errors of misspelling and punctuation errors. These errors are the
result of either wrong transfer of L1 spelling and punctuation rules into L2 or the arbitrary
nature of L2 spelling and punctuation rules. Interestingly enough, errors resulting out of the
second source, are found similar to those made by NS of a given TL.
2.5.3.2. Text Errors
Text errors are divided into lexical errors and grammar errors. According to James
(1998:142), “text errors arise from ignorance and misapplication of the „lexico-grammatical‟
rules of the language”.
2.5.3.3. Lexical Errors
Lexical errors are of great interest for both error analysts and teachers. In the recent years,
however, lexis has enjoyed a special status in the ESL/EFL research because of the central
role that lexis (vocabulary) plays in L2 acquisition. By analysing learners‟ lexical errors, it
has become possible to discover how vocabulary is acquired and what strategy learners apply
in learning them. According to James (1998:143), lexical errors are found to be the most
frequent category of errors. Thus, there are therefore good reasons to undertake lexical EA.
Such EA studies have revealed the following categories of lexical errors. First, misselection,
these errors result when learners misselect the right form of a word because there are many
words which are similar. For instance, learners usually confuse prize with price and
stationery with stationary. James (ibid.) gives the following erroneous sentences to
exemplify the point in question: „He wanted to cancel his guilt‟ (conceal), „It was a genius
29
diamond‟ (genuine). The second category is misformation. These types of errors emerge
when learners either borrow and/or literally translate from their L1 or coin new words when
they do not have the right word to use. The third type is distortion. While the misformation
errors are mainly „interlingual‟, distortion errors are „intralingual‟. Learners may omit a
morpheme (e.g. int(e)resting), overinclude one (e.g. din(n)ing room), misselect (e.g.
delitouse/delicious), misorder (e.g. littel/little), or even blend (e.g. the deepth of the ocean
[depth+deep]). The fourth type of lexical errors is semantic errors, they occur when learners
confuse the meanings and uses of lexical terms (words). For instance, learners may use a
more general term when a more specific one is demand by the context (e.g., Capitalism made
America big [powerful]), or the reverse (e.g. all villagers vote for their candidate [citizens]).
They may also use a word in a wrong collocation (e.g. tall hair (long), make a try (have)).
2.5.3.4. Grammar Errors
According to James (1998:154), Grammar errors can occur in two levels of language:
morphology and syntax. Morphology errors are of different types and occur in almost all the
five lexical word types: nouns, verbs, adjectives, adverbs and prepositions. Learners are
found to make noun morphology errors such as omitting the plural –s as in „four book‟
(books) and overincluding the plural –s as in „every teachers‟ (teacher) and confusing the
form:sound relationship of some morphemes as in „menshen‟ (mention). Verb morphology
errors include: (1) omission or overinclusion of third personal singular –s as in „he teach
English‟ (teaches) and „students in my class organises trips‟ (organise); (2) confusion with
verb endings as in „I was write‟ (writing); (3) misselection of the correct verb form that is
demanded by context as in „we go shopping last weekend‟ (went); and (4) confusion with
the regular and irregular verbs as in „I bringed all my papers‟ (brought) and „he dove into
the water‟ (dived). The following erroneous sentences exemplify the two types of adverb
30
morphology errors: „I will visit you soonly‟ (soon); „I am particular interested in..‟
(particularly). Finally, error like „a colourfuller scene‟ (colourful), exemplifies the adjective
morphology errors.
Let us now move to the second type of grammar error, namely, syntax errors. These are
errors that affect texts larger than the word level, namely, phrase, clause, sentence and
ultimately paragraphs, James (1998:156). Errors at the phrase level turn out to be quite
problematic since an error can occur in a prepositional phrase (PP), adjective phrase (AP),
noun phrase (NP), verb phrase (VP) and these are found in a „larger‟ noun phrase (NP). In
this case an (AP) error can also be an (NP) one at the same time. Some possible errors at this
level are: wrong article use, article omission, article overinclusion, misordering of modifiers
within the (NP) and a combination of these. Clause errors involve the ways in which phrase
- themselves well-formed – operate in clauses (James, ibid.). Five possible errors are
identified here: (1) the phrase in question may be overused in the cause, (2) it is omitted, (3)
it is misordered, (4) it is misselected, and (5) it is a blend of two or more of the four. Errors
at the sentence level involve the selection, order and combination of clauses into larger units
and so possible errors would result from either misselection, misorder and/or wrong
combination of phrases in the sentence in question. Finally, errors at the paragraph level
involve the misuse and/or omission of cohesive devices that connect sentences and ideas
within the paragraph to produce a cohesive text.
So for I have presented three linguistic taxonomies. The first two were proposed by Dulay, et
al (1982) while the third is proposed by James (1998). It is useful to note here that James
(1998) seems to draw upon Dulay‟s, et al (1982) taxonomy which is obvious in the use of
nearly the same labels for error types and subtypes. I shall now proceed to present the two
remaining taxonomies of Dulay, et al (1982).
31
2.5.4. Comparative Analysis Taxonomy
This is the third type of taxonomies presented by Dulay, et al (1982). The classification of
errors in a comparative taxonomy is based on “comparisons between the structure of L2
errors and certain other types of constructions” (ibid.:163). In EA literature, L2 learners‟
errors have most been compared to errors made by children as they acquire their L1, and to
equivalent phrases or sentences in learners‟ mother tongue. The former comparison yields in
„developmental errors‟ while the latter yields in „interlingual errors‟. Yet, there still remain
errors that could be classified equally well under both of the previously mentioned categories
(these are labelled „ambiguous errors‟) and others that are not of either type (these are
labelled „other errors‟). For the sake of brevity, the reader is referred to Dulay, et al
(1982:163-188) for more details on this taxonomy.
2.5.5. Communicative Effect Taxonomy
The type of errors classified under this taxonomy are those perceived as having effect on the
listener or reader. It therefore distinguishes those errors which cause miscommunication and
those do not (Dulay, et al 1982:189). Research has shown that certain types of errors make a
critical difference as to whether or not the listener or reader comprehends the speaker or
writer‟s intended message. A distinction was then made by Burt and Kiparsky (1975)
between two types of errors according to their domain of effect: global errors, which affect
the overall sentence organisation; and local errors which affect single element or constituent
in the sentence (ibid.:73). It has been found that global errors, as in „English language use
many people‟, significantly hinder successful communication while local errors, as in „He
very busy now‟ usually do not hinder communication.
32
In summary, so far, I have attempted to present different taxonomies that account for various
types of errors that ESL/EFL learners make due to different reasons. However, I believe that
these taxonomy, while useful, are still inadequately sufficient enough to account for other
types of errors such as induced errors and errors caused by avoidance behaviour of some
learners. Induced errors are those caused by either wrong introduction, misleading
presentation order, wrong definition, wrong elicitation techniques, etc. of language items in
classrooms (Stenson, 1974). Latter on, Kleinmann (1983) conducted a study to examine the
avoidance behaviour in adult second language acquisition. He found that some learners do
avoid using particular structures as to not make errors since these constitute great difficulty
for them.
At last, it is worth mentioning that the practice of error classification, while useful, is never
all easy; a considerable number of errors could not be classified at all. Duskova (1983), in a
study conducted to examine the errors of fifty Czech adult learners, reported that “of the 48
errors made in word order, it was possible to divide 31 into four subgroups; the remaining 17,
however, defied all attempts at classification, being unique in character, nonrecurrent, and not
readily traceable to their sources”, (P.218).
2.6. Error Frequency, Error Density and Error Gravity
Having examined different error types, levels and classifications, it would prove fruitful to
examine some error features such as frequency, density and gravity, which will be presented
respectively in the present section.
2.6.1. Error Frequency
33
When conducting EA, one of the important stages is to count the number of errors and error
types in a given corpus in order to arrive at some statistical differences between groups of
learners in terms of the number and types of errors each group has made. Such information
gained from token-count of errors serves two related purposes: theoretical and pedagogical.
Theoretically, knowing what error type constitutes the most frequent among other types and
their relative proportion of a given group(s) of learners allows researchers account for areas
of difficulties for these learners and what items of the TL happen to be acquired before others
(late versus early acquisition). Practically, on the other hand, it would be useful for teachers
to know what area of the language turns out to be problematic for their learners and so take
decisions about their teaching priority and what needs to be focused on.
Saying that, however, the question of whether we should count error types or error tokens
(the number of occurrence of the same error) arises as with regards to the issue of counting
errors. Let us consider the following examples cited in James (1998:117): „He play tennis
very well but football he play badly‟, „She play tennis very well and also she swim very
well‟. In the first example, the second „play‟ is obviously a lexical replica of the first and so,
according to James (ibid.) we consider them as one error. In the second, by contrast, there
are two errors since „play‟ and „swim‟ are not lexical replicas. This criteria, I believe, is not
very satisfactory since the cognitive process underlying the two errors is plainly the same and
this very learner failed to apply the same rule namely, the verb takes an -s when the subject is
singular in the present time.
2.6.2. Error Density
While frequency concerns with how many times the same error is repeated in a given text
produced by a learner, density is calculated by counting how many different errors occur per
34
unit of text (James, 1998:211). Having set the difference between frequency and density, it
is fruitful to note that texts with a high error density present the reader with a greater problem
than with error frequency (ibid.). That is, because in a high frequency error text, when the
same error type has been encountered once or twice, the reader would learn to accommodate
it and adjust his/her reading accordingly, but in texts with a high error density, on the other
hand, readers are encountered with a new error type every time which makes it really difficult
to fully comprehend the intended meaning.
2.6.3. Error Gravity (EG)
Intuitively, not all errors have the same effect on communication and thus errors vary in
terms of “seriousness”. Error gravity then is a measure of the effect that learners‟ errors have
on communication. There are different factors affecting error gravity. James (1998:206)
presents the following criteria for EG: error frequency and density (mentioned above),
grammaticality, comprehensibility, communicativity and noticeability.
Experience and observation show that teachers differ in their judgement of learners‟ errors.
It is hypothesised that teachers‟ error gravity judgement would tend to reflect the frequency
with which a given error occurs (Mattar, 1989:383). This, however, turns out to be
ungereralisable since for other teachers, particular error types are more gravy than others
regardless to frequency (see for instance Roberta, et al 1984).
2.7. Common errors made by Arab EFL/ESL learners
Coincidentally with the aims of the present EA study, which investigates the common errors
made by Omani ESL learners, I found it fruitful to shed some light on the common errors
35
made by EFL/ESL Arab learners as to contrast them with those made by their Omani
counterparts.
One of the EA studies that the present research is drawn upon is a cross-sectional study
conducted by Mattar (1989) which is aimed at investigating writing errors made by adult EFL
learners in Bahrain. Among the many findings of his study, and what concerns us most in
this section, is the following list of the most common areas of difficulty that Bahraini learners
are found to face when learning Englsh:
1. The passive voice construction;
2. Question formation;
3. The English relative clauses;
4. Articles;
5. Auxiliary verbs (mainly auxiliary /be/;
6. The copula;
7. Prepositions;
8. Verb usage (mainly wrong tense sequence across clauses)
(Mattar, 1989:385)
Similar findings were also reported by Mukattash (1984:354) who examined the avoidance
strategies of (4835) Jordanian students. He stated that “Arabs have a tendency to use active
sentences even in contexts that require the use of the passive” (P.345).
In his 1986‟s study, Mukattash reported three major error types of adult Jordanian students.
They are as follows: (1) errors in the verbal system, these include: tense, phase, aspect, voice,
and be-deletion; (2) errors in relative clause formation, these include: relative pronoun
deletion, pronominal reflexes, relative pronoun replacement, and non-restrictive clauses; (3)
36
new error types which include: use of „that‟ in sentence initial position, placement of the
„subordinate clause‟ after the subject (NP), subject repetition, and „whichz‟-deletion.
In short, the above mentioned studies along with many others (e.g. Mukattash, 1981; Scott
and Tucker, 1974; Schachter, 1983) have shown different error types with various
frequencies of occurrence. It would be of great interest, however, to see how the findings of
the present EA study would vary (and/or support) the findings of these studies.
2.8. Error Correction (EC) and Teacher response to students’ writing
errors
As with reference to the third study question (set out in chapter 1), it would be of value to
make sense of the literature on the different approaches to error correction and the
effectiveness of different ways teachers employ in responding to their students‟ writing
errors.
A number of recent studies have brought attention to the central role that error correction
plays in the process of second language learning (e.g., Long, 1983; Chaudron, 1983;
Hedgcock and Lefkowitz, 1994; Johnson 1988; etc.). The principal assumption underlying
these studies has been twofold: to identify how and when teachers should correct their
learners‟ errors and mistakes and how effective error correction is. According to James
(1998:235), “one of the purposes of doing EA is to identify the principles which should guide
effective error correction. Principled EC is applied EA”.
Three levels of corrections can be identified in the literature: feedback, correction and
remediation (James, ibid.). Feedback involves teachers to simply inform the learners that
there is an error and leaving them to discover it and correct it themselves. Correction, by
37
contrast, involves providing the learner with the necessary information or treatment that
would lead them to correct their errors. Here, the teacher may specify how and where,
suggest alternative and give hints that lead to the correct form. In the remedation level, the
teacher does not simply identify the error and suggest ways of correcting it, but provides
comprehensible information that allow learners to revise or reject the wrong rule they are
operating with when produced the error. In other words, encourage them to reconstruct their
mental representation of the TL rule so that the error type or token does not recur.
Hammerly (1991: Ch.9) refers to these levels by suggesting a distinction between deep and
surface correction. For him, surface correction involves mare editing or putting right but
does not address the source of the problem. The deep correction, by contrast, addresses the
source of the error, gives students clear picture of why a bit of language is the way it is and
helps learners to “reorganise their cognitive structures” (ibid.:93).
The many studies that have been conducted to date have shown that EC is necessary and has
proved useful in improving the quality of students‟ writing (see for example Fathman and
Whalley, 1990). However, there are a number of principles that need to be considered for an
effective EC. I shall highlight the most relevant three presently. Firstly, EC should be
effective. For EC to be effective, teachers, according to James (1998:249), need to use
correction techniques that bring about improvements in accuracy and at the same time
efficient, i.e., require less time and effort in the teacher‟s part and have the desired effect on
students. Secondly, EC should be sensitive. This is an important principle since not all
learners react similarly positive to their teachers‟ correction. Many authors have suggested
an individualised EC or a whole-class correction. This would definitely personalise EC and
makes it a non-threatening experience. Both Hammerly (1991) and Edge (1989) emphasise
the notion of a „humanised‟ EC. „Conferencing‟ is another technique to avoid threatening EC
38
and to encourage student-teacher active interaction and cooperation. This last notion of
cooperation between teachers and learners has also been pointed out by Dheram (1995:160).
Thirdly, EC should match students’ preferences. Studies on EC have indicated that
students‟ preferences for certain type of correction cannot be ignored. Hedgcock and
Lefkowtz (1994), for instance, found that EFL learners prefer a form-focused feedback on
their writing while EEL learners prefer a content-oriented correction. Cohen and Cavalcanti
(1990:175) in another study noted that a clear agreement between teacher and student as to
what will be commented on and how is a missing ingredient of a typical approach to EC.
In short, EC has gained a special interest among researchers who have investigated different
ways of providing feedback to students‟ writing. Studies on EC suggest that EC is required
by learners themselves and essential especially in those cases where the language processing
task is difficult and so self-correction, although encouraged, is not possible. Hence, EC
needs to be handled effectively and sensitively.
2.9. Summary
This chapter sets the scene to the present EA study and provides readers with the basic
information on what errors mean, how they are significant and to whom, what their types and
categories are and how they can be best corrected. The chapter has also attempted to make
sense of the literature on the basic terms and metalanguage that are recurrent in the literature
of EA and are essential for the basic understanding of the subject matter.
Errors have been defined in terms of two distinct criteria: (1) in terms of who makes them
and (2) in terms of their deviance from the standard acceptable forms of TL. Errors made by
EFL/ESL learners are apparently different from those produced by their NS counterparts.
The former error type may be referred to as either competence errors or performance errors
39
while the latter group would only produce performance errors (Corder, 1974). Learners‟
interlanguage deviance is also judged according to some measures such as grammaticality,
acceptability and correctness (James, 1998).
As the traditional CA has started to lose its credibility in the late 1960s, researchers have
started to devote more interest to the EA movement which has succeeded in providing valid
explanations to different types of learners‟ errors and have paved the way to the study of
learners‟ interlanguage. These studies have significantly shifted people‟s attitudes towards
errors and have made it possible to classify learners‟ errors in categories and subcategories
according to different criteria such as their sources, the underlying cognitive process and their
effects on communication. Other types of studies have been devoted to study the teachers‟
noticeability of their learners‟ errors and the levels of error gravity some of these errors have
(e.g., Roberta, et al, 1984; Mattar, 1989).
One of the deficiencies in the literature of EA, which has also stimulated the present EA
study, is the lack of studies that exclusively deal with errors made by Arab ESL learners in
general and Omani learners in particular. The available studies (e.g., Mattar, 1989;
Mukattash, 1981, 1984, 1986; Scott and Tucker, 1974; Schachter, 1983) have made good
account of the error types and difficulty areas encountered by Arab learners, but more studies
that deal with variety of learners from different countries and with different dialects would
certainly prove useful.
Lastly, One of the purposes of doing EA is to identify the principles that provide guidelines
for effective error correction. It is widely accepted among authors that arbitrary and
unsystematic correction should be avoided, instead, teachers‟ feedback, as James (1998:249)
40
asserts, should be focused, sensitive, and relatively effective taking into account learners‟
individual differences and some group factors such as level of attainment in the SL/FL.
Chapter Three
Study Design
3.0. Introduction
The present chapter sets out the overall design of the study. It specifies the research method
used, study population, research instruments and procedures for data analysis.
In the literature of EA, three types of studies can be identified: proportion studies, quasi-
proportion studies, and occurrence studies. In proportion studies, errors in an entire body of
speech or writing are classified and counted, enabling the researcher to state in quantitative
terms the relative proportion of each error type. In the quasi-proportion study type, by
contrast, errors are analysed and classified but not counted, permitting qualitative estimates,
but not quantitative statements about the proportion of each error type. The occurrence
studies, the researcher reports only the occurrence of particular error type, but no attempt is
made to address proportion (Dulay, et al 1982:174).
For the present EA study, an approach based on the first type of studies, namely the
proportion studies, have been chosen and designed because these types of studies speak most
directly and precisely to the issues of error proportion and error types and enable a valid and
reliable comparison between groups of learners. They also prove useful and relevant to the
aims and objectives of the present study aims set out in chapter one. The briefly summarised
occurrence of errors, which is offered by the other types of studies, does not serve the aims of
the present EA study.
41
3.1. Research Method
In keeping with the nature and aims of the present study, a proportion case study was found
to best fulfil these aims compared to other research methods such as surveys and
experiments. Bell, 1984 (cited in Dheram 1995:62) asserts that in view of the explanatory
nature of investigation, case study was decided to be the most appropriate research strategy.
It is also the researcher‟s view that a case study approach contributes fresh insights into the
ESL/EFL pedagogy and allows collaborative effort between researchers, teachers and
learners. Case study has been defined in the Longman Dictionary of Language Teaching and
Applied Linguistics (1992:47) as:
“The intensive study of an aspect of behaviour, either at one period in time or over a
long period of time … The case study method provides an opportunity to collect
detailed information which may not be observed using other research techniques …
and is usually based on the assumption that the information gathered on a particular
individual, group, community etc., will also be true of the other individuals, groups or
communities.”
In specific terms, a proportion case study approach was used to investigate the frequencies
and types of writing errors made by first year students from two different levels of
proficiency at Sultan Qaboos University (SQU) of Oman. The detailed description of the
study population will be the focus of the next section.
3. 2. Study Population and Sample
All first year students who were enrolled in the foundation intensive programme in English
(see below) in the Language Centre at SQU together with their eight writing teachers
42
constituted the study population. From these students, a total number of 60 students and 8
teachers was chosen to represent the study sample.
3.2.1. The Student sample
A total of 60 first year students doing their foundation course in English at the university
were chosen by the researcher from the study population as a sample for the present study.
These students were in four different levels or bands, ranging from 1 to 4, according to their
performance in the Placement Test that they sat at the beginning of the academic year. For
the purpose of this study, a sample of 30 students from band 2 and another 30 students from
band 4 (a total of 60 students) were selected as respondents. Students in band 2 were in a low
level of proficiency as opposed to those students in band 4 who were in a relatively higher
proficiency level. The students‟ level of proficiency (the band they belong to) in this context
was decided by their performance in the placement test. The rationale behind choosing
students in these levels was twofold. In the first place, there was a clear difference in the
level of proficiency between students in band 2 and students in band 4 which would
definitely serve the aim of the study – comparing and contrasting the type and frequency of
writing errors made by students in the two different levels of proficiency. Secondly, students
in band 2 and 4 were chosen in particular by the researcher because of the significant
difference in proficiency level between band 2 and 4 as opposed to students in bands 1 and 2,
on one hand, and 3 and 4, on the other. In other words, students in band 1 were very similar,
in terms of level of proficiency, to those in band 2 and likewise there was no significant
difference between students in band 3 and 4 in terms of proficiency level. Therefore, both
bands 1 and 3 students had been excluded when the study was designed for the sake of
significant, reliable and valid results. To be more accurate in the description of the study
43
sample, band 2 students, according to the international standards, were in a pre-intermediate
level while band 4 students were in an upper-intermediate level.
The 60 students representing the study sample were male and female Omani students aged
between 19 and 21 years and coming from different regions of the country. No attempt was
made to divide the students into groups according to their gender since the gender factor was
not a variable in the present study, though, it would be interesting to see how the gender
factor would influence the type and frequency of errors. These students had had twelve years
of formal education and nine years of ELT instruction before the university level.
3.2.2. The Teacher Sample
As stated above, eight teachers were also selected to represent the teacher sample of the
study. These were from various nationalities native and non-native speakers of English and
all of them (except for one) had had more than six-year experience as ESL teachers. These
teachers had been chosen in particular to represent the teacher sample for the sake of valid
and reliable responses to the questionnaire (see below), since they were involved in teaching
writing to students in bands 2 and 4.
3.3. Research Instruments
Two methods were used in the present study for the purpose of data collection from the
selected sample: a writing task and a questionnaire. With reference to the first and second
questions of study stated in chapter one, the purpose of the writing task was to gather
information about the type and frequency of errors made by the students from both groups in
writing. The questionnaire was directed to the writing teachers to elicit further quantitative
data on the commonest error types and qualitative data on the way they respond to the
44
students‟ writing errors, which provided answers to the third question of the study (see
chapter one).
3.3.1. The Writing Task
The writing task was an expository essay. The students were instructed to express their
attitudes as to whether they prefer to continue their study in Oman or abroad after they had
finished their secondary school (see the writing task in appendix 1). The chosen topic for the
writing task had been decided on based on its familiarity to the students in both levels of
proficiency. Additionally, the researcher had the belief that the more students are involved in
the topic and the more use of their personal experience could be made, the more creative their
writing would be. Thus, with the chosen topic, students will find themselves involved in, and
use their personal experience to defend their own opinion towards, the topic.
In order to reduce the students‟ anxiety and write at their best capacity, the writing teachers
who carried out the writing task with their students were instructed, in an explanatory letter
from the researcher, to make it clear to them that the task was not intended to function as a
writing test (see appendix 2 for the teachers‟ letter). In addition, students had been given
time (about 15 minutes) to discuss the topic of the writing task in groups and after that as a
class before they started writing their composition individually. They were given 30 minutes
to write their essays, but extra time were given when required. The rationale for such
procedures was to ensure that students would share ideas with each other and thus enrich
their essays and minimise the possibility that some students might not have much to write
about. This will definitely serve the overall aim of the study – gathering as much of the
students‟ errors in writing as possible. Noticeably, these procedures were followed with
students in both band 2 and 4.
45
The researcher had found the above detailed technique to elicit the actual students‟ errors in
writing the most logical, reliable and valid technique compared to others such as interviews,
questionnaires and document collection. The difficulty with interviews and questionnaire as
techniques for data collection of this kind was that some students were not actually aware of
their errors and thus could not talk about, and/or discuss in writing, their errors if they were
to be interviewed or asked to respond to a questionnaire. In addition, the validity and
reliability of the study design and its methodology would have been questioned had the data
on the students‟ errors collected been based on what the students had produced in the
Placement Test at the beginning of the academic year. In fact, the errors made by the
students in the writing component of the placement test would not precisely represent the
actual errors of the students at the time when the study was conducted bearing in mind that
their linguistic competence, including their writing skills, would have improved since then.
The method was also reliable in a sense that if the students were to write another essay on a
similar topic, the same patterns of errors would be expected. It is worth noting in this regard
that students were not allowed to use dictionaries as they might be devised as a resource for
spelling and word use. This way ensured the „authenticity‟ of the errors collected since the
errors found in the students‟ scripts represented the actual status of the students‟ competence
in English and also enhanced the validity and reliability of the writing task as a data
collection tool to this type of data. In addition, Students in both groups were also given the
same writing task for the sake of valid and reliable comparison of the two groups. In short,
the writing task was valid, by definition, since it elicited what it had been designed for and
reliable since it would elicit nearly the same error types and of the same frequency of
occurrence had it been administered to the same students later in the time.
46
Oppenheim (1992) has stressed the importance of confidentiality, anonymity, and ethicality
in all aspects of any research. In the present study, the students were informed that the
writing task would not be corrected and used by their teachers so as not to worry about its
impact on their course grades. They were also informed that whatever they write would be
treated confidentially and would be only used by the researcher for research purposes.
Names were not required on the scripts so as to ensure the anonymity of writers.
3.3.2. The Teacher Questionnaire
As with regards to the third question of the study (see chapter one), a questionnaire was
designed to be completed by the writing teachers who were involved in teaching writing to
bands 2 and 4 (see a sample of the questionnaire in appendix 3). The purpose of the
questionnaire was threefold. Firstly, to gather additional quantitative data on the most
common errors made by the students in both levels of proficiency from the teachers‟
perspectives. For this purpose, the teachers were asked to list all the error types and
categories made by their students when writing in English. This question yielded extra
quantitative data to that provided by the analysis of the students‟ scripts. The second purpose
of the questionnaire was to measure the teachers‟ reaction towards the same error types (see
hypothesis two in chapter one). In other words, the interest was to find out what error types
constituted the highest gravity rate and so noticed by all teachers (for error gravity, see
chapter two section 2.6.3, page 34). To achieve such purpose, a text produced by one of band
2 students was attached to the questionnaire for teachers to mark. Thirdly, the questionnaire
also gathered information on the method used by teachers when responding to their students‟
errors in writing, which, in turn, answered the third study question. To achieve this purpose,
the teachers were provided by an open-ended question which required them to fully describe
their techniques in responding to their students‟ errors as they mark the students‟ drafts. In
47
addition, by looking at the text marked by the teachers, it was possible to gain some
knowledge on the approach followed by every individual teacher when responding to his/her
students‟ errors. The rationale for using a questionnaire, as opposed to other techniques of
gathering data such as interview, was to give teachers more time to think of their students‟
errors in writing and perhaps consult some of their documents and students‟ drafts, which
would be apparently impossible to do had the teachers being interviewed. In addition to that,
the open-ended questions used in the questionnaire were found to most suit the nature of data
required by the researcher as opposed to the closed questions. It is worth quoting Oppenheim
(1992:112) in his argument in favour of this type of questions. He asserts that:
“Open or free questions are not followed by any kind of choice, and the answers have
to be recorded in full … The chief advantage of the open question is the freedom it
gives to the respondents. Once they have understood the intent of the question, they
can let their thoughts roam freely, unencumbered by a prepared set of replies. We
obtain their ideas in their own language, expressed spontaneously, and this
spontaneity is often extremely worthwhile as a basis for new hypothesis”.
The questionnaire itself was not piloted. However, its validity and reliability was ensured via
various procedures. Every single part was carefully designed to elicit the exact data type
required by the researcher, and teachers would have no difficulty in understanding and
responding to the questions. Furthermore, the researcher made himself available to the
teachers should they have any inquiries regarding the questionnaire. No inquiries were raised
by the teachers which, in turn, enhanced the validity and reliability of the instrument.
As with regards to anonymity and confidentiality, the teachers‟ were not required to put their
names or any other signs which might lead to their identity. Additionally, the confidentiality
of the information they provided was also ensured. It was made clear to the teachers that the
information they write would be treated confidentially and would not be used beyond the
research purposes.
48
In short, the procedures and data collection techniques used in the present study have been
designed in a way that was consistent with, and enabled the fulfilment of, the study aims.
3.4. Data Analysis
After the writing task had been administered to the student sample, all the scripts were
handed to the researcher together with the teacher questionnaires and were brought back to
Bristol to be analysed. The students‟ scripts were first analysed followed by the
questionnaires.
3.4.1. Corpus Analysis
A corpus of 60 scripts written by students from bands 2 and 4 was carefully marked in order
to elicit the number and frequency of errors and error types made by the students from each
group. The scripts of both groups were marked separately to allow for a systematic and valid
comparison of the two groups in terms of error number of occurrence and error type in a way
which best answer the study questions.
The literature contains a number of studies which compare two or more groups of learners in
terms of error frequency and error types. (See for example Mattar 1989, who investigated the
number and nature of errors made by adult EFL learners in Bahrain; Mukattash 1984, who
investigated grammar errors of 4835 Arab learners in Jordan; Scott and Tucker 1974, who
examined the written and spoken language of 22 Arabic-speaking students at the American
University of Beirut). Researchers in these studies and others used different methods to both
elicit and classify students‟ errors. The present EA study, as stated in the introduction in
chapter one, does not attempt to explain learners‟ errors, i.e., trace back each single error type
49
to its source, rather, it describes the nature and type of learners‟ errors, since the two tasks are
not of the same nature and so have to be dealt with separately (Dulay, et al, 1982:145). To be
consistent with the descriptive nature of the present EA study, and for the sake of valid
answers to the study questions, I found it useful to adopt a descriptive taxonomy which
allows for a clear description of errors that affect different linguistic levels such as word,
sentence, paragraph, etc. James‟ (1998) descriptive taxonomy of errors, which was
highlighted in chapter two section (2.5.3), was found to best serve the aims and the analysis
approach of the present study.
Drawing upon James‟ taxonomy, I have developed a similar approach to describe and
classify the errors made by students in both bands 2 and 4. In this approach, errors are
classified based on the level of language that they affect. Two levels are identified:
substance level and text level. Within the substance level, errors are classified as either
misspelling errors or punctuation errors. The text level errors are of two sorts: lexical errors
and grammar errors. The lexical errors may be either semantic, when the used word does not
match the context; misselection, when students misselect a word from a set of words that
differ slightly from each other in terms of form; or distortion, when students fail to form the
word correctly. The grammar errors are of two types: morphology and syntax. The
morphology errors occur in the noun, verb, adjective or adjective, while the syntax affect the
levels of phrase, clause, sentence and paragraph. The third level of analysis in James‟
taxonomy, which is the discourse level, was excluded from the adapted version of the
taxonomy since it deals with errors that are beyond the paragraph level. It is fruitful to note
in this regard that the students who represented the study sample, and especially band two
students, were not yet introduced, at the time when the writing task was administered, to
essay writing which requires the knowledge of connecting paragraphs together and arranging
50
them in a logical order to produce a coherent essay. For this reason I found it of no value to
consider errors that occurred at beyond-paragraph level. The adopted model of analysis can
be best understood visually as a chart like the following one which summarised the error
levels, types and sub-types.
Substance level Text level Blend
misspelling
punctuation
Lexical errors Grammar errors
semantic misselection distortion Morphology syntax
n
o
u
n
v
e
r
b
a
d
j
e
c
t
i
v
e
a
d
v
e
r
b
p
h
r
a
s
e
c
l
a
u
s
e
s
e
n
t
e
n
c
e
p
a
r
a
g
r
a
p
h
Figure1: model of analysis
Using the above detailed model of analysis, band 2 and 4 students‟ scripts were marked
separately and the elicited errors were entered into a grid (see appendix 6) which was
prepared by the researcher based on the categories and sub-categories of the model used to
classify the errors. The use of the grid had facilitated the marking process and made it easier
to draw conclusions about student errors from each band based on the number and types of
error made by each individual student. It is worth noting here that every script was assigned
a number from 1 to 30. The purpose of this numbering system was to enter the script
numbers into the grid matrix under the category or subcategory that the extracted error
belonged to. By using such technique, access to a particular script that contained examples of
a specific error type became easier since the grid provided information on both the error type
51
and its location within the whole corpus. In addition, symbols such as (sp), (pun), (sem),
(miss), (dis), (ph) were also used when marking the scripts. These symbols served two major
purposes: they identified the error location and specified the category and subcategory of the
identified error.
Two major considerations were taken into account when marking the corpus. Firstly, when
marking the students‟ scripts, the focus was on form not on content and organisation. That is,
students were neither penalised for how strong and convincing their arguments were, nor for
how well they had organised and written their papers. Secondly, for each student‟s paper,
error types were counted not error token. In other words, if the same error type made by a
student more than once, then it was counted as one error (for argument on this very issue, see
chapter two section 2.6.1).
3.4.2. Questionnaire Analysis
For the same reason stated above regarding the use of the grid when marking the student
scripts, a different grid was also devised when the teacher questionnaires were analysed (see
appendix 3). The questionnaire grid was developed based on the sections and questions of
the questionnaire. Data extracted from the questionnaires was entered into the grid matrix
horizontally according to the question they belonged to and vertically according to the
questionnaire number. Again this technique made it possible to look at, compare and draw
conclusions on data from eight questionnaires using the grid.
3.5. Summary
52
To sum up so far, the present chapter highlighted the design of the study. It specified the
research method used, study population, research instruments and finally detailed the analysis
approach and procedure.
A case study approach was used to investigate the writing errors made by high proficiency
and low proficiency first year students at Sultan Qaboos University of Oman. To answer the
study questions and fulfil its aims, 60 first year students doing their intensive foundation
course in English at the Language Centre were chosen as a sample to the present EA study.
By means of a questionnaire and writing task, data were gathered on the type and frequency
of errors made by students in band 2 and 4. In addition, information on how teachers respond
to their students‟ writing errors was also gathered by the questionnaire. A model based on
James‟ (1998) taxonomy of errors was developed to classify and describe the errors extracted
from the students‟ scripts. The number and types of these errors will be presented in the next
chapter, which will be devoted to the presentation and discussion of the findings.
53
Chapter Four
Findings and Discussion
4.0. Introduction
In chapter two, I have presented the recent discussion and arguments in the field of EA
together with the findings of various studies in the field which both informed the design and
the analysis procedures of the present study, which was the focus of chapter three. The
present chapter, however, is devoted to report on the findings revealed by the analysis of the
data collected from the field via the means of a writing task and questionnaire.
The two parts that make up the present chapter present and discuss the findings from the two
study instruments, the writing task and the questionnaire. These findings will be presented
separately and respectively in the chapter. In the first part, which occupies two thirds of the
chapter, I shall present and discuss the findings revealed by the analysis of the writing task
administered to the students in both groups, while in the second part, which occupies the last
third of the chapter, I shall deal with the findings revealed by the analysis of the
questionnaire which was filled out by the writing teachers. It is worth noting at this point that
errors are classified and labelled according to the framework developed and explained in the
previous chapter.
54
Before proceeding to the presentation and discussion of the findings, it would be of value, at
this point, to revisit the three study questions and the two underlying hypotheses which
stimulated the present study since reference will be made to these from time to time in the
course our discussion of the findings. The three questions that initiated the study are:
1. What are the most common errors made by high proficiency and low proficiency first
year students in writing?
2. What is the relationship between the students‟ level of proficiency and the type and
frequency of errors they make?
3. How do teachers respond to their students‟ writing errors?
The underlying hypotheses that the findings seek to test are:
The type and frequency of errors will vary with the level of proficiency.
All teachers will respond similarly to the same error types.
4.1. Findings of the Writing Task
To help the reader easily follow up with and digest the dense information presented in this
part, I found it useful to divide the present part into three sections. In the first section, I shall
present and discuss the number, types and subtypes of errors made by band 2 students. I
shall secondly present the number, types and subtypes of errors made by band 4 students. In
the last section, I shall attempt to compare the number of errors and error types of both
groups.
4.1.1. General Observations
55
After the systematic marking of the corpus of 60 texts written by both bands 2 and 4 students,
it was possible to make the following six general observation regarding the test length,
organisation and the nature of errors.
1. Text length
Since there was no limitation imposed on the students regarding the length of the text they
were asked to produce, variation in the text length of bands 2 and 4 was expected. It was
easy to notice, even before beginning to mark the texts, that band 4 students produced longer
texts than did band 2 students. Statistically speaking, an individual band 2 student produced
an average of 180-word long texts, while an individual band 4 student produced an average
of 285-word long texts. In other words, band 4 students produced texts that were 22.6%
longer in words than texts produced by band 2 students. It is clear then that the text length
correlates positively with the students‟ level of proficiency.
2. Text Organisation
Although errors at the organisational level, such as paragraphing, are out of the scope of the
present study, it is worth noting however that band 4 students produced multi-paragraph texts
whereas the majority of band 2 students, on the other hand, produced single-paragraph texts.
3. Classification Difficulty
While classifying errors according to the analysis model explained in chapter three section
(3.4.1), some errors were found to belong to more than one category at the same time. For
instance an error may be classified as a lexical error such as semantic, misselection or
distortion errors, but it can also belong to grammar errors such as morphology or syntax, as in
the following example: “Students who study abroad have to wash clothes and clean desh”.
56
Here, the word „desh‟ could be classified as a distortion error since the vowel „e‟ has been
used instead of „i‟, which is a lexical error. In addition, the structure of the sentence requires
the plural form of the word, which is a noun error belongs to the morphology level. For this
reason, I found it useful to add a third category to the analysis model which have been
developed and explained in chapter three and call it blend. This category contains errors that
can belong to more than one category simultaneously. More examples of this type are given
below. It is worth noting in this regard however that error classification is never an easy task
especially in studies, like the one at hand, which do not examine particular error types.
4. Error Types
No major difference was found regarding the types of errors made by one group compared to
those of the other. In other words, the same error types were found to occur in both groups‟
scripts. The significant difference however was in the number of errors rather than in error
types. I shall reserve the term “error density” to our discussion of the number of errors since
the term often overlaps with another very commonly used one, namely “error frequency”.
The difference between the two terms was highlighted previously in chapter two, (see section
2.6). I stated elsewhere (see page,53) that error types are counted not error tokens. Put
differently, the interest was in the number of times different error type occurs rather than the
number of times the same error is repeated in the text.
5. Reoccurrence of error types in the tests
I also found that each error type is a candidate for reoccurrence in the scripts of both groups
and with the same present frequency rate if the students are to extend the length of their texts.
57
6. The Nature of Errors
The majority of errors made by both groups can be classified as local errors, as opposed to
global ones. This distinction between local and global errors was first introduced into the EA
literature by Burt and Kiparsky (1975) to judge errors according to their effect on
comprehensibility and communicativity, as mentioned earlier in chapter two section (2.5.5).
What that entails is that students are able to express themselves and communicate their
meanings without any hindrance in communication. However, two sentences were found in
two of band 2 students‟ scripts which are incomprehensible and ambiguous. These are: “One
the other hand, I have a look multi-cultral foigen” and “Considiring, if I have alook multi-
cultral forign”. In these two examples, the overall meaning of the two sentences remain
ambiguous even if the misspelled words hade been correctly spelled. One effective way of
dealing with such erroneous sentences is to talk to the students and ask them to explain their
intended meaning. Such strategy will not only give teachers access to what hypotheses the
students have about the target language, but will also provide information about the
developmental stages of the students‟ interlanguage. As with regard to error correction, such
errors require a deep correction rather than a surface treatment and mere identification of the
error to the students (Hammerly 1991). Nevertheless, in most cases the reader find no
difficulty in understanding the meaning intended by the students bearing in mind the context
of task and the students‟ level of proficiency.
These are the most salience observations regarding the texts produced by the students in both
groups. I shall now present the number and types of errors made by band 2 and band 4
students respectively with more specificity.
4.1.2. Errors Made by Band 2 Students
58
In this section I shall present and discuss all the errors and error types found in the corpus of
30 scripts written by band 2 students per an average of 180-word long texts, the average text
length of band 2 students. The total number of errors found in band 2 corpus was 449 errors
classified according to the level of their occurrence into two main levels: substance level (6.7
%) and text level (91.8 %). The remaining errors (1.5 %) are classified as blend errors as
they belong to more than one category at the same time. The following table summarises the
number of errors at each level and their percentages.
Level of Error Number of Errors Percentage of Error
Substance Level 30 6.7%
Text Level 412 91.8%
Blend 7 1.5%
Total 449 100%
Figure 2: number and percentages of errors at the two main levels of analysis, band 2.
The fact that nearly 92% of the students‟ errors occurred at the text level and only 6.7% at the
substance level, lends further support to the above observation regarding the nature of the
students‟ errors (see section 6 page,59). It therefore follows that the low frequency of errors
at the substance level suggests that the students are able to communicate meaningfully with
their readers, their writing teachers in this case, without any serious communication
deficiencies. This very observation will also be made when presenting band 4 errors.
However, one should be careful when passing such generalisation because observation and
59
experience show that in some cases students do write unintelligible sentences which teachers
find difficult to understand especially if they contain many blend errors.
Having presented the number of errors and their percentages at the two major levels of
analysis, namely levels of substance and text, I shall now present the number of errors
together with illustrative examples that occurred at the sublevels. It is fruitful to note at this
stage that all of the examples cited below are taken from the actual data produced by the
students.
I. Substance Level Errors
According to the model of analysis proposed in chapter three section 3.4.1, two error types
are identified at this level, namely, misspelling and punctuation. I will be dealing with these
error types individually.
1. Misspelling
Th types of error under this category are similar to those labelled in other publications as
„spelling mistakes‟, which will be dealt with under „distortion‟ (see section 1 page, 64) but
with a slight difference, as we shall see. To illustrate this difference between the two types,
let us consider “out side”, “alot” and “contry”. In the first example, students have „split‟ the
compound word and in the second „fused‟ the constituent word, whereas in the third, they
failed to comply for the missing vowel “u” in their attempt to produced „country. (The terms
split and fuse are used after James, 1998:131). The first two examples are misspelling errors
whereas the third is a distortion.
The total number of misspelling errors in band 2 corpus was 9 representing 2% of the 449
errors extracted from the corpus. There are two types of misspelling errors which both
60
occurred in band 2 data: overuse and underuse of the space between compound words.
Drawing on James‟s (1998) labels, I shall call the former type “split” which includes
examples like out side, a bout, a broad; and the later “fusion” and includes examples like
alot. Finally, it is worth noting before leaving this point that non of the EA studies that have
been reviewed for this research have counted for errors of this nature.
2. Punctuation
This was the second type of the substance level errors. It involves violation of the
punctuation rules of the language. The total number of punctuation errors found in band 2
corpus was 21 accounting for 4.7% of the total errors found in the corpus. The most frequent
punctuation errors found in band 2 scripts were as follows. (X indicates the position of the
omitted element).
Comma omission, as in: First of all X studying in Oman …”
Comma overinclusion, as in: “I think, to study abroad.”
Full stop omission at the end of the sentence.
Uncapitalise country names, as in “uk” and “the usa”.
Using full stop where a comma is required, as in: “If I go to study in a foreign university.
I will live …”.
Viewing the misspelling and punctuation errors from the error:mistake perspective, one could
confidently suggest that students are would be able to self-correct these error types if their
attention has been drawn to them. They are errors of performance rather than of competence
and so, to be more precise in labelling errors, they are mistakes not errors. The error:mistake
distinction was first introduced into the modern debate by Corder (1967) who associates
errors with failures in „competence‟ and mistakes with failures in „performance‟.
61
Interestingly enough, no difference was found in the types of punctuation errors made by
band 2 and band 4 students. However, as we shall see, band 4 students made more
punctuation errors than band 2 students did when errors of both groups per 70 words were
compared.
It is useful to summarise the number and percentages of errors at the substance level in a
table as follows.
Substance error type Number Percentage
Misspelling 9 2%
Punctuation 21 4.7%
Total 30 6.7%
Figure 3: number and percentages of the errors at the substance level, band 2.
II. Text Level Errors
This level is divided into two other sublevels in which 91.8% of the errors found in band 2
corpus occurred. These are lexical level error and grammar level errors. As it is the case
with band 4 students, as we shall see, band 2 students make more grammar errors than ones.
The number of errors found at the lexical level is 148 while 264 errors are found at the
grammar level, as shown in the table below.
Text error type Number Percentage
Lexical errors 148 33%
Grammar errors 264 58.8%
62
Total 412 91.8%
Figure 4: number and percentages of errors at the text level, band 2.
A. Lexical errors
Lexical errors are those who affect an individual lexis (word). A word become erroneous for
three reasons: distortion, formal or semantic misselection. These three reasons form the
bases for the three types of errors at the lexical level, which I shall highlight presently.
1. Distortion
Distortion involves various processes that result in a wrongly spelled word. Students may
omit, substitute, and /or overuse consonants and vowels while writing. The total number of
distortion errors in band 2 scripts was 70 errors representing 15.6% of the total errors in the
corpus. In general, Band 2 students were found to make distortion errors such as the
following:
Vowel omission, as in: religin, smok, explan, therefor
Vowel misplacement, as in: foriegn.
Vowel/consonant substitution, as in: deffirent, thier.
Approximate “y” omission, as in: studing.
It is fruitful to note that the last kind of distortion errors, namely, omission of the
approximate „y‟, occurred at a relatively high frequency level in band 2 scripts especially
when the word „study‟ was used in the –ing form or as a gerund. The result of such omission
was „studing‟.
2. Formal Misselection
63
This type of errors emerges when there are two or more forms of a word and learners fail to
select the correct form which is demanded by the context such as price and prize (see chapter
two section 2.5.3.3). Band 2 students made a total of 37 errors of this type (8.2%). The
following are illustrative examples of this error type.
Her religion may be effected (affected).
I want to continue my study in British (Britain).
To see how people life (live) in other countries.
…and people will suspect (respect) him.
…to know their culture and traditional (traditions).
I will speak it very will (well).
I don‟t thing (think) of my family.
3. Semantic Errors
This error type occur when learners confuse the meanings and/or uses of certain words which
results in words being used in a wrong collocation or context, as stated in chapter two section
(2.5.3.3). The total number of semantic errors was 41 accounting for 9.1% of the total errors
extracted from band 2 corpus. Examples of this error type include:
I will have to travel single (alone).
Teachers in universities teach hard (very well).
I can get more marks (high/higher).
…because they find some reasons (have).
People will handle me as foreigner (treat).
You will obtain difficult days alone (face/live).
...but it will learn the students…(teach).
64
According to James (1998:143), lexical errors are found to be the most frequent category of
errors, thus, there are therefore good reasons to undertake lexical EA. Yet, most of the EA
studies seem to focus on grammar errors and ignore the lexical ones (see for example Scott
and Tucker, 1974; Mukattash,1986; Duskova, 1983).
Before moving to discuss grammar errors, it is useful to summarise the number and
percentage of the lexical errors in a table such as the following.
Type of lexical error Number Percentage
Distortion 70 15.6%
Misselection 37 8.2%
Semantic 41 9.1%
Total 148 33%
Figure 5: number and percentages of lexical errors, band 2.
B. Grammar Errors
Grammar errors are of special interest not to error analysts only, but to language teachers and
learners in general. Teachers often complain about their students‟ repetition of the same
grammar errors even in higher proficiency levels, as it is evident from the findings of the
present study. Consequently, many EA studies have been devoted to analyse students‟ errors
in grammar in an attempt to find out what areas of grammar turn out to be the most
problematic for students so teachers can focus on them, prepare teaching materials that
provide students with enough practise on these structures and plan any necessary remedial
work. However, some of the grammar errors „fossilise‟ in the students‟ interlanguage and so
are made by students in relatively higher proficiency levels (Selinker, 1972). The literature
of EA contains many studies that are devised to examine students‟ grammar errors, to
65
mention some: Matter, 1989; Scott and Tucker, 1974; Mukattash, 1986, 1981, 1984;
Duskova, 1983; Kharma, 1981. As a matter of fact, the interest in these very studies stems
from the fact that they are of relevance to the present EA study not only because they
examine grammar errors, but also because the student population was Arab ESL/EFL
learners. Yet, the findings of individual studies remain limited to the student population
involved and any attempts of generalisation should be carefully approached.
Grammar errors in the present study are perceived as affecting two levels of language,
namely, morphology and syntax, as suggested by the analysis model developed and explained
in chapter three section (3.4.1.). As for the quantitative aspect of the grammar errors in the
present study, a total of 264 errors were counted in band 2 students‟ corpus representing
58.8% of the total errors in the corpus. The morphology errors accounted for 21.6% and the
syntax errors for 37.2%, as shown in the table below.
Grammar error type Number Percentage
Morphology errors 97 21.6%
Syntax errors 167 37.2%
Total 264 58.8%
Figure 6: number and percentages of grammar errors, band 2.
1. Morphology Errors
66
The difference between morphology and syntax errors is that errors of the former type affect
individual words while errors of the latter affect more than a single word. Noticeably, the
figures in the table above suggest that band 2 students have more problems with syntax errors
than with morphology. I shall deal with the morphology errors first leaving the discussion of
the syntax errors to section 2 (see page 72 below). The morphology errors are of four types:
verb, noun, adjective and adverb. I shall deal with these next.
a. Noun Morphology Errors
Band 2 students were found to make more noun errors than other types of errors within the
morphology level. Their total number was 55 with a percentage of 12.2% of the 449 errors
extracted from band 2 corpus. Basically, three noun errors are observed in band 2 scripts,
they were:
Plural –s deletion, as in: for two reason (reasons); the custom (customs) don‟t allow me
to…; I need them for many thing (things).
Plural –s overinclusion, as in: … better than to learn one languages (language).
Misformation of the plural of irregular nouns, as in: many students go to study in many
countrys (countries).
An observation is also made regarding the proportion of the three types of noun errors. Band
2 students tend to delete the plural marker from nouns more than to overuse it. This
observation lends further support to the findings of Scott and Tucker (1974) who examined
the grammar errors of 22 Arabic-speaking students at the American University of Beirut.
Their findings fall in consistency with the present study‟s regarding the error under
discussion except for error type which does not occur in neither band 2 nor band 4 scripts,
namely, the use of non-count nouns with the plural marker (ibid.:90). Drawing on Corder‟s
67
(1967) distinction between „performance errors‟ and „competence errors‟, Scott and Tucker
(1974:90) labelled errors in this category as “performance mistakes”.
b. Verb Morphology Errors
This error type was less frequent than the previous one. A number of 38 verb errors was
counted in band 2 scripts (8.4%). The types of errors under this category are:
Third person singular –s omission, as in: the study in Oman take (takes) long time; it
improve (improves) information.
Third person singular –s overinclusion, as in: I prefers (prefer) to continue my study…
Model misselection, as in: …and must (have to) do things yourself; my family don‟t
(doesn‟t) allow me to travel alone.
Model overinclusion, as in: I will be can learning another language (I will be learning).
Auxiliary misselection, as in: I am (I don‟t) feel homesick.
Auxiliary overinclusion, as in: I will have improve my language by…(I will improve).
Auxiliary omission, as in: students prefer to study abroad when they graduated from high
school (have graduated).
Verb misformation after models, as in: they will died (die).
“to” omission from infinitive forms, as in: I prefer study abroad (to study); I have many
reasons support my opinion (to support).
“to” overinclusion in infinitive forms, as in: …but they couldn‟t to continue their study
(couldn‟t continue).
Other researchers, who investigated errors made by Arab learners, have also reported similar
findings regarding verbal errors. Mukattash (1986), who investigated the interlanguage of 80
fourth-year Arab students at the university of Jordan, for instance, concluded that verbal
errors, like the ones discussed above, and relative clauses “have a high frequency of
occurrence in the students interlanguage and a high tendency to fossilise” (P.189). Similar
findings were also reported by Mukattash in his 1984‟s study. The term „fossilisation‟ was
first introduced in the literature by Selinker (1972). The only problem with Mukattash‟s
68
study is that no statistical representation of the errors was attempted and so no comparison
seems to be possible between his findings and those of the present study.
Another findings on students‟ difficulty with verbal errors were reported by Duskova (1983)
who examined the errors made by fifty Czech adult learners of English. Model verbs, third
person singular –s, and verb tense are only examples of the errors found in the Czech
students‟ written language.
In their investigation of the writing errors collected from 22 Arabic-speaking students, Scott
and Tucker (1874) found that verb errors were among those errors which students most often
deviate from the standard (P.75). It would be exceedingly interesting to compare the verbal
error types of the present study and those of by Scott and Tucker (ibid.) since the level of the
students in both studies is the almost the same, first year students in their lower intermediate
to intermediate level. They found that about 19% of all finite verbs used were erroneous.
These errors involved omission, substitution and redundant use of auxiliary and copula; third
person singular –s omission; wrong verb tense; incorrect form use of finite verb and subject-
verb agreement.
c. Adjective Morphology Errors
Two incidences were found in band 2 corpus in where the students made an adjective error.
The two cases involved adjective/adverb confusion as in “A comfortable and easily (easy)
life” and adjective misformation as in “comfortablly” (comfortable). The adjective errors
represented only 0.4% of the total errors.
d. Adverb Morphology Errors
69
Again only two incidences were found in band 2 corpus in which the student substituted an
adjective with an adverb. It seems that the students at this level of proficiency find it difficult
to distinguish the usage of both adjectives and adverbs and so they end up using an adjective
where an adverb is required and vice versa. As it is the case with the adjective errors, the
adverb errors represented 0.4% of the total errors.
A very important observation in relation with lower frequency of an error type in the
students‟ writing has been made by Duscova (1983). He stresses that “lower frequency of
error need not necessarily mean that the point in question is less difficult” (P.218).
Regarding the error types under discussion, adjective and adverb errors, their low frequent of
occurrence in the students‟ scripts does not necessarily mean that the students have no
problems using them effectively, on the contrary, they may consciously „avoid‟ using
structures containing adjectives and/or adverbs because they are not yet confident enough to
use them correctly. This notion of avoidance strategy which is employed by learners as one
of the strategies of learning a second/foreign language has been widely dealt with in the
literature. James (1998:173), for instance, associates avoidance with learners‟ ignorance of
the TL structure(s). Many studies have concluded that there is a correlation between
avoidance behaviour and item difficulty in the TL such as Mattar, 1989; Kleinmann 1983 and
Mukattash 1984.
As we have covered the four types of morphology errors, it is fruitful to summarise the
number and percentages of each type in a table as follows.
Morphology error type Number Percentage
Noun morphology errors 55 12.2%
Verb morphology errors 38 8.4%
70
Adjective morphology errors 2 0.4%
Adverb morphology errors 2 0.4%
Total 97 21.4%
Figure 7: number and percentages of morphology errors, band 2.
2. Syntax Errors
As stated earlier in section 1 (see page, 68) errors in the syntax level does not affect only a
single word, but units larger than the word, namely, phrases, clauses, sentences and
ultimately paragraphs. All errors of the four subtypes are counted and classified as errors of
phrase, clause, sentence and paragraph. The total number of syntax errors was 167 (see
figure 6 above) accounting for 37.2% of the total errors found in band 2 corpus.
a. Phrase Errors
This category includes errors that occurred at the phrase level. Errors at this level were found
to occur at a relatively higher level of frequency when compared with other error types
within the same level, i.e., syntax. A number of 98 phrase errors were found in band 2 scripts
accounting for 22% of the whole errors in the corpus. Errors in this category includes:
Article omission, as in: there is economical (an economical) reason why I want to
continue my study in Oman; to see things from past (the past); we can use internet (the
internet).
Article overinclusion, as in: The studying abroad…(studying); meet a new (new) people.
Overinclusion of “for”, as in: my parents don‟t allow for me to…(allow me).
Pronoun misselection, as in: I want to continue their (my) study abroad; a certificate
from him (his) country.
Pronoun omission, as in: I prefer to continue study (my study) in Oman.
71
Preposition omission, as in: to go England (to England); I want to know features foreign
(of foreign) cultures; to know a lot of information people (about people); Oman has
abilities to provide us (provide us with) whatever we need.
Preposition overinclusion, as in: I will miss to my (miss my) family; I want to continue my
study in abroad (abroad).
Preposition misselection, as in: I will learn languages very well for (in) these countries;
in the other hand (on); I don‟t think to (of) my family.
Lack of agreement between adjectives and their nouns, as in: in this countries (these);
this advantages (these); some of this reasons (this).
Word order in phrase, as in: in the East Middle (Middle East).
It is worth noting, however, that no significant difference was found in the „type‟ of phrase
errors made by students in bands 2 and 4. The noticeable difference was in the number of
times the phrase errors occurred in the data.
The high frequency of occurrence of these error types in particular leads us to conclude that
the students are either in ignorance of the TL rules or have half-learned these rules, as
suggested by James (1998), and so they are not errors of performance, rather, they are errors
of competence, as suggested by Corder (1967). Teachers have to pay serious attention to
these types of errors since they require a „deep‟, as opposed to surface‟ correction
(Hammerly, 1991).
These error types are by no means peculiar or unexpected features of students‟ interlanguage
regardless to their background. Findings of other studies regarding these error types will be
presented later in this section after the presentation of the three other types of syntax errors,
which I shall deal with below.
b. Clause Errors
72
These errors involve violation of clause construction rules in English. They were less
frequent than phrase errors. The total number of clause errors was 52 representing 11.6% of
the total errors found in band 2 corpus. As it is the case with phrase errors, clause errors are
also of different types. These are:
Subject omission, as in (the letter X will be used to indicate the position of the deleted
element): …such as how X cook their food (how they cook); also X might (I might) learn
many useful things.
Subject overinclusion, as in (the underlined subjects are overincluded): I believe that
anyone when he wants (anyone wants) to study outside he should; if she travels she will
be (it will be) difficult for her; they want to be near their homes and they to avoid (and to
avoid) facing problems.
Be omission as a full verb, as in: Another reason X that I can learn English; because it X
very important; The second reason X I will ….
Be overinclusion, as in: I want to be study (want to study) near my family.
Main verb omission, as in: …so you will X new places.
Word order, as in: if all we go abroad (if we all); I can visit also many places (I can also
visit); …and meet some people famous (some famous people).
„and‟ overinclusion when listing, as in: …have new friends and their career and their
customs.
„who‟ omission in relative clauses, as in: There are a lot of people X want (who want) to
study abroad.
Apostrophe omission in possession structure, as in: you will know some of other X culture
and tradition (other‟s).
c. Sentential Errors
Errors at the sentential level involve either fragment constructions or misarrangement of
clauses in the sentence by omitting connectives. 16 errors of this type were found in band 2
73
scripts. They accounted for only 3.5% of the total errors in the corpus. The main error types
that appeared in the data were:
Misarrangement of clauses in the sentence or connective omission, as in: I prefer that
because I want to be near my family because I need them for many things;
then we can get more information because Oman opened SQU in 1986 so it does not have
good information but we can say that SQU is a very good university because it…;
In my opinion, there are many reasons study in the UK is better than study in Oman;
I have accustomed myself with some customs in my country I can not change.
Fragment constructions, as in: And there are a lot of disadvantages. Because you find
hard life without your family.
d. Paragraph errors
Generally speaking, no serious errors were found in band 2 scripts at the paragraph level.
Sentences within paragraphs were well ordered and various conjunctions were used, except
in a single incident.
To sum up, the statistics of the syntax errors are represented in the table below.
Syntax error type Number Percentage
Phrase errors 98 22%
Clause errors 52 11.6%
Sentence errors 16 3.5%
Paragraph errors 1 0.2%
Total 167 37.2%
Figure 8: number and percentages of syntax errors, band 2.
74
It would be profitable to compare the present findings on syntax errors with the similar error
types found in other studies. As stated earlier in section “a” (see page 73), these error types
are found to be made by students from different backgrounds and surprisingly at relatively
higher levels of proficiency. Schachter (1983:358), who examined the production of relative
clauses in five language groups, reported that relative clauses are a common problem for
Persian, Arab, Chinese and Japanese students. (The fifth group was native speakers of
English used as a control group). Interestingly enough, the Arab and Persians learners in this
study were found to produce the highest percentage of erroneous relative clauses compared to
the other two groups.
Another study that reports on errors at the phrase and clause levels is the one by Duscova
(1983) who examined the errors made by adult Czech students. Errors like article omission,
lack of agreement between adjective and its noun, be omission, the use of which instead of
who and word order are bound to appear in all students‟ scripts.
The findings reported by Mattar, 1989; Mukattash, 1986 and Scott and Tucker, 1974 are of
special interest to the present study since they all deal with errors made by Arab learners of
English. What concerns us most in this section is the findings of these studies on the
syntactic errors. In Mattar‟s (1989) study, students were found to have problems with the
relative clause structure involving the repetition of objects and subject within the same
clause. These two error types accounted for 43% of the relative clauses in the students‟
composition (P.381). This very error type has been discussed in the present study in section
“b” above (see page 74). Another error type reported by Mattar in relation with relative
clauses is the replacement of relative pronouns such as /whose/ and /where/ with
inappropriate pronoun such as /which or /who/. Similar observation is also made regarding
band 2 students in the present study and the findings by Mukattash (1986) who, too,
75
investigated Arab students‟ errors. As Mattar puts it “such findings seem to indicate that
Arab learners find the relative pronouns /whose/ and /where/ more difficult to use than
pronouns such as /which/ and /who/”. As with regard to other error types at the syntax level,
Scott and Tucker (1974) also reported that Arab students having difficulties with preposition
use (22.9%), articles (8%) and sentence combining (no percentage given). It is worth noting
that other error types, such as question formation errors, which are reported in the above
mentioned studies, were not observed in the data of the present study due to the nature of the
elicitation techniques, i.e., expository writing on a given topic. However, the very rare
occurrence of such errors in the data does not indicate, as suggested by Kleinmann
(1983:364), that the structure in question is less difficult.
III. Blend Errors
So far I have presented and discussed the error types that occurred in two major levels of
analysis, namely the levels of substance and text. Among the error types that were found in
band 2 scripts (and band 4 as well) are those which can belong to more than one level at the
same time. I labelled these errors as Blend. Their occurrence in the data accounted for 1.5%
of the total errors. Examples of this type include:
Distortion + plural –s omission, as in: …have to wash desh (dishes).
Word order + „to‟ omission + article omission, as in: In my opinion, prefer people
continue their study (some people prefer to continue).
4.1.3. Errors Made by Band 4 students
The most striking finding of the present study, as mentioned earlier in section 4 (see page
58), is that no significant difference has been noticed between band 2 and 4 students in terms
76
of error types, i.e., all the error types and subtypes that occurred in band 2 data and have been
discussed earlier have also been observed in band 4 scripts, but with variation in the
frequency of occurrence of each error type. For this reason, I have found it of no value to re-
present the error types made by band 4 students. In our treatment of band 4 students, only the
number of error types and subtypes at each level will be presented. For example and
discussion of these error types, the reader is referred to section (4.1.2). The following table
shows the number and frequency of band 4 errors in the three major levels of analysis.
Level of Error Number of Errors Percentage of Error
Substance Level 95 15.3%
Text Level 511 82.4%
Blend 14 2.3%
Total 620 100%
Figure 9: number and percentages of errors at the two main levels of analysis, band 4.
From an inspection of the number of errors of both groups shown by tables 2 and 9, one
would be led to suspect that band 4 students made more errors than band 2 did. However,
such conclusion turns out to be wrong when text length, discussed in section 1 (page, 57), is
taken into consideration. I shall revisit this point after the presentation of band 4‟s error
types and subtypes together with their percentages.
77
Error level/type
Number Percentage Total
Misspelling 41 6.6% 95 (15.3%)
Punctuation 54 8.7%
Lexical errors 225 36.3% 511 (82.4%)
Grammar errors 286 46.1%
Semantic errors 46 7.4% 225 (36.3%)
Misselection errors 57 9.2%
Distortion 122 19.7%
Morphology errors 105 17% 286 (46%)
Syntax errors 181 29%
Noun errors 28 4.5%
105 (17%)
Verb errors 72 11.6%
Adjective errors 4 0.6%
Adverb errors 1 0.2%
Phrase errors 102 16.5% 181 (29%)
Clause errors 62 10%
Sentence errors 16 2.6%
Paragraph errors 1 0.2%
Figure 10: number and percentages of error types and subtypes, band 4.
Let us now turn our attention to the fact that band 4 students made more errors than band 2.
These errors however were counted per text length in each group. In section 1 (see page 57
above), I have stated that an individual band 2 student produced a text that was 180-words
long in the average while the average text length in words produced by an individual band 4
student was 285. We can argue here that since band 4 students produced longer texts than
band 2 did they would be expected to make more errors also. The findings in its present
shape make it impossible to draw any valid comparison between the two groups as to answer
the second question of the study (see introduction to this chapter). For a valid comparison of
this type to take effect, the compared texts have to be of the same size. Consequently, I
78
found it profitable, as to answer the second study question, to consider the number of error
types and the density of each type in each band per 70 words, the shortest text produced by
band 2 students. The following table and chart show the results of such calculation.
Error type Error density per 70 words (band 2)
Error density per 70 words (band 4)
Misspelling 8 8
Punctuation 13 15
Semantic 18 12
Misselection 23 17
Distortion 39 35
Noun 25 11
Verb 27 23
Adjective 2 1
Adverb 0 0
Phrase 64 35
Clause 38 16
Sentence 6 6
Paragraph 0 0
Blend 4 7
Total 267 186
Figure 11: error types and their density per 70 words, bands 2 and 4.
79
Figure 12: A chart shows the error types and their density per 70 words, bands 2 and 4.
The above visual presenting of the data makes it possible to draw the following conclusions:
1. Students in both bands made a total of 453 errors (59% by band 2 and 41% by band 4).
2. The average number of errors made by an individual band 2 student per 70 words was
(9). For an individual band 4 student, the average number of errors is (6.2).
3. Band 2 students made 18% as more errors as band 4 students did.
4. Again no difference was observed in terms of error types in both groups.
5. Within a 70–word long text, both bands 2 and 4 made no adjective or adverb errors.
6. The striking finding is that band 4 students made less errors than band 2 students did in
all error types except in punctuation (13 by band 2, but 15 by band 4) and blend (4 by
band 2, but 7 by band 4). Experience and observation show that students are able to self-
correct such minor errors once their attention has been drawn to them and so they are not
errors of competence, rather, as Corder (1967) views them, they are errors of
performance.
0
10
20
30
40
50
60
70
density o
f error typ
e sper 7
Error types per 70 words
Band 2 Band 4
80
7. By turning the attention to the second study question (see introduction to this chapter),
we can conclude, based on the above findings, that there is a high correlation between
students‟ level of proficiency and the number and frequency of errors they make.
8. As with regard to the first hypothesis, the findings of the present study suggest that the
error types do not vary with the proficiency level, however the error density does vary
with the level of proficiency.
4.2. The Questionnaire Findings
In this part of the chapter, I will be presenting and discussing the findings of the analysis of
the eight questionnaires that were completed by the writing teachers. The three sections that
make up this part will cover the following areas: the error types as reported by the writing
teachers, teachers‟ reaction to certain error types in a given text and the teachers‟ approaches
in responding to their students‟ errors in writing.
4.2.1. Teaching Experience
The analysis of the questionnaire revealed that six of the teachers have spent more than six
years as EFL/ESL teachers. Only one of band two teachers have spent between 1 to 3 years
in the field and the eighth teacher provided no response to the question.
4.2.2. Error types reported by the writing teachers
81
This part of the questionnaire (see appendix 3) aims at eliciting what the writing teachers
view as the most common error types made by students in bands 2 and 4. The teachers gave
the following error categories as the most recurrent ones in their students‟ writing:
Error types reported by band 2 teachers Error types reported by band 4 teachers
Wrong form / tense Tense
Fragments / incomplete sentences Fragments
Wrong word use Word choice
Wrong article use / lack of article Articles
Spelling Spelling
Quantifiers / articles * Language control
Connectives / conjunctions Missing words
Word order Word order
Singular / plural Number
* punctuation Auxiliary use
* Pronoun use * Clauses: relative / comparative /adverbial
* Repetition Part of speech
* Reference * Infinitives
Preposition use Preposition use
* Active / passive Transition words
Figure 13: error types as reported by bands 2 and 4 teachers. The * marked types are
reported to be made by one group but not by the other.
The majority of the error types that were reported by the eight writing teachers have been
observed in the students‟ scripts of both bands. (Note that some of these categories have been
labelled differently in the analysis model used in this study). However, some of the error
types that occurred in the data have not been reported by the teachers. For example, the most
obvious missing category in the above error list of band 4 is punctuation, which occurred 15
82
times per 70 words in band 4 scripts compared to 13 times in band 2 scripts. Additionally,
errors at the clause level have not been mentioned by band 2 teachers although band 4
teachers have. Nevertheless, they did mention other categories such as word order, repetition
and reference, which are examples of the error types that affect the phrase and clause level. It
is also of value to note in this regard that some error types, like passive sentences, for
instance, have occurred very infrequently in the data but have been mentioned by the
teachers. This phenomenon can be explained with relation to the elicitation technique
adopted for the present study, namely, expository writing on a given topic. Additionally, this
very phenomenon observed in the present study lends further support to what other EA
studies have found in relation with the Arab learners‟ avoidance of the passive constructions,
see for instance, Mattar, 1989; Kleinmann, 1983; Mukattash, 1984. Although the
organisational errors are out of the scope of the present study, it is fruitful to note, however,
that one band 4 teacher mentioned the lack of textual organisation such as paragraphing as
one of the problems in band 4 writing.
4.2.3. Teachers’ response to their students’ errors in writing
It was my intuition that the eight teachers will respond similarly to the same error types (see
hypothesis 2 page 4). However, in their response to the errors contained in the text given in
part two of the questionnaire (see appendix 3), teachers of both bands showed different
methods when responding to the same error types. To exemplify the point under discussion,
let us consider particular error types from the text which was attached to the questionnaire for
the teachers to mark and see how many teachers, out of the eight, will mark these errors as
deviations.
Error type Example
Line No. Number of teachers marked as deviant
83
Spelling Continu 1 7/8
Forgin 4 7/8
Country 4 8/8
Fragment Because I have… 1-2 7/8
Verb omission important reason, X if you 3 8/8
Another reason X that 6 6/8
Article omission in forgin contury 3-4 7/8
Will be important 10 8/8
Word misselection Another 6 8/8
Esteem 11 8/8
Collocation High university 4 7/8
Figure 14: teacher responses to certain error types. The „X‟ shows the position of the omitted
verb.
The fact that not all teachers responded to the same error type similarly can be explained on
the basis of two concepts that have been widely discussed in the literature of EA, namely,
Error Noticeability (James, 1998:218) and Error Gravity (James, 1998; Mattar, 1989:383;
Roberta, et al 1984), see chapter two section 2.6. According to James (1998:218), some
errors are by their nature unnoticeable and so go unnoticed while others occur very frequently
in the text and rarely go unnoticed. He called the former type “covert errors” and reserved
the term “overt errors” to the latter. As for the second possible explanation to the
phenomenon that not all teachers responded similarly to the same error types, it can be
hypothesised that some teachers view certain error types as less serious than others and so it
is worthwhile focussing the students‟ attention on such errors that hinder communication and
of a relatively high frequency of occurrence in the text. In a similar study, Roberta et al,
(1984) examined the faculty response to the written errors of students at the sentence level.
Their findings fall in agreement with those of the present study regarding the fact that
84
teachers judge the seriousness of certain errors differently. In conclusion, the present
findings tend to falsify hypothesis 2 set out in the introduction (see page 4).
As with regard to the general approach available for teachers in responding to their students‟
errors in writing, the analysis of eight responses given by the writing teachers in the third part
of the questionnaire revealed systematic variations in the method followed by band 2 and 4
teachers.
The intensive examination of the feedback methods used by the eight teachers in responding
to the text in the questionnaire revealed that band 2 teachers did not only locate the error
position on their students‟ drafts by using marking symbols above each error, but tended to
provide corrections as well. On the other hand, band 4 teachers appeared to provide lesser
help depending on the nature of the error and whether the students are able to self-correct
their errors once their attention has been drawn to them, as commented by one of the band 4
teachers. Finally, it is worth presenting the techniques that the writing teachers revealed
using to follow-up with their students‟ writing. Three out of the eight teachers (two band 2
teacher and one band 4) reported using peer marking or correction. Harmer (1991:147)
argues that student-student correction proves useful in helping students reflect on their own
errors as well as on the other‟s and self-correct them. He proceeds that such approach can in
itself be classed as a communicative activity. Another technique that has been reported by
the teachers is conferencing. Three band 4 teachers and, surprisingly enough, only one of
band 2 teacher reported that they often sit with their students after class or invite them to their
offices to discuss the students‟ errors and provide feedback on how to improve their writing
skills.
85
As I have pointed out in chapter two, the literature on error correction to date suggests that
there is no taken-for-granted way that best helps students overcome their errors and mistakes
and consequently improve their writing. Feedback types and modality such as when, in terms
of time and students‟ level, to provide feedback, form/content focussed feedback,
negative/positive feedback, marginal/terminal feedback, intensive extensive feedback,
process/product focussed feedback and finally implicit/explicit feedback, have no apparent
contributions to differences in overall writing quality (Hedgcock and Lefkowitz 1994:144).
Paradoxically, other studies, see for example Fathman and Whalley (1990), proved evident
that form and content-focussed feedback help students improve their grammatical accuracy
and the content of their compositions respectively.
4.3. Summary
The present chapter reports on the findings of the two research instruments, the writing task
and questionnaire. The systematic marking of a corpus of 60 scripts written by 60 first year
students at two different levels of proficiency revealed the most common error types and their
frequency of occurrence as made by the students in writing, which answered the first study
question set out in chapter one. When compared in terms of the error types and their density
per 70 words, students in bands 2 and 4 were found to make the same error types but differ in
the density of each error. In quantitative terms, band 2 students were found to make more
errors than band 4 students do. It has been hypothesised that the number and types of errors
will vary with the level of proficiency (see hypothesis 1 introduced in chapter 1). The
findings of this study indeed provided sufficient evidence in support of only the first part of
the hypothesis: the error frequency. There was not significant statistical difference between
band 2 and 4 students in terms of error types. The findings from the questionnaires suggested
86
that the students‟ level of proficiency determined the approach used by teachers in their
response to the students‟ errors.
Chapter Five
Conclusions, implications and recommendations
5.0. Introduction
The underlying objective of the present EA research, which deals with Arab ESL learners‟
errors in writing, is to find out what patterns of error types will emerge from the analysis of
87
students‟ writing. The basic assumption is that familiarity with the error types that are
actually made by the students in their attempt to learn a second language is a very useful
guide for determining the difficulties that these students have in learning that language and,
consequently form the basis for the sequence and emphasis of teaching and error correction
in ESL/EFL classrooms. Generally speaking, the present EA study focuses on the linguistic
classification of errors made by 60 students at two different levels of proficiency based on the
level that each error occurs at. The results of the present empirical research are detailed and
discussed in the previous chapter, however, the main conclusions and implications regarding
the present study are introduced in this chapter. Several recommendations for further
research on Arab learners‟ errors will also be made in a later stage of the chapter.
5.1. Major Conclusions and implications
As explained in chapter one (see section 1.3), no attempt was made in the present EA study
to explain the errors found in the students‟ writing. Relating every error type to its source
was not attempted here based on the fact that “the accurate description of errors is a separate
activity from the task of inferring the sources of those errors” (Dulay et al, 1982:145).
Consequently, all the conclusions drawn in this chapter will be based on the directly
observable characteristics of the errors collected from the data.
5.1.1. Text length and organisation
The findings of the present study suggest that there is a correlation between the students‟
level of proficiency and the text length. As a matter of fact, the writing task that was
administered to the students in both bands 2 and 4 did not impose any text length on the part
of students as explained in chapter 4 (see page57). However, the analysis of the students
writing revealed that band 4 students produced texts that were 22.6% longer than those
88
produced by band 2 students. What this entails is that the more proficient the students are the
richer and longer texts they produce. Additionally, almost all band 2 students produced
single-paragraph texts, except for a few. These very observations lead us to conclude that no
adequate and sufficient focus has been paid to the organisational aspect of the students‟
writing at this level. From my experience as a teacher at the Language Centre, where the
present empirical research took place, students are introduced to essay writing as early as
they pass their first level of proficiency (band 1) and so by the time when the writing task
was administered students should have acquired the basic principles of essay writing. These
conclusions have direct implications for classroom practice. In the absence of students‟
adequate knowledge of academic writing skills, I would suggest that teachers should focus on
these skills in class and raise the students‟ awareness of the importance of such aspects of the
academic writing. What this entails is that teachers should design tasks that help students
develop more sophisticated writing skills such as, outlining, summarising, reporting and
arguing, topic-oriented paragraphing and synthesising. On the material writers‟ part, these
skills have to be incorporated into the curriculum and tested as basic components of the
course as a whole, especially with band 2 students in regard to this study. In the light of the
present findings regarding the organisation and content of the students‟ writing at SQU, I
would suggest a content-based approach to the academic writing (Shih, 1986). In such
approach, the emphasis is on writing from sources (reading, lectures, discussions) and on
synthesis and interpretation of information to be studied in depth. As Shih (1986:620) puts it,
“The use of writing tasks which follow from, and are integrated with, the listening and
reading of academic material is a defining characteristics of content-based approaches to
academic writing”. The assumption is that extended study of a topic precedes writing helps
students maintain active control of ideas and extensive processing of new information. In a
word, the two approaches to academic writing instruction that have been mentioned above,
89
namely, skill-focused and content-based approaches, are believed to have important
contributions to the development of students‟ academic writing, especially in a context like
the one at hand.
5.1.2. Student Level and Error Types
The assumption is that error types and frequency will vary with the students‟ level of
proficiency (see hypothesis one chapter one). However, the findings of the present EA study
suggest that only error frequency, or error density when taken text length into account, seems
to vary with proficiency. The examination of error types of students from both levels (bands)
reveals no difference regarding the error types. In the light of these findings, we can
conclude that error types do not vary with proficiency as it is suggested in chapter one. This
finding lends further support to those of Mukattash (1986) who examined the errors of 80
four-year Arabic-speaking students. He found that a great number of the error types, such as
errors in the verbal system and in relative clauses, reoccurred in test II after being recorded in
test I, although these errors types had been deeply discussed with the students in class before
the administration of the second test (P.189). The fact that students in band 4 do not differ in
terms of error types from students in band 2 has its own implications for writing teachers as
well as material writers and test developers. On the part of the writing teachers, students
have to be consciously encouraged to recognise their own errors and correct them. In other
words, students have to be involved in the process of error detection and correction. This
could be done by dividing students in groups of three or four and present each group with a
text written by one of the students in the class and ask them to identify and correct all the
error types in that text. Teachers could also modify the error types so as to focus the
students‟ attention on particular error types that are prevalent among the students. Such
consciousness raising tasks would help the students reflect on heir own errors and
90
consequently improve their writing. Such observation has also made by Mattar (1989) who
examined the EFL writing materials of students in Bahrain. On the part of material writers
and developers, the writing materials should benefit from the findings of the many EA
studies like the present one and incorporate the findings of these studies when designing
writing materials and tests. For instance, the same error detection and correction task
suggested above could also be incorporated in exams to test students‟ ability to recognise and
self-correct errors.
5.1.3. The Nature of Errors
As explained in section 6 (see page 58), the majority of errors that have been found in the
students‟ scripts could be classified, drawing on Burt and Kiparsky (1972), as local errors, as
opposed to global ones. This leads us to conclude that students do not have serious problem
in writing meaningful sentences, however, more attention to the local errors also needs to be
paid. Again, a reflective approach to dealing with students‟ errors which involves the
students taking part in the detection and correction of their own errors is also suggested in
relation with the point under discussion.
5.1.4. Errors at the Substance Level
The substance level contains misspelling and punctuation errors, as detailed in section I (see
page 61). Although errors at this level accounted for only 4.6% of the errors in band 2 data
per 70 words, they constitute real problems for students even at higher levels of proficiency.
This fact becomes more evident when compared to band 4 errors at this level which
accounted for 5.7% in the same text length. These findings, thus, suggest that teachers have
high expectation of students at band 4 which lacks deep examination of the actual students‟
competencies in writing. This expectation has led teachers to „marginise‟ such error types
91
assuming that students are able to self-correct such errors and will learn how to minimise
them as they become more proficient. The findings of the present empirical EA research
seem not to fall in agreement with such assumption.
5.1.5. Errors at the Text Level
As figures 4 and 9 show, most of the students‟ errors are in lexis and grammar. The lexical
errors account for 33% of the total errors made by band 2 students while the grammar errors
account for 58.8%. As with regards to band 4 students, their lexical errors represent 36% of
the total errors while the grammar errors represent 46%. What these percentages suggest is
that students at both levels of proficiency have real problems in these two areas. To highlight
these areas of difficulties, it is worth listing all the error types in their categories that have
been extracted from the data and found to account for most of the students‟ difficulties in
learning English at least in the present context. These areas include:
Semantic: selection of words that mismatch the context such as using “learn” where
“study” is required.
Formal misselection: using words that morphologically look similar but have different
function such as “price” and “prize”; “thing” and “think”; “effect” and “affect”, etc.
Distortion: vowel/consonant misplacement such as “forigeon”;
- vowel omission such as “explan”;
- approximate “y” omission such as “studing”.
Noun errors: plural –s deletion;
- plural–s overinclusion;
- ignorance of the plural form of irregular nouns, e.g., “countrys” as the plural form of
„country‟.
92
Verb errors: third person– s deletion;
- third person –s overinclusion;
- model misselection;
- model overinclusion;
- wrong tense;
- wrong verb form after auxiliary;
- auxiliary omission;
- auxiliary overinclusion;
- auxiliary misselection;
- „to‟ omission;
- „to‟ overinclusion.
Adjective/adverb errors: use of adjective and adverb interchangeably.
Phrase errors: Article omission;
- Article overinclusion;
- Overinclusion of “for”;
- Pronoun misselection;
- Pronoun omission;
- Preposition omission;
- Preposition overinclusion;
- Preposition misselection;
- Lack of agreement between adjectives and their nouns;
- Word order in phrase.
Clause errors: subject omission;
- Subject overinclusion;
- Be omission as a full verb;
- Be overinclusion;
- Main verb omission;
- Word order;
- „and‟ overinclusion when listing;
- „who‟ omission in relative clauses;
93
- Apostrophe omission in possession structure.
Sentence errors: misarrangement of clauses in the sentence;
- connective omission.
Similar error types have also been reported in other EA studies which investigated writing
errors of adult ESL/EFL learners from different backgrounds, e.g., Richards, 1974;
Mukattash, 1986; Kharma, 1981; Mattar, 1989; Scott and Tucker, 1974. The findings of the
above mentioned studies as well as of the present one provide ample evidence that there is a
lot which needs to be done in order to minimise such errors from our students‟ writing.
These findings call upon a collaborative effort of the material writers, test developers,
teachers and even the students themselves to closely investigate and analysis these errors
bearing in mind the useful and practical information that EA provides. Such investigation can
informally be carried out by teachers in their own classrooms using the action research
technique (Nunan, 1989) to fulfil such goal. These findings have also several implications
for language teaching methodology. Many researchers have recommended involving the
learners in the process of detecting and correcting their own errors such as Mattar (1989:
354) and Harmer (1991:147). Additionally, in the light of such findings, writing teachers
have to focus their teaching as well as their students‟ attention to these errors and adapt
materials and techniques that help students overcome such errors.
5.1.6. Teacher Response to Student Errors
The underlying objective of the present piece of research is to find out how the writing
teachers respond to their students‟ errors in writing. The assumption is that such data would
provide useful information on how the writing teachers respond to their students‟ errors in
writing which will ultimately make it possible to evaluate such methods of EC in the light of
94
the recent research in the field and the findings of the present study. It has been hypothesised
that teachers will react similarly to the same error type (see hypothesis one in chapter one).
The findings suggest however that teachers differ in their „noticeability‟ of the same error
type and their responses vary with the students‟ level of proficiency. As figure 14 suggests,
errors do not have the same gravity in attracting the teachers‟ attention. We can conclude
then that the „overt errors‟ are more likely to be „noticed‟ by teachers and thus corrected than
„covert errors‟ that often pass unnoticed (James, 1998:218). In terms of the teachers‟
response to the error after it has been noticed, the data suggest that band 2 teachers use more
correction symbols and in most cases provide the students with the correct form of the
erroneous words more than band 4 teachers do. As pointed out in chapters two and five, no
particular feedback type and mode seem to have an effect on the students‟ overall writing
quality. We can thus recommend that teachers should adopt an approach with is compatible
with the students‟ level and best help them become more conscious of their own errors. In
addition, the correction method should stem from a deep understanding of the students‟ level
so the adopted approach for EC will neither be demotivating by over-correcting every single
error nor deal with errors so superficially that students do not get enough feedback on their
errors.
5.2. Limitations and Recommendations for Future Research into EA
5.2.1. Investigation of Error Sources
The descriptive nature of the present EA study makes it impossible to make any valid
suggestion regarding error sources of Arabic-speaking learners in Oman. Nevertheless, the
findings of the present study can form the basis for future research in the field of EA
specially researches which involve Arab learners. Taking such limitation into account, it
95
would be extremely useful to carry out more EA studies which not only describe and classify
the errors of Arab learners, but seek after the potential sources of these errors. Such studies
should thus account for all the possible error causes that are widely discussed in the literature
such as mother tongue interference (interlingual errors), ignorance or incomplete learning of
the TL rule (intralingual error), teaching methodology and/or learning materials (induced
errors), etc. Such studies would provide evidence in favour of or against the Contrastive
Analysis Hypothesis (see chapter two). Put differently, the careful analysis of the findings of
such studies would make it possible to test the validity of the CA hypothesis which attributes
learners‟ errors to MT interference in the TL.
5.2.2. The current writing skill syllabus
One of the limitations of the present EA study is that no attempts have been made to evaluate
the existing writing skill syllabus. The assumption is that some of the error types that have
been found in the students‟ data can be related to the insufficient treatment of such errors by
the materials as well as by the evaluation methods adopted by teachers. These error types are
referred to as „induced errors‟ and have been introduced into the literature by Stenson (1974).
In the light of such assumption, it would be fruitful to conduct further studies that focus on
induced errors and involve a thorough evaluation of learning materials, evaluation methods
and teaching methodologies.
5.2.3. Investigating more English language structures
Any small-scale research has potential limitations. No single investigation can cover all
aspect of language. The present EA study is no exception. Although several important
language items and structures were investigated in this study, several more still await
empirical investigation. I would suggest that further studies involving Arab learners should
96
be devised to investigate structures like passive voice (present, past, progressive and perfect),
past tense (simple, continuous, perfect), reported speech, and the like. To do so, these studies
have to adopt different study design methods such as observation and experiment along with
case study as it is the in the present study. Data collected techniques should also vary with
the study aim. I would recommend that future EA studies should device different data
collection tools in order to elicit as much data from students as possible. These may include
verbal report, note taking, introspection and translation.
References
97
Burt, M. and Kiparsky, C. (1975) „Global and local mistakes‟. In Schumann, J. and Stenson,
N. (Eds.)(1975) New frontiers in second language learning, Newbury House Publishers,
Massachusetts.
Chaudron, C. (1983) „A descriptive model of discourse in the corrective treatment of
learners‟ errors‟. In Robinett, B. and Schachter, J. (Eds)(1983) Second language learning:
Contrastive analysis, error analysis and related aspects, The University of Michigan Press,
USA.
Cohen, A. and Cavalcanti, M. (1990) „Feedback on conpositions: teacher and student verbal
reports‟. In Kroll, B. (Ed)(1990) Second language writing, Campridge University Press.
Corder, S.P. (1967) „Significance of learners‟ errors‟. IRAL. No.5. Reprinted in Richards, J.
(Ed.)(1974) Error analysis, Longman, Great Britain.
Corder, S.P. (1971) „diosyncratic dielect and error analysis‟, IRAL, Vol.IX/2. Reprinted in
Richards, J. (Ed.)(1974) Error analysis, Longman, Great Britain.
Corder, S.P. (1974) „The significance of learners‟ errors‟. In Richards, J. (Ed.)(1974) Error
analysis, Longman, Great Britain.
Corder, S.P. (1975) „Error analysis, interlanguage and second language acquisition‟ (Survey
article), Language teaching and linguistics abstracts, Vol. 8. Reprinted in Kinsella, V.
(Ed.)(1978) Language teaching and linguistics: Surveys, Cambridge, Cambridge University
Press.
Corder, S.P. (1981) Error analysis and interlanguage,Oxford: Oxford University Press.
Dheram, P. (1995) „Feedback as a two-bullock cart: a case study of teaching writing‟, ELT
Journal, Vol. 49/2, Oxford University Press.
Dulay, H., Burt, M and Krashen, S. (1982) Language Two, Oxford University Press, New
York.
Duscova, L. (1983) „On sources of errors in foreign language learning‟. In Robinett, B. and
Schachter, J. (Eds)(1983) Second language learning: Contrastive analysis, error analysis and
related aspects, The University of Michigan Press, USA.
Edge, J. (1989) Mistakes and correction, Longman, London.
Ellis, R. (1986) Understanding second language acquisition, Oxford: Oxford University
press.
Fathman, A. and Whalley, E. (1990) „Teacher response to student writing: focus on form
versus content‟. In Kroll, B. (Ed)(1990) Second language writing, Campridge University
Press.
Hammerly, H. (1991) Fluency and accuracy, Multilingual Matters LTd., Clevedon.
98
Harmer, J. (1991) The practice of English language teaching, Longman, London.
Hedgcock, J. and Lefkowitz, N. (1994) „Feedback on feedback: Assessing learner receptivity
to teacher response in L2 composing‟, Journal of Second Language Writing, No.3/2, 141-
163.
James, C. (1998) Errors in language learning and use, Longman, London.
Johnson, K. (1988) „Mistake correction‟, ELT Journal, Vol. 42/2, Oxford University Press.
Kharma, N. (1981) „Analysis of the errors committed by Arab university students in the use
of the English definite/indefinite articles‟, IRAL, Vol. XIX/4.
Kleinmann, H. (1983) „Avoidance behaviour in adult second language acquisition‟. In
Robinett, B. and Schachter, J. (Eds)(1983) Second language learning: Contrastive analysis,
error analysis and related aspects, The University of Michigan Press, USA.
Long, M. (1983) „Teacher feedback and learner error: mapping cognitions‟. In Robinett, B.
and Schachter, J. (Eds)(1983) Second language learning: Contrastive analysis, error analysis
and related aspects, The University of Michigan Press, USA.
Mattar, H. (1989) „A cross-sectional error analysis study of the common writing errors made
by adult Arabic-speaking EFL learners in Bahrain‟, Unpublished Ph.D. dissertation,
University of East Anglia.
Mukattash, L. (1981) „WH-Questions in English: A problem for Arab Students‟, IRAL, Vol.
XIX/4, 317-332.
Mukattash, L. (1984) „Contrastive analysis, error analysis, and learning difficulties‟. In
Fisiak, J. (1984) Contrastive Linguistics, Mouton Publishers: Berlin.
Mukattash, L. (1986) „Persistence and Fossilisation‟, IRAL, Vol. XXIV/3, 187-203.
Nemser, W. (1971) „Approximative systems of foreign language learners‟, IRAL, Vol. 9/2.
Reprinted in Richards, J. (Ed.)(1974) Error analysis, Longman, Great Britain.
Nunan, D. (1989) Understanding language classrooms: a guide for teacher initiated action,
New York, Prentice-Hall.
Oppenheim, A. (1992) Questionnaire design, interviewing and attitude measurment, Pinter
Publishers, London.
Richards, J. (Ed.)(1974) Error analysis, Longman, Great Britain.
Richards, J., Platt, J. and Platt,H. (Eds)(1992) Dictionary of language teaching and applied
linguistics, Longman, London.
Roberta, V., Daisy, M. and Fredrick, L. (1984) „Error gravity: A study of faculty opinion of
ESL errors‟, TESOL Quarterly, Vol. 18/3, 427-439.
99
Schachter, J. (1982) „An error in error analysis‟. In Robinett, B. and Schachter, J.
(Eds)(1983) Second language learning: Contrastive analysis, error analysis and related
aspects, The University of Michigan Press, USA.
Schumann, J. and Stenson, N. (Eds.)(1975) New frontiers in second language learning,
Newbury House Publishers, Massachusetts.
Scott, M. and Tucker, R. (1974) „Error analysis and English-language strategies of Arab
students‟, Language Learning, Vol. 24/1, 69-97.
Selinker, L. (1972) „Interlanguage‟, IRAL, Vol. 10/3. Reprinted in Richards, J. (Ed.)(1974)
Error analysis, Longman, Great Britain.
Shih, M. (1986) Content based to teaching academic writing, TESOL Quarterly, Vol.20/ 4,
P.617-648.
Stenson, N. (1974) „Induced errors‟. In Burt, M. and Kiparsky, C. (1975) „Global and local
mistakes‟. In Schumann, J. and Stenson, N. (Eds.)(1975) New frontiers in second language
learning, Newbury House Publishers, Massachusetts.
Taylor, G. (1986) „Errors and Explanation‟, Applied Linguistics, Vol.7/2, Oxford University
Press.
Zamel, V. (1985) „Responding to student writing‟, TESOL Quarterly, Vol. 19/1, 79-101.