Advanced Writing in English as a Foreign Language: A Corpus-Based Study of Processes and Products

 2
curriculum development, writing instruction procedures, assignment and course goals, feedback 
types, revision strategies, and the role of peer revision.  
On the corpus linguistics pane, the following areas will be considered: the theory of performance and 
competence, theoretical arguments for and against corpus evidence, the nature and empirical use of 
corpus matter, and data-driven learning that exploits both native and learner corpora.  
Chapter One sets the context of the study by providing a description and analysis of theoretical issues 
and empirical research in the fields of contrastive rhetoric, writing pedagogy and materials 
development. It presents the outcomes of continued cooperation between the teaching profession and 
academia. After evaluating the claims product- and the process-oriented writing instruction has 
made, it concludes by setting the research agenda for integrating learner writing development 
procedures with the method of corpus linguistics.  
Chapter Two then pursues how this can be done by presenting relevant corpus linguistic research. 
After an analysis of the underlying theoretical considerations and a historical overview of the 
development of the corpus linguistic method, it aims to provide a detailed explanation of variables in 
corpus planning, development and manipulation. The chapter ends with the discussion of the specific 
nature of learner corpora, the development of which represents an exciting new vista in both 
language pedagogy and corpus linguistics. By presenting the composition and application of the 
International Corpus of Learner English, the chapter concludes by explaining that the study of 
learner scripts can contribute to enhancing the authenticity of writing pedagogy.  
Following the reviews, Chapter Three presents, employing a mix of qualitative and quantitative data, 
the writing pedagogy procedures at the English Department of Janus Pannonius University, focusing 
on the Writing and Research Skills courses I taught in the past three years.  
Chapter Four presents compositional details of the JPU Corpus and the results of empirical research. 
It examines linguistic data drawn from the corpus and a set of examples of the pedagogical 
exploitation of that data for writing skills development. As will be seen, the largest Hungarian EFL 
learner corpus offers opportunities to describe the lexical and text organization patterns of written 
learner discourse. Another contribution of this chapter to the field is the collection of concordance-
based descriptions and evaluations of learner scripts, which allow for the development of study 
guides for individual and group use.  
The study's conclusions suggest that the JPU Corpus has the potential for further nationwide, and 
possibly international, cooperation between corpus linguists and writing professionals.  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 3
Chapter One: Issues in Writing Pedagogy: A 
Review of the Literature  
Writing, because it allows us to represent to ourselves our learning, our ways of 
making meaning, teaches us the most profound lesson about how we read, write, and 
use language, about what it means to know. (Zamel, 1992, p. 481) 
Introduction 
Writing is among the most complex human activities. It involves the development of a design idea, 
the capture of mental representations of knowledge, and of experience with subjects. The 
interlocking processes of writing by novice and expert authors have been studied by such diverse 
disciplines as cognitive psychology, stylistics, rhetorics, text linguistics, critical literary theory, 
hypertext theory, second language acquisition, and writing pedagogy. From such a wealth of 
approaches and themes, this dissertation will be concerned with what is immediately relevant to the 
teaching and learning of writing in EFL at advanced levels.  
This chapter proposes to set the context of investigating written learner English at university level. A 
descriptive and analytical undertaking, such a project needs to be informed by general second 
language acquisition theory, research design considerations and specifically by the results of research 
in writing pedagogy. I will present the theoretical framework of my study and then review the 
literature that has shaped the present project.  
The chapter is divided into six sections. In the first, a general introduction to second language 
acquisition (SLA) research and writing theory will set the context of the issues considered in this 
dissertation (1.1). The notions and practice of product- and process-oriented writing instruction will 
feature in the next section (1.2). Narrowing down the scope of investigation, the following section 
aims to systematize what is known about the practice of writing pedagogy (assignments, course 
goals, and writing instruction procedures, in 1.3). Focusing on the interaction between teacher and 
learner, and among students, section 1.4 will elaborate on revision strategies, and the role of peer 
revision. The literature review will then present the theory and practice of feedback students receive 
on their scripts (1.5). The concluding section (1.6) will synthesize the most important strands of the 
literature.  
I hope that after this discussion, the present research agenda for integrating learner writing 
development with the method and findings of corpus linguistics will be made explicit.  
 
 
 
 
 
 
 
 
 
 
 
 
 4
1.1 SLA research and writing theory 
1.1.1 Theory and practice in language education 
In reviewing and critiquing SLA research traditions and trends, Ellis (1998) pointed out that much of 
the effort was either theoretical or pedagogical. He argued for a model whereby the communication 
between researchers and teachers can take the form of one of three types: research informing 
pedagogy, research informed by pedagogy, and research and pedagogy interacting to address 
theoretical and practical concerns, and emphasized the importance of the last approach. He also 
argued that any SLA theory can only be applicable by language pedagogy if it is relevant to it (Ellis, 
1995): the goals of the theory must be compatible with the aims of teaching.  
A similar proposal was made by Brumfit (1995) in the discussion of teacher professionalism and 
research. Offering his views on British educational policy and on the needs for integrating global 
SLA research with local observations, he suggested that for classroom practice descriptions to be 
significant, one needs to consider the common variables in different language teaching contexts (p. 
41). Specifically, Brumfit suggested that educational research needs a systematic program, rather 
than focusing on fragmented projects.  
Three strands of investigation were suggested (Brumfit, 1995, pp. 39-40). The first ought to describe 
classroom practice so that events, attitudes and policies are spelled out. The second should take on to 
explain what was found in the first phase: drawing on the data gathered, theory needs to construct 
models to be able to adequately structure that knowledge. Third, studies directed at the pedagogical 
processes need to extract what ought to take place in education from what is happening there. 
Brumfit argued that these three approaches will enable empirical research to establish the program 
orientation.  
For the field to arrive at valid conclusions on the acquirer of language, Larsen-Freeman (1991) 
pointed out the importance of studying and describing the learner. Reviewing research into the 
differential success of acquiring a second language (L2), she critiqued findings related to variables 
such as learner age, language learning aptitude, attitude and motivation, personality, cognitive issues 
and learning strategies. She concluded that future research and language education will need to 
corroborate findings and test such hypotheses as the following: Learning is a gradual process; it is 
not linear; unless learners are ready to proceed to new phases of learning, no long-term acquisition 
takes place. In a discussion of instructed SLA research, Larsen-Freeman and Long (1991) called 
attention to the need to study the ways in which instruction affects SLA. For this process to be 
studied, they suggested that linguistic input sequence and frequency have to be operationalized, 
together with those tasks that learners are exposed to in the classroom. By studying these variables, 
SLA theory can integrate action research findings initiated by the teacher (Larsen-Freeman & Long 
1991, p. 327), a proposal similar to that made by Ellis (1995; 1998) and Dörnyei (1997).  
In many ways, the tasks the language educator faces in teaching and in initiating research and the 
tasks in which learners perform have common features. Both aim to integrate what is known with 
what is being learned about the situation or the language item being studied. Yet there are crucial 
differences, too. In a discussion of the interface between language learning theory and practice, 
Prabhu; (1995) offered a four-component model to describe this relationship. These are the ideational 
(concepts and processes of language learning), operational (pedagogical practice), ideological (social 
variables), and managerial (pedagogical decision-making). As far as the operational module is 
concerned, Prabhu pointed out the contrast between teaching and learning, saying that while teaching 
can be planned and sequenced, learning follows a route based on mental processes that are difficult 
to observe.  
 5
1.1.2 The Input Hypothesis 
However, there is a growing body of research evidence on the rate of acquisition and the optimal 
conditions for successful acquisition to occur. In this area, the work by Krashen has shown direction. 
The Input Hypothesis (Krashen;, 1985) claims that to ensure long-term success in language 
acquisition, there must be comprehensible input. The theory comprises five hypotheses, of which the 
Monitor Hypothesis and the Affective Filter Hypothesis are particularly relevant for writing research.  
The Monitor Hypothesis is concerned with language production--the ability to use language is a 
result of competence based on acquisition, while learning acts to enable speakers and writers to 
"change the output of the acquired system before [they] speak or write" (Krashen, 1985, p. 2). For 
this monitor (or editor) to operate, Krashen hypothesized, the user needs to be aware of the 
importance of accuracy, and the rule stating correct forms should be present. The Affective Filter 
Hypothesis states that for comprehensible input to become intake, a mental block should be lowered: 
this can occur when the speaker is self-confident, and when a potential failure to produce the 
necessary language is not seen as a risk. Krashen added that for the filter to be down, the speaker 
must focus on the message. This model of language acquisition was partly based on Krashen's survey 
and evaluation of theoretical work in applied linguistics, and on investigations of skill-specific 
empirical research, also motivating subsequent work on the implications of the hypothesis in 
language education. Particularly relevant of these studies is his summary of writing research 
(Krashen, 1984) and a recommendation for a reading-based program (Krashen, 1993).  
In the writing study, Krashen (1984) hypothesized that his generic SLA hypothesis of 
comprehensible input held for the development of writing skills, suggesting that extended reading 
was necessary for organizational and grammatical improvement to occur. He analyzed a wealth of 
case studies that confirmed the hypothesis: the acts of planning, rereading, scanning, revising for 
clarification occurred significantly more often and with better results in good writers who also 
reported pleasure in reading. Also, while less able writers were shown to have much more difficulty 
in transferring what is known as writer-based prose to reader-based prose, more apt writers had less 
difficulty to consider readers' needs. Krashen concluded that although formal instruction of sentence-
level rules can help improvement in writing, for significant and successful writing development to 
occur, this may only be a complement to receiving comprehensible input via reading.  
In the reading-focused work, Krashen (1993) presented the framework and application of a program 
that allows the extensive use of what he called "free voluntary reading." Investigating the 
relationship between writing instruction and learning, he reported that because the rules of formal 
writing are far too complicated to learn, style does not result from more writing practice but from 
more reading. Opposing the view that "we learn to write by actually writing" (Krashen, 1993, p. 73) 
he claimed that improved writing quality, and the ensuing discovery of one's own style, is a result of 
frequent reading.  
1.1.3 Writing theories 
For decades, the most influential paradigm of writing was contrastive rhetoric, proposed by Kaplan 
(re-assessed in 1983). The contrastive rhetoric tradition focused on the product of writing and 
established prescriptive approaches to the teaching of writing. Kaplan claimed that in English, 
writers tended to develop their thoughts in a linear fashion, advancing a thesis, forwarding 
supporting evidence in sequentially presented topic sentences, developed in unified paragraphs. The 
aim of writing pedagogy was to compare and contrast the text organizing patterns in the L1 and L2 
and thus facilitate acknowledgment of differences. The primary technique in the classroom was 
imitating paragraphs so that the patterns were practiced. Raimes (1991) noted that this tradition was 
the dominant approach up to the mid-70s, when the focus shifted to the writer and the context of 
 6
writing, and thus to a more process-oriented analysis of writing and writing pedagogy. The latter 
trend also coincided with greater emphasis on language as communication, focusing teachers' 
attention away from form as prescribed by controlled-traditional rhetoric to collaboration between 
teacher and student, and among the students themselves.  
Particularly influential was the work of Hayes and Flower (1980) and Flower and Hayes (1981), who 
developed a cognitive theory of writing processes, eliciting information directly from writers via 
think-aloud protocols and observations (Grabe & Kaplan, 1996, p. 91). They proposed a model that 
was based on three tenets:  
 
 
• processes of writing, such as planning, organizing, reviewing, and evaluating, often interact 
with each other;  
• writing follows a goal the writer is aware of;  
• processes are performed differently by experienced and inexperienced writers. 
The theory identified a task environment (made up by a rhetorical problem and text produced), the 
three major components of the writing process (generating, translating and reviewing), each of which 
is controlled by a monitor. In proposing this model, Flower and Hayes also generated much needed 
empirical research.  
One result of this research was that the use of protocols came under heavy criticism: it was argued 
that the validity of the model that relied on writers aiming to explain what they were doing while 
they were engaged in writing was limited. In response to the need for theory building and for 
validating theory in research, Bereiter and Scardamalia (1987) offered a new perspective: instead of 
bringing together factors characteristic of novice and expert writers, they suggested that different 
models can describe different levels and contexts. Basically, their two-process theory aimed to 
explain how and why differences occur in inexperienced and experienced writers' performance.  
Two models make up the theory. The first is called "knowledge-telling," which involves the 
processes of inexperienced writers, and the second is "knowledge-transforming." In both, the writer 
considers three main factors: knowledge of content, knowledge of discourse, and ideas of a writing 
assignment. However, the first is primarily a step-by-step operation that is engaged as the writer 
collects material and lexis, whereas the second includes the writer's identification of a unique 
problem and goal so that the writing becomes essentially a process to solve the rhetorical problem. 
The first model describes the less experienced writer, whereas the second the expert writer. How one 
proceeds from one level to another, however, was not shown explicitly.  
According to Silva (1990) the development and pedagogical application of these cognitive models 
meant a decreasing concern with error in English as a Second Language (ESL) and EFL. The 
emerging paradigm of the process approach called for a much more positive and encouraging setting, 
a workshop-like environment (p. 15). Still, as Leki noted (1991), contrastive rhetoric still has much 
to offer to language teachers: The information a contrastive analysis reveals of L1 and L2 text 
structures can contribute to what teachers and students regard as successful communication (p. 137).  
In the nineties, one can witness a wide variety of writing pedagogy and research, applying and 
critiquing both major traditions. As noted by Raimes (1991), the field has come to acknowledge the 
complexity of the composing process, with individual research projects focusing on the central issues 
of form, the writer, content, and the reader (p. 421): an ethnography of writing is being produced 
(Silva, 1990; Atkinson & Ramanathan, 1995; and Leki, 1995 are among the recent examples of such 
endeavors). This recognition has a number of implications for theory and practice: the field has to 
gather more data on novice and expert student writers' performance, on the writing processes applied