2.3 EFL writing research

In order to better understand the nature of EFL writing, the necessity of investigating EFL writing processes could never be overestimated. Hence cognitive processes involved in EFL writing were first explored in this section. Meanwhile, the affective aspects of EFL writing were also examined, given the potential influence of affect on the writing product. Moreover, with the aim of facilitating writing task design and enlightening data analysis later in this study, an analysis of the EFL writing construct was carried out to arrive at the operational construct of EFL writing. Therefore, this section explores the nature of EFL writing ability from three perspectives: the cognitive processes, the affective aspects, and the components of the construct of EFL writing.

2.3.1 Cognitive processes of EFL writing

Although EFL writing process research adopts similar research approaches and focuses to those in L1 writing research, research on EFL writing process is rather inadequate, with only a few studies undertaken so far (e.g., Arndt, 1987; Kamimura, 1996; Sasaki, 2000; Victori, 1999). Notwithstanding the paucity of research on EFL writing, researchers have shown EFL writing process, in its entirety, seem comparable to the L1 writing process. As indicated in the models previously established for L1 writing such as Flower and Hayes (1981), the composition of a text in English as a foreign language also involves three major stages, i.e., planning, while-writing, and revising. It should be noted that this division of the writing process into three stages is an oversimplified model because, although the act of writing in general involves these three stages, the process of composition is not in a simple sequential order. In fact, each of these mental acts may occur at any time in the composing process, rather than in clear-cut stages (Hedge, 1988). For example, writers may constantly plan and revise as they compose. In other words, the act of developing and refining one's own goals is not limited to the planning stage, but is intimately bound up with the ongoing, moment-to-moment process of composing. On the other hand, it should be noted that various cognitive activities are not randomly distributed over the writing process, although the writing process is recursive in nature. Rather, certain activities are more likely to occur at some stages than others (van den Bergh & Rijlaarsdam, 1999). So breaking the writing process into stages is reasonable and desirable for instruction and assessment purposes.

As the L1 writing process research indicated (e.g., Flower & Hayes, 1981; Grabe & Kaplan, 1996; Hayes, 1996; White & Arndt, 1991), three major sub-processes are identified: planning, while-writing, and revising. The subprocess of planning determines the selection and order of presentation of information, including activities such as interpreting the writing assignment, generating and organizing ideas, and setting goals. While planning occurs both before the initiation of text production and throughout the whole writing and revising processes, since every act of writing involves planning at least to some degree, it is impossible to detach every act of planning from the whole process even with more advanced technologies (Butcher & Kintsch, 2001). Thus, only the planning that occurs before the initiation of text production is under the researchers' scrutiny. The second phase is while-writing in which the abstract ideas hatched in the planning process are transformed into words and sentences. The common practice in writing research is to label the period between planning and the beginning of revising as the sub-process of while-writing. The sub-process of revising deals with text evaluation, changing and editing, which usually begins after a draft is completed. Similar to planning, this division of the revision phase is relatively rough because it does not include on-going revision that takes place during the while-writing phase.

Although the three major sub-processes are identified, this division is neither neat nor constant for any writer. In fact, the writing process is highly susceptible to various types of factors (Kellogg, 1996; McCutchen, 1996), for example, the writer's writing proficiency (Bereiter & Scardamalia, 1987;Flower et al.,1990)and the writer's memory(Hayes,1996).Given the huge differences between L1 writers and EFL writers, we could expect that there might be marked differences between L1 and EFL writing processes despite the general similarities in terms of the three major sub-processes. This possibility deserves investigation. The review of the relevant literature helps us to identify the following two sources of differences.

On the one hand, a writer's knowledge influences his writing. As discussed above, during planning, writers, activated by the topic, resort to long-term memory (LTM) to retrieve and organize ideas. According to psychologists, LTM is the base where knowledge is stored, including topic knowledge, audience knowledge, and genre knowledge (Hayes, 1996). Thus, knowledge from LTM is necessary to planning, and is believed to affect the procedural plans and content of writing. EFL writers might lack various types of knowledge compared to their L1 counterparts, such as knowledge of how language functions in discourse and in society (Weigle, 2005). For instance, they may not have the awareness of the social and cultural uses of writing in the foreign language, the appropriate ways in which various functions can be expressed in writing, or the expectations of readers from a different culture. Thus, EFL writers are more likely to be constrained by their inadequate English proficiency during the planning process. As Silva (1993) indicates, ESL writers spend less time planning and employ fewer strategies than native speakers do. EFL writers might be more limited than ESL writers.

On the other hand，a writer's language proficiency affects his writing. As we know，planning activities occur at the very beginning of the writing process，and still continue during the writing sub-process. The writer has to alter and generate ideas continuously as the text expands because the written text restrains the ideas and language to be generated. Modifying and generating ideas in accordance with the expanding text during the while-writing sub-process burdens a writer's cognitive capacity (Flower &Hayes，1981). Therefore，a full command of the language is essential for writers to write fluently. If the writer devotes constant attention to demands such as spelling and grammar，the task of translating，i.e.，“putting ideas into visible language” (Flower & Hayes，1981: 373)，can interfere with the macro planning of what one wants to express. Worse still，there exists a structural gap between ideas and language because ideas in memory are not structured in a linear way as human languages are (Beaugrande，1984，cited in Bryson et al.,1991).Thus,translating ideas into written language can be a cognitively demanding task for L1 writers and is exacerbated when the writer writes in a foreign language. As some researchers (e.g.，Lay，1982;Uzawa & Cumming，1989) indicated，constraints on EFL writers due to their linguistic deficiency could never be ignored. For example，they are conceivably less fluent and accurate in word selection and syntactic choices (Weigle，2002).

In fact, despite the apparent similarities between L1 and EFL writing, for EFL writers, all stages of composing are likely to be slower and more laborious (Williams, 2005). This slow and laborious pace may be mainly due to their comparatively limited English language proficiency, as EFL writers seem to devote much attention in their writing process to finding resources such as appropriate words. Additionally, encoding ideas into written texts disrupts lengthy searches for appropriate lexical and syntactic choices. It is also likely that EFL writers experience a kind of overload when they write. EFL writers, because they have to pay more attention to language as they write, are unable to devote much attention to macro planning of their writing, given the limited capacity of their working memory. It is also possible that EFL writers, like their ESL counterparts reported by researchers (e.g., Raimes, 1985; Silva, 1993), also spend less time reviewing and revising their work, compared to native speakers. Besides, it is likely that many, or perhaps most, EFL writers use their L1 at some point during composing, making the encoding process less automatic. These claims and generalizations, however, need to be confirmed by scientific research.

In short, the writing process is complex and recursive, involving the continuous interaction of several activities which occur at any time during the whole writing process. Although investigations into L1 and L2 composing process have found them to be similar in some aspects, significant differences nevertheless exist (e.g., De Larios et al., 1999;Sasaki, 2000; Silva, 1993). For EFL writers, the whole process of EFL writing is possibly much more laborious and recursive because of EFL writers' lack of knowledge and deficiency in foreign language proficiency.

2.3.2 Affective aspects of EFL writing

As the focus of teaching approaches has shifted from the teacher and teaching methodology to the learner and learning process since the 1970s, scholars considered individual factors residing within the learner that may help to account for huge differences in learning results. Particularly, affect has been highlighted as being of special significance for understanding students' learning (Arnold, 1999). In tandem with the significant role of affect in language teaching, some researchers recognized the paucity of research on the effects of attitudes toward and beliefs about writing, acknowledging that the act of writing was not entirely cognitive, but also affective (Boscolo, 1995; Cleary, 1991; McLeod, 1987). Therefore, this part is devoted to a brief discussion of the affective factors that are involved in writing, and to the investigation of one affective factor, situational interest, that is most likely to exert some influence on test takers' writing performance.

Although various affective factors are involved in writing, the commonly researched ones are writing anxiety/apprehension, motivation, and belief. Despite the fact that these factors are intertwined and interrelated in ways that make it impossible to isolate completely the influence of any one factor, the present study focuses on those factors which can facilitate writers to give their best effort in writing assessment. Here, we do not mean to deny the fact that helping some heavily anxious writers reduce their writing apprehension can also help them display their best writing, as ample research evidence has confirmed (e.g., Daly & Miller, 1975a, 1975b; Horwitz et al., 1986). However, the focus of the present study is on writing assessment, and hence test anxiety is somewhat unavoidable. So, this research does not take into consideration writing apprehension. In view of Chinese college students' lack of interest in English writing and the significance of interest in fostering intrinsic motivation to learn, which in turn facilitates learning, the present study mainly focuses on the effect of writing tasks on students' interest in writing. In particular, it concentrates on“situational interest”, which is spontaneous in nature and is closely related to specific tasks, as discussed later in this section.

The importance of interest in language education has long been recognized (Hidi, 1990; Krapp et al., 1992). Recently it has been conceptualized in educational psychology as a psychological state and/or as an individual disposition (Hidi, 2001). According to Ainley (2006:398), interest is “the feeling of being animated and enlivened and triggers actions that are aimed at expanding knowledge and experience of the target content”. Since Krapp et al. (1992) proposed to differentiate between individual or personal interest and situational interest, most researchers (e.g., Hidi et al., 2002; Schraw et al., 1995) adopted this classification. Individual interest is “a person's relatively enduring predisposition to reengage particular content over time” (Hidi & Renninger, 2006: 113). It is regarded as “a long-term investment or deep-seated involvement in the target field” (Alexander, 2004: 286) and thus is quite stable. Situational interest, compared with individual interest, is quite mutable and defined as “the focused attention and the immediate feelings triggered by the situation” (Ainley, 2006: 394). It could be generated by certain conditions and/or stimuli in the environment and represents a relatively immediate affective reaction that may not last long (Hidi & Anderson, 1992). That is, once situational interest is triggered, it may or may not be maintained (Harackiewicz et al.,2000).

Given the nature of situational interest, which is assumed to be spontaneous, malleable, and context-specific, and the focus of the present study, we mainly concentrated on test takers' situational interest in different writing tasks, thereby examining the possible relationship between test takers' situational interest and their writing test performance.

Recognizing the flexibility of situational interest compared with individual interest, a growing body of researchers have documented specific stimulus characteristics likely to be associated with arousal of situational interest (e.g., Hidi & Renninger, 2006; Schraw & Lehman, 2001; Schraw et al.,1995;Wade,2001)and also investigated how situational interest can be utilized to contribute to improved motivation and learning (Ainley et al.,2002;Bergin,1999;Cordova&Lepper,1996;Hoffmann et al.,1998). Research indicated that situational interest could be triggered through attention to the way learning is presented.Ainley et al. (2002) used an interactive computer program to measure various aspects of students' interest in a text. They reported that both situational and individual factors contributed to the students' interest in a particular topic, and that topic interest influenced the students' affect, which in turn influenced their persistence. In analyzing structural features of text materials such as coherence and vividness, Schraw and Lehman (2001) provided some guidance concerning features that were known to trigger and maintain situational interest in reading a text. Other researches also showed that text variables such as coherence (Wade, 1992), character identification and suspense (Jose & Brewer, 1984), and the concreteness and imageability of salient text segments(Sadoski et al.,1993)can increase situational interest. After reviewing the history of interest research and empirical studies, Schraw et al.(2001)provided three ways to increase situational interest in the classroom: increasing students' autonomy, engaging students with better texts, and helping students to process information at a deeper level.

In light of the related literature, we can see that previous studies on situational interest primarily focused on text features that helped increase students' interest while reading. Few studies, however, have been conducted to explore the ways to increase writers' situational interest in writing tasks. The present study, therefore, focuses on the effects of situational interest on test-takers' writing performance, and particularly, whether the contextualized writing task would trigger Chinese college students' situational interest in writing.

2.3.3 The componential construct of EFL writing

So far, the process of EFL writing has been reviewed from both a cognitive and an affective perspective. Improving pedagogical and testing practices, however, requires a product-oriented investigation of the nature of EFL writing, because studies should be designed to search for evidence of the existence of separable components in language ability and to clarify the roles played by each component. A componential model should, therefore, be developed to explain or assess EFL writing ability, i.e., what kind of skills/strategies and knowledge a writer needs in order to accomplish a writing task? Once identified, these skills which will inform the design and development of the writing task. Since EFL writing ability is an instance of communicative language use (Bachman & Palmer, 1996; Grabe & Kaplan, 1996), this study assumes the theory of communicative competence as a framework to analyze what is involved in the performance of EFL writing.

“Communicative competence” was proposed by Hymes (1972), who distinguished himself from other researchers by shifting from a narrow focus on language as a formal system to a much broader view of language as communication. Hymes (1972) defined communicative competence as the ability to use and understand language appropriately in a variety of situations. Later on, a group of researchers devoted themselves to the study of communicative competence, among whom the most renowned were Canale and Swain (1980). The researchers advanced a four-subcomponent framework of communicative competence, including: ① grammatical competence, which covers “all knowledge of lexical items and of rules of morphology, syntax, sentence-grammar, semantics and phonology” (Canale& Swain, 1980: 29); ② discourse competence, or the ability to deal with extended use of language; ③ sociolinguistic competence, or knowledge of rules of language use in terms of what is appropriate to different types of interlocutors, in different settings, and on different topics; and ④ strategic competence, or the ability to compensate in performance for incomplete or imperfect linguistic resources in target language.

Drawing on the work of Hymes (1972), Canale and Swain (1980), Savignon (1983), and other linguists, Bachman (1990) proposed a model of communicative language ability (CLA) which consists of three parts:language competence, strategic competence and psychophysiological mechanisms. Language competence entails organizational competence which subsumes grammatical competence and textual competence, and pragmatic competence which contains illocutionary competence and sociolinguistic competence. Strategic competence refers to the mental capacity for implementing or executing the components of language competence in effective communication. Psychophysiological mechanisms are “the neurological and psychological processes involved in the actual execution of language as a physical phenomenon” (Bachman, 1990: 84). It needs to be made clear that the psychophysiological mechanisms, though involved in any linguistic behavior, are covert and hard to detect unless with sophisticated facilities and, therefore, cannot be teachable or inferred from test takers' performance.

Later, Bachman and Palmer (1996) further refined the Bachman (1990) model and proposed the multi-componential model of communicative language ability, setting language use within particular situations and conceptualizing language use as “complex and multiple interactions among the various individual characteristics of language users, on the one hand, and between these characteristics and the characteristics of the language use or testing situation, on the other” (Bachman & Palmer, 1996: 62). The researchers restricted the individual characteristics to those which are relatively predictable and have been better understood, including personal characteristics (such as age, sex, and native language), topical knowledge，affective schemata, and language ability.

Bachman and Palmer (1996) conceptualized language ability as an internal construct made up of language knowledge and strategic competence. Language knowledge involves organizational knowledge, or knowledge of language structure, and pragmatic knowledge which is related to how individuals communicate meaning and produce appropriate sentences or text. Organizational knowledge is further divided into grammatical knowledge (i.e., how individual sentences are organized) and textual knowledge (i.e., how sentences are organized into texts on a discoursal level). Pragmatic knowledge contains functional knowledge, referred as the knowledge about how language is used to achieve a variety of communicative and illocutionary functions, and sociolinguistic knowledge, known as the knowledge of using language to suit the social settings. Strategic competence is described as a set of metacognitive strategies which help learners set goals in a language use or test task, assess what needs to be done, and formulate a plan for doing it.

Bachman and Palmer (1996) made great contributions to the development of the theory of CLA. However, it should be noted that their model is generic in nature and for language use in general. It seems difficult to apply the model directly to the construct of writing, although significant insights can be obtained from the model. More specifically related to the nature of writing is the framework proposed by Grabe and Kaplan (1996) who perceived writing as communicative language use and specified a comprehensive and detailed taxonomy of writing.

The Grabe and Kaplan (1996) model gave considerable attention to the linguistic knowledge base, as Figure 2.1 illustrates. According to these researchers, language competence is made up of three competences:linguistic (grammatical), discourse, and sociolinguistic. These competences were minutely specified in the taxonomy presented in Grabe and Kaplan (1996: 217-22). One hallmark of their taxonomy is the special attention given to “audience”, an essential concept for writers (Reid, 1988), by adding a further category of “further audience considerations” aside from the widely agreed components of language competence. This is in concert with the current notion of writing as a communicative act. Another difference between the Bachman and Palmer (1996) framework and that of Grabe and Kaplan (1996) lies in the status of functional knowledge. In the Bachman and Palmer (1996) framework, functional knowledge, together with sociolinguistic knowledge, constitutes pragmatic knowledge, whereas in the taxonomy proposed by Grabe and Kaplan (1996), functional use of written knowledge is part of the component of sociolinguistic knowledge.

More recently, Weir (2005) provided a theoretical socio-cognitive framework for an evidence-based validity approach, attempting to reconfigure validity as a unitary concept and to show how its constituent parts interact with each other. The framework is socio-cognitive in that the abilities to be tested are mental constructs which are latent and within the brain of the test taker (the cognitive dimension); and the use of language in performing tasks is viewed as a social rather than purely linguistic phenomenon. This framework identifies six key types of validity evidence that need to be collected at each stage in the test development process: including test taker, context validity, theory-based validity, scoring validity, consequential validity, and criterion-related validity. The pictorial representation depicts how these components fit together both temporally and conceptually. Here, priority is given to the theory-based validity for validating writing tests given the focus of this section is on the construct of EFL writing.

According to Weir (2005: 110-1), establishing theory-based validity in writing is concerned with evaluating the activation of executive resources and executive processes prompted by the task.Executive resources involve language knowledge and content knowledge.Language knowledge comprises grammatical knowledge, discoursal knowledge, functional (pragmatic knowledge), and sociolinguistic knowledge. Content knowledge may already be possessed by the candidate through developed schemata or might be available in information supplied in task input. Executive processes refers to cognitive processing and includes the procedures of goal-setting,topic and genre modifying,generating,organizing,translating and reviewing.A detailed presentation of the components of the theory-based validity is displayed in Table 2.1.

Later on，Shaw and Weir (2007) introduced the term “cognitive validity” to replace the term “theory-based validity” in Weir (2005). According to Shaw and Weir (2007: 34)，“the cognitive validity of a writing task is a measure of how closely it represents the cognitive processing involved in writing contexts beyond the test itself，i.e. in performing the task in real life”. Therefore，the mission of language test designers is much clearer，i.e. “to be aware of the cognitive processing that underpins equivalent operations in real-life language use” (Shaw & Weir，2007: 6). Drawing extensively on the literature of writing process research，Shaw and Weir (2007) proposed a framework of cognitive validity parameters in writing，displayed in Table 2.2，which consisted of six major procedures.

Macro-planning refers to gathering ideas and identifying major constraints.Organization means ordering the ideas.Micro-planning refers to focusing on the part of the text that is about to be produced.Translation is the process of converting propositional content held in the abstract form into linguistic forms.Monitoring involves checking the mechanical accuracy of spelling, punctuation, and syntax, and/or examining the text to determine whether it reflects the writer's intentions and fits the developing argument structure of the text. Revising includes making corrections or adjustments to those unsatisfactory aspects of the text.

So far, we have reviewed the components of language competence in general and components of writing ability in particular. While language knowledge is made up of all the above components, the specific instances of language tasks will determine which areas of language ability (i.e., language knowledge and strategy competence) are engaged. Different language tasks, therefore, involve different instances of language use and tap into different aspects of language ability. Thus, language ability has to be defined in a way appropriate for the prospective situation (Weir & Shaw, 2005). Since the present study was conducted in the setting of Chinese undergraduates learning English, the specific purpose of the test was, therefore, to assess the ability of the Chinese undergraduates to write in English.

On the basis of the previous research, this study arrived at a set of components of EFL writing competence, presented in Table 2.3, to guide the design and development of EFL writing test tasks as well as the rating scale. Given the fact that writing not only involves linguistic knowledge but also entails gathering ideas and information, analyzing and organizing this information, and presenting it in a way that effectively communicates ideas to the readership (Williams, 2005), this framework, therefore, considers EFL writing as involving both executive resources, i.e., language, and executive processes. Knowledge consists of language knowledge (grammatical knowledge, textual knowledge and sociolinguistic knowledge) and content knowledge (both internal and external). In light of writing as a communicative act, special attention is paid to sociolinguistic knowledge (i.e., individual's knowledge about how to use written language appropriately to address certain audience and achieve a variety of communicative purposes in social settings), as indicated in the rating scale developed for this research. Thus, in the present study EFL writing ability encompasses the grammatical, textual, and sociolinguistic aspects of the written language, and the interaction among these three components.

Drawing on the previous research on the writing process and especially on the executive processes proposed by Weir (2005) and on the cognitive validity parameters put forward by Shaw and Weir (2007), the present study regarded executive processes and strategies involved in EFL writing as consisting of three major procedures, i.e. planning, text production, reviewing and revision. Planning entails interpreting the task (i.e., taking into consideration audience, purpose, occasion, and so on), goal-setting, and idea generating (such as brainstorming and free writing). At the stage of text production, writers move from an internal private representation, which is abstract and only understood by him or her, to its expression in the public shared code of language by means of selecting words, syntactic structures, etc. and ordering the ideas. In addition, writers monitor the whole process of text production, for instance, after writing a word, a sentence, a paragraph or a complete text. As a result of monitoring, writers will review or revise their texts via evaluating (audience, purpose, and occasion appropriateness), revising (content, structure, etc.), and editing (grammar, spelling, mechanical conventions, etc.).

In sum, the present study considers EFL writing competence as comprising both executive resources and executive processes. However, the list is not meant to be exhaustive since no operational construct definition can fully reflect everything in such a complex process as EFL writing.