2.1 Mental operations in writing

Writing，a complex communicating act，involves an interaction between the writer and the reader whereby the former pays careful attention to the need of the audience to make sure his or her written product turns out comprehensible，persuasive，or informative—whatever the writing context requires.Since the primary goal of the present study is to examine the subjects’strategy use in integrated writing tasks，as suggested by Manchón et al.（2007），

“the study of writing strategies should be viewed within a wider research movement known as‘process writing’，which emerged in the field of native language（L1）writing with the aim of gaining insights into the mental actions writers engage in while composing.”（p.229）

we should start from the review of mental operations in writing while synthesizing a relatively comprehensive theoretical framework for this study.

2.1.1 Cognitive operations in writing

As noted earlier，researchers in English writing studies“have come to recognize the complexity and multidisciplinary nature of writing research and teaching”（Matsuda 2006：23）.In the 1960s，researchers on L1 writing concentrated almost exclusively on the formal characteristics of writing.Writing has often been perceived as three distinct and linear phrases，i.e.，pre-writing，writing，and revising，with the aim of producing an accurate and clear written product.In the 1970s and 1980s，however，writing has been conceptualized as a complex and nonlinear cognitive activity incorporating a set of processes and strategies（Emig 1971；Flower&Hayes 1980）.

In the past decades，much research has been conducted to explore cognitive operations in writing.Cognitively-oriented studies on L2 writing has made considerable use of L1 writing models（Flower&Hayes 1981；Bereiter&Scardamalia 1987a），according to which writing activity entails an interaction of cognitive processes and mental representations that writers engage in to formulate，articulate，and refine their ideas while composing a text.

In the field of L1 writing，on which L2 writing scholars draw and build，the most particularly influential model of cognitive composing has been the one proposed by Flower&Hayes（1981），which was an entire shift from previous approaches that generally viewed the writing process as three ordered stages：pre-writing，while-writing，and after-writing.They were among the originators who examined the cognitive dynamics in writing as well as the nature of great variety that writing processes entails.

As indicated in Figure 2.1，writers cognitive model of writing comprises four active components—（a）task environment，referring to“everything outside the writer’s skin”，（b）the writer’s long-term memory，consisting of“knowledge of topics and knowledge of writing plans”，and（c）the composing processes（planning，translating，reviewing，and monitoring）（ibid.208-209）.According to them，writers must solve rhetorical problems：writing purpose，audience and specific individual goal.They found that successful writing relies on writers’capability of defining the problem and that experienced writers frequently define and redefine their goals as well as consider and reconsider the ways they expect to affect their audience（Flower&Hayes 1977）.Generally，writers represent their text in mind，which helps create and organize ideas，and then convert them into language，making conscious metalingusistic choices.Furthermore，they claimed that the sub-processes are exceedingly interactive and that the writer needs a monitor exerting effective control over these processes to select the most useful strategies at the appropriate time and to oversee their application.Knowing what，when and how to engage in different cognitive processes towards successful task performance is what metacognitive skills require.They，additionally，insisted that the cognitive processes of writing are of somewhat hierarchical nature（Flower&Hayes 1981）.

A hierarchical，highly embedded organization in which any given processes can be embedded within any other；that the act of composing itself is a goal-directed thinking process，guided by the writer’s own growing network of goals；and that writers create their own goals by generating both high-level goals and supporting sub-goals.（p.366）

Some researchers，interestingly，challenged this point by maintaining cognitive processes do not necessarily occur in time sequence.For instance，evaluation is not necessarily followed by revision（Breetvelt et al.1994；Negretti 2009）.Despite some criticism，the research conducted by Flower and Hayes helped set the evident association between cognitive strategies and writing performance.Moreover，Hayes（1996）updated the original framework postulated by Flower and Hayes（1980）by incorporating the writer’s motivation，affect，as well as both short-term memory and long-term memory.

New insights regarding conceptual changes and sophistication in writing processes came from Bereiter and Scardamalia（1987）who posited that differences in ability may be due to different writing processes.They proposed two developmental models：knowledge telling usually followed by less skilled writers，and knowledge transforming often followed by skilled writers.The first knowledge-telling model incorporates steps allowing writing without“external support”，which was considered quite similar to oral language presentation，for it does not need much planning or goal setting.The second knowledge-transforming model emphasized problem-solving during the writing process.They conceptualized“problem space”as the one in which the writer often alters or modifies his or her thinking while composing，and it is in this space that knowledge is transformed.Bereiter and Scardamalia’s models reiterated that while writing，writers are in reflective thinking processes whereby they are requested to solve both content and rhetorical problems.They validated their writing process models with empirical studies that shed light on（a）the differences between sophisticated and less sophisticated writers；（b）variability of processing demands for writing tasks；（c）significance of strategic planning and processing；（d）need for planning as moving beyond content generalization，and（e）necessity to develop self-regulating，evaluating and self-reflecting competences.

The cognitive models of writing have primarily examined the mental processes and representations writers go through while writing.In terms of their relationship，Roca de Larios et al.（2008）elaborated as follows：

Processes operate on these representations so as to transform an input representation into an output representation according to the goals writers set for themselves.Processes are also characterized by their dynamism in the sense that writers continually shift among them，for instance，among the planning of ideas，their evaluation，or their transcription.（p.31）

Thus，this permeability of processes makes writing cyclical in nature and allows the existence of an executive component that enables the writer to exert control over the succession and organization of operations that constitute the writing process（Bereiter&Scardamalia 1987a；Flower&Hayes 1981）.As the writing process proceeds，therefore，writers are supposed to adapt to the frequently changing situation（task environment）by adjusting their operations and strategies（van den Bergh&Rijlaarsdam 2001）.

Based on these two models，many processes and strategies have thus been investigated and reported in both cognitive psychology and L1 writing，which have contributed to L2 writing and strategy use research.The previous studies primarily focused on such cognitive processes as comprehending the task requirements（Purpura 1997，1998，1999）and retrieving information from the memory while performing a specific task.They also investigated cognitive strategies including translating（O’Malley&Chamot 1987；Oxford 1990），inferencing（Hayes 1996），clarifying，organizing information，composing（Odell&Goswami 1982；Witte 1987），and revising.However，the aforementioned research failed to explicate the underlying metacognitive mechanism exerting conscious control over cognitive processes or strategy use，which will be discussed next.

2.1.2 Metacognition

The term，metacognition，originally coined by Flavell in 1976，pertains to one’s ability to think about how he or she has learned or can learn best，which was composed of four components，i.e.，“knowledge about cognition that may help the learners improve their learning process，knowledge about oneself，knowledge of the learning task and knowledge of strategies available to complete that task”（Flavell 1976：232）.Flavell believed that learners are supposed to take conscious steps to exert effective control over their cognition for the purpose of improving understanding and retention.He further theorized that metacognition entails two general dimensions that take place when learners attempt to fulfill a task：metacognitive knowledge and metacognitive experiences.Metacognitive knowledge is typically concerned with learners’acquired knowledge or beliefs regarding cognitive processes，which comprises three categories：person variables，task variables，and strategy variables；it reflects one’s beliefs about tasks and their demands as well as strategies and their potential effectiveness.According to Flavell（1976），what distinguishes metacognitive knowledge from that stored in the long-term memory lies in how the information is employed，consciously or unconsciously，and when this knowledge is invoked by a certain activity or experience（p.907）.Metacognitive experiences refer to conscious ideas and thoughts about cognitive endeavor taken，mainly involving one’s use of metacognitive strategies or metacognitive regulation，which play an important role in planning，evaluating，or revising knowledge and strategies.

During the past decades，metacognition has got such synonyms as self-management（O’Neil&Speilberger 1979），metamentation（Bogdan 2000）or meta-learning（Cross&Steadman 1996），among which the most generally accepted rationale of metacognition is the one postulated by Brown（1987）.He proposed that metacognition entails two components：metacognitive knowledge of cognition as well as metacognitive regulation of cognition.The former，frequently called metacognitive awareness，refers to the extent to which learners are able to be conscious of and understand their processes and ways of thinking and learning；regulation cognition pertains to how learners make use of metacognitive awareness to manipulate their own thinking and learning.While some researchers argued that these two components are associated and overlapping during self-regulation（e.g.，Sperling et al.2004），it seems to suggest that metacognitive awareness might be a pre-requisite for metacognitive regulation.

Campbell and Dickinson（1992）thought of the term metacognition as several processes including“awareness of one’s preferred mode of learning，commitment and persistence in tasks，goal setting，attitude towards learning，risk-taking and paying attention”（p.149）.The core component of their interpretation of metacognition was“reflective analysis”whereby a learner could contemplate how he or she manages a particular learning situation and would thus develop personal learning preferences.Hofer（2004）asserted that metacognitive processes and behaviors involved monitoring and judgment and that learners were capable of activating their metacognitive awareness in the knowledge construction process.Moreover，he converted Flavell’s original concept of metacognition，“thinking about thinking”，into“knowing about knowing”（p.44）.He contended that the learners’belief about how knowing occurs，what is regarded as knowledge and where it resides，how knowledge is constructed and assessed may exert effects on their comprehension and cognitive processes.

The emergence and development of metacognitive theory has further promoted our understanding of the importance of reflection on human thought by providing increasingly clear pictures of how learners’awareness and control over their cognitive activities.Nevertheless，metacognitive theory poses epistemological and conceptual challenges.For instance，Wellman（1983）argued that the concept of metacognition itself was rather vague，potentially including a bunch of such intuitionally-identifiable activities in learning process as planning，monitoring，self-questioning，self-correcting，etc.Furthermore，some researchers indicated that it is ambiguous where to draw a discriminating line between cognition and metacognition（Brown et al.1983）.Bonner（1988）claimed that many definitions and classifications of metacognition，no matter how different from one another，are all associated with the learning process in which monitoring strategies are frequently used and that most researchers integrated the two approaches into a single perspective，emphasizing the significance of cognitive states and processes as well as control of metacognitive activities（Paris&Winograd 1990）.

In terms of the critical understanding noted above，Brown（1987）and Flavell（1976）argued that when a learner is studying a chapter in a science text，“ask yourself questions about the chapter might function either to improve your knowledge（a cognitive function）or to monitor it（a metacognitive function）”（Flavell 1976）.Moreover，Lazear（1991）made an important distinction by stating：

If the literal meaning of metacognition was thinking about thinking，then as far as we know，human creatures are the only ones that have this self-reflective ability，that is，the ability to step back from a situation in which they are involved and watch themselves.This capacity gives us an immense amount of freedom because in self-awareness lies the possibility of conscious change.We can observe our thinking patterns and actions in a given situation and learn from them.We can alter both our thinking patterns and our behavior for greater effectiveness the next time we are in a similar situation（p.76）.

The distinction between them is clearly informed here.Cross（1987）also claimed that metacognition was proved most effective when introduced to adolescents and young adults because they had capacity for self-awareness，self-reflection and self-evaluation.

Constructivism locates cognition and understanding within the individual by emphasizing the close association between self-reflection and knowledge construction as well as the extent to which it could contribute to the metacognitive skills.The advocators of Constructivism encourage learners actively to be involved in reflective and recursive behaviors for the purpose of attaining mastery.

Some research in very recent years has also proved the significant role of metacognition in language achievement in either ESL or EFL context.For instance，Pishghadam and Khajavy（2013）explored the role of both metacognition and intelligence in foreign language achievement among 143 Iranian EFL learners by using SEM as the major research instrument.They found that intelligence accounts for 12.2%of the variance and metacognition，17.6%of the variance in foreign language achievement.Thus，they concluded that metacognition plays an even more important role than intelligence as a predictor of foreign language achievement.Their findings confirmed the claims held by Chamot（2001）that metacognition acts as one of the primary factors in determining the effectiveness of learning another language.The major pertinent threat to its interpretive effects，nonetheless，consists in that the research just concentrated on the learners’total foreign language achievement rather than the independent components of language abilities，let alone the integrated language skills.Further research is needed to explore the role of metacognition with regard to the EFL subskills or integrated language abilities.

Kim（2013）investigated the metacognitive knowledge in L2 writing by using both quantitative and qualitative methods，finding that its components encompass“knowledge of 1）task，2）personal learning process，3）strategy，4）text and accuracy，5）problem solving，and 6）discourse features”（p.23），among which metacognitive knowledge of strategy use and metacognitive knowledge of problem solving act as significant predicators of L2 writing performance.The study made encouraging progress in analyzing the role of metacognition in independent language skill，i.e.，L2 writing.However，little is known regarding how their metacognition works when subjects perform integrated test tasks.

2.1.3 Metacognitive strategies in writing

ESL/EFL writers need to draw on a wide repertoire of knowledge and strategies while accomplishing a certain writing task.Canale（1983）suggested that strategic competence is the knowledge or awareness of strategies that either enhances or repairs communication.Grabe and Kaplan（1996）established a model of writing in which the“language competence”consists of three components：linguistic（grammatical），discourse，and sociolinguistic，which constitute the linguistic resources that can be activated by（learner-internal）metacognitive and（learner-external）social-cultural variables related to the writing context.In other words，strategic competence in their writing model“is accounted for primarily by the metacognitive processing through‘verbal working memory’”（Grabe&Kaplan 1996：236）.Similarly，as delineated by Hyland（2003），in order to successfully perform English writing tasks，L2 writers require grammatical competence，discourse competence，sociolinguistic competence，and strategic competence.He asserted that a writer，when faced with a communicative problem，improvises his or her solution by using a“variety of communicative strategies”appropriately（Hyland 2003：32）.This has been echoed by Celce-Murcia and Olshtain（2005），who further posited that language learners’strategic competence determines the extent to which they can deploy the knowledge and resources at their disposal to achieve their expected communicative purposes.

Of particular importance has been Bachman’s definition of language ability as“a capacity that enables language users to create and interpret discourse”，which embraces two components，i.e.，“language knowledge and strategic knowledge”（Bachman 2010：33）.Language knowledge includes organizational knowledge，i.e.，how learners produce sentences or texts with grammatical accuracy；and pragmatic knowledge，i.e.，how individuals produce sentences or texts with contextual appropriateness.Strategic knowledge is conceptualized as a set of metacognitive strategies，helping L2 learners formulate a goal in a language use domain or test task，assess what needs to be accomplished，and make a plan for doing it.Language use in an assessment context is interpreted as“the interaction between the subjects’language knowledge and topical knowledge with the specific task，mediated by the metacognitive strategies and affective schemata”（ibid：209）.The word“mediate”here serves as a significant notion in understanding the role of cognitive or metacongitive strategies in the context of task-based language assessments.In other words，without activating cognitive and metacognitive strategies，a test-taker will not possibly be able to successfully perform a task and hence the study on strategy use for a specific language test task，particularly the integrated writing task type，is in its own right quite essential.Bachman’s model of communicative language ability suggested the significance of studying strategy use particularly in an academic assessment context.

Writing strategies were typically viewed as processes，and they were all considered cognitive，which lead to the pervasive use of obscure terms in the previous studies of writing strategy use，but some of them failed to draw a distinction between processes and strategies.Neither did they consider cognitive and metacognitive strategy use as different constructs.The present study makes a distinction between them on the basis of previous research concerning learner strategies as well as educational psychology.Processes and strategies should be viewed as two distinct concepts.Processes are generally seen as an inclusive term involving the individual stages of various mental activities in information processing（e.g.，comprehending，supplying，and activating of information from memory）.Strategies refer to an individual’s conscious and tactful management of specific tactics（e.g.，inferencing and revising）that learners employ to facilitate achieving a particular goal（Seliger 1984；Bialystok 1990）.Processes are mainly automatic，requiring little resources，whereas strategies serve as a complementary system which exerts control over various processes consciously for the purpose of enhancing effectiveness in performing language tasks.For instance，when a learner’s general approaches are not adequate to address the complexity of a problem，intentional strategies will be used to settle the situation.Researchers on strategy use reiterated that the main distinction between processes and strategies consists in automaticity and purposefulness（Chamot 1987；Cohen 1998）.

Given the complex nature of writing，a distinction should also be made between cognitive and metacognitive strategy use.It is generally agreed among most language learning strategy researchers and educational psychologists that cognitive strategy use entails the strategies learners directly apply when performing the language task at hand，whereas metacognitive strategy use is seen as mechanism of the higher-order thinking that exerts control over language learning，use and task-taking process（O’Malley&Chamot 1990；Purpura 1999；Rubin 1981）.This was also supported by other researchers who defined metacognitive strategies as learners’use of self-management techniques to oversee and manage their cognitive goals in the contexts of both language learning and language testing（Brown et al.1983）.In other words，cognitive strategies may be considered directly and positively related to test performance，while metacognitive strategy use exerts an executive control over cognitive strategy use to accomplish the communicating goal（Purpura 1997，1998，1999）.

In order to investigate metacognitive strategy use in integrated writing tasks，it is necessary to first elaborate on the role played by metacognition in writing practice.Researchers have proved that metacognition plays a significant role in not only learning in general sense but also specific subject areas，such as mathematics，problem solving，and reading and writing（Veenman et al.2006）.Given the fact that writing is a significant self-reflective and agentive instrument for learners concerning their own understanding of a topic by allowing them to express their understandings，confusions，and feelings about it（Bangert-Drowns，et al.2004），metacognition is highly integral to the writing process.Emig（1971）proved to be one of the earliest studies investigating metacognitive strategy use in writing in the history of relevant research，which elaborated that the act of writing comprises the writer’s goal-setting，idea-organizing，self-monitoring，and evaluating processes，requiring that he or she gives proper attribution to metacognition（which she defined until 1977）.The operation of metacognition in writing，furthermore，she held，was dependent upon both spheres of human brain.A key metacognitive strategy articulated by Emig was“self-provided feedback”，which meant the learner self-monitored and revised what had been written to improve the quality of writing.Additionally，Devine（1993）held that“metacognitive variables play an even more important role than linguistic competence in successful second language writing”（p.116）.

More recent studies have，furthermore，found that different mental activities have dominant effects in specific sub-processes of writing and that even exactly the same activities，such as self-monitoring，may have quite different functions in each process of writing.This has also been evidenced by Hacker et al.（2009）when they argued that writing can be considered applied metacognition.

“Reading，re-reading，and reviewing are monitoring strategies of our own thoughts.Editing，drafting，idea generation，word production，translation，diagnosing and revision are used as control strategies of our own thoughts.The monitoring and control of our own thinking is metacognition.Writing is applied metacognition”（p.161）.

They acknowledged the importance of metacognitive monitoring and control in writing process.The research conducted by Myhill&Jones（2007）indicated the extent to which metacognition affects every stage of writing process，ranging from the overall planning stage，the language choices while making ideas explicit into words，to the self-monitoring and revising processes during and after the writing act.It has been evidently proved that metacognition plays such an important role in promoting writing proficiency and cognitive maturity，for it makes it possible for learners to self-regulate seemingly unconscious thinking processes.Research in terms of educational psychology also supports similar opinions.Zimmerman and Bandura（1994）investigated the writing process of English college students；they found that the extent to which writers could effectively fulfill the writing task somehow depends more on their metacognitive and self-regulating efforts than on their linguistic capabilities.

Kasper（1997）investigated the effects that the three components of metacognitive knowledge（i.e.，person，task and strategy）may have on EFL writing performance.He attempted to figure out whether these variables have an equivalent effect upon writing performance，whether they affect performance equally across different levels of language proficiency and how metacognitive knowledge develops when writers improve their linguistic competency.The researcher also emphasized the significance of promoting EFL students’metacognitive awareness in writing instruction.

Each of the cognitive processes is managed or regulated by a set of“control”strategies.In the context of ESL/EFL writing，metacognitive strategies are believed to exert an executive control over writing performance.Despite the fact that they do not transform written products directly，metacognitive strategies are usually applied to manage the writers’cognitive strategy use as they perform the writing task（Bachman&Palmer 1996；Brown et al.1983；Faerch&Kasper 1983；O’Malley&Chamot 1990）.Purpura（1996）depicted metacognitive strategy use as a set of techniques entailing planning，considering the audience，monitoring as well as evaluating.

The first strategy，planning，involves making decisions of allocating resources，ordering the steps to follow while performing the writing task，and setting the intensity or speed at which the writer should work in various stages（Kluwe 1982）.Planning also involves setting writing focus，goals，content，and even transforming knowledge in response to the requirements of a specific writing task or social context（Flower&Hayes 1980）.Planning not only occurs at the very beginning of any writing activity but recurs throughout the performing process.Some researchers distinguished process planning from text planning.The former pertains to how the writer plans to take the writing task，whereas the latter is concerned with how the written text can be produced in terms of content，form，and its anticipated impact（Hayes&Nash 1996）.

Another metacognitive strategy is writers’consciousness of considering the audience.Writing properly for an audience requests an awareness and sensitivity about whom the writer is writing for and how to accomplish the writing goal.It is widely accepted that this strategy enhances the quality of the written text.While adjusting the writing to meet the audience’s expectations can be viewed as a cognitive strategy，realizing the fact that this is an important strategy itself is metacognitive，which is a significant conceptual distinction.For instance，sophisticated writers sometimes imagine for themselves a reader with down-to-earth interests and needs.This representation not only helps them make an appropriate response to the real rhetorical situation but allows them to generate ideas for their own written text（Flower 1981）.The metacognitive strategy of taking into account the needs of the intended audience allows the writer to use higher rhetorical skills for text generation（Tsai 2004）.With this strategy use，writers shape the content，tone，language，and style of writing to meet the audience’s requirements.

Monitoring proves to be a significant metacognitive strategy applied not only for checking the process of the writing but for making intended adjustments to the writing product itself.The ability to monitor one’s writing is considered one of the major differences across proficiency levels（Beach 1976；Tsai 2004）.Tsai（2004）made a distinction between micro-and macro-level monitoring.Micro-level monitoring pertains to invoking grammatical knowledge，whereas macro-level monitoring entails invoking discourse-level knowledge to supervise whether the argumentation is constructed logically，whether the transitional devices used can give the reader correct signposts appropriately，and whether the essay is coherently organized.

Monitoring is a strategy used when the writer is engaged in a writing task，while evaluating can be viewed as the one applied after the task is completed.Expert writers frequently take the inner measure of appropriateness，accuracy，and adequacy to evaluate their written products by making assessment about the language used，ideas，tone，and register concerning the requirements of the task.According to Flower&Hayes（1981），evaluating is a strategy writers apply while reading what they have already written for the purpose of either stimulating further writing or revising the text systematically.Kroll（1979）found that expert writers take their time to evaluate the written texts by means of a conscious examination of what they have written.The strategy of evaluating often promote further planning，idea generating，revising，and editing（Witte&Cherry 1994）.