1.1 Rationale for the study
Second language writing research is characterized by its complexity and multidisciplinary nature（Matsuda，2006）.Most researchers acknowledge that writing can be essentially described as“a‘more’and‘better’incorporation of various separate attributes that make up language performance”（Byrnes 2002：37）.In particular，the communicative competence framework，established by Hyland（2003），highlighting that ESL learners require at least grammatical competence，discourse competence，sociolinguistic competence，and strategic competence for the purpose of writing successfully in English.This framework“sheds light on the varying needs of students as they progress from Novice to Distinguished levels of proficiency”（Leaver&Shekhtman 2002：10）.The development of learners’communicative competence has long been perceived as the fundamental objective of ESL/EFL teaching.In achieving such a goal，the integrated-skill approach has increasingly been favored by both language instructors and assessors because of its simulation of authentic language use in real life，its stimulation of learners’strategy use，and its encouragement of learners’acquisition of content knowledge by means of multiple methods（Grabe&Stoller 2001）.
With regard to instruction，language instructors have attempted to simulate authentic language use in classroom settings in order to help learners better prepare for language tasks in real-life settings.Specifically，when learners are requested to perform complicated tasks in academic contexts，language instructors have to develop their integrated language skills by implementing instructions in which different language components are combined.
Similar benefits are applicable to performance assessment.In writing assessment，Weigle（2004）claimed that writing is seldom done in isolation，but is essentially conducted as a response to source texts instead.In recent years，furthermore，conventional impromptu writing-only tasks have been criticized as an inauthentic task type that cannot provide a valid assessment of academic writing ability in real language contexts（Hamp-Lyons&Kroll 1996；Leki&Carson 1997；Weigle 2002，2004）.As a response to this long-standing criticism，the use of integrated writing tasks for assessing writing ability in large-scale language assessment programs has been increasing in recent years（e.g.，General English Proficiency Test[GEPT]，Canadian Academic English Language Assessment[CAEL]，Internet-based Test of English as a Foreign Language[TOEFL iBT]，Certificate of Proficiency in English[COPE]），often as a substitute for impromptu writing-only tasks.The primary rationale for the integrated writing task type consists in the general assumption that writing test tasks for academic purposes should be associated with what learners are supposed to do in real language contexts in the sense that students in colleges or universities are frequently expected to write for the purpose of demonstrating their writing proficiency in different skill areas，including listening or reading（Cumming 2013；Weigle et al.2013；Leki，Cumming，&Silva 2008）；therefore，for academic purposes，the writing ability construct should comprise the ability not only to generate coherent written texts as measured by the conventional writing-only task type but also to compose written responses that are purposefully and appropriately related to ideas and information from given academic sources（Plakans 2009；Cumming 2013）.Research interest has accordingly been growing，as proved by a body of studies on integrated test tasks in recent years（e.g.，Cumming et al.2005；Asención 2004；Gebril 2009；Plakans 2008，2009；Yang&Plakans 2012；Kim 2013）.The intense research interest in the promising task type could also be evidenced by the fact that one of the most notable academic periodical in the field of language testing，Language Assessment Quarterly，published a special issue in 2013，addressing both theoretical rationales and empirical evidence for integrated writing test tasks.
Additionally，the past decades have witnessed the shift of research paradigm from how second language or foreign language should be taught to what makes a good language leaner，which has promoted researchers’interest in language learning strategies invoked by ESL/EFL learners and their effects on the development of language proficiency（e.g.，O’Malley&Chamot 1987；Wenden 1987；Abraham&Vann 1987；Purpura 1999）.Just a few researchers（e.g.，Tsai 2004；Ou 2013），however，have systematically concentrated on the writing strategies of good writers or what strategies help writers develop their writing skills.As claimed by both Victori（1995）and Beare（2000），it is the strategy use that primarily distinguishes successful from less successful writers，which has also been supported by Sasaki（2000）who found that expert and novice writers made use of different strategies or used the same ones but in different ways.Furthermore，it has been found that test-taking strategy research provides insights for test validation.In order to achieve a better understanding of the construct of test tasks，language practitioners should，as Bachman（2002）suggested，
collect data on the processes or strategies that test-takers use in responding to assessment tasks（e.g.，verbal protocols，observations，questionnaires，interviews，discourse that is created in the assessment process，such as speech or writing samples，physiological and neurophysiological responses）and utilize appropriate qualitative analyses to investigate the ways in which test-takers process language assessment tasks（p.470）.
In this sense，many areas concerning integrated writing tasks remain to be more fully explored.In particular，there is insufficient empirical evidence about how subjects use different strategies while performing those tasks.Further exploration is needed in terms of the extent to which subjects’strategy use interacts with their reading-to-write test performance.
Three research issues underlie the present study：（1）a lack of empirical research on strategy use reported by subjects while performing integrated writing tasks in Chinese EFL contexts；（2）a lack of information concerning the interrelatedness between EFL university students’strategy use and their performance on integrated writing tasks across scoring levels；（3）the scarcity of research employing a mixed-method approach for investigating strategy use.
First，it is noteworthy that there has been a lack of empirical study focusing specifically on subjects’reported strategy use while performing an integrated reading-to-write task particularly in Chinese EFL contexts.While numerous previous studies explored how language learners employ strategies in an attempt to acquire a language skill，very few focused on their strategy use in assessment contexts（Tsai 2004；Yang 2009；Yang&Plakans 2012）.Additionally，according to Macaro（2007），learners’combined uses of a strategy with others could improve performance in a given task.There is，thus，an obvious lack of systematic research addressing the issue of how EFL subjects actually deploy meaning-constructing，metacognitive and cognitive strategies when engaged in an integrated reading-to-write task.
Second，much in-depth empirical information is needed to reveal the interrelatedness between EFL university students’strategy use and their performance on integrated writing tasks across scoring levels.It has been believed that the use of strategies（e.g.，metacognitive type）awakens learners’awareness to recognize learning contexts and processes and thus leads to more efficient learning and improved performance（Anderson 2002）.Thus，we essentially need more substantial evidence to address the complexities of the relationship between subjects’strategy use and their performance.
Third，concerning research methodology，most of the previous studies have adopted a single-method approach：either quantitative（e.g.，Cumming et al.2005；Yang&Plakans 2012）or qualitative research method（e.g.，Delaney 2008；Plakans 2006，2008）.There is a need for explicit and systematic use of mixed-method designs to concurrently collect different but complementary evidence which allows“triangulation”to investigate subjects’strategy use corresponding to interactive processes of writing from sources.
The purpose of this study is to explore the nature of subjects’strategy use construct in their completion of an EFL integrated reading-to-write task as represented by its generic components on the basis of the evidence from both questionnaires and qualitative think-aloud protocol（TAP）data.It also aims at specifying differences in terms of strategy use across scoring levels as revealed in TAPs and follow-up interviews.
With regard to data，the study involved four types of data：1）integrated writing test performance derived from 212 subjects；2）a questionnaire focusing on strategy use issued to these subjects immediately after the integrated writing test in attempt to elicit factors contributing to the construct of subjects’strategy use；3）16 subjects’TAPs while performing the integrated writing task recorded，transcribed and coded to derive information from which an inference concerning their strategy use could be made；4）follow-up interviews to the 16 TAP participants conducted with the intention of exploring their utilization of a variety of strategies.
The data were analyzed from the following four perspectives：1）Multi-facets Rasch Model（MFRM）was used to analyze the facets of the test that may contribute to variability of the test scores；2）exploratory factor analyses were conducted to identify variables from questionnaire data with which generate initial models of strategy use；3）structural equation modeling was utilized to measure the quality of these models.4）16 participants’TAP reports and follow-up interview transcripts were coded and qualitatively analyzed by means of Nvivo 8.0.The qualitative data elicited from TAPs and interviews were used to triangulate quantitative findings and provide further information about the strategy use of subjects across groups of different scoring levels.