Investigating source use, discourse features, and process in integrated writing tests, A Gebril, L Plakans

Tags: Discourse Features, source texts, A. Gebril, academic writing, reading-writing, writing assessment, tasks, source text, proficiency levels, global warming, written products, Standard deviation, proficiency, writing process, reading ability, integrated tasks, independent tasks, undergraduate students, National Center for Policy Analysis, textual features, writing proficiency, global satellite data, independent writing, Oxford University Press, Modern Language Journal, L2 proficiency, writing instruction, writing activities, University students, writers, Cambridge University Press, language assessment, Journal of Second Language Writing, reading
Content: Spaan Fellow Working Papers in Second or foreign language Assessment Copyright © 2009 Volume 7: 47­84 English Language Institute University of Michigan www.lsa.umich.edu/eli/research/spaan Investigating Source Use, Discourse Features, and Process in Integrated Writing Tests Atta Gebril United Arab Emirates University Lia Plakans The University of Texas at Austin ABSTRACT As integrated writing tasks appear in more assessments of writing for academic purposes, research on their validity and usefulness is needed to assist users in interpreting scores. This study investigated features of writing from integrated reading-writing tasks as well as the process writers' used to complete them. One hundred and thirty-one undergraduate students at a United Arab Emirates university completed a reading-writing task on the topic of global warming followed by a questionnaire on their writing process. Their writing was scored and divided into three levels of writing proficiency. The written products were analyzed for discourse features, including lexical sophistication, syntactic complexity, accuracy, and fluency. They were also analyzed for source use by determining the amount of source use, direct (quotation) or indirect integration, and verbatim source use. All of these features became dependent variables in oneway analyses of variance to find if they differed significantly across the writing proficiency levels. The process-questionnaire items that related to source text use were also analyzed across the proficiency levels. The results indicated significant differences across levels for a number of discourse and source-use features; however, follow-up analysis indicated that the differences were greater between the lowest level and the upper two levels. The upper levels were not significantly different in terms of the writing features, which suggests that the writing at the higher levels was distinguished by other aspects such as organization, content, or coherence, which were on the rating rubric. The process results reveal all writers reporting used the source texts for ideas and forming opinions. Differences did appear in difficulties experienced by writers, coursework on integrated writing, and borrowing/citation from the readings. The implications for the study hold that integrated writing may be more distinguished at lower levels for discourse features, but other textual features such as cohesion, content, or organization differentiate higher-level writing. Further, a construct of integrated writing should include reading proficiency and knowledge about integrating reading with writing. 47
48 A. Gebril & L. Plakans Integrated reading-writing tasks are increasing in popularity and either replacing or complementing writing-only independent tasks used in assessing academic writing. The integrated tasks are seen to have more authenticity (Feak & Dobson, 1996; Weigle 2002, 2004) and may provide test takers with content, lowering anxiety and creativity demands on writing (Plakans, 2008; Read, 1990). These benefits extend to potentially positive washback effects in academic writing classrooms. (Cumming, Grant, Mulcahy-Ernt, & Powers, 2004). However, problems exist for these tasks, such as development, plagiarism, and constructrelated validity. In terms of validity, more is needed to understand what scores from integrated tasks infer about English language writing ability. One component of a validity argument is "explanation," which requires evidence such as the connection between an expected score and academic language proficiency (Chapelle, Enright, & Jamieson, 2008). Some research has looked at discourse features and process in writing assessment with relation to proficiency (Engber, 1995; Ishikawa, 1995; Jarvis et al., 2003; Machon et al., 2000; Ortega, 2003; Sasaki, 2000); however, most studies addressed the relationship based on impromptu writing tasks. Although holistic scores on independent and integrated tasks have been found to correlate (Brown, Hilgers, & Marsella, 1991; Gebril, 2006; Lewkowicz, 1994), the written products have been shown to have significantly different discourse features (Cumming, Kantor, Baba, Erdosy, Eouanzoui, & James, 2005, 2006). Thus, it may be inconclusive to base score interpretations for integrated tasks on research from independent writing tasks. To fill this gap, our research explored the connections between writers' scores on integrated test tasks and features of the written products as well as the process in completing the tasks with a group of writers from the Middle East. In a study of TOEFL prototype tasks, Cumming et al. (2005, 2006) investigated the writing features of three kinds of tasks: writing-only, reading-writing, and listening-writing. They analyzed and compared the written products across tasks and across proficiency levels emphasizing the importance of such analyses: A related issue concerns knowing if and how the written discourse may vary in the written compositions produced by examinees at different score levels on the test. The discourse of written texts cannot be assumed consistent for examinees with differing levels of proficiency in English, so consideration also needs to be given to how the written discourse of examinees varies in particular tasks with their English proficiency. This information is needed to verify, or refine, the scoring schemes being developed to evaluate examinees' performance on these writing tasks. (2005: 8­9). In their study, they found that, across proficiency levels, differences occurred in both essay length and number of clauses, indicators of syntactic complexity. The higher proficiency writers also used more variety in their word choice and scored higher on a grammatical accuracy scale. Lastly, the higher proficiency writers wrote longer compositions than the other groups, a finding that has appeared in other studies of writing assessment (Watanabe, 2001). Due to the use of sources in integrated tasks, score interpretations should also consider how writers integrate the reading material. A few research studies have looked at integration
Investigating Source Use, Discourse Features, and Process in Integrated Writing Tests 49 style in these tasks. Watanabe (2001) studied the use of summary, paraphrase, and quotation in writers' products. He discovered that quotation was used most, with some instances of paraphrasing and quotation. However, his analysis did not extend to the relation between these occurrences and writers' proficiency. Cumming et al. (2005, 2006) did consider this aspect of integration style finding summarizing more common for higher proficiency writers, along with better integration of the readings. Those writers who were midrange relied more on paraphrasing and verbatim source use, while the least proficient writers used sources least. The authors suggested that this last group struggled more with comprehension, thus did not use the content from source texts in their essays. Another line of research has focused on writers' verbatim source or plagiarism (Campbell, 1990; Currie, 1998; Deckert, 1993; Johns & Mayes, 1990; O'Connor, 2003; Pennycook, 1996; Rinnert and Kobayashi, 2005; Scanlon & Neumann, 2002; SutherlandSmith, 2005), suggesting that there is a relation between proficiency and verbatim source use. Johns and Mayes (1990) studied the writing of nonnative writers at two proficiency levels on a summary task. The findings showed no significant difference between the two proficiency groups in their textual borrowing in terms of distortion or sentence replication. However, other studies have found that proficiency impacts source use (Campbell, 1990; Currie, 1998; Cumming et al., 2005). Campbell concluded, "From this study it is seen that language proficiency affects the use of information from background reading text in academic writing" (p. 224). Thus, the impact of L2 proficiency on verbatim source use is possible, and the inconsistent results might be attributable to how researchers define and measure proficiency. Studies of L2 writers' source use outside of the assessment field have focused on either ESL students who study in North America or EFL students studying in South Asia. We are not aware of any studies that discuss the issue of source integration and plagiarism with students from the Middle East. In addition to uncovering relationships between integrated writing test scores and the written products, studying the test taker's process in completing these tasks is also important in explaining some of the differences or similarities found in the written products and to provide validity evidence. Studying process allows test developers to understand if the test measures accurately what we want to measure; in other words, validity (Ascensiуn, 2005; Cohen, 1998, 2007; Cohen & Upton, 2007; Plakans, 2008; Rupp, Ferne, & Choi, 2006). Bachman (2004) made this clear by stating: "There are two aspects of test performance that we need to investigate in our evaluation of test usefulness: the processes or strategies test takers use in responding to specific test tasks and the product of those processes or strategies.... In order to evaluate the usefulness of a given test, we need to investigate both aspects" (p. 5). Many of the recent process studies have analyzed data to learn about writer characteristics and strategies used in the process of writing. Research by Cumming (1991), Cohen (1994), Bosher (1998), and Sasaki (2000) has addressed these issues in L2 writing. In their studies, both writing-only and reading-to-write tasks are used. They include variables related to writer characteristics such as L2 proficiency, writing skill, writing expertise in L1 and L2, and instruction. This research on writer characteristics has clarified some important points in the L2 writing process. First of all, writing expertise has a strong impact on the L2 writing process; this impact is separate from L2 proficiency and perhaps more central. In terms of strategies used in the process of L2 composing, writers with higher L2 proficiency and more expertise and skill use more strategies.
50 A. Gebril & L. Plakans To build on previous research with integrated writing tasks, this study explored how discourse features, source text use, and process interact with writing proficiency in readingwriting tasks for students in a Middle Eastern context. The following research questions guided this study of integrated tasks: 1. Does writing proficiency level affect discourse features? 2. Does writing proficiency level affect use of source texts and verbatim source use? 3. How does writing proficiency level impact writers' process of source use? The results hold implications for both writing assessment and instruction. First of all, this study improves our understanding on how Middle Eastern students integrate information from sources in their writing. As mentioned earlier, this population is not given adequate attention in writing assessment research. Secondly, the research yields useful results concerning the writing features and use of sources in the performance of students at different proficiency levels. This information is useful in interpreting test scores from integrated tasks as well as for development of rating rubrics for such tasks. Lastly, the study provides some insight into the processes that L2 test takers follow while writing based on source texts; an area that needs further investigation as suggested by research (Cumming et al., 2005). This information will be helpful in making informed decision in test development, test use, and writing instruction, such as in creating rating scales or determining areas for curricular focus in teaching academic writing. Method This study was conducted with participants from a college of humanities and social sciences in a public university at the United Arab Emirates (UAE). The students who participated in this study represent a number of majors, including applied linguistics, general linguistics, translation studies, communications studies, geography, urban planning, and social work. Two instruments were developed and implemented: an integrated reading-writing task and a process questionnaire. Once the participants completed the tasks, the writing was scored and analyzed for discourse features as well as source use. The items on the questionnaire regarding source use were also processed to look at how writers reported on their source use. This section will detail these methods. Participants Students participating in the project were undergraduate students enrolled in humanities and social sciences programs. Courses at the university are taught in English; therefore, participants hold a level of English sufficient for university study because students go through a rigorous language program before choosing a major. Most students had taken the International English Language Testing System (IELTS), a measure of English language proficiency. The average mean score on IELTS was 4.82 with the mode of 4.5 (See Table 1 for participants' IELTS scores). It is important to mention that those students took the IELTS test after finishing their general education classes and before joining their majors. Most of those students took the IELTS test at least one year before data collection, and some of them took it two to three years earlier. So, because these students' proficiency level was much better than what the IELTS scores reflect, caution is warranted when interpreting their
Investigating Source Use, Discourse Features, and Process in Integrated Writing Tests 51 language scores. Data were collected from the women's campus, thus all participants were female. Initially, 139 students wrote on the tasks, but eight essays were removed from the data set because they were incomplete.
Table 1. Participants' IELTS Scores
Mean
4.82
Standard Error
0.04
Median
4.75
Mode
4.50
Standard Deviation 0.47
Instruments The Integrated Task The task for the study was developed to have some characteristics of authentic academic writing. Thus, we chose an argumentative essay prompt and two short reading passages presenting opposing points of view. The argumentative mode was selected based on the recommendations of a number of researchers (Plakans, 2008; Gebril, 2006; Melenhorst, 2006; Cumming et al., 2005). For example, Melenhorst (2006) argues that argumentative texts are approached successfully by students. The topic addressed in this prompt--global warming--was selected with the idea that it would hold students' interest, holds clear positions that writers can choose, and has possible source texts that included some solid evidence. Once the topic and format were determined, we spent time selecting the reading passages. Authenticity was again attempted, by finding sources from published magazines. One source was written by a credible scientist, while the other source was attributed to Reuters, so an author's name was added to increase the ease of citation for students. Two passages were selected following Lewkowicz (1994) recommendations to use more than one text. The texts were modified slightly for readability as well as parallel length and difficulty (see Table 2 for more information). In addition, following Carrell's recommendation (1987) to consider factors such as the nature of students, background knowledge, and cultural issues, the task was given to five experts, who were faculty members at the university where data were collected. They felt that the source texts were too long, some words too difficult, and instructions needed clarity. These issues were addressed in revision.
Table 2. Descriptive Data of the Reading Passages
Flesch-
Kincaid Flesch
Grade Reading
Level Ease Words Sentences
Text 1
12
47.9 275
11
Text 2
12
33.4 275
15
Words per sentence 25 18.3
Characters per word 1356 1437
52 A. Gebril & L. Plakans This revised version of the task was piloted with 46 undergraduate students from the same university, all of whom had English language levels similar to the research population. They also answered questions following the writing session about the task difficulty. Their responses and writing products were considered for revisions to the tasks. The biggest challenges for writers were difficult vocabulary in the source texts, lacking ideas for their arguments, and organizing ideas. In their writing, some students wrote summaries rather than essays. Time was also attended to in the pilot. Based on this session, the source texts were modified to make the vocabulary less difficult; the instructions and prompt were revised for clarity and offered clearer positions on the issues; and it was determined that the task should have an one-hour limit. The final integrated tasks are included in Appendix A. The Process Questionnaire This study employed a questionnaire to elicit writers' process of composing the tasks. Given the large number of writers and our previous use of concurrent verbal protocols (Plakans, 2008), a questionnaire seemed feasible and had potential to add new information about writers' process in integrated tasks. The questionnaire was developed to have a five-point Likert scale format presenting statements about the task and process followed by: 1 (strongly disagree) through 5 (strongly agree). Initially, 55 items were developed with the intention of selecting the best items through a multistage process of piloting. In addition, some questions were included to collect demographic and background information as well as six open-ended questions for qualitative analysis. The statements in the questionnaire were based on research on L1 and L2 writing processes (Anderson, 1991; Esmaeili, 2002; Grabe & Kaplan, 1996; Spivey, 1984, 1990, 1997; Watanabe, 2001) and our own previous research on the writing and reading processes in integrated tasks (Plakans, 2008, in press). To design the statements, seven interviews of writers who had completed an integrated task were reviewed for process statements. In addition, three writers' think-aloud transcriptions during an integrated task were analyzed. The questionnaire included statements to consider the research question proposed by this study, but also attempted to provide more information such as writers' attitudes toward the tasks, task effect, topic effect, planning before writing, and comprehension of reading. Based on these areas of interest and prior research, the main themes of the questionnaire included: reading effect, topic effect, task effect, writing process, validity, reading-to-write vs. writing-only, and source text use. The last area will be the focus of this report. The next step was to give the questionnaire to four experts: two who have extensive experience with questionnaires in second language acquisition research and two with backgrounds in academic writing. Their feedback provided suggestions on revisions and additions of items as well as formatting and scoring advice leading to a revision of the questionnaire. This draft of the questionnaire was completed by one student as a think-aloud protocol task following an integrated writing session. This pilot provided information on clarity, ease of answering, relation to writing task, and any other potentially problematic issues with the questionnaire. The transcription and analysis of the think aloud resulted in rewording several questions for more specificity. The questionnaire was given to another student to complete after an integrated task, followed by an interview to find out more specifically how the
Investigating Source Use, Discourse Features, and Process in Integrated Writing Tests 53 questionnaire represented her process. This session also led to rewording, clarification, and addition/deletion of several items. The next step was to pilot both the questionnaire and the task with 46 students at the university in the UAE. To investigate reliability of the questionnaire, Cronbach's Alpha was used and yielded a reliability coefficient equal to (0.87), which is a relatively high value. Fourteen items did not perform well when investigating item-total correlations and consequently were deleted. The final version of the questionnaire included 41 items (see Appendix B). Procedures Testing Session Several testing sessions were held during the spring of 2008 in a classroom at the university. Participants were first provided an information sheet describing the study, the procedures for the writing session, and information on their rights as participants. This information was also given verbally by proctors. If participants agreed to continue, they were given the writing task. Verbal instructions also were given with the written prompt. Then they received the process questionnaire, which they were asked to complete after they finished writing their essays. Students were told they would have one hour to complete the task. Each writer was given an identification number; no names were included with their work to assure confidentiality. Scoring Essays were rated using a holistic scale adopted from the integrated tasks in the TOEFL iBT but revised for this study based on a rating session with the pilot data to incorporate the following changes: (a) clarifying development and organization in the rubric to match the prompt evaluation criteria, (b) providing instructions for raters on scoring essays with plagiarism or with no source use, and (c) removing any references to listening features (the TOEFL rubric is used with essays written based on both reading and listening sources). From the pilot rating, essays were chosen to represent each score as profile and training essays. Two raters conducted the rating. Both were experienced ESL teachers who had also taught academic writing. They attended a rater training session conducted by one of the researchers. In the session, the study's purpose was discussed, and they were acquainted with the task and the rubric. Then the raters and the trainer rated one essay together followed by a discussion of scoring. The same process was followed for three more essays. Then raters scored five essays independently followed by discussion and calibration. This procedure was continued until both raters felt comfortable with the rating scale. Each essay was scored by both raters, with an interrater reliability coefficient of r = 0.75. No ratings were more than one level apart; on essays where the raters disagreed, a third rating was done by one of the researchers. In addition to providing a score for each essay, raters were asked to note their level of confidence with the score from 1 = no confidence, 2 = confident, to 3 = very confident. For both raters, this measure revealed that the raters ranged from highly confident to confident about the scores assigned as shown in Table 3.
54 A. Gebril & L. Plakans
Table 3. Raters' Confidence Indices
Standard
Mean
Error
Rater 1
2.83
0.03
Rater 2
1.88
0.04
Median 3 2
Mode 3 2
Standard Deviation 0.38 0.53
Once the rating was completed, the scores were used to group the writers into three levels of proficiency for the statistical analysis (see Table 4 for scoring and level information).
Table 4. Writing Scores and Proficiency Level
Scores
Number of Participants
Level 1
1­2
61
Level 2
3
49
Level 3
4­5
21
Analysis The essays and questionnaires were analyzed to answer the three research questions. First, discourse features were analyzed in the essays, followed by a comparison across the three levels of writing proficiency. Then, essays were analyzed for source text use and again compared across proficiency levels. Finally, items from the process questionnaire were grouped to consider aspects of source use across the writing levels. Each of these steps will be described in more detail in this section. Discourse Features Writing research uses many different methods to identify discourse features (Hinkel, 2002). In this study, we adopted guidelines by Cumming et al. establishing that indicators for such analysis should meet the following requirements (2005: 8­9): 1. include a range of discourse features including lexical, syntactical, rhetorical, and pragmatic characteristics 2. be applied reliably and meaningfully 3. possibly show differences between compositions scored at different scale levels We choose several discourse features as a focus for the analysis: lexical sophistication, syntactic complexity, grammatical accuracy, and fluency. However, before analyzing these separate features, some general characteristics of the essays needed tabulation: T-units, the number of sentences, the number of words, and characters. T-units were marked for all essays by two raters, while the other characteristics were calculated using a Mircosoft Word feature. Lexical Sophistication. This feature of discourse has been defined as the size of a writer's productive vocabulary based on the written essay (Wolf-Quintero, Inagaki, & Kim, 1998). Average word length was used to establish lexical sophistication based on prior research ( Biber, 1988; Cumming et al. , 2005; Engber, 1995; Frase, Faletti, Ginther, & Grant, 1999; Grant & Ginther, 2000). The proposed study adopted the definition used by Cumming
Investigating Source Use, Discourse Features, and Process in Integrated Writing Tests 55 et al. (2005: 9) for average word length, "the number of characters divided by the number of words per composition." Microsoft Word was employed in this study to calculate the targeted vocabulary measure. Syntactic Complexity. This feature measures "the range of forms that surface in language production and the degree of sophistication of such forms" (Ortega, 2003: 492). This feature particularly has had many different methods employed in previous research. We followed Ortega's advice (2003) to include several measures for this discourse feature and adapted methods used by Cumming et al. 2005; Henry, 1996; Homburg, 1984; Perkins, 1980; Tedick, 1990: 1. the mean number of T-units per sentences 2. mean length of T-units Grammatical Accuracy. This feature is a standard assessment of language proficiency and has gained increased interest in second language acquisition and L2 writing assessment (Polio, 1997). However, it remains somewhat elusive because of the difficulty in precisely measuring linguistic accuracy in a quantifiable way. Most of the quantitative measures used to judge linguistic accuracy, such as error counts (Fischer, 1984; Zhang, 1987; Carlisle, 1989; Kepner, 1991) and error count with classification (Bardovi-Harlig & Bofman, 1989; Chastain, 1990; Frantzen, 1995; Kobayashi & Rinnert, 1992; Kroll, 1990) have been problematic due to either the unreliability of coding schemes or not accounting for the error severity (Polio, 1997). For these reasons, this study used a holistic rating of grammatical accuracy adopted from Cumming et al. (2005), that was based on a scale used by Hamp-Lyons and Henning (1991): 1. many severe errors, often affecting comprehensibility 2. some errors but comprehensible to a reader 3. few errors, and comprehensibility seldom obscured for a reader Two raters used this scale to rate all essays for grammatical accuracy. The inter-rater reliability coefficient for these ratings was r = 0.94. Fluency. As is common in most studies of writing features, fluency was determined through word count, which has been successful in differentiating proficiency levels in many studies of second language writing (e.g., Cumming et al., 2005; Hirano, 1991; LarsenFreeman, 1978; Tedick, 1990). Source Text Use To answer question three, the use of source texts was analyzed in the essays. This variable required defining kinds of source text use and verbatim source use. Initially, a set of ten essays was selected and each T-unit was coded for summary, paraphrase, and direction quotation. A commercial Internet program called Turnitin.com was used to identify verbatim source use across all 131 essays. Based on this analysis, several decisions were made. First, we found that the distinction between summary and paraphrase was difficult for raters and often caused disagreement. As a solution, the two kinds of source use were combined as "indirect" source use and considered in contrast to direct use. There was some concern with the use of Turnitin.com because it was not clear how the program was identifying verbatim
56 A. Gebril & L. Plakans source use, and it searched the entire Internet as well as its essay base to find verbatim source use. As the two source texts were from Internet sources, many matches came up that were not relevant to verbatim source use in the writing sessions of this study. To resolve this problem, the raters coded "direct" source use as "no quotation" or "with quotation" to identify verbatim source use. The raters defined direct use as others have done (Cummings et al., 2005), but counting strings of three or more words from the original as verbatim source use. Following these guidelines, raters retrained on a small set of essays to achieve an inter-rater reliability of coefficient r = 0.97, then the raters coded the rest of the essays. Statistical Analysis of Discourse Features and Source Use Since the main purpose of this study was to investigate the relationship between one independent variable with three proficiency levels and a number of dependent variables, discourse features, a one-way analysis of variance (ANOVA) was used to analyze the results. In case of a statistically significant overall ANOVA, follow-up tests using the Tukey HSD formula were carried out to investigate which pairwise comparisons were significant. In addition, the researchers followed this analysis by a check on the homogeneity of variance, which is an assumption of one-way ANOVA. In case of violations of assumption, the BrownForsythe procedure and Dunnette's C follow-up test, which do not assume homogeneity of variance, were used to check the accuracy of results yielded from the ANOVA procedures that require this assumption. Furthermore, the researchers included descriptive statistics for the different discourse features. The following is a list of the dependent variables included in the analysis: 1. average word length 2. the mean number of T-units per sentences 3. the mean number of words per T-unit 4. total number of words 5. grammatical accuracy 6. indirect source use 7. direct source use with quotation marks 8. direct source use no quotation marks 9. total source use Processes The questionnaire developed included items addressing many areas of interest, but this report will focus on those relating to the initial research question regarding source use in integrated tasks. These 14 questions were grouped into subcategories so that several items addressed each (shown in Table 5). For each item the mean, standard deviation, median and mode were calculated at each writing proficiency level. The only qualitative data in the study came from the open-ended interview questions. These questions were analyzed using a semi-inductive method (Charmaz, 2004). First, the responses to the questions were divided across the writing proficiency levels. Then, within each level, a line-by-line coding was conducted to find patterns/themes emerging from the responses. Categories were defined by this initial coding and a second coding was conducted for these issues. This second coding was conducted both within levels and across levels. These categories were integrated with the four groups of questionnaire items to synthesize the
Investigating Source Use, Discourse Features, and Process in Integrated Writing Tests 57 quantitative and qualitative results. This process resulted in a more in-depth understanding of process to amplify the quantitative data from the questionnaire.
Table 5. Questionnaire Item Categories
Category
Question
Source text used for ideas
15. I used some of the ideas from the readings in my essay. 29. I used examples and ideas from the readings to support my argument in my essay. 32. I reread the readings while I was writing to find ideas to put in my essay.
Source text used in organizing
31. I used the readings to help organize my essay. 23. The reading helped me choose an opinion on the issue.
Integration process
14. I used some words from the readings when I wrote. 36. I paraphrased the reading in my writing. 38. I copied phrases and sentences directly from the reading into my essay. 35. I used the authors' names in my essays
Knowledge of
39. I have learned how to use reading with my writing in a
integrating reading-
class.
writing
40. I have learned about plagiarism.
41. I feel comfortable using other people's ideas in my
writing
General reading-writing 43. Explain any difficulties that you had with reading or
process
writing in this task.
(open-ended question) 44. If you used the readings for your writing, please describe
how you used them.
Results Discourse Features and Proficiency A one-way ANOVA was used to investigate the relationship between the independent variable, writing proficiency, and discourse features (shown in Table 6). The independent variable included three proficiency levels: levels 1, 2, and 3. This classification, as described in the methodology section, was obtained based on holistic ratings of students' essays. As for the dependent variables in this study, the researchers identified the following discourse features: grammatical accuracy, fluency, syntactic complexity, and lexical sophistication. In addition, the analysis looked into the relationship between proficiency and another dependent variable, which is the integration style of reading sources in students' writing.
58 A. Gebril & L. Plakans
Table 6. One-way ANOVA Results of Discourse Features across the Three Proficiency Levels
df
SS
MS
F
P
2
Fluency
Between Groups Within Groups Total
2.00 373729.48 128.00 416810.90 130.00 790540.38
186864.739 3256.335
57.385 43.800
.000 .000
.473
Grammatical Accuracy
Between Groups Within Groups
2.00 128.00
7.328 32.672
3.664 .255
14.355 .000
.183
Total
130.00
40.000
Syntactic Complexity
Words per T-unit
Between Groups Within Groups
2.00
34.552
128.00 1263.094
17.276 9.868
1.751 .178 .027
Total
130.00 1297.646
Mean # of T-units per sentence Between Groups Within Groups Total Lexical Sophistication Average Word Length Between Groups Within Groups Total Source Use Direct Source Use Between Groups Within Groups Total
2.00 128.00 130.00 2.00 128.00 130.00 2.00 128.00 130.00
3.2 289.686 292.887 .096 7.796 7.892 2.894 113.839 116.733
1.6 2.263
.707 .495 .011
.048 .061
.789 .456 .012
1.447 .889
1.627 .201 .025
Direct Without Quotation Between Groups Within Groups Total
2.00 128.00 130.00
2.180 311.927 314.107
1.090 2.437
.447 .640 .007
Indirect Source Use Between Groups Within Groups Total
2.00 373729.48 128.00 416810.90 130.00 790540.38
43.905 4.350
10.094 .000 9.955 .000
.136
Total # of Source Use Between Groups Within Groups Total
2.00 128.00 130.00
88.283 894.648 982.931
44.142 6.989
6.315 .002 .09
Fluency Fluency was estimated based on the text length (the total number of words in each essay). As shown in Table 6 and Figure 1, there were statistically significant differences in
Investigating Source Use, Discourse Features, and Process in Integrated Writing Tests 59 text length across the three proficiency levels (F (2, 128) = 57.385, p < .001, 2 = .473). Since Levene's Homogeneity of variance test revealed that the population variances across the three groups were not equal and therefore the homogeneity of variance assumption was violated, the researchers used the Brown-Forsythe formula to check the accuracy of the previous results. The Brown-Forsythe formula yielded similar results (F = 43.80, p < .001), which confirms the results obtained when homogeneity of variance is assumed. In general, this result showed that L2 writers produce longer essays when their proficiency improves.
Table 7. Pairwise Comparisons of the Fluency Variable
M
SD
Level 1
Level 2
Level 1
135.8033 40.88105
Level 2
226.1633 63.77119 -90.35999*
Level 3
268.8571 77.88792 -133.05386* -42.69388*
* statistically significant difference using Tukey HSD (P =.05)
Follow-up tests were conducted to assess pairwise differences among the three proficiency levels using the Tukey HSD test. Table 7 provides the means and the standards deviations of the three levels and also the results of the pairwise comparisons. As shown in Table 7, there were statistically significant differences between the mean of Level 1 and both the means of the other two groups. In addition, when the means of both Level 2 and Level 3 groups were compared, significant results were also obtained. Another follow-up test, Dunett's C, which does not assume homogeneity of variance was employed and yielded similar values to Tukey's.
Figure 1. Text Length across the Three Proficiency Levels
60 A. Gebril & L. Plakans
Grammatical Accuracy Grammatical accuracy was calculated based on a three-point holistic rubric used by two independent raters. As shown in Table 6, the ANOVA results yielded statistically significant differences across the means of three proficiency levels (F (2, 128) = 14.355, p < 0.001, 2 = .183). As shown in Table 8, the mean ratings of grammatical accuracy improved as the proficiency level got higher: (M = 1.7541 for the Level 1, M = 2.1633 for Level 2, and M = 2.333 for Level 3). In order to investigate whether these means are statistically different, the Tukey HSD test was employed. The results illustrated that the mean of the Level 1 group was statistically different from the other two groups. However, Table 8 shows that the mean of the advanced group is higher than that of the intermediate group; the pairwise comparison did not yield any statistically significant differences between the two means. This result showed the grammatical accuracy of Level 1 students improving when they move to the next higher level, while this is not the case in the other two groups
Table 8. Pairwise Comparisons of the Grammatical Accuracy Variable
M
SD
Level 1
Level 2
Level 1
1.7541
.53714
Level 2
2.1633
.47201
-.40917*
Level 3
2.3333
.48305
-.57923* -.17007
* statistically significant difference using Tukey HSD (P =.05)
Lexical Sophistication Average word length was used to measure lexical sophistication of students' essays based on the reading-to-write task. As shown in Table 6, results of the one-way ANOVA demonstrated no statistically significant differences in lexical sophistication across the three proficiency levels (F (2,128) =.789, p=.456). This result is due to the similar values of the means across the three proficiency groups (M = 4.63 for Level 1, M = 4.76 for Level 2, and M = 4.71 for Level 3).
Syntactic Complexity Syntactic complexity was calculated based on the average number of words in each T- unit and the mean number of T-units per sentences. As shown in Table 6, the ANOVA results showed no statistically significant differences in the mean number of words in a T-unit across the different proficiency levels (F (2, 128) = 1.751, p =.178). This result was expected given the similar values of means obtained from students at different proficiency levels (M = 13.3256 for Level 1, M = 14.1535 for Level 2, and M = 14.6448 for Level 3). As for the second syntactic complexity measure, the mean number of T-units per sentences, the results found no statistical significant differences across the three proficiency levels, which is similar to the results yielded in the analysis of the mean number of words in each T-unit.
Source Use in Reading-to-Write Tasks In order to investigate the source use patterns in reading-to-write tasks across different proficiency levels, the following indicators were used:
1. direct source use with quotation 2. direct Source use without quotation
Investigating Source Use, Discourse Features, and Process in Integrated Writing Tests 61
3. indirect source use 4. total number of source use in each essay
The first feature addressed in this section is the direct use of the sources using quotations marks. The ANOVA results showed no statistically significant differences in the use of quotations across different proficiency levels (F (2, 128) = .447, p = .64). The second feature, direct source use without quotations, also revealed no statistically significant differences (F (2, 128) = 1.627, p = .201) as shown in Table 6. Results of the indirect source use yielded statistically significant differences among Level 1, Level 2, and Level 3 students (F (2, 128) = 10.094, p < .001, 2 = .136). Levene's homogeneity of variance test showed that the population variances were not equal. Given this issue, the Brown-Forsythe formula was used to verify the accuracy of the previous results. Brown-Forsythe formula yielded very similar results (F = 9.955, p < .001) and consequently the null hypothesis was rejected. Following the rejection of the null hypothesis, the Tukey HSD and Dunnett's C procedures were used to compare the means of the three groups. Results of the pairwise comparisons, in Table 11, indicated that Level 2 and Level 3 students showed similar indirect use while the mean of Level 1 students was statistically different from the means of those two groups. The last feature addressed in this section is the total amount of source use in students' essays. This feature was calculated by tallying the number of source use incidents based on data obtained from the last three features: direct source use, direct source use without quotations, and indirect source use. Analysis of variance found statistically significant differences in the total number of source use incidents across the three proficiency levels (F (2, 128) = 6.315, p =.002, 2 = .09) indicating that the proficiency level affected the amount of source use in students' writing. The follow-up test demonstrated that there were statistically significant differences only in source use between the means of Level 1 and both Level 2 and Level 3 students as shown in Table 11. However, there were no significant differences between the means of the Level 2 and Level 3 groups in the total source use. This result is expected given the fact that both groups had similar means (M = 4.7143 for Level 2 and M = 4.8095 for Level 3).
Table 11. Pairwise comparisons of the source use indicators
Indirect Source Use
Proficiency Level Level 1 Level 2 Level 3
M 1.9016 3.5510 3.5238
SD 1.57803 2.63044 1.93956
Level 1 1.64938* 1.62217*
Proficiency Level Level 1 Level 2 Level 3
Total Source Use
M
SD
3.0984
2.05425
4.7143
3.09570
4.8095
2.80391
Level 1 -1.61593* -1.71116*
* statistically significant difference using Tukey HSD (P =.05)
Level 2 .02721 Level 2 -.09524
62 A. Gebril & L. Plakans Writers' Reported Source Use and Proficiency Level In addition to exploring writers' differences in discourse features and source text use, a questionnaire was used to delve into how their processes differed. Particularly this study focused on the use of the source texts. In this section, the results from analyzing the questionnaire will be detailed in terms of source text use for ideas and organization; integration processes; knowledge of integrating reading and writing; task difficulties; and integration style. Source Text used for Ideas and Organizing Consistent with other studies of integrated task processes, source texts were seen by writers at all levels as a resource for ideas (Plakans, 2008). Table 12 shows that all levels agreed that they used some of the ideas from the readings in their essays, with Level 2 having a slightly higher mean and most writers choosing strongly agree. In the open-ended questionnaire many writers reported using ideas from the texts; for example, a writer at Level 2 said, "I used this article to increase my information and I take some sentences to help me when I wrote essay and I benefit from ideas I read it."
Table 12. Question 15: "I used some of the ideas from the
readings in my essay."
Level 1
Level 2
Mean
3.57
3.80
Standard deviation
1.06
1.17
Median
4
4
Mode
4
5
Level 3 3.62 1.24 4 4
Question 29 was similar to question 15, but with a more specific purpose for using the source text ideas: supporting an argument. The task instructions asked writers to use the source texts in supporting their ideas, which encourages this strategy. All writers agreed that to some extent they used the ideas in the source texts as examples, as Table 13 illustrates. Interestingly, many at Level 1 and Level 3 strongly agreed with this, while most writers at Level 2 fell in the middle between agreeing and disagreeing with the statement. For Level 2 this contrasts with their answer to question 15, suggesting that they may be using the source texts to generate their own ideas, but not necessarily connecting author's ideas to their essays.
Table 13. Question 29: "I used examples and ideas from the
readings to support my argument in my essay."
Level 1
Level 2
Level 3
Mean
3.33
3.45
3.57
Standard deviation
1.34
1.28
1.29
Median
3
3
4
Mode
5
3
5
Investigating Source Use, Discourse Features, and Process in Integrated Writing Tests 63 Question 32 addressed whether the writers' returned to the source texts for ideas while they wrote rather than just reading them in planning. Table 14 shows that across the levels, writers did not report agreement or disagreement with this statement, which suggests that it might have occurred but only occasionally. However, at Level 3 most writers strongly disagreed with this statement. Several explanations could be given for this: they understood the reading more fully after reading one reading or they were less dependent on it for ideas as they wrote.
Table 14. Question 32: "I reread the readings while I was writing
to find ideas to put in my essay."
Level 1
Level 2
Level 3
Mean
3.20
3.00
3.05
Standard deviation
1.27
1.35
1.60
Median
3
3
3
Mode
3
3
1
Using source texts to organize essays during planning has been found in previous studies of integrated task process (Plakans, 2008). On the questionnaire, writers reported somewhat different uses of this strategy across levels as shown in Table 15. At Level 1 and 2, most writers agreed that they use the readings when organizing, while at Level 3 most strongly disagreed with this statement. The means across groups fell slightly into the disagreement range. These findings might suggest that some writers are using the source texts for more varied purposes than idea generation, however, not to the benefit of their score.
Table 15. Question 31: "I used the readings to help organize my essay."
Level 1
Level 2
Level 3
Mean
2.95
2.98
2.5
Standard deviation
1.33
1.37
1.29
Median
3
3
3
Mode
4
4
1
In contrast to using the readings for organizing, many writers at all levels used the readings to help choose an opinion. This questionnaire item was included in the category of organizing, as it would lead to selecting a topic sentence. However, it may also fall into the "generating ideas" category, which was strong for most writers. Table 16 displays this across writers. Most writers at Level 3 strongly agreed that they used the source texts to develop their opinions. Also, writers at levels 1 and 2 agreed with this statement. Because the readings presented two opinions on the topic, this strategy may be related to the task structure as well as the provision of source texts. One writer mentioned this in response to an open-ended question: "I read first and underline the main ideas that I understood. Then, I built my own opinion about this issue."
64 A. Gebril & L. Plakans
Table 16. Question 23: "The reading helped me choose an opinion
on the issue.
Level 1
Level 2
Level 3
Mean
3.74
3.65
3.52
Standard deviation
1.15
1.23
1.33
Median
4
4
4
Mode
4
4
5
In summary, writers at all levels reported using the source text to gain ideas and to decide on a position for their argument. The high- and low-level writers reported that they used ideas from the readings for support. The high level marked lower use of rereading the source texts, while writing and using the source text to help them organize. Integration Process As shown in Table 17, for all levels, there was a moderate level of agreement that they used words from the readings as they wrote. More agreement seemed to come from Level 2, as the mode was 5 and the median 4. In their open-ended responses, writers at levels 1 and 2 mentioned this integration process at the word level with statements like, "I get words from reading" or "I get ideas and some vocab."
Table 17. Question 14: "I used some words from the readings
when I wrote."
Level 1
Level 2
Level 3
Mean
3.54
3.63
3.3
Standard deviation
1.25
1.42
1.22
Median
4
4
3.5
Mode
4
5
4
When asked about paraphrasing from the source texts, shown in Table 18, writers did not report heavy use of this integration style. At all levels the means were on the side of disagreement, with the Level 3 writers slightly but not significantly lower.
Table 18. Question 36: "I paraphrased the reading in my writing."
Level 1
Level 2
Level 3
Mean
2.97
2.90
2.7
Standard deviation
1.21
1.44
1.49
Median
3
3
3
Mode
3
3
3
In terms of direct source use, most writers strongly disagreed that they copied phrases or sentences directly from the source texts. Table 19 shows that this finding occurred at all
Investigating Source Use, Discourse Features, and Process in Integrated Writing Tests 65
levels. This result suggests that writers were not aware of verbatim source use or did not wish to report that they had used this strategy.
Table 19. Question 38: "I copied phrases and sentences directly
from the reading into my essay."
Level 1
Level 2
Level 3
Mean
2.59
2.42
2.38
Standard deviation
1.44
1.49
1.56
Median
2
2
2
Mode
1
1
1
More difference occurred across levels when writers reported on citing authors of the source texts in their essays. The means across levels, displayed in Table 20, were somewhat similar, falling slightly below agreeing that they used this process. However, the mode for Level 3 was strong agreement with the statement regarding use of authors' names. For this question, there were also overall larger deviations from the mean suggesting a range of responses from the writers.
Table 20. Question 35: "I used the authors' names in my essays."
Level 1
Level 2
Level 3
Mean
2.67
2.96
2.85
Standard deviation
1.7
1.77
1.93
Median
2
3
2
Mode
1
1
5
In summary, some differences in integration process occurred across levels. The most striking difference was in citation of the source text authors with Level 3 writers strongly agreeing that they cited. The Level 2 writers reported using the words from the source texts more. All levels were neutral about copying words or phrases from the text. Knowledge of Integrating Reading-Writing While not directly addressing the use of source texts in the writing, the items regarding prior knowledge and experience with source text use can inform results regarding source use. These items explored how familiar writers were with the issues around Academic Reading and writing. Question 39 asked directly about previous learning in classes, showing that all levels reported some agreement with the statement (see Table 21). Level 1 and Level 2 had similar neutral responses to this item, while most writers in Level 3 reported that they strongly agreed with this statement. Their exposure to using reading with writing might have aided them in the task and led to a higher score.
66 A. Gebril & L. Plakans
Table 21. Question 39: "I have learned how to use reading with
writing in a class."
Level 1
Level 2
Level 3
Mean
3.36
3.49
3.52
Standard deviation
1.18
1.21
1.36
Median
3
3
4
Mode
3
3
5
Question 40 delved more specifically into integration practices by referring to plagiarism. As Table 22 shows, the responses from Level 3 writers were strongly in agreement with this statement, which could explain the results in the written products and their agreement with the statement about citation. Level 1 and 2 writers also slightly agreed but fell mostly in the middle.
Table 22. Question 40: "I have learned about plagiarism."
Level 1
Level 2
Level 3
Mean
3.10
3.37
4.0
Standard deviation
1.25
1.38
1.30
Median
3
3
5
Mode
3
3
5
Writers were also asked about their comfort level with using others' ideas in their
writing. As Table 23 shows, all levels were similar in response to this item, with no strong
agreement or disagreement. More writers at Level 2 reported disagreement about comfort than
the other levels.
Table 23. Question 41: "I feel comfortable using other people's
ideas in my writing."
Level 1
Level 2
Level 3
Mean
3.20
3.00
3.10
Standard deviation
1.11
1.34
1.26
Median
3
3
3
Mode
3
2
3
Task Difficulties In analyzing the writers' responses about task difficulties, some differences appeared across levels. Both the Level 1 and Level 2 writers frequently mentioned the vocabulary in the reading being too challenging. Level 3 writers did not remark on this, but instead had a different vocabulary-related issue. They felt that they did not have the scientific terminology to write on the topic. For example, one writer said, "I did not prepare for writing scientific essay. The time was not enough. I forgot some scientific words and how to spell them." Level 2 writers also responded that spelling was difficult for them, while Level 1 and Level 3 writers did not mention it at all.
Investigating Source Use, Discourse Features, and Process in Integrated Writing Tests 67 Level 1 and Level 2 writers felt that the task was difficult because they did not have enough information about the topic, did not understand the topic, and did not have many ideas to write about it. They also stated that organizing the essay was difficult for them. The Level 3 writers did not mention these difficulties, but instead commented on the time limit and the low interest level of the topic. Integration Style Writers' comments on integrating the readings are included with the summary of the quantitative data; however, some comparison from the qualitative analysis also revealed interesting patterns. The Level 1 writers reported mainly using the source texts for ideas and to "get words." Level 2 also mentioned gaining ideas and information from the reading, but also commented on using the source text to support their opinion. This use was also found in Level 3 comments. For both Level 2 and Level 3, a small number of writers noted that they did not use the sources in their writing. Level 3 writers answered the question somewhat more specifically by writing that they summarized the texts or used quotations. For example, one writer said, "I summarize the ideas in my own words and also I quote." Summary of Key Findings When writers reported on source use in their composing processes, overall the differences in proficiency levels were not huge. All writers reported using the source texts to generate ideas and opinions. The higher level writers seemed less dependent on the source texts; they reported rereading them less and reported lower rates of using the source texts for organizing their essays. The lower levels reported more difficulty with the vocabulary in the readings. All levels showed writers were neutral about copying directly from the source texts; however, the higher level writers reported more learning and knowledge about integrated reading-writing. These process results show that the writers did not report different processes when asked direct questions about it; however, they revealed some patterns that suggests higher scoring writers had more knowledge and sophistication with integrated reading-writing as well as possibly better comprehension of the readings. Discussion and Implications In this study, a number of discourse features were analyzed to investigate how these features behave across different proficiency levels. In addition, the study attempted to examine if certain features could play a greater role in characterizing a certain level. Furthermore, the questionnaire data were used to check whether L2 writers at different proficiency levels use similar or different strategies while working on reading-to-write tasks. The results showed statistically significant overall ANOVA across the three proficiency levels in fluency, grammatical accuracy, indirect source use, and total source use. These results exhibited effect sizes that ranged from medium to large values. However, the analysis of some other features, such as lexical sophistication, syntactic complexity, direct source use, and direct source use without quotations, did not yield significant differences. With regard to overall source use, results showed that students with higher proficiency generally exhibited more source use than those with low proficiency. Results of the fluency variable analysis demonstrated that L2 writers produce longer essays when proficiency increases. Fluency was the only variable that exhibited significant
68 A. Gebril & L. Plakans differences across the three groups. In addition, it produced the largest effect size among all the discourse features analyzed in this study. In general, L2 writing studies (e.g., Cumming et al., 2005; Grant & Ginther, 2000) yielded similar results. Grammatical accuracy also was another variable that produced significant results. However, pairwise comparisons showed that there were no significant differences between Level 2 and Level 3. This result may suggest that grammatical accuracy is a main factor when scoring essays at lower levels. Once proficiency increases, grammatical accuracy does not take precedence and other discourse features are at play. Analysis of syntactic complexity did not yield any significant results. The absence of significant differences across the three proficiency levels in syntactic complexity is in disagreement with the results of Cummings et al. (2005) who found significant results in the mean number of words per T-units at different proficiency levels. But, Cumming et al. (2005) found no significant differences in the amount of subordination measured by the mean number of clauses per T-unit. The Current Research used a coordination measure, the mean number of T-units per sentence (TU/S), not a subordination indicator. However, the current results are in agreement with a number of studies that showed no significant differences in syntactic complexity. For example, Hillock (1986) concluded that there is no consistent association between holistic ratings of writing and syntactic complexity of a written text. In addition, Ortega (2003) argued that studies establishing proficiency groups based on holistic ratings usually yield homogenous results across compared levels, which leads to a prevalence of statistically nonsignificant differences. She adds that it is not appropriate to equate syntactically complex writing with good writing. Another issue that may interpret the absence of nonsignificant differences in syntactic complexity is cross-rhetorical transfer. The students participating in this study speak Arabic as a native language. Consequently, it might be the case that they transferred a number of rhetorical strategies from Arabic. Cross-rhetorical influences in L2 writers with Arabic background would be an area of research that needs to be investigated in future studies. With regard to lexical sophistication, results yielded no significant differences across the three proficiency levels. This result is in disagreement with the study of Cumming et al. (2005) who found significant differences in average word length across different proficiency levels. However, results of Cumming et al. (2005) yielded a low effect size (2 p= .03). One possible reason for this result may be due to lexical borrowing from the source readings. It is expected that students would integrate words and phrases appearing in the reading sources in their writing. Therefore, this practice might have improved vocabulary sophistication across the three proficiency profiles. This conclusion is in agreement with East (2006) who found improvement in lexical sophistication when students used dictionaries. In addition, average word length does not take into account lexical accuracy, a linguistic feature that affects writing quality. Engber (1995: 194) found that `lexical variation without error correlated more highly with test scores than lexical variation with error.' The source-use analysis yielded a number of interesting results. For example, students at higher proficiency levels showed more overall source use compared to low-proficiency students. This result could be attributed to the complex and demanding nature of source integration. Cumming et al. (2005, p. 34) argue that effective reading-based writing requires "appropriate and meaningful uses of and orientations to source evidence, both conceptually (in terms of apprehending, synthesizing, and presenting source ideas) and textually (in terms of stylistic conventions for presenting, citing, and acknowledging sources)." Their study
Investigating Source Use, Discourse Features, and Process in Integrated Writing Tests 69 found the essays at lower score levels had less verbatim source use with integrated writing tasks. They concluded that lower proficiency writers struggled to comprehend readings in the integrated tasks, which led to their using them less. In the questionnaire, this level mentioned using the readings for ideas, but it also commonly cited difficult vocabulary as a challenge in completing the task, which suggests that reading comprehension may be related to lower source. So, probably there is a proficiency threshold that students should pass before successfully integrating reading sources in their writing. This conclusion was reiterated by a number of low-proficiency students in their responses to the open-ended questionnaire questions. Those students had trouble with numerous vocabulary items and structures in the two passages. Others referred to the overall difficulty of the two reading texts. When asked about their awareness of plagiarism and source integration issues, the majority of students across the three proficiency levels confirmed their attentiveness to these matters. However, the results demonstrated that those students often used inappropriate textual borrowing. One possible reason behind this behavior could be attributed to the students' lack of declarative knowledge about plagiarism, although they expressed the opposite in their questionnaire responses. This behavior is expected in the questionnaire methodology when dealing with a thorny issue, such as plagiarism, for social desirability reasons. In addition, the participants were undergraduate students with little experience in academic writing, so they may lack knowledge of textual borrowing strategies. This conclusion is supported by the results of a study conducted by Campbell (1990) who found frequent copying practices among both native and nonnative undergraduate writers. A more politically correct interpretation for this contradiction could be linked to patchwriting (Pecorari, 2003; Howard, 2001; Angeґ lil-Carter, 2000), which refers to the unintentional copying from source . For example, Howard (2001: 1) argues that these practices are caused by: uneven reading comprehension: the student doesn't fully understand what she is reading and thus can't frame alternative ways for talking about its ideas. Or the student understands what she is reading but is new to the discourse. She merges her voice with that of the source to create a pastiche over which she exercises a new-found control. In this quotation, Howard suggests that the declarative knowledge should not necessarily be the cause of inappropriate textual borrowing, but rather procedural knowledge. This was evident in the current study because students were instructed not to use patchwriting. However, this problem was common in many of the essays across the different proficiency levels. It is important that inappropriate textual borrowing be interpreted within a culturally situated framework. The educational context in which L2 writers operate is critical in understanding why they use sources improperly. A number of studies (e.g., Pennycook, 1996; Matalene, 1985) referred to these cultural aspects of plagiarism. For example, Pennycook (1996), who worked with Chinese students, concluded that many of those students had a different conceptualization of plagiarism from the Western paradigm. Some of them did not think of unattributed textual borrowing as plagiarism. Another study carried out by Kirkland and Saunders (1991) found that students in countries or cultures that embrace rote learning and memorization tend to use inappropriate textual borrowing. Based on the researchers' experience, teaching L2 writing at the undergraduate level in the UAE and the Middle East in
70 A. Gebril & L. Plakans general, writing instruction is still heavily focused on independent writing tasks. University students usually work on writing activities that target development of ideas, organization, mechanics, and more importantly grammar. There are a number of reasons behind this trend. First, undergraduate students at the UAE spend the first two years in general English classes where priority is given to basic writing skills. In addition, students are required to take the IELTS test before starting courses in their majors. The IELTS test mainly focuses on independent writing tasks, and that is why writing instructors work hard to prepare students for this task type. Given this context, few activities in these classes, if any, provide students with opportunities to work on reading-to-write tasks. As a result, students are not taught how to integrate reading sources and consequently are not aware of strategies used to integrate sources in their writing. With the current changes in writing assessment, it is hoped that beneficial washback would be achieved in L2 writing classes in the UAE context with more time assigned to source-based writing, and, more importantly, to how to integrate these sources in students' writing. These results hold a number of implications for language assessment in terms of construct definition, rating scales, and task development. A construct for writing or for integrated reading-writing has not be decided on in the field (Grabe & Kaplan, 1996; Hirvela, 2004). By investigating discourse features, source use, and process in writing for an integrated task, this study provided some insight into what might be included in an integrated readingwriting construct or, at the very least, what should be considered when interpreting scores from integrated tasks. For the lower level writers, different factors may be at work in scores than for the higher level writers because discourse features such as fluency and grammatical accuracy were more defining and limiting issues for their writing. The Level 1 writers may have been hindered more by general L2 proficiency or L2 reading ability, which impacts the writing features as well as their use of source texts in writing and their reading of source texts. Higher level writers may have been more facile with these features and have had more exposure to integrated writing or issues in academic writing such as plagiarism. These issues suggest that a construct for integrated reading-writing includes ability to control and create language elements of the written product, knowledge and use of integrating sources text, as well as general L2 proficiency and L2 reading ability. The results might suggest that there is a developmental difference in the weighting of the issues for the process and scoring of reading-writing tasks, with more attention on proficiency and discourse features at lower levels and more application of knowledge on reading-writing and source use at higher levels. These implications are speculative and will need more study with different populations and different tasks. Because construct should shape a rating scale for a writing test, the issues mentioned in the above paragraph might need consideration in scale development. The scale used in this study included issues of source use; however, the scoring was holistic, which causes difficulty in identifying how much source use impacted the final rating. The use of sources, as well as reading proficiency, may be features that raters need to consider in rating integrated tasks. In addition, the finding that discourse features and source use did not differentiate the upper two levels indicates that other features in the rating scale were distinguishing them, particularly organization, coherence, and content. For the lower levels, the writers may not have written enough to create cohesive, organized arguments. Other explanations may be that the raters focused more on language errors and use of source texts at the lower levels because more errors occurred, or the language errors may have prevented the raters from comprehending
Investigating Source Use, Discourse Features, and Process in Integrated Writing Tests 71 aspects of the essay such as organization and coherence. Certainly, these issues reveal the challenges of rating integrated tasks, and decidedly recommend further study of what raters attend to when rating such tasks, as well as the rubrics used in rating. The rating of integrated tasks is an area needing more study. Because this study only focused on one task, implications for developing integrated tasks are somewhat limited. However, the writers' questionnaire responses regarding their process suggest that several considerations need to be made by developers. Specifically, the selection of reading texts is no small matter. Writers need to be able to comprehend sources in order to integrate them. Many writers used the source texts for information and generating ideas, which indicates that for fairness across writers, readings should be accessible. If challenging texts are used, then scores need to include reading proficiency in the interpretation. More research is needed about the reading issue and readability levels of texts, as well as the role of reading for different kinds of integrated tasks; for example, summary writing, which might hold even stronger ties to reading ability. Limitations Several limitations in this study need to be acknowledged. First of all, the study used one task in order to control task effect. However, this decision creates issues about the topic and task. In the process questionnaire, writers indicated that they had different levels of interest, knowledge, and motivation for writing on this topic, which may have impacted their writing. The use of an argumentative essay is also a limitation, as many different kinds of integrated tasks occur in writing tests and in academic classrooms. Another decision was to focus on a homogenous group of writers to provide clearer results and implications. However, this narrow scope limits implications from the study. Caution should be used in applying the findings to other culture groups or different first-language writers. Lastly, the use of a process questionnaire may have limited the information on writers' source use. Questionnaire and survey data relying on self-reporting and, in this study, self-reflection on processes may have been hard for writers to access after writing, and thus hindered their ability to accurately or confidently report on their source use when they wrote. Conclusion Integrated tasks provide an interesting, innovative, and authentic means to assess academic writing. Writers and test developers see them as a potential solution to a number of dilemmas with impromptu independent tasks. This study investigated several aspects of the products and processes from writers completing integrated tasks in order to learn more about them. The findings show interesting differences between low and high proficiency writers, as well as revealing the integral relationship of the reading texts and writing in such tasks. More questions arose in this study and will require attention for the field to continue using integrated tasks and for test users to understand the resulting scores.
72 A. Gebril & L. Plakans Acknowledgements This study was made possible by a Spaan Fellowship for Studies in Second or Foreign Language Assessment at the University of Michigan. We would also like to thank the expert scholars who provided input in task and questionnaire development, as well as our raters and participants. The United Arab Emirates University was helpful and receptive in allowing our study to be conducted at their institution. References Angeґ lil-Carter, S. (2000). Stolen language? Plagiarism in writing. London: Pearson Education. Ascenciуn, Y. (2005). Validation of reading-to-write assessment tasks performed by second language learners. Unpublished doctoral dissertation, Northern Arizona University, Flagstaff. Anderson, N.J. (1991). Individual differences in strategy use in second language reading and testing. Modern Language Journal, 75(4), 460­472. Bachman, L. F. (2004). Statistical Analysis for Language Testing. Cambridge, UK: Cambridge University Press. Bardovi-Harlig, K., & Bofman, T. (1989). Attainment of syntactic and morphological accuracy by advanced language learners. Studies in Second Language Acquisition, 11, 17­ 34. Biber, D. (1988). Variation across speech and writing. Cambridge: Cambridge University Press. Bosher, S., (1998). The composing process of three Southeast Asian writers at the postsecondary level: An exploratory study. Journal of Second Language Writing, 7(2), 205­ 241. Brown, J. D., Hilgers, T., & Marsella, J. (1991). Essay prompts and topics: Minimizing the effect of mean differences. Written Communication, 8, 533­556. Burnett, S. (2001). Myths of global warming. Accessed on December 15, 2007, from http://www.ncpa.org/ba/ba230.html Campbell, C. (1990). Writing with other's words: Using background reading texts in academic compositions. In B. Kroll (Ed.), Second language writing (pp. 211­230). Cambridge UK: Cambridge University Press. Carlisle, R. (1989). The writing of Anglo and Hispanic elementary school students in bilingual, submersion, and regular programs. Studies in Second Language Acquisition, 11, 257­280. Carrell, P. (1987). Readability in ESL. Reading in a Foreign Language, 4(1), 21­40. Chapelle, C., Enright, M. & Jamieson, J. ( 2008). Score interpretation and use. In C. Chapelle, M. Enright & J. Jamieson (Eds.), Building a validity argument for the Test of English as a foreign language (pp. 1­25). New York: Routledge. Charmaz, K. (2004). Grounded theory. In N. Hesse-Biber & P. Leavy (Eds.), Approaches to qualitative research: A reader on theory and practice (pp. 496­521). Oxford, UK: Oxford University Press. Chastain, K. (1990). Characteristics of graded and ungraded compositions. Modern Language Journal, 74, 10­14.
Investigating Source Use, Discourse Features, and Process in Integrated Writing Tests 73 Cohen, A. D. (1994). English for academic purposes in Brazil: the use of summary tasks. In C. Hill & K. Parry (Eds.), From testing to assessment: English as an international language (pp. 174­204). New York: Longman. Cohen, A. D. (1998). Strategies and processes in test-taking and SLA. In L. R. Bachman & A. D. Cohen (Eds.), Interfaces between second language acquisition and language testing research (pp. 90­111). New York: Cambridge University Press. Cohen, A. D. (2007). The coming of age of research on test-taking strategies. Language Assessment Journal, 3(4), 307­331. Cohen, A. D., & Upton, T. A. (2007). "I want to go back to the text"; Response strategies on the reading subtest of the new TOEFL. Language Testing, 24(2), 209­250. Cooper, T.C. (1976). Measuring written syntactic patterns of second language learners of German. Journal of Educational Research, 69, 176­183. Cumming, A. (1991). Writing expertise and second language proficiency. language learning, 39, 81­141. Cumming, A., Kantor, R., Powers, E., Santos, T., & Taylor, C. (2000). TOEFL 2000 writing framework: A working paper. (TOEFL Monograph, No. MS-18 RM-00-05). Cumming, A., Grant, L., Mulcahy-Ernt, P., & Powers, D. (2004). A teacher-verification study of speaking and writing prototype tasks for a new TOEFL. Language Testing, 21(2), 107­ 145. Cumming, A., Kantor, R., Baba, K., Erdosy, U., Eouanzoui, K., & James, M. (2005). Differences in written discourse in independent and integrated prototype tasks for next generation TOEFL. Assessing Writing, 10, 5­43. Cumming, A., Kantor, R., Baba, K., Erdosy, U., Eouanzoui, K., & James, M. (2006). Analysis of discourse features and verification of scoring levels for independent and integrated tasks for the new TOEFL. (TOEFL Monograph No.MS-30 Rm 05-13) Princeton, NJ: ETS. Currie, P. (1998). Staying out of trouble: Apparent plagiarism and academic survival. Journal of Second Language Writing, 7, 1­18. Deckert, G. (1993). Perspectives on plagiarism from ESL students in Hong Kong. Journal of Second Language Writing, 2(2), 131­148. East, M. (2006). The impact of bilingual dictionaries on lexical sophistication and lexical accuracy in tests of L2 writing proficiency: A quantitative analysis. Assessing Writing, 11, 179­197. Engber, C. (1995). The relationship of lexical proficiency to the quality of ESL compositions. Journal of Second Language Writing, 4, 139­155. Esmaeili, H. (2002). Reading-to-write tasks and ESL students' reading and writing performance in an English language test. Canadian Modern Language Review, 58, 599­ 622. Feak, C. & Dobson, B. (1996). Building on the impromptu: Source-based academic writing assessment. College ESL, 6(1), 73­84. Ferris, D. (1994). Lexical and syntactic features of ESL writing by students at different levels of L2 proficiency. TESOL Quarterly, 28(2), 414­420. Fischer, R. (1984). Testing written communicative competence in French. Modern Language Journal, 68, 13­20. Frantzen, D. (1995). The effects of grammar supplementation on written accuracy in an intermediate Spanish content course. Modern Language Journal, 79, 329­344.
74 A. Gebril & L. Plakans Frase, L., Faletti, J., Ginther, A., & Grant, L. (1999). Computer Analysis of the TOEFL Test of Written English (TOEFL Research Report No. 64). Princeton, NJ: Educational Testing Service. Grabe, W. & Kaplan, R. (1996). Theory and practice of writing. New York: Longman. Gebril, A. (2006). Independent and integrated academic writing tasks: A study in generalizability and test method. Unpublished doctoral dissertation, The University of Iowa, Iowa City. Gebril, A. (in press). Score generalizability of academic writing tasks: Does one test method fit it all? Language Testing. Grant, L., & Ginther, A. (2000). Using computer-tagged linguistic features to describe L2 writing differences. Journal of Second Language Writing, 9, 123­145. Hamp-Lyons, L., & Henning, G. (1991). Communicative writing profiles: An investigation of the transferability of a multiple-trait scoring instrument across ESL writing assessment contexts. Language Learning, 41, 337­373. Henry, K. (1996). Early L2 writing development: A study of biographical of essays by university-level students of Russian. The Modern Language Journal, 80, 309­326. Hillock, G. ( 1986). Research on writing composition: New directions for teaching. Urbana, Illinois: Eric Clearinghouse on Reading and Communication Skills / National Conference on Research in English. Hinkel, E. (2002). Second language writers' texts. Mahwah, NJ: Lawrence Erlbaum. Hinkel, E. (2005). Handbook of research in second language teaching and learning. New York: Routledge. Hirano, K. (1991). The effect of audience on the efficacy of objective measures of EFL proficiency in Japanese university students. Annual Review of English Language Education in Japan, 2, 21­30. Hirvela, A. (2004). Connecting reading and writing in second language writing instruction. Ann Arbor: The University of Michigan Press. Homburg, T. J. (1984). Holistic evaluation of ESL compositions: Can it be validated objectively? TESOL Quarterly, 18, 87­107. Ho-Pong, L. (1983). Using T-units measures to assess writing proficiency of university ESL students. RELC Journal, 14, 35­43. Howard, R.M. (2001) Plagiarism: What should a teacher do? Paper presented at the Conference on College Composition and Communication, Denver, CO. http://wrthoward.syr.edu/Papers/CCCC2001.html, accessed December 27, 2008. Ishikawa, S. (1995). Objective measurement of low-proficiency EFL narrative writing. Journal of Second Language Writing, 4(1), 51­69. Jarvis, S., Grant, L., Bikowski, D., (2003). Exploring profiles of highly rated learner compositions, Journal of Second Language Writing, 4(12), 377­403. Johns, A., & Mayes, P. (1990). An analysis of summary protocols of university ESL students. Applied Linguistics, 11(3), 253­271. Kantz, M (1990). Promises of coherence, weak content, and strong organization: An analysis of students' texts. In L. Flower, V. Stein, H. Ackerman, M. Kantz, K. McCormick, & W. Peck (Eds.), Reading-to-write; Exploring a cognitive and social process (pp. 76­95). New York: Oxford University Press. Kepner, C. (1991). An experiment in the relationship of types of written feedback to the development of second-language writing skills. Modern Language Journal, 75, 305­313.
Investigating Source Use, Discourse Features, and Process in Integrated Writing Tests 75 Kobayashi, H. & Rinnert, C. (1992). Effects of first language on second language writing: Translation versus direct composition. Language Learning, 42, 183­215. Kroll, B. (1990). What does time buy? ESL student performance on home versus class compositions. In B. Kroll (Ed.), Second language writing: Research insights for the classroom (pp. 140­154). Cambridge: Cambridge University Press. Larsen-Freeman, D. (1978). An ESL index of development. TESOL Quarterly, 12(4), 75­84. Lewkowicz, J. A. (1994). Writing from sources: Does source material help or hinder students' performance? In M. Bird et al. (Eds.), Language and learning: Papers presented at the Annual International Language in Education Conference. ERIC Document (ED 386 050). Kirkland, M., & Saunders, M. (1991). Maximizing student performance in summary writing: Managing cognitive load. TESOL Quarterly, 25, 105­121. Manchуn, R., Murphy, & Roca de Larios, J. (2000). An approximation to the study of backtracking in L2 writing, Learning and instruction, 10, 13­35. Matalene, C. (1985). Contrastive rhetoric: An American writing teacher in China. College English, 47, 789­807 Melenhorst, M. (2006) . Highlighting professional writing: On screen note-taking as part of writing-from-sources by professionals. Unpublished doctoral dissertation, The University of Twente, The Netherlands. O'Connor, S. (2003). Cheating and electronic plagiarism: Scope, consequences and detection. Educause, 47(7), 18­24. Ortega, L. (2003). Syntactic complexity measures and their relationship to L2 proficiency: A research synthesis of college-level L2 writing. Applied Linguistics, 24(4), 492­518. Pecorari, D. (2003). Good and original: Plagiarism and patchwriting in academic second language writing. Journal of Second Language Writing, 12(4), 317­345. Pennycook, A. (1996). Borrowing others' words: Text, ownership, memory and plagiarism. TESOL Quarterly, 30, 201­230. Perkins, K. (1980). Using objective methods of attained writing proficiency to discriminate among holistic evaluations. TESOL Quarterly, 14, 61­69. Perlman, M. (2008). Finalizing the test blueprint. In C. Chapelle, M.K. Enright, & J. Jamieson (Eds.) Building a validity argument for the Test of English as a Foreign Language (pp. 227­258). New York: Routledge. Plakans, L. (2008). Comparing composing processes in writing-only and reading-to-write test tasks. Assessing Writing, 13, 111­129. Plakans, L. (in press). Discourse synthesis in integrated second language assessment. Language Testing. Polio, C. (1997). Measures of linguistic accuracy in second language writing research. Language Learning, 47, 101­143. Read, J. (1990). Providing relevant content in an EAP writing test. English for Specific Purposes, 9, 109­121. Reuters. (2005). Scientists Say Global Warming is Undeniable. Accessed on December 15, 2007, from http://www.abc.net.au/news/newsitems/200502/s1306233.htm Rinnert, C., & Kobayashi, H. (2005). Borrowing words and ideas: Insights from Japanese L1 writers. Journal of Asian Pacific Communication, 15(1), 31­56.
76 A. Gebril & L. Plakans Rupp, A. A., Ferne, T., & Choi, H. (2006). How assessing reading comprehension with multiple-choice questions shapes the construct: a cognitive processing perspective. Language Testing, 23(4), 441­474. Sasaki, M. (2000). Toward an empirical model of EFL writing processes: An exploratory study. Journal of Second Language Writing, 9(3), 259­291. Scanlon, P. M. and Neumann, D.R. (2002). Internet plagiarism among college students. Journal of College Student Development, 4(3), 374­385. Silva, T. (1993) Toward and understanding of the distinct nature of L2 writing: The ESL research and its implications. TESOL Quarterly, 27(4), 66­86. Spivey, N. (1984). Discourse synthesis: Constructing texts in reading and writing. Outstanding Dissertation Monograph. Newark, DE: International Reading Association. Spivey, N. (1990). Transforming texts: Constructive processes in reading and writing. Written Communication, 7(2), 256­287. Spivey, N. (1997). The constructivist metaphor: Reading, writing, and making of meaning. San Diego: Academic Press. Sutherland-Smith, W. (2005). Pandora's box: Academic perceptions of student plagiarism in writing. Journal of English for Academic Purposes, 4(1), 83­95. Tedick, D.J. (1990). ESL writing assessment: Subject-matter knowledge and its impact on performance. English for Specific Purposes, 9, 123­143. Watanabe, Y. (2001). Read-to-Write tasks for the assessment of second language academic writing skills: Investigating text features and rater reactions. Unpublished doctoral dissertation, University of Hawaii. Weigle, S. (2002). Assessing writing. Cambridge: Cambridge University Press. Weigle, S. (2004). Integrating reading and writing in a competency test for non-native speakers of English. Assessing Writing, 9, 27­55. Wolfe-Quintero, K., Inagaki, S., & Kim, H.Y. (1998). Second language development in writing: Measures of fluency accuracy and complexity. Honolulu: University of Hawai'i at Manoa. Zhang, S. (1987). Cognitive complexity and written production in English as a second language. Language Learning, 37, 469­481.
Investigating Source Use, Discourse Features, and Process in Integrated Writing Tests 77 Appendix A Writing task Read the question below, then, read the two passages to get more information about the topic. Write an essay on the topic giving your opinion in response to the question. Typically, an effective response will contain a minimum of 300 words. Your writing will be scored based on how well: - your ideas are explained - the readings support your argument - you organize your essay - you choose words - you use grammar - you spell and punctuate Some people believe that global warming is damaging our planet. Others believe that global warming is not a serious problem. Which point of view do you agree with? Why? Give reasons and support your writing with examples. IMPORTANT! PLEASE READ CAREFULLY BEFORE WORKING ON THE TASK: *- **TUhesetwsoppeacsisfaicgersesahsoounlds haenldp ydoeutagielst stoomseuipdpeoasrtabyoouutrthaentsowpiecr.. - You may go back to the passages to check information while writing. - You can borrow ideas and examples from the text. However, you should mention the author's name if you do so. - Also, if you take exactly the same phrases or sentences mentioned in the passage, put them between two inverted commas (" ").
78 A. Gebril & L. Plakans Reading (1): Scientists Say Global Warming is Undeniable (Adapted from an article by David Smith, Reuters, 2005) Scientists have confirmed that climate change is being caused by human activity. A number of studies looking at the oceans and melting ice leave no doubt that it is getting warmer, people are to blame, and the weather is going to suffer. Tim Barnett, who is a famous global warming researcher, indicates that new computer models that look at ocean temperatures instead of the atmosphere show the clearest signal yet that global warming is well under way. Mr. Barnett said that earlier climate models based on air temperatures were weak because most of the evidence for global warming is not in the air. Other researchers found clear effects on climate and animals. For example, Ruth Curry, who is from an important oceanographic institute, said changes in the water cycle affects the ocean and, ultimately, climate. She said the changes were already causing droughts in the United States, and Greenland's ice cap. Sharon Smith of the University of Miami found melting ice was taking with it plants that are an important base of the food supply for many animals. And the disappearing ice meant big Institution found that melting ice was changing the animals such as polar bears and seals were losing their homes. Given all these serious problems caused by global warming and the way humans have abused the earth, governments must act immediately to save our planet. The future of this planet depends on our actions and any delay would result in serious problems. Reading (2): Myths of Global Warming (Adapted from an article by Sterling Burnett, a Senior Fellow for the National Center for Policy Analysis, 2001) There is no scientific agreement that global warming is a problem or that humans are its cause. Even if current predictions of global warming are correct, much of the environmental policy now proposed is based on wrong theories. First, there is a wrong belief that the earth is warming. While ground-level temperature suggests the earth has warmed between 0.3 and 0.6 degrees since 1850, reliable global satellite data show no evidence of warming during the past 18 years. In addition, scientists do not agree that humans affect global climate because the evidence supporting that theory is weak. Some people also think that the government must act now to stop global warming. However, a 1995 analysis by supporters of global warming theory concluded that the world's governments can wait up to 25 years to take action with no bad effect on the environment. In short, our policymakers need not act immediately. The government has time to gather more data, and industry has time to develop new ways of reducing its influence. Supporters of the theory of human-caused global warming also argue that it is causing and will continue to cause all environmental problems. Many famous scientists reject these beliefs. Sea levels are rising around the globe, though not equally. In fact, sea levels have risen more than 300 feet over the last 18,000 years. Contrary to the predictions of global warming theorists, the current rate of increase is slower than the average rate over the 18,000-year period.
Investigating Source Use, Discourse Features, and Process in Integrated Writing Tests 79
Appendix B ID Number _____________________
Reading-to-write process questionnaire General Instructions The writing assignment you just completed is called a reading-to-write task, and we are interested in knowing what you thought about it. This is not a test, so there are no `right' answers, just think back and remember the best you can. Thank you very much for your help.
Opinions about the writing task Read the statements below and circle your feelings on the scale from 1 to 5. "1"= strongly disagree and "5" strongly agree
I. Writing process
Strongly disagree
Strongly agree
1. I understood the instructions for the writing task.
1234 5
2. I found global warming interesting to write about.
1234 5
3. It was clear to me how to complete the task.
1234 5
4. I have strong opinions on global warming.
1234 5
5. I have written papers on global warming before today 1 2 3 4 5
6. I made an outline before I started writing.
1234 5
7. I wrote a list of my ideas before I started writing.
1234 5
8. I did all of my planning before writing.
1234 5
9. I planned before and during writing.
1234 5
10. I wrote a good essay.
1234 5
11. The essay I wrote shows my ability as a writer.
1234 5
12. The readings helped me to write better.
1234 5
13. I liked reading about the topic before writing.
1234 5
14. I used some words from the readings when I wrote. 1 2 3 4 5
80 A. Gebril & L. Plakans
15. I used some of the ideas from the readings in my essay.
1234 5
16. I have worked on a reading and writing task like this one before.
1234 5
17. This seemed like a writing assignment that I would 1 2 3 4 5 have in an university class.
18. I think my score will be high on this task.
1234 5
19. I revised my writing before I finished.
1234 5
20. It was easy to support my opinion on this topic.
1234 5
21. I didn't have any trouble in writing this essay.
1234 5
II. Reading in the writing task
22. I looked back at the readings often while I was writing. 1 2 3 4 5
23. The reading helped me choose an opinion on the issue. 1 2 3 4 5
24. The readings were interesting to me.
1234 5
25. I could understand most of the words in the readings.
1234 5
26. I could understand the ideas in the readings.
1234 5
27. I have read articles on global warming in English before today.
1234 5
28. I have read articles on global warming in Arabic before today.
1234 5
29. I used examples and ideas from the readings to support my argument in my essay.
1234 5
30. I used the readings to help me get ideas on the topic. 1 2 3 4 5
31. I used the readings to help organize my essay.
1234 5
32. I reread the readings while I was writing to find ideas to put in my essay.
1234 5
Investigating Source Use, Discourse Features, and Process in Integrated Writing Tests 81
33. I could have written this essay without the readings. 1 2 3 4 5
34. I reread the readings while writing because I forgot what they were about.
1234 5
35. I used the authors' names in my essays.
1234 5
36. I paraphrased the reading in my writing.
1234 5
37. I used only my own ideas in my writing, nothing from the reading.
1234 5
38. I copied phrases and sentences directly from the reading into my essay.
1234 5
39. I have learned how to use reading with my writing 1 2 3 4 5 in a class.
40. I have learned about plagiarism.
1234 5
41. I feel comfortable using other people's ideas in my writing.
1234 5
IV. Comments Please write responses to the following questions. 42. Did you have enough time to complete the task? If not, how much more time would you have liked?
43. Explain any difficulties that you had with reading or writing in this task.
44. If you used the readings for your writing, please describe how you used them.
45. Would you prefer writing with reading on a test to a writing test without reading? Why or why not?
46. What would you tell a teacher who is thinking about using reading-to-write tasks in his/her classes?
47. Is there anything you would like to comment on about this task?
II Background information 48. How may years have you been at the university?
49. What is your major?
50. Have you had previous writing courses?
82 A. Gebril & L. Plakans If yes, please list them. 51. What is your level of experience with academic writing (circle best answer) No experience a little some a lot very experienced 52. What is your level of English language skill (circle best answer) low so-so medium high very high 53. If you have taken IELTS, what was your score? 54. If you have taken IELTS, when did you take it last? (circle best answer) Less than a year 1 year 2 years 3 years 4 years More
Investigating Source Use, Discourse Features, and Process in Integrated Writing Tests 83
Appendix C
Reading-to-write scoring rubric
(Adapted from the scoring rubric of the TOEFL iBT integrated task)
Score
Task Description
5 A response at this level: · successfully presents their ideas in relation to the relevant information presented in the reading sources. · is well organized with well-developed content · occasional language errors that are present do not result in inaccurate or imprecise presentation of content or connections. 4 A response at this level: · is generally good in coherently and accurately presenting their ideas in relation to the relevant information in the reading texts , although may have inaccuracy, vagueness, or imprecision in connection to points made in the readings. · has clear organization and logical development. · more frequent or noticeable minor language errors; such errors do not result in anything more than an occasional lapse of clarity or in he connection of ideas. 3 A response at this level · conveys some relevant connection to the reading, but only vague, global, unclear, or somewhat imprecise connection to points made in the reading . · development is somewhat limited, but some specific support for their argument is provided. · occasionally lacks cohesion but has a basic organizational structure. · includes errors of usage and/or grammar that are more frequent or may result in noticeably vague expressions or obscured meanings in conveying ideas and connections. 2 A response at this level · contains some relevant information from the readings, but is marked by significant language difficulties or by significant omission or inaccuracy of important ideas from the readings · lacks logical organizational coherence and development. Ideas are very general and lack specific details in support. · contains language errors or expressions that largely obscure connections or meaning at key junctures, or that would likely obscure understanding of key ideas for a reader not already familiar with the topic. 1 A response at this level · provides little or no meaningful or relevant coherent content from the readings and does not follow an organization pattern or develop content. · includes language that is so low and it is difficult to derive meaning.
84 A. Gebril & L. Plakans 0 A response at this level · either merely copies sentences from the reading, rejects the topic, not connected to the topic, is written in a foreign language, or is blank. Notes: -- If only one phrase or sentence is copied from the reading, do not assign a "0" but base the rating on the rest of the essay. -- If language use & development are at a certain level but the readings have not been included (directly or indirectly), add "NS" to the score -- If ideas from the readings are used but not cited, do not rate lower unless the writer is copying directly (use NC marking).

A Gebril, L Plakans

File: investigating-source-use-discourse-features-and-process-in-integrated.pdf
Author: A Gebril, L Plakans
Published: Tue Apr 28 09:00:09 2009
Pages: 38
File size: 0.5 Mb


ABOUT THE PRODUCTION, 27 pages, 0.53 Mb

, pages, 0 Mb
Copyright © 2018 doc.uments.com