Introduction
This study focuses on the test development and validation issues identified by the IELTS Joint Research Committee's Call for Proposals 2007/2008 Specifically, it explores the cognitive processes of IELTS test takers during Academic Writing Task One (AWT1) The research examines these cognitive processes at two key stages: before and after participants undergo targeted training aimed at enhancing their performance in AWT1 Additionally, this study addresses concerns related to the impact of testing.
“test preparation practice” – another area of interest identified by the IELTS Joint Research
This section outlines the theoretical foundation for the study, focusing on how graph features and test takers' "graphicacy" (Wainer, 1992, p 16) influence their AWT1 performance Further details regarding the research design and methodology will be discussed starting in Section 2.
Dearth of research into test takers’ cognitive processes of completing AWT1
In IELTS AWT1 tasks candidates are asked to “describe some information (graph/chart
Candidates are advised to allocate 20 minutes to describe data, processes, or events in their own words, aiming for a minimum of 150 words Their assessment focuses on the ability to organize and present information effectively, compare data, and explain how something functions, as outlined in the IELTS Handbook 2006.
AWT1 is an integrated writing task that requires candidates to comprehend and accurately re-present information from a graph in written English Various terms, including chart, graph, and diagram, have been used interchangeably in research (Friel, Curcio, & Bright 2001; Fry 1981; Wainer 1992) In the context of applied cognitive psychology, which underpins this research, the term "graph" will be used to encompass all related terms, including table, chart, and diagram.
Graph comprehension is essential for successfully completing writing tasks, as variations in graph input and differing levels of familiarity among candidates can undermine the validity of AWT1 in assessing academic writing skills Notably, only two research projects funded by IELTS have explored these concerns (Mickan, Slater, & Gibson 2000; O'Loughlin & Wigglesworth 2003), with Mickan et al addressing this topic as a minor aspect of their studies.
A study by (2000) focused on how test takers interpreted AWT1 task prompts and planned their writings, but did not consider the impact of graph characteristics on the writing process or outcome O’Loughlin and Wigglesworth (2003) explored how the difficulty of AWT1 tasks was influenced by the quantity and presentation of information in graphs Their product-oriented analysis revealed that writings based on graphs with less information were linguistically more complex than those based on graphs with more information, regardless of participants' language proficiency However, they found no significant differences in task difficulty across varying graph information It's important to note that their study utilized only three types of graphs (statistical table, bar chart, line graph) and focused on two specific topics, raising questions about the generalizability of their findings Future research could investigate whether similar results would emerge with different graph types and topics.
Research on the cognitive processes involved in Academic Writing Task 1 (AWT1) is notably limited compared to the extensive studies conducted on IELTS Academic Writing Task 2 While graphs have long been used as prompts in language tests, recent investigations into their impact on test-takers' performance are just beginning to surface Current studies primarily focus on the outcomes rather than the cognitive processes of participants when engaging with graphical information This gap highlights the necessity for deeper insights into how different presentations of graph data influence test-takers' understanding and performance, as well as the broader implications for language testing validity and the concept of graphicacy—defined as proficiency in interpreting quantitative information presented graphically.
Theories of graph comprehension in cognitive psychology and their implications for
implications for research into integrated writing tasks using graph prompts in language tests
The AWT1 integrated writing task requires understanding and re-presenting information from graphs in coherent written form This study focuses on two key research areas: graph theories in cognitive psychology and second/foreign language writing processes, with an emphasis on the former due to its underrepresentation in related studies We will examine essential factors influencing graph comprehension from a cognitive perspective, contrasting it with the view that graphs derive meaning from their contextual use The implications of these theories will be explored in relation to language test development and validation, particularly for IELTS AWT1.
In cognitive psychology, various models of graph comprehension have been proposed, including those by Carpenter & Shah (1998), Freedman & Shah (2002), and Hollands & Spence (1998, 2001) Key frameworks also include contributions from Guthrie, Weber, & Kimmerly (1993), Kửrner (2004), and Lohse (1993) Additional insights can be found in the reviews by Pinker (1990) and Shah & Hoeffner (2002), which summarize these models and their implications for understanding how individuals interpret graphical information.
The knowledge-based construction-integration model by Shah and colleagues is particularly relevant for research on integrated writing assessments using graphs as prompts, as it parallels Kintsch’s Construction-Integration model of text comprehension Graph comprehension is a specialized form of text comprehension that involves a sequence of interactions between conceptual and perceptual processes According to Carpenter and Shah, this includes pattern-recognition, interpretive processes for deriving qualitative and quantitative meanings, and integrative processes that connect these meanings to inferred referents from labels and titles Freedman emphasizes that graph comprehension is influenced by various factors, such as the display characteristics of the graph, cognitive demands, and the importance of graphical elements, along with a viewer’s domain knowledge, graphical literacy, and scientific reasoning skills.
& Trafton 2004, 2006)], and incremental interactions between these factors
Further empirical evidence supports this knowledge-based model of graph comprehension
Research by Meyer, Shinar, and Leiser (1997) highlights the varying efficiencies of line graphs, bar graphs, and tables in information processing, revealing that bar graphs excel in reading exact values and identifying maxima, while line graphs are better for discerning trends Participants often described data contrasts in bar graphs using terms like higher and lower, whereas they used trend-related terms such as rising and falling for line graphs This suggests that inherent features of different graph types influence how information is interpreted Additionally, Fischer, Dewulf, and Hill (2005) found that horizontal bar graphs and negative number graphs demand more decision-making time compared to their vertical and positive counterparts Pinker (1990) emphasized that the ease of interpreting graphs varies based on the type of information being extracted Carpenter and Shah (1998) noted that even simple graphs require complex cognitive processes Zacks and Tversky (1999) confirmed a tendency for readers to make discrete comparisons with bar graphs and assess trends with line graphs, coining the terms "bar-line message correspondence" and "bar-line data correspondence," which align with Tversky's (1995) principles of cognitive naturalness in graphic communication.
A knowledge-based model of graph comprehension highlights the differences between novices and experts, influencing their ability to understand graphs Friel, Curcio, and Bright (2001) identified four key factors affecting statistical graph comprehension: (a) the purpose of using graphs, such as for analysis or communication (Kosslyn 1989), (b) task characteristics (Simkin & Hastie, 1987), (c) discipline characteristics like data spread, type, size, and graph complexity, and (d) reader characteristics, including prior knowledge and biases (Vernon 1946) These factors align with the knowledge-based construction-integration model Empirical evidence supports the significant impact of reader characteristics on comprehension, with Carpenter and Shah (1998) asserting that individual differences in graphic knowledge are as crucial as the graph's properties This relationship also extends to students' academic achievements.
Bengtsson and Ottosson (2006) noted that graphic knowledge had the strongest correlation with mathematic/science achievements
Roth (2002) emphasized the complexity of graphic knowledge, urging a shift from mere familiarity to a broader understanding of how knowledge, experience, and expertise influence graph comprehension He classified graph reading into three categories: “transparent,” “competent,” and “problematic.” In “transparent reading,” familiar graphs act as a clear lens into known concepts, allowing readers to connect directly with the content without distraction “Competent reading” requires more effort as readers engage with less familiar graphs and their subjects Conversely, “problematic reading” arises when readers struggle with unfamiliar graphs or concepts, leading to difficulties in understanding and focusing on the graph's structure rather than its intended meaning.
Research on the impact of visual inputs, such as graphs and diagrams, on second or foreign language test performance is limited, despite their widespread use as prompts Most studies have focused on listening and speaking assessments, with minimal attention to writing evaluations Notably, the TOEFL Program temporarily halted the use of chart-graph formats in its Test of Written English due to findings indicating that this format led to higher mean scores Evidence suggests that test takers' familiarity with graphs influences their performance, as seen in Xi's (2005) study, which demonstrated varying effects of graph familiarity on speaking test scores These individual differences raise concerns about the validity and fairness of assessments that incorporate graphic inputs.
O’Loughlin and Wigglesworth (2003) highlight the need for a deeper understanding of the IELTS AWT1's validity, noting that previous studies have not adequately explored how graph characteristics and test takers' graphicacy interact to affect performance To fully assess the validity of AWT1 tasks, it is essential to analyze not only the written responses but also the cognitive processes of test takers during the tasks This comprehensive approach has guided the focus of the current research.
Research aims
This study aims to investigate the cognitive processes of IELTS test takers during the Academic Writing Task 1 (AWT1) using graphical prompts, focusing on the validity of AWT1 through a dynamic lens rather than a static one By examining cognitive processes at various stages—before and after test preparation—the research seeks to uncover the influence of preparation activities on task performance and address validity and fairness concerns associated with intensive AWT1 preparation Additionally, the study will explore the varying effects of different preparation methods on test takers' cognitive processes, providing critical insights into the complexities of these cognitive dynamics (Carpenter and Shah 1998).
Research questions
The study addressed the following research questions (RQ):
RQ1: To what extent are there differences in the candidates’ cognitive processes due to different AWT1 prompts?
RQ2: To what extent are the candidates’ cognitive processes affected by their graphicacy?
RQ3: To what extent are the candidates’ cognitive processes related to their writing abilities?
RQ4: To what extent are the candidates’ cognitive processes influenced by test preparation opportunities offered to them by the research team?
Approach
This research utilized a case study and grounded approach to gain in-depth insights into the cognitive processes involved in taking AWT1 We view the case study method as a strength, allowing for the collection of rich, detailed data that large-scale quantitative studies often overlook This approach not only enhances our understanding but also facilitates the development of hypotheses for future research across diverse contexts, including participants with varying language backgrounds and graph familiarity levels Additionally, we observed that most studies on test-taking processes in language testing have traditionally relied on different methodologies.
The current research stands out by utilizing a longitudinal design, collecting data from the same participants at multiple time points, which enhances our understanding of the cognitive processes involved in AWT1 tasks, unlike the previous "one-off" approach.
We followed the University of Bristol’s research ethics and data protection guidelines and regulations during the whole process of this project Written consent was obtained from each participant
We set out our research methodology below.
Participants
This research was conducted at a prominent university in China, known for its high volume of undergraduate and postgraduate students taking the IELTS academic module annually By selecting this university, the study aims to provide valuable insights into the validity of AWT1, benefiting IELTS partners more than a smaller institution would With the support of the Foreign Affairs Office, the Graduate School, and the student Society of International Communication and Exchange, participation calls were disseminated across the university's administrative intranet and public communication platform, attracting significant interest from over 380 students and one faculty member Anticipating a large turnout due to the appeal of free AWT1 training, the researchers faced the challenge of selecting a limited number of participants based on criteria such as gender, department (science, social sciences, or arts), and academic status (undergraduate or postgraduate) The selection process involved carefully reviewing applicants' personal details, which were later verified through a graph familiarity questionnaire to ensure accuracy.
Undergraduate students must be in their final year of study, or fourth year for those in seven-year medical programs Master's students are required to be in their second year, although some programs may extend to three or four years Doctoral candidates can be in any year of their academic journey.
Participants in this research must have no prior experience with the IELTS test and are planning to take it by the end of 2008 or early 2009 This criterion is expected to enhance their engagement in the project, and only those who provided this information in their application were selected.
After applying our selection criteria, we identified 121 students in our database, comprising 21 doctoral students, 28 undergraduates, and 72 master's level students We sent a follow-up email to assess their availability from late October to early November, which helped gauge their commitment to the research project This process narrowed the pool to 53 students who were available daily that week From this group, we randomly selected 24 participants and created a waiting list to accommodate potential dropouts A few students later notified us of their inability to attend the first session for various reasons, allowing us to contact those on the waiting list, ultimately ensuring 24 participants were present when data collection began (see Table 1).
The study involved 24 students, comprising 13 females and 11 males from various university departments, including economics, finance, management, psychology, and engineering Among the participants, 12 were undergraduates and 12 were postgraduates, including five PhD candidates, with a distribution of 17 in science and engineering and 7 in social sciences and arts This selection mirrored the interests of 380 students who expressed a desire to participate Notably, none had taken official IELTS tests, but all intended to do so by early 2009 Each participant was assigned a unique code, ranging from A to X, for reference in the report.
Table 1: Cross-tabulation of 24 participants by gender and academic status
At the beginning of Stage 2 data collection, five participants (Students F, M, O, P, and V) withdrew from the study, followed by one additional dropout (Student A) at the start of Stage 4 Consequently, the final dataset comprised 18 participants, including 11 females and 7 males, with 11 undergraduates and 7 postgraduates, among whom were 2 PhD candidates.
Table 2: Cross-tabulation of the final set of 18 participants by gender and academic status
Data collection
This research collected both qualitative and quantitative data at five stages with different research focuses and instruments, as summarised below in time order
STAGE 1: Baseline data collection and think-aloud training
1 At the first meeting, the participants were briefed about the purpose of the project and the schedule of data collection Written consent was obtained from each participant (see Appendix 1)
The IELTS Academic Writing Tasks 1 and 2, designed by the research team in accordance with the IELTS Handbook (2006), assess participants' writing skills without the influence of verbal thought processes In Task 1, participants were instructed to summarize and compare the key features of a line graph that illustrates UK greenhouse gas emissions by various end users from 1990 onward.
In 2003, the AWT2 task prompted participants to express their level of agreement or disagreement with the statement: "Once children start school, teachers have a greater influence on their intellectual and social development than parents."
The graph familiarity questionnaire was administered to assess participants' understanding and experience with graphs, as outlined by Xi (2005) This self-evaluation tool included questions regarding participants' academic use of graphs, familiarity with various graph types, and their skills in graph comprehension, interpretation, and written communication The first 11 questions gathered personal information, including gender, IELTS experience, specialization, and year group To clarify the term "graph," the questionnaire provided Chinese explanations and examples, such as bar graphs, line graphs, pie charts, statistical tables, and flow charts.
Participants received training on the think-aloud method, initially practicing with simple mathematics word problems before progressing to an IELTS AWT1 task They were shown an example of think-aloud protocols to enhance their understanding of its significance Participants were encouraged to express their thoughts in English or Chinese, depending on their comfort level.
STAGE 2: First round of collection of the participants’ cognitive processes (ie, pre-training)
At the onset of Stage 2 data collection, participants received additional training and practice in the think-aloud protocol using the initial AWT1 task They commenced the AWT1 tasks only after feeling comfortable with the think-aloud method To minimize any additional information processing burden, all graphs utilized in the AWT1 tasks were designed in 2-D format (Carswell).
2 Administration of the first set of four AWT1 tasks, printed in colour (see Appendix 5, Set A) These graphic prompts included
– a simplified diagram/drawing showing the changes of the landscape or layout of an area from 1937 to 1995 (hereafter A-layout)
– a numerical table showing the top countries from which the USA imported crude oil (hereafter A-oil)
– a stacked bar chart reporting the instruction hours of children in OECD countries (hereafter A-instruction)
– a line graph describing the UK CO2 emissions trend (hereafter A-UKCO2)
In the AWT1 study, participants were assigned tasks in a random order, with each task lasting 25 minutes to accommodate the additional thinking-aloud requirement beyond the IELTS's recommended 20 minutes Participants verbalized their thought processes while completing the writing tasks, using digital voice recorders to capture their reflections To prevent interference and ensure individual recordings, several measures were implemented to minimize the influence of one participant on another.
– only four participants a time were allowed into a very big classroom – they were seated in the four corners of the classroom, and
– at no point did participants have the same writing prompt
Field notes on how the participants were engaging with the tasks were taken by the first author
STAGE 3: Training on AWT1 tasks
In a concise training session lasting over two hours, the first author provided participants with an overview of the IELTS test, focusing on AWT1 task prompts and rating criteria Utilizing recorded think-aloud protocols from Stage 2, the session illustrated the cognitive strategies employed by participants and highlighted common pitfalls to avoid, particularly the importance of refraining from drawing unwarranted personal conclusions based on data presented in graphs Participants were reminded of the dos and don’ts recommended by Cambridge ESOL, along with key vocabulary for describing trends and comparisons To enhance their familiarity with AWT1 tasks, they were given 12 diverse graph-based tasks for practice Additionally, each participant received a formal training booklet summarizing these essential points, available upon request from the authors.
STAGE 4: Second round of collection of the participants’ cognitive processes (ie, post-training)
In Stage 2, participants were familiarized with the think-aloud method, allowing them to articulate their thoughts while completing a second set of four AWT1 tasks This time, no additional training on the think-aloud technique was provided, as the participants had already gained experience with this data collection tool.
– a diagram showing the sequence of the events leading to the broadcast of a documentary by a TV programme (hereafter B-broadcast)
– a numerical table and the related coloured world map showing the amount of CO2 emissions by the top eight countries (hereafter B-map)
– column and pie charts describing grams of CO2 emissions per passenger/km of different transport methods and the EU funds spent on them (hereafter B-EUfund),
– a line graph showing the individuals viewing share of the major TV channels in the
Two students were asked to evaluate their training and experiences related to the AWT1 project, providing suggestions for its preparation Although this evaluation was not part of the original research proposal, it is essential for understanding how the training influenced their cognitive processes during AWT1 tasks (Research Question 4) The students completed a questionnaire (refer to Appendix 6) immediately after the final AWT1 task or at home, returning it the following day during their interviews.
Due to six dropouts, we modified our original plan to individually interview 50% of the 24 participants based on gender, graphicacy, and writing abilities, opting instead to interview all 18 remaining participants The interviews focused on three key areas: (i) how different graphic prompts affected their cognitive processes, (ii) the relationship between their familiarity with graphs and their comprehension and writing skills, and (iii) any changes in cognitive processes resulting from training Each semi-structured interview, lasting between 45 to 68 minutes, was conducted in English and/or Chinese as appropriate and was audio-recorded.
This research involved five distinct stages, utilizing a layered and progressive approach to data collection It assessed participants' academic writing performance without think-aloud protocols, graph familiarity, and comprehension in Stage 1 Stages 2 and 4 focused on AWT1 performances with various graph prompts while participants thought aloud Stage 3 provided training on AWT1 to the participants Finally, Stage 5 consisted of interviews aimed at exploring the participants' cognitive processes during AWT1 tasks.
The data were analysed first of all to understand the participants’ graph familiarity and their writing performances under the normal examination conditions and under think-aloud conditions (4.1)
Section 4.2 presents an overview of the participants’ cognitive processes Section 4.3 addresses the four research questions one by one.
Participants’ graph familiarity and English writing abilities
Participants’ graph familiarity
In Stage 1 of the study, data on graph familiarity was collected from 24 participants using a questionnaire, resulting in a total of 32 questions contributing to their graph familiarity score Responses to most questions were aggregated, with higher scores indicating greater familiarity, while negative questions were recoded for consistency Reliability analysis of the 32-item scale yielded a high Alpha of 0.948, confirming the consistency of participant responses The scores ranged from a minimum of 96 to a maximum of 182, with a mean of 143.8, indicating that participants were generally familiar with graphs Approximately one-third of the participants frequently utilized specialized software for graph production, highlighting their engagement with graph-related tasks The data distribution approached normality, suggesting a solid understanding of graph concepts among the participants.
A significant portion of participants, specifically 11 out of 24, frequently required graph production, while 6 out of 24 often needed to interpret graphs For additional insights into the participants' familiarity with graphs, please refer to Appendix 8, which details the frequency statistics of their responses.
Figure 1: Graph familiarity of the 24 participants
A preliminary analysis was conducted to explore potential differences in graph familiarity between male and female participants The findings revealed no statistically significant differences in graph familiarity, with both groups displaying similar mean scores.
11 male participants = 151.4, std deviation.0; mean of 13 female participants (n) 7.5, std deviation$.7 (t=1.52, df", n.s.)
After excluding six participants who dropped out, we conducted separate analyses on the remaining 18 participants, which yielded very similar results (see Figure 2) The mean score was 143.6 with a standard deviation of 24.96, and the difference between female and male participants was found to be insignificant.
Figure 2: Graph familiarity of the 18 participants
Below we report the participants’ AWT1 writing performance under normal examination conditions and when thinking aloud.
Participants’ writing performances under normal examination conditions
In a controlled examination setting, participants completed two writing tasks, AWT1 and AWT2, which were then processed and evaluated by the research team using IELTS writing band descriptors Each researcher independently assessed two-thirds of the randomly assigned written scripts to maintain a double-blind marking process If there was a discrepancy greater than one point between the two initial scores, a third rater was brought in to evaluate the writing The final score was determined either by the third rater's average of the first two marks or by the average of the two closest scores if the third mark differed significantly.
In cases where the three raters did not reach an agreement, face-to-face moderation discussions were conducted The initial consensus among the raters prior to moderation was unsatisfactory, primarily due to varying interpretations of the first set of rating criteria for AWT1.
Script ID Rater 1 Rater 2 Rater 3
Table 3: Double blind marking scheme
During the moderation meetings, we collaboratively reviewed and marked the scripts, which was crucial for enhancing rater reliability This moderation exercise significantly improved the average inter-rater reliability, achieving a Cronbach’s Alpha score above 0.83.
The AWT2 scripts exhibited a high level of inter-rater consistency, with Cronbach’s Alpha scores of 0.843 between Rater 1 and Rater 2, 0.881 between Rater 1 and Rater 3, and 0.793 between Rater 2 and Rater 3 Only two scripts had a rating difference greater than 1, which prompted the involvement of a third rater The performances of participants in these two writing tasks are detailed below.
N Minimum Maximum Mean Std Deviation
Table 4: Performance in AWT1 and AWT2 without think-aloud
In Table 4, the average performance of participants in AWT1 and AWT2 under standard examination conditions without think-aloud was 5.69 (standard deviation = 0.67) and 6.32 (standard deviation = 0.76), respectively The subsequent section details the participants' performances across eight tasks while utilizing think-aloud conditions.
Participants’ writing performances under think-aloud conditions
The research team conducted a double-blind assessment of the writings from eight tasks completed under think-aloud conditions, following the same marking procedure as the initial AWT1 task under standard examination settings The inter-rater reliability, measured by Cronbach’s Alphas, yielded satisfactory results: 0.69 for A-instruction scripts, 0.86 for A-layout scripts, 0.85 for A-oil scripts, 0.90 for A-UKCO2 scripts, 0.78 for B-broadcast, 0.74 for B-EUfund, 0.71 for B-map, and 0.79 for B-viewing The participants' performances across these tasks are detailed in Table 5.
Task A: UK CO 2 Line graph 18 4.75 7.25 6.03 150 635
Task B: broadcast Diagram of sequence of events 17 4.75 7.25 6.00 165 679
Task B: EU fund Bar graph and pie chart 17 4.75 7.00 6.04 144 595
Task B: map World map and statistical table 18 5.25 7.25 6.36 107 456
Table 5: Performance in the eight AWT1 tasks with think-aloud
A student, referred to as #T, articulated her thoughts verbally instead of producing a written response to the AWT1 tasks While her data was excluded from the performance analysis of the participants, it was considered in the examination of cognitive processes.
As shown in Table 5, the participants’ performance ranged from around 5 to 7.5, with mean scores from 6 to 6.5.
Developing a working model of cognitive processes for AWT1 tasks
Comprehending non-graphically presented task instructions
Participants began by reading the non-graphical task instructions prior to engaging with the graphs, but their attention varied across different sections of the text They strategically allocated their time, notably skipping the first line, which indicated a 20-minute time frame, and the last line, which required a minimum of 150 words, despite both being highlighted in bold This behavior suggests that participants were already familiar with these standard task requirements, leading them to overlook these specific instructions.
Participants frequently revisited two key sentences during their tasks, which served as introductory summaries The first sentence outlined the content of the graph, such as “The following graph shows the UK CO2 emissions by end users from 1970 to 2004” for Task A, and “The following map and table show the amount of CO2 emission by the top 8 countries” for Task B This summary provided clarity on the graph's topic, aiding participants in processing the information However, it acted merely as an entry point; participants still needed to identify the main features and make comparisons For instance, Participant K noted in an interview that the summary-like sentence in Task A guided their focus but did not encompass all necessary details for analysis.
The task instructions specify "different age groups," prompting the need to explore and highlight the distinctions among these groups After detailing the overall duration of the instructional hours, it remains unclear whether it is essential to address the variations between countries within these age categories.
In Section 4.2.3 we report on how the non-graphic task instructions were re-presented or re-produced in the actual writings
The consistency of the instruction to "summarise the information by selecting and reporting the main features, and make comparisons where relevant" across all AWT1 tasks led to unexpected participant behavior Instead of ignoring this directive as they did with the time allowance and expected length, participants exhibited increased re-reading of this sentence during the second set of tasks This change was linked to the brief training session, which emphasized the importance of describing rather than interpreting data Participants, initially inclined to analyze graphs based on prior knowledge, became more aware of their primary task—summarising and comparing the data—after the training Consequently, re-reading the instruction served as a reminder to focus on summarisation and comparison, minimizing personal interpretation.
Comprehending graphic information
Participants first reviewed the non-graphic task instructions before analyzing the graphs, focusing on identifying key features and making comparisons.
The cognitive transition from non-graphic to graphic information is a natural process that involves several iterative steps This includes reading, understanding, identifying key features, and interpreting graphs Ultimately, these processes lead to the re-presentation and reproduction of information in continuous written discourse, particularly in English as a foreign language Think-aloud protocols reveal the various activities that play a role in graph comprehension, aligning with Curcio's findings.
(1987) three levels of graph processing and comprehension: read the data, read between the data, and read beyond the data
Participants engaged in a systematic approach to interpreting graph data, beginning with localized searches for specific information and expanding to broader trends and patterns They continually referenced their existing knowledge on topics such as CO2 emissions and environmental protection, as well as their understanding of graph types and conventions The symbols and configurations used in graphs were crucial for readers, guiding their information processing—such as fact retrieval from line and bar graphs, trend identification, and proportion judgments from pie charts This indicates that the design and conventions of graphs significantly shape comparison methods, aligning with the cognitive principles of graphic communication Furthermore, participants' graphicacy, explanatory skills, and scientific reasoning abilities played a vital role in their cognitive processes and overall performance in task completion.
Re-presenting graphic and non-graphic information in continuous discourse
The iterative cognitive processes involved in completing AWT1 tasks encompass various factors, such as reading non-graphic instructions, interacting with graph elements (including titles, axes, data sources, legends, and color schemes), understanding graphs, extracting key data points, and making comparisons Participants were engaged in planning and organizing content for writing tasks, ensuring linguistic accuracy, and self-monitoring their work The knowledge construction-integration model of graph comprehension effectively explains the cognitive processes during these tasks, highlighting the unique aspect of AWT1, which requires candidates to reproduce both graphic and non-graphic information in a continuous written format in English This writing process not only facilitates better interaction with graphic information but also enhances overall graph comprehension, demonstrating the mutual benefits of comprehending and reproducing graphic data.
AWT1 tasks involve knowledge telling, where candidates reproduce meanings from task prompts, allowing them to utilize lexical and syntactic elements in their writing The significance of summary-like sentences in non-graphic task instructions is evident in participants' responses For instance, in Task A: UKCO2, all 19 participants began their writing with similar summary sentences, such as “The line chart illustrates the trend of the UK CO2 emissions by different end users between 1970 and 2004 in million tonnes carbon equivalent” and variations thereof This pattern underscores the effectiveness of clear introductory statements in conveying the main trends in the data.
Participants consistently began their writings by closely mirroring the summary sentence from the task instructions, stating, “It’s a graph which shows the UK CO2 emission by end user from 1970 to 2004.” This approach was uniform across all eight tasks When interviewed about their choice to start with this sentence, participants expressed that it served as an authoritative and effective introduction, setting the stage for the details that followed in their paragraphs.
Summary of the working model
The cognitive processes involved in AWT1 tasks are illustrated in Figure 3, which features a vertical timeline indicating the time allocated for these tasks The left column highlights the average duration of three key processes, revealing an overlap between "graph comprehension" and "reproducing graph comprehension in continuous written form in English." This overlap suggests that some students enhance their graph understanding through writing, rather than adhering to a linear comprehension process The initial stage of "comprehending non-graphically presented task instructions," particularly the summative introductory sentence, significantly influences subsequent graph comprehension This comprehension is shaped by the graph's display characteristics, the traits of the readers, their interactions, and the cognitive demands of the writing tasks In the final process of reproducing graph comprehension in written discourse, crucial elements include writing and revising main graph features, making relevant comparisons, expressing personal interpretations, and continuously monitoring and self-evaluating their writing.
Figure 3: A working model of cognitive processes for taking IELTS AWT1 tasks
Comprehending non-graphically presented task instructions
Reading time allowance (often skipped)
Reading expected length (often skipped)
Graph comprehension (e.g reading x-axis, y-axis, legends, recognising patterns and relationships between graphical components) Prior content knowledge
Prior knowledge of graph conventions
Graphicacy: experience and expertise of graph reading
Explanatory and other scientific reasoning skills
Re-producing graph comprehension in continuous written form in English as a foreign language Display characteristics of graphs including:
Types of graphs (e.g bar, line, map, diagram) and their presentational conventions
Relative importance and relationships of graphical components
Purposes of understanding the graphs and the cognitive demands of the writing task
Writing/revising main features of graphs
Writing/revising comparisons, trends, and predictions, based on different types of graph conventions
Making personal interpretations (sometimes) Constantly monitoring and self-evaluating
The English writing abilities of participants are crucial, as they significantly influence performance in AWT1 tasks At this stage, more generalized second language writing models are increasingly relevant and applicable.
This working model will be used to guide our analyses of the empirical data to address the four research questions in Section 4.3.
Addressing the four research questions
Research question 1: To what extent are there differences in the candidates’
processes due to different AWT1 prompts?
The cognitive processes involved in completing AWT1 tasks can be categorized into three essential stages: understanding non-graphic task instructions, interpreting graphic information, and reproducing both graphic and non-graphic information in continuous discourse To explore the impact of various graphic prompts on these cognitive processes, we utilized think-aloud protocols, interviews, and AWT1 scripts as our primary data sources.
In our study, we utilized various graphic prompts to enhance understanding, including bar graphs for Task A (Instruction) and Task B (EUfund), a linear layout drawing for Task A (Layout), statistical tables for Task A (Oil) and Task B (Map), and line graphs for Task A (UKCO2) These visual aids were essential in presenting data effectively and facilitating analysis.
B:Viewing), diagram of the sequence of events (Task B:Broadcast), pie chart (Task B:EUfund), and world map (Task B:Map) Two of the tasks used two types of graphic prompts In Task B:EUfund both bar graph and pie chart were used, and in Task B:Map, world map and statistical table
In section 4.2.1, it was observed that participants generally processed non-graphic task instructions similarly, with the exception of Participant T, who began by analyzing graphic information during Task A: Layout The key differences in cognitive processes emerged when participants engaged with graphic prompts and attempted to reproduce the main features of the graphic information in written discourse.
Participants exhibited significant variability in performance across eight tasks, as indicated by a Chi-square analysis (p