1. Introduction

This chapter introduces Xu-argument-based continuation writing (XBCW) as a novel reading-to-write task in second language acquisition that effectively enhances vocabulary, phraseological knowledge, syntactic development, discourse structure, and overall writing quality among L2 learners. It highlights the insufficient exploration of real-time processing dynamics in XBCW, noting the lack of empirical evidence on learners’ moment-to-moment attentional patterns when engaging with source texts. Unlike summary writing or argumentative synthesis, XBCW demands learners to continue an incomplete narrative while maintaining thematic and stylistic coherence, requiring deep attention to linguistic forms and discourse features. The study addresses this gap by employing eye-tracking technology to capture fine-grained, non-interruptive eye-movement data, allowing the investigation of detailed stages of text processing from lexical access to semantic integration. This approach aims to clarify the unique cognitive mechanisms at play in XBCW, thereby advancing the understanding of how reading and writing interplay in task-based language learning.

2. Literature Review

This chapter discussed the theoretical foundations and empirical evidence supporting the use of xu-argument-based continuation writing (XBCW) in second language (L2) learning. XBCW fosters close interaction between reading comprehension and language production, helping learners reconcile their typically weaker productive skills with stronger receptive abilities, a process termed the “leveling effect.” It also addresses the challenge of applying static linguistic knowledge to dynamic communicative contexts by providing linguistic and contextual support through the source text, enabling learners to produce highly contextualized language output. Empirical research has demonstrated that intensive engagement with the source text correlates with better alignment and improved writing quality. Studies further suggest that XBCW enhances various linguistic competencies, including vocabulary, phraseology, syntax, and discourse, supporting sustained L2 development. However, the cognitive mechanisms underlying these positive outcomes remain underexplored.

This chapter examined the role of source-text processing in XBCW, emphasizing its dual nature as a receptive and productive task that requires initial comprehension of the source followed by selective integration of its content in writing. Previous research on reading-to-write tasks illustrated how source-text interaction influences writing outcomes, although most focused on traditional tasks such as summary writing or argumentative synthesis rather than continuation writing. Recent studies in the Chinese EFL context revealed that frequent rereading supports coherence and alignment by fostering an interplay between reading and writing. Investigations combining think-aloud data with interviews showed that learners dynamically modulate their attention to meaning, form, and discourse structure across different writing phases. Eye-tracking research further linked greater visual attention to source-text features with stronger linguistic alignment and improved writing performance. Despite these advances, research predominantly relied on introspective methods with limited temporal precision and scarce use of eye-tracking technology. Moreover, existing eye-tracking studies mainly explored attention to n-gram patterns, leaving holistic source-text processing largely unexamined. To overcome these gaps, the current study utilizes eye-tracking to provide real-time, fine-grained observations of how L2 learners allocate attention to the source text during continuation writing, aiming to elucidate the cognitive mechanisms involved.

This chapter introduced eye-tracking as the primary methodological tool for investigating source-text processing in XBCW. Eye-tracking is praised for its non-invasive nature and capacity to provide insights into internal cognitive processes during language tasks, often described metaphorically as a “window into language and cognition.” Its advantages include allowing naturalistic reading conditions with full-text presentation, high temporal resolution for capturing moment-to-moment processing variations, and objective, continuous data collection without disrupting the reading or writing processes. These strengths make eye-tracking a robust and promising method for studying real-time language comprehension and production. The chapter outlined key eye-tracking metrics used to infer cognitive processing stages. Early measures such as first fixation duration (FFD) and first pass duration (FPD) capture initial, shallow processing like lexical access and word recognition during the reader’s first encounter with a word or area of interest. Late measures including total fixation duration (TFD) and go-past time (GPT) index deeper, post-lexical processes such as semantic integration, reanalysis, and ambiguity resolution, reflecting readers’ higher-order comprehension efforts. These measures together offer a comprehensive framework for analyzing dynamic eye movement patterns and mental processes during source-text reading and continuation writing tasks.

3. The Current Study

This chapter introduced an eye-tracking study designed to explore how Xu-Argument-Based Continuation Writing (XBCW) affects learners’ processing of source texts, examining eye-movement patterns during reading (pre-writing) and writing phases. Participants were assigned to continuation writing, summary writing, or reading comprehension tasks, matched on English proficiency. Seventy-five Chinese university English majors participated, with no significant differences in proficiency among groups.

The source text, a 299-word adapted narrative called “Face to Face with Polar Bears,” was selected for its ease of reading and included five pseudo-words replaced for experimental purposes, which did not impede comprehension. Eye movements were tracked using the Tobii Pro Spectrum, a screen-based eye-tracker that allowed natural reading without physical restraints. Stimuli were presented on a 23.8" display, and eye gaze was recorded at 1,200 Hz, processed via the Tobii I-VT algorithm.

Each participant completed the experiment individually in a controlled environment. After calibration, they first read the source text silently at their own pace. Following this, only participants in the continuation writing and summary writing groups wrote by hand on paper, as a pre-study survey showed varied typing skills, making handwriting a more uniform mode of response. During writing, participants could revisit the on-screen text freely.

Eye-tracking data were carefully collected, excluding participants with less than 85% sampling rates during reading, resulting in a final sample of 66. Data from word-based areas of interest across the entire text were analyzed. During reading, four eye-tracking metrics—two early and two late stage indicators—were gathered to assess how task demands influenced reading. During writing, total duration and fixation counts on the source text revisits were recorded.

Statistical analysis involved preliminary tests for normality and variance homogeneity, revealing non-normal distributions despite homogeneity. Consequently, nonparametric tests were used: Kruskal-Wallis tests compared the three groups during reading, with Dunn-Bonferroni post hoc tests to identify specific differences, while Mann-Whitney U tests compared the two writing groups during writing. All analyses were conducted in R using relevant statistical packages, with mean ranks reported to clarify effect directions.

4. Results

This chapter described the eye-tracking results during the reading and writing phases of a study comparing continuation writing, summary writing, and reading comprehension tasks. During the reading phase, early eye-tracking measures showed no significant differences in first fixation duration among the groups. However, first pass duration was significantly longer for both continuation writing and summary writing groups compared to the reading comprehension group. Late eye-tracking measures revealed that total fixation duration and go-past time were significantly greater for continuation writing and summary writing groups than for the reading comprehension group. Furthermore, the continuation writing group exhibited longer total fixation duration and go-past time than the summary writing group.

During the writing phase, participants in the continuation writing and summary writing groups were allowed to revisit the source text while completing their tasks. Analysis of eye-tracking data indicated that continuation writers had significantly longer total fixation durations and a higher number of fixations on the source text than summary writers, suggesting greater visual engagement with the source during writing. These findings highlight differential processing patterns depending on the writing task, with continuation writing eliciting more intensive and prolonged visual attention to the source text both before and during writing.

5. Discussion

This chapter discussed how xu-argument-based continuation writing (XBCW) influences real-time reading processes among Chinese EFL learners by comparing continuation writing, summary writing, and reading comprehension groups via eye-tracking data. During the initial reading phase, first fixation duration showed no significant differences across groups, indicating similar basic lexical access and word recognition efficiency regardless of subsequent task demands. However, first pass duration was longer in continuation and summary writing groups compared to reading comprehension, reflecting deeper initial processing as learners activated broader semantic networks to prepare for generative writing tasks. This difference demonstrates that reading purpose modifies early processing, with writing tasks requiring more extensive consideration of text content beyond mere recognition.

Late eye-movement measures revealed that continuation writing elicited the highest cognitive demands, as participants needed to construct a flexible mental situation model and generate coherent, novel narratives. This deep engagement involved extensive inferencing, knowledge integration, and metalinguistic reflection to maintain linguistic consistency and stylistic alignment with the source text. The increased fixation durations and frequency indicated iterative processes of reading, internalizing, planning, and self-monitoring. Continuation writing amplified the functions of output hypothesis, including noticing linguistic gaps, hypothesis testing in expression, and metalinguistic reflection on form and cohesion. The results outlined a cognitive continuum, with reading comprehension as the least demanding, summary writing intermediate through content condensation and reformulation, and continuation writing as the most complex due to creative production and rigorous alignment requirements.

In the writing phase, eye-tracking data showed that participants in the continuation writing task revisited the source text more frequently and for longer durations than those in summary writing. This finding confirmed prior hypotheses that successful continuation requires constant checking of coherence between original and generated content, thus fostering intense interaction between comprehension and production. The increased revisiting reflects learners’ need to verify details, maintain semantic and stylistic consistency, and support complex content generation under L2 constraints. Overall, the study highlighted how task demands systematically shape reading and writing processes, with continuation writing imposing the highest cognitive load through integrated comprehension and generative production activities.

6. Conclusion

This chapter discussed how continuation writing significantly enhances learner-text interaction, as shown by increased total fixation duration and longer go-past time during reading, indicating greater cognitive engagement before writing begins. It introduced findings that during writing, learners frequently revisit and re-read the source text, demonstrating sustained interaction throughout the task. Eye-tracking data provided strong empirical support for the intensified relationship between L2 production and comprehension. The chapter also noted limitations, including the study’s focus on Chinese college students majoring in English, which may restrict generalizability. It acknowledged the influence of text type and complexity on eye-movement patterns and the absence of real-time writing process data due to technical constraints. Finally, it suggested future research directions to include diverse learner populations, real-time writing metrics, and investigation of the link between eye movements and written output quality.