Coherence and Cohesion in L2 Academic Writing

It is well-known that L2AW is key in teacher education, requiring the ability to master linguistic accuracy and rhetorical conventions at an advanced level to articulate complex ideas, develop arguments, and engage critically with formal discourse. As an interdisciplinary field drawing on applied linguistics, rhetoric, composition, education, and anthropology, L2AW addresses how to teach, learn, and write in an additional language (Tardy, 2025), a process that presents unique challenges compared to writing in an L1. In Chile, both foreign and Chilean college students may indeed face particular challenges. The former need to communicate competently in Spanish (the country’s official language) as a vehicular language (Pastor Cesteros & Ferreira Cabrera, 2018). The latter are required by national standards to achieve C1 level in English (according to the Common European Framework of Reference for Languages [CEFR]) before graduation. However, their texts in the foreign language often lack fluency and effectiveness (Silva, 1993) since they struggle with grammatical precision, lexical variety, and coherent organization, which contributes to the perception of L2AW as an onerous endeavor (Lin & Morrison, 2021; Pineteh, 2013).

Such difficulties seem greater for PETs, who must also demonstrate textual organization competence, genre awareness, and rhetorical precision (Lloyd, 2007; Sparks et al., 2014) as well as skills in synthesizing sources, maintaining coherence, using evidence persuasively, and, as in the case of our participants, mastering citation conventions when they work on their undergraduate thesis and other academic tasks. In fact, L2AW, rather than sentence-level accuracy, is more about critical thinking, analysis, and information synthesis (Castillo-Martínez & Ramírez-Montoya, 2021). It requires developing a specific style marked by a formal tone, discipline-specific terminology, and adherence to genre conventions (Septiwan & Hafizh, 2021). Since the development of coherence and cohesion in L2AW is closely linked to learners’ linguistic competence and their awareness of discourse structure, let us examine both in more detail.

The role of coherence and cohesion as prime textual features is acknowledged in theoretical and empirical work (Hyland, 2006; Lee, 2002; McNamara et al., 2010). Presenting ideas in a clear and logical manner is key to helping readers understand the writer’s train of thought, which is central to effective writing (Brown, 2022; Chen & Cui, 2022). Coherence involves macro-organization of ideas relevant to the topic with clear development (Lee, 2002; Reinhart, 1980). Meanwhile, cohesion requires the use of lexico-grammatical devices—such as conjunctions, reference, substitution, ellipsis, and lexical connections, including pronouns—that link ideas across clauses and sentences to glue paragraphs together (Halliday & Hasan, 1976; McCarthy, 1991; Samuels, 2024; Saputra & Hakim, 2020). Empirical research suggests that these devices—particularly reference, conjunction, and lexical cohesion—may predict fluency and writing quality (Tian et al., 2021). Indeed, discourse connectors provide the logical scaffolding for constructing persuasive arguments (Hyland, 2005), with more coherent L2 texts using more varied and appropriate connectors (Schiftner-Tengg, 2022). Thus, mastering the use of connectors compensates for limitations in vocabulary and syntax, enabling writers to communicate complex relationships effectively. Without such tools, writing seems disjointed, hindering comprehension and rhetorical impact (Halliday & Hasan, 1976).

Interestingly, texts that display cohesion might still not achieve coherence (Lee, 2002; Nilopa et al., 2017), as many learners overuse or misuse cohesive markers, resulting in technically cohesive texts that lack logical flow. McNamara et al. (2010) argue that cohesion is not merely about surface-level textual links but about underlying semantic and conceptual connections supporting overall comprehension. As Crossley et al. (2016) state, both local (e.g., within-sentence) and global (e.g., paragraph-level) cohesion contribute to text coherence. Their computational analyses of learner corpora evidenced that more proficient writers were more purposeful and strategic in their use of cohesive devices. For Mallia (2017), achieving coherence requires an understanding of rhetorical structure and awareness of audience and purpose, while Hyland (2011) posits that students learn what counts as good writing through an understanding of their discipline and the conventions and genres regarded as representative.

Hence, explicit instruction that targets both coherence and cohesion (Al Shamalat & Ghani, 2020; Bahaziq, 2016; Bui, 2022; Grabe & Kaplan, 1996; Kadiri et al., 2016) is important to align L2 learners’ writing with disciplinary expectations that will enable them to effectively engage with academic communities. Indeed, explicit instruction is key to improving students’ writing skills, enabling them to produce more organized, logically structured texts (Nilopa et al., 2017). Riazi’s (1997) study on Persian ESL learners found that their use of lexical cohesion and reference increased as their proficiency improved. Conversely, overuse of certain cohesive devices, particularly conjunctions, among L2 students led to mechanical or unnatural-sounding texts, as Liu and Braine (2005) found when comparing texts written in English by Chinese students and native speakers. This suggests that, while cohesion may be taught explicitly, coherence seems to result from cognitive processes and rhetorical awareness (Connor & Johns, 1990; Hyland, 2003) that can be targeted through reflection.

Consequently, teaching coherence and cohesion poses significant challenges, particularly in ESL/EFL contexts, since students may lack the language resources or rhetorical knowledge to organize ideas logically and fluently (Hinkel, 2001). Instructional strategies aimed at improving coherence and cohesion typically include the use of model texts, rhetorical moves, and explicit instruction in cohesive devices; however, the literature points to approaches that integrate form-focused instruction with higher-order skills such as idea development and discourse organization. Wette (2017) stresses logical progression and thematic development in writing tasks to enhance both discourse features. Knoch et al. (2015) allude to the importance of raising students’ awareness of textual organization through targeted instruction on genre-based writing, metacognitive strategies, and feedback mechanisms. They argue that L2 learners tend to make more informed linguistic and structural choices when exposed to scaffolded instruction, aiding in their understanding of coherence and cohesion. From the teacher’s perspective, writing instruction may overemphasize cohesive devices while neglecting idea development and discourse structure, thereby hindering genuine coherence (Ferris & Hedgecock, 2005; Silva, 1990). It seems sensible to teach coherence and cohesion as integral components of academic literacy in authentic writing tasks scaffolded by strategies that foster metacognitive awareness and reduce anxiety.

Peer Feedforward and Metacognition

Studies seeking to support learning during the writing process have largely focused on written corrective feedback (Bitchener & Ferris, 2012; Ferreira, 2017, 2022; Sheen, 2011), mostly by teachers. Fewer studies deal with prospective formative approaches (Contreras & Zúñiga, 2017), peer feedback, or agency (Farini & Scollan, 2023) through collaborative writing. Peer feedback, or peer assessment, is a well-known collaborative approach in which learners judge each other’s work using given criteria, encouraging dynamic idea exchange and evaluative comments. This structured arrangement enables learners to assess and elaborate on their peers’ work and engage in discussions that enhance understanding and skill development (Topping, 2017). In L2AW, peer feedback’s unique contribution lies in promoting collaborative learning and audience awareness while reducing the perceived authority gap of teacher feedback (Nelson & Carson, 1998; Ruegg, 2015). Zaccaron and Xhafaj (2020) observed that students’ uptake of teacher feedback (which the participants assumed was peer feedback) was “quite similar to the ones who received peer-feedback” [sic] (p. 50). This prompted the authors to disguise the teachers’ comments on lexis and grammar as peers’ comments (Zaccaron & Xhafaj, 2024), confirming their hypothesis that “the social representations of teachers and peers seem to bias their feedback processing” (p. 59)—authority and trust do play a role. When learners act as assessors, peer feedback enhances writing self-efficacy and promotes self-regulatory behaviors (Lee & Evans, 2019). Farini and Scollan (2023) posit that facilitating agency—the synergy between one’s action and participation in social interactions—can help legitimate learners as “authors of knowledge” (p. 13). This is evident in Gambhir and Tangkiengsirisin’s (2017) study on peer feedback and peer feedforward towards EFL argumentative writing, where comments in the form of praise, advice, and critique—including acknowledgment of progress for improvement and advice towards future tasks (a form of peer feedforward)—enabled learner uptake in a subsequent assignment, resulting in improved learner goal setting and positive perception of both strategies, indicating that receiving and giving both peer feedback and peer feedforward helped improve argumentative essay writing.

Similarly, Ubilla Rosales and Gómez Álvarez (2015) and Ubilla Rosales et al. (2017, 2020) applied cognitive and sociocultural principles of second language acquisition to design a learner-centered blended course for collaborative writing to improve Chilean PETs’ argumentative essays, including peer-review cycles (as we understand feedforward now, though not coined at that time) while performing drafter, reviewer, and editor roles in triads. They found improvements in the L2AW process and text quality, a significant correlation between participants’ perceptions of collaboration and written performance self-assessments, and evidence of the importance of using online writing tools in explicit instruction of writing as a process. Álvarez and Difabio de Anglat (2017) noted that peer feedback enhanced reflective practice and improved writing at different educational levels.

With increased interest in strategies aiming at engaging students actively in the writing process by means of formative assessment, collaboration, and reflection, peer feedforward emerges as an alternative encompassing them all. In a systematic review on ESL/EFL learner gains from online peer feedback (OPF) in English writing, Cao et al. (2022) found that both writing cohesion and coherence improved in terms of local error correction and global text revisions. They also found that students perceived that OPF improved the flow, organization, and transitions of their essays, and helped them focus on local aspects such as grammar, sentence structure, and vocabulary. More recently, empirical research points at peer feedback and peer feedforward as tools not just for error correction but for enhancing critical evaluation skills and deeper awareness of writing issues (Yang & Zhang, 2023), encouraging planning and reflection to enable learners’ self‑regulation in their revision process (Chen et al., 2023; Li & Hebert, 2024; Yang & Zhang, 2023), and developing feedback literacy to empower learners to assess and act on peer comments strategically (Weng et al., 2024).

When Carless (2007)) and Carless et al. (2006) introduced the concept of feedforward in education as learner-oriented assessment, they meant guidance offered before or during the writing process. Sadler (2010) contributed to this paradigm shift of learners actively developing evaluative judgment skills to become self-regulated learners with capabilities of assessing and improving their own work while also engaging in purposeful peer assessment during the learning process. In contrast to feedback, peer feedforward is designed to inform future performance and help writers anticipate challenges for strategic decision-making while writing (Carless & Boud, 2018). Understood as a feedback cycle, feedback loop, or feedback spiral (Carless, 2020), what makes feedforward innovative is the timing of its delivery: during the writing process and with a focus on future action (Carless, 2006; Carless & Boud, 2018; Orsmond et al., 2013; Sadler, 2010). We may add the potential to reflect on the writing process before summative assessment (i.e., assessment as learning; Yu, 2024).

Although a comprehensive review of assessment paradigms is beyond the scope of this work, suffice it to stress that assessment of learning typically centers on summative outcomes and grading; assessment for learning emphasizes formative, process-oriented feedback aimed at improvement; and assessment as learning takes a step further by engaging learners in reflection as they set goals, monitor progress, and participate in self-assessment (Yu, 2024). By actively engaging in assessment for and as learning, PETs may improve their own L2AW as they develop the evaluative and reflective skills needed in their teaching practice. In the context of peer feedforward, assessment as learning is particularly relevant, as it supports learners’ development of self-regulation and metacognitive strategies (Lee & Evans, 2019), allowing students to receive and critically evaluate peers’ comments and apply them thoughtfully while developing their rhetorical understanding and L2AW competence (Yu & Lee, 2016). Research suggests that peer feedforward can bridge writing instruction and professional development by helping learners consider real audiences and clarify their rhetorical intentions (Nelson & Carson, 1998; Ruegg, 2015), which are essential for PETs. When implemented through peer interaction, feedforward may serve a dual function—scaffolding writing development while encouraging critical reflection and evaluative judgment—providing students with opportunities for active learning as they give and receive constructive comments and suggestions on writing-in-progress.

Peer feedforward seems most effective when embedded in a pedagogical model integrating metacognition (Flavell, 1976, 1979; Hart, 1965), that is, learners’ awareness and regulation of their own cognitive (and, in this case, writing) processes. Chen and Cui (2022) posit that dialogic peer engagement promotes metacognitive growth and writing awareness. As part of learning processes, O’Malley and Chamot (1990) identify three types of metacognitive strategies: planning (organizing actions to perform a task), monitoring (awareness of what one is doing during a task), and evaluating (judging what has been done). To these, Cross and Vandergrift (2018) add problem solving (identifying and solving comprehension problems). In writing, evidence suggests that metacognitive strategies enhance coherence by promoting intentional revisions and reflective writing decisions (De Mello et al., 2023; Rosdiana et al., 2023). Yu and Lee (2016) claim that deciding whether to incorporate peer suggestions into one’s own revisions is itself a metacognitive exercise. Indeed, it requires learners to assess the relevance, validity, and potential impact of feedback, mobilizing critical thinking, rhetorical judgment, and reflective practice. Mohamad and Tasir (2023) emphasize the importance of reflective questioning to help students assess whether their ideas are logically connected and clearly communicated, connecting metacognition to the act of providing feedback to others, as it requires learners to engage in reflection, compare their own writing with peer texts, and evaluate the effectiveness of various rhetorical choices.

In sum, the literature points to the advantages of blending peer feedforward and metacognitive instruction to support learners in developing cohesive and coherent texts, as well as the reflective skills needed for long-term academic and professional writing, and for effective L2AW instruction. Consequently, our research question is: How can peer feedforward—centered on coherence and cohesion—support metacognitive awareness and writing development among PETs?

Context and Participants

This study was conducted within the context of a required advanced L2AW course offered by the first author and taught by a trained teaching/research assistant, at a publicly-supported private Chilean university. Data were collected for three weeks on a course management platform (Canvas) in a blended class that included a four-hour-per-week face-to-face writing lab, during the seventh semester of a five-year TEFL program, whose final year is devoted to writing an undergraduate thesis and conducting the professional teaching practicum for licensing. The course focused on developing students’ ability to write academic texts aligned with international standards, with emphasis on coherence, cohesion, and rhetorical structure.

Although 35 Chilean PETs from intact classes initially signed up for the study via an online Google Forms form, 30 (nine men and 21 women) completed the writing tasks within the allotted period. These participants were native speakers of Spanish with a B2 level of English (according to the CEFR), as measured by a home-based IELTS mock-examination at the end of semester six. They had previously completed coursework in English grammar, composition, and language pedagogy, and received instruction in reflective writing and peer feedforward as part of the course introduction. The participants were familiar with the online platform, which was used throughout the study for submitting reflections and engaging in peer feedforward on the discussion board. This context provided an authentic environment to explore how peer feedforward, as a pedagogical strategy, could support the development of L2AW skills among PETs.

Analysis

The written corpus for analysis was downloaded from the course management platform. Each writing cycle involved two tasks: (a) writing a 200-word reflection journal on personal writing development, and (b) providing peer feedforward on two classmates’ reflections. Feedforward was in the form of comments of praise, advice, acknowledgment of progress, suggestions for improvements, and clarification questions. Participants had several means on which to base their comments, including exemplary texts to set clear expectations, content regarding academic writing, coherence and cohesion, the teacher’s instructions they could use as a checklist, and the IELTS band descriptors (British Council, IDP: IELTS Australia, & Cambridge Assessment English, 2023) as a guide for their self- and peer-review process. The instructions explicitly stated that this process was not about opinions but about text quality appraisal. Once participants received peer comments, they individually revised their drafts for improvement before submitting them to the teacher for scoring. Since peer feedforward was not a common practice in previous courses and learner autonomy needs scaffolding, we thought involving the teacher as a final assessor would serve as a step towards agency in collaborative writing. The score was given strictly for adherence to task instructions and helping students transition from teacher-only feedback into student-based revision. The teacher did not participate in the discussion board to avoid influencing the participants’ ideas. Only when students were off task did the teacher add a metacognitive comment to the score to redirect their task compliance; no comments on coherence, cohesion, or language use were provided, since the focus was on peer feedforward. This recursive process was designed to foster metacognitive engagement and support collaborative improvement of the task at hand, not subsequent writing tasks.

We used RStudio for statistical analysis and Python to generate figures. Qualitative coding and interpretation were conducted manually. After thematic coding of both peer comments and reflection entries, we analyzed the data quantitatively to identify patterns of improvement in coherence and cohesion. Quantitative data resulted from the categorization of peer feedforward across weekly cycles based on the linguistic focus of each comment (i.e., coherence and cohesion, vocabulary and lexical resources, or grammatical accuracy), extracted from peer responses to reflection journals submitted to Canvas. The categories that emerged from qualitative data analysis were confirmed through repeated discussions among researchers to ensure consistency. This approach enabled a week-by-week tracking of shifts in attention and writing focus, capturing the emergence of metacognitive awareness. By triangulating peer feedforward, self-reflections, and textual revisions, we were able to map patterns of development over time, particularly in relation to coherence and cohesion. This integrative analysis revealed how peer feedforward supported both the identification of areas of improvement and the internalization of organizational strategies in L2AW.

The study was approved by the institutional research ethics committee. Participants signed a consent form, which included a general description of the study and an invitation to participate voluntarily and anonymously. Participants who did not consent to the study or did not complete the tasks were excluded.