Meaning-oriented understanding of L2 academic writing development with task-type repetition

The study addresses how L2 academic writing develops from a meaning-oriented perspective under task-type repetition (TtR). Prior research has emphasized form-focused CAF (Complexity, Accuracy, Fluency) measures, leaving meaning-oriented development underexplored. Given evidence that single tasks may not trigger qualitative changes in learners’ L2 systems, the study focuses on repetition of task types. It leverages Systemic Functional Linguistics (SFL), particularly grammatical metaphor (GM) and nominalization, to examine meaning-making in academic writing discourse, which typically prefers noun-dense, abstract language and intra-clause reasoning. The study aims to fill two gaps: (1) provide a meaning-oriented view of L2 writing development under TtR, and (2) trace longitudinal developmental trajectories rather than cross-sectional pre/post snapshots. Two research questions guide the work: (1) What trajectory may cohesion-/nominalization-based meaning-making follow along TtR? (2) What noteworthy changes emerge in cohesion-/nominalization-based meaning-making throughout TtR?

Task repetition (TR) has shown facilitative effects in oral L1/L2 development, theorized via limited attentional resources: familiarity saves attentional capacity, improving performance. Writing, with permanence and less time pressure, may also benefit. Prior L2 writing TR studies largely report CAF gains (accuracy, fluency, lexical resources, quality), and more recent work examines TR’s resonance with variables (task structure, corrective feedback), but still mainly form-focused. A meaning-oriented approach is thus needed. SFL frames language as meaning-making across strata; GM involves incongruent realization of meaning (e.g., processes reconstrued as nouns). Academic writing’s density and abstraction rely heavily on nominalization, making it a strong lens on meaning-oriented development. The study adopts Liardét’s tripartite framework for nominalization: metaphorical control (extent of incongruent reconstrual), metaphorical enrichment (semantic alternatives and density), and logogenetic impact (textual/cohesion effects: anaphoric reconstrual, elaborated nominal groups, causal metaphorical nets, meaning accumulation). Coh-Metrix LSA indices (LSASS1, LSASSP, LSAGN) align with local and global semantic overlap and cohesion. The study aims to reveal fine-grained, individual developmental idiosyncrasies rather than generalizable group effects.

Design and context: A semester-long EFL academic writing course for 25 MTI students at a Chinese university. Weekly submission of an essay (200–500 words) to an online co-editable file; in-class oral feedback was provided on form and meaning, but there was no direct instruction on GM/nominalization. Tasks: Ten home assignments over ten weeks constituted task-type repetition (TtR): same task type with different content. Each prompt asked students to (1) describe a public sign with problematic Chinese–English translation, (2) provide expository/evaluative/analytic comments on the inappropriateness, and (3) conclude with a proposed translation, eliciting a consistent genre staging: DESCRIPTION – EXPOSITION – CONCLUSION (typically three paragraphs). Participants: Group-level analysis included all 25 students (Male=7, Female=18, Mean age=24). For ontogenetic trajectories, three focal students differed by proficiency: Lee (low; TEM-8 fail; ≈CEFR B1), Monica (intermediate; TEM-8 pass; B2), Jack (high; >80/100 TEM-8; C1). Instruments and data: - Coh-Metrix 3.0 LSA indices: LSASS1 (adjacent sentence similarity), LSASSP (all sentence pairs within a paragraph), LSAGN (given-to-new information per sentence), computed for weekly corpora (group) and each focal student. - Nominalization analysis: Two coders manually identified nominalization cases across the three genre stages using derivation and agnation criteria (Ravelli, 1988, 1999). Inter-coder reliability Cronbach’s alpha=0.817. Frequency counts considered ideationally distinct nominalizations; repeated instances of the same nominal form within a stage counted once. Qualitative analysis focused on typical cases at key timings and frequently repeated forms. - Interviews: 11 semi-structured follow-up interviews (≈15 min each) explored writers’ intentions and decision-making behind nominalization and cohesion choices. Dataset: 248 essays, 97,924 words total (Mean=394.9; SD=1.87; Min=207.9; Max=581.9). Analysis: LSA trajectories for group and individuals were compared diachronically to identify key timings with noticeable changes (start T1, intermediate around T4 and T8, and end T10). Nominalization frequencies and case analyses were aligned to these timings and genre stages. Ethical approval and informed consent were obtained; no direct GM instruction ensured naturalistic data.

• Developmental trajectory: Staged, not linear. Four key timings were used for interpretation: T1 (start), around T4, around T8, T10 (end). LSA indices rose and fell across sessions rather than accumulating monotonically. - Proficiency–TtR resonance in cohesion: • Early session (T1–~T4): Low-proficiency student (Lee) outperformed both the intermediate (Monica) and high (Jack) in inter-sentential cohesion and semantic overlap; LSASS1 increased to ~0.26 and LSASSP to ~0.31; LSAGN remained above class averages. Jack was lowest on LSASSP/LSAGN and below general level on LSASS1 until after T3. • Middle session (~T4–~T8): Intermediate student (Monica) generally led in inter-sentential cohesion and semantic overlap; LSASS1 peaks ~0.22 (T5) and ~0.24 (T7), while Lee and Jack fluctuated around or slightly below general-level scores. - Nominalization by genre stage: • DESCRIPTION: Nominalization was absent initially; it appeared at T4 (Jack) and T5 (Lee, Monica). Debut suggests growing genre awareness. Fossilization observed: Monica repeated a nominal relative construction “...the intention of which is to ...” from T5–T9; Lee used “translation” as a recurring nominalization. Interviews indicated choices driven by perceived academic style and attempts to enhance cohesion via relative clauses (syntactic complication mediating anaphora). Jack showed greater metaphorical enrichment with diverse nominalizations (e.g., abbreviation, significance, expression, inappropriateness), though sometimes misaligned with the descriptive stage, reflecting evolving genre knowledge. • EXPOSITION: More dynamic and higher nominalization frequencies; middle session (~T4–~T8) peaked. Jack maintained >10 distinct nominalizations across T4–T8 (13 at T8, highest); Monica and Lee fluctuated around 6 and 5 respectively (both peaking at T4: 9 and 7). Wider subgenre affordances mitigated fossilization seen in DESCRIPTION. Evidence of nuanced gains in metaphorical enrichment and control: e.g., Lee’s progression from “the understanding of …” (T4) to “comprehension” (T10); Jack’s shift from “mistranslation” (T1) to “inappropriateness” and “inaccuracy” (T4). Intermediate metaphorical control patterns were rare overall. • CONCLUSION: Jack used nominalization more frequently overall (e.g., 4 cases at T4; none at T2, T9), while lower-proficiency students often used none (Lee absent in 7/10 tasks). Frequencies for Jack and Monica rose after T2, peaking at T4, then declined, indicating non-linear change. Interviews and case analyses showed initial misuse of nominalization for analytic/argumentative purposes in conclusions (e.g., “alternative,” “relation,” “contrast,” “faithfulness,” “easiness”), later corrected by T10 to more conclusive, genre-appropriate nominalizations. - Cohesion and logogenetic impact: Attempts at anaphoric reconstrual were sometimes replaced by syntactic strategies (e.g., Monica’s relative clause), suggesting individual mediation of anaphora. Evidence of elaborated nominal groups and causal metaphorical nets appeared in some cases (e.g., “a polished alternative … can result from …”), offering enriched semantic density but occasionally misaligned with the conclusion genre. - Quantitative anchors: 248 essays; 97,924 words; inter-coder reliability Cronbach’s alpha=0.817; notable LSA values for Lee early (LSASS1 ~0.26; LSASSP ~0.31); peak nominalization counts in EXPOSITION (Jack 13 at T8; Monica 9 at T4; Lee 7 at T4).

The findings address the research questions by showing that under TtR, meaning-making development in L2 academic writing follows staged, non-linear trajectories that resonate with learners’ proficiency and evolving genre knowledge. Early advantages in local/global cohesion for lower-proficiency writers suggest that increased task familiarity can free attentional resources for cohesion. Mid-session advantages for intermediate writers indicate shifting resource allocation and consolidation effects. Nominalization patterns reveal that genre knowledge—both rhetorical purpose and formal conventions—mediates where and how learners deploy grammatical metaphor. TtR afforded iterative reflection, prompting learners to re-evaluate descriptive versus expository versus conclusive purposes and to attempt denser, more abstract construals of meaning. Misalignments (e.g., analytic nominalizations in conclusions) decreased over time, indicating growing genre control. The observed fossilization at DESCRIPTION highlights how repeated tasks can also channel attention toward syntactic routines, leading to formulaic nominalizations; yet richer subgenre demands in EXPOSITION diversified meaning-making and reduced fossilization. Individually mediated anaphora (via relative clauses) shows that learners may pursue cohesion through syntactic complexity rather than fully controlled GM, underscoring the need for instruction that links nominalization choices to cohesive functions and genre moves. Overall, the study supports TtR as a pedagogical affordance that can elicit reflective adjustments in meaning-making strategies aligned with academic discourse demands.

Using SFL-based nominalization (Liardét’s framework), Coh-Metrix LSA indices, and interviews, the study shows that L2 academic writing development under task-type repetition is staged and proficiency-sensitive. Key contributions: (1) empirical evidence of non-linear, staged trajectories in cohesion and nominalization use across TtR with identifiable key timings (T1, ~T4, ~T8, T10); (2) demonstration that evolving genre knowledge resonates with meaning-making decisions, explaining debut, absence, and correction of nominalization across DESCRIPTION, EXPOSITION, and CONCLUSION; (3) identification of fossilization at DESCRIPTION and individually mediated anaphoric strategies; (4) nuanced improvements in metaphorical enrichment and control. Pedagogically, instructors should time interventions to developmental stages and exploit TtR to foster reflection on genre conventions and meaning–form alignment. Future work should broaden meaning constructs beyond nominalization, refine mappings between Coh-Metrix indices and specific meaning-making activities, increase TtR frequency and participant diversity to test generalizability, and examine a wider array of genres and topics.

• Scope of meaning: The study operationalized a limited facet of meaning (nominalization/GM), leaving other meaning dimensions unexplored. - Coh-Metrix mapping: While LSA indices relate to cohesion/overlap, clearer pedagogical interpretations linking indices to specific meaning-making activities are needed. - Design constraints: Institutional scheduling limited TtR frequency and sample range; not reaching massed repetition may limit effects. Findings are case-specific rather than broadly generalizable. - Task-type constraints: Repeating the same task type with varying content may constrain ideational attempts; effects of nominalization may differ with more diverse genres/topics. - Attention–familiarity dynamics: Task familiarity may reallocate attention toward syntactic routines, potentially fostering fossilization; this dynamic warrants further theoretical and empirical exploration.