Akademik Veri Yönetim Sistemi
17th International Conference on Corpus Linguistics, Madrid, İspanya, 10 - 12 Haziran 2026, ss.20-21, (Özet Bildiri)
Grounded in Hyland's (2005) interpersonal model of metadiscourse, this study employs a mixed methods design to investigate how interactional metadiscourse is deployed across three writer groups: L1 Turkish academics writing in Turkish, L1 English academics, and L2 English academics (Turkish scholars writing in English). The corpus comprises 739 research articles and 180 doctoral dissertations drawn from three disciplines (i.e., English Language Teaching, Linguistics, and Literature) supplemented by semi-structured interviews with 66 domain experts.
The study analyses five interactional metadiscourse categories (hedges, boosters, attitude markers, self-mentions, and engagement markers) across 18 sub-categories. Frequency values are standardised per 1,000 words to control for text-length variation. Group differences are assessed using Kruskal-Wallis and Mann-Whitney U tests, with effect sizes computed via Cohen's d.
Quantitative findings reveal that L1 Turkish academic writing is significantly lower in interactional metadiscourse density across all five categories compared to both English groups. The sharpest contrast appears in engagement markers, indicating that Turkish academic discourse structurally marginalises reader-involvement devices, such as directives and shared-knowledge appeals. Self-mentions show a near fivefold gap between L1 English and L1 Turkish, consistent with Turkish academic writing traditions that favour passive constructions and authorial effacement. Notably, sub-category analysis reveals two exceptions: L1 Turkish writers surpass both English groups in amplifier adverbs and possessive adjectives, pointing to language-specific morphological influences on metadiscourse expression.
Comparison between L1 and L2 English groups reveals partial convergence: L2 writers reach L1 norms for boosters and attitude markers, yet remain behind on hedges, self-mentions, and engagement markers, suggesting that certain pragmatic competencies are late-acquired features of academic discourse socialisation.
Qualitative interview data contextualise these corpus findings and reveal several cross-cultural patterns. Both groups report a near-universal absence of formal metadiscourse instruction, having acquired these competencies primarily through extensive reading. Both also share a preference for balanced metadiscourse use, viewing booster overuse as a sign of inexperience and hedging as an indicator of epistemic maturity. Significant differences emerge regarding first-person use: L1 English academics frame it as an epistemological norm, while L2 Turkish academics report institutional pressures (e.g., manuscript rejections from Turkish journals that discourage authorial visibility). A recurring theme among L2 writers is an internalised self-censorship mechanism, shaped by Turkish academic culture and a French intellectual tradition that prizes elaboration over reader engagement. Uniquely among L2 participants, authorial voice is increasingly positioned as evidence of human labour and authenticity in an era of AI-generated content.
These findings carry implications for contrastive rhetoric and EAP pedagogy. The observed differences cannot be reduced to linguistic transfer alone; they reflect culturally and institutionally shaped discourse identities. The study argues that academic writing instruction in both Turkish and English must move beyond generic form-focused approaches to address discipline-specific metadiscourse strategies, the risks of over- and under-assertion, and the structural conditions that constrain authorial voice in non-Anglophone academic contexts.