Changes in the ways authors refer to themselves: a diachronic study of self-mention in English research articles

The study examines how authors present themselves (self-mention) in English research articles over time and across disciplines. Self-mention includes first-person pronouns (FPP), third-person nouns (TPN, e.g., “the author”), and abstract subjects (AS, e.g., “this paper”). Self-mention contributes to authorial identity, credibility, and reader engagement, yet diachronic changes have been underexplored, with prior work largely focusing only on first-person pronouns. The purpose is to investigate diachronic (1970–2019) and cross-disciplinary (soft sciences: Applied Linguistics, Sociology; hard sciences: Biology, Electrical Engineering) changes in overall self-mention and its three sub-categories in single-authored research articles. Research questions: (1) What are the diachronic changes in the use of self-mention and its three sub-categories in English research articles from 1970 to 2019? (2) What are the cross-disciplinary differences in the use of self-mention and its three sub-categories over the same period.

Prior research has explored: (a) discourse functions of self-mention (e.g., explaining procedures, stating purpose, arguing, hedging, signaling contribution), demonstrating its multifaceted rhetorical roles; (b) cross-language variation showing higher frequencies of first-person pronouns in English articles than in Spanish, Italian, Turkish, or Chinese; (c) L1 vs. L2 writer differences, with English L1 writers generally preferring first-person pronouns more than L2 writers; and (d) cross-disciplinary variation, often finding more self-mention in soft sciences than hard sciences, different section distributions, and varied roles of first-person pronouns. Limited diachronic studies exist and have largely focused on first-person pronouns; for example, Hyland and Jiang (2018) found decreasing I/we in Applied Linguistics but increases in Sociology, Biology, and Electrical Engineering (1965–2015), while other case studies show increasing self-mention over time for specific writers or in abstracts. The current study extends this literature by analyzing all three sub-categories (FPP, TPN, AS) diachronically across four disciplines with a larger corpus (1200 articles).

Design and corpus: A diachronic corpus study of 1200 single-authored research articles from 12 leading journals in four disciplines—soft sciences: Applied Linguistics (Applied Linguistics, TESOL Quarterly, Language Learning, Foreign Language Annals) and Sociology (Sociology of Religion, Social Problems, British Journal of Sociology); hard sciences: Biology (Journal of Radiation Research, Journal of Experimental Biology, Biology of Reproduction) and Electrical Engineering (IEEE Transactions on Automatic Control, IEEE Journal of Solid-State Circuits, IEEE Transactions on Information Theory). Articles span 1970–2019, divided into five decades. Sampling: each year, two research articles were randomly selected from each journal, yielding 1200 total. Only IMRDC sections were included; only single-authored articles were used. The corpus totals ~7.32 million tokens (soft sciences: 4,251,498; hard sciences: 3,068,572). Operationalization: Self-mention categorized per Liu (2011) into (1) first-person pronouns (I, me, my, mine, we, us, our, ours, myself), (2) third-person nouns referring to the author/research team (e.g., the author, the researcher, this laboratory, the research group), and (3) abstract subjects (e.g., this/the paper/study/article/thesis/research/essay). Additional expressions were integrated after preliminary manual analysis. Extraction and validation: Instances were retrieved with AntConc 3.4.3. All annotations were manually checked in context to ensure accuracy (e.g., removing non-referential “i” as a serial marker). Normalization and measures: For each article, normalized frequencies per 10,000 words were computed due to length variation: NF = (raw frequency / tokens) × 10,000. Percentages of sub-categories were computed as NF(sub-category)/NF(all self-mention). Statistical analysis: Mixed effects models in R 4.3.0 assessed diachronic trends. Writers were random effects (grouped by discipline), with fixed effects of discipline and year (time). Analyses proceeded in three stages: (1) overall trends across all disciplines; (2) separate models for soft vs. hard sciences, including decade-wise normalized frequencies; (3) separate models for each sub-category within soft vs. hard sciences, with decade-wise percentages. p-values assessed significance of diachronic differences.

Overall trends (all 1200 articles): - Self-mention increased over time (Time estimate ≈ 0.376; p=0.006). First-person pronouns increased modestly overall (Time ≈ 0.292; p=0.030). Third-person nouns decreased (Time ≈ −0.0175; p=0.006). Abstract subjects increased strongly (Time ≈ 0.101; p=2.22e-08). - Composition over decades (percent of self-mention): FPP declined slightly from 88.0% (1970s) to 85.8% (2010s); TPN from 1.4% to 1.0%; AS rose from 10.6% to 13.2%. Cross-disciplinary trends in self-mention: - Soft sciences: significant decline (Time ≈ −0.566; p=0.008). Decade-wise NF per 10k: 83.69 (1970s), 80.12 (1980s), 91.78 (1990s), 85.81 (2000s), 43.29 (2010s). - Hard sciences: significant increase (Time ≈ 1.043; p=4.24e-11). Decade-wise NF per 10k: 41.78, 45.48, 54.08, 72.57, 83.66. Sub-category trends by discipline (mixed effects results): - First-person pronouns (FPP): Soft sciences decreased (Time ≈ −0.436; p=0.046). Hard sciences increased (Time ≈ 1.019; p=4.13e-11). Percentages: SS 88.8%→77.8%; HS 85.9%→91.0% (1970s→2010s). - Third-person nouns (TPN): Soft sciences showed no significant change (Time ≈ −0.010; p=0.353); percentage rose from 1.1% to 2.4%. Hard sciences decreased (Time ≈ −0.025; p=0.001); percentage fell from 2.2% to 0.5%. - Abstract subjects (AS): Increased in both SS (Time ≈ 0.155; p=1.12e-06) and HS (Time ≈ 0.048; p=0.006). Percentages: SS 10.1%→19.8%; HS 11.9%→8.5%. Notable decade dynamics: - SS self-mention rose most from 1980s→1990s (+~15%) but fell sharply from 2000s→2010s (~−50%). - HS self-mention increased most from 1990s→2000s and peaked in the 2010s; FPP proportion in HS reached ~91% in the 2010s. Interpretive highlights: - Soft science authors increasingly prefer subtler self-representation (more AS and relatively more TPN by percentage) while reducing overall self-mention and FPP. - Hard science authors increasingly adopt direct self-representation via FPP, reducing reliance on TPN and AS by percentage.

The findings answer the research questions by showing clear diachronic and disciplinary divergences in self-mention practices. Overall, authors increasingly reference themselves, primarily through more FPP and AS and fewer TPN; however, this aggregate masks contrasting disciplinary trajectories. Soft sciences exhibit a decline in overall self-mention and FPP use, alongside rising proportions of AS (and modestly of TPN), suggesting a strategic move toward more impersonal yet still self-referential framing that balances authority and objectivity. In contrast, hard sciences increase both overall self-mention and FPP, while reducing TPN and the percentage of AS, indicating a shift toward more visible authorial presence and direct engagement. These patterns likely reflect evolving disciplinary norms and perceived communicative needs: soft science authors, sensitive to critiques of subjectivity and disciplinary conventions, appear to temper overt self-insertion; hard science authors, supported by robust empirical data and responding to calls for clearer authorial presence and credit, increasingly foreground themselves via FPP. The results refine prior assumptions that soft sciences uniformly use more self-mention than hard sciences by revealing recent reversals and nuanced shifts in sub-category choices, thereby updating understanding of authorial identity construction in academic discourse.

This diachronic, cross-disciplinary corpus study (1970–2019) shows that self-mention practices in English research articles have evolved markedly and differently across fields. Overall, self-mention, FPP, and AS increased, while TPN decreased. Disciplinarily, soft sciences reduced overall self-mention and FPP and increased reliance on AS (and relatively on TPN), presenting authors more subtly. Hard sciences increased overall self-mention and FPP while reducing TPN and the proportional use of AS, foregrounding more direct authorial presence. These findings offer an updated map of authorial identity strategies across disciplines and decades and have pedagogical implications for guiding novice writers in calibrated self-representation. Future work should integrate qualitative analyses of discourse functions and contextual factors and triangulate with insights from sociolinguistics and cognitive linguistics, as well as author interviews, to better understand motivations behind these shifts.

The study is primarily quantitative and descriptive, focusing on frequencies and percentages without in-depth qualitative analysis of discourse functions or contextual factors. It does not incorporate interviews or triangulate with sociolinguistic or cognitive perspectives. Only single-authored articles and specific sections (IMRDC) were included, potentially limiting generalizability to co-authored works or other sections (e.g., abstracts). The corpus covers four disciplines and 12 journals, which, though sizable, may not represent all fields or journals.