Evaluating a corpus-based intervention in medical English writing: a mixed-methods study

Abstract

Background

Recent work in medical education has begun to explore how language teaching may contribute to the humanistic aims of health professions training. While technology-assisted language learning (TALL) offers one possible pedagogical space, and corpus-based approaches provide structured access to authentic language use, less attention has been given to how corpus-based TALL may support students’ written representation of humanistic and professional values in clinical and ethical contexts.

Methods

Grounded in experiential and constructivist views of learning, this quasi-experimental mixed‑methods study explored on a corpus-based intervention implemented in a technology-assisted language learning environment at a medical university in Southeast China. Sixty medical students were assigned to experimental and control groups. Quantitative analyses compared pre‑ and post‑test writing performances through rubric scores and frequency tracking of humanistic expression within and between groups. Qualitative content analysis of writings examined changes in the discursive realization of clinical, ethical, and humanistic concerns. Semi‑structured interviews with experimental group students explored students’ reflections on how corpus-based learning influenced their written articulation of ethical responsibility, sensitivity to patients’ perspectives, and discursive engagement with clinical and ethical issues.

Results

After controlling for pre-test scores, the experimental group achieved significantly higher post-test writing scores than the control group (F(1,57) = 16.72, p < 0.001, η² = 0.23; adjusted means: 79.21 vs. 66.38). Frequency tracking further showed significant gains in clarity, accuracy, appropriateness, and use of disciplinary language in the experimental group, with large effect sizes for the first three dimensions (all p < 0.001); improvement in the integration of ethical concepts was smaller and not statistically significant. Qualitative analysis nevertheless indicated more explicit written representation of ethical considerations in clinical scenarios, and interview data suggested increased awareness of how professional values and ethical concerns are conventionally framed in medical discourse. The control group, by contrast, showed limited change in both overall writing performance and the written representation of ethical and professional concerns.

Conclusions

Findings provide empirical support for embedding corpus consultation within technology-enhanced, task-oriented medical English instruction. The intervention appears to have been most effective in strengthening linguistic precision, professional register, and the explicit framing of ethical issues in writing. It therefore offers a modest methodological contribution by showing how the written representation of humanistic values can be examined with identifiable discursive patterns. It also contributes to ongoing discussions on integrating technological and linguistic dimensions within bio‑psycho‑social approaches to medical education.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12909-026-09745-z.

Background

Humanistic education, as informed by the work of Abraham Maslow and Carl Rogers, is generally understood as a student-centered approach that values holistic development across intellectual, emotional, and personal domains, while emphasizing intrinsic motivation, dignity, and human flourishing [1, 2]. Recent literature has increasingly emphasized the importance of humanistic education, including literature, philosophy, and the arts [3, 4], in medicine, particularly for fostering empathy, cultural sensitivity, and ethical awareness as part of medical students’ professional formation [5, 6]. This emphasis has become more visible as healthcare education has increasingly engaged with bio-psycho-social understandings of care [7], prompting growing interest in pedagogies that can support the expression of humanistic values within professional communication [8]. In China, where reflective and patient-oriented discourse practices have not always been strongly foregrounded in medical training [9], recent educational and health reforms have called for closer integration of humanistic values into medical education [6, 10–12].

Within this context, language education deserves closer attention. Empathy, narrative competence, and global perspectives are increasingly discussed as important aims of medical education [13–15], and language learning may offer one practical site in which these aims can be developed through engagement with clinical and social issues [16]. Writing is especially important in this context, because it makes visible how professional identity, ethical positioning, and value orientation are constructed through language use [17, 18]. Medical English curricula that engage with medical humanities may therefore help students examine how patient-centered and ethically attentive positions are represented in discourse. At present, however, medical English instruction often remains oriented toward technical terminology and examination performance [19]. This leaves continuing practical and pedagogical questions about how discourse-focused learning, humanistic reflection, and sustained instructional support can be brought together in a systematic way [9, 19].

Against this backdrop, educational technologies are increasingly discussed in relation to accessibility, participation, and instructional innovation in medical English teaching [20, 21]. In particular, technology-assisted language learning (TALL), which uses digitally mediated and context-responsive tools to support learning in clinically meaningful settings, has been associated with greater participation, immediacy, and applied understanding [22–24]. Corpus-based and data-driven approaches have also attracted attention for their potential to help learners notice authentic language patterns and rhetorical positioning in disciplinary discourse [25, 26]. In this study, corpus-based approaches refer to the pedagogical use of a structured collection of authentic, discipline-relevant texts to examine recurring lexical and discourse features [27–30], thereby supporting learners’ analytical reflection on collocations, stance markers, evaluative language, and genre conventions in ways that may strengthen their writing awareness and effectiveness [31].

Within medical English education, corpus-informed instruction can give students access to authentic disciplinary texts and support close attention to how patient-centered and ethically attentive perspectives are realized linguistically [28]. It may also help learners observe recurring evaluative and relational language and draw on these resources in their own writing [30]. Yet existing studies tend to focus either on technological efficacy in language learning [14, 32–39] or on standalone humanities modules within medical curricula [3, 4]. Empirical research rarely examines how corpus-informed language pedagogy may influence students’ written articulation of humanistic perspectives within medical contexts [25, 40–42]. Recent scholarship on educational technology has highlighted both the opportunities and the limitations associated with digitally assisted learning environments. While many studies focus on vocabulary acquisition, communication skills, or learner autonomy, less attention has been paid to how technological tools may support discourse-level awareness in professional writing. Concerns regarding over-reliance on automated systems and reduced critical engagement further underscore the importance of pedagogical designs that foreground analytical reflection [43–45]. In medical education specifically, corpus-based learning is often discussed as a useful way of supporting terminology acquisition through authentic and data-driven engagement with language, yet current evidence remains concentrated on vocabulary retention and offers only limited insight into how such pedagogy may shape students’ written articulation of humanistic understanding in response to healthcare concerns [46].

The literature suggests a need to examine whether corpus-based TALL can support not only linguistic performance in medical English writing, but also students’ written representation of humanistic and professional values in clinical and ethical contexts. The present study therefore addressed the following research questions:

• RQ 1: In what ways does a corpus-based TALL intervention influence medical students’ English writing performance?

• RQ2: How is the discursive realization of humanistic values reflected in students’ case-based medical English writing following participation in a corpus-based TALL environment, and how do students describe their experience of this writing process?

Methods

Ethics

This research received ethical approval from the Human Research Ethics Committee (HREC) at The Education University of Hong Kong (2021-2022-0393) for the project focusing on teachers’ assessment strategy in English as a foreign language teaching in China’s mainland universities. As part of the ongoing research series, this research highlights new technology in building upon and expanding the scope of research. The approval, valid from September 1, 2022, to September 1, 2023, explicitly covers investigations explored in the current study. Research activities including data collection and analysis, were conducted strictly within the validity period of the aforementioned ethical approval and align closely with the approved research themes. Ethical considerations approved for the original study remain pertinent and are adhered to in this expanded research context. In addition, permission to conduct interviews was granted through an official approval letter from the institution where the participants study. Prior to the interviews, each participant was provided with an informed consent form via email, which was thoroughly discussed to address any inquiries and ensure understanding. All participants provided their verbal consent to participate. To maintain participant anonymity and validate the use of direct quotes, member checks were conducted as recommended by Thomas [47].

Theoretical underpinning and the research design

A corpus-based intervention was developed within a technology-assisted language learning (TALL) environment to support medical English writing. The pedagogical design drew on Rogers’ experiential learning theory [48] and Bruner’s constructivist perspectives on learning [49]. Rogers foregrounds the use of authentic materials, learner reflection, and the teacher’s facilitative role, whereas Bruner emphasizes scaffolded discovery and the active construction of knowledge. Together, these perspectives provided the pedagogical rationale for the structure and implementation of the intervention. Guided by these principles, corpus-based data-driven learning was operationalized through the integration of corpus tools, such as AntConc, with institutional learning platforms. Corpus materials were uploaded to Chaoxing, an institutional learning management platform, to support task organization and assignment submission, while iwrite, an automated writing evaluation platform, was used for the pre- and post-test writing tasks.

To examine the effects of the intervention, the study adopted a mixed-methods, quasi-experimental design. Participants’ pre- and post-test writings were assessed using the Medical English Humanistic Writing Rubric, a theoretically informed analytic framework developed from dimensions of the Humanistic Professional Awareness Scale for Healthcare Students and Providers (HPAS-HSP) [50] and principles adapted from the REFLECT rubric [51]. The quantitative strand examined changes in writing performance and the use of humanistic language through rubric-based comparisons of pre- and post-test scores and frequency analysis of humanistic expressions in context [52]. The qualitative strand provided explanatory depth through content analysis of students’ written texts and thematic analysis of semi-structured interviews, offering insight into how participants engaged with, interpreted, and experienced the intervention.

Setting and characteristics of participants

Conducted from September to November, 2022 at a comprehensive medical university in Southeast China, our study involved 60 first-year undergraduate medical students, aged between 18 and 20 years (M = 18.9, SD = 0.6). They were enrolled in a Level A English class based on comparable TOEFL iBT-adapted entrance examination scores and were from multiple medical majors. Participation was voluntary with assured confidentiality.

Convenience sampling was adopted due to instructional access to intact cohorts, meaning classes were randomly assigned into two groups, each consisting of 30 medical students. The experimental group of students received the intervention. To ensure educational equity, the control group, initially receiving traditional instruction, was later provided several sessions of the learning materials and activities embedded in the intervention. Both classes followed the same institutional English syllabus and used the same textbook. Concrete demographic information could be found in Table 1.

Table 1.

Demographic characteristics of participants

Major	Experimental (n = 30)	Control (n = 30)	Background Information
Clinical Medicine	8	8	Freshmen, possessing some basic English writing skills; having started acquiring some medical knowledge through professional medical courses; having chances for beginning clinical practice; holding the need to integrate professional knowledge with English writing, important for their learning and future career
Pharmacology	7	6
General Medicine	5	6
Nursing	4	3
Optometry	4	4
Biomedical Engineering	2	3

Before the intervention, the writing section of the entrance examination was re-evaluated among the 60 students for equivalence in English writing proficiency at baseline for this research (Table 2). Inter-rater reliability was high (Cronbach’s α = 0.92). Independent samples t-tests indicated no significant differences between the experimental and control groups in overall writing scores (t(58) = − 0.83, p = 0.412, Cohen’s d = 0.15). In addition, a pre-intervention questionnaire was adopted in order to assess students’ self-reported writing experience and prior exposure to medical humanistic expression learning. The instrument demonstrated acceptable internal consistency (Cronbach’s α = 0.79). Independent samples t-tests revealed no statistically significant group differences in reported writing experience (t(58) = 0.61, p = 0.54) or baseline medical knowledge (t(58) = 0.48, p = 0.64).

Table 2.

Baseline English writing proficiency comparison

Group	n	Mean	SD	t	df
Experimental group	30	76.2	3.502	-0.83	58.12
Control group	30	75.8	3.598

Procedure: the corpus-based TALL intervention as the experiment

The eight-week intervention consisted of 16 sessions (two 45-minute lessons per week, 720 min in total) and was implemented through a five-stage data-driven learning approach [37]: training, analysis, integration, application, and outcome reflection. In line with Rogers’ framework, the intervention incorporated authentic medical texts, structured opportunities for learner reflection, and ongoing teacher feedback to facilitate learning throughout the instructional process. Authentic materials were drawn from a self-developed Medical Humanities Contextual Terminology Corpus (MHCTC) with 217,000 characters, 80 high-frequency English medical terms and multi-word expressions. Two textbooks, both titled Medical Humanities English (East China Normal University Press; Fudan University Press), and English-language medical articles from China Daily (2012–2022) formed the core contents. Informed by the Association for Medical Education in Europe (AMEE), the Association of American Medical Colleges (AAMC), and the General Medical Council (GMC), the MHCTC was categorized into three broad domains—Medical Humanities and Ethics, Doctor-Patient Relationships and Communication, and Medical Culture and Social Practice—each further subdivided into topics reflecting multifaceted medical communication. It was uploaded to the Chaoxing platform.

High-frequency terms, as defined by Nation (2001) [53], are those that occur frequently in a corpus and hold functional and communicative importance in real-world contexts. In this study, terms such as empathy, autonomy, beneficence, informed consent process, dignity, and shared decision-making were identified through frequency analysis with priority given to their relevance in medical humanities. Beyond term identification, the corpus was designed to support students’ contextualized use of disciplinary language in the discussion of medical and ethical concerns. The terms were cross-referenced with the Corpus of Global Web-Based English (GloWbE) [54] to validate their representativeness in authentic medical discourse. In practice, a random sample of 25 terms was selected for validation testing, focusing on their normalized frequency (M = 45.6 occurrences per million words, SD = 12.4) and distribution across clinical case reports, medical journals, and patient communication scenarios. They were further reviewed by three medical English specialists and two medical humanities experts from the same university for humanistic relevance and communicative value.

Bruner’s constructivist orientation informed the scaffolded structure of corpus-based data-driven learning activities. The first 2 weeks’ of training of using AntConc for frequency analysis involved students employing the MHCTC to generate frequency lists of modal verbs (e.g., might, must, should, could) paired with key medical terms and multi-word expressions related to humanistic care across genres like diagnostic statements, medical case reports, and patient communication texts. Week 3 tasks focused on exploring modal verbs, enabling students to examine how they convey degrees of certainty and uncertainty. The concordance function enabled students to explore words and phrases extracted from the MHCTC in a Key Word in Context (KWIC) format. For instance, in Week 5, students analyzed lexical bundles commonly found in medical case reports (e.g., it is recommended that, the patient was advised to, further examination revealed) to understand how such expressions enhanced cohesion, readability, and an authoritative tone in professional communication. In Week 7, students used concordance analysis to investigate phrases related to patient autonomy, such respect for autonomy, informed consent, and shared decision-making, further connecting linguistic patterns to ethical communication in medical practice (Fig. 1). Collocation analysis using MHCTC data provided context-specific word pairings in medical texts. In Week 4, students examined verbs commonly used in passive constructions (e.g., was administered, was observed, was concluded) and compared them to active constructions (e.g., the researchers observed, the study concluded), gaining insight into how voice and tone influence clarity and objectivity. In Week 8, students explored collocations related to confidentiality, such as maintain confidentiality and breach of confidentiality, using MHCTC case studies to explore the ethical and legal implications of these practices. In Week 6, the MHCTC was the primary resource for analyzing rhetorical moves in medical opinion articles. Students used concordance tools to examine structural patterns, including transitions, hedging devices, and stance markers (e.g., however, therefore, it is argued that), and the collocation function to identify persuasive word pairings (e.g., strong evidence supports, it is widely accepted), gaining insight into how language constructs arguments and conveys professional authority.

Fig. 1.

Chaoxing platform interface of writing assignment

Therefore, throughout the intervention, emphasis was placed on the articulation of humanistic values through disciplinary discourse features, including evaluative positioning, stance marking, and context-sensitive disciplinary language use [55–57].

Data collection and analysis

Writing empowers students to clearly express their thoughts, ideas, and learned knowledge. In this study, we define writing proficiency as the ability to effectively and ethically use domain-specific language in medical communication to articulate complex ideas, clinical knowledge, and professional responsibilities within appropriate ethical and interpersonal contexts [58–60]. This construct focused humanistic orientation on how values such as empathy, ethical responsibility, and sensitivity to others’ perspectives became visible in written communication [61].

As above mentioned, the written representation of humanistic and professional values in medical English writing was assessed using the Medical English Humanistic Writing Rubric developed with reference to the HPAS-HSP [50] and the REFLECT rubric [51], which informed four dimensions of assessment: clarity of communication, accuracy of information, audience appropriateness, and ethical considerations. Expert review was incorporated during rubric refinement. Three experienced medical English teachers reviewed the corpus-informed disciplinary terms and scoring descriptors, and two also took part in the re-evaluation of the entrance examination writing samples. Interrater agreement was high (Cohen’s κ = 0.87), with strong internal consistency (Cronbach’s α = 0.92). The rubric also demonstrated satisfactory content validity (CVI = 0.85) and low rater variability (SD = 0.12). Scores were assigned on a 100-point scale: 40 points for clarity of communication, 30 for accuracy of information, 20 for audience appropriateness, and 10 for ethical considerations.

Pre- and post-test writing tasks (Appendix B)—Handling a Medication Oversight and End-of-Life Care Decisions—were completed in class using iwrite (Fig. 2), incorporating active learner engagement strategies (Branch et al., 2001) to promote reflective writing that fosters attentiveness, critical thinking, and deeper understanding [61]. Under the assessment criteria by the Medical English Humanistic Writing Rubric, clarity of communication was evaluated through sections like Oversight Description (pre-test) and Patient’s Condition and Current Treatments (post-test), highlighting effective medical information delivery; accuracy of information assessed precise use of medical terminology in Immediate Actions (pre-test) and Life-Extending Treatments (post-test); appropriateness for audience measured tone and language adaptation for professional and lay audiences in Communicating with the Patient (pre-test) and Discussing the Patient’s Wishes (post-test), and ethical considerations were examined through Ethical Dilemmas sections, with attention to how autonomy, trust, and professional responsibility were discursively represented. Altogether, 120 pieces of writing were collection from both groups.

Fig. 2.

An experimental group student’s post-test writing shown on iwrite platform

Two members of the research team independently scored all writing samples using the rubric, and discrepancies were resolved through discussion. Post-test writing scores were compared between groups using one-way analysis of covariance (ANCOVA) in SPSS version 27.0, with group as the fixed factor, post-test score as the dependent variable, and pre-test score as the covariate to adjust for baseline differences in writing performance. Assumptions for parametric analysis were checked before the main analysis. The data did not show substantial deviation from normality, as indicated by non-significant Shapiro–Wilk tests (p > 0.05) and acceptable skewness and kurtosis values (− 0.90 to + 0.85) [62]. Homogeneity of variance was supported by Levene’s test (p > 0.05), linearity was verified through scatterplots, and no extreme outliers were identified beyond ± 3 standard deviations [63]. An a priori power analysis conducted in G*Power 3.1 [64] indicated that a sample of 60 participants would provide adequate power (> 0.80) to detect an effect of f = 0.28 at α = 0.05 [65].

To complement rubric-based scoring, frequency tracking analysis was performed on the pre- and post-test writings to identify changes in the representation of key humanistic discourse features in context [51, 52]. Drawing on predefined analytic categories, the analysis focused on clarity, accuracy, appropriateness, the use of disciplinary language, and the integration of ethical concepts. This procedure enabled a corpus-informed examination of how students’ writing developed in terms of linguistic precision, professional appropriateness, and value-oriented expression over time.

For the content analysis on pre- and post-test texts, two researchers independently coded the texts and jointly reviewed an initial subset to calibrate interpretations and refine category definitions. Intercoder reliability on coding framework was assessed using Cohen’s κ to determine whether it could be applied consistently beyond chance, with acceptable agreement achieved (Cohen’s κ = 0.785) [66–67]. Discrepancies were resolved through discussion until consensus was achieved. For example, statements such as “I understand how patients feel when…” or “It is important to acknowledge their emotional needs” were coded as empathetic positioning in discourse, whereas phrases such as “Doctors should always prioritize honesty…” or “Maintaining confidentiality is essential…” were coded as discursive expressions of ethical responsibility.

The three dimensions of the HPAS-HSP [50] informed the development of the semi-structured interview guide (Appendix C) for 30 students from the experimental group. Although the interview guide was prepared in English, the interviews were conducted in Chinese to facilitate fuller expression and a more natural communicative environment. Each interview lasted approximately 40 min and was audio-recorded with participants’ consent.

Interview questions were designed to elicit students’ reflections on how corpus-based learning shaped their written expression of humanistic and professional values across the three HPAS-HSP dimensions. Questions related to personal integrity and accountability explored how students articulated ethical responsibility in writing, including references to patient confidentiality and social justice. For example, participants were asked: “Can you discuss a case or scenario from your studies where ethical dilemmas were presented? How did the resources guide you in considering patient confidentiality and social justice in your decision-making?” Questions addressing sensitivity to others examined how students described attending to patients’ perspectives and emotional contexts in written communication. One representative question was: “Could you describe an example from your reading materials that highlighted effective empathetic communication between a healthcare provider and a patient? How has this example influenced your approach to patient interactions?” The analytic focus was on participants’ accounts of how such examples informed their written communication in professional contexts. Questions concerning students’ discursive engagement with clinical and ethical issues focused on how they organized, synthesized, and communicated complex medical and ethical concerns in discourse. For instance, participants were asked: “From your corpus-based learning, what insights have you gained about critical thinking in medical practice? Can you provide an instance where a text helped you understand a complex medical issue or improve your communication with peers?”

Interviews were transcribed verbatim and analyzed thematically following Braun and Clarke’s (2006) six-phase procedure [68]. Consistent with their emphasis on theoretical positioning, the analysis was conducted at the semantic level, focusing on participants’ explicit descriptions of their writing practices and learning experiences. The analysis adopted a hybrid deductive–inductive strategy: the three HPAS-HSP dimensions served as initial analytic domains for coding, while additional subthemes were generated inductively when participants’ accounts extended beyond or complicated these domains. The unit of coding was a meaningful segment of text, such as a phrase, sentence, or short passage conveying a distinct idea relevant to the research questions. To refine the coding framework, two researchers independently coded 20% of the transcripts and developed a codebook aligned with the three analytic dimensions, including operational definitions and illustrative examples. Segments reflecting more than one construct were double-coded where analytically justified, and codes were applied only when directly supported by statements.

Results

Descriptive statistics for pre- and post-test writing scores are presented in Table 3. The two groups did not differ significantly in pre-test scores, t(58) = − 0.32, p = 0.752. After controlling for pre-test scores, one-way ANCOVA showed a significant effect of group on post-test writing scores, F (1, 57) = 16.72, p < 0.001, η² = 0.23. Students in the experimental group achieved higher post-test scores than those in the control group. The adjusted post-test mean was higher for the experimental group (M_adj = 79.21) than for the control group (M_adj = 66.38).

Table 3.

Pre- and post-test writing scores for experimental and control groups

Group	N	Pre-test Mean	SD	Pre-test range	Post-test Mean	SD	Post-test range
Experimental group	30	65.30	9.25	48–82	79.49	8.14	62–94
Control group	30	64.81	10.13	45–84	66.10	9.54	50–84

Frequency tracking analysis was conducted to examine pre- to post-test changes in the representation of value-oriented discourse features in students’ writing (Table 4). In the experimental group, significant increases were observed in clarity, accuracy, and appropriateness of linguistic expression, all with large effect sizes (d = 1.182, 1.242, and 1.149, respectively; p < 0.001 for all). These findings indicate improved linguistic precision and a more professionally appropriate tone in students’ written responses. A moderate increase was also found in the use of disciplinary language (d = 0.641, p = 0.002), whereas the increase in the integration of ethical concepts was smaller and did not reach statistical significance (d = 0.392, p = 0.087). No significant pre- to post-test changes were found in the control group across the tracked categories.

Table 4.

Frequency analysis of key terms in pre- and post-test writings

Category	Group	Pre Mean	SD	Post Mean	SD	t (29)	p	Effect size (Cohen’s d)
Clarity	Experimental	3.271	1.342	5.869	1.457	6.452	< 0.001	1.182
	Control	3.289	1.335	3.167	1.361	-0.487	0.630	-0.088
Accuracy	Experimental	2.873	1.215	5.667	1.382	6.784	< 0.001	1.242
	Control	2.883	1.217	2.917	1.223	0.121	0.905	0.020
Appropriateness	Experimental	3.127	1.291	6.003	1.509	6.319	< 0.001	1.149
	Control	3.150	1.298	3.257	1.308	0.301	0.765	0.057
Use of disciplinary language	Experimental	2.774	1.183	4.396	1.327	3.491	0.002	0.641
	Control	2.780	1.186	2.807	1.191	0.095	0.926	0.015
Integration of ethical concepts	Experimental	2.403	1.154	3.467	1.216	2.123	0.087	0.392
	Control	2.407	1.142	2.323	1.147	-0.248	0.806	-0.045

Qualitative analysis of the pre- and post-test writings (Table 5) indicated more pronounced changes in clarity and linguistic precision than in the textual integration of ethical considerations or disciplinary language. In the experimental group, vague expressions such as “Doctor must [be] careful with patient” were reformulated into more precise statements, for example, “The doctor must handle patients with care.” Minor but meaningful refinements were also observed, including the revision of “Patient was given wrong diagnose” to “The patient received an incorrect diagnosis.” These shifts reflect greater precision in linguistic form and alignment with professional written conventions, consistent with the quantitative patterns reported in Table 4.

Table 5.

Qualitative coding of pre- and post-test writings

Category	Group	Pre-test example	Post-test example	Observed discursive change
Clarity	Experimental	“Doctor must [be] careful with patient.”	“The doctor must handle patients with care.”	More precise and structured expression of clinical responsibility
	Control	“Surgery is hard and risky.”	“Surgery [could be] very difficult and danger.”	Generalized or loosely formulated descriptions persisted
Accuracy	Experimental	“She make[s] the cardiac arrest to stop.”	“She performed CPR to resuscitate the patient.”	More precise medical expression and stronger grammatical control
	Control	“Patient was given wrong diagnose.”	“Wrong diagnosis was said to about the patient.”	Continued lexical and syntactic inaccuracies
Appropriateness	Experimental	“Doctor must fight with diseases.”	“Doctors must combat diseases effectively.”	Greater alignment with professional register
	Control	“Doctor need to be more nicer to patients.”	“Doctors needs to be nicer.”	Ongoing inconsistencies in tone and formality
Use of disciplinary language	Experimental	“The heart stop working.”	“The patient experienced cardiac arrest.”	Broader deployment of field-appropriate language
	Control	“Blood pressure [is] high.”	“Patient has high blood pressure now.”	Limited expansion of specialized lexical resources
Ethical considerations	Experimental	“Doctor always do[es] right thing by patient.”	“Doctors must consider ethical implications when treating patients.”	More explicit written representation of ethical dimensions
	Control	“It is doctor job to care.”	“Doctor [should] always care for patients.”	Minimal elaboration of ethical aspects in writing

Across categories, the experimental group demonstrated more consistent discursive realization of professional tone and greater precision in the use of medical and clinical expressions in the post-test writings. In contrast, the control group showed limited observable change. For instance, clarity remained uneven, ethical considerations were minimally elaborated, and adjustments in terminology and appropriateness were modest. In terms of clarity, experimental group responses showed more structured and explicit written representation of clinical actions and responsibilities, while the control group maintained generalized or loosely formulated descriptions. For accuracy, experimental group revisions reflected greater terminological precision and correct medical usage, whereas recurrent lexical and grammatical errors persisted in the control group texts. Regarding appropriateness, experimental group writings displayed more stable adoption of professional register and conventional medical phrasing, aligning with norms of clinical communication. Control group texts continued to exhibit inconsistencies in tone and formality. For the use of disciplinary language, experimental group responses incorporated a broader range of field-specific expressions, such as “cardiac arrest” replacing “heart stop working.” This pattern suggests expanded discursive resources for representing medical scenarios in writing. The control group showed minimal expansion in specialized vocabulary. With respect to ethical considerations, experimental group writings included more explicit written representation of ethical implications within clinical scenarios, for example, references to considering ethical implications during treatment decisions. Control group responses tended to restate general responsibilities without further elaboration.

Thematic analysis of the interview data, informed by Braun and Clarke’s approach [69–71] and a constructionist view of language [49, 70, 72], identified three interrelated patterns in students’ medical English writing within the TALL context (Table 6). The first theme, Ethical discourse exploration, used corpus tools to explore and apply language for discussing ethical issues in written tasks. Under the code “Corpus exploration,” students reported identifying unfamiliar ethical terms and examining their usage patterns in authentic medical texts. One student noted: “When I started to look at these texts with the corpus software, it was like a whole new world opened up to me.” (Student 5). Another stated: “Through analyzing collocations, I studied how certain ethical phrases were actually used in real medical contexts.” (Student 9). These accounts reflect engagement with lexical patterns and usage conventions relevant to the written representation of humanistic values. The code “Context application” captured how students incorporated ethically relevant language into their analyses of medical cases. As one student explained: “I could really see how the ethical terms could be used to describe the dilemmas.” (Student 12). Another commented: “Discussing the ethical dimensions of the texts with my classmates broadened my perspective.” (Student 17). These reflections indicate increased attention to the discursive realization of ethical considerations in written case analysis. Under “Reflective writing,” students described incorporating corpus-informed language resources into their own texts. One participant stated: “I found myself using those key terms and phrases in my case analyses.” (Student 20). This suggests greater precision in the articulation of ethical viewpoints in writing.

Table 6.

Thematic analysis of students’ learning experiences in the TALL environment

Theme	Codes	Illustrative Excerpts of sample responses	Analytical focus
Ethical discourse exploration	Corpus exploration	“By AntConc, I realized how many ethical terms, like ‘beneficence’, in materials, it was eye-opening.” “Analyzing collocations, I understand the use of terms in medical ethics contexts.”	Engagement with ethical language and usage patterns in medical texts
	Context application	“Identifying ethical terms in case materials helped me interpret dilemmas and propose solutions.” “We learned to apply some terms in real-world contexts, which are important in ethical decision-making. My understandings are deepened”	Application of ethical language resources in written case analysis
	Reflective writing	“Some key terms improved my ability to express ethical viewpoints.” “I and my writing became more thoughtful with ethical principles that we explored through corpus work.”	Greater precision in the articulation of ethical viewpoints
Professional values integration	Ethical language analysis	“Corpus analysis provided a richer understanding of medical ethics. I feel I can communicate complex ethical stances more effectively.”	Alignment of written discourse with professional principles
	Professional values integration	“I learned to see my future professional identity and understand the core medical values better.”	Discursive positioning within professional contexts
Enriched Learning Experience	Learning experience	“Corpus tools and case studies balanced language and ethical learning for real-world challenges.” “Discussions and teacher feedback improved my writing and understanding of ethics.”	Structured learning activities supporting written representation
	Instructional support	“Step-by-step guidance made corpus analysis easier and provided meaningful insights.” “Feedback helped improve my writing and my ability to discuss ethical topics thoughtfully.”	Instructional scaffolding for discourse production

The theme Professional values integration emerged in relation to students’ references to professional discourse and core medical values. Students reported that corpus-informed analysis supported their ability to formulate written statements aligned with professional principles. For example: “Corpus analysis provided a richer understanding of medical ethics. I feel I can communicate complex ethical stances more effectively.” (Student 15). Another student noted: “I learned to see my future professional identity and understand the core medical values better.” (Student 21). In this context, references to identity were treated analytically as discourse positioning within written texts rather than as evidence of attitudinal change.

The final theme, Enriched learning experience, reflects students’ perceptions of the structured instructional design within the TALL setting. Codes “Learning experience” and “Instructional support” highlighted the role of corpus tools, case-based tasks, peer discussion, and feedback. Students commented: “Corpus tools and case studies balanced language and ethical learning for real-world challenges.” (Student 19). “Step-by-step guidance made corpus analysis easier.” (Student 22). “Feedback helped improve my writing and my ability to discuss ethical topics thoughtfully.” (Student 25).

Discussion

Interpreted through Rogers’ experiential learning theory, our findings underscore the importance of learning experiences that are professionally meaningful, contextually relevant, and grounded in students’ active engagement with authentic problems. In the present intervention, case-based writing tasks and corpus consultation created such conditions by situating students in clinically relevant scenarios and encouraging them to explore how ethical and humanistic concerns are expressed in authentic disciplinary discourse. From this perspective, humanistic learning was experienced through purposeful engagement with meaningful communicative situations.

The findings also resonate with Bruner’s constructivist perspective on learning, which highlights meaning-making through guided inquiry, scaffolded discovery, and the progressive refinement of representation. In the present study, corpus tools, rubrics, and structured writing tasks served as pedagogical scaffolds that supported students in noticing discourse patterns, comparing rhetorical options, and refining their written choices through revision. From this perspective, the discursive realization of humanistic values can be understood as a constructivist process in which learners actively build professional meaning through interaction with texts, tools, and task contexts. Whereas Rogers’ perspective helps explain why ethically and professionally meaningful engagement matters, Bruner’s perspective clarifies how such engagement can be pedagogically organized and sustained.

Corpus-based TALL and medical English writing performance (RQ1)

Regarding the first research question, the findings suggest that the corpus-based TALL intervention was associated with measurable improvements in the clarity, accuracy, and appropriateness of students’ medical English writing. These gains are consistent with previous research showing that corpus-informed instruction can strengthen domain-specific writing and enhance learners’ awareness of disciplinary discourse [73–75]. However, our study further suggests that such improvement is linked to the pedagogical affordances created when corpus consultation is integrated with case-based analysis and guided writing. In this task-oriented medical English context, corpora functioned not as standalone linguistic resources, but as tools that supported students in making more contextually appropriate and professionally aligned writing choices [76, 77].

The unevenness of these gains provides a more differentiated understanding of the affordances and limitations of corpus-based TALL. Gains in specialized medical language and explicit ethical framing were more modest, suggesting that corpus tools may more readily support syntactic precision and conventional professional phrasing than the development of specialized medical lexis and ethically inflected expression [28, 37, 38]. One possible explanation is that these dimensions require repeated contextual use, interpretive support, and sustained opportunities for uptake [29, 30]. Ethical framing also involves context-sensitive judgments about stance, tone, and professional value positioning, which are closely related to the situated nature of English for medical purposes communication [31]. The findings therefore point to an important limitation of short-term intervention: although corpus tools can facilitate noticing and revision, they may not by themselves be sufficient for developing more conceptually and rhetorically complex dimensions of medical writing [39]. Such results are also consistent with implementation challenges reported in efforts to integrate interdisciplinary content into medical English education curricula [18, 78, 79].

The results also contribute to ongoing discussions about the pedagogical scope of TALL. Earlier studies have focused on general language gains in computer assisted environments [32, 37, 38, 80, 81]. By embedding corpus consultation within structured medical case tasks, the present study illustrates how such environment can be operationalized as TALL to address profession-specific writing demands in medical education. This application aligns with arguments for the normalization of technology in language pedagogy [32] and our study further demonstrates how corpus tools can function as routine components [20, 32, 81, 82] of medical education in TALL for strengthening the written representation of ethical and professional concerns in medical English writing. Its contribution appears most robust at the level of linguistic precision and professional register, while students’ fuller integration of specialized medical language and ethical framing may benefit from sustained curricular integration.

Corpus-based TALL and the discursive realization of humanistic values (RQ2)

Concerning RQ2, the findings suggest a meaningful development in students’ greater inclusion of ethically relevant expressions at the discourse level, moving from generalized statements of responsibility toward more specific and precise articulations of ethical viewpoints in clinical scenarios. We regard such noteworthy results because writing in medical English involves not only terminological accuracy, but also the ability to express disciplinary stance in ways that are contextually appropriate and professionally meaningful [76, 77]. In this sense, the present study extends the contribution of corpus-based TALL beyond lexical or formal gains alone, showing its potential to support the textual representation of ethical stance in profession-specific writing.

A likely explanation lies in the design of the intervention in integrating ethical expressions and related terminology into corpus exploration, collocation analysis, and guided case writing to generating observable language patterns in medical discourse rather than treating ethical content as an added topic, thus creating opportunities for students to see how humanistic values can be articulated through professional medical language. Our corpus-informed design is consistent with arguments that medical English instruction should be built around recurrent patterns of disciplinary language use and high-value vocabulary in authentic texts [75], and it also accords with evidence that corpus-assisted methods can promote deeper processing and more sustained lexical learning than lecture-based instruction alone [46]. This may help explain why the present findings appear more productive than the uneven outcomes often associated with attempts to integrate medical humanities into medical English teaching without sufficient linguistic scaffolding [18, 79].

The thematic analysis further re-framed corpus consultation from its traditional role as a lexical aid [20, 26, 27] into a mediating tool for translating ethical concerns into disciplinary discourse, echoing Bruner’s view of learning as a mediated and scaffolded meaning-making process, which helps explain how such support enabled students to render ethical concerns in language more explicitly and contextually. Such a role points to an emerging awareness among students of how ethical principles can be articulated through medical English writing beyond the level of general professional intention, contributing to a shift in attention from ethical development as an internal disposition to ethical expression as an observable textual practice.

While earlier scholarship in medical humanities education has often emphasized empathy or professional formation [9, 83], and related work has stressed the importance of humanistic quality cultivation and the integration of medical humanities into language education [3, 14, 18], the present findings suggest that humanistic values can also be traced in the rhetorical and lexical choices through which students frame clinical responsibility [31, 74].

Pedagogically, and more broadly, the study invites a view of ethical language not as a spontaneous by-product of professional maturation, but as a teachable form of disciplinary discourse featuring visibility and reusability under corpus-based TALL [24, 31]. In this sense, corpus-based TALL begins to address the gap between ethical instruction as curricular intent and its concrete realization in medical communication, particularly when corpus-informed activities are integrated into medical English pedagogy in structured and context-sensitive ways [46, 75].

Unexpected patterns and theoretical implications

Two unanticipated patterns merit closer theoretical attention. First, the experimental group’s post-test writing showed a level of contextual precision in expressing medical and ethical concerns that had not been fully anticipated. The change was reflected in the increased use of disciplinary and ethical lexis and in the more situated ways these expressions were incorporated into clinical scenarios. Terms such as autonomy and informed consent appeared in case-based statements concerned with explaining treatment options, communicating risks, and respecting patients’ decisions in specific clinical situations. This suggests that structured corpus consultation within a TALL environment may support context-sensitive meaning-making alongside linguistic refinement [84–86]. Second, the interview data indicated that students engaged with corpus exploration in highly individualized ways when examining how ethical principles are conventionally articulated in medical discourse. They described tracing collocations, comparing authentic usage patterns, and revising wording in response to corpus evidence. This observation adds to computer-assisted language learning research on learner agency and exploratory practice [87, 88] by showing how these processes can be carried forward into observable textual performance.

Placed in relation to previous studies, these findings reveal both continuity and a notable point of departure. They are consistent with research showing that interdisciplinary curriculum design in medical education can support the integration of professional and ethical dimensions into learning [13, 89]. They also speak to a longer-standing concern in the literature: the difficulty of embedding medical humanities within English instruction in ways that yield consistent pedagogical effects [9, 78]. Against this background, the present results are significant for the form that these effects took. The study did not only indicate increased attention to ethical themes. It also identified relatively stable patterns in the written articulation of ethical and clinical concerns. A plausible explanation is that corpus analysis was not treated as an isolated language-support technique, but was tied directly to case-based writing tasks. This appears to have created a more explicit pathway through which students could render professional values in concrete and context-sensitive disciplinary writing. Therefore, it could be established that instructional effects became visible because ethical content was organized as analyzable, writable, and assessable discourse.

This point can be taken further in relation to corpus-based language learning. Authentic language resources have often been understood as supporting learners’ engagement with culturally and professionally situated discourse practices [90, 91]. The present findings add a more specific dimension to this view. They suggest that corpus-informed tasks may also support the written enactment of humanistic values when ethical and professional language is treated as a discourse resource that students can examine, compare, and revise in response to disciplinary context. The emphasis, in this sense, falls on the textual processes through which values become communicatively meaningful in writing.

This also bears on a persistent issue in medical humanities education, where assessment has long been recognized as difficult [3, 4, 92–95]. The study makes a modest methodological contribution by operationalizing the written representation of humanistic values through corpus-informed rubrics and textual analysis. What this offers is a way of approaching discursive realization as something that can be traced in observable linguistic and rhetorical patterning. This does not resolve the broader difficulty of assessing humanistic development, but it does make one important dimension of that development more accessible to systematic inquiry.

A related implication concerns the design of technology-enhanced instruction. Research on technological pedagogical and content knowledge has consistently stressed the need for meaningful integration between digital tools and teaching practice [96, 97]. The present findings are in line with this position, while also indicating more concretely what such integration required in this context. Effective TALL design appeared to depend on close alignment among corpus tools, disciplinary case-based writing tasks, and explicit criteria for the textual expression of humanistic concerns. Ongoing professional development and contextual adaptation therefore remain important, since coherence across technological resources, task design, and educational aims is unlikely to be sustained without deliberate instructional support [78, 98].

Toward a technologically mediated bio–psycho–social framework in medical English education

Contemporary medical education continues to face the consequences of a predominantly technology-centered training paradigm, in which communicative, ethical, and humanistic dimensions of practice often receive less systematic attention than biomedical knowledge and technical competence [2, 7, 8, 99–101]. This tension is particularly visible in China and other English as a foreign language contexts, where the integration of medical humanities into disciplinary curricula remains uneven and where medical English courses are still frequently treated as instrumental language training rather than as sites of professional formation. Within such conditions, one persistent challenge is how to translate broad humanistic goals into observable, teachable, and assessable forms of learning. The present study addresses this challenge by showing that attention to the written representation and discursive realization of humanistic values can provide a concrete entry point for integrating bio–psycho–social perspectives into medical English education.

The bio–psycho–social model has long been proposed as a corrective to narrowly biomedical conceptions of healthcare [2]. The empirical evidence from this study suggests that this framework can be pedagogically operationalized when TALL is aligned with case-based disciplinary tasks, corpus resources, and explicit evaluative criteria. More specifically, the findings indicate that corpus-based TALL does not support humanistic learning simply by virtue of technological access. Its contribution lies in creating a mediated instructional space in which students can examine how ethical and professional concerns are linguistically formulated, compare alternative ways of expressing such concerns, and revise their own writing in light of disciplinary expectations [84–86]. In this sense, the study offers empirical support for a technologically mediated bio–psycho–social framework in which humanistic values are not treated as abstract ideals external to language learning, but as discursive resources that become visible and negotiable through writing.

This framework has direct pedagogical implications for corpus-based TALL in medical education. First, it suggests that corpus tools should be used not only to improve lexical accuracy or genre appropriateness, but also to help students notice how empathy, responsibility, respect, and patient-centered reasoning are expressed in authentic medical discourse. Second, writing tasks need to be anchored in clinically and ethically situated cases so that students engage with language as part of professional judgment rather than as decontextualized linguistic performance. Third, assessment should move beyond holistic impressions of “humanistic quality” and instead incorporate explicit criteria that capture how values are represented through textual and discursive choices. Finally, instruction should include guided comparison, reflection, and revision, so that students are supported in analyzing corpus evidence and transforming that evidence into contextually appropriate written expression.These principles position corpus-based TALL as a pedagogical design for making the language of medical humanity more teachable, visible, and assessable.

The framework also carries broader curricular and institutional implications. If medical English is to contribute meaningfully to medical humanities education, humanistic objectives cannot remain peripheral statements in curriculum documents [102–105]. They need to be embedded in tasks, assessment practices, and teacher preparation. From this perspective, the study points to the importance of curricular coherence across platform design, case selection, writing expectations, and evaluative standards. Institutions seeking to strengthen humanities integration may therefore need to support not only access to digital tools, but also the development of assessment frameworks and teacher expertise [96, 97] that enable ethical and professional language to be taught as part of disciplinary communication. In this sense, the implications are policy-relevant: they speak to how medical English may be repositioned within contemporary medical education as a legitimate site for the cultivation and assessment of professionally meaningful discourse practices.

Viewed in this way, medical English is not simply a supporting language subject, nor is technology merely an instructional supplement. Rather, language education becomes one of the spaces in which bio–psycho–social orientations are enacted through discourse, and digital mediation becomes meaningful when it helps learners examine, articulate, and revise the language of professional values [106, 107]. This reframing offers a more grounded account of how corpus-based TALL can contribute to contemporary medical education, in which experiential relevance, constructivist scaffolding, and disciplinary discourse practice converge under the technologically mediated bio–psycho–social framework proposed here.

Conclusion, limitations and future directions

This study examined how a corpus-based TALL intervention influenced medical students’ writing performance and the written representation of humanistic and professional values in medical English. The findings suggest that structured corpus-based pedagogy, when integrated with case-based writing tasks, can support more precise articulation and more contextually grounded humanistic expression in students’ written production. In addition to measurable gains in writing performance, the study highlights the value of examining how ethical and professional concerns are rendered linguistically visible and offers a workable analytical perspective for investigating how humanities-oriented aims may be expressed within medical English education. Within this framework, TALL environments can be understood as pedagogical spaces where humanistic values, once realized discursively, enters writing through discourse work and becomes partly assessable, a process we argue depends on instructional alignment instead of simple tool use. This perspective positions medical English writing as a site where disciplinary knowledge, ethical communication, and digital mediation intersect in contemporary medical education. A technologically mediated bio–psycho–social view of medical English education strengthens such perspective by providing instructionally structured conditions for humanistic and professional values to be discursively realized, made visible in writing and incorporated into pedagogical and evaluative practice.

Several limitations should be acknowledged. The eight-week duration and the single-institution context may constrain the transferability of the findings. Longer-term studies are needed to determine whether these patterns are sustained over time. The study focused primarily on written production and interview reflections. While these data provide insight into the articulation and written representation of humanistic values, they do not capture how such discourse practices operate in spoken interactions, clinical simulations, or authentic healthcare settings. Future research could incorporate multimodal data, including classroom dialogue, simulated patient communication, behavioral or attitudinal assessments, and assessment scripts, to examine how discursive patterns evolve across contexts. In addition, the participants were first-year medical students who had not yet entered clinical training. At this stage, opportunities for authentic professional communication remain limited. Longitudinal research following students into later phases of medical education would provide a clearer understanding of how early exposure to TALL-mediated corpus analysis shapes subsequent disciplinary writing and professional discourse practices. Future investigations could also explore cross-institutional and cross-cultural implementations to examine contextual variation in the discursive realization of humanistic expression. Comparative designs may clarify how curricular alignment, assessment frameworks, and technological infrastructure influence outcomes. Further attention to instructor preparation in technology-enhanced language pedagogy may also deepen understanding of how TALL environments can be sustainably integrated within medical curricula. This need may become more salient as AI-supported writing tools enter language education and begin to shape how students plan, draft, and revise disciplinary texts [43, 106, 107]. Future research could therefore examine how corpus-based TALL interacts with emerging AI-mediated writing support, with particular attention to students’ engagement, emotional experience, and developing AI literacy [45, 107]. In medical education, this question carries additional significance because the expanding use of artificial intelligence has also renewed attention to the place of medical humanities in professional formation [21].

Authors’ contributions

Y.P. and L.W. were responsible for conceptualizing this study and engaging in data collection. All authors (Y.P., Y.Z. and L.W.) were responsible for data analysis and writing the final manuscript. All authors have reviewed and approved the final manuscript.

Funding

Not applicable.

Data availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was reviewed and approved by the Human Research Ethics Committee (HREC) at The Education University of Hong Kong (2021-2022-0393) in view of the retrospective nature of the study and all the procedures being performed were part of the routine care. Written informed consent was obtained from all participants prior to data collection. Measures were taken to protect the security and confidentiality of the recorded data. Personal information was separated from the dataset through a coding system. Access to the data was restricted to the researchers. In reporting the findings, all identifying information was removed to ensure participant confidentiality.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.