This study investigates how Artificial Intelligence (AI) can enhance two mission-critical linguistic competencies

linguistic performance, and technological readiness. Findings reveal that intranet-based AI tools significantly

improve cadets’ accuracy in domain-specific terminology and increase confidence in delivering English

command phrases. AI-generated feedback reduces speaking anxiety, while adaptive modules provide targeted

practice aligned with operational tasks such as radar coordination and artillery command. However,

implementation remains constrained by teachers’ limited AI literacy, security restrictions, and insufficient

infrastructural support. To address these issues, the study proposes a three-layer model combining pedagogical

adaptation, secure technological mediation, and institutional capacity building. The results provide a

contextualized framework for integrating AI into ESP instruction in high-security military environments.

Keywords: Artificial Intelligence, English for Specific Purposes, Military English, Adaptive Learning, Defense

Education

coordination, and technological command readiness (Dudley-Evans & St John, 1998).

At the ADAFA in Vietnam, English training plays a vital role in preparing cadets for air defense operations.

However, instruction remains largely teacher-centered and focused on grammar translation, with limited

opportunities for communicative practice or exposure to authentic operational discourse (Xuan Mai & Thanh

Thao, 2022). Security restrictions also limit cadets’ access to online learning resources and real-time language

support (Xu, 2024). These structural constraints hinder the development of two mission-critical linguistic

competencies in military English: (1) technical vocabulary accuracy and (2) spoken command performance

under time pressure. These competencies were selected as the focus of the present study because they directly

influence operational clarity during radar coordination, command issuing, and joint-force communication.

AI-based systems have the potential to address these barriers by offering immediate corrective feedback,

This study addresses this gap by examining how AI can enhance military ESP instruction at ADAFA, with a

specific focus on improving cadets’ mastery of technical terminology and spoken command performance. It

explores instructors’ and cadets’ perceptions of AI-based learning, evaluates its effects on motivation and

linguistic accuracy, and considers the institutional conditions needed to support secure implementation. By

contextualizing AI integration within the discipline, hierarchy, and digital restrictions of a military academy, this

research contributes both theoretical insights and practical implications for AI-enhanced ESP education in

defense environments.

AI has become a transformative force in language education, thanks to its ability to process learner data at scale,

deliver adaptive feedback, and simulate communicative contexts with high accuracy (Edmett et al., 2023). AI-

centered pedagogy, where technology supports autonomy, reflection, and differentiated instruction.

Within English for Specific Purposes (ESP), scholars have documented the potential of AI in creating

domainspecific learning pathways. He, Zhang, and Huang (2025) argue that AI platforms generate customized

tasks targeting specialized terminology and communicative situations, allowing learners to engage with language

forms directly connected to their professional fields. Adaptive learning engines can detect varying proficiency

levels within the same class, delivering individualized tasks that prevent advanced learners from being held back

while supporting those who require remediation. Edmett et al. (2023) further note that automated writing

evaluation, speech analytics, and AI-assisted vocabulary trainers enable teachers to devote instructional time to

developing strategic communication skills rather than correcting mechanical errors. Such capabilities are

particularly relevant for ESP domains that demand high levels of precision, clarity, and discipline-specific

form of ESP due to its direct connection to mission execution, command coordination, and international defense

engagement. Studies on military ESP highlight that learners must master specialized lexical sets, such as

airdefense terminology, artillery commands, radar reporting structures, and perform spoken communication tasks

in real or simulated operational contexts, often under time pressure (Xuan Mai & Thanh Thao, 2022). These

linguistic demands go beyond general English or civilian ESP, as miscommunication may compromise

operational safety.

Despite the clear pedagogical relevance of AI for supporting accuracy and real-time performance, the

technological integration of AI in military ESP remains limited. In global defense systems, AI has been primarily

applied to tactical simulations, unmanned systems, and strategic decision-support algorithms (Rashid et al.,

2023). In contrast, applications focusing on linguistic competence, especially those aimed at improving spoken

hesitation in adopting AI tools that require data exchange with external servers.

cadets, such as anxiety when delivering spoken commands, difficulty recalling complex terminology, or lack of

exposure to authentic operational communication.

Thus, while the broader literature demonstrates AI’s potential to individualize instruction, enhance performance

analytics, and support specialized vocabulary learning, little is known about its pedagogical, technological, and

institutional integration within military ESP contexts. The present study extends existing research by focusing

on the ADAFA in Vietnam and examining how AI can address two mission-critical linguistic competencies: (1)

technical air-defense vocabulary accuracy, and (2) spoken command performance under time-pressured

conditions. By situating AI within the constraints of a secure military environment, this study offers a

contextualized understanding of how adaptive technologies can support operationally relevant language learning

while adhering to institutional discipline and security requirements.

where quantitative data alone may not fully reflect institutional barriers or pedagogical constraints. As Creswell

and Plano Clark (2018) argue, combining quantitative and qualitative strands enables researchers to triangulate

findings, enhance validity, and gain a more comprehensive understanding of complex educational phenomena.

The case study design was chosen because ADAFA represents a unique instructional setting characterized by

restricted internet access, strict confidentiality protocols, and a highly structured learning culture. These features

require closer contextual examination than survey-based or experimental designs typically allow. The integration

of AI technologies within such constraints necessitates analyzing both learner responses and institutional

dynamics, justifying a context-embedded case study approach.

Participants included 60 cadets and five ESP instructors from the Faculty of Foreign Languages at ADAFA. The

challenges and institutional constraints. Participation was voluntary for both groups, and anonymity was

To capture multiple dimensions of AI-assisted learning in the military context, the study employed two

complementary instruments: a structured questionnaire for cadets and a semi-structured interview guide for

instructors.

A 20-item Likert-scale questionnaire (1 = Strongly Disagree; 5 = Strongly Agree) was administered offline to

comply with the Academy’s security regulations. The questionnaire items were adapted from validated scales

used in previous AI-in-education research (Edmett et al., 2023; Mizumoto, 2023). The decision to use 20 items

was intentional: fewer than 20 would not adequately cover the constructs relevant to military ESP, while more

items would extend testing time and disrupt cadets’ tightly scheduled training duties. The items measured four

Qualitative data from interviews were transcribed and analyzed using Braun and Clarke’s (2023) six-phase

communicative tasks. Military English relies on short, formulaic utterances delivered under time pressure; thus,

the opportunity to rehearse commands with immediate, private feedback was highly valued by cadets. This

finding aligns with Mizumoto’s (2023) notion of AI as a facilitator of metacognitive regulation: cadets monitored

their pronunciation accuracy, tracked progress, and self-corrected before speaking publicly. The psychological

effect is noteworthy. In a hierarchical, error-sensitive environment, AI provided a low-pressure space that helped

learners gain confidence, a factor rarely highlighted in civilian ESP literature but central to military

communication training.

The strong engagement generated by AI-supported simulations demonstrates the importance of embedding

However, the findings also reveal institutional and technological constraints that shape AI adoption. Restricted

internet access, essential for military security, limits the use of commercial AI tools, mirroring concerns raised

by Rashid et al. (2023). Instructors’ apprehension about data handling shows that AI solutions must be

pedagogically effective while conforming to defense confidentiality standards. These requirements differentiate

military ESP from civilian contexts and highlight the need for local hosting, data sovereignty, and system

transparency.

Instructor readiness emerged as another mediating factor. Although instructors recognized the value of AI,

uneven levels of AI literacy led to inconsistent implementation. This confirms Edmett et al.’s (2023) observation

that teachers play a central role in AI-enhanced pedagogy; however, the military context amplifies this issue, as

institutional alignment, a key element of the study’s conceptual framework.

Overall, the findings reaffirm the pedagogical potential of AI while demonstrating that the unique demands of

military contexts mediate its effectiveness. AI clearly supports more precise vocabulary learning, more confident

spoken command performance, and more authentic engagement with operational tasks. Yet these benefits depend

on secure infrastructures, teacher readiness, and policy-level coordination to ensure that integration is both

pedagogically meaningful and institutionally feasible.

The findings offer several implications for military ESP instruction. Pedagogically, the improvements in

vocabulary mastery and spoken command performance suggest that AI should be deliberately integrated into

existing curricula rather than treated as optional support. Adaptive exercises can reinforce foundational

system developers is crucial to ensure that AI platforms meet both pedagogical and security standards.

Institutionally, the variation in instructor readiness underscores the need for targeted professional development.

Training should focus on interpreting AI analytics, integrating feedback into lesson design, and managing learner

interactions with AI. Additionally, clear institutional messaging is needed so cadets and instructors understand

AI as a supportive tool rather than a replacement for human authority. Establishing policies on acceptable use,

data protection, and instructional expectations will help ensure consistent and responsible implementation.

Together, these implications show that effective AI adoption in military ESP requires strategic pedagogical

Building on the patterns identified in the findings and the interpretive insights presented in the discussion, this

alignment to ensure both effectiveness and feasibility.

simulationbased practice. Accordingly, this layer positions AI as an integrated component of the curriculum

as essential mediators, consistent with Edmett et al. (2023), who interpret AI-generated data, guide students

through feedback, and contextualize machine recommendations within operational communication norms.

The second layer, technological security and infrastructure, emerges directly from the institutional and

technological constraints identified in both the findings and the discussion. Given that ADAFA operates in a

restricted digital ecosystem, the framework emphasizes the development of locally hosted, intranet-based AI

systems capable of delivering adaptive learning without relying on external cloud services. This layer integrates

military security requirements with pedagogical needs by outlining a secure architecture involving encrypted

data storage, role-based access permissions, and stable offline functionality. Rather than limiting innovation,

these constraints shape a technology ecosystem purpose-built for military education, one that protects sensitive

information while still enabling features that cadets and instructors found most beneficial, such as pronunciation

therefore, prioritizes structured professional development, focusing on interpreting AI analytics, designing

AIsupported lessons, and guiding learners’ interaction with automated feedback. It also calls for the Academy to

articulate explicit policies on acceptable use, data handling, and pedagogical expectations so that AI adoption

does not depend on individual instructor initiative. Clear communication regarding the supportive, not

evaluative, role of AI can help mitigate cultural reservations associated with hierarchical military learning

environments. Finally, periodic pilot testing and iterative revisions ensure that AI tools evolve in tandem with

curricular demands and institutional priorities.

In short, these three layers form a cohesive framework that aligns pedagogical practice, technological design,

and institutional governance. AI integration at ADAFA will be most effective when these layers operate in

concert: when adaptive learning is embedded within instruction, when secure infrastructures enable reliable use,

framework, therefore, provides a structured pathway for adopting AI in a manner that is pedagogically

regulated military environment.

The findings highlight three key outcomes. First, adaptive AI tools effectively reinforced discipline-specific

terminology, enabling cadets to review and internalize core lexical items essential for operational

communication. Second, AI-driven pronunciation and fluency feedback contributed to greater accuracy and

confidence in spoken command delivery, helping cadets rehearse mission-critical utterances in a low-pressure

environment before applying them in classroom drills. Third, simulation-based tasks increased engagement by

embedding language practice within realistic operational scenarios, thereby narrowing the longstanding gap

between classroom instruction and field communication.

Beyond reporting these outcomes, the study contributes to current scholarship by demonstrating how

pedagogical affordances associated with AI, such as personalization, real-time feedback, and task authenticity,

advanced or cloud-based AI applications. Future studies could expand the sample across multiple military

branches, examine long-term proficiency development, or explore how AI can support additional ESP

competencies such as listening comprehension, written reporting, and cross-cultural communication in

multinational operations. Experimental or longitudinal designs may also yield more profound insights into the

developmental trajectory of AI-assisted learning over time.

In conclusion, the study affirms that AI represents a promising and feasible pathway for modernizing language

training in defense settings. When guided by secure infrastructure, informed pedagogical practice, and

institutional alignment, AI can play a strategic role in preparing cadets for the linguistic demands of

contemporary military cooperation and technological operations.

Research designs: Quantitative, qualitative, neuropsychological, and biological (2nd ed., pp. 65–81).

3. American Psychological Association.

4. Chen, H. (2024). Innovative approaches in English language teaching: Integrating communicative

methods and technology for enhanced proficiency. Communications in Humanities Research, 32, 214–

220.

5. Creswell, J. W., & Plano Clark, V. L. (2018). Designing and conducting mixed methods research (3rd

ed.). SAGE Publications.