language learning – Panta Rhei Enterprise

Ten skill and knowledge areas for (language) teaching with GenAI

We propose 10 areas for language teachers to develop fundamental knowledge and skills in. While these do not constitute an exhaustive list, they cover a wide range of the main applications of GenAI for language education. We suggest that teachers devote a few hours to exploring each as soon as possible, reflecting critically on how they may be incorporated in current teaching contexts. Additional informed work can then proceed more organically as noted in the following section on our proposed reimagining of continuing professional development. For teachers who have already been using GenAI regularly and have confidence in certain areas, this section can serve to highlight potential gaps.

This is part of a draft of an article I wrote with Phil Hubbard. He wrote this part. In this paper, we are proposing a way in which teachers can organize their own professional development (PD) in the context of the rapid expansion of Generative AI.
We call this PD sustained integrated PD (GenAI-SIPD). Sustained because it is continuous and respectful of the other responsibilities and commitments teachers have; integrated because the PD activities are an integral part of what teachers do anyway; the teacher retains control of the PD process.

The full article is available as open access:
Hubbard, Philip and Mathias Schulze (2025) AI and the future of language teaching – Motivating sustained integrated professional development (SIPD). International Journal of Computer Assisted Language Learning and Teaching 15.1., 1–17. DOI:10.4018/IJCALLT.378304 https://www.igi-global.com/gateway/article/full-text-html/378304

Adapting and creating prompts. A GenAI prompt is the set of instructions provided by the user to get the GenAI to generate the desired output. Prompt templates for particular purposes provide useful starting points, and then teachers can adapt these as needed. The act of adapting predefined prompts or creating novel ones is often referred to as prompt engineering. This is typically an iterative process, where an initial prompt is followed by adjustments to arrive at the desired outcome. Importantly, that outcome should always be checked. For an example of an introduction to prompts, see Heaps (2024). This source distinguishes strategies for “one big prompt,” chaining prompts for dividing a complex task into parts, and question and answer prompts.
GenAI ethics. There are a number of ethical areas surrounding the use of GenAI, ranging from what the underlying language model, to issues of what constitutes acceptable use by teachers and learners, privacy issues, safety issues, credibility issues, equity issues, issues in output bias, and even considerations of environmental impact. Some of the information teachers and learners need in order to make ethical decisions is easily accessible, but some is hidden within the language of user agreements and default settings of the controls. Many institutions already have guidelines. For example, the Ethics Institute at Michigan State University maintains a site that provides ethical guidelines and examples for teaching with AI: https://ethics.msu.edu/gen-ai/teaching. A recent paper by Ohashi and Hubbard (2025) discusses GenAI ethics in depth in the context of language education and offers a set of emerging principles to guide language teachers in ethical use.
GenAI chatbots. AI chatbots have been around for many years in language learning (Coniam, 2008), but GenAI ones are relatively new. They are a quintessential application of AI for language education as the vehicle for interaction (the student’s L2) is also the target of learning. There are at least three main uses for a GenAI chatbot. One is to interact with the chatbot in the target language to assist with content searches or (allowed) translation of texts beyond the learner’s level for some ancillary purpose (see point 6). A second use is for the learner to produce target language texts that the learner then edits, keeping transparent both the original AI-generated and adapted version, or the teacher can use GenAI to produce examples that the learner then renders into coherent prose and incorporates in their work. Third, a chatbot can be a conversation partner, though as noted previously, the GenAI chatbot itself is not negotiating meaning. However, their use in principle allows a simulated interaction that can provide both direct conversational practice and learner-produced material for AI evaluation and feedback. Elements such as perceived usefulness and ease of use have been shown to be important considerations in selecting and employing GenAI chatbots (Zou et al., 2024). As reported in various studies (Huang & Zou, 2024; Peng & Liang, 2025; Wang et al., 2024) for some students GenAIs like ChatGPT as conversation partners can lower affective barriers and improve willingness to communicate.
GenAI translation. A special case for language education is the impact of GenAI on machine translation (MT). As noted earlier, the availability and appropriate use of MT in language learning has been an issue for several decades. How can it serve as an aid to learning rather than as a shortcut for students that is detrimental to the goal of learning? LLMs like ChatGPT can act as translators and in the role of a conversational agent may be useful for tasks like providing a needed word or phrase as part of a natural communicative interaction rather than as a dedicated translation tool. There is a general sense of the need for teachers to develop tasks and to help learners to understand ways in which translation can be helpful.
Course and lesson creation. A major area for teachers is the potential of GenAI to save time producing course curricula and lesson plans as well as generating lesson content targeted to a specific teaching context. For example, Contact North’s AI Assistant at https://contactnorth.ca/news/introducing-ai-teaching-assistant-pro-a-free-personal-teaching-aide can produce a draft course curriculum, complete with lesson outlines and course notes, simply by having a user fill out an online form. How efficient and effective this might be for a language class as opposed to a physics class remains an open question. However, even in its present state, GenAI is worth experimenting with as a tool for generating lesson ideas and sample content. Another element of lesson creation is the potential of GenAI to personalize learning so that students could work with materials and learning paths tailored to their current level, personal interests, and learning styles. This is frequently mentioned as an affordance of AI but remains an underexplored area. Using GenAI tools to generate additional differentiated tasks for individual students and small sub-groups as well as useful examples and bases for student exploration and discussion has been suggested as a way forward (Mittal et al., 2024).
Classroom assessment. Gen-AI based assessment is a popular but challenging application because its accuracy is often questionable. However, there are two areas where such assessment may be well-supported. The first is in the simple assessment of areas like reading or listening comprehension via quizzes. A variety of options exist for generating such quizzes, such as an open-access one from the Canadian non-profit Contact North: https://www.aiteachingassistantpro.ca/. As with all GenAI applications, it is important for teachers to review the output carefully and make edits before releasing the result to students. Recent research suggests that GenAI is good at writing multiple choice questions but has significant issues with providing plausible distractors (Chun & Barley, 2024). It is of utmost importance to consider pedagogical and ethical consequences. When a teacher generates a quiz, it is more likely that students can generate the answers, and it is questionable how a teacher should react to generated answers, when students are answering a generated question. The second is formative assessment and feedback for learners on their own work, especially in writing. Studies looking at this application for writing have noted that the feedback is primarily focused on surface issues but that students still find it useful. Finally, GenAI applications can be used for summative or proficiency assessment, though extra care must be taken in checking their work.
Student misconduct. Soon after ChatGPT 3.5 went public in November 2022, the most visible concern involved students using GenAI to complete assignments and assessments with the assistance of this tool. There were efforts to outright ban ChatGPT and its cousins because there was confusion about how to employ it in a way that didn’t constitute cheating or otherwise subverting learning objectives and the goal of course assessment (e.g., Banks, 2023). For entities at a number of levels, from government, to institutions, to individual teachers, the primary responses to this issue have been 1) to provide guidelines for what constitutes acceptable and unacceptable use in course assignments and assessments and 2) to develop strategies to limit the risk of students violating those guidelines, such as flipped learning (Pratschke, 2024) and assessing process rather than product (Schulze, 2025a). GenAI has the potential to be a valuable partner to students in language classes but only if they use it to support rather than interfere with learning.
Learner training for AI. Closely tied to the setting of guidelines for what constitutes acceptable use of GenAI is the training of learners in that acceptable use. And beyond just acceptable use, teachers can train learners in effective use so that the GenAI becomes a positive participant in language learning tasks without compromising the outcome (Wang & Stockwell, 2024). It is just as important for learners to become proficient in GenAI for their language learning as it is for teachers, a point that has been made for technology more generally in language learning (see e.g., Hubbard, 2013; Lai & Lin, 2012). Learner peers can be helpful as well in this process through sharing relevant strategies.
Awareness of functionalities of general vs. targeted GenAIs. Much of our discussion here has revolved around general purpose GenAI applications like ChatGPT or Gemini, but it is also common to find applications that embed GenAI, often using a smaller language model. Godwin-Jones (2023) notes that creating these has become more accessible since descriptions of the desired functions of a “narrow app” can now be done by prompting in plain language rather than needing programming skills. He cites the example of Lan and Chen (2024), who reported on a project where they created a dedicated GenAI tutor to support learners’ story writing. As using such apps often involves the same privacy and safety considerations as the general GenAI apps, it is important for teachers to be aware of what underlies these narrow apps and how user data is treated.
Awareness and control of other forms of AI. As we noted earlier, GenAI is the most visible AI at the moment, but there are other AI implementations relevant for language learning. It is easy in some instances to distinguish GenAI from other forms of AI when using branded GenAI apps like ChatGPT or Gemini. In other cases, the form is not clear. Automatic Writing Evaluation, for instance, relied on other types of AI well before the appearance of GenAI (Lu, 2019). Also, the same type of AI applications that are used to create individual user models for personalized recommendations by streaming services like Netflix and online shopping sites like Amazon have the potential to personalize language instruction as well through student modeling.

We proposed 10 key areas for teachers to develop foundational knowledge and skills in as soon as possible so that they can make informed decisions regarding GenAI use in their teaching contexts. We have not provided details of either the precise level of knowledge and skills, nor information beyond a few examples of where and how to gain that knowledge and skills. Those decisions need to be contextualized and consider what a teacher already knows and can do. However, a useful informal rubric for a teacher might be “How well can I explain what GenAI is with respect to the preceding 10 key areas to another teacher or to my students?” Giving these explanations to students is an excellent example of a teachable moment, and helping colleagues with GenAI is exemplary collaborative professional development.

References

Banks, D. C. (2023, May 18). ChatGPT caught NYC schools off guard. Now, we’re determined to embrace its potential. Chalkbeat. https://www.chalkbeat.org/newyork/2023/5/18/23727942/chatgpt-nyc-schools-david-banks/

Chun, J. Y., & Barley, N. (2024). A comparative analysis of multiple-choice questions: ChatGPT-generated items vs. human-developed items. In C. A. Chapelle, G. H. Beckett, & J. Ranalli (Eds.), Exploring artificial intelligence in applied linguistics (pp. 118–136). Iowa State University Digital Press. https://doi.org/10.31274/isudp.2024.154.08

Coniam, D. (2008). Evaluating the language resources of chatbots for their potential in English as a second language. ReCALL, 20(1), 98–116. https://doi.org/10.1017/S0958344008000719

Godwin-Jones, R. (2023). Emerging spaces for language learning: AI bots, ambient intelligence, and the metaverse. Language Learning & Technology, 27(2), 6–27. https://doi.org/10125/73585

Heaps, T. (2024). Generative artificial intelligence: Practical uses in education. Open Education Manitoba. https://pressbooks.openedmb.ca/aiineducation/chapter/writing-and-refining-prompts/

Huang, F. & Zou, B. (2024). English speaking with artificial intelligence (AI): The roles of enjoyment, willingness to communicate with AI, and innovativeness. Computers in Human Behaviour. 159. 108355.https://doi.org/10.1016/j.chb.2024.108355

Hubbard, P. (2013). Making a case for learner training in technology-enhanced language learning environments. CALICO Journal, 30(2), 163–178. https://doi.org/10.11139/cj.30.2.163-178

Lai, C., & Lin, X. (2012). Strategy training in a task-based language classroom. The Language Learning Journal, 43(1), 20–40. https://doi.org/10.1080/09571736.2012.681794

Lan, Y. J., & Chen, N. S. (2024). Teachers’ agency in the era of LLM and generative AI. Educational Technology & Society, 27(1), I–XVIII.

Lu, X. (2019). An empirical study on the artificial intelligence writing evaluation system in China CET. Big Data, 7(2), 121–129.

Mittal, U., Sai, S., Chamola, V., & Sangwan, D. (2024). A comprehensive review on generative AI for education. IEEE Access, 12, 142733–142759. https://doi.org/10.1109/ACCESS.2024.3468368

Ohashi, L. (2024). AI in language education: The impact of machine translation and ChatGPT. In P. Ilic, I. Casebourne, & R. Wegerif (Eds.), Artificial intelligence in education: The intersection of technology and pedagogy (pp. 289–311). Springer. https://doi.org/10.1007/978-3-031-71232-6_13

Pratschke, B. M. (2024). Generative AI and education: Digital pedagogies, teaching innovation and learning design. Springer Nature.

Schulze, Mathias (2025). The impact of artificial intelligence (AI) on CALL pedagogies. In Lee McCallum & Dara Tafazoli (eds) The Palgrave Encyclopedia of Computer-Assisted Language Learning. Palgrave Macmillan, Cham. https://doi.org/10.1007/978-3-031-51447-0_7-1.

Wang, C., Zou, B., Du, Y., Wang, Z. (2024) The impact of different conversational generative AI chatbots on EFL learners: An analysis of willingness to communicate, foreign language speaking anxiety, and self-perceived communicative competence, System, 127,103533. https://doi.org/10.1016/j.system.2024.103533

Wang, Y., & Stockwell, G. (2024). Training to use machine translation for vocabulary learning. In M.F. Teng, A. Kukulska-Hulme, J.G. Wu (Eds.) Theory and practice in vocabulary research in digital environments (pp. 132-153). Routledge.

Zou, B., Wang, C., Yan, Y., Du, X., & Ji, Y. (2024). Exploring English as a foreign language learners’ adoption and utilisation of ChatGPT for speaking practice through an extended Technology Acceptance Model. International Journal of Applied Linguistics (pp. 1-16). https://doi.org/10.1111/ijal.12658

Education and AI: Tool versus tutor

Of course, a language teacher is more than a benevolent conversation partner. In AI, an intelligent tutoring system (ITS) would be more akin to a language teacher than a chatbot would. An ITS consists of three interacting components (see Heift & Schulze, 2007):

The expert model, which captures the domain knowledge or the information that students should learn;
The tutor model, which makes decisions about the instructional sequences and steps as well as appropriate feedback and guidance for the group as a whole and for individual students;
The student model, which records and structures information about the learning progress and instruction received, domain beliefs and acquired information, as well as the learning preferences and styles of each student.

This is part of a draft of an article I wrote with Phil Hubbard. In this paper, we are proposing a way in which teachers can organize their own professional development (PD) in the context of the rapid expansion of Generative AI.
We call this PD sustained integrated PD (GenAI-SIPD). Sustained because it is continuous and respectful of the other responsibilities and commitments teachers have; integrated because the PD activities are an integral part of what teachers do anyway; the teacher retains control of the PD process.

Only if the sole learning objective is conversational ability, can one assume that the LLM has elements of an expert model. The other two models, however, cannot be mimicked by a GenAI tool. Consequently, teachers still have to teach – determine instructional sequences, time appropriate feedback, remember and work with an individual student’s strengths and weaknesses – also when using GenAI tools in various phases of the learning process. GenAI tools can provide multiple ideas for engaging learning activities, texts for reading with a ready-made glossary, or drafts of an entire unit or lesson plan. However, it is the teacher who must understand, select, adapt, and implement them. The entire teaching process and its success are still the responsibility of the teacher.

Grammar teaching in antiquity — Grammar teaching in Ancient Rome (generated by ChatGPT 5.1)

In an educational institution, teachers can meet this responsibility because learners normally trust their expert knowledge, because teachers have been trained, certified, and frequently evaluated. The same is not (yet) true of GenAI tools. They have been trained through machine learning, but their semantic accuracy and pragmatic appropriateness have often been found lacking. The generated text is plausible, but not necessarily factually correct or complete. This way, GenAI output is an insufficient basis for successful learning. This becomes apparent not only when one tries out a GenAI tool in the area of one’s own expertise, but also when one looks back on what teachers have said about the various levels of trustworthiness of internet texts, which also formed the basis for the machine learning for LLMs, for the last thirty years: sources have to be checked and validated. In machine learning for LLMs, the texts and sources are not checked nor validated. This can impact the content accuracy of LLM output. Of course, learners cannot be expected to check the accuracy of information they are only about to learn; believing the truth value of the information is a prerequisite for learning. Critical analysis and questioning the information learnt is always a second step. Also, first studies have emerged that show that GenAI can create the illusion of knowing and thus of learning (Mollick, 2024); consequently, chatbots are not always a tool for successful learning.

The main thing to remember is: these GenAI chatbots are a tool and not a tutor – more like a hammer than an artisan, more like a dictionary than an interpreter, and more like an answering machine (remember those?) than a teacher.

References

Heift, Trude and Mathias Schulze (2007). Errors and intelligence in CALL: Parsers and pedagogues. Routledge.

Mollick, E. (2024). Post-apocalyptic education: What comes after the homework apocalypse. https://www.oneusefulthing.org/p/post-apocalyptic-education

Translation and AI: Separated by a common language

In the interaction with a chatbot, one can change the language or prompt the machine to reply in another language than that of the prompt or request a translation of a text generated previously. It is therefore not surprising that dedicated machine translation (MT), such as Google Translate and DeepL, relies on LLMs and thus artificial neural networks in the way that GenAI chatbots do. MT was there at the beginning of AI (see above) and was rooted in symbolic approaches, using grammatical rules and lexical items. MT output had to be post-edited by human translators as a matter of course. Such post-editing tasks and the critical reading and analysis of MT output have also been used in language learning with some success (e.g., Niño, 2009). Post-editing was necessary and language learning with MT was useful, because the linguistic accuracy of the MT output was such that it needed post-editing, and it contained more errors than most language learners would make. Today the MT output has high levels of linguistic accuracy and complexity similar to the turns generated by GenAI chatbots. Based on our impression over the years, we would submit that MT output is usually more complex and accurate than the writing of many language learners.

Machine translation in the Cold War. Generated by ChatGPT 5.1

Ohashi (2024) states, “Numerous studies have been conducted on the use of MT in language education” (p.292) recently and discusses several review and literature studies. Lee and Kang (2024) conclude from their study “that MT helped students deliver their meaning, reduce grammatical errors, find appropriate vocabulary, and use expressions and sentence structures beyond their current levels” (p. 12) Yet they also admit that the improved accuracy of the translated texts is not necessarily an indication of successful and sustained language learning.

Schulze (2025a) highlights some of the problems of generating written texts in the language being learned: the speed of the text generation does not encourage planning, thinking, and intentional engagement and the plausibility of the machine output makes processes of checking and correction very difficult if not impossible, especially for learners who use GenAI tools habitually. This applies in equal measure to GenAI MT. As a matter of fact, the conundrum of GenAI as a powerful tool in multilingual communication becomes clearer when one looks at MT. The speed and linguistic accuracy of translation makes it feasible to have a GenAI-tool-mediated conversation at almost normal interactional speed, with both sides only producing and receiving text in their first language and not being able to check the communicative adequacy and felicity of either. Language teachers will have to determine their stance vis-à-vis MT and, more importantly, the continued motivation to go through the long process of learning another language in times of instant results from machine translation and chatbots.

Summing up this excursion into GenAI, we can conclude with the essence of the seven lessons in Schulze (2025b) and state that

This new technology facilitates the exposure to rich and authentic language and,
GenAI potentially enriches learning with additional opportunities of communicative interaction and language use, because it offers a new way of communicating;
And yet necessary processes of appropriate error correction and feedback as well as documenting learner behavior and dynamic individualization cannot be performed using GenAI tools and remain the responsibility of the teacher.

References

Lee, S. M., & Kang, N. (2024). Effects of machine translation on L2 writing proficiency: The complexity, accuracy, lexical diversity, and fluency. Language Learning & Technology, 28(1), 1–19. https://doi.org/10.1016/j.langlt.2024.73585

Niño, A. (2009). Machine translation in foreign language learning: Language learners’ and tutors’ perceptions of its advantages and disadvantages. ReCALL, 21(2), 241–258. https://doi.org/10.1017/S0958344009000172

Schulze, M. (2025a). The impact of artificial intelligence (AI) on CALL pedagogies. In L. McCallum & D. Tafazoli (Eds.) The Palgrave Encyclopedia of Computer-Assisted Language Learning. Palgrave Macmillan, Cham. https://doi.org/10.1007/978-3-031-51447-0_7-1.

Schulze, M. (2025b). ICALL and AI: Seven lessons from seventy years. In Y. Wang, A. Alm, & G. Dizon (Eds.), Insights into AI and language teaching and learning. (pp. 11–31) Castledown Publishers. https://doi.org/10.29140/9781763711600-02