Computer-assisted language learning and AI

Artificial intelligence (AI), in particular the areas of natural language processing and student modeling (Heift & Schulze, 2007; Schulze, 2008), have played a role in CALL – this sub-area is commonly referred to as Intelligent CALL or ICALL – for almost 50 years. For a small group of researchers, pedagogical issues of contingent corrective feedback and learner guidance and help were often at the center of attention. Since November 2022 and the release of ChatGPT 3.5, AI and especially the branch of generative AI (GenAI) has garnered widespread media attention and the interest not only of language teachers worldwide. Generative AI refers to AI systems that generate texts (and programming code, music, images, and video) that are based on the data on which they have been trained. In the context of language learning, the focus of these systems’ use shifted from providing corrective feedback and learner help to the generation of texts: from reading texts for students to lesson plans and quiz questions, from students’ discussion board posts and essays to interactive dialogs and word lists or other learning aids. What remained the same is that the learners are interacting with the computer system – this has been labeled Tutorial CALL (Heift & Schulze, 2015; Hubbard & Bradin-Siskin, 2004; Schulze, 2024) – rather than they are communicating with other people via the digital device.

This radically new and powerful GenAI technology has been met both with enthusiasm and caution. “Artificial Intelligence (AI) presents enormous global opportunities: it has the potential to transform and enhance human wellbeing, peace and prosperity. … Alongside these opportunities, AI also poses significant risks, including in those domains of daily life.”  (The Bletchley Declaration by Countries Attending the AI Safety Summit, 1-2 November 2023, 2023). Some leading entrepreneurs have argued that “AI will finally solve all of humanity’s problems” (Harari, 2024, p. xx); others have been more skeptical and “warned the public that AI could destroy our civilization” (ibid.). Also for the role of GenAI in the classroom and in education in general, different predictions have been made: teachers were cautioned to ignore the hype (Langreo, 2024), guided to embrace the potential (Banks, 2023), or warned of “language education [being] on the brink of singularity” (Obari, 2024). Various departments and ministries of education and educational authorities have issued policies and guidelines. For example, the US Department of Education – Office of Educational Technology give “Emphasize humans in the loop” (2023, p. 53) as their first and strongest recommendation. In 2023 and 2024, conferences on computer-assisted language learning and on applied linguistics had a markedly higher number of presentations on topics related to AI than in years prior. Organizations of and for world language teachers have been offering resources, webinars, and courses about GenAI in language education (e.g., ACTFL, 2024). The assumption is that GenAI is impacting pedagogy, curricular decisions, and day-to-day language teaching methodology in CALL and will continue to do so (Godwin-Jones, 2024). This exploration and the discussion are just beginning in 2024 for CALL. The emerging literature on GenAI in education shows that the appropriate and successful integration of GenAI tools requires teachers

• to understand the basics of these powerful AI technologies and their capabilities and limitations,
• to be reflective on the use of AI tools in their own work, their teaching, and their students’ learning, and
• to be aware of the role of GenAI in different communities and its impact on society.

To shed more light on the impact of artificial intelligence technologies on CALL, we will focus on GenAI since late 2022 in the context of developing written proficiency.

This blog post is an excerpt from the manuscript for 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. 
In 2024, I wrote this encyclopedia entry as my first attempt of gaining a better understanding of what was going on after GenAI burst into Language Education. 

AI in CALL – what changed

Artificial intelligence as a field of research and development started in 1956 (McCarthy et al., 1955). A first AI-based application for language learning was published in 1978 (Weischedel et al., 1978). Its further development for language education went largely unnoticed among teachers and applied linguists for almost 50 years. This began to change in late 2022. The underlying technology had changed and the results in, what computer scientists call, natural language understanding and generation improved dramatically. The field moved from symbolic natural language processing (NLP), which in ICALL used hand-programmed grammatical rules to parse learner texts, via statistical NLP to artificial neural networks and (deep) machine learning (see below).

The GenAI systems and apps with their enormous capacity to generate well-formed texts are based on large language models (Naveed et al., 2024). These models, which are multi-dimensional artificial neural networks, are built through machine learning, essentially a statistical analysis of very large amounts of texts from the internet. This analysis hardly relies on linguistic knowledge and language rules, as previous ICALL systems did. The patterns of language in text are not captured in grammatical and other language rules, but are stored in artificial neural networks, whose nodes have probabilistic weights. Generated texts appear well-formed, such that users almost automatically imbue them with rich meaning, when reading them. The GenAI tool only computed the form – what we read, hear, or see – and not the meaning – what we understand, feel, and believe. In other words, these models, which have been called “stochastic parrots”(Bender et al., 2021), generate plausible language, but do not understand what they generated. And the chatbots have no intention to produce meaning. Students are essentially asked to put meaning to the texts they had generated to get new information. This becomes even more complicated when the interaction of a user – student or teacher – with an AI tool is not rooted in a basic understanding of the ways, capabilities, and limitations of the underlying technology. It is this new aspect of computer literacy that is central in the meaningful use of GenAI in language education.

Before exploring this use, one more facet of the generation of the large language models is important: Large language models rely on large collections of texts from the internet. Yet most websites, and hence texts, on the internet are in English (about 50% (Wikipedia contributors, 2024)). The commonly taught languages follow (Spanish = 5.9%; German = 5.4%; French = 4.3%). Only a few less commonly taught languages have a share of 1% of websites or more: Japanese (4.9%), Russian (4.1%), Portuguese (3.7%), Italian (2.6%), Dutch (2.1%), Turkish (1.9%), Polish (1.8%), Persian (1.3%), Chinese (1.2%), and Vietnamese and Indonesian (1.1% each). A determining factor of the quality of text-generating AI tools is the number of texts and tokens that can be used for machine learning and the creation of a large language model. Whereas the 15 languages listed here cover approximately 85% of all websites, the remaining approximately 6,000 languages have far fewer texts as the basis for a large language model. Especially for some endangered languages with few speakers and often an oral culture, large language models are hardly feasible (Trosterud, 2023). This has important consequences for CALL, in that teachers of these languages often cannot rely on GenAI tools in the teaching and learning.

Kern (2024, p. 516) argues in his position paper on 21^st-century technologies in language education that

Our profession is truly at an inflection point. In the face of technologies that seem to provide the support people need to function reasonably well in another language, we urgently need to articulate and communicate the value of language study in a social context, identify what technology offers that is positive for language education, rethink how we organize our teaching in light of technology’s affordances, and be clear about what technology cannot do.

The impact of GenAI on CALL is complex and multidirectional, and thus different areas of language learning and teaching in general and of CALL specifically will be affected. These areas can be labeled (1) Language use (2) Learner motivation, (3) Learner agency, and (4) Assessment. Questions and challenges for CALL pedagogy that are emerging in these four areas will be sketched in turn.

References

ACTFL. (2024). Making AI Work for Language Teachers. Retrieved Sep 30 from https://www.actfl.org/professional-learning/discover-series/making-ai-work-for-language-teachers

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

Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021). On the Dangers of Stochastic Parrots. In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency (pp. 610-623). https://doi.org/10.1145/3442188.3445922

The Bletchley Declaration by Countries Attending the AI Safety Summit, 1-2 November 2023. (2023). London: Government of the United Kingdom Retrieved from https://www.gov.uk/government/publications/ai-safety-summit-2023-the-bletchley-declaration/the-bletchley-declaration-by-countries-attending-the-ai-safety-summit-1-2-november-2023

Godwin-Jones, R. (2024). Distributed agency in second language learning and teaching through generative AI. 5-31. https://hdl.handle.net/10125/73570

Harari, Y. N. (2024). Nexus. A brief history of inrmation networks from the stone age to AI. Random House.

Heift, T., & Schulze, M. (2007). Errors and Intelligence in CALL. Parsers and Pedagogues. Routledge.

Heift, T., & Schulze, M. (2015). Tutorial CALL. Language Teaching, 48(4), 471–490.

Hubbard, P., & Bradin-Siskin, C. (2004). Another Look at Tutorial CALL. ReCALL, 16(2), 448–461.

Kern, R. (2024). Twenty‐first century technologies and language education: Charting a path forward. The Modern Language Journal, 108(2), 515-533. https://doi.org/10.1111/modl.12924

Langreo, L. (2024). Don’t buy the AI hype, learning expert warns. Education Week. https://www.edweek.org/technology/dont-buy-the-ai-hype-learning-expert-warns/2024/08

McCarthy, J., Minsky, M. L., Rochester, N., & Shannon, C. E. (1955). A Proposal for the Dartmouth Summer Research Project on Artificial Intelligence. Retrieved Sep 30 from http://www-formal.stanford.edu/jmc/history/dartmouth/dartmouth.html

Naveed, H., Khan, A. U., Qiub, S., Saqib, M., Anwar, S., Usman, M., Akhtar, N., Barnes, N., & Mian, A. (2024). A Comprehensive Overview of Large Language Models. https://dx.doi.org/10.48550/arxiv.2307.06435 http://arxiv.org/pdf/2307.06435

Obari, H. (2024). Language learning at the brink of singularity: AI’s impact on educational paradigms. In Proceedings of the International CALL Research Conference, 2024.

Schulze, M. (2008). Modeling SLA Processes Using NLP. In C. Chapelle, Y.-R. Chung, & J. Xu (Eds.), Towards Adaptive CALL: Natural Language Processing for Diagnostic Assessment. (pp. 149-166). Iowa State University. https://apling.engl.iastate.edu/wp-content/uploads/sites/221/2015/05/5thTSLL2007_proceedings.pdf

Schulze, M. (2024). Tutorial CALL — Language practice with the computer. In R. Hampel & U. Stickler (Eds.), Bloomsbury Handbook of Language Learning and Technologies (pp. 35–47). Bloomsbury Publishing.

Trosterud, T. (2023). CALL for all languages? Why languages differ and what consequences that has for CALL. Keynote presentation Eurocall 2023, Reykjavik.

US Department of Education – Office of Educational Technology. (2023). Artificial intelligence and the future of teaching and learning.  Retrieved from https://tech.ed.gov/ai-future-of-teaching-and-learning/

Weischedel, R. M., Voge, W. M., & James, M. (1978). An Artificial Intelligence Approach to Language Instruction. Artificial Intelligence, 10, 225-240.

Wikipedia contributors. (2024). Languages used on the internet. In Wikipedia. https://en.wikipedia.org/wiki/Languages_used_on_the_Internet