3. Varied interaction in language-learning tasks
The human-machine conversation often works because we are used to adhere – even if the machine cannot and is not — to Grice’s four maxims of conversation (Grice, 1975): quantity (be informative), quality (be truthful), relation (be relevant), and manner (be clear). Interaction in dialog works because readers look at the mathematically compiled output of the GenAI and assume that it is informative, truthful, and relevant. Due to its generation of linguistically accurate and plausible text, the GenAI appears to be clear. Communicative interaction proceeds successfully as long as the human reader does not detect that the machine output is not truthful or factually accurate because of, for example, hallucinations (Nananukul & Kejriwal, 2024) or errors or is not relevant because of misinterpreting an ambiguity in different contexts (e.g., when asked about bats, giving information about the mammal rather than the intended sports instrument).
And here is lesson #3 of a short series. Part 0 gives a historical introduction. Lesson #1 focuses on the necessary exposure to authentic language and whether this can be done with GenAI. Lesson #2 looked at communication in context, which is central in language learning. And now we are turning to the role of interaction in language learning with GenAI.
Besides these hurdles, GenAIs have become interesting verbal interactants in language education. On the other hand, ICALL systems, mainly due to their limited language coverage (see above), have provided limited interaction. Systems with or without AI worked with branching trees and canned text, for example in Quandary, a software of the Hot Potatoes suite (Arneil & Holmes, n.d.), which does not have NLP built in. Other systems were more like Chatbots whose conversation was limited to one topic or topic area (e.g., Underwood, 1982). Such CALL chatbots were inspired by Weizenbaum’s Eliza (for his reflection see Weizenbaum (1976)) and SHRDLU (Winograd, 1971) and often relied on regular expressions (Computer Science Field Guide, n.d.) and keyword searches. More sophisticated NLP was employed in the interactive games Spion (Sanders & Sanders, 1995) and Kommissar (DeSmedt, 1995). These early examples of the direct interaction of a learner with a machine with some AI capabilities, especially a level of NLP, show that GenAI has opened a door to the possibility of many more complex and comprehensive verbal interactions and role plays in a variety of languages.
My inspiration for this title came from the book
Snyder, T. (2017). On tyranny: Twenty lessons from the twentieth century. Tim Duggan Books.
I am sharing these early drafts of a book chapter I published in
Yijen Wang, Antonie Alm, & Gilbert Dizon (Eds.) (2025),
Insights into AI and language teaching and learning. Castledown Publishers.
https://doi.org/10.29140/9781763711600-02.
Of course, language learning tasks (see Willis (1996) for an early introduction to the now commonly applied Task-based Language Teaching) are not only rooted in conversations and role plays. GenAI can also generate model answers for different task components or be employed for brainstorming first ideas in the pre-task steps, for example. This was impossible with the ICALL systems based on symbolic NLP and (limited) expert systems. A discussion of the affordances and challenges of this powerful generation of (partial) task outcomes and components both by the student or the teacher is beyond the confines of this chapter, but it is an area within the application of GenAI in language education that is in urgent need of discussion. This agentive collaboration in dialog, possible scaffolding, and student guidance can either support or hinder and even prevent learning.
References
Arneil, S., & Holmes, M. (n.d.). Quandary. Retrieved January 17 from https://hcmc.uvic.ca/project/quandary/
Computer Science Field Guide. (n.d.). Regular expressions – Formal Languages. Retrieved January 27 from https://www.csfieldguide.org.nz/en/chapters/formal-languages/regular-expressions/
DeSmedt, W. H. (1995). Herr Kommissar: An ICALL Conversation Simulator for Intermediate German. In V. M. Holland, J. D. Kaplan, & M. R. Sams (Eds.), Intelligent Language Tutors: Theory Shaping Technology (pp. 153-174). Lawrence Erlbaum Associates.
Grice, H. P. (1975). Logic and Conversation. In D. Cole & J. Morgan (Eds.), Syntax and Semantics: Speech Acts (pp. 41-58). Academic Press.
Nananukul, N., & Kejriwal, M. (2024). HALO: an ontology for representing and categorizing hallucinations in large language models. Proc. SPIE 13058, Disruptive Technologies in Information Sciences VIII, 130580B (6 June 2024),
Sanders, R. H., & Sanders, A. F. (1995). History of an AI Spy Game: Spion. In V. M. Holland, J. D. Kaplan, & M. R. Sams (Eds.), Intelligent Language Tutors: Theory Shaping Technology (pp. 141-151). Lawrence Erlbaum Associates.
Underwood, J. H. (1982). Simulated Conversation as CAI Strategy. Foreign Language Annals, 15, 209-212.
Weizenbaum, J. (1976). Computer Power and Human Reason: From Judgment To Calculation. W. H. Freeman.
Willis, J. R. (1996). A framework for task-based learning. Longman.
Winograd, T. (1971). Procedures as a representation for data in a computer program for understanding natural language. https://hci.stanford.edu/winograd/shrdlu/AITR-235.pdf
Panta Rhei Enterprise 