Plenary session by Danny Norrington-Davies and Richard Chinn
Emergent language: How we see it and what it can be
In today’s plenary session, Danny Norrtington-Davies and Richard Chinn introduce emergent language and share some updates on their latest insights.
They begin by mentioning the theoretical influence on their teaching: Scott Thorbury and Luke Medding’s Teaching Unplugged (2009). This is manifested in three tenets of teaching: conversation-driven, materials-light, and focus on emergent language.
Danny Norrtington-Davies and Richard Chinn define emergent language as any unplanned language that is needed or produced by the learners in meaning-focused interactions. Some examples of such language are errors, communication breakdowns, unclear expressions, and extension (e.g. excellent word choice, appropriate style, etc.). Emergent language need not always contain negative examples, but it is by no means part of the target language of a lesson.
They move on to suggest some criteria that teachers can use to prioritise certain kinds of emergent language. These include communication breakdowns, language or interaction skills related to the teaching context, repetitive mistakes, high-frequency expressions, task-specific language, and language that is judged to be new, interesting or useful to the learners.
To find out what emergent language means across various teaching and learning contexts, Danny Norrington-Davies and Richard Chinn spoke with a cohort of teachers. They shared the main themes in their findings:
- Teachers can create the conditions for language to emerge in meaningful interactions by extending lead-in discussions, using personal response questions, and asking open-ended follow-up questions.
- Teachers can use communicative tasks that do not specify any target language as it otherwise limits the room for focussing on emergent language.
- Topics that lie beyond a prescribed textbook syllabus can create the conditions for language to emerge.
Teachers can adopt the view of their role as helping learners to express their ideas effectively in their own ways, rather than trying to elicit what learners are expected to say according to the syllabus. Nevertheless, it does not necessarily mean a lack of consideration of the complexity of language for the learners.
They also mentioned some of the challenges that the teachers had faced when dealing with emergent language, such as different learning backgrounds or clashing beliefs about teacher and learner roles. Some learners may feel embarrassed to speak, whereas others adopt a defensive or confrontational attitude towards feedback as they see it as an affront to them.
Finally, Danny Norrington-Davies and Richard Chinn share an update on the practice of dealing with emergent language. They see it as a fluid concept that may contain different kinds of language depending on the specific classroom context. What is crucial, however, is the idea of eliciting the learners’ intended meaning by the words they have chosen to use. When dealing with emergent language, teachers are encouraged to develop their reactive skills [or responsive teaching skills] by reflecting on their own beliefs, on whether meaning has genuinely been explored, and on feedback from the learners on their learning.
- - - - -
Talk by Shaun Wilden
Unseen insight: Teacher reflection with AI mentoring
Shaun Wilden (IATEFL Vice-President) traces back to the reflective practice of journalling as the starting point of his experiment with artificial intelligence (AI). In his experiment, he uses AI tools as a partner or collaborator to aid the teacher in the process of self-reflection.
He uses Gibb’s Reflective Cycle to inform his experiment. The assumption is that ChatGPT has a comprehensive knowledge of English Language Teaching (ELT) theories as well as a thorough understanding of practices.
The two pillars of his experiment are AI mentoring and narrative self-reflection. This is also compatible with self-observation, which is especially useful for teachers working in contexts without organisational or mentoring support. Before teachers use ChatGPT, they need the following data:
- Detailed description of their classroom context [role]
(e.g. You are an English for Academic Purposes tutor teaching a university-level pre-sessional class.) - Transcript of the recorded lesson, with any sensitive information about the teacher and the students removed. For example, each student can be labelled anonymously with a letter or number.
- Teacher’s self-evaluation or journal entry of the lesson
The following steps are followed in the evaluation process:
- The teacher asks ChatGPT a (Socratic) question about the main learning aim of the lesson.
- The teacher asks ChatGPT another (Socratic) question about what it thinks the self-observation aim is.
- After ensuring that ChatGPT aligns with the teacher in their idea about the aims, the AI chatbot is asked to act as a lesson observer. It should analyse both the transcript and the teacher’s self-evaluation or journal entry.
- The teacher compares their own self-evaluation (what is thought to have happened) to ChatGPT’s observations (what actually happened).
Shaun Wilden concludes by suggesting that while ChatGPT can draw teachers’ attention to some overlooked aspects of the lesson, it remains a mirror of the teacher’s input, rather than a collaborator in its true sense. Nevertheless, there is still value in the teacher receiving non-judgmental feedback from ChatGPT. As most formal observations tend to happen once a year, ChatGPT can provide a sense of continuity in teachers’ self-observation journeys.
- - - - -
Talk by Chris Lewis
Belief and growth: Acknowledging principles in teacher development
Chris Lewis (EC English) invites the audience to consider why some teachers flourish in their professional development, while others seem to be resistant to growth.
He draws on Simon Sinek’s work, which centres around the ‘why’ behind what one decides to do. Although the ‘what’ and the ‘how’ are most often featured in our own thinking, it is more important to be aware of the driving force behind our actions or decisions.
In the context of teachers’ professional development, Chris Lewis suggests that we may not always be aware of the changes or developments in our practices over time. It would, therefore, be beneficial to examine our own teaching beliefs and principles on a regular basis. He shares an idea for encouraging teachers to do that in in-service training (inset) sessions. Teachers are given certain statements about teaching and learning, some of which may be deliberately provocative. Consequently, any outstanding points in their discussion can form the basis for future training sessions.
- - - - -
Workshop by Beatrice Segura Harvey
Beyond prompts: The great experiment with agentic AI
Beatrice Segura Harvey (Freelance | Nile ELT) demonstrates how we can harness the power of agentic AI to write materials for teaching and learning.
She defines agentic AI as a system in which various AI-powered agents or tools work behind the scenes to generate output for an AI chatbot (e.g. ChatGPT, Gemini, Claude, etc.). In this context, she sees large language models (LLMs) as a jack of all trades whose data supports different AI-powered agents.
She demonstrates how the AI chatbot Claude can write a textbook unit in under a few hours. Using this chatbot, she enlists ten additional AI-powered agents, each of which is tasked with a different aspect of textbook design. For each aspect, the following data needs to be fed to the corresponding AI-powered agent as input:
- Working documents (e.g. CEFR statements)
- Agent roles
- Curriculum outline
- Guardrails, i.e. limiting criteria to prevent AI-powered agents going off-topic
- Constraints
She divides the materials writing process into three phases: content, rendering (design), and quality control.
During her demonstration, Claude AI initially fails to produce her desired output. This shows that technology is not yet fully reliable. Nevertheless, it has the potential to increase our efficiency in certain aspects of materials writing.
- - - - -
Forum on AI with purpose: Practical frameworks for language classrooms
In the context of higher education, Nazli Deniz Barutçuoğlu (MEF University, Türkiye) proposes a tried-and-tested framework for integrating AI into teaching and learning.
She outlines the different types of lesson tasks according to their cognitive demands:
- Foundational tasks: AI-free
(e.g. activating prior knowledge, brainstorming ideas) - Structured tasks: AI-enhanced, but limits should be imposed on input prompts to prevent outsourcing of learners' cognition
(e.g. comparing information*, refining ideas, explaining ideas, providing critiques) - Open exploration tasks: AI-transformed
* Learners can make comparisons between their output and the output generated by an AI chatbot. [This echoes Mark Smith’s idea about cognitive processes in his talk on Day 1.]
She also suggests that effective prompts usually include the four elements: role, context, task, and control (bounds). [This echoes Peter Lucantoni and Emir Aydin’s workshop on Day 1.]
Learners can be scaffolded in writing effective prompts, moving from the teacher’s prompt as a model, through analysis and evaluation of sample prompts, to learner-generated prompts.
Rasha Halat (Lebanese International University) introduces two models of pedagogical design: TPACK and SAMR.
TPACK stands for the optimal area of a Venn diagram consisting of:
- Content: target concepts for learning or learning outcomes
- Pedagogy: teaching strategies based on the content
- Technology: knowledge of how AI works
She advocates that the most effective use of AI in the classroom is one that is strategic and ethical.
SAMR, which represents four increasing levels of AI integration, stands for:
- Substitution: using AI to complete lesson tasks with zero learner input
- Augmentation: using AI to enhance human ideas
- Modification: using AI to transform, revise or change human output
- Redefinition: AI-facilitated learning through questioning and prompting
In lesson planning, she suggests that teachers can consider the following elements:
- Outcome: content aims and learning aims
- Task: lesson task or activity
- SAMR: cognitive demands of AI integration
- TPACK: value for learning with AI use
- Guardrails: controls and bounds
Consequently, TPACK can be used as a checklist to align with the learning outcome of a lesson, whereas SAMR can be used to target the cognitive depth of a task.
Irme Fekete (Budapest University of Economics and Business, Hungary) presents several classroom ideas for EFL teaching and learning. To elaborate on Nazli Deniz Barutçuoğlu’s idea, Irme Fekete mentions that learners can compare the similarities and differences in both versions, as well as identify anything incoherent in AI-generated output.
Another idea involves the learners’ use of AI tools to generate images. Each learner receives a different part of a reading text, which is used individually for image generation. After that, the images of various parts are put together in a picture story. Learners are asked to discuss how representative the AI-generated pictures are of the reading text, including any inherent bias with AI tools.
His final idea is similar to the one above, but it involves the learners’ group discussion of AI-generated podcasts instead of images. Learners use Google LMNotebook or NoteGPT to create a podcast, which they subsequently listen to and evaluate its credibility.
- - - - -
Talk by Anna Lyons
How to help students avoid translation-app dependence
Anna Lyons (International House London) outlines several common reasons for the learners’ overreliance on translation apps in their learning process. These include low confidence, the lack of language skills, insufficient language knowledge, and pure convenience.
She first highlights some beneficial uses of translation apps, such as the mediation of certain concepts for the learners, comparative analysis between their first language (L1) and English (L2), clarification of language, and modelling judicious use of such apps.
She moves on to share what she has done to tackle the issue of overreliance.
- Learners need to deduce the overall idea of a reading text by using the words they know. For example, they match the ideas of the text to pictures. In a jigsaw reading task, learners may be allowed to look up a few words in the dictionary. These words are used productively in a later stage, such as by creating a poster to illustrate the ideas of the reading text.
- Collaborative writing activities without resorting to the use of translation apps. This has the benefit of peer teaching.
- Learners do a writing task in English. They do the same task again but write it in their L1 before they translate their written work into English. After that, they compare both versions and notice any differences in terms of vocabulary and sentence structures.