Abstract:

Chatbots are increasingly being used in language courses to create opportunities for individualized speaking practice. However, the usefulness of such practice depends not only on whether students produce grammatically accurate language, but also on whether they can sustain a purposeful professional exchange. In investor-style conversations, this means responding to questions directly, developing relevant information, and keeping the pitch aligned with the expectations of the funding scenario. This study therefore looks closely at what happens when engineering students interact with a chatbot investor and the conversation begins to move off track.

The study is based on interaction data from engineering students who used a Poe-based chatbot as preparation for live, assessed funding pitches. The chatbot was designed to play the role of a potential investor evaluating students’ innovation proposals. Using the systemic functional linguistic concept of field, or “what the conversation is about,” the analysis focuses on moments when students lost control of the conversational focus. These moments are described as field shifts: points in the interaction where the student’s response no longer fully matched the investor’s question, the proposal content, or the communicative purpose of the pitch.

The analysis identified several recurring patterns in how conversations shifted. Some students Zoomed Out, offering broad or generic answers instead of proposal-specific responses. Others produced a Wide Miss, responding in a way that moved away from the core issue raised by the investor. In other cases, students gave Blurred Shot responses, where potentially relevant ideas were included but presented in a fragmented or difficult-to-follow manner. These shifts were not simply problems of vocabulary or grammar. They reflected difficulties in managing relevance, role expectations, and the unfolding logic of the professional exchange.

By examining these shifts, the study shows how chatbot interaction data can make visible the communicative challenges students face during simulated professional speaking tasks. The presentation argues that field shifts can be used as diagnostic indicators for teaching and feedback, helping instructors move beyond general comments such as “be clearer” or “give more detail.” Instead, teachers can help students notice how and when their responses drift, what effect this has on the interlocutor, and how they might keep the conversation more closely aligned with the task. The session will be relevant to language teachers, communication instructors, and instructional designers interested in using AI-mediated interaction to support professional communication development.

Keywords:

English for Engineering, spoken interaction, chatbots, field shift.