AI Twin: Enhancing ESL Speaking Practice through AI Self-Clones of a Better Me

TL;DR

This work introduces AI Twin, a personalized self-clone that rephrases a learner’s utterances into more fluent English and delivers them in the learner’s own voice to support ESL speaking practice. A within-subject study with 20 adult Korean ESL learners shows that in-conversation rephrasing enhances emotional engagement and, when aligned with the Ideal L2 Self, increases motivation compared with explicit feedback and non-personalized rephrasing. The approach integrates voice cloning, real-time reformulation, and context-aware prompting to preserve conversational flow while providing subtle corrective cues. These findings suggest that aspirational AI self-clones can meaningfully augment affective support and motivation in AI-mediated language learning, with implications for flexible design and ethical considerations in self-representational technologies.

Abstract

Advances in AI have enabled ESL learners to practice speaking through conversational systems. However, most tools rely on explicit correction, which can interrupt the conversation and undermine confidence. Grounded in second language acquisition and motivational psychology, we present AI Twin, a system that rephrases learner utterances into more fluent English and delivers them in the learner's voice. Embodying a more confident and proficient version of the learner, AI Twin reinforces motivation through alignment with their aspirational Ideal L2 Self. Also, its use of implicit feedback through rephrasing preserves conversational flow and fosters an emotionally supportive environment. In a within-subject study with 20 adult ESL learners, we compared AI Twin with explicit correction and a non-personalized rephrasing agent. Results show that AI Twin elicited higher emotional engagement, with participants describing the experience as more motivating. These findings highlight the potential of self-representative AI for personalized, psychologically grounded support in ESL learning.

Figures (7)

• Figure 1: System flow of AI Twin. The process begins with initialization, where the learner registers their voice to create a personalized clone. During practice, the conversation cycle repeats: the learner's speech is transcribed using ASR, reformulated by a LLM, and synthesized in the learner's cloned voice. The AI interlocutor then generates a response based on this reformulated speech, enabling interactive conversation practice.
• Figure 2: Rephrasing process in AI Twin. The learner's utterance, together with the ongoing dialogue context, is passed into the system prompt for the large language model (LLM). The LLM then generates a rephrased version of the learner's response, which is returned as clearer, more fluent English.
• Figure 3: Examples of the three study conditions. After each learner utterance, (1) Explicit Feedback provides direct correction, (2) AI Proxy offers rephrasing in a neutral synthetic voice, and (3) AI Twin (our approach) delivered the rephrased utterance in the learner's own cloned voice. Samples shown are drawn from the study data.
• Figure 4: Illustration of the AI Twin interface used in the study. Learners engaged in voice-based interaction by recording their speech and listening to the system's responses. Within study sessions, practice was situated in goal-oriented conversations, and after each learner utterance AI Twin provided feedback through rephrased speech. The interface text was originally in Korean; English translations are shown here for clarity.
• Figure 5: Distribution of participants' CEFR proficiency levels (N=20)