Generative Artificial Intelligence in the Assessment of Young Multilingual Learners’ Productive Language Skills

• In Event: AI Innovations in Language and Literacy Education for K-12 Multilingual Learners

Abstract

Generative artificial intelligence (GenAI) is rapidly transforming educational assessment, with growing interest in its use for evaluating young learners’ productive language skills, i.e., speaking and writing. While the use of GenAI has gained momentum in higher education and adult learning environments, its application in K–8 language assessment—particularly for multilingual learners—remains underexplored (Huang & Yan, forthcoming; Williamson & Eynon, 2020). As educational researchers and practitioners seek developmentally appropriate, valid, scalable, and equitable assessment solutions, GenAI presents both promising innovations and pressing challenges. This presentation highlights four emerging applications of GenAI in assessing productive language skills in young learners: (1) content generation, (2) test administration, (3) automated scoring, and (4) assessment preparation.

1. Content and Item Generation: With the right pedagogical and linguistic prompts, GenAI tools such as OpenAI’s GPT-4 and Google’s Gemini can produce developmentally appropriate and culturally responsive assessment content (Ijiri & Healy, 2026). For example, AI can generate picture-based tasks, age-appropriate stories, or writing prompts that reflect diverse cultural experiences, such as a Día de los Muertos celebration or Lunar New Year festivities. These tools can also create parallel forms of tasks with controlled variation in linguistic complexity and theme, supporting differentiated instruction and adaptive assessment models.

2. Gamified, Automated Test Delivery: Child-centered interfaces that leverage GenAI, such as animated avatars, story-based scenarios, or game-like environments can enhance student engagement during assessments (Lamrani & Abdelwahed, 2020). AI-powered chatbots can simulate natural peer or teacher interactions and guide students through spoken or written tasks in playful, low-pressure settings. These systems can also adapt in real time based on learners’ responses, adjusting task difficulty or scaffolding based on need. Such personalization may be especially important for young learners who experience anxiety or have limited exposure to formal testing environments.

3. Scoring of Productive Language Skills: The use of AI to score young learners’ speaking and writing is an area of growing research interest. AI models are increasingly adept at analyzing features such as oral fluency, lexical richness, and syntactic variety in children’s speech (Hsieh & Wang, 2019) and writing samples (Suhan & Wolf, 2025). Despite these advancements, concerns about fairness and algorithmic bias persist, especially for emergent multilingual learners from racially and linguistically diverse backgrounds. Ensuring transparency and equity in scoring systems must remain a central priority.

4. AI-Supported Assessment Preparation: In addition to formal assessment, GenAI tools can serve as conversational partners and feedback engines for low-stakes practice. Platforms like ChatGPT and Curipod enable interactive role-playing (e.g., describing a pet or buying fruit at a market), providing instant, tailored feedback to help learners rehearse expressive skills. Early findings suggest that AI-based dialog tools improve L2 learners’ oral fluency and narrative cohesion with repeated use (Wiboolyasarin et al., 2025).

In sum, GenAI introduces powerful opportunities for evaluating and supporting productive language skills among young multilingual learners. This presentation will explore emerging use cases, current limitations, and future directions for responsible integration of GenAI into K–8 assessment practices via a human-centered approach (Miao & Holmes, 2023; UNESCO, 2021).

Author

• Becky H. Huang, The Ohio State University