AI evidence extraction

Unpacking EFL teachers’ perceived opportunities and challenges of multimodal AI-driven language testing

Ali Derakhshan, Gurpinder Singh Lalli, Yujong Park · Language Testing in Asia · 2026

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

5/10

Relevance

1/4

Quality (LMQS)

Evidence

Citations

543.12

FWCI

Top 10%

Impact

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.1186/s40468-026-00445-5

Methodology & findings

Study design

Phenomenological qualitative study using online focus group interview with thematic analysis

Sample

N = 10, 2 groups

Primary method

Thematic analysis of qualitative data from focus group interview. Grounded in Expectancy-Value Theory (EVT) as the theoretical framework.

Main result

The study identified four perceived opportunities: "fostering the provision of personalized assessment, offering instant feedback, facilitating the assessment of productive skills, and improving efficiency, consistency, and reliability." Conversely, participants identified four significant challenges "specifically concerning the complication of construct validity, the potential for algorithmic bias, difficulties in score interpretation, and the necessity for robust technical infrastructure."

Reports effect sizes.

Research paradigm

Interpretivist/Constructivist

Author conclusions

The findings "offer implications for EFL teachers, language testers, teacher educators, and policymakers, highlighting the bright and dark sides of integrating multimodal AI tools into L2 testing and assessment practices."

Risk of bias

Selection bias: participants were experienced EFL teachers only; Self-selection bias: voluntary participation in focus group; Social desirability bias: group discussion setting may influence responses; Limited sample representation: single focus group with ten participants; Selection bias: Only experienced EFL teachers were recruited; sample of 10 participants may not represent broader EFL teacher populations; Single focus group methodology may be susceptible to groupthink or dominant personalities influencing responses; Geographic and institutional representation not specified, limiting generalizability; Selection bias - voluntary participation in focus group; Small sample size (n=10); Single data collection method; Potential groupthink in focus group setting

Limitations

The paper states this is a qualitative study with a limited sample size of ten EFL teachers collected through a single online focus group interview, which may limit generalizability
As a qualitative phenomenological investigation, the findings are context-dependent and may not be transferable across different educational settings or cultures.

Open questions raised

The authors note that while AI has transformed L2 testing, "empirical research into how EFL practitioners perceive the specific opportunities and challenges of these tools remains limited," which was the gap this study aimed to address. Future research should explore implementation strategies, address the identified challenges, and investigate teacher training needs for multimodal AI integration.
The study addresses a gap where "empirical research into how EFL practitioners perceive the specific opportunities and challenges of these tools remains limited."
The authors address a gap where "empirical research into how EFL practitioners perceive the specific opportunities and challenges of these tools remains limited."

Data: not_statedCode: not_statedExtracted from: pdfAgreement 70%

Explore related topics

AI Literacy Training AI Governance Qualitative