Technology-enhanced content and language integrated learning: A systematic review of empirical studies

Hu, Chen, Li, and Wang review how technology has been used in content and language integrated learning, or CLIL, where students learn a subject through a non-native language. The paper addresses an important AIED question: does adding digital technology help learners manage the dual demands of understanding subject content and developing language? Rather than assuming that technology is beneficial, the authors map what has been studied, which tools have been used, what outcomes have been measured, and which designs offer useful examples for practice.

The review searched Web of Science for English-language SSCI journal articles published from 2014 to 2023. From 390 records, the authors retained 20 empirical studies after title, abstract, and full-text screening. Two researchers screened and coded the studies independently, resolving disagreements through discussion; coding agreement was high, with a Kappa value of 0.91. The review classified education level, learning mode, subject domain, learners' first language, language of instruction, language skills, technology type, and cognitive or affective outcomes. For comparative evidence, it also synthesized nine experiments from seven controlled studies using a random-effects model.

The evidence base is concentrated. Fourteen of the 20 studies were conducted in Asia, five in Europe, and one in South America. Three quarters involved tertiary education, and 90 percent used English as the language of instruction. Productive language skills such as speaking and writing received more attention than reading and listening, although many papers did not specify a language-skill focus. The authors identify seven technology categories: telecollaboration tools, online learning platforms, multimedia, virtual learning environments, social-network applications, language tools, and intelligent conversation tools. Telecollaboration was most common, while intelligent chatbots appeared in only one included study.

The outcome evidence is more cautious than the technology range might suggest. All 20 studies reported cognitive outcomes, commonly language knowledge or subject-content knowledge, while eight examined affective outcomes. Only one addressed collaboration and communication, and none assessed higher-order thinking as an outcome. Across the controlled studies, the estimated effect on content knowledge was small at 0.29, with a 95 percent confidence interval from -0.07 to 0.66. The estimated effect on language achievement was 0.16, with a confidence interval from -0.03 to 0.35. Because both intervals include zero and the number of experiments is small, neither estimate provides firm evidence of a positive overall effect.

Three exemplary cases make the design implications more concrete: cooperative digital storytelling in a Taiwanese primary school, an educational chatbot in a Greek secondary-school history unit, and a cloud learning environment in a mainland Chinese university course. Across these examples, technology is most useful when it supports rich input, scaffolding, interaction, purposeful learner output, higher-order activity, and balanced development of both content and language. The chatbot case also showed that novelty or student preference did not automatically produce better achievement.

For Hong Kong educators, the review argues for design discipline. Technology should serve explicit content and language goals, with scaffolds that manage the cognitive load of learning subject matter through another language. Evaluation should include motivation, self-efficacy, interaction, and durable language development, not only immediate knowledge tests. The review is limited to English-language SSCI articles in one database, includes only 20 studies, and groups some EMI work with CLIL, so its estimates should not be treated as final. Its strongest contribution is a practical warning: effective TECLIL depends on pedagogy and evidence, not the presence of a digital tool alone.