Testing and Assessment
ALTE Conference 2026: A Look at Current Trends

Experts from research and professional practice gathered to discuss key future-oriented issues in foreign-language testing and assessment, under the conference theme Responsible Language Assessment to Empower Learners. The event centred first on the responsibility that comes with language testing, since recognised language certificates can open doors to employment and higher education. Ensuring fair assessment practices and outcomes is a shared responsibility among all stakeholders, particularly test providers and (political) decision-makers.

Secondly, the conference placed a strong emphasis on learners themselves. Learner self-efficacy is becoming an increasingly important focus. The growing availability of tools and apps for self-directed learning, along with new opportunities to interact with artificial intelligence (AI) in the language-learning process, means learners are expected to take responsibility for their own learning (learner agency). This personal responsibility should be actively fostered, for instance by German language teachers in the classroom or through appropriate assessment formats.

The conference’s overarching theme was divided into four thematic strands:

• Responsibility and Social Justice in Language Assessment
• Learner Agency in Language Assessment
• Responsible AI & Tech in Language Assessment to Empower Learners
• Wider Communication Skills to Empower Learners

Perspectives on responsible language assessment

The three keynote speakers offered deeper insights into the different aspects of the conference theme.

Dr Jennifer Randall provided important perspectives on the area of social justice, calling for a radical rethinking of the validity of assessments. Her presentation questioned the fairness of our understanding of valid assessments, particularly for marginalised groups of learners. She followed this with suggestions for making the concept of validity more equitable, for example, when defining what a test is intended to measure or when developing assessment content.

Dr Diane Larsen-Freeman spoke in her presentation about promoting learner autonomy – an increasingly important competency in a world that is being rapidly reshaped by AI. She introduced methods of integrating learner agency into language assessments, such as giving learners a choice of tasks in assessment components like writing and speaking, or allowing learners to control the pausing and replaying of audio recordings in listening comprehension tasks.

On the topic “Do Large Language Models (LLMs) Understand Language?”, Dr Gino Roncaglia took conference participants on an enlightening historical journey through the development of LLMs, focusing specifically on bias, hallucinations and social sustainability. He explained how language models can systematically favour or distort certain groups, perspectives and information (bias), and generate false or inaccurate content (hallucinations). When such phenomena occur in the context of AI-based language assessment, they can lead to unfair outcomes. The presentation also highlighted the growing role LLMs are playing in education, including teaching, learning and assessment.

The keynote speeches introduced the conference’s three main themes and outlined key considerations for responsible language testing and assessment.

AI topics take centre stage

The conference programme, comprising 140 presentations, panel discussions and workshops, reflected the current trends shaping specialist discourse in the field. Interest in integrating AI into language learning and assessment remains strong. Presentations on the thematic strand “Responsible AI & Tech in Language Assessment” introduced approaches, studies and concrete examples of how AI can be used in language assessment.

One example is the AI-assisted, (semi-)automated creation of assessment materials and relevant quality assurance measures. A second trend to emerge from the conference presentations was the automated assessment of written assignments and spoken responses using AI systems. Moritz Dittmeyer (Goethe-Institut) described the possibilities and limitations of the AI-assisted analysis of written texts in language teaching using the intelligent correction assistant INKA. Such feedback on texts as part of formative assessment can support the learning process and promote learner autonomy. Formative assessment is the gathering of information about learners’ progress during the learning process, for example through exercises or learning journals, with a focus on feedback.

“The AI-assisted development of assignments and assessment materials is already a highly dynamic field. When it comes to actually assessing – that is, evaluating, correcting and making decisions about language proficiency – greater caution is needed. The underlying AI technologies have advanced sufficiently; the key challenge now is to use them to develop smart, robust and pedagogically sound applications. In assessment contexts especially, validity, fairness, transparency and accountability are key. The journey from promising prototypes to production-ready applications is therefore a challenging and resource-intensive once.

The keen interest shown in INKA at the conference showed me that there is a real need for pragmatic solutions that can be integrated into the learning process – systems that do not replace teachers but provide concrete support, improve the quality of feedback and leave the final decision to humans. Especially when resources are limited, such smart approaches can offer real added value.”

Dr Moritz Dittmeyer

Using technology in language assessment also makes it possible to analyse data from live examinations. This is of particular interest to large-scale test providers, especially as a means of detecting attempts at cheating during exams. The measures presented are a response to the worldwide increase in cheating in high-stakes examinations.

Experts agree that AI must be used responsibly in language assessment and remain subject to human oversight. This “human-in-the-loop” approach allows humans to retain control over key decisions, such as the final assessment of a written test or the evaluation of an AI-generated text’s quality in relation to a given assignment.

Helping learners develop a broad range of communication skills

This thematic strand broadened the perspective to include aspects of language acquisition that go beyond purely linguistic skills. These include mediation activities and strategies, plurilingual repertoires and metacognitive skills for regulating one’s own learning, for instance through individual learning strategies. Presentations in this strand explored a wide range of such skills and how they might be integrated into teaching and assessment.

On the subject of mediation, Swapna Kulkarni-Ajgaonkar and Afsheen Jivani (Goethe-Institut Pune) presented a qualitative study conducted during language courses in Pune and outlined the opportunities and challenges associated with integrating mediation activities.

“In the future, we would like to integrate more authentic, everyday assignments into our teaching. This could include activities that require learners to summarise information for a certain target audience, bring together content from different sources or mediate between people with different linguistic backgrounds.

We also see interesting options for end-of-course assessments. Short mediation tasks could help assess learners’ ability to communicate more authentically.”

Afsheen Jivani und Swapna Kulkarni-Ajgaonkar

Insights for the future of language assessment

The ALTE Conference 2026 demonstrated that responsible language assessment remains a dynamic field, one in which new technological possibilities, societal demands and scientific insights must be continually realigned. The future of language assessment will be shaped not by technological innovation alone, but by how responsibly these innovations are used by humans to effectively support the learning process.