LLMs have revolutionized the development of interactional artificial intelligence (AI) systems, due to their accessibility to the general public. Significant advances continue to be observed in fields and applications such as multimodal conversational agents, emotionally and socially aware dialogue, hyper-personalized and context-adaptive interaction, human–AI collaboration and symbiotic AI, multi-party and social conversation modelling, immersive and embodied conversational systems, and responsible, controllable, and domain-aware interaction. The use of these interactional AI systems is increasingly widespread, as these models have produced remarkable achievements in several benchmarks. State-of-the-art NLP systems achieve impressive performance, but remain prone to brittleness in language understanding (LU); they often lack robust modelling of communicative intentions, pragmatic inference, and context-sensitive meaning, which are central to many EMNLP tasks (dialogue, QA, safety, content moderation, etc.). This raises doubts about the extent to which such systems can really understand human language(s).

The rapid deployment of LLMs raises urgent questions about interpretability, trust, and accountability of systems that produce linguistically plausible but potentially misleading outputs. In this scenario, this dedicated workshop provides a venue for cross-fertilisation between formal/theoretical work and applied NLP, in line with EMNLP’s focus on cutting-edge, high-impact research in NLP.

The 3rd Language Understanding the Human-Machine Era workshop (LUHME 2026) focuses on how meaning is represented, inferred, and negotiated across human and machine environments, with particular emphasis on logic-aware, pragmatics-sensitive, and socially grounded approaches to LU in NLP. The workshop brings together researchers in (computational) linguistics, formal semantics/pragmatics, dialogue and discourse, cognitive science, and AI safety/ethics to discuss what it means for systems to understand language in situated, interactive contexts.

Topics of interest

Submissions are invited on (but not limited to):
- Language understanding in LLMs
- Language grounding
- AI‑mediated communication in high‑stakes domains (law, health, finance, governance)
- Psycholinguistic approaches to LU
- Discourse, pragmatics and LU
- Intent detection
- Computational treatment of speech acts, dialogue, and communicative intentions in interaction
- Conversation analysis, narrative progression, and argumentation
- Turn‑taking, repair, and alignment in human–AI interaction
- Evaluation of LU
- Human vs. machine LU
- Machine translation/interpreting and LU
- Multimodality and LU
- Socio-cultural aspects in LU
- Effects and risks of language misunderstanding
- Manifestations of language (mis)understanding
- NLU and toxic content
- Ethical issues in LU
- Distributional semantics and LU
- Linguistic theory and LU by machines
- Linguistic, world, and commonsense knowledge in LU
- Role of language professionals in the LLM era
- Understanding language and explainable AI
- Use of LLMs in generating, analysing or evaluating linguistic data

We invite researchers interested in the intersection of language understanding and the effective use of language technologies in human-machine interaction to submit papers and participate in LUHME 2026.

For further information, please visit https://luhme.up.pt/