MINT: Multimodal Interaction in Face-to-Face Dialogue is a new workshop on modelling, generating, and understanding multimodal verbal and non-verbal behaviour in face-to-face communication.

Human communication is inherently multimodal: meaning is conveyed not only through words and speech, but also through prosody, gesture, facial expression, gaze, and body pose. Yet much multimodal NLP still treats these signals as peripheral context or as outputs generated only after linguistic content has already been decided. MINT addresses this gap by focusing on multimodal interaction as a core part of communication.

The workshop aims to bring together researchers across computational linguistics, NLP, computer vision, cognitive science, HCI, and robotics to study the resources, modelling approaches, processing methods, and evaluation strategies needed for face-to-face dialogue.

We are particularly interested in work that connects insights about human multimodal communication with the development of AI systems that interact with humans or simulate human-like behaviour. This includes research on new datasets, multimodal learning methods, embodied agents, social robots, and evaluation frameworks for naturalistic interaction.

We invite submissions to MINT: Multimodal Interaction in Face-to-Face Dialogue. The workshop brings together researchers from computational linguistics, NLP, computer vision, HCI, robotics, and cognitive science working on multimodal face-to-face communication.

We welcome work on topics including:

• computational models that integrate verbal and non-verbal cues such as speech, text, gesture, facial expression, gaze, and body pose;
• cognitive and linguistic insights about face-to-face communication that can inform AI systems;
• multimodal datasets with synchronised speech, video, and motion data;
• evaluation methods for multimodal interaction;
• applications and tools for embodied conversational agents, social robots, annotation, and behavioural analysis.

We welcome both archival and non-archival contributions. Details on submission routes, formatting, review, and presentation are provided below.

Important dates

Deadlines are at 11:59 PM Anywhere on Earth (AoE).

Submission information

Papers should be prepared using the official ACL formatting guidelines and the official ACL style files.

Submission channels and portals

MINT will accept submissions through two channels:

• Direct submission. The dedicated OpenReview portal can be found here https://openreview.net/group?id=EMNLP/2026/Workshop/MINT. Archivable papers submitted through this channel will be reviewed by the MINT programme committee (submission deadline: July 8th, 2026).
• ACL Rolling Review (ARR). Authors may submit through ARR and commit their paper together with the ARR reviews to MINT at https://openreview.net/group?id=EMNLP/2026/Workshop/MINT_ARR_Commitment. Please consult the ARR author guidelines for current procedures and requirements. The pre-reviewed ARR commitment deadline is August 24th, 2026.

Paper categories and length

MINT welcomes both archival and non-archival papers:

• Archival papers. Submissions must be anonymous and report original, unpublished research to appear in the workshop proceedings. These may be submitted as long papers (up to 8 pages plus references) or short papers (up to 4 pages plus references).
• Non-archival papers. Non-anonymous submissions reporting previously published work, preliminary research, or demos to be presented at the workshop and not published in the MINT proceedings.

Anonymization

Archival submissions should be anonymized for double-blind review. Please remove author names, affiliations, acknowledgements, and other identifying information from the manuscript and supplementary material where appropriate. Non-archival submissions do not need to be anonymous.

Cross-submission

We allow cross-submissions to other venues. However, to be included in the proceedings, authors of accepted papers must withdraw them from any other venue where they remain under consideration

Reviewing criteria

We aim to provide three reviews for each archivable direct submission. Non-archival  submissions will be evaluated primarily for relevance to the workshop theme and their potential to stimulate discussion. Submissions may be desk-rejected if they don’t fall within the workshop’s scope or violate basic scientific integrity guidelines (e.g., fabricated references to inexistent work).

Presentation

Accepted contributions will be invited for presentation at the MINT workshop co-located with EMNLP 2026 in Budapest, Hungary. Presentation formats may include posters and selected talks, depending on the final program.

We thank the Multimodal Language Department (MLD) at Max Planck Institute for Psycholinguistics, for sponsoring this workshop.