RoGameMM — Open-Access Components

RoGameMM is a multimodal corpus of spontaneous Romanian narrative speech (gaming narratives) annotated for speech disfluencies and their co-occurring bodily behaviours: head movement, gaze, gesture, and adaptors. This record contains the openly licensed components of the corpus (CC BY 4.0): ANNOTATION_RoGameMM.txt — the annotation manual, documenting the full pipeline from GDPR-compliant data collection through Whisper transcription, Praat alignment at the inter-pausal unit (IPU) level (Ro-Phon-based conventions), disfluency annotation, and two-pass multimodal annotation in ELAN. Speaker1_RoGameMM.TextGrid — Praat TextGrid for Speaker 1 (~64 min) with three tiers: enriched orthographic transcription (ORTHO), IPU segmentation, and a closely aligned English translation. RoGameMM_template_protocol.etf / .eaf — ELAN template implementing the tier hierarchy, linguistic types, and controlled vocabularies of the multimodal annotation scheme, developed jointly with Maria Graziano (Lund University Humanities Lab). RoGameMM_sample_4min.eaf — a fully annotated four-minute sample (speech, disfluency, head, gaze, gesture, and adaptor tiers) illustrating the scheme in use; distributed without media. Data_Use_Agreement.pdf — the Data Use Agreement governing access to the restricted components. The audio and video recordings contain personal data (identifiable faces and voices) and are therefore deposited separately under restricted access; access requires an approved request and a signed Data Use Agreement. This pilot was developed within the project Pauses, Boundaries, and Meaning: Inter-Pausal Units as Discourse Structuring Devices in Romanian Spontaneous Speech (PBM), ICUB, University of Bucharest, Young Researchers Grant no. 13137/16.12.2025.