Questions in the TED-Multilingual Discourse Bank and the development of an annotation scheme

Bibliographic Data

ID21226990
AuthorsDeniz Zeyrek (0000-0001-9248-0141, Middle East Technical University), Amália Mendes (0000-0003-1020-8633, University of Lisbon)
Year2025
Volume11
Issues2
Pages215-228
Publication date2025-12-25
Peer ReviewedYes
Open AccessYes
TypeARTICLE
VenueLinguistics Vanguard (JOURNAL)
Journal identifiersISSN: 2199-174X • E-ISSN: 2199-174X
PublisherWalter de Gruyter GmbH (PUBLISHER • DE)
DOI10.1515/lingvan-2025-0035
OpenAlexW4411130979
LanguageEN
Citations received3
References cited20

In this paper, we analyze question-answer pairs and stand-alone questions within the spoken discourse of TED Talks, specifically focusing on the TED-Multilingual Discourse Bank. Our aim is to reveal various characteristics of questions through an annotation approach. We have developed a taxonomy, referred to as the TAQ-TED, to categorize types of questions and examine their information transfer and dialogue control functions, drawing on the Dynamic Interpretation Theory++ taxonomy of dialogue acts and their attribution in accordance with the Penn Discourse Treebank annotation guidelines. We outline the taxonomy, present our annotation results, and provide a preliminary cross-linguistic analysis comparing English questions with their Turkish and Portuguese translations. The TAQ-TED represents a promising initial framework for annotating questions in monologic discourse across multiple languages

Annotation · Linguistics · Computer Science · Mathematics · Multi-Agent Systems and Negotiation · Natural Language Processing Techniques · Philosophy · Speech and dialogue systems · Artificial Intelligence

Tools

Open DOIOpen Access