Roeland Ordelman
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Roeland Ordelman has a total of 102 co-authors. Here are the top co-authors by number of shared publications:
Franciska de Jong
16 shared publications- Gareth J. F. Jones
ADAPT Centre, Hamilton Institute, Maynooth University, Maynooth, Ireland
8 shared publications Martha Larson
Radboud University
8 shared publications- Khiet P. Truong
University of Twente, Enschede, Netherlands
6 shared publications Maria Eskevich
EURECOM, Sophia Antipolis, France
6 shared publicationsRobin Aly
University of Twente
6 shared publications
Latest Publications
- Open Access
- 13 August 2026
- Published by Cogitatio in Media and Communication
- Vol. 14
Twi-XL: An Infrastructure for Cross-Media Research in the Netherlands
In the domain of media and communication research, copyright and privacy issues provide significant obstacles to achieving the strategic goals of open science. Investigating the role of different media in shaping public debates is especially difficult, as research across different media collections and platforms is not yet systematically supported by open research infrastructures on a global scale. Moreover, the private ownership of crucial communication spaces such as social media results in unreliable access opportunities for researchers and unclear conditions regarding the completeness and integrity of data where platform owners provide their own archives. This also creates obstacles for users without a technical background who are interested in gathering information, exploring, or making sense of big data. In response to these challenges, a consortium of two universities (the University of Amsterdam and the University of Groningen), two national archives (the National Library of the Netherlands and the Netherlands Institute for Sound & Vision), and SURF as a national research IT supporter created Twi-XL: Designed to facilitate cross-media research, it provides a research environment in which users of varying computational skill levels can apply automated methods to a range of archival collections, including websites, public broadcasts, radio and podcast transcripts, and over 10 years of Dutch tweets. The societal relevance of Twi-XL lies in creating a user-friendly tool that can be used by researchers to analyze societal phenomena across social media and traditional media, capturing how public debate unfolds across interconnected media environments. We describe and demonstrate the functional logic of Twi-XL with two use cases focused on: (a) social media affordances and users’ practices of news sharing, and (b) the broader public debate on sexual misconduct. Going beyond the dominance of the English language in cross-media research, our results not only showcase the advantages of developing cross-media infrastructures specific to the Dutch language, but also how Twi-XL can be utilized in qualitative and quantitative research. Embedding Twi-XL in the broader discussion on transparency, reproducibility, and accessibility, we discuss the implications and feasibility of such infrastructures, particularly for analyzing social media content.
- 13 August 2026
- Published by Cogitatio in Media and Communication
- Vol. 14
Open AccessTwi-XL: An Infrastructure for Cross-Media Research in the Netherlands
In the domain of media and communication research, copyright and privacy issues provide significant obstacles to achieving the strategic goals of open science. Investigating the role of different media in shaping public debates is especially difficult, as research across different media collections and platforms is not yet systematically supported by open research infrastructures on a global scale. Moreover, the private ownership of crucial communication spaces such as social media results in unreliable access opportunities for researchers and unclear conditions regarding the completeness and integrity of data where platform owners provide their own archives. This also creates obstacles for users without a technical background who are interested in gathering information, exploring, or making sense of big data. In response to these challenges, a consortium of two universities (the University of Amsterdam and the University of Groningen), two national archives (the National Library of the Netherlands and the Netherlands Institute for Sound & Vision), and SURF as a national research IT supporter created Twi-XL: Designed to facilitate cross-media research, it provides a research environment in which users of varying computational skill levels can apply automated methods to a range of archival collections, including websites, public broadcasts, radio and podcast transcripts, and over 10 years of Dutch tweets. The societal relevance of Twi-XL lies in creating a user-friendly tool that can be used by researchers to analyze societal phenomena across social media and traditional media, capturing how public debate unfolds across interconnected media environments. We describe and demonstrate the functional logic of Twi-XL with two use cases focused on: (a) social media affordances and users’ practices of news sharing, and (b) the broader public debate on sexual misconduct. Going beyond the dominance of the English language in cross-media research, our results not only showcase the advantages of developing cross-media infrastructures specific to the Dutch language, but also how Twi-XL can be utilized in qualitative and quantitative research. Embedding Twi-XL in the broader discussion on transparency, reproducibility, and accessibility, we discuss the implications and feasibility of such infrastructures, particularly for analyzing social media content.
- Open Access
Self-supervised learning of speech representations with Dutch archival data
This paper explores the use of Dutch archival television broadcast data for self-supervised learning of speech foundation models, specifically wav2vec 2.0. We first study data quality assumptions for pre-training, and show how music, noise and speaker overlap affect SSL convergence and downstream fine-tuning performance. Secondly, we explore effectively pre-processing strategies to convert the noisy broadcast dataset into a qualitative dataset for pre-training, by using Whisper and WhisperX. Thirdly, we compare mono-lingual and multi-lingual pre-training with equivalent amounts of data, and show that mono-lingual pre-training is more robust to out-of-domain data. Lastly, we achieve a state-of-the-art LARGE wav2vec 2.0 model for the Dutch language, by a continuation of pre-training a wav2vec 2.0 XLS-R model checkpoint with our 55k hour archival dataset.
Open AccessSelf-supervised learning of speech representations with Dutch archival data
This paper explores the use of Dutch archival television broadcast data for self-supervised learning of speech foundation models, specifically wav2vec 2.0. We first study data quality assumptions for pre-training, and show how music, noise and speaker overlap affect SSL convergence and downstream fine-tuning performance. Secondly, we explore effectively pre-processing strategies to convert the noisy broadcast dataset into a qualitative dataset for pre-training, by using Whisper and WhisperX. Thirdly, we compare mono-lingual and multi-lingual pre-training with equivalent amounts of data, and show that mono-lingual pre-training is more robust to out-of-domain data. Lastly, we achieve a state-of-the-art LARGE wav2vec 2.0 model for the Dutch language, by a continuation of pre-training a wav2vec 2.0 XLS-R model checkpoint with our 55k hour archival dataset.
- 25 March 2025
- Published by Springer Nature
- p. 417-430
Ask and You Shall Find: How Suggestions by a Conversational Robot Assist Children with Information Search
- 25 March 2025
- Published by Springer Nature
- p. 417-430
Ask and You Shall Find: How Suggestions by a Conversational Robot Assist Children with Information Search
- Open Access
- 16 September 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-4
‘Uhm... Are you sure?’ An Exploratory Study of Trust Indicators in Robot-Directed Child Speech
In order to calibrate children’s trust in robots toward appropriate levels in the interaction, reliable trust measures are necessary. Current trust measures are not suitable for measuring children’s trust in a real-time manner. While speech from adult speakers has proven to contain information on their trust, this paper presents a first exploration investigating whether these results hold up in the context of a child-robot interaction. Fifty-eight conversations between children and robots were recorded (N=29), evoking high and low trust moments in the interaction. Correlation tests showed no (strong) predictors of children’s trust in their speech. Limitations and possibilities of how to advance the investigations of an automatic trust measure for child-robot interaction are discussed.
- 16 September 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-4
Open Access‘Uhm... Are you sure?’ An Exploratory Study of Trust Indicators in Robot-Directed Child Speech
In order to calibrate children’s trust in robots toward appropriate levels in the interaction, reliable trust measures are necessary. Current trust measures are not suitable for measuring children’s trust in a real-time manner. While speech from adult speakers has proven to contain information on their trust, this paper presents a first exploration investigating whether these results hold up in the context of a child-robot interaction. Fifty-eight conversations between children and robots were recorded (N=29), evoking high and low trust moments in the interaction. Correlation tests showed no (strong) predictors of children’s trust in their speech. Limitations and possibilities of how to advance the investigations of an automatic trust measure for child-robot interaction are discussed.
- Open Access
- 14 July 2024
- Published by Elsevier in International Journal of Child-Computer Interaction
- Vol. 41
Does conversation lead to better searches? Investigating single-shot and multi-turn spoken searches with children
- 14 July 2024
- Published by Elsevier in International Journal of Child-Computer Interaction
- Vol. 41
Open AccessDoes conversation lead to better searches? Investigating single-shot and multi-turn spoken searches with children
- 23 March 2024
- Published by Springer Nature
- p. 144-151
A Conversational Robot for Children’s Access to a Cultural Heritage Multimedia Archive
- 23 March 2024
- Published by Springer Nature
- p. 144-151
A Conversational Robot for Children’s Access to a Cultural Heritage Multimedia Archive
- Open Access
- 1 January 2024
- Published by Elsevier in SSRN Electronic Journal
Does Conversation Lead to Better Searches? Investigating Single-Shot and Multi-Turn Spoken Searches with Children
Interactive speech technologies such as Siri and Amazon Echo allow people to search for information via speech. These systems generally take a single statement as input, which may not be enough to fully understand the user's need. More interactive, multi-turn conversations, have been proposed to establish more details of users' information needs. A conversational approach can particularly benefit non-standard audiences, such as children who require more support when formulating queries. We set up a study with 32 children comparing multi-turn conversations with single-shot (or Query-Response) interactions. We looked at the search need descriptions in both conditions and children's experiences while searching. Additionally, we derived design guidelines for conversational agents for children's information searching. Findings indicate that more elaborate descriptions of children’s information needs (more keywords) were gathered by the conversational interaction, leading to more focused search results compared with the Query-Response interaction. Further analysis of children’s responses yielded insights into their preferences for conversing with speech-based interfaces.
- 1 January 2024
- Published by Elsevier in SSRN Electronic Journal
Open AccessDoes Conversation Lead to Better Searches? Investigating Single-Shot and Multi-Turn Spoken Searches with Children
Interactive speech technologies such as Siri and Amazon Echo allow people to search for information via speech. These systems generally take a single statement as input, which may not be enough to fully understand the user's need. More interactive, multi-turn conversations, have been proposed to establish more details of users' information needs. A conversational approach can particularly benefit non-standard audiences, such as children who require more support when formulating queries. We set up a study with 32 children comparing multi-turn conversations with single-shot (or Query-Response) interactions. We looked at the search need descriptions in both conditions and children's experiences while searching. Additionally, we derived design guidelines for conversational agents for children's information searching. Findings indicate that more elaborate descriptions of children’s information needs (more keywords) were gathered by the conversational interaction, leading to more focused search results compared with the Query-Response interaction. Further analysis of children’s responses yielded insights into their preferences for conversing with speech-based interfaces.
- Open Access
- 19 April 2023
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Children’s Trust in Robots and the Information They Provide
Previous work has shown that children tend to trust embodied conversational agents such as social robots. Also, that children have difficulty assessing the credibility of information. The study reported in this paper addresses how children’s attitudes toward and trust in a robot affect their acceptance of information provided by the robot. We conducted a within-subjects study (N=30) where children engaged with a ‘trustworthy’ versus an ‘untrustworthy’ robot. Due to the pandemic period, this interaction was carried out via video call. The children played a quiz with the robot where we measured whether they accepted the information provided by the robot. Results show that the manipulation of trustworthiness was successful. We did not find evidence for a causal relationship between trust in the robot and acceptance of the information. Furthermore, semi-structured interviews offered a more in-depth understanding of how children perceived the two different robots and their preference for the trustworthy robot.
- 19 April 2023
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Open AccessChildren’s Trust in Robots and the Information They Provide
Previous work has shown that children tend to trust embodied conversational agents such as social robots. Also, that children have difficulty assessing the credibility of information. The study reported in this paper addresses how children’s attitudes toward and trust in a robot affect their acceptance of information provided by the robot. We conducted a within-subjects study (N=30) where children engaged with a ‘trustworthy’ versus an ‘untrustworthy’ robot. Due to the pandemic period, this interaction was carried out via video call. The children played a quiz with the robot where we measured whether they accepted the information provided by the robot. Results show that the manipulation of trustworthiness was successful. We did not find evidence for a causal relationship between trust in the robot and acceptance of the information. Furthermore, semi-structured interviews offered a more in-depth understanding of how children perceived the two different robots and their preference for the trustworthy robot.
- 31 December 2016
- Published by Springer Nature
- p. 280-292
Multimodal Video-to-Video Linking: Turning to the Crowd for Insight and Evaluation
- 31 December 2016
- Published by Springer Nature
- p. 280-292
Multimodal Video-to-Video Linking: Turning to the Crowd for Insight and Evaluation