This projects aims to investigate (1) how the emotional states of speakers influence the language they produce and (2) how the influence of emotion on language production can be modeled in computational tools for affective natural language generation. Specifically, we ask both whether content selection ("deciding what to say") and message formulation ("deciding how to say it") are affected by emotional appraisals of the language user. This will be studied in a series of experiments, zooming in on referential communication; for example, we study whether happy speakers are more likely to describe people based on positive properties ("the smiling girl") even though other properties such as hair color ("the blond girl") would be equally effective from the addressees perspective. The experimental findings will feed into the development of a novel automatic news report generation tool that can adjust the message depending on the emotion it is intended to reflect (e.g., a report of the same match from the perspective of a losing team can be different from the report for the winning team). The application is developed with support from two parties from industry (Sanoma Media, in The Netherlands, and ARRIA, in the UK) that have expressed an interest in tools for the automatic generation of affective language. The findings from all three projects will be relevant for psycholinguistics, affective scientists, computational linguistics, and those working in the area of affective natural language generation. All our results will be made publicly availably (open source, open access).

In the NWO DATA2PERSON project BLISS (Behaviour-based Language-Interactive Speaking Systems), researchers from the Center for Language and Speech Technology (CLST) of Radboud University worked together with colleagues from Human Media Interaction (HMI) of the University of Twente, the company Games for Health (GfH) and the company ReadSpeaker on developing a voicebot that can communicate with people in Dutch about their daily lives and well-being. The BLISS bot can have personalized conversations about hobbies, favorite activities and interests that give users insights into their own lives, such as insights into their social network and their passions.

Although vulnerable workers that share a disability with other intersectional characteristics (e.g., migrant background and/or low socio-economic status) are increasingly employed, many are given short-term, precarious contracts. To make up for this these precarious forms of employment, we aim to design and implement online labor platform (OLP) cooperatives, and underlying matching algorithms, to foster the sustainable employment and inclusion of individuals with a disability. We do so, by combining insights from the engineering and social sciences and by involving individual workers, employers and government bodies.

Detoxification of substance use is an important step in addiction recovery. Patients often prefer detoxification within their own living environment. To facilitate safe and comfortable outpatient detoxification, close monitoring and rapid medical response are essential to mitigate severe withdrawal. Detox@Home supports this with a digital platform for self-report measures, telemonitoring of withdrawal severity, communication with therapists and eHealth support. We investigate patients’ and clinicians’ support needs during detoxification, build a digital platform and test this in clinical practice. Our team with patient representation, clinicians, researchers, a healthcare insurer and technological company ensures an efficient, safe and accessible outpatient detoxification.

We aim to acquire the iCub 2.0 robotic platform to investigate cognitive robotics, neurorobotics, social robotics and their integration. Cognitive robotics allows us to investigate cognitive abilities and the grounding of conceptual representations by means of the physical interactions of the iCub with its environment. Neurorobotics allows us to investigate neuronal learning processes based on these interactions. Social robotics allows us to investigate the social interactions between humans and robots, and the factors that determine the acceptance of robots by humans. The combination of cognitive robotics, neurorobotics and social robotics provides unique possibilities for the integration of these fields. The iCub allows us to test and investigate cognitive architectures and mechanisms by the direct execution of behavior in a physical environment, not just by simulation or analysis. Furthermore, the mechanisms involved concern the entire process from perception to action, instead of just a part of it (leaving open the question of how that part would fit in with the rest of the process). Comparing the behavior of the robot with that of humans allows us to test our models of the cognitive mechanisms that humans apply in similar tasks. We will use the robot to learn conceptual representations of its environment through the coordinated use of its senses and limbs. Understanding how the robot learns from interacting with its environment informs us about how humans could learn from similar experiences. The new capabilities of the iCub 2.0 (e.g., force control, optical flow, touch abilities) allow us to learn and investigate a range of novel conceptual representations. The successful development of robots will critically depend on whether humans accept robots in their environment and on how humans are able to interact with them in a safe and purposive manner. Conversely, the robot will have to learn its social behavior by interacting with humans. The iCub allows us to test and investigate human-robot dialogue based on the accurate detection of human non-verbal and verbal social behaviors and the real time behavioral responses of the robot. The iCub will allow us to address societal relevant contexts of robot applications such as health and care. The infrastructure for the iCub is provided for by the campus of the University of Twente, which guarantees the optimal use of the highly multidisciplinary research platform offered by the iCub. We provide access to and use of the iCub for other national research institutes.