Periodic Reporting for period 2 - SCIL (The role of social cues for infant word learning)
Période du rapport: 2018-04-01 au 2019-03-31
Babies start knowing the meaning of some words by 6 months of age, and by age two they know hundreds. Many researchers have suggested that social factors are central for this rapid acquisition. The mechanisms through which they support the learning process are, however, still poorly understood. SCIL aims at better understanding these mechanisms, focusing on the role of two salient elements of social interactions: The interactivity and human-likeness of an interaction partner. We study whether babies learn better from an interactive than a non-interactive teacher, and how the teacher’s human-likeness influences this process.
Understanding how the social environment shapes language learning is crucial for improving our recommendations for caregivers and educators. SCIL specifically contributes to recommendations pertaining to the role of interactive screen media in language education.
Understanding how the social environment shapes language learning is crucial for improving our recommendations for caregivers and educators. SCIL specifically contributes to recommendations pertaining to the role of interactive screen media in language education.
We developed an innovative gaze-contingent eye-tracking paradigm including a virtual agent and tested toddler’s language learning across a series of studies, varying the virtual agent's contingent reactivity and its human-likeness.
Across studies, we find that contingent reactivity improves the learning of novel word-object associations, but only if the virtual agent retains a certain degree of human-likeness. These results provide new insight into the minimum amount of social cues infants need in order to learn from an interaction partner. These results have been presented at numerous international conferences and as invited talks in labs in Europe, the US, and Japan. One of the studies is published, another under review, and two more pending write-up
Across studies, we find that contingent reactivity improves the learning of novel word-object associations, but only if the virtual agent retains a certain degree of human-likeness. These results provide new insight into the minimum amount of social cues infants need in order to learn from an interaction partner. These results have been presented at numerous international conferences and as invited talks in labs in Europe, the US, and Japan. One of the studies is published, another under review, and two more pending write-up
This project has contributed to a better and more nuanced understanding on how social cues in the environment influence language acquisition. Social cues can augment learning even for a non-human agent, a finding highly relevant for our understanding of the mechanisms through which the social environment supports language acquisition, and for research on the use of interactive screen media. Given the overwhelming rise of interactive smartphone and tablet applications targeted at infants and labeled as educational, but largely untested, we are convinced that research into sensible use of such opportunities is indispensable.