Martin Heckmann, Holger Brandl, Xavier Domont, Miguel Vaz, Jens Schmüdderich, Bram Bolder, Frank Joublin, Christian Goerick, "Language Acquisition Embedded into Tutor-Robot Interaction", Proc. ACORNS Workshop Computational Models of Language Evolution, Acquisition and Processing, 2009.

Children acquire language to a large extend in the interaction with their caregivers. Inspired by this observation we develop computational models and artifacts for the acquisition of language in an interactive scenario. The artifact bootstraps its representations with little a priori knowledge and can be taught by a human tutor. In this framework we investigate different aspects of the speech acquisition process. This encompasses the learning of speech features, word and sub-word units as well as the production of acquired speech units. As speech features we explore a set of hierarchical spectro-temporal features which are learned in an unsupervised fashion based on the observed speech data. Phone-like speech units emerge from an unsupervised clustering process. These phone-units can then be used to bootstrap word learning in an interactive scenario where a tutor shows a visual property and at the same time utters a corresponding speech label. Thereby an auditory attention system and predefined key phrases trigger the learning behavior. Finally the learned units can also be reproduced.