Body-Mind-Language: Embodied Cognition in Natural Language

Vortrag von Dagmar Gromann

Veranstaltungsort: APB 3027
Beginn: 23. November 2017 um 09:20
Ende: 23. November 2017 um 10:50
Forschungsgruppe: Computational Logic
Event series: KBS Seminar
iCal

Embodied cognition starts from the assumption that conceptual structure in our minds derives from sensorimotor experiences. Cognitive linguistics has provided compelling evidence that semantic structure in natural language reflects that conceptual structure arising from our embodied experience in the world. Thus, natural language provides an excellent source of knowledge to study embodied cognition. To capture this cognitive conceptual structure, a set of spatio-temporal building blocks called image schemas was introduced by George Lakoff and Mark Johnson. Detecting image schemas in natural language can provide further insights into how we encode embodied experiences in our communication and potentially contribute to research on conceptual understanding and symbol grounding in cognitive systems. However, due to their abstract nature and lack of formalization they are difficult to detect in language. In this talk, I will first briefly introduce the general idea of image schemas, which has been presented to this group before by my co-author Maria Hedblom, and then proceed to machine learning techniques to extract them from multilingual text. Furthermore, I will provide a short overview on existing approaches and our ongoing work on formalizing image schemas.