Object recognition in infancy: towards an embodied skeletal shape model of word identification

This paper presents results of observations of a small sample of four two-year old children as they were introduced to written word cards as named objects. This research sought to explore the notion that very young children can identify printed words as objects in a similar way to identifying other objects in their environment. Our findings suggest that reading, or word identification, at its most fundamental level is object recognition, and support theories of infant object recognition that utilise non-variant shape skeletons whereby individual features can be recognised. Additionally, our findings support the synchronised naming of objects as they are perceived by the infant as a natural learning mechanism. We propose that, aligned with theories of embodied cognition, children can learn to identify written words, where these words are within a child’s existing vocabulary, using their complete repertoire of experience from being in the world associated with the printed word.