This thesis examines the role of vision in language development by focusing on: first, the understanding blind infants have of objects, actions/events and the way they start to talk about these aspects of their environment; and, second, the ways visual information contributes to conceptual and lexical development in sighted infants. Until recently, research has predominantly focused on infants' understanding of objects and their understanding of actions/events has been neglected. Since individuals who are blind predominantly have access to temporal, rather than spatial infomation and so are better able to process information about actions and events rather than objects, this bias seems to have led to the conclusion that an absence of visual information results in a cognitive deficit. Six blind/severely visually impaired infants and their sighted controls were studied for around a year using a range of quasi-experimental, parental report and observational techniques. The studies found little difference between the blind and sighted infants in the age of onset or rate at which first words are produced. However, blind infants were found to be delayed in the age at which they were able to comprehend and produce labels for objects and they produced few words for concrete, discrete objects. The finding that the blind infants were able to categorize objects/actions as well as generalise and extend their words calls into question Dunlea's (1989) claim that an absence of visual information leads to a cognitive deficit. It is argued that blind infants can make their way into language using a route which is merely one end of a spectrum of routes used by sighted infants. Implications are discussed for theories of lexical development (multiroute model, developmental lexical principles framework and the social-pragmatic framework) as well as for possible strategies facilitate conceptual and lexical development in blind/SVI and sighted infants.