My research deals with the neural mechanisms responsible for understanding and producing language, with an emphasis on multiple partially redundant pathways. The study of multiple pathways for information processing is essential to future developments in stroke rehabilitation, as functional and structural assessments can be made of an individual's capacity to exploit spared pathways to recover cognitive and linguistic abilities. Behavioural intervention strategies could be tailored to an individual's post-stroke neuroanatomical status for maximal effect. My work has explored the potential of magnetoencephalography as a mapping tool in neurolinguistics, providing the spatial and temporal resolution necessary to measure the involvement of specific neural pathways on a time scale relevant to everyday language use. More recent work aims at a closer integration between the fields of cognitive neuroscience and computational linguistics. Assessment of language function in neurological patients has traditionally depended on behavioural tests administered by a trained professional under somewhat artificial conditions. Although these tests provide invaluable information, they do not capture more subtle aspects of variability in language use, especially those associated with milder forms of impairment as present in dementia and traumatic brain injury patients. Recently numerous tools have become available for automated analysis of naturalistic language samples that are easily obtained from patients. Studies have revealed quantifiable aspects of language use associated with normal aging and neurological impairment, but the relationship of these factors to neuroanatomical pathways remains to be explored. Current studies in my lab will develop the potential of computational linguistics tools to contribute towards improved diagnosis and treatment of cognitive impairment in stroke and progressive disorders.