Most patients with aphasia recover language function to some degree, but there is significant variability in the speed, nature, and extent of language recovery that occurs in the months following a stroke. Currently, there is no effective method available to predict how an individual with aphasia will recover. The most accurate predictors of recovery to date rely on global outcome measures which do not provide relevant information regarding recovery of symptoms that are targeted in therapy. The overall goal of this thesis was to examine whether language-related brain activity predicted recovery of specific symptoms in aphasia. The first aim was to gain a clearer understanding of the neural mechanisms which underpin word processing in healthy individuals.Brain activity associated with spoken word processing was investigated firstly in a group of healthy young adults, then in a healthy older cohort in order to ascertain whether task performance was preserved in ageing and if so, whether a preserved performance was underpinned by age-related changes in the neural mechanisms. The second main aim was to investigate the neural substrates which underpin language processing in the subacute (2-6 weeks) and chronic (6 months) phases of aphasia recovery and to determine whether the timing and extent to which these mechanisms are recruited contributes to a good or poor language outcome. An auditory lexical decision task with a novel pseudoword condition was employed to investigate language recovery mechanisms and was selected to ensure that patients with aphasia in the subacute stage would be able to perform the task with sufficient accuracy. The real word stimuli consisted of concrete and abstract items. The stimuli were presented in an event-related fMRI task to young healthy controls, older healthy controls, and patients with aphasia during the subacute and chronic phases post-stroke. A range of behavioural language measures was administered to all three cohorts and at both time points for patients with aphasia.In the healthy young group, concrete words elicited increased activity in a widely distributed network of brain regions including bilateral angular gyrus (AG), left posterior cingulate and left dorsolateral prefrontal cortex, compared to abstract words. Meanwhile, no region showed increased activity for abstract compared to concrete words, rather differences were only observed in bilateral AG when abstract words were contrasted with pseudowords. In the healthy older group, region of interest (ROI) results identified that a preserved performance was associated with an altered pattern of left hemisphere activity in the inferior and middle frontal regions, the AG and the fusiform gyrus.Differences between the healthy older and healthy young group were identified in key language regions, with the left inferior frontal gyrus (IFG) showing increased activity for abstract and pseudowords compared to concrete words while increased activity in the left AG was observed in the younger group only.ii In patients with a...