BACKGROUND: Around one-quarter of children attending language support services have difficulty in retrieving words. Therapy studies with such children have shown that both semantic and phonological techniques can improve word finding. A new approach to intervention is described using a computerized aid that converts letters into sound cues. AIMS: (1) To assess the word-finding abilities in a small series of children with word-finding difficulties. (2) To investigate whether intervention with a computerized aid can improve children's ability to find words. (3) To evaluate the effect of therapy on connected speech (using the Test of Word Finding in Discourse) and in relation to children's, parents' and other practitioners' views (using a questionnaire). (4) To understand the outcome in relation to an interactive model of speech production. METHODS & PROCEDURES: Five children (aged 6:10-10:7) were assessed on a variety of background measures. Following baseline assessment, two sets of items were included in therapy: a research set (with matched control items) and a further set selected for their functional relevance for each child. Intervention using the computerized aid took place once a week for half a term (6 weeks). OUTCOMES & RESULTS: The children differed from one another in their language development and non-verbal abilities, but all had difficulties in word retrieval. All children showed significant improvement in naming intervention items after therapy. This effect was maintained half a term later. Despite very different profiles, the children benefited similarly from the intervention, suggesting a common processing source for change. After the intervention, two children showed a significant reduction in word-finding behaviours (e.g. use of substitutions, repetitions and empty words) on the Test of Word Finding in Discourse. The views of participants, parents and professionals were varied but generally positive. CONCLUSIONS: The main effect of therapy can be understood, in relation to an interactive model of speech production, as strengthening links from meaning to form. Use of the aid in therapy improved word finding for targeted items and this occurred within a clinically realistic timeframe. The finding that the gains in word retrieval maintained suggests that the intervention can be functionally useful if target items are chosen appropriately. Measurable change for two children in connected speech suggests wider benefits can also occur. Overall, the aid can be a useful additional tool for therapists/teachers working with children with this difficulty.