The question of how words from different languages are represented and accessed in bilingual speakers is the focus of much debate in the psycholinguistic literature. In this thesis, we aimed to contribute to a better understanding of the processes that underlie naming and translation abilities in bilingual speakers. Our goals were 1) to clarify the role of cognateness across tasks, languages and populations in relation to interactive activation models of bilingual lexical production and 2) to evaluate models that posit direct lexical links between words in two languages by examining the extent of semantic involvement across tasks. These questions were studied by collecting converging evidence from younger and older neurologically healthy participants and from brain-damaged participants with word finding deficits. The key results are as follows: 1) In healthy participants, robust cognate facilitation effects were present across tasks, languages and age groups. 2)Cognate effects were stronger in translation than in naming in healthy participants. 3)In aphasic participants, a consistent cognate advantage was observed when naming pictures in the weaker language, but less so when naming pictures in the strongest language or in translation. 4) Treating words in one language generalised to cognate words, with some generalisation to untreated tasks. 5) Aphasic participants produced fewer semantic errors in translation than in naming. Overall, this study clarifies the role of cognateness in bilingual language production. It is the first to examine cognateness effects in a within-subject design, using the same stimuli across tasks and participant groups in an attempt to resolve some of the inconsistencies in prior research, which may be related to variations in experimental protocols across studies. It is also the first to use converging evidence from aphasia in an integrated study. Our findings support interactive activation models of the bilingual lexicon and dual-route models of translation.