The nature of the lexical selection process in bilingual spoken word production is one of the pending questions of research on bilingualism. According to one view this competitive process is language-specific, while another holds that it is language-nonspecific (i.e., lexical competition is cross-linguistic). In recent years, research on bilingual language production has seen the rise of a third view that postulates that lexical selection is in fact dynamic and may function as language-specific or nonspecific depending on a number of factors. The aim of the present study was to investigate the lexical selection process among moderately proficient bilinguals whose two languages are typologically distant: Tunisian Arabic and French. The picture-word interference task was used in two experiments where moderately proficient Tunisian Arabic (L1)-French (L2) bilinguals were asked to name pictures in their L2 while ignoring auditory distractors (semantic, phono-translation, phonological, or unrelated) in their L2 (Experiment 1) or their L1 (Experiment 2). Thus, the language context was entirely monolingual in Experiment 1 and bilingual in Experiment 2. In Experiment 1, only a phonological facilitation effect was observed. In Experiment 2, interference was found in the phono-translation, semantic, and phonological conditions. Taken together, these results indicate that cross-language competition occurs among moderately proficient Tunisian Arabic-French bilinguals only in a bilingual context (Experiment 2) as indexed by the phono-translation interference effect observed. Our findings are in line with the recent hypothesis that lexical selection is a dynamic process modulated by factors like language similarity, language proficiency, and the experimental language context.