How flexible is bilingual language control and how does it adapt to the linguistic context of a conversation? We address this by looking at the pattern of switch costs in contexts involving mostly the use of a dominant or non-dominant language. This linguistic context affected switching patterns: switching was equally costly for both languages in a dominant (L1) context, while switching was harder for the weaker language in the non-dominant (L2) context. Also, naming latencies for each language were affected by the linguistic contexts: only the dominant L1 context led to slower latencies for the dominant language. This latter finding was also present when looking at the LPC component, which may reveal differences in the way inhibitory control is applied depending on the linguistic context. These results reveal that the bilingual language control system is flexible and that it adapts to the linguistic context in which the speaker is placed.