This paper argues that some of the patterns seen in aphasia may reflect difficulties in the cognitive preparations for language. In particular, some individuals might be unable to carry out processes of 'Thinking for Speaking' (Slobin 1996), which frame thoughts for language production. Evidence to support this proposal is presented, together with signs that such thinking can be assisted with cues and therapy. It is argued that these preliminary data need to be pursued via a more comprehensive investigation of thinking therapy.