There is substantial previous research which suggests alignment, or co-ordination, on many levels between two partners in a dialogue. This report examines alignment in language in relation to participants’ scores on the Autism-Spectrum Quotient, a questionnaire which indicates degrees of autistic traits. We report an experiment investigating two language tasks, using a confederate scripting design, in order to determine if score on the AQ affected participants’ alignment. We found that score on the AQ had a significant effect on lexical alignment, with participants who scored highly aligning less with a partner. A marginal effect, which was just below significance, was found between the AQ and alignment on the syntactic task, which suggests that in this task participants who scored highly on the AQ also aligned less syntactically. From these results we conclude that alignment in language is governed by a mediated account and that participants in a dialogue take beliefs about a partner into account when choosing which language to use. We specifically promote a social affect theory behind alignment, with participants choosing words in order to be socially appropriate and to increase positive feeling between them and their partner.