Genetic factors contribute significantly to autism susceptibility, but the heterogeneity of autism and autism spectrum disorders (ASD) poses a challenge for genetic studies. Research from several groups including our own suggests that studying cognitive or behavioral components of autism, or endophenotypes may aid in identification of more homogeneous subgroups and hasten the identification of genetic loci underlying this condition. We have hypothesized that speech and language related endophenotypes may underlie several of the suggestive or probable autism linkage peaks and worked in the AGRE sample to identify one major and several minor linkage peaks related to speech and language delay in autistic siblings in this large autism cohort (AGRE). This work was performed using retrospective medical and family history questionnaire data, or ADI-R data. Thus, although these findings represent a very promising first step, they rely on relatively basic phenotypic assessment that could be significantly enhanced by prospective language assessments and approaches based in modern cognitive neuroscience. Here we take this work to its next logical step in a synergistic fashion, by refining language related endophenotypes that contribute to autism susceptibility using prospective data and subjects from the infant sibling project (Project 1) in this ACE center. In addition we will use subjects and data collected in Projects 3 and 5 to relate genetic risk variants to specific brain circuits probed by functional imaging. First, we will perform independent confirmation of the speech and language related loci identified in AGRE using the prospectively collected cohort of 280 infant-sib/ autistic child pairs studied for language and social communication development in Project 1. This provides the advantage of testing the language-related loci in an independent sample that is in which receptive and expressive language and speech assessment will be directly measured, rather than depending entirely on retrospectively collected historical data. In addition, the availability of data on both receptive and expressive language will allow significant refinement of the specific language phenotype underlying these loci. Multivariate analysis will be used to explore the relationships among social communication, language, and outcomes of the children in Project 1, and to explore the extent to which identified genetic loci have pleiotropic effects. A highly focused study of candidate gene association will be conducted using SNP genotyping and family based association methods. This work will initially focus on CNTNAP2, in which we have identified association under the 7q language QTL in ASD families; the infant sibling cohort here will be used for confirming and refining this association. In addition, we will assess the impact of common variants in CNTNAP2 on several language related fMRI tasks, connecting specific genetic risk variants to specific underlying brain circuitry that is affected in ASD for the first time.