Autism spectrum disorder is a heritable neurodevelopmental condition that displays heterogeneity in both presentation and etiology and it often presents with concomitant communication difficulties. The hypothesis behind the New Jersey Language and Autism Genetic Study is that genetic heterogeneity for component phenotypes of autism spectrum disorder may be reduced relative to the disorder as a whole. We previously published an initial phase of this study with family recruitment that used very restricted inclusion/exclusion criteria for both autism and language deficits in other family members. Here we present an expanded sample that includes a wider range of phenotypic presentations in the autism and language domains. We found that our previous findings on 15q and 16q, connecting autism spectrum disorder and oral/written communication, are only relevant to the narrow autism spectrum disorder and language impairment phenotypes (with posterior probability of linkage of 57% and 33%, respectively) though addition of families did reduce both critical regions. After variant prioritization and additional filtration based on segregation patterns and functional annotations, we determined a set of 10 and 6 top candidate risk genes with strong association to language impairment and reading impairment, respectively. The top candidate genes include both genes previously implicated in neurodevelopmental disorders (e.g.,