We investigated the effect of incorporating information about proband and parental structural language phenotypes into linkage analyses in the two regions for which we found the highest signals in our first-stage affected sibling pair genome screen: chromosomes 13q and 7q. We were particularly interested in following up on our chromosome 7q finding in light of two prior reports of linkage of this region to developmental language disorder, since one of the diagnostic criteria for autism is absent or abnormal language development. We hypothesized that if the language phenotype were genetically relevant to linkage at the chromosome 7q locus, then incorporating parents phenotypes would increase the signal at that locus, and most of the signal would originate from the subset of families in which both probands had severe language delay. The results support these hypotheses. The linkage signals we obtained on chromosome 7q as well as at least one signal on chromosome 13q are mainly attributable to the subgroup of families in which both probands had language delay. This became apparent only when the parents' history of language-related difficulties was also incorporated into the analyses. Although based on our data, we were not able to distinguish between epistasis or heterogeneity models, we tentatively concluded that there may be more than one autism susceptibility locus related to language development.