This research brief argues that existing autism diagnostic and phenotypic instruments contain language-relevant evidence that is often dispersed across deficit-coded domains — echolalia, pragmatic impairment, adaptive communication, developmental delay — without being recognised as a coherent pattern. Rather than treating Gestalt Language Processing (GLP) as requiring wholly new data collection, the brief proposes a construct-validity agenda for re-reading item-level variables within established autism datasets, including SPARK, for GLP-proxy signals. Published as part of the Towcester Abbey Research Brief / Occasional Paper Series.