Existing accounts of language in autism have tended to focus on English and have rarely progressed beyond broad characterizations of pragmatic difficulty. This volume changes both. Drawing on a clinically defined corpus of Japanese-speaking individuals with ASD and combining Systemic Functional Linguistics with machine learning, it identifies 46 lexicogrammatical discriminators of ASD and non-ASD discourse, with 80% classification accuracy. Thirty-eight of these discriminators are interpreted across twelve cognitive-functional domains, including weak central coherence, joint attention, predictive processing, executive function, and working memory, revealing how recurrent grammatical choices relate to neurocognitive profiles. Analyses span a broad range of lexicogrammatical categories, including clause complexes, process types, appraisal, modality, evidentiality, negotiating particles, and sound-symbolic expression. The result is the first book-length, empirically grounded account of lexicogrammatical variability in ASD, opening new directions for diagnostic assessment and language intervention. Essential reading for researchers and clinicians in autism studies, human neuroscience, clinical linguistics, and neurodevelopmental disorders.