Abstract
Kazakh is a Turkic language with agglutinative morphology and relatively free word order. This feature makes it ideal for studying how syntax and semantics interact in language processing. Unlike languages with fixed word order, Kazakh signals semantic roles mainly through case affixes, which challenges traditional grammatical models. To investigate the interaction between syntax and semantics in Kazakh, we combined corpus-based semantic annotation with neurophysiological data. Our corpus comprises 1,200 sentences from classical Kazakh literature by Abai, Zhumabaev, and Auezov, annotated using the UCCA and PropBank frameworks, while metaphors were identified via the Metaphor Identification Procedure VU (MIPVU). Additionally, we performed a meta-analysis of 15 event-related potential (ERP) and fMRI studies on Turkic languages conducted between 2010 and 2024 to support our findings. Results show that approximately 98.3% of semantic roles (e.g., agent, patient) remain identifiable across varied word orders, demonstrating strong semantic stability despite syntactic variation. Based on these findings, we propose the Cognitive–Semantic Matching Model (CSMM), a generative-cognitive framework in which grammatical affixes and conceptual metaphors work together to support comprehension. This framework integrates generative syntax with cognitive semantics and offers insights relevant to linguistic theory, cognitive neuroscience, and natural language processing for agglutinative languages.