Negotiating the convex decreasing relationship: how clause type and register moderate the syntax-semantics interface in complex sentences

This study investigates the nonlinear relationship between syntactic complexity, measured as Mean Dependency Distance (MDD), and Semantic Density (SD) in English complex sentences. Drawing on a corpus of 4,346 sentences from the Brown Corpus and operationalizing SD through distributional semantics, we identified a convex decreasing trajectory characterized by a decline in semantic density as syntactic complexity increases, but at a diminishing rate that eventually reverses into a slight upturn at extreme MDD values. Clause type differentially conditions this relationship. Nominal clauses exhibit the strongest linear association, consistent with their role as primary carriers of dense propositional content, whereas relative clauses display distinct nonlinear saturation effects. Furthermore, the relationship varies by register: the Learned and Press categories exhibit significant curvilinear patterns that persist after controlling for sentence length, while in Fiction the effect largely dissipates. These findings challenge purely linear models of syntactic-semantic mapping and suggest that the interaction between structural elaboration and meaning concentration is conditional upon both grammatical context and communicative function.