This study examines sentence-final expressions in a corpus of 8,232 Korean and Japanese newspaper editorial headlines drawn from six major daily newspapers, three in each language, published between February 2022 and April 2025. It proposes a dual-classification framework that combines Layer 1, consisting of discourse functions (deontic, descriptive, assertive, interrogative, and directional), with Layer 2, consisting of morphosyntactic realization types. The overall distribution differed significantly between the two languages (χ²(4)=454.54, p<.001, Cramér's V=0.235). Japanese headlines were heavily concentrated in the deontic function (47.2%), whereas Korean headlines were distributed more evenly across the deontic (37.9%), descriptive (24.6%), and assertive (21.6%) functions.
Correspondence analysis revealed that the cross-linguistic dimension accounted for 83.7% of the total inertia, confirming that differences between the two languages outweigh variation among individual newspapers within each language. A sensitivity analysis that collapsed the five categories into four, and then into three, confirmed that this cross-linguistic difference does not depend on the particular category scheme used (χ²=271.28, Cramér's V=0.182 for the four-category scheme; V=0.181 for the three-category scheme). At Layer 2, the pattern of morphosyntactic concentration and dispersion was reversed depending on the function: Korean converged on a single form in the deontic function while Japanese showed greater dispersion, whereas the opposite pattern appeared in the interrogative function. These findings indicate that the asymmetry in sentence-final expressions between the two languages is not unidirectional but function-dependent, reflecting the degree of internal differentiation within each grammatical subsystem.