Production-Comprehension Asymmetry in the Processing of Multiple Adjuncts in Korean and its Psycholinguistic Implications.

This study examined word-order preferences in Korean sentences containing temporal and locative adjuncts, focusing on the interplay between "TIME-before-PLACE (TbP)" and "LONG-before-SHORT (LbS)" preferences. While LbS is well established for Korean argument structures, its role in adjunct processing remained unclear. Experiment 1 investigated production via sentence elicitation with constituents of varying lengths. Results revealed a robust TbP preference (85%) dominating over LbS: 68% of sentences maintained TbP order even when locative adjuncts were lengthened-contrasting sharply with the 33% canonical ordering observed for lengthened arguments in prior research. This suggests that adjunct ordering is governed by semantic-structural principles rather than dependency minimization. Experiment 2 employed eye-tracking during reading. Neither TbP nor LbS independently affected reading times. However, a significant interaction emerged across go-past time, second-pass reading time, and total reading time. Processing difficulty increased when a longer constituent preceded a shorter one only under PLACE-before-TIME order; when TbP was maintained, SHORT-before-LONG (SbL) was preferred over LbS. Processing costs were greatest when both factors were simultaneously violated (Short PLACE-Long TIME condition). These findings demonstrate a clear production-comprehension asymmetry. In production, TbP dominance reflects semantic-structural organization prioritizing temporal framing before spatial specification. In comprehension, sensitivity emerges from the joint operation of two adjunct-specific principles: temporal adjuncts, as event-external discourse anchors, are processed most efficiently when sentence-initial, while locative adjuncts, as event-internal quasi-arguments, benefit from verb-proximal positioning. This contrasts with argument processing, where both modalities show consistent LbS preferences, revealing qualitative differences between argument and adjunct processing mechanisms.