2018 Sat Session A 0900

Saturday, November 3, 2018 | Session A, East Balcony | 9am

The interaction of word order, case-marking, and verbal morphology in children’s comprehension of suffixal passives in Korean
G. Shin, K. Deen

For children to comprehend passives in languages like Korean, they draw on knowledge of word order, case-marking, and verbal morphology (Table 1). In this study, we systematically obscure cues from these sources and measure comprehension of canonical and scrambled passives. We find that children initially rely on a comprehension strategy that expects the first noun in a sentence to be an agent and to be nominative-marked. This parsing preference (akin to Huang et al., 2013), we argue, is the source of difficulty with passives in languages like Korean, and perhaps beyond.

Actives in Korean typically occur with the agent first, case-marked nominative, and the theme second, case-marked accusative. A verb carries no dedicated active morphology per se. Suffixal passives in Korean occur with the theme initial, case-marked nominative, the agent occurs second, case-marked dative; the verb carries dedicated passive morphology. Pre- verbal elements in both patterns can also be scrambled. Thus, passives pose a particular challenge in Korean since there are three cues to be integrated in order for a comprehender to correctly identify a pattern as passive: word order, case-marking, and verbal morphology.

We manipulated these three cues in Experiments 1-3 through picture selection tasks. We specifically manipulated test sentences by obscuring parts of the sentences (Experiments 2-3; Table 2). Experiment 1 (baseline, 30 3-4yr-olds, mean=4;1, 19 5-6yr-olds, mean=6;1; 20 adult controls) compared canonical and scrambled passives with nothing obscured.

Participants heard a sentence and matched it to its corresponding picture. Experiment 2 (28 3-4yr-olds, mean=4;13, 18 5-6yr-olds, mean=6;14; 20 adult controls) assessed word order effects on sentence comprehension by obscuring case-marking and verbal morphology.

Participants were told that the speaker of test sentences was hungry and was eating food, chewing at various points in the sentences. In doing so, yum-yum sounds strategically occurred to obscure case markers and verbs. Participants thus only heard two nouns in between yum-yum sounds, with word order as the only cue. Experiment 3 (30 3-4yr-olds, mean=4;1, 19 5-6yr-olds, mean=6;1; 20 adult controls) measured the effect of case-marking on children’s understanding of passives by obscuring verbal morphology: the speaker was sick and coughed at every verb, masking the verb, but leaving case markers intact.

Results. Experiment 1: 3-4yr-olds performed at chance on canonical and scrambled passives whereas 5-6yr-olds showed accuracy significantly above chance in both conditions.

Experiment 2: 3-4yr-olds continued to show chance performance, but 5-6yr-olds’ selection of agent-first pictures reveals the development of a strong agent-first preference. Experiment 3: both child groups significantly preferred the reading of sentences where the first noun was the agent, showing that case-marking alone is insufficient to sway children from their agent- first preference.

Discussion. The results show that, whereas children do employ verbal morphology to inform their interpretation of passives (comparing the results from Experiments 1 and 3), they still have a strong preference for interpreting the first noun as an agent. We conclude that this strong preference, coupled with the difficulty of integrating three cues online, conspire to pose difficulties for the comprehension of passives for children.