Pre- and Post-head Processing for Single- and Double-Scrambled Sentences of a Head-Final Language as Measured by the Eye Tracking Method

Using the eye-tracking method, the present study depicted pre- and post-head processing for simple scrambled sentences of head-final languages. Three versions of simple Japanese active sentences with ditransitive verbs were used: namely, (1) $\text{ SO }_{\!1\!}\text{ O }_{\!2\!}\text{ V }$ canonical, (2) $\text{ SO }_{\!2\!}\text{ O }_{\!1\!}\text{ V }$ single-scrambled, and (3) $\text{ O }_{\!1\!}\text{ O }_{\!2\!}\text{ SV }$ double-scrambled order. First pass reading times indicated that the third noun phrase just before the verb in both single- and double-scrambled sentences required longer reading times compared to canonical sentences. Re-reading times (the sum of all fixations minus the first pass reading) showed that all noun phrases including the crucial phrase before the verb in double-scrambled sentences required longer re-reading times than those required for single-scrambled sentences; single-scrambled sentences had no difference from canonical ones. Therefore, a single filler-gap dependency can be resolved in pre-head anticipatory processing whereas two filler-gap dependencies require much greater cognitive loading than a single case. These two dependencies can be resolved in post-head processing using verb agreement information.