The availability of sloppy interpretation for null arguments differs across languages. This difference is a challenge for child learners because they are unlikely to receive sufficient input that provides clear evidence about the available interpretation. Theories suggest that knowing the presence/absence of agreement could help solving this learning problem. While languages like Japanese that lack agreement allow argument ellipsis (hence the sloppy reading is available), languages with a rich agreement system like Spanish do not. This study explores the utility of this correlation as a cue for learners to infer the available interpretation of a null argument. We show that Japanese adults who learned semi-artificial Japanese that has object-verb agreement are more likely to accept the strict reading than the ones who learned only an artificial singular/plural marker attached to an object. Our results provide evidence that agreement may play a role in learning the interpretation of null arguments.

Previous experiments have demonstrated that Japanese children can use the number of arguments and the case markers tolearn novel verbs. However, these cues are mostly omitted in child-directed speech. We revisit this gap between the abilityof children to use syntactic cues and the deficiency of such input by examining a different mode of input in the form ofpicture books. We built a Japanese picture book predicate-argument structure corpus containing annotations of predicate-argument structure and non-linguistic information. The analyses show that Japanese picture books contain more overtarguments and accusative case markers, and that these cues have significant influence on the prediction of verb transitivity.In addition, this study demonstrates that non-linguistic information (animacy and the numbers of potential referents) couldhelp predict transitivity if learners are able to use these cues to infer the presence of null arguments.