David Katten, Rice University
Predicting the ditransitive: An experimental study
In this paper I explore the role that information status (Chafe 1994) plays in online processing, specifically how the given-new contrast might be used to predict upcoming argument structure. An eyetracking experiment shows that listeners are able to anticipate not only upcoming thematic roles but argument structure based on which referents had been mentioned in the preceding discourse, as well as the verb’s bias toward one argument structure over another.
Altmann & Kamide (1999) have demonstrated that listeners in an experimental setting are able to identify direct objects faster when there is only one plausible referent available. Using eyetracking, they found that in sentences like “The child will eat the cake,” listeners began to look to the cake before the offset of the verb. For sentences like “The child will move the cake,” where there are several possible on-screen referents that are compatible with “move,” listeners waited to hear the direct object before looking at it. Because the selectional restrictiveness of the verb permitted listeners to identify a referent before they heard it, the phenomenon is called prediction.
The question I address is whether or not information status may serve as a similar cue for argument structure. The recipient argument in a ditransitive construction (see below) is given in discourse (Thompson 1995). Therefore, the parameter of given participant – theme or recipient – should help guide a listener to predict which participant will be in the postverbal slot.
· The teacher assigned the student a book. (ditransitive)
· The teacher assigned a book to the student. (dative)
Because of the serial nature of speech, the listener does not know (in real-time) which argument structure will follow the verb. As ditransitives always have previously mentioned recipients, it was hypothesized that not mentioning a possible recipient would block a ditransitive anticipation and prime the dative. This would be evident if looks to the theme increased over the verb and looks to the recipient decreased (when there was no previously mentioned theme). In selecting the verbs to be used, I chose 10 verbs from a cline of lexical bias ranging from 5% to 50% ditransitive, as measured by Argaman (2003). The experiment tested if these two factors could influence online processing during the verb – (1) whether the theme or recipient was previously mentioned and (2) the overall lexical bias of the verb toward either construction. Two example stimuli follow.
-
A baker is working in his store.
-
A customer walks in to meet him.
-
Now, he will bake a cake for her / Now he will bake her a cake
-
Please click on the cake.
In analyzing the data, I focused on fixation proportions (i.e. what the listeners were looking at on average) only during the verb. Results showed that listeners were able to identify the argument that immediately followed the verb only when the verb was biased toward the ditransitive and the recipient was given. However, there was a marginal effect when the verb was biased toward the dative and the theme was given. The hypothesis regarding cases where the recipient was not previously mentioned priming the dative was not borne out.
This suggests that prediction is manifested differently in discourse situations than in isolated sentences. It also shows that listeners are able to use the information status of an argument as a cue for determining which argument structure is upcoming. This finding has repercussions for not only sentence processing, but for grammar and syntax as well. These results show that grammatical theories, while also accounting for the semantics of argument structure, must also attend to discourse status as a productive influence on “core” grammar.
References
Altmann, Gerry & Yuki Kamide. “Incremental interpretation at verbs: restricting the domain of subsequent reference.” Cognition 73. pp. 247-64. 1999.
Argaman, Vered. Lexical semantics and argument structure in sentence processing. Ph.D. Dissertation: Northeastern University. 2003.
Chafe, Wallace. Discourse, consciousness, and time. Chicago: University of Chicago Press. 1994.
Thompson, Sandra. “The iconicity of ‘dative shift’ in English: Considerations from information flow in discourse.” In M. Landsberg (ed.) Syntactic Iconicity and Linguistic Freezes: The Human Dimension. Pp. 155-75. 1995.
