Britta Stolterfoht,1 Anja Hahne,1 Angela D. Friederici1 & Markus Bader2
1 Max Planck Institute of Cognitive Neuroscience, Leipzig & 2 University of Massachusetts at Amherst
German, in contrast to English, has a relatively free word order. For instance, the first determiner phrase (DP) after the complementizer can be either the subject or the object of the complement clause as a result of case ambiguity (nominative vs. accusative). Sentences can be constructed which are ambiguous until the number information of the finite verb appears:
| (1) | a. | Maria hat gesagt, | daß | [die Mutter(nom) | die Kinder(acc) | beschaeftigt hat]F |
| Mary has said, | that | the mother | the children | occupied has | ||
| "Mary said that the mother occupied the children." |
||||||
| (1) | b. | Maria hat gesagt, | daß | die Mutter(i,acc) | [die KINDER(nom) ]F [ti] | beschaeftigt haben. |
| Mary has said, | that | the mother | the children | occupied have | ||
|
"Mary said that the children occupied the mother." |
||||||
| (2) | a. | Maria hat gesagt, | daß | [sie(nom) | die Kinder(acc) | beschaeftigt hat]F |
| Mary has said, | that | she | the children | occupied has | ||
|
"Mary said that she occupied the children." |
||||||
| b. | Maria hat gesagt, | daß | [sie(i,acc) | die Kinder (nom) [ti] | beschaeftigt haben]F | |
| Mary has said, that she | the children | occupied have | ||||
|
"Mary said that the children occupied her." |
||||||
The object-first order in (1b) and (2b) results not only in a more complex syntactic structure, but in (1b) also in a change of focus structure accompanied by a prosodic change. Results of reading experiments, investigating these ambiguities, showed clear garden path effects for sentences like (1b) (Bader & Meng, 1999). These effects were clearly smaller in sentences with a pronominal object like (2b).
It is assumed that a prosodic contour is projected during silent reading which may influence or interact with syntactic processing (Implicit Prosody Hypothesis, Fodor 2001). Thus, one reason for the processing difference in scrambled sentences with pronouns in contrast to referential DPs could be, apart from the necessary syntactic reanalysis in both sentence types, the additional revision of the prosodic structure.
To investigate this hypothesis, we carried out two reading experiments using event-related brain potentials (ERP) to differentiate between processes of syntactic reanalysis on the one hand and the revision of the prosodic structure on the other hand. The data analysis showed two different components in Experiment 1: An early positivity (300-400 ms) which is interpreted as reanalysis of the syntactic structure, and a right anterior negativity (400-600 ms) which might be seen as the correlate of a prosodic revision process.
In Experiment 2, we eliminated the difference in the prosodic structure between the two sentence types by insertion of a focus particle in front of the second DP, thus preventing the parser from a prosodic structure revision. If the right anterior negativity in Experiment 1 is correctly correlated with the revision of the prosodic structure, it should disappear in Experiment 2. As predicted, there was no right anterior negativity, whereas the early positivity (300-400 ms) was still elicited and followed by a later positive component (500-900 ms). We are currently running a third experiment to clarify the appearance of this late syntactic process in the second experiment. Taken together, our results show that prosodic and syntactic processes can be differentiated in the ERP. Furthermore, the results support the assumption that implicit prosody plays a crucial role during silent reading.
References
Bader, M., & Meng, M. (1999). Subject-object ambiguities in German embedded clauses: An across-the-board comparison. Journal of Psycholinguistic Research, 28 (2), 121-143.
Fodor, J. D. (2001). Prosodic disambiguation in silent reading. Paper presented at the conference 'Prosody in Processing', Utrecht.