Ralf Rummer,1 Johannes Engelkamp1 & Lars Konieczny2
1 Saarland University, 2 University of Freiburg
In a series of four experiments, we investigated the role of subordination in retaining and processing causal sentences (1-2).
Weil er Hans begegnet war, rief er Ralf an.
Because he Hans met had, called he Ralf up.
"Because he met Hans, he called Ralf."
Er war Hans begegnet, deshalb rief er Ralf an.
He had Hans met, therefore called he Ralf up.
"He met Hans, therefore he called Ralf."
In Experiment 1, we demonstrated that immediate recall performance for auditorily presented subordinate causal sentences (two sentences in each trial) is better than for co-ordinate sentences. Furthermore, subordinate structures showed no difference between recall of clause 1 and clause 2, whereas recall of clause 1 was much better than for clause 2 in coordinate constructions.
This subordination effect can be explained by the assumption that subordinate sentences are represented more hierarchically than co-ordinate sentences. Therefore, chunking and thus retaining the content of subordinate clauses is easier than for co-ordinate clauses. There are, however, three other possible explanations which are addressed in the subsequent experiments.
Experiment 2 tested whether the subordination effect is due to the different prosodic structures of subordinate and co-ordinate sentences. To address this question, we presented our subjects with the same sentences as in Experiment 1. However, the sentences were auditorily presented, but without prosodic information. If the prosody hypothesis holds true, the subordination effect should disappear under this condition. In contrast to this assumption, the results were the same as in the first experiment.
Some models of syntactic complexity (e.g., Gibson, 1998) predict that processing German subordinate sentences should be harder than processing co-ordinate structures. Better recall of subordinates could thus be due to information being encoded more thoroughly here. This explanation was addressed in a self-paced reading experiment (exp 3) using the same sentences as in the two preceding experiments. Although the subordination effect could be replicated in recall performance, reading times were even longer for co-ordinate structures. These results clearly exclude the mental effort hypothesis.
The last alternative explanation attributes memory differences to the fact that the critical causal conjunction word was presented earlier in subordinate than in co-ordinate sentences. Experiment 4 tested this explanation by changing the position of the causal conjunction within the sentence. It was demonstrated that the subordination effect is unaffected by the position on which the causal information is provided. In sum, our data support the assumption that syntactic subordination facilitates both processing and memory, as reflected by reading time and memory performance.