Wittenberg, E. 1 & Freywald, U. 2
1 University of California, San Diego
2 Universität Potsdam
The semantics of clausal connectors is often ambiguous: For example, 'although' can be concessive (p despite q: [I?ll take another coffee]p although [I already had one]q) or corrective (not p, because q: [I?ll take another coffee.]p Although, [I already had one]q). How do comprehenders know which meaning was intended?
One hypothesis is that the standard interpretation (concessive) is more likely when the connected clause is syntactically dependent (matrix-subordinate clause structure); another, that world knowledge determines interpretation. We tested this in two studies, using German, which allows for manipulating two factors usually associated with syntactic independence: verb placement (verb-last vs. verb-second, indicating subordinate vs. main clause syntax), and phonological separation, operationalized via punctuation (comma vs. full stop, representing non-final vs. final boundary tones).
384 participants saw 48 biclausal statements with 4 connectors, and judged the likelihood of p. Experiment 2 differed from Experiment 1 in that all content words were replaced by nonsense words, to eliminate the effect of world knowledge.
Results confirm the expected effect of syntactic dependence, with stronger effect sizes in Experiment 2, but no significant difference between experiments. This suggests that speakers use cues of syntax and prosody to draw inferences on unusual word meanings.