Figure 3: Accommodation study, Tiemann et al. 2011
The previous study on verification did not address what happens if a comprehender can neither verify nor falsify the presupposition in the preceding context. This next study compares the verify and falsify contexts with a neutral context. Tiemann et al.'s prediction had been that, if the look-up/fail-to-find/add sequence is harder than the look-up/check process, sentences whose presupposition must be accommodated should yield the most difficulty. There were three conditions: the two from the Verification study plus an additional neutral context:
Context sentence variants:
Context sentence: verify
Susanne hat dieses Jahr bereits rote Handschuhe gekauft.
Susanne had this year already red gloves bought.
Susanne had already bought red gloves this year.'
Context sentence: falsify
Susanne hat bisher nie rote Handschuhe gekauft.
Susanne had until now never red gloves bought.
Susanne had never bought red gloves until now.'
Context sentence: neutral
Inge hat bisher nie rote Handschuhe gekauft.
Inge had until now never red gloves bought.
Inge had never bought red gloves until now.'
Target sentence (always the same):
Target sentence: trigger
Heute hat Susanne wieder rote Handschuhe gekauft und sie gleich angezogen.
Today has Susanne again red gloves bought and then immediately put them on.
'Today, Susanne bought red gloves again and put them on right away.'
The acceptability ratings, shown in Figure 3 below, did not show the predicted pattern. The neutral condition, which had been predicted to be the most difficult, received better ratings than the falsify condition. The verifying condition did, however, receive the highest ratings, as predicted.
Figure 3: Accommodation study, Tiemann et al. 2011
The reading time results were more in line with the predictions: Considering the reading times for the whole sentence (which includes the post-critical region), the neutral condition yielded longer reading times than the falsifying condition, with the verifying condition in between.
To sum up, Study 1 (with non-triggers being easier than triggers which are easier than unacceptable words) can be taken to show that triggers signal to the comprehender that work will be required to confirm that the context contains the presupposition. At the critical word, where verification is required, Study 2 confirmed that comprehenders slow down at the critical word, taking longer if the context does not contain the presupposition. Lastly, the post-critical region was not as clear-cut: The ratings showed that accommodating was actually easier than falsifying, whereas the reading times showed that accommodating is harder than falsifying. Puzzles like these should be treated as invitations to do follow-ups (e.g., dissertation projects!).
What to take from these studies? The first point is simply methodological: that researchers have options in how they assess language processing (offline vs. online) and how they design experiments (comparing apples and oranges vs apples and apples). Secondly, there was variation across triggers: some yielded stronger effects (e.g., the verifying/falsifying conditions differ more for the additive particle "again" than the definite "his"; see Figure 2). And lastly, apropos the question of modularity that was raised at the beginning of this section, there is the following key point:
KEY POINT: Tiemann et al.'s study shows that comprehenders don't wait to do pragmatics.
Tiemann, S., Schmid, M., Bade, N., Rolke, B., Hertrich, I., Ackerman, H., Knapp, J., & Beck, S. (2011). In Reich, Ingo et al. (eds.), Proceedings of Sinn & Bedeutung 15, pp. 581–595. Universaar – Saarland University Press: Saarbrücken, Germany, 2011.