On the processing of explanations in discourse: Evidence from eyetracking during reading

Torgrim Solstad
Berlin, Zentrum für Allgemeine Sprachwissenschaft

Interpreting causal statements in discourse requires not only the proper identification of causes and effects, but also the ability to judge the plausibility of the causal relation relative to a given context. Consider (1):

(1) Weil [Peter sich an die Aufbauanleitung hielt]_cause, [ging die Spülmaschine kaputt]_effect.
‘Because [Peter followed the assembly instructions]_cause, [the dishwasher broke down]_effect.’

In isolation, (1) seems anomalous. World knowledge predicts an effect to the contrary: Following the instructions usually prevents a device from breaking down. Given its incompatibility with world knowledge, the sequence in (1) can at most constitute a partial explanation (Halpern and Pearl 2005; Pearl 2009). If no additional cause is given, uttering (1) may be assumed to lead to rejection or abductive inferencing based on world knowledge (Aliseda 2006).

However, once (1) is embedded in a larger context, the anomaly may disappear. One can think of a number of situations in which (1) could make sense. Consider, for instance, the complex cause in (1′) consisting of the cause provided in (1), labeled cause2, and an additional cause1. This complex cause would seem to fully explain the effect in (1):

(1′) Weil [die Aufbauanleitung einen Fehler enthielt]_cause1 und [Peter sich an die Anleitung hielt]_cause2, [ging die Spülmaschine kaputt]_effect.
‘Because [the assembly instructions contained an error]_cause1 and [Peter followed the instructions]_cause2, [the dishwasher broke down]_effect.’

How can we model partial and full explanations, what are the inferential processes they trigger and how are they processed in discourse? The discussion of the first two of these questions will be accompanied by the presentation of evidence from an experiment involving eyetracking during reading, showing that world knowledge has an immediate effect on the processing of causal relations. Furthermore, we will also see that differences between two types of abductive reasoning patterns are reflected in different temporal profiles when inferring from partial to full explanations during text comprehension.

If you would like to meet the speaker, please contact Vera Demberg.

References:

Aliseda, Atocha (2006). Abductive Reasoning: Logical Investigations into Discovery and Explanation. Dordrecht: Springer.
Halpern, Joseph Y. and Judea Pearl (2005). Causes and Explanations: A Structural-model Approach. British Journal for the Philosophy of Science 56(4): 843-911.
Pearl, Judea (2009). Causality: Models, Reasoning, and Inference. Cambridge: Cambridge University Press.