Summary: | We describe two experiments investigating the comprehension of different types of Wh-questions in neurotypical adults (Experiment 1) and adults with Broca’s aphasia (Experiment 2). Consider as examples:
Two mailmen and a fireman got into a fight yesterday afternoon.
1a. Who pushed the fireman yesterday afternoon? – Subject-extracted Who
1b. Who did the fireman push ___ yesterday afternoon? – Object-extracted Who
2a. Which mailman pushed the fireman yesterday afternoon? – Subject-extracted Which
2b. Which mailman did the fireman push ___ yesterday afternoon? – Object-extracted Which
There is evidence from the linguistic and psycholinguistic literatures that suggest Which-questions are more difficult to understand than Who/What-questions and within those, that object-extracted are more difficult than subject-extracted. We used a unique eye tracking-while listening method where listeners were presented with sentences like (1) and (2) above while gazing at a three-figure picture (e.g., a picture of a mailman pushing a fireman who is pushing another mailman); we measured gazes to the referents in the pictures across the time-course of the sentences, and also collected accuracy and response time data to answer the questions (by button press). We examined four specific hypotheses: Discourse, Memory Retrieval, Word Order, and Intervener. The Discourse hypothesis suggests that Which-questions should be more difficult to process than Who-questions because the former is required to refer to an individual taken from a set of entities previously mentioned in the discourse (Donkers & Stowe, 2006; Shapiro, 2000). The Memory Retrieval hypothesis makes the opposite claim; Which-questions, unlike Who-questions, contain specific information in the Wh-phrase that should speed memory retrieval (Hofmeister, 2007). The Word Order hypothesis suggests that, regardless of question type (Which or Who), object-extracted questions should be more difficult to understand than subject-extracted questions because the former are in non-canonical word order. Finally, the Intervener hypothesis suggests that only object-extracted Which-questions should be problematic, particularly for those participants with language disorders (e.g., Friedmann & Novogrodsky, 2011). An intervener is an NP that has similar properties to other NPs in the sentence, and thus results in similarity-based interference. Only object-extracted Which-questions contain an intervener (e.g., the fireman in (2b), which interferes with the chain consisting of the displaced Which-phrase, Which mailman, and its direct object gap occurring after the verb).
Briefly here, only the Intervener Hypothesis was supported by our rich data set, and this was observed unambiguously for our participants with Broca’s aphasia. As an example (see Figure 1), we observed significantly greater proportion of gazes to the incorrect referent (i.e., the intervening NP) in the object-extracted Which- relative to Who-questions beginning in the Verb-gap time window and extending throughout the remainder of the sentence and into the response period following the sentence. These patterns indicate lasting similarity-based interference effects during real-time sentence processing. The implications of our findings to extant accounts of sentence processing disruptions will be discussed, including accounts that root sentence comprehension impairments to memory-based interference.
|