Abstract for Irina Sekerina's talk

Eye movements of agrammatic aphasics in processing of non-canonical sentences in German
Irina Sekerina (College of Staten Island & Graduate Center, CUNY)
December 2, 2008 (Tuesday)
6:30 PM - 8:00 PM; Room 7102, CUNY Graduate Center

Broca aphasics are often at chance at comprehension of reversible non-canonical sentences (Grodzinsky 1995, Burchert et al. 2003). The Trace Deletion Hypothesis (TDH, Grodzinsky 1995) argues that patients’ errors are caused by an agent-first heuristic in assigning thematic roles to NPs. However, in an eye-tracking study, Dickey et al. (2007) found that it wasn't always the case: Aphasics exhibited normal-like online processing of English wh-questions in the correct trials but their eye movements were deviant in the wrong ones.

We conducted a sentence-picture-matching experiment with 8 controls and 7 German Broca aphasics using German canonical SVO sentences (1) and non-canonical OVS sentences (2).

(1) Der Sohn fängt den Vater. 'The son is catching the father. '
(2) Den Vater fängt der Sohn. Lit. 'The father, the son is catching.'

Two pictures, the correct one (agent and patient acting in accordance to the sentence) and the foil (semantically reversed action), were presented side-by-side simultaneously with the spoken sentence. Accuracy, RTs, and eye-movements were recorded and analyzed using mixed effects models. We investigated the critical theoretical question: Do aphasic patients process non-canonical OVS sentences differently from controls?

Results: For accuracy, controls were at ceiling for both conditions (98% for SVO, 95% for OVS) while patients were impaired for SVO (80%) and at chance for OVS (46%). Patients were twice as slow as controls but SVO was processed faster than OVS in both groups. Controls' eye movements reflected a preference for the correct picture from the der/den-NP region onward in both conditions. Patients' fixation patterns in the canonical SVO condition were very similar to the controls. When we analyzed all non-canonical OVS trials -- correct and wrong ones -- together, as is typical in aphasia studies, we observed a persistent preference for the foil picture suggestive of the application of an agent-first heuristic. However, when following Dickey et al. (2007), we looked at the correct OVS trials separately, we found that patients, just like controls, showed an early preference for the correct picture. Only in the incorrectly answered trial did we find the difference between the 2 groups. Controls initially processed the OVS sentences as if they were SVO, with the case information temporarily ignored, but very soon thereafter they engaged in reanalysis at the verb that was successfully completed during the subject NP. In contrast, patients processed the OVS sentence deterministically. Once they adopted the incorrect analysis of the object NP as the subject, they failed to succeed in reanalysis. This resembles the difference found between adult and child processing of sentences with PP-attachment ambiguities in English (Trueswell et al., 1999), and can be taken as new evidence for the slowed processing hypothesis in aphasia (Dickey et al., 2007).

[The research reported in talk presentation was undertaken in collaboration with Sandra Hanne, Shravan Vasishth, Frank Burchert, and Ria De Bleser (Universität Potsdam).]