David S. Race
University of Southern California (BA - Psychology)
University of Wisconsin-Madison (PhD - Experimental Psychology)
-Sentence Production & Comprehension
Most recent publication
Race, D., Ochfeld, E., & Hillis, A.E. (in press). Lesion Analysis of Cortical Regions Associated with the Comprehension of Nonreversible and Reversible Yes/No Questions. Neuropsychologia.
Neurobiology of language Conference 2010
An on-line investigation of lexical and syntactic processing during sentence comprehension: Monitoring eye-movements in a sentence-picture matching task Race, D. and Hillis, A. Johns Hopkins University School of Medicine, Department of Neurology, Baltimore, MD, USA.
In the present experiment we measured eye-movements during a sentence-picture matching task to investigate, in patients with left-hemisphere lesions, lexical and syntactic processing over the time-course of the sentence. To minimize the inﬂuence of semantic knowledge, we used shapes (e.g. square, cross), in place of animate actors (Chaterjee , Maher, Gonzalez Rothi, & Heilman, 1995; Schwartz, Saffran, & Marin, 1980). Measuring eye-movements provides ﬁne-grain information regarding how participants incorporate relevant lexical and syntactic information (Dickey, Choy & Thompson, 2007). If participants are quick to use this information, we expect signiﬁcantly more looks to the Target during regions of disambiguation. The goal is to use this data to create a more detailed description of the degree to which lexical and syntactic factors inﬂuence processing difﬁculty, at select points in the sentence, over various syntactic structures. In the sentence-picture matching task, parti cipants selected one of four pictures described by a sentence. Each picture depicted an event (e.g. kicking), between two shapes. An arrow placed between the shapes represented the verb such that the action began at the Agent and ended at the Patient. Each display of four pictures consisted of a Target picture, a Competitor that shared at least one shape with the Target, and two distracters that did not share any shapes with the target or competitor. We manipulated Sentence-Type (Active, Passive, Subject-Cleft, Object-Cleft), and Contrast, which refers to the region in the sentence where the Target picture becomes unambiguous, (NP1, NP2, Verb-Nonreversible, Verb-Reversible). In effect, NP1 and NP2 test lexical processing, while Verb-Nonreversible and Verb-Reversible test syntactic processing, (Figure 1 depicts the Contrast conditions for the sentence, The square kicked the cross). Preliminary results indicate that controls were both more accurate and generally more likely to use lexical and syntactic information earlier than patients. Controls were over 90% accurate while patients were 73% accurate (58% Reversible, 79% in the other three Contrast conditions). For the factor Sentence-Type, controls had a signiﬁcantly higher proportion of looks to the target, during the region of disambiguation, in the Passive, Subject-Cleft and Object-Cleft conditions, while for the patients there was no difference in any of the conditions. For the factor Contrast, controls had a signiﬁcantly higher proportion of looks to the target, in the Verb-Reversible and Verb-Nonreversible conditions, while again there was no difference in any condition for the patients. In sum, while both controls and patients eventually had a higher proportion of ﬁxations to the target, controls were more likely to look at the target shortly after hearing the disambiguating information. The results indicate that even when patients make the correct choice, it takes them longer to incorpo rate relevant lexical and syntactic information. This highlights the claim that comprehension difﬁculties can result from multiple underlying problems. In the future, we plan to perform a more detailed analysis of performance across Sentence-Type and Contrast. Furthermore, we plan to analyze the relation between lesion site and performance.
American Neurological Association 2011
On-line Lexical-Semantics in the Semantic Variant of Primary Progressive Aphasia
Patients with the Semantic Variant of Primary Progressive Aphasia (PPA-SV) have semantic deficits in language. In a word-picture matching study, we used eye-tracking to investigate whether PPA-SV patients suffer from lexical-semantic interference during target selection.
Participants selected one of four pictures (target, competitor, two unrelated) upon hearing its name. The target-competitor relation was either: coordinate (bee-mosquito), associative (cow-farm), or linguistic (butter-fly). We measured accuracy and proportion of fixations to each picture (which should increase with its level of semantic activity). PPA-SV performance was compared with PPA-nonSV patients and controls.
Accuracy scores were high for controls (98%) and PPA-nonSV (93%) and PPA-SV (91%). For each condition, all groups had more fixations to the target than to the competitor and unrelated pictures (p < .05). Controls and PPA-nonSV patients had more fixations to the competitor than the unrelated pictures in the Coordinate condition only (p < .05). In contrast, SV patients had more fixations to the competitor in both the coordinate and associative conditions (p < .05).
These results suggest that the deficits in PPA-SV at least partially stem from increased semantic interference (both coordinate and associative) over the time-course of processing.
American Academy of Neurology 2011