Brain and Language 103 (2007) 8–249 www.elsevier.com/locate/b&l
Automatic thematic role priming of related verbs in younger and older adults Lisa A. Edmonds
*
Communication Sciences and Disorders, University of Florida, P.O. Box 117420, Dauer 351, Gainesville, FL 32611, USA
Verbs are central to the semantics and syntax of a sentence via their relationship with their arguments/thematics. Online priming studies in young adults have found that verbs (arrest) provide immediate access to typical agents (policeman) and patients (criminal) by generating expectancies from a verb and its related thematic roles (Ferretti, McRae, & Hatherell, 2001) and vice versa (McRae, Hare, Elman, & Ferretti, 2005). These findings indicate a neural co-activation whenever either a verb or its closely related noun is selected in a relevant context. The specific aims of the current investigation are to (1) replicate agent and patient to verb priming results of McRae et al. (2005) in young adults (YA) and (2) investigate agent and patient to verb priming patterns in older adults (OA). It is hypothesized that older adults will show the same priming pattern as YA’s, but they will exhibit overall slower reaction times. Methods Participants Twenty-one young adults (YA) (M = 21.5 years) and 17 older adults (OA) (M = 72.5 years, minimum age = 60 years) participated. Participants were right-handed English speakers. OA’s scored within normal limits on tests to screen for cognitive and linguistic disorders (Cognitive Linguistic Quick Test, Helm-Estabrooks, 2001) and semantic processing impairments in nouns (Pyramids and Palmtrees, Howard & Patterson, 1992) and verbs (Kissing and Dancing, Bak & Hodges, 2003). Stimuli One set of animate agent-verb pairs and one set of inanimate patientverb pairs were developed. To create related pairs, a group of 13 undergraduates completed two questionnaires (Ferretti et al., 2001). Participants rated how common it is for someone do something (agent condition) and for something to receive the action of something (patient condition) on a scale of 1 (not common) to 7 (very common). Thirty-one related agentverb pairs (thief/stealing) (M = 6.73, SD = 0.14) and 26 related patientverb pairs (coffee/brewing) (M = 6.67, SD = 0.19) were chosen. Unrelated pairs were created by shuffling the related pairs (thief/winning, pizza/brew-
*
Fax: +1 352 846 0243. E-mail address:
[email protected]
doi:10.1016/j.bandl.2007.07.048
ing). Filler pairs with different verbs and nonword pairs with present progressive inflection (barber/prafing) were also developed for both conditions. The ratio of target pairs to foil pairs was 2:3. The experiment was developed on (DirectRT Empirisoft, 2004) and run on a Dell computer with 17-in. monitor. Trials were presented in random order. Each trial consisted of (1) a focal point (#) presented in the center of the screen for 500 ms, (2) the prime (thief) for 200 ms, (3) a mask (&&&&&&&&&) for 50 ms, and (4) the target (stealing) until the participant decided whether the target was a word or a nonword. The inter-stimulis interval (ISI) was 250 ms, and the intertrial interval (ITI) was 1500 ms. Procedure Modifications to McRae et al. procedures were made in anticipation of running this experiment with persons with aphasia. Modifications included use of the left hand and a lexical decision task rather than oral reading. These modifications were not expected to affect the results (Kiran & Thompson, 2003). Instructions were to read the first word silently and to decide if the second word was a real word or not. Participants pressed a ‘‘y’’ to indicate yes and an ‘‘n’’ to indicate no (letters were taped on neighboring keyboard keys) Order of condition (agent and patient) was counterbalanced across participants. Results Response latencies greater than 3 SD’s from the mean (<1% of all responses for both groups) were replaced by the upper limit value. Incorrect responses (<1% of all responses for both groups) were replaced by the average reaction time (RT) for that participant. See Fig. 1 for RT results for both conditions and groups. Agent condition Reaction times (RT) for the related and unrelated pairs were analyzed by paired t-test for each group. For the YA group, RT was significantly shorter (t(650) = 3.596, p = .000) for related agent-verb pairs (M = 642.37 ms, SD = 226.33) than for the unrelated verb-agent pairs (M = 683.07 ms, SD = 235.51). For the OA group, RT was significantly shorter (t(464) = 1.876, p = .03) for related pairs (M = 954.28 ms, SD = 364.07) than for the unrelated pairs (M = 996.58, SD = 418.34). An unpaired t-test with Bonferroni correction was conducted to examine RT across groups.
68
Abstract / Brain and Language 103 (2007) 8–249 had significantly faster RT’s (t(1592) = 19.691, p < .0001) than the OA’s (M = 930.96 ms, SD = 327.90).
1050 REL
Reaction time (Msec)
1000
UR
950
Discussion
900
Both the YA and OA groups exhibited faster reaction times for agents and patients paired with related verbs as compared to unrelated verbs, indicating an automatic co-activation of related verbs when typical agents or patients are activated. These results replicate the findings of McRae et al. (2005) in YA’s despite methodological modifications and indicate that OA’s maintain normal thematic role-verb processing with overall slower RT’s than YA’s. This work provides a paradigm for testing automatic processing of verb-thematic processing in single words in persons with aphasia.
850 800 750 700 650 600 YA agents
YA patients
OA agents
OA patients
Groups
Fig. 1. Reaction times for related and unrelated agent-verb and patientverb pairs for young adults (YA) and older adults (OA).
Young Adults (M = 659.43 ms, SD = 236.11) had significantly faster RT’s (t(1753) = 21.313, p < .0001) than the OA’s (M = 943.78 ms, SD = 349.63). Patient condition RT’s across participants were significantly shorter (t(545) = 3.997, p = .000) for related patient-verb pairs (M = 645.6 ms, SD = 223.03) than for the unrelated verb-patient pairs (M = 690.26, SD = 248.22). For the OA group, RT’s across participants were significantly shorter (t(389) = 3.975, p = .000) for related patient-verb pairs (M = 901.54 ms, SD = 269.31) than for the unrelated patient-verb pairs (M = 981.52, SD = 389.51). An unpaired t-test with Bonferroni correction was conducted to examine RT across groups Young adults (M = 668.00 ms, SD = 238.13)
References Bak, T., & Hodges, J. R. (2003). ‘Kissing and Dancing’—a test to distinguish the lexical and conceptual contributions to noun/verb and action/object dissociation. Preliminary results in patients with frontotemporal dementia. Journal of Neurolinguistics, 16, 169–181. Direct RT Research Software (2004). New York, NY: Empirisoft. Ferretti, T. R., McRae, K., & Hatherell, A. (2001). Integrating verbs, situation schemas, and thematic role concepts. Journal of Memory and Language, 44, 516–547. Helm-Estabrooks, N. (2001). Cognitive Linguistic Quick Test. United States of America: The Psychological Corporation. Howard, D., & Patterson, K. (1992). The Pyramids and Palm Trees Test. London: Harcourt Assessment. Kiran, S., & Thompson, C. K. (2003). Effect of typicality on online category verification of animate category exemplars in aphasia. Brain and Language, 85, 441–450. McRae, K., Hare, M., Elman, J. L., & Ferretti, T. (2005). A basis for generating expectancies for verbs from nouns. Memory and Cognition, 33, 1174–1184.