Abstract
This study examined three processes crucial to reading comprehension (semantic access, integration, and inhibition) to identify causes of comprehension impairment. Poor comprehenders were compared to chronological-age controls and vocabulary-age (VA) controls. When listening to homonym primes (“bank”) versus unrelated primes, controls were faster to name pictures related to dominant (money) and subordinate (river) meanings at 250 ms interstimulus interval (ISI) but only showed dominant priming at 1,000 ms ISI, whereas poor comprehenders only showed dominant priming. When listening to subordinately biased sentences ending in homonyms (“John fished from the bank”) versus control sentences, all groups showed priming when naming subordinate (appropriate) pictures at 250 ms ISI: VA controls and poor comprehenders also showed priming when naming dominant (inappropriate) pictures. At 1,000 ms ISI, controls showed appropriate priming, whereas poor comprehenders only showed inappropriate priming. These findings suggest that poor comprehenders have difficulties accessing subordinate word meanings, which can manifest as a failure to inhibit irrelevant information.
ACKNOWLEDGMENTS
This research was funded by a 1+3 Economic and Social Research Council studentship to Lisa Henderson. We thank the children and schools who participated and Charles Hulme and Hannah Nash for their valuable comments.
Notes
1One problem for “multiple-access” views of word recognition is the “backward priming effect.” On this view, priming effects observed for multiple meanings at short ISIs may emerge because participants, upon encountering the target stimulus, develop a backward inference to the preceding prime. It has been argued that, unlike the lexical decision task, the naming task is not as susceptible to backward priming effects. CitationSeidenberg, Waters, Sanders, and Langer (1984) measured both naming and lexical decision times for the second words of asymmetrically related word pairs, such as “stick – lip,” which were highly related only in the backward direction (SOA = 500 ms). Only lexical decision revealed backward priming effects. Hence, the naming task was used here in an effort to reduce the influence of backward priming.
2We first carried out a four-way mixed-design ANOVA including ISI as an additional within-subject factor. There was a significant ISI × Association × Relatedness interaction, F(1, 48) = 10.83, p < .01, ηp 2 = .18, which justifies our decision to separate the analysis by ISI condition and test specific hypotheses about each ISI in turn. The ISI × Association × Relatedness × Group interaction was not significant, F(2, 48) = 0.17, p > .05, ηp 2 = .01.
3We first carried out a four-way mixed-design ANOVA including ISI as an additional within-subject factor. There was a significant main effect of ISI, F(1, 48) = 10.16, p < .01,ηp 2 = .18, and significant interactions between Context × Appropriateness, F(1, 48) = 4.19, p < .05,ηp 2 = .08; Context × ISI, F(1, 48) = 17.60, p < .001, ηp 2 = .27; Context × ISI × Group, F(2, 48) = 3.37, p < .05, ηp 2 = .12; and Context × Appropriateness × Group, F(2, 48) = 2.91, p < .05, ηp 2 = .11, which justifies separate ANOVAs to test specific hypotheses about each ISI.