Contextual Constraints in Situation Model Construction: An Investigation of Age and Reading Span

ABSTRACT

This study examined the effects of age and reading span on the ability to use contextual constraints during language comprehension. Older and younger participants listened to sentences over headphones and named pictures that appeared subsequently. Older adults named pictures faster when the preceding sentence context matched rather than mismatched the shape of the depicted object, but younger adults showed less of a match advantage. This effect of contextual match was especially pronounced in older high-span participants, consistent with models of cognitive aging in which surface level processing declines in older adulthood whereas processing at the situation model level remains intact. Results suggest that the practiced ability to immediately construe word meanings and activate the appropriate stored representations is preserved, if not strengthened throughout the lifespan.

Keywords:

• Language comprehension
• Age-related expertise
• Perceptual simulations
• Sentence processing
• Embodied cognition

ACKNOWLEDGEMENTS

We thank Brittney Dodson, Marissa Mainwood, Michelle Peruche, Nickesha Pitter, Greg Smith, and Adam Summeralls for their assistance in the data collection. We also thank several anonymous reviewers for their helpful comments on an earlier version of this manuscript. This research was supported by grant MH 63972 from the National Institute on Mental Health, and the data were collected at the Florida State University in Tallahassee, FL.

Notes

¹ Bad trials removed by group and condition. High span/young: match = 4.1%, mismatch = 4.8%; High span/old: match = 4.4%, mismatch = 5.8%; Low span/young: match = 4.6%, mismatch = 3.8%; Low span/old: match = 5.3%, mismatch = 6.8%

Outliers removed by group and condition. High span/young: match = 5.2%, mismatch = 4.1%; High span/old: match = 6.6%, mismatch = 5.8%; Low span/young: match = 5.9%, mismatch = 3.8%; Low span/old: match = 6.8%, mismatch = 5.7%

²Only the poorly recognized pictures (chewed gum) were removed in both their match and mismatch conditions, but not their corresponding well-recognized pictures (stick of gum). This was done to preserve as many items as possible in the analysis. The three removed items fell evenly across the 4 lists, resulting in the removal of one item per match condition and 1 item per mismatch condition on each list, except that no items were removed in the match condition of list A or the mismatch condition of list D. Thus, the counterbalancing of the lists was relatively preserved, and the resulting lists after removal of the poorly recognized picture items were as follows.

• List A: 14 match items, 13 mismatch items

• List B: 13 match items, 13 mismatch items

• List C: 13 match items, 13 mismatch items

• List D: 13 match items, 14 mismatch items

Percentages for outliers and bad naming trials do not include these 3 bad items.

³For counterbalanced designs only subject analyses should be performed (Raaijmakers, 2003; Raaijmakers et al., 1999). Raaijmakers et al. (1999) explain that there is no need to include both subject and item analyses in cases where item variability is experimentally controlled by matching or counterbalancing. Specifically, they state that ‘it is not required to do separate analyses over subjects and items in order to test the effect of the treatment factor. The expected means-squares, under the assumption that List is a random effect, show that the treatment effect can always be tested directly using the mean-squares obtained from the standard subject analysis (averaging over items)' (p. 425).