Reading in Healthy Aging: Selective Use of Information Structuring Cues

ABSTRACT

Previous research has shown that information referring to a named character or to information in the main clause of a sentence is more accessible and facilitates the processing of anaphoric references. We investigated whether the use of such cues are maintained in healthy aging. We present two experiments investigating whether information contained in the main clause of a sentence is more accessible compared with information contained in the subordinate clause. We present two further experiments that explored whether proper names act as controllers of discourse focus. Experiment 1 showed that information contained in the subordinate clause of a sentence decreased the processing efficiency of anaphoric references (more so for older adults), and Experiment 2 found that main clauses facilitated probe recognition. Experiments 3 and 4 showed that named characters increased the processing efficiency of anaphoric references and facilitated probe recognition (younger adults only), whereas older adults displayed a primacy effect.

Acknowledgments

This work was supported by an Economic and Social Research Council (U.K.) Ph.D. scholarship awarded to J.M.P.

Notes

¹ Covariate analysis was carried out to rule out whether any age differences was due to the higher level of education or predicted IQ and showed no reliable effects on the dependent variables within each age group (Fs < 1); this was the case for all four experiments. Previous research has shown that older adults tend to yield superior performance on the National Adult Reading Test compared with younger adults, which requires pronunciation of irregular spelled words such as superfluous (e.g., Burke & Shafto, 2008; Uttl, 2002).

² There was a high proportion of correct answers to the comprehension questions referring to the main clause (younger adults 92.2%; old adults 93.6%) and subordinate clause conditions (younger adults 89.1%; older adults 89.3%) for both age groups. None of these numerical differences was statistically reliable.

³ Word frequency and length might also influence the likelihood of recognizing a probe; these were matched by determining how frequently each word occurred in the British National Corpus (900 million words). The mean log frequencies of occurrence per million was 3.03 (SE = .15) and the words were approximately equated for length by the number of letters (5.63).

⁴ We correlated the performance on the span measures and total reading times with the probe accuracy scores. None of the correlations were statistically reliable (ps > .05) for all measures, ass was the case for Experiments 2 and 4.

⁵ This was assessed by pretesting for gender differences on a scale of 0 (typically male) to 7 (typically female). Five participants rated the perceived gender differences between the role descriptions (typical male mean = 1.8, typical female mean = 4.8). Roles were chosen that showed a strong bias toward either a male or female being in that occupation.

⁶ There was a high proportion of correct answers to the comprehension questions referring to the named character (younger adults 91.8%; older adults 87.9%) or role (younger adults 89.1%; older adults 90.3%) for both age groups. None of these numerical differences was statistically reliable.

⁷ Word length could influence the likelihood of recognizing a probe; the critical nouns were matched across conditions for length, and no significant differences were found. The length of the nouns in the name conditions averaged 6.86 (min. = 4, max. = 11) characters, whereas the nouns in the role condition averaged 7.11 (min.= 5, max.= 11) characters.