Strategy use in the reading span test: An analysis of eye movements and reported encoding strategies

Abstract

Strategy use in the traditional reading span test was examined by recording participants’ eye movements during the task (Experiment 1) and by interviewing participants about their strategy use (Experiment 2). In Experiment 1, no differences between individuals with a low, medium, and high span were observed in how they distributed processing time between task elements. In all three groups, fixation times on words up to the to-be-remembered (TBR) word became shorter and the time spent on the TBR longer as memory load in the task increased. The results of Experiment 2, however, show that span groups differ in the use of memory encoding strategies: individuals with a low span use mainly rehearsal, whereas individuals with a high span use almost exclusively semantic elaboration. The results indicate that the use of elaborative strategies may enhance span performance but that not all individuals are necessarily able to use such strategies efficiently.

Acknowledgements

Parts of the data were presented at the 13th Conference of the European Society for Cognitive Psychology in Granada, Spain, in September 2003; at the 46th Annual Meeting of the Psychonomic Society in Toronto, Canada, in November 2005; and at the 3rd European Working Memory Symposium in Genoa, Italy, in June 2006.

We would like to thank Miia Juntunen for her help in preparing the Finnish materials for the span task.

Notes

¹The present study focused on how participants divided processing time between different regions of the reading span test materials, not on the component processes related to reading, such as word recognition. Thus, some modifications were made to the standard procedures typical for psycholinguistic research. First, in the present study a skipped word was assigned a gaze duration value of 0 ms. In psycholinguistic eye-movement research it is generally preferable to analyse skip rates separately and to treat skips as missing data in fixation time analyses. However, because we were interested in how participants divide their processing time between the different regions of the reading span test sentences, a composite measure of fixation time is more informative than separate measures of skip rates and gaze durations. Due to the same reason, data from all words that were in the middle of the sentence, including function words, were included when computing the average fixation time for the middle of the sentence. Thus, the words in the different sentence regions were not matched for frequency or length, which is why one should not directly compare the fixation times between the different sentence regions (both word frequency and length have a significant influence on eye fixation times during reading; see Hyönä & Olson, 1995; Rayner, 1998). Moreover, the first word and the TBR were not matched for these factors (the mean length and lemma frequency per million were 9.00, SD=3.41 and 595, SD=1178 for the first word in set-size 4; 8.75, SD=2.45 and 56, SD=6 for the TBR word in set-size 4; 8.13, SD=2.47 and 701, SD=1041 for the first word in set-size 5; and 7.73, SD=1.49 and 56, SD=8 for the TBR word in set-size 5). It is also typical that the sentence-final words attract longer gaze durations due to the sentence wrap-up effect (Rayner, Kambe, & Duffy, 2000). Thus, the effects of sentence region are only interesting if they emerge in interactions with the other variables of interest (memory load and reading span group).

²Due to multiple levels of memory load, it would have been possible to also test for other types of contrasts (e.g., quadratic or cubic). However, as we had no a priori predictions for such effects, they were not tested.