Delineating the contribution of long-term associations to immediate recall

Abstract

In the present study we examined the contribution of semantic associative links to short-term recall performance by using the separation effect first introduced in free recall studies (Glanzer, 1969). Pairs of associated words were inserted in the to-be-remembered lists. In two experiments associated words were better recalled than non-associated words, and were better recalled when they were adjacent in the list than when they were separated by one non-associated item. In addition results showed that forward associative links among pair members were as beneficial to immediate serial recall as backward associative links. Finally the benefit of associative links among pair members was observed with both forward and backward recall.

Keywords:

• Immediate serial recall
• Associative links
• Backward recall

This research was supported by discovery grants from the Natural Sciences and Engineering Research Council of Canada to Jean Saint-Aubin and to Katherine Guérard. Special thanks are due to Olivia Beaudry for running Experiment 2 and to Annie Roy-Charland for hosting the experiment in her laboratory.

This research was supported by discovery grants from the Natural Sciences and Engineering Research Council of Canada to Jean Saint-Aubin and to Katherine Guérard. Special thanks are due to Olivia Beaudry for running Experiment 2 and to Annie Roy-Charland for hosting the experiment in her laboratory.

Notes

¹ The labels cue and target are used for sake of comparability with studies using semantic associates, although in the current experiments using an immediate serial recall task, both pair members have to be recalled and none of them is provided at recall by the experimenter.

² The proportion of correct recall for non-associated words in lists at Distance 0 and at Distance 1 were very similar (.484 at Distance 0 and .478 at Distance 1), and did not significantly differ, F(1, 47) = 0.23, p = .63. Consequently non-associated words from both conditions were merged together.

³ In the Distance 0 condition both pair members could be presented in serial position 3, 4 and 5. Since the same effects were observed whether we considered only these three serial positions or all possible serial positions—that is, positions 2, 3, 4, and 5 for the first pair member and positions 3, 4, 5, and 6 for the second pair member—we present analysis based on all possible serial positions.

⁴ The analyses were restricted to target words because the same target words were used in all three conditions, the only difference across conditions being the associative link with the preceding word, called the cue word. As mentioned in the method section, across the three conditions (forward link, backward link, and weak link), cue words were equated on many dimensions, but they were not equated with the target word. Consequently, it was not possible to compute pair member analyses as in Experiment 1 in which, across participants, the same target words were used as first and second pair member.