Testing the associative-link hypothesis in immediate serial recall: Evidence from word frequency and word imageability effects

Abstract

Two immediate serial recall experiments were conducted to test the associative-link hypothesis (Stuart & Hulme, 2000). We manipulated interitem association by varying the intralist latent semantic analysis (LSA) cosines in our 7-item study word lists, each of which consists of high- or low-frequency words in Experiment 1 and high- or low-imageability words in Experiment 2. Whether item recall performance was scored by a serial-recall or free-recall criterion, we found main effects of interitem association, word imageability, and word frequency. The effect of interitem association also interacted with the word frequency effect, but not with the word imageability effect. The LSA-cosine×word frequency interaction occurred in the recency, but not primacy, portion of the serial position curve. The present findings set explanatory boundaries for the associative-link hypothesis and we argue that both item- and associative-based mechanisms are necessary to account for the word frequency effect in immediate serial recall.

Acknowledgements

This research was supported in part by a Presidential Fellowship awarded to the first author by the University at Albany, State University of New York. We thank Steve Lewandowsky, Jean Saint-Aubin, and an anonymous reviewer for their comments on the earlier draft of this paper and Rebekah Feinman and Caitlin Gorr for their help with data collection.

Notes

¹The episodic and pre-existing interitem associations could also be qualitatively different. While Stuart and Hulme's (2000) pre-experimental familiarisation procedure builds strong yet temporary episodic associations between words, Hulme et al.'s (2003) use of LSA-cosine indexes non-episodic linguistic associations between words that have developed over time. The latter conceptualisation seems to be more closely related to Deese's (1960) original proposal.

²In some (168 out of 2744, 6%) of our individual word pairs, the values of LSA-cosine were negative (ranged from –.01 to –.07). None but one of the mean LSA-cosines in each word type stated in Appendix B changed when we computed the means based on the signed, rather than absolute, values. (The exception was LF-low LSA-cosine lists—the LSA-cosine changed from .04 using signed values to .05 using absolute values.) Hence, we only presented the mean LSA-cosine based on the signed values of LSA-cosine.

³Another way to manipulate the word frequency (or word imageability) and LSA-cosine was to counterbalance the same set of high and low frequency/imageability words across two levels (high vs. low) of LSA-cosine manipulation. This could completely eliminate any confound between low vs high LSA-cosine in each level of word frequency/imageability manipulation.

⁴Nevertheless, in a trial-by-trial analysis, Saint-Aubin, Ouellette, and Poirier (2005) reported that the facilitation in recalling in order the categorically similar items, relative to the categorically dissimilar items, was constant across trials in their experiment.

⁵When we restricted our analyses of the HF-high LSA-cosine and LF-low LSA-cosine lists, we did not obtain the Frequency×Serial Position interaction, F(1, 27) = 1.62, MSE=.02, which was reported in some (e.g., Hulme et al., 1997), but not all of the prior studies (e.g., Saint-Aubin & Poirier, 2005). Previous research reported that the word frequency effect became stronger as a function of serial positions. After controlling the interitem association, we found a Frequency×Serial Position interaction for high LSA-cosine lists, F(1, 27) = 2.23, MSE=.02, but not for low LSA-cosine lists, F(1, 27) = 1.12, MSE=.02. However, as depicted in Figure 1, the word frequency effect became weaker as a function of serial positions for these high LSA-cosine lists. Yet this latter pattern was also reported in studies using the non-spoken recall modality but not controlling the interitem association (e.g., Saint-Aubin & LeBlanc, 2005).

⁶In Poirier and Saint-Aubin (1996), their “LF” words were far lower in frequency (only the words with occurrence of 1 per million) than the ones we used in the present study (the words with occurrence between 1 and 77 per million). Thus, we considered their “medium-frequency” words to be more suitable for this comparison. We considered the findings of Poirier and Saint-Aubin's phonologically dissimilar lists because none of our word lists consisting of phonologically similar items and the findings of Saint-Aubin and Poirier's (2005) and Stuart and Hulme's (2000) unfamiliarised HF and LF word lists because we did not use the pre-experimental familiarisation procedure in our experiment.

⁷Because Allen and Hulme (2006), Bourassa and Besner (1994), and Walker and Hulme (1999) formed their word lists by randomly drawing items from a limited word pool in each condition, we computed the mean LSA-cosine of each item with respect to all of the other items in the same pool and then found the mean LSA-cosine for HI and LI words across all of their corresponding items.