An item-level analysis of lexical-semantic effects in free recall and recognition memory using the megastudy approach

Abstract

Psycholinguists have developed a number of measures to tap different aspects of a word's semantic representation. The influence of these measures on lexical processing has collectively been described as semantic richness effects. However, the effects of these word properties on memory are currently not well understood. This study examines the relative contributions of lexical and semantic variables in free recall and recognition memory at the item-level, using a megastudy approach. Hierarchical regression of recall and recognition performance on a number of lexical-semantic variables showed task-general effects where the structural component, frequency, number of senses, and arousal accounted for unique variance in both free recall and recognition memory. Task-specific effects included number of features, imageability, and body-object interaction, which accounted for unique variance in recall, whereas age of acquisition, familiarity, and extremity of valence accounted for unique variance in recognition. Forward selection regression analyses generally converged on these findings. Hierarchical regression also revealed that lexical variables accounted for more variance in recognition compared with recall, whereas semantic variables accounted for more unique variance above and beyond lexical variables in recall compared with recognition. Implications of the findings are discussed.

Exploring word memorability: How well do different word properties explain item free-recall probability?

“…The animacy and usefulness correlations are higher in magnitude than any of word properties that had been considered in the analyses reported by Lau et al (2018).…”

“…I would like to thank Mike Kahana and his lab for generously making the PEERS data freely available for others to use; I similarly would like to thank Mabel Lau for providing the item-wise results from Lau et al (2018). I would also like to thank Daniela Palombo for feedback on an earlier version of the manuscript.…”

“…Here I use data from the Penn Electrophysiology of Encoding and Retrieval Study (PEERS) study to examine word memorability by estimating free-recall probability for words from a database of 1638 words, in a sample of 147 young adults. While a handful of studies have investigated the influence of individual word properties on free recall (e.g., Christian et al, 1978;Rubin, 1980;Rubin & Friendly, 1986;Nairne et al, 2013;Lau et al, 2018), they did not consider the range of semantic properties examined here and were conducted in smaller databases of words. These findings were then replicated in a second dataset (Lau et al, 2018) of 532 words from a sample of 116 young adults.…”

Vector-Space Models of Semantic Representation From a Cognitive Perspective: A Discussion of Common Misconceptions

“…In some descriptions, DSMs are presented as (refined) word co-occurrence measures (Barsalou, 1999; Dreyer & Pulvermüller, 2018; French & Labiouse, 2002; Goldberg, Perfetti, & Schneider, 2006; Lau, Goh, & Yap, 2018; McKoon & Ratcliff, 1998; Meteyard, Cuadrado, Bahrami, & Vigliocco, 2012). For example, Dreyer and Pulvermüller (2018) describe DSMs as “distributional learning of word-word correlations from texts” (p. 66), and Kowialiewski and Majerus (2018) explicitly use LSA cosine similarities as a word co-occurrence measure, stating that “LSA measures the extent to which two words co-occur within similar contexts using large corpora” (p. 70).…”

Effectiveness of the method of loci is only minimally related to factors that should influence imagined navigation

“… 2. After exclusions, the words CEMENT, CLOCK, DRESSER, LAUNDRY, PIGEON, SHOWER, STATUE, STOOL, STOVE, STRAW, TOILET, and TRASH were removed because these named objects were found in one of the virtual environments. Numerous other uncontrolled word characteristics could influence probability of recall (e.g., Lau, Goh, & Yap, 2018), and random sequences of words might carry idiosyncratic meaning or vary in ngram frequency. However, because word lists were constructed in a new random order for each participant, this is expected to operate as random error, not likely to be systematically related to our Environment variable. …”