Abstract
The affective attribute in free recall was investigated in relation to the order in which words are recalled. A factorial analysis of variance was employed to determine the effect of certain emotionally arousing words on the recall of these words by 24 black and 24 white subjects. The most important result was that black subjects held back affective words significantly more often than white subjects. No significant contextual effect of experimenter's race was observed. The study shows the importance of the affective attribute on the order of free recall.
A number of studies of free recall have demonstrated a high correlation between probability of recall and order of recall (Marbe's Law). For example, items from the initial and terminal positions of a study list are generally recalled both more frequently and earlier (the primacy and recency effects, respectively) than items from intermediate positions (Bousfield, Whitmarsh, & Esterson, 1958; Deese & Kaufman, 1957; Gorfein, Abak, Phillips, & Squillace, 1976). On the other hand, Battig, Allen, and Jensen (1965) presented evidence from three separate multiple-trial free-recall experiments that subjects tend to recall newly learned items prior to items that have been recalled correctly on previous trials (old items). Battig, et al.‘s findings have been supported by several studies (Battig & Slaybaugh, 1969; Steinmetz & Battig, 1969;’ Roberts, 1969). Such a subject strategy is inconsistent with the assumption that order of recall is an index of relative item strength, e.g., the spew hypothesis.
Underwood (1969) has pointed out that some words are probably associated with certain affective responses which are nonverbal, even though words could be used to describe these generated responses. He hypothesized that a memory may consist, in part, of nonverbal and contextual attributes. However, few empirical studies have examined the effects of such contextual cues and affective meanings on free recall. Kayton and Koh (1975) found significant differences in the recall of pleasant and unpleasant words among abnormal populations. Nonverbal attributes were observed to be utilized better in situations of free recall than serial recall (Paivio, et al., 1975). Grimmett (1975) and others have demonstrated that both conceptual and acoustic properties of verbal materials affect free recall. Despite the paucity of such research, it appears that an extension of the hypothesis about the effect of priority in terms of Underwood's attribute theory is possible.
The present study was designed to investigate the influence of the affective attribute on free recall. Specifically, the study attempted, using a series of single-word lists, to answer the following questions: (a) Do affective items on a free recall list influence the order of recall of newly learned items? (b) Do contextual and classification variables, e.g., race of the experimenter, race and sex of the child, interact with affective items to affect their order of recall?
It was expected that the degree of emotional arousal of words would strongly affect the order in which they are recalled. The prediction was that highly emotional words would be held back during free recall. It was also hypothesized that black subjects would suppress these words more often than white subjects. This was expected since the words were selected to be mainly arousing to black subjects. In addition, the race of the experimenter should add to the emotional content of these words. It was anticipated that the words would induce more emotional reaction when a white experimenter presented the lists to a black subject than when a black person was conducting the experiment. Hence, the affective words should be held back most often in cross-racial conditions.
Method
The subjects were 24 black and 24 white sixth grade students who participated in the experiment during school hours. The racial groups were equally divided by sex. Half of each racial group was presented the lists by a black, male experimenter and half by a white, male experimenter.
All of the subjects received four trials on two lists. Each list contained 20 two-syllable nouns. The experimental list contained 10 ‘affective’ words which had racial connotations, e.g., nigger, honkey, and were considered to be more emotionally arousing to blacks than whites. These words were chosen from an unpublished list compiled for a study conducted in the black community of Chicago, dealing with the classification of words by their emotional content. The remaining words on the list were neutral in affectivity and had little associative value. These ‘neutral’ words were all drawn from the Thorndike-Lorge frequency count. The control list contained a set of 10 ‘group’ words which were categorical in nature but not affective. This list also contained 10 noncategorical ‘neutral’ items similar to those on the experimental list. Since the affective words were categorical in nature, the control list was used to account for the effects of clustering on the order of recall.
The subjects were tested individually. Each child was allowed 1 min. in which to recall orally as many words as possible. Each list was presented to a subject in different orders four times. Half of the subjects were given the affective lists first and the other half the control lists first. The order of the words on each trial presentation was randomized. The subjects were randomly assigned to a black or a white experimenter as they entered the laboratory. The lists were presented to the subjects orally by the experimenter at a 2-sec. rate. Each study trial was immediately followed by a 1-min. period of free recall. The interval between the end of the test trial and the beginning of the next study trial was 10 sec. for both groups.
A recall rank score was computed for each word by assigning positive values to all words recalled before the median and negative values after the median. The earlier a word was recalled the higher its recall rank. Recall order scores were computed separately for the affective, group, and neutral words by adding together the ranks for these words on all of the lists.
Results and Discussion
An initial t test was computed to examine for sex differences in holding back affective words. As no significant effect of sex was observed (t = 0.83, df = 46), the male and female subjects were pooled for all further analyses.
A 2 × 2 analysis of variance design (subject's race by experimenter's race) was employed with the order score for the affective words minus the order score for the group words as the dependent variable. A summary of the results of this analysis is shown in Table 1. The emotional content of the affective words appeared to be an important factor in the recall orderings of both black and white subjects. This was indicated by mean negative difference in the recall ranks for both groups (M = −7.02 for blacks; M = −2.35 for whites). However, as shown in Table 1, this suppression of affective words was on the average significantly more apparent for blacks than whites. This was anticipated, since the words were designed to be more emotional for black children.
Comparisons of Recall Ordering of Affective Minus Group Words by Experimenter's and Subject's Race
The main effect of examiner's race was not significant. Contrary to expectations, black subjects reported the affective words later in recall with the black (M = −4.81) than with the white experimenter (M = −3.86). However, the interaction between the race of the experimenter and the race of the child also did not reach significance. Apparently, the experimenter in this testing situation cannot be considered a significant contextual cue.
In general, the results indicated that the affective attribute of words can be an important determinant in the order of free recall. However, the mediating effects of subject and experimenter variables, such as race and sex, were inconsistent. Further study is necessary of additional sets of affective words and contextual cues which may interact with these stimuli to affect the memory process.