Effect of contextual meaningfulness on the rated meaningfulness and recall of words

Effect of contextual meaningfulness on the rated meaningfulness and recall of words 1 HARRIETT AMSTER UNIVERSITY OF CALIFORNIA, BERKELEY In three lists containing the same subset of 15 standard words of medium rated meaningfulness (m') and 15 context words of varying m', recall of the context words increased directly with their m'. Recall of the standard words did not vary reliably, but tended to be most affected by contextual meaningfulness under number-assignment instructions under which recall increased with the m' of the context items. In contrast to the tendency toward a direct variation of recall with contextual m', the m' of the standard words was found to vary inversely with the m' of the context words, replicating Amster & Battig (1965). In an earlier study (Amster & Battig, 1965), verbal contexts which varied in meaningfulness were found to affect both rated meaningfulness (m') and associative productivity (m) of a standard subset of items within the lists, but expected effects of contextual meaningfulness on recall were not observed. The present experiment constitutes a further study of contextual effects on recall, and also permits replication of the earlier results on m'. Contextual m' was expected to affect the free recall of other items within the list, but might depend on the type of instructional set employed. Amster & Battig (1965) found increased contextual m' to decrease the m' of standard words while it facilitated m. These conditions of contextual m' enable determining whether recall is dependent on rated or actual meaningfulness, Le., if recall is dependent on m, the recall of a subset of items should be improved under the contextual conditions which increase the m of these items. If dependent on m', recall should be improved under the opposite conditions. A subset of standard items appeared in three lists differing with respect to the m' of the other (contextual) items. Recall followed m' ratings or a condition in which the same words were rated for evaluative connotation or were assigned numbers with no basis specified by E. Amster & Battig (1965) may have failed to find significant recall results because of the prior task, to produce associations to the words in the list. These associations may have been strengthened by virtue of their overt verbalization and could have constituted interference at recall. This explanation is consistent with an earlier failure to find facilitation in recall following the production of associations (Mandler & Campbell, 1957). The present experiment also controlled for pleasantness, known to affect recall (Amster, 1964), and item order. Psychon. Sci., 1966, Vol. 4 METHOD Subjects The 108 Ss included 63 elementary pSyChology students from Contra Costa Junior College, and 45 educational-psychology students at the University of California at Berkeley. Additional Ss were randomly discarded so that the 12 Ss in each condition included equal proportions of seven and five from the respective student populations. Desien The nine conditions represented a 3 by 3 factorial design with the three types of rating instructions (m', pleasantness, and unspecified assignment of numbers) combined with the three lists varying in contextual m'. All lists contained the same 15 standard words ofmediumm' and 15 context words which were either relatively high, medium, or low in m'. Four different orders within the list were each used for an equal number of Ss. For each order the standard words were presented in the same ordinal poSition for each of the three lists. Materials Prior to the experiment, the 60 CVC words used by Amster &. Battig (1965) along with 60 additional CVC words of comparable m' levels, were rated for pleasantness on a nine-point scale by 30 educational-psychology students at the University of California at Berkeley. One set of 15 standard words with a median m' value of 3.54 and 15-word sets of higher (4.21), equal (3.53), and lower (2.95) m' context words, which were equated andhighiy homogeneous in neutral evaluative connotations were selected. Procedure All Ss were tested in Intact classes, with all nine conditions administered in every class. Each S was given a small spiral notebook to be used for his ratings, along with a three-page booklet containing instructions, a list of words appropriate to his conditions, and a response sheet for recall. For the m' rating, S rated each word as the "number of things or ideas which it makes you think of," on a scale ranging from none(l) tovery very many(7). For the evaluative rating, Ss were asked "to rate each word as to how pleasant it is" on a scale from very very unpleasant (l)very very pleasant (7). For the number control, Ss were told "to simply write down a number from one to seven next to each word" and that "there is no right or wrong about it." Ss were further instructed to record the time when they finished their ratings and then to study the words in preparation for the recall test in which "you will be asked to write down the ones you remember." After the 7.5 mln. allotted for the rating task, the Ss were given 5 mln. to record as many words from the list as they could in whatever order they came to mlnd and not to try to write the words in a space corresponding to its position in the orignal list. RESUL TS An analysis of variance of the rat i n g s(see Table 1) in the m' and number-assignment conditions revealed a significant interaction of Context List with Word Type (standard or context), F=10.39, df=2/66, p< .01, and an interaction of these variables with Instruction, F =9.34, df", 2/66, P < .01. The latter reflects the absence of any consistent effect of either List or Word Type under the number-assignment condition. However, replicating Amster & Battig (1965), standard words were rated reliably lower in the high m I context than in the low m I context. According to the Scheffe contrast, mean differences (Table 1) as great as .56 are sig151 Table 1. Mean Ratings of Standard and Context words in Three Lists under Meaningfulness Instructions Type of word High m' Medium m' Low m' Standard Context 3.14 4.18 3.87 3.96 4.13 3.23 nificant beyond the .05 level •. A similar analysis for the pleasantness and number-assignment conditions revealed only. that significantly larger numbers were used for rated pleasantness (2.14) thannumber-assignment (1.97), F=5.34, df=1/66, p< .05. As shown by the overall means in Table 2, recall of the context words varied directly with their m' but the corresponding variation in recall of standard words was not consistent, producing a significant interaction between these two variables, F= 3.13, df = 2/72, P < .05. Recall of the standard words varied directly with contextual m' under the number-assignment condition, but deviated from this pattern under pleasantness instructions. Although the triple interaction was n.s., a more stringent test, an analysis of the recall of standard words only, revealed a trend (F=2.16, d (...truncated)