3
Effects of semantic and acoustic relatedness on free recall in normal children and retardates* JACK J. ZUPNICKt and WILLIAM E. FORRESTER State University of New York, College at Oswego, Oswego, N.Y. 13126 Third-grade children and matched-MA retardates were presented a list of semantically related, acoustically related, or unrelated words for free recall. Results showed that the normals recalled and clustered significantly more than the retardates. Although both groups of Ss showed a significantly greater tendency to cluster semantically than acoustically, only the normals given the semantic list cIustered in amounts signifieantly above chance levels. Signifieant list differences in terms of amount reealled indicated an inhibitory effect of acoustic relatedness. An extensive amount of research has shown that semantic relatedness of verbal items, as determined by free-association norms or category norms, affects the composition of S's output during free recall. The finding that semantically related items separated during input tend to occur together in clusters during output is well known and easily replicated with normal adult Ss. Studies using normal adult Ss (e.g., Underwood, 1964, pp. 62-65) have also shown that amount of free recall is significantly greater for semantically related than for unrelated words. The possibility that clustering might occur when items are acoustically, rather than semantieally, related was examined by Bousfield & Wicklund (1969). These investigators, using college student Ss and a list consisting of 12 pairs of rhyming words, did indeed find significant "acoustic" clustering. Two subsequent studies have compared the effects of.semantic and acoustic relatedness on free recaU using a mixed-list design (Forrester & King, 1971) and a between-Ss design (Forrester, in press). The two studies were quite consistent in showing significant clustering of semantically related items only. In addition, amount recalled was significantly greater for semantically related items than for acoustically related or unrelated control items. There were no significant differences *We wish to thank Drs. H. H. Spitz and J. J. Winters. Jr •• for tbeir critical comments with regard to the de5ign and procedures of tht present study. Any ßawa remaininl in ihe study are not tbeir responsibility. We are also grateful io the Johnstone Training and Research Center. BordentoWD. N .J., for providing the reiarded S5 and to Dr. C. R. Otis and Mr. T. J. Poddi of the Oswego Campus School for providing the normal Ss. tNow at the University of South Dakota. Vennillion, S. Dak. 57069. 188 in recall of the latter two classes of items. The conclusion reached from the two preceding studies is that acoustic relatedness has no effect on amount of free recall. In contrast to Bousfield & Wicklund's (1969) conclusion, acoustic relatedness does not appear to be an effective basis for organization of verbal items-at least for normal adult Ss. One purpose of the present study was to determine if acoustic relatedness might be a more effective variable in free recall with Ss of lower mental age (MA) or IQ than those used in the above studies. The Ss used in the study, therefore, included normal third-grade children and retarded Ss matched with the normals in terms of MA. The possibility that acoustic relatedness might affee! free recall with retarded Ss was previously examined by Reiss (1968). Since Reiss did not include a control list of unrelated items, however, he could not determine the absolute effects of acoustic relatedness on amount recalled. Furthermore, the effect of acoustic relatedness on clustering could not be fully evaluated in Reiss's study, since blocked rat her than randomized presentation was used in two of bis lists, while clustering of the third, a mixed list which did involve randomized presentation, was evaIuated by formulas which are inappropriate for a mixed list. An attempt was made to eliminate such deficiencies in the present study. With respect to normal children, there are no available data on acoustic-relatedness effects in free recall. There is some evidence, however, which suggests that acoustic characteristics of words are more salient than semantic features for these Ss. Studies using second- and third-grade children, for example, have shown that these Ss make more false-recognition errors for acoustically related, than for semantieally related, words (Bach & Underwood, 1970; Felzen & Anisfeld, 1970). In addition to the question of semantic and acoustic effects, the present study was designed to provide a direct comparison of normals' and retardates' cIustering and recall. Although previous c1ustering studies have included retardates and matched·MA normals as Ss, the results have been somewhat equivocal. Gerjuoy & Spitz (1966), for example, found no differences in either clustering or amount recalled, while in a comparable study by Gerjuoy & Alvarez (1969), normals were found to recall significantly more and to cluster in higher amounts than the retarded Ss. In the two preceding studies, conclusions regarding retardates' and normals' cIustering differences were reached through indirect means rather than by direct statistical analyses. The approach used in these studies was to evaluate the absolute amounts of clustering in the two S groups and then infer a difference or lack of difference between the two. The shortcomings of this approach are most clearly demonstrated in the results reported by Fagan (1969), who examined subjective organization (SO) in retardates and matched-MA normals. While Fagan's normal Ss showed significant SO and the retardates did not, a direct. comparison o f the two groups showed no significant difference. The present study was designed to a110w for a similar analysis of clustering differences. SUBJECTS The Ss were 36 normal third-grade children (mean age = 9.39 years; mean IQ = 110.31) and 36 institutionalized retardates (mean age = 15.54 years; mean IQ = 66.08). The former were obtained from the State University of New York at Oswego Campus School and the latter from the Johnstone Training Center in Bordentown, N.J. Twelve Ss (6 males, 6 females) from each population were assigned randomly to each of the three experimental conditions (semantic, acoustic, or controllist). MATERIALS The learning materials consisted of three Iists of 12 words each, with all words having Thorndike-Lorge frequencies of A or AA. The lists were equated in terms of overall frequency. One list (List S) consisted of three semanticaIly categorized sets of items with 4 words per category. A second list (List A) consisted of three sets of acoustically related items with 4 words per set. A total of 12 "unrelated" words comprised the third list Psychon. Sei., 1972, Vol. 26 (4)

Effects of semantic and acoustic relatedness on free recall in normal children and retardates

Embed Size (px)

Citation preview

Page 1: Effects of semantic and acoustic relatedness on free recall in normal children and retardates

Effects of semantic and acoustic relatedness on free recall in normal children

and retardates*

JACK J. ZUPNICKt and WILLIAM E. FORRESTER State University of New York, College at Oswego, Oswego, N.Y. 13126

Third-grade children and matched-MA retardates were presented a list of semantically related, acoustically related, or unrelated words for free recall. Results showed that the normals recalled and clustered significantly more than the retardates. Although both groups of Ss showed a significantly greater tendency to cluster semantically than acoustically, only the normals given the semantic list cIustered in amounts signifieantly above chance levels. Signifieant list differences in terms of amount reealled indicated an inhibitory effect of acoustic relatedness.

An extensive amount of research has shown that semantic relatedness of verbal items, as determined by free-association norms or category norms, affects the composition of S's output during free recall. The finding that semantically related items separated during input tend to occur together in clusters during output is well known and easily replicated with normal adult Ss. Studies using normal adult Ss (e.g., Underwood, 1964, pp. 62-65) have also shown that amount of free recall is significantly greater for semantically related than for unrelated words.

The possibility that clustering might occur when items are acoustically, rather than semantieally, related was examined by Bousfield & Wicklund (1969). These investigators, using college student Ss and a list consisting of 12 pairs of rhyming words, did indeed find significant "acoustic" clustering. Two subsequent studies have compared the effects of.semantic and acoustic relatedness on free recaU using a mixed-list design (Forrester & King, 1971) and a between-Ss design (Forrester, in press). The two studies were quite consistent in showing significant clustering of semantically related items only. In addition, amount recalled was significantly greater for semantically related items than for acoustically related or unrelated control items. There were no significant differences

*We wish to thank Drs. H. H. Spitz and J. J. Winters. Jr •• for tbeir critical comments with regard to the de5ign and procedures of tht present study. Any ßawa remaininl in ihe study are not tbeir responsibility. We are also grateful io the Johnstone Training and Research Center. BordentoWD. N .J., for providing the reiarded S5 and to Dr. C. R. Otis and Mr. T. J. Poddi of the Oswego Campus School for providing the normal Ss.

tNow at the University of South Dakota. Vennillion, S. Dak. 57069.

188

in recall of the latter two classes of items. The conclusion reached from the two preceding studies is that acoustic relatedness has no effect on amount of free recall. In contrast to Bousfield & Wicklund's (1969) conclusion, acoustic relatedness does not appear to be an effective basis for organization of verbal items-at least for normal adult Ss.

One purpose of the present study was to determine if acoustic relatedness might be a more effective variable in free recall with Ss of lower mental age (MA) or IQ than those used in the above studies. The Ss used in the study, therefore, included normal third-grade children and retarded Ss matched with the normals in terms of MA.

The possibility that acoustic relatedness might affee! free recall with retarded Ss was previously examined by Reiss (1968). Since Reiss did not include a control list of unrelated items, however, he could not determine the absolute effects of acoustic relatedness on amount recalled. Furthermore, the effect of acoustic relatedness on clustering could not be fully evaluated in Reiss's study, since blocked rat her than randomized presentation was used in two of bis lists, while clustering of the third, a mixed list which did involve randomized presentation, was evaIuated by formulas which are inappropriate for a mixed list. An attempt was made to eliminate such deficiencies in the present study.

With respect to normal children, there are no available data on acoustic-relatedness effects in free recall. There is some evidence, however, which suggests that acoustic characteristics of words are more salient than semantic features for these Ss. Studies using second- and third-grade children, for example, have shown that these Ss make more

false-recognition errors for acoustically related, than for semantieally related, words (Bach & Underwood, 1970; Felzen & Anisfeld, 1970).

In addition to the question of semantic and acoustic effects, the present study was designed to provide a direct comparison of normals' and retardates' cIustering and recall. Although previous c1ustering studies have included retardates and matched·MA normals as Ss, the results have been somewhat equivocal. Gerjuoy & Spitz (1966), for example, found no differences in either clustering or amount recalled, while in a comparable study by Gerjuoy & Alvarez (1969), normals were found to recall significantly more and to cluster in higher amounts than the retarded Ss.

In the two preceding studies, conclusions regarding retardates' and normals' cIustering differences were reached through indirect means rather than by direct statistical analyses. The approach used in these studies was to evaluate the absolute amounts of clustering in the two S groups and then infer a difference or lack of difference between the two. The shortcomings of this approach are most clearly demonstrated in the results reported by Fagan (1969), who examined subjective organization (SO) in retardates and matched-MA normals. While Fagan's normal Ss showed significant SO and the retardates did not, a direct. comparison o f the two groups showed no significant difference. The present study was designed to a110w for a similar analysis of clustering differences.

SUBJECTS The Ss were 36 normal third-grade

children (mean age = 9.39 years; mean IQ = 110.31) and 36 institutionalized retardates (mean age = 15.54 years; mean IQ = 66.08). The former were obtained from the State University of New York at Oswego Campus School and the latter from the Johnstone Training Center in Bordentown, N.J. Twelve Ss (6 males, 6 females) from each population were assigned randomly to each of the three experimental conditions (semantic, acoustic, or controllist).

MATERIALS The learning materials consisted of

three Iists of 12 words each, with all words having Thorndike-Lorge frequencies of A or AA. The lists were equated in terms of overall frequency. One list (List S) consisted of three semanticaIly categorized sets of items with 4 words per category. A second list (List A) consisted of three sets of acoustically related items with 4 words per set. A total of 12 "unrelated" words comprised the third list

Psychon. Sei., 1972, Vol. 26 (4)

Page 2: Effects of semantic and acoustic relatedness on free recall in normal children and retardates

Table 1 Means and Standard DeviatioDS for Total Correct Responses ACIOSS Five Trials

List

Ss Semantic Acoustic Unrelated

Normals Mean 43.75 36.00 38.67 sn 4.41 6.83 6.39

Retardates Mean 36.50 30.17 36.83 sn

(List U). Within each of the lists, care was taken to minirnize word relationships along dimensions other than the one appropriate for a given list. The one exception was List A, where acoustic and formal similarity were confounded. The effeet of formal similarity was presumed to be minimal, however, since all materials were presented and recalled orally.

The specific words used in List S were: dog, horse, cat, sheep, milk, bread, cake, corn, leg, ear, nose, and mouth. The words used in List A were: fear, year, clear, near, land, hand, band, sand, my, pie, cry, and tie. List U words were: sheep, bread, clear, land, nose, cry, boat, ten, gray, chair, hat, and gun.

PROCEDURE All Ss were tested individually. Each

S was instructed to repeat list items aloud as they were presented and to recall as many items as possible following presentation of the list. Oral presentation and recall were used in all conditions. Items were presented at a 3-sec rate, with a total of 2 min allowed for S's free recall. Five trials were given to S, with a different randomization of items on each. Within a given randomization, no two items from the same category appeared in successive positions of the list.

RESULTS

6.29 6.96 7.61

by Bousfield & Bousfield (1966). Under this method, each S's c1ustering index is presented by the difference between the observed number of stimulus category repetitions, O(SCR), and tbe expected stimulus category re petitions, E(SCR). The latter is determined by the formula [(m,2 + m,2 . + m 3

2 )/n] -1, where m represents the number of items recalled in eacb of tbe three categories and n represents tbe total number of list items recalled by S. Maximum values of O(SCR) and E(SCR) in tbis study were 9.00 and 3.00, respectively.

Differences between observed and expected SCRs were evaluated by means of one-tailed t tests (df = 11). As shown in Table 2, significant amounts of clustering were found primarily for the normals given List S. Significant t8 ranged from t = 2.02 on Trial 3 to t = 4.93 on TrialLThe only trial on wh\ch tbe normal children did· not show significant semantic clustering was Trial4 (t = 1.55, p< .10). Acoustic clustering, on the other hand, reached a significant level only for the normals on TrialS (t = 6.35, p< .005). The retarded Ss showed no significant clustering in either type of list (all ts< 1.00). An analysis of variance performed on the clustering data (omitting Condition U) reflected the same general findings as in the analysis of amount of recall. Significant effects of S population [F(1,44)=10.92, p<.005], lists [F(1,44)=4.10, p< .05], and trials [F( 4,176) '" 3.17, p< .05] were obtained with no significant interactions. In short, normals clustered significantly more than did

retardates, while both types of Ss showed a significantly greater tendency to cluster List S t~an List A.

DISCUSSION The present results indicate that the

superiority of semantic relatedness over acoustic relatedness, insofar as list organization is concemed, is not limited to normal adult Ss. For both normal ehildren and retardates, the present results show a relatively greater tendency to cluster items semantically than acoustically. In an absolute sense, however, only the normals given the semantic list showed a significant amount of clustering.

The findings that normals recalled significantly more and clustered in significantly greater amounts tban did matched-MA retardates are consistent with the results of Gerjuoy & Alvarez (1969). In the present study, however, clustering differences were determined by direct statistical analysis rather than by the indirect methods used by the latter investigators. The lack of a significant Lists by S Populations interaction in the present study, moreover, indicates that the normals were superior to the retardates in terms of acoustic as weIl as semantic clustering.

In contrast to results obtained witb normal adult Ss, semantic relatedness did not facilitate recall in the present study. The finding that semantic-list recall was superior to the aeoustic list, however, is in agreement with the findings of Reiss (1968), wbo used retarded Ss only. By including a control list of unrelated items, the present study extends Reiss's imdings to show an actual inhibitory effeet of acoustic relatedness.

Tbe finding of a significant deficit in acoustic-list recall suggests that young children and retardates may indeed be influenced by the acoustic features of verbal items to a greater extent than are normal adults. Findings obtained by Forrester (in press) and Forrester & King (1971) indicate that normal adult Ss do not use aeoustic relatedness as a basis for list organization but, instead, treat aeoustieally related words as

An analysis of variance on number of words correctly recalled showed significant main effects of S population [F(1,66) = 10.47, p< .005], lists [F(2,66) = 7.24, p < .005 J, and trials [F{4,264) = 67.75, p < .001]. No inter­actions were significant. A1though tbe normals performed significantly better than tbe retardates overall, rates of learning for these two groups did not differ significantly. A Newman-Keuls analysis sbowed the significant F associated with list differences to be due to the superiority of List S over List A (p < .01) and tbe superiority of List U over List A (p< .05). The difference between Lists Sand U was not significant. These results, discounting tbe trials factor, may be seen in Table 1.

Table 2

Absolute amounts of clustering were evaluated by formulas suggested

Psychon. Sei., 1972, Vol. 26 (4)

Ss

Normals

Retardates

*p < .05,

Observed and

List

Semantic 0 E

Acoustic 0 E

Semantic 0 E

Acoustic 0 E

tp < .005

Expected Stimulus Category Repetitions

Trials

1 2 3 4 5

2.08t 3.08t 3.33* 2.67 3.421" 1.39 2.21 2.46 2.35 2.27 .75 1.50 2.00 2.25 3.17t

1.16 1.41 1.75 1.94 1.98

1.25 1.83 1.50 2.17 2.25 1.15 1.70 1.83 1.74 2.02

.42 1.17 1.50 1.17 1.67

.71 1.35 1.30 1.8Q 1.53

189

Page 3: Effects of semantic and acoustic relatedness on free recall in normal children and retardates

essentially unrelated items. The inhibitory . effeet of aeoustic relatedness on children's and retardates' recall, on the other band, suggests tbat these Ss may attempt to store items by their acoustie features. When Sattempts to decode these aeoustically coded materials, however, aeoustic eonl"usion (i.e., interference) may arise and inhibit recall. The alternative possibility, that interference oceurs during eneoding rather than during decoding, seems less likely in the present study, since Ss were required to repeat items aIoud dUring list presentation and were eorreeted if errors were made.

REFERENCES BACH. M. J.. " UNDERWOOD. B. J.

190

Developmental changes in memory attributes. Journal of Educational Psychology. 1970. 61. 292-296.

BOUSFIELD. A. K •• & BOUSFIELD. W. A. Measurement of clustering and of sequential constancies in expected free recall. Psychological Reports. 1966. 19. 935-942.

BOUSFIELD. W. A •• " WICKLUND. D. A. Rhyme as adeterminant of clustering. Psychonomic Science. 1969. 16. 183-184.

FAGAN. J. F. Free recalliearning in normal and retarded children. Journal of Experimental Child PSYchology, 1969.8, 9-19.

FELZEN. E •• " ANISFELD, M. Semantic and phonetic relations in tbe false recognition of words by tbUd· and sixth-grade children. Developmental Psychology, 1970, 3, 163-168.

FORRESTER, W. E. The effects of semantic and acoustic relatedness on free recall in a between-subjects design. Psychological Reports, in press.

FORRESTER, W. E .. " KING, D. J. The effects of semantic and acoustic relatedness on free recall and clustering. Journal of Experimental Psychology, 1971,88, 16-19.

GERJUOY, I. R., " ALVAREZ, J. M. Transfer of leaming in associative clustering of retardates and normals. American Journal of Mental Deficiency. 1969,73,733-738.

GERJUOY, I. R., " SPITZ, H. H. Associative clustering in free recall: InteUectual and developmentaI variables. American Journal of Mental Deficiency. 1966. 70, 918-927.

REISS, P. Rhyming and conceptual relationships in a free·recall task with retardates. American Journal of Mental Deficiency, 1968,73,81-85.

UNDERWOOD, B. J. The representativeness of rote verbalteaming. In A. W. Metton (Ed.), Catellories o( human learninll.

New Volk: Academic Press, 1964.

Psychon. Sei., 1972, Vol. 26 (4)