Recall of common and uncommon words from pure and mixed lists

Recall of common and uncommon words from pure and mixed lists

JOURNAL OF VERBAL LEARNING AND VERBAL BEHAVIOR 19, 240-245 (1980) Recall of Common and Uncommon Words from Pure and Mixed Lists VERNON H . G'REGG AND...

484KB Sizes 0 Downloads 18 Views

JOURNAL OF VERBAL LEARNING AND VERBAL BEHAVIOR 19, 240-245 (1980)

Recall of Common and Uncommon Words from Pure and Mixed Lists VERNON H . G'REGG AND DANIELLE C . MONTGOMERY

Birkbeck College, University of London AND DOLORES CASTAlqO

University of Valencia Recall of high- and low-frequency words in the conventional free recall paradigm was compared with recall of the same words when subjects were required to count backward before and after the presentation of each word. The addition of this distractor task was associated with a reduction in the high-frequency advantage otherwise found with pure lists containing only high- or low-frequency words. This finding is attributed to the disruption of organizational processes. In contrast, the low-frequency advantage found with conventional presentation of mixed lists, containing high- and low-frequency words, was not reduced by distraction. These findings indicate that the frequency effects obtained with mixed and pure lists have different origins.

Free recall of lists containing only highfrequency words (Kueera & Francis, 1967; Thorndike & Lorge, 1944) is generally superior to recall of lists composed of only low-frequency words (e.g., Deese, 1960; Duncan, 1974; Gregg, 1976; Sumby, 1963). In contrast to the results obtained with pure lists ( c o n t a i n i n g o n l y high- o r lowfrequency words), low-frequency Words are more likely to be recalled than highfrequency words from mixed lists (containing equal numbers of high- and lowfrequency words). This has been shown by direct comparisons of recall from mixed and pure lists reported by Duncan (1974), Gregg (1976), and May and Tryk (1970). In addition, Gregg (1976) has drawn attention to the evidence for low-frequency superiority that is revealed by detailed consideration of the experiments reported by Matthews (1966) in which mixed lists were employed. Apparently contradictory results were obtained by Raymond (1969). She employed mixed lists containing either high- and low-frequency words, or highand low-frequency nonsense trigrams, and found that high rather than low frequency was associated with superior recall. However, Raymond failed to analyze the sepa240 0022-5371/80/020240-06502.00/0 Copyright~) 1980by AcademicPress, Inc. All rightsof reproductionin any formreserved.

rate effects of frequency on words and trigrams thus leaving open the possibility that the main effect of frequency contained a small low-frequency advantage with words which was offset by a larger high-frequency advantage with trigrams. Gregg and Casta~o (Note 1) have checked on this possibility and obtained a low-frequency advantage with mixed lists of words under conditions similar to those used by Raymond. Thus, the reversal of the frequency effect with mixed lists relative to pure lists is a robust phenomenon. The effect of word frequency on recall of pure lists has largely been attributed to differences in interitem semantic relationships and the opportunities these create for effectively encoding the words. For example, Deese (1960) attributed the effect to differences in normative associations between w o r d s b e c a u s e he f o u n d , f o r highfrequency lists at least, that recall was related to an index of interitem associative strength (IIAS). Also, he found no relationship between word frequency and recall when the index of IIAS for each list was zero. Postman (1970) took a different approach and attempted to identify the basis of the high-frequency advantage in terms of

WORD FREQUENCY AND RECALL

241

" a d o p t e d c h u n k s " (Tulving & Patkau, employed by Deese (1960) for pure lists is 1962). He obtained larger subjective or- based on single response, unidirectional asganization (SO) scores (Tulving, 1962) for sociative norms (Russell & Jenkins, 1954). high- than for low-frequency lists and was As Matthews, Marcer, and Morgan (1964) able to conclude that the high-frequency have argued, this method may underestiadvantage reflects the ease with which or- mate the associative relations effectively ganizational units may be developed from used by subjects. The same point can be the two types of list. Likewise, Sumby made with respect to Postman's (1970) as(1963) attributed the superior recall of sessment of IIAS. The SO measure of subhigh-frequency words to the ease with jective organization has also been criticized which meaningful relationships between because it taps only a relatively simple level them are discovered and utilized. He did so of organization (see, e.g., Monk, 1976; because changes in serial position curves Pellegrino, 1971). Such limitations in meaover successive learning trials were similar sures oflIAS and SO may account for some to t h o s e o b t a i n e d with lists of high- apparent inconsistencies in the results of s e q u e n t i a l a p p r o x i m a t i o n - t o - E n g l i s h ; Deese and of P o s t m a n . For example, low-frequency lists showed changes similar whereas Deese (1960) found no relationship between recall and word frequency when to low-approximation sequences. Although these three accounts of the fre- I1AS was zero, Postman (1970) obtained a quency effect differ in several ways they significant frequency effect with lists of are similar in one respect at least. Accord- minimal levels of IIAS. Also, despite this ;ng to encoding specificity theory (see, e.g. latter finding by Postman, the correlations lving, 1974) interitem relationships between individual SO scores and recall which facilitate recall must be encoded at were not significant. While more elaborate methods may help the time traces are stored. This seems to apply for normative associations (Tulving resolve the problems of assessing subjec& Thomson, 1973), and for organization tive organization (Monk, 1976) and asgenerally (see Postman, 1972). It is possible sociative relationships (Matthews et al., to argue from this theory that these three 1964) the present study represents a differaccounts of the frequeficy effect all require e n t m e t h o d o l o g i c a l a p p r o a c h . T h e the subjects to appreciate and encode use- through-list distractor paradigm, introful interitem relationship during presenta- duced by Whitten and Bjork (Note 2) was tion of the lists. employed because it discourages organizaIt seems plausible that the low-frequency tional activity and the appreciation of advantage in the recall of mixed lists might interword relationships. This paradigm inalso be explained by opportunities for or- volves separating to-be-remembered items, ganization created by interitem relation- or groups of items, in the presentation seships within such lists. As far as we are quence by periods of distractor activity aware this potential explanation has not such as backward counting. The parabeen tested, and possible tests which digm also includes instructions which merely follow those procedures used with encourage subjects to attend to each item pure lists and referred to earlier are of only while it is presented. To the extent that questionable value. Apart from any possi- these instructions are successful they ble problems in applying organizational should reduce simultaneous rehearsal of measures separately to high- and low- to-be-remembered words, decrease the frequency words in mixed lists, the mea- subject's appreciation of interword resures of IIAS and SO may not be com- lationships, and so inhibit the active organipletely informative about the interitem re- zation of words into higher-order units at lationships utilized by subjects in either input. If the high-frequency advantage in pure pure or mixed lists. The index of IIAS

242

GREGG, MONTGOMERY, AND CASTAlqO

lists is correctly understood in terms of organizational processes taking place during list presentation it should be attenuated in the t h r o u g h - l i s t d i s t r a c t o r p a r a d i g m . Likewise, if the low-frequency advantage in mixed lists is attributable to organizational processes it too should be attenuated by through-list distraction. These predictions were tested by comparing the effect of word frequency on recall when conventional presentation is employed with that obtained when through-list distraction is added, for both pure and mixed lists. METHOD

Subjects. Forty-eight subjects from the Birkbeck College Subject Pool took part in the experiment. They all had English as their first language. Design and materials. A four-way factorial design was employed. Method of Presentation (with or without through-list distraction) and Frequency (high or low frequency of occurrence) were within-subject variables. Type of List (mixed or pure) and Order of Testing (with or without throughlist distraction in the first block of trials) were between-subject variables. The word lists were taken from those used by Gregg and Castaflo (Note 1) which gave rise to a significant high-frequency advantage in recall of pure lists and a smaller low-frequency advantage in mixed lists. These lists were c o n s t r u c t e d from 216 bisyllabic nouns, all judged by the experimenters to have imageable referents. Half were high-frequency words having frequencies of at least 40 per million in the G count of Thorndike and Lorge (1944). The remaining words were low frequency (1 per million or less). Each word was printed on a white card for visual presentation. Each of the nine high-frequency lists contained 12 words drawn randomly from the pool without replacement and with the restriction that no list contained phonemically similar words. The words were randomly o r d e r e d within lists. Nine lowfrequency pure lists were similarly constructed. The 18 pure lists were randomly

allocated to three sets P(A), P(B) and P(C) with the constraint that each set contained three high- and three low-frequency lists. Three sets each of six mixed lists were cons t r u c t e d by amalgamating half a pure h i g h - f r e q u e n c y list w i t h h a l f a lowfrequency list. Mixed set M(A) was constructed from the lists in set P(A), M(B) from P(B), and M(C) from P(C). The order in which high- and low-frequency words occurred was pseudorandom, the restriction being that no more than three words from the same frequency class should occur in immediate succession. Each subject was tested on two list sets, one with and one without through-list distraction, the latter to be referred to as conventional presentation. Each subject in the pure list condition was yoked with one in the mixed list condition with respect to list sets. For example, for each subject who received sets P(A) and P(B) there was a subject who received sets M(A) and M(B). List sets were then balanced across the remaining experimental conditions. Procedure. Subjects were randomly allocated to the four groups corresponding to the combination of the between-subject variables of Type of List and Order of Testing. They were tested individually in a single session lasting about 50 minutes. Separate instructions were given before each part of the experiment which contained one practice trial with words of medium frequency and six experimental trials. A 3-minute rest period separated the two parts of the experiment. In the conventional presentation condition, the instructions emphasized that the subjects could use any mental strategy which might help them r e m e m b e r the words. Each trial started with the command Ready followed by manual presentation of the words at a 2-second rate. A 13th card containing a three-digit number was included in each list and subjects were required to vocalize the digits as soon as they were presented and to count backwards from the number by threes, aloud, and as fast as possible. The 10-second period of

243

WORD FREQUENCY AND RECALL

counting was terminated by the command Recall and 60 seconds was allowed for written, free recall. In the through-list distraction condition the instructions stressed the need for the subject to perform the counting task as fast and as accurately as possible. Also, they were to c o n c e n t r a t e on each word only when it was actually being presented. Each trial started with the command Ready followed by presentation of a three-digit n u m b e r printed on a card. The subjects were required to vocalize the number immediately and then count backwards, by threes, aloud and as fast as possible for 10 seconds. A 100-millisecond visual signal indicated the end of the counting period at which point a word was presented. The end of the 2-second presentation period was marked by the visual signal and another three-digit number presented by placing it on top of the previous card. The 12th word in each list was followed by a 10-second period of counting which was terminated by the auditory c o m m a n d Recall. Subjects were allowed 60 seconds for written, free recall. RESULTS

The mean proportions of words recalled are shown in Table 1. The data are collapsed o v e r O r d e r of Testing since this variable had no important consequences. T h e r e s u l t s s h o w t h a t the large highf r e q u e n c y advantage obtained with pure lists, u n d e r c o n v e n t i o n a l presentation is virtually eliminated when through-list distraction is present. In mixed lists, by con-

trast, a modest low-frequency advantage under conventional presentation is slightly increased by through-list distraction. There is a generally lower level o f recall with through-list distraction than with conventional presentation. Analysis of variance tests were carried out based, for convenience, on the total number of words recalled in each condition. S e p a r a t e analyses o f pure and mixed lists allowed the separate Frequency × Method of Presentation int e r a c t i o n s f o r e a c h t y p e o f list to be evaluated. A n a l y s i s o f data f r o m the p u r e lists s h o w e d that n e i t h e r the main e f f e c t o f Order of Testing nor any of the interactions involving it were significant (p > . 1 in all cases). The effect of F r e q u e n c y was significant, F(1,22) = 15.59, p < .001, MSe = 8.08. Performance under through-list distraction was significantly worse than under standard presentation, F(1,22) = 22.87, p < .001, MSe = 27.11. Importantly, the interaction between F r e q u e n c y and Method of Presentation was significant, F(1,22) = 9.89, p < .01, MSe = 8.15. The data for mixed lists was analyzed in the same manner. The main effect of Frequency was significant, F(1,22) = 4.76, p < .05, MSe = 13.29. Performance was significantly better with conventional presentation than with t h r o u g h - l i s t d i s t r a c t i o n , F(1,22) = 12.33,p < .01,MS~ = 21.62. The interaction between Frequency and Method of Presentation was not significant (F < 1), but the low-frequency advantage was increased slightly by the presence of through-list distraction. The only signifi-

TABLE 1 FREE RECALL PROBABILITIES Pure lists

High

Mixed lists

frequency

Low frequency

High frequency

Low frequency

Conventional presentation

.56

.45

.48

.52

Through-list distraction

.37

.35

.38

.44

244

GREGG, MONTGOMERY, AND CASTAlqO

cant term involving order of testing was the interaction with Method of Presentation. This simply reflects a general decline in performance over trials. DISCUSSION

The pattern of recall obtained with conventional presentation shows a large highfrequency advantage with pure lists and a smaller low-frequency advantage with mixed lists. These results are almost identical to those obtained by Gregg and Castafio (Note 1) and are broadly in line with other previous research (e.g., Duncan, 1974; Gregg, 1976). Adding through-list distraction s u b s t a n t i a l l y r e d u c e d the highfrequency advantage with pure lists, but slightly increased the low-frequency advantage with mixed lists. It also produced a general reduction in performance to levels comparable to those obtained in other studies employing similar distraction procedures (Gardiner & Gregg, 1979; Tzeng, 1974). This general drop in performance may, in part, be due to factors not directly concerned with the discouragement of organizational activity, such as overall increases in retention interval. The different consequences of adding through-list distraction for recall of pure and mixed lists strongly suggests that the frequency effects obtained with conventional presentation have different origins. They show also that through-list distraction does not merely have a general, detrimental effect on recall (cf. Poltrock & Macleod, 1977). The pure list results are consistent with the view that the high-frequency advantage obtained with conventional presentation is a consequence of organizational activity, and that through-list distraction effectively discourages such activity. The present study, then, achieves the aims of implicating organizational activity as the origin of the high-frequency advantage, while employing a single-trial recall procedure and avoiding the problems of directly assessing interitem relationships or degrees of organization. Although the present method does not discriminate between

the positions of Postman (1970), Deese (1960), and Sumby (1963) with respect to the critical aspects of pure lists which facilitate the high-frequency advantage, the results are compatible with all three. There is no apparent reason to suppose that the discouragement of organizational activity associated with the reduction of the high-frequency advantage with pure lists did not also occur with mixed lists. Thus, the finding that the low-frequency advantage with mixed lists was not reduced by through-list distraction suggests it is independent of organizational activity. How, then, is it to be accounted for? One possibility is that recall of the low-frequency list members interferes with recall of the highfrequency members. This suggestion may be considered attractive because it places the locus of the effect at retrieval rather than at encoding. However, Duncan (1974) has been able to dismiss this suggestion. He ensured that each to-be-remembered word began with a unique initial letter so that recall of either high- or low-frequency words could be cued with the appropriate letters. His results showed superior recall of lowfrequency words from mixed lists even though a high-frequency advantage was obtained when the same words were cued from pure lists. May and Tryk (1970) placed the locus of the effect at the encoding stage by suggesting that low-frequency words, by virtue of their unfamiliarity, are more attentiongaining than the high-frequency words with which they are mixed. Although May and Tryk did not elaborate their attentiongaining explanation it seems that a plausible argument along these lines is that subjects judge the low-frequency words more difficult to remember than high-frequency words and, when able to do so, devote more time to them. The present results are inconsistent with such a suggestion because the through-list distractor paradigm encourages processing of each and every item singly. Hence, such biases in the allocation of processing time should be reduced by this procedure and the low-frequency ad-

WORD FREQUENCY AND RECALL

vantage reduced accordingly. Thus, the results indicate that neither biases in the allocation of processing time, nor the exploitation of organizational opportunities offered by m i x e d lists will e x p l a i n the l o w frequency advantage. The findings reported here indicate that the frequency effects found with pure and mixed lists have different origins. While the basis of the effect with pure lists appears to be the ease with which inter-item relationships can be exploited by organizational activity, the underlying cause(s) of the effect with mixed lists remains unresolved. REFERENCES DEESE, J. Frequency of usage and number of words in free recall: The role of association. Psychological Reports, 1960, 7, 337-344. DVNCAN, C. P. Retrieval of low-frequency words from mixed lists. Bulletin of the Psychonomic Society. 1974, 4, 137-138. GARDINER, J. M., & GREGG, V. H. When auditory memory is not overwritten. Journal of" Verbal Learning and Verbal Behavior, 1979, 18, 705719. GREGG, V. Word frequency, recognition and recall. In J. Brown (Ed.), Recall and recognition. London: Wiley, 1976. KU(SERA, H., & FRANCIS, W. N. Computational analysis of present-day american english. Providence, R.I.: Brown Univ. Press, 1967. MATTHEWS, W. A. Continued word associations and free recall. Quarterly Journal of Experimental Psychology, 1966, 17, 31-38. MATTHEWS, W. A., MARCER, D., & MORGAN, E. Word association hierarchies and free recall. Journal of Verbal Learning and Verbal Behavior, 1964, 3, 371-375. MAY, R. B., & TRYK, H. E. Word sequence, word frequency, and free recall. Canadian Journal of Psychology, 1970, 24, 299-304. MONK, A. F. A new approach to the characterization of sequential structure in multi-trial free recall using hierarchical grouping analysis. British Journal of Mathematical and Statistical Psychology, 1976, 29, 1-18. PELLEGRINO, J. W. A general measure of organization in free recall for variable unit size and internal sequential consistency. Behaviour R e s e a r c h Methods and Instrumentation, 1971, 3, 241-246. POLTROCK, S. E., & MACLEOD, C. M. Primacy and

245

recency in the continuous distractor paradigm. Journal of Experimental Psychology: Haman Learning and Memory, 1977, 3,560-571. POSTMAN, L. Effects of word frequency on acquisition and retention under conditions of free-recall learning. Quarterly Journal of Experimental Psychology, 1970, 22, 185-195. POSTMAN, L. A pragmatic view of organization theory. In E. Tulving and W. A. Donaldson (Eds.), Organization of memory. New York: Academic Press, 1972. RAYMOND, B. Short-term storage and long-term storage in free recall. Journal of Verbal Learning and Verbal Behavior, 1969, 8, 567-574. RUSSELL, W. A., & JENKINS, J. J. The complete Minnesota norms for responses to 100 words from the Kent-Rosanoff association test. (Tech. Report No. 11). University of Minnesota. ONR Contract N8 ONR-66216, 1954. SUMBY, W. H. Word frequency and serial position effects. Journal of Verbal Learning and Verbal Behavior, 1963, 1, 443-450. THORNDIKE, E. L. & LORGE, I. The Teacher's Handbook of 30,000 Words, New York: Columbia Univ. Press, 1944. TULVING, E. Subjective organization in free recall of "unrelated" words. Psychological Review, 1962, 69, 344-354. TULVING, E. Recall and recognition of semantically encoded words. Journal of Experimental Psychology, 1974, 102, 778-787. TULVING, E., & PATKAU, J. E. Concurrent effects of contextual constraint and word frequency on immediate recall and learning of verbal material. Canadian Journal of Psychology, 1962, 16, 83 - 95. TULWNG, E., & THOMSON, D . M . Encoding specificity and retrieval processes in episodic memory. Psychological Review, 1973, 80, 352-373. TZENG, O. J. L. Positive recency effect in a delayed free recall. Journal of Verbal Learning and Verbal Behavior, 1973, 12, 436-439. REFERENCE NOTES 1. GREGG, V. H., & CASTANO, D. Recaerdo de palabras inglesas comunes y no comunes procedentes de listas puras y mixtas. Paper presented at Twenty-second Meeting, Spanish Psychological Society, Valencia, April, 1978. 2. WH~TXEN, W. B., & BJORK, R. A. Test events as learning trials: The importance of being imperfect. Paper presented at Midwestern Mathematical Psychology Meetings, Bloomington, Indiana, April 1972. (Received October 25, 1979)