Journal of Experimental Child Psychology 111 (2012) 552–560

Contents lists available at SciVerse ScienceDirect

Journal of Experimental ChildPsychology

journal homepage: www.elsevier .com/locate/ jecp

Brief Report

Rehearsal dynamics in elementary school children

Martin Lehmann ⇑, Marcus HasselhornGerman Institute for International Educational Research, D-60486 Frankfurt, GermanyCenter for Individual Development and Adaptive Education of Children at Risk (IDeA), D-60325 Frankfurt, Germany

a r t i c l e i n f o

Article history:Received 27 May 2011Revised 28 October 2011Available online 22 December 2011

Keywords:RehearsalFree recallMemory strategiesMemory developmentEpisodic memoryEpisodic association

0022-0965/$ - see front matter � 2011 Elsevier Indoi:10.1016/j.jecp.2011.10.013

⇑ Corresponding author at: German Institute forE-mail address: lehmann@dipf.de (M. Lehmann

a b s t r a c t

Several studies on free recall suggest that processes responsible forrecall are analogous to processes responsible for rehearsal. In chil-dren, the relationship between cumulative rehearsal and recallperformance has been proven to be critical; however, the locus ofthe effect of rehearsal is not yet fully understood. To unfold themechanisms that come into play in an overt rehearsal free recalltask, we assessed rehearsal and recall sequences in childrenbetween 8 and 10 years of age. These sequences give informationabout the context in which items are repeated and rearrangedthroughout the list and subsequently recalled. Rehearsal sequencesconsisted mainly of items from neighboring list positions in theiroriginal temporal order. The same characteristics were true forrecall sequences. Qualitatively, order effects during study andrecall did not differ over age groups. However, in older childrenwho were using cumulative rehearsal more intensively, successiverehearsal and recall of items in their original order was more pro-nounced. Therefore, we suggest that a main feature of item rehear-sal with regard to facilitating recall is the strengthening ofinteritem associations based on the temporal order within a listand that this characteristic develops with age.

� 2011 Elsevier Inc. All rights reserved.

Introduction

Research on rehearsal strategy development has stressed the importance of rehearsal style, namelyactive rehearsal, for subsequent recall performance. The source responsible for the effects of active re-hearsal, however, largely remains open. One reason for the lack of understanding rehearsal–retrievalcorrespondences is that children’s dynamic processes involved in studying and subsequently recalling

c. All rights reserved.

International Educational Research, D-60486 Frankfurt, Germany.).

M. Lehmann, M. Hasselhorn / Journal of Experimental Child Psychology 111 (2012) 552–560 553

a list of words are often neglected. The current study analyzed these processes by exploring possibleage-related differences in rehearsal sequences during presentation and subsequent recall order in anovert rehearsal free recall task.

The implementation of the overt rehearsal procedure by Rundus and Atkinson (1970; see alsoRundus, 1971), who asked participants to study aloud during list learning, provided information abouthow often, when, and in what combination items were studied. Today, rehearsal development is char-acterized as a change in quality (rehearsal style) and quantity (intensity of use). Between the presen-tation of successive words within a list, young children mainly use passive forms of rehearsal by eitherrepeating a presented word only once (labeling) or repeating it several times (single word rehearsal).Children between 8 and 10 years of age mainly use an active rehearsal style (cumulative rehearsal)and repeat different items together (Lehmann & Hasselhorn, 2007; Ornstein & Naus, 1978). In addi-tion, rehearsal development seems to entail a gradual change from an intensive use of labeling towarda dominant use of cumulative rehearsal. In their longitudinal study, Lehmann and Hasselhorn (2007)discovered that young children mainly started memorizing the list by using cumulative rehearsal butshifted to labeling each newly presented item early in the list. When children grew older, they againstarted with cumulative rehearsal but continued to do so for a prolonged period of time and usedlabeling late in the list (if at all). In accordance with the development of rehearsal behavior on the pas-sive–active dimension, recall performance also changes with age; the more actively children rehearse(Ornstein & Naus, 1978) and the more intensively they use cumulative rehearsal (Lehmann & Hassel-horn, 2007), the better is their recall performance.

Considering the processes of study and recall, Laming (2006, 2008), in providing an algorithm topredict recall on the basis of vocalizations during rehearsal, suggested that rehearsal and recall mightbe regarded as similar; the sequence of item rehearsals is a copy of the sequence of the elements inmemory, and the sequence of recalls is a copy of recorded items in memory. Accordingly, cumulativerehearsal of several items implies retrieval of previously presented items (miniature free recall) inaddition to the actually presented item (Tan & Ward, 2000; Ward, Woodward, Stevens, & Stinson,2003). The resulting rehearsal sequences seem to operate on preceding rehearsal sequences and areassumed to be systematic. For instance, Kahana (1996) showed that successive recall of items is morelikely to come from nearby serial positions than from remote serial positions and is also reflected inshorter interresponse times (contiguity effect) (for an overview, see Kahana, Howard, & Polyn, 2008).Hence, if temporal order helps participants to recall items in a specific manner (e.g., Sederberg, Miller,Howard, & Kahana, 2010), this should also be the case when retrieval for rehearsal takes place. In areanalysis of their longitudinal data, Lehmann and Hasselhorn (2010) demonstrated that the contigu-ity effect was present in children’s recall sequences, developing with age. However, it is uncertainwhether contiguity in recall is based on similar order effects during rehearsal.

Because rehearsal seems to continuously assemble item sequences during study, we addressed twounresolved issues in the current study: Can the concept of free recall processes (i.e., contiguity effect)in children be transferred to their rehearsal behavior, and are there age-related differences duringstudy that, in turn, result in recall differences?

Method

Participants

Participants in the study were 19 second-graders (7 girls and 12 boys, mean age = 8.0 years,SD = 4 months), 18 third-graders (8 girls and 10 boys, mean age = 9.0 years, SD = 5 months), and 18fourth-graders (8 girls and 10 boys, mean age = 10.2 years, SD = 5 months) from the suburban areaof Munich, Germany. All of the children took part voluntarily (with parental permission) and were pri-marily from Caucasian families with average incomes.

Materials and procedure

All children were presented with an overt rehearsal free recall task and were tested individually.Both their study and recall behaviors were videotaped and digitally recorded for subsequent analyses.

554 M. Lehmann, M. Hasselhorn / Journal of Experimental Child Psychology 111 (2012) 552–560

All children were given the same three lists of words to memorize and to subsequently recall. Thethree distinct lists consisted of 12 highly concrete familiar items from different semantic categories.Each list contained only 1 item from each category, and the three lists were parallel with regard tothe number of syllables.

At the beginning of each trial, children were informed that they would be acoustically and visuallypresented with a number of items to be memorized and then asked to subsequently recall these itemsin any order they wished. In addition, children were asked to do all of their thinking/studying aloudwhile each list item was presented (see also Lehmann & Hasselhorn, 2007). The lists of items were pre-sented acoustically from a CD and by simultaneously placing 5 � 5-cm picture cards, with the namesof the objects printed underneath their pictures, in front of the children. Each individual card was vis-ible for an interval of 8 s until the subsequent item was acoustically presented and the card was re-placed by the subsequent picture card. After the last presentation interval, the picture cards wereremoved and children were asked to recall as many items as possible. The recall phase ended whenchildren were not able to recall any further item (new or already named) for a 30-s interval. To ensurethat children were able to meet the task requirements and to familiarize them with the overt studyfree recall procedure, two practice trials (consisting of 6 items each) were administered.

Analysis procedure for study and recall processes

In an overt rehearsal free recall task, the traditional approach to quantifying age-related differencesin study behavior is the examination of rehearsal set sizes, that is, the number of different items stud-ied within an interstimulus interval. Rehearsal sets give information about whether children repeatseveral items together (cumulative rehearsal) or repeat single items only once (labeling) or severaltimes (single word rehearsal). Age-related differences in recall are classically quantified by the inspec-tion of serial position curves. In collapsing over input and output positions, however, rehearsal set sizeand serial position curve discard information about dynamic sequential effects, that is, item-by-itemcontingencies. To quantify these contingencies, we adopted the conditional response probability as afunction of lag (lag–CRP) measure introduced by Kahana (1996); lag–CRP gives information abouthow one recall follows another. Lehmann and Hasselhorn (2010) demonstrated that the lag–CRP isa valid measure for recall processes in children. Regarding rehearsal in children, however, we knowlittle about the processes and routines that items undergo when entering memory for subsequent re-call. In the following sections, we demonstrate how patterns of recalled items are quantified on thebasis of the lag–CRP and how we transfer the lag–CRP measure on study processes.

Conditional response probability during recallThe CRP quantifies the order in which items are recalled and is plotted as a function of the distance

(lag) between 2 items within the list reinstated during recall, that is, the probability of recalling itemi + lag after recalling item i. The maximum lag is defined by the range of the list. In our case, using listsconsisting of 12 items each, there are 11 possible backward transition lags (–1 to –11) and 11 possibleforward transition lags (+1 to +11). The calculation of the CRPs as a function of lag (i.e., lag–CRPs)works as follows. For each list of each child, we move stepwise through each recall transition. By doingso, the numerator value that matches the actual transition is incremented and the denominator valuesmatching the set of all possible recall transitions are incremented. Finally, for each child, the lag–CRPfor each possible lag is calculated by dividing the number of transitions made to that lag (numerator)by the number of transitions that could have been made to that lag (denominator) (see also Lehmann& Hasselhorn, 2010; Sederberg et al., 2010).

Conditional response probability during studyTransferring the concept of the lag–CRP to study processes is based on the following assumption:

Interstimulus Intervals in an overt rehearsal free recall task that are used to rehearse previously pre-sented items can be considered as short retrieval periods (Tan & Ward, 2000). Hence, to quantify ordereffects within rehearsal, we analyzed each interstimulus interval as if it were a recall interval. For eachinterstimulus interval, we adopted the same calculation procedure as described above with the differ-ence of an increasing list length for each subsequent interstimulus interval because of each additional

M. Lehmann, M. Hasselhorn / Journal of Experimental Child Psychology 111 (2012) 552–560 555

item presentation. As a consequence, the possible transition lag size also increases. For instance, forthe second interstimulus interval the current list length is 2 and the possible transition lags are –1and +1, and for the third interstimulus interval the current list length is 3 and the possible transitionlags are –2, –1, +1, and +2.

Conditional response latency during studyAnother aspect concerning dynamic processes in overt rehearsal free recall is the variation of inter-

response times (IRTs) between successive responses. As defined by Howard and Kahana (1999), theso-called conditional response latency as a function of lag (lag–CRL) displays the mean IRT between suc-ceeding recalls of items from serial positions i and i + lag. Analogously, the lag–CRL during study dis-plays the mean IRT between successive rehearsals of previously presented items from serial positions iand i + lag.

Scoring of response order and response latencies during study and recallAnalysis of both items’ rehearsal order during each interstimulus interval and items’ recall order

during the recall phase was conducted via inspection of children’s videotaped performance in theovert rehearsal free recall task. Analysis of the IRTs was realized via a computer-based visual displayof the digital audio signal. Two independent raters agreed on approximately 92% of the assessment ofnumber and order of vocalized items and on approximately 89% of the assessment of the time be-tween each recall. Disagreements were resolved through mutual inspection of the recordings anddiscussion.

Results

Results are presented in two sections. The first section displays evidence of age-related differencesin dynamic study processes, and the second section presents age-related differences in dynamic recallprocesses. Furthermore, analyses of the relationship between both processes are reported. Becausepreliminary analyses did not reveal any effects regarding children’s gender, the data reported belowhave been collapsed across gender. Unless stated otherwise, all differences were significant at thep < .05 level.

Study

Strategic behaviorThe finding of age-related differences in rehearsal behavior (Lehmann & Hasselhorn, 2007), namely

predominant use of passive rehearsal (labeling and single word rehearsal) in younger children and pre-dominant use of cumulative rehearsal in older children, seems to be very important for the analysis ofstudy dynamics; active rehearsal, but not passive rehearsal, allows individually ordering the list itemsduring study and, therefore, allows temporal contiguity when memorizing the items. Table 1 reveals anage-related decrease of labeling/single-word rehearsal and an age-related increase of cumulativerehearsal condensed over all lists. Other/not-observable behavior decreased as a function of grade.An analysis of variance (ANOVA) with list (1, 2, and 3) and strategic behavior (labeling/single-word

Table 1Means (and standard deviations) for percentages of strategic behavior and recall performance for Grades 2, 3, and 4.

Grade Strategic behavior (%) Recall

Labeling/single-word rehearsal Cumulative rehearsal Other

2 60.9 (38.8) 20.6 (31.5) 18.5 (34.1) 4.9 (1.0)3 65.7 (36.6) 32.2 (37.7) 2.1 (7.2) 5.7 (1.1)4 33.4 (35.6) 62.6 (35.0) 4.0 (8.1) 6.3 (0.9)

Note. Strategic behavior was identified over all three lists for Interstimulus Intervals 2 to 12. For more detailed information onstrategy scoring, see Lehmann and Hasselhorn (2007).

556 M. Lehmann, M. Hasselhorn / Journal of Experimental Child Psychology 111 (2012) 552–560

rehearsal, cumulative rehearsal, and other) as within-subject factors and grade (2, 3, and 4) as the be-tween-subject factor revealed a significant main effect of strategic behavior, F(2, 104) = 19.15, p < .001,g2

p = .27, and a significant Strategic Behavior � Grade interaction, F(4, 104) = 5.17, p < .001, g2p = .17.

Bonferroni post hoc comparisons indicated that children in Grade 4 used more cumulative rehearsaland, in turn, used less labeling/single-word rehearsal than the younger students (all ts > 2.60); thatis, they rehearsed different items together more often.

Conditional response probability during studyBecause there were very few interfering repetitions (0.04, 0.04, and 0.05 for Grades 2, 3, and 4,

respectively) and a total of only four intrusions during study, repetitions and intrusions were excludedfrom further analyses. The first interstimulus interval was also ignored because here only a single itemwas available for study and transition was impossible. As mentioned above, rehearsal behavior seemsto change throughout the list. For analyzing order effects during study, we subdivided the list intothree sections: Interstimulus Intervals 2–5, 6–9, and 10–12.

Based on the children’s rehearsal order, Figs. 1A to 1C show the lag–CRP curves for each gradeas a function of section. The lag–CRP curves reveal that rehearsals of items over and withinsections are much more likely to come from forward adjacent serial positions than from eitherbackward adjacent serial positions (asymmetry effect) or remote serial positions. However, olderchildren seem to reveal higher lag–CRPs for forward adjacent recalls throughout the list. To quan-titatively assess age-related differences in the lag-recency effect, we averaged within each of thethree sections and differentiated between a forward adjacent transition (lag +1) and remote tran-sitions (forward and backward transitions with lags +3, +4, +5 and –3, –4, –5, respectively). Thisprocedure is very similar to Kahana, Howard, Zaromb, and Wingfield (2002) and Lehmann andHasselhorn (2010), and it takes into consideration the asymmetry effect in associative processesin free recall.

The observations in Figs. 1A to 1C were confirmed by an ANOVA with grade (2, 3, and 4) as the be-tween-subject factor and lag (forward adjacent and remote), list (1, 2, and 3), and section (2–5, 6–9,and 10–12) as within-subject factors. This analysis revealed significant main effects of grade, F(2,52) = 6.22, p < .01, g2

p = .19, lag, F(1, 52) = 47.00, p < .001, g2p = .48, and section, F(2, 104) = 33.27,

p < .001, g2p = .39, which were qualified by significant interactions of Grade � Lag, F(2, 52) = 5.24,

p < .01, g2p = .17, Lag � Section, F(2, 104) = 42.86, p < .001, g2

p = .45, and Grade � Lag � Section, F(4,104) = 2.85, p < .05, g2

p = .10. No other effect was significant. Bonferroni post hoc comparisons sup-ported the observation that children’s CRPs for adjacent forward rehearsals were larger than thosefor remote rehearsals (all ts > 2.95). Hence, the temporal proximity of list items seems to exert a stronginfluence on rehearsal order within the interstimulus Intervals. In addition, comparisons indicatedthat successive rehearsals of forward adjacent items were more frequent within the first section com-pared with later list-learning behavior (all ts > 3.13) and that this effect was larger and even extendedto the second and third section in fourth-graders (all ts > 2.62).

Strategic behavior and lag–CRP during studyAs displayed in Figs. 1A to 1C, age-related differences seemed to exist in the contiguity effect on a

quantitative level but not on a qualitative level, with all three age groups tending to rehearse forwardadjacent items one after another. Because older children were more often engaged in cumulative re-hearsal than younger children, we analyzed the relationship between strategic behavior and meanlag–CRPs over Interstimulus Intervals 2 to 12. We found significant correlations between lag–CRPsfor forward adjacent transitions and percentages of cumulative rehearsal, r(18) = .79, r(17) = .81,and r(17) = .64 for Grades 2, 3, and 4, respectively. Correlations between lag–CRPs for remote transi-tions and percentages of cumulative rehearsal were significant for Grade 2, r(18) = .48, and Grade 3,r(17) = .66, but not for Grade 4, r(17) = .37. In sum, when children from Grades 2 and 3 implementedcumulative rehearsal, they successively rehearsed items from adjacent positions but also from remotepositions. When fourth-graders implemented cumulative rehearsal, however, they mainly rehearseditems from adjacent list positions successively and less frequently rehearsed items from remote posi-tions successively, thereby maintaining the original order of presentation.

M. Lehmann, M. Hasselhorn / Journal of Experimental Child Psychology 111 (2012) 552–560 557

Conditional response latency during studyFig. 1D displays the mean IRTs of successively rehearsed items within the interstimulus Intervals.

The lag–CRL curves reveal short IRTs for adjacent rehearsals (particularly forward) in comparison withlonger IRTs for remote rehearsals. The observations were supported by a two-way ANOVA that in-cluded the between-subject factor grade (2, 3, and 4) and the within-subject factor lag (forward adja-cent and remote). There was a significant main effect of lag, F(1, 26) = 55.65, p < .001, g2

p = .68, but notof grade or of the Lag � Grade interaction. Bonferroni post hoc comparisons revealed that all agegroups rehearsed items from forward adjacent serial positions faster than they rehearsed items fromremote serial positions (all ts > 3.69). Thus, despite the stronger tendency of fourth-graders to studysuccessive items one after another, children of all grades were equally fast in rehearsing the itemsfrom forward adjacent positions successively.

In sum, two major findings can be recapped regarding rehearsal processes during study. First, lag–CRP and lag–CRL seem to be apt for displaying rehearsal dynamics in an overt rehearsal free recalltask. When analyzing the order of rehearsals within Interstimulus Intervals, the lag–CRP provides ahighly appropriate measure of study dynamics characterized by episodically organized interitem asso-ciations. Second, and more important, temporally defined interitem associations within cumulativerehearsal seem to develop with age; already in second- and third-graders, the implementation ofcumulative rehearsal was characterized by the contiguity effect, but not exclusively so. In these chil-dren, consecutive rehearsals of items from remote serial positions, and therefore a larger permutationof items from diverse lags, also took place. However, in fourth-graders, who used cumulative rehearsalmore extensively, retrieval of neighboring items for consecutive rehearsal predominated.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

-5 -4 -3 -2 -1 0 1 2 3 4 5

Con

ditio

nal r

espo

nse

prob

abili

ty

Lag

2

3

4

(A) Intervals 2-5 Grade

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

-5 -4 -3 -2 -1 0 1 2 3 4 5

Con

ditio

nal r

espo

nse

prob

abili

ty

Lag

2

3

4

(C) Intervals 10-12 Grade

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

-5 -4 -3 -2 -1 0 1 2 3 4 5

Con

ditio

nal r

espo

nse

prob

abili

ty

Lag

2

3

4

(B) Intervals 6-9 Grade

0

2

4

6

8

10

12

-5 -4 -3 -2 -1 0 1 2 3 4 5

Con

ditio

nal r

espo

nse

late

ncy

(s)

Lag

2

3

4

(D) lag-CRL Grade

Fig. 1. Rehearsal dynamics. (A–C) Mean conditional response probability curves as a function of lag (lag–CRP) during study forInterstimulus Intervals 2–5 (A), Interstimulus Intervals 6–9 (B), and Interstimulus Intervals 10–12 (C) for second-, third-, andfourth-graders. (D) Conditional response latency curves as a function of lag (lag–CRL) during study for Interstimulus Intervals 2to 12 for second-, third-, and fourth-graders.

558 M. Lehmann, M. Hasselhorn / Journal of Experimental Child Psychology 111 (2012) 552–560

Recall

Recall performanceAn ANOVA with list (1, 2, and 3) as the within-subject factor and grade (2, 3, and 4) as the between-

subject factor showed that recall increased with age, F(2, 52) = 8.84, p < .001, g2p = .25, whereas neither

the effect of list nor the interaction was significant. Bonferroni post hoc comparisons revealed that,regarding their recall performance, only a marginal difference was found between second- andthird-graders, t(35) = 2.21, and no difference was found between third- and fourth-graders. However,fourth-graders recalled significantly more items than second-graders, t(35) = 4.39 (see Table 1).

Conditional response probability during recallFor the analyses of lag–CRPs during recall, recall order of the items was examined. Children of all

ages produced few repetitions (0.1, 0.1, and 0.2 for Grades 2, 3, and 4, respectively) and few intrusions(0.1, 0.1, and 0.1 for Grades 2, 3, and 4, respectively) early in recall. Most of the repetitions and intru-sions occurred when no novel item was recalled. Therefore, repetitions and intrusions were excludedfrom analyses. Figs. 2A to 2C display the lag–CRP curves separately for each grade for output positions1 to 4, that is, transitions 1–2, 2–3, 3–4, and 4–5, respectively. At later output positions, the youngestchildren especially recalled too few correct items and, therefore, did not provide sufficient data forreliable lag–CRPs (see also Howard & Kahana, 1999; Kahana et al., 2002). Across panels A to C inFig. 2, children of all age groups revealed the same tendency to start recall with a transition towarda neighboring item (lags –1 and +1) with a strong bias toward a forward transition. Especially inGrades 2 and 3, however, transitions at later output positions are less straightforward and seem tocome from randomly varying lags. In fourth-graders, transitions at later output positions retain theshape of a pronounced contiguity effect. This is confirmed by an ANOVA with grade (2, 3, and 4) asthe between-subject factor and lag (forward adjacent and remote), list (1, 2, and 3), and output posi-tion (1, 2, 3, and 4) as within-subject factors. This analysis revealed significant main effects of grade,F(2, 52) = 6.55, p < .01, g2

p = .20, lag, F(1, 52) = 40.97, p < .001, g2p = .44, and output position, F(3,

156) = 8.03, p < .001, g2p = .13, which were qualified by significant interactions of Grade � Lag, F(2,

52) = 3.88, p < .05, g2p = .13, Lag � Output Position, F(3, 156) = 7.02, p < .001, g2

p = .12, List � Output Po-sition, F(6, 312) = 2.59, p < .05, g2

p = .05, and Lag � List � Output Position, F(6, 312) = 3.32, p < .01,g2

p = .06. No other effect was significant. Bonferroni post hoc comparisons revealed that, overall, thedifference between forward adjacent and remote transitions increased both with age (all ts > 2.67)and with earliness of output position (all ts > 2.31). This effect was strongest in the first list as com-pared with the other lists (all ts > 2.21). In addition, especially fourth-graders displayed these strongeradvantages of forward adjacent transitions compared with second-graders, t(35) = 3.18, and were fur-thermore able to maintain these forward adjacent transitions over multiple output positions (1–3) (allts > 2.87).

(A) 2nd grade Output position (C) 4th grade(B) 3rd grade

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

-5 -4 -3 -2 -1 0 1 2 3 4 5

Con

ditio

nal r

espo

nse

prob

abili

ty

Lag

1

2

3

4

-5 -4 -3 -2 -1 0 1 2 3 4 5Lag

-5 -4 -3 -2 -1 0 1 2 3 4 5Lag

Fig. 2. Recall dynamics: Conditional response probability curves as a function of lag (lag–CRP) for Output Positions 1, 2, 3, and 4during recall for second-graders (A), third-graders (B), and fourth-graders (C).

M. Lehmann, M. Hasselhorn / Journal of Experimental Child Psychology 111 (2012) 552–560 559

Consequences of contiguity in study and recall

In a final step, we analyzed the relationship between dynamics in overt list learning, recall dynam-ics, and recall performance. Correlation analyses revealed that a high occurrence of the contiguity ef-fect during study was accompanied by a high mean contiguity effect for output positions 1 to 4 only atGrade 4, r(17) = .51, and not at Grade 2, r(18) = .21, or Grade 3, r(17) = .26. Hence, older children wereable to effectively use the episodic information during study to start recall and to retain it in a contig-uous manner. In addition, correlation analyses of lag–CRPs within Output Positions 1 to 4 and recallperformance again turned out to approach significance only at Grade 4, r(17) = .40, p = .10, and notat Grade 2, r(18) = .37, or Grade 3, r(17) = .07. In sum, fourth-graders not only retrieved neighboringitems for episodically driven rehearsal but also initiated and pursued recall in an episodically drivenfashion, and in doing so they were more likely to recall more items.

Discussion

Cumulative rehearsal has been shown to be critically important in enhancing recall performance. Inthe current study, we adopted the lag–CRP measure (Kahana, 1996) to analyze rehearsal dynamics inchildren, that is, the order in which children were vocalizing items for memory. We found character-istics in rehearsal that are highly typical for recall dynamics and are related to recall performance (e.g.,Kahana & Miller, in press), that is, the contiguity effect with a strong asymmetry effect component.Taking into account previous studies analyzing rehearsal–recall dynamics (e.g., Kahana, 1996; Laming,2006; Laming, 2008; Sederberg et al., 2010; Tan & Ward, 2000), we suggest that facilitating associativeprocesses based on temporal proximity, or rather on the preexisting list structure (Howard & Kahana,2002), seems to constitute a major impact of cumulative rehearsal on children. During the course of alist-learning process, children seem to arrange sequentially presented items in a temporally drivenmanner, with each rehearsal of an item serving as a cue for the following retrieval for rehearsal.The resulting string of items then provides a foundation for the actually presented item that is tobe added. Hence, when the recorded string of items is retrieved, children seem to extend their rehear-sal sequence again according to the temporal order of presentation. In children, however, the contigu-ity effect during study displays age-related variations. Contiguity effects turned out to be high early inthe list and to decrease thereafter (especially in younger children). This is consistent with age-relatedvarying rehearsal behavior (Lehmann & Hasselhorn, 2007). A larger and prolonged contiguity effect inolder children seems to reflect active rehearsal behavior characterized by retrieving items for rehear-sal on the basis of temporal associations. In younger children, the same effect seems to decline becauseof their failure to retrieve the string of earlier presented items.

The striking similarities between rehearsal and recall dynamics and their age-related differencesare also supported by contiguity during study and contiguity early in recall in fourth-graders. Again,one might suppose that immediate recall is just another retrieval sequence analogous to the previousretrieval for rehearsal during study (Howard, Venkatadass, Norman, & Kahana, 2007; Lehmann & Has-selhorn, 2010; Ward et al., 2003).

In sum, quantifying the development of rehearsal dynamics in children by applying the lag–CRPmeasure (Kahana, 1996) seems to be promising. It provides additional insights into rehearsal develop-ment beyond rehearsal style and intensity of rehearsal strategy use; it tells us that when childrenimplement a more complex rehearsal strategy, they rather rely on temporal order within a list of unre-lated material.

References

Howard, M. W., & Kahana, M. J. (1999). Contextual variability and serial position effects in free recall. Journal of ExperimentalPsychology: Learning, Memory, and Cognition, 25, 923–941.

Howard, M. W., & Kahana, M. J. (2002). A distributed representation of temporal context. Journal of Mathematical Psychology, 46,269–299.

Howard, M. W., Venkatadass, V., Norman, K. A., & Kahana, M. J. (2007). Associative processes in immediate recency. Memory andCognition, 35, 1700–1711.

Kahana, M. J. (1996). Associative retrieval processes in free recall. Memory and Cognition, 24, 103–109.

560 M. Lehmann, M. Hasselhorn / Journal of Experimental Child Psychology 111 (2012) 552–560

Kahana, M. J., Howard, M. W., & Polyn, S. M. (2008). Associative retrieval processes in episodic memory. In H. L. Roediger (Ed.),Learning and memory: A comprehensive reference (pp. 476–490). Oxford, UK: Academic Press.

Kahana, M. J., Howard, M. W., Zaromb, F., & Wingfield, A. (2002). Age dissociates recency and lag-recency effects in free recall.Journal of Experimental Psychology: Learning, Memory, and Cognition, 28, 530–540.

Kahana, M. J., & Miller, J. F. (in press). Memory, recall dynamics. In H. Pashler (Ed.), Encyclopedia of the mind. Thousand Oaks, CA:Sage.

Laming, D. (2006). Predicting free recalls. Journal of Experimental Psychology: Learning, Memory, and Cognition, 32, 1146–1163.Laming, D. (2008). An improved algorithm for predicting free recalls. Cognitive Psychology, 57, 179–219.Lehmann, M., & Hasselhorn, M. (2007). Variable memory strategy use in children’s adaptive intratask learning behavior:

Developmental changes and working memory influences in free recall. Child Development, 78, 1068–1082.Lehmann, M., & Hasselhorn, M. (2010). The dynamics of free recall and their relationship to rehearsal between 8- and 10-years

of age. Child Development, 81, 1006–1020.Ornstein, P. A., & Naus, M. J. (1978). Rehearsal processes in children’s memory. In P. A. Ornstein (Ed.), Memory development in

children (pp. 69–99). Hillsdale, NJ: Lawrence Erlbaum.Rundus, D. (1971). Analysis of rehearsal processes in free recall. Journal of Experimental Psychology, 89, 63–77.Rundus, D., & Atkinson, R. C. (1970). Rehearsal processes in free recall: A procedure for direct observation. Journal of Verbal

Learning and Verbal Behavior, 9, 99–105.Sederberg, P. B., Miller, J. F., Howard, W. H., & Kahana, M. J. (2010). The temporal contiguity effect predicts episodic memory

performance. Memory and Cognition, 38, 689–699.Tan, L., & Ward, G. (2000). A recency-based account of the primacy effect in free recall. Journal of Experimental Psychology:

Learning, Memory, and Cognition, 26, 1589–1625.Ward, G., Woodward, G., Stevens, A., & Stinson, C. (2003). Using overt rehearsals to explain word frequency effects in free recall.

Journal of Experimental Psychology: Learning, Memory, and Cognition, 29, 186–210.