matemal regulation of children's problem-solving behavior ... · matemal regulation of...

Matemal Regulation of Children'sProblem-solving Behavior and Its Impacton Children's Performance

Lisa S. FreundUniversity of Maryland Baltimore County

FREUND, LISA S. Maternal Regulation of Children's Problem-solving Behavior and Its Impact onChildren's Performance. CHILD DEVELOPMENT, 1990, 61, 113-126. This study focused on a socialinteraction theory of the development of cognitive self-regulation. Specifically, the e£fect of mother-child interaction on the child's ability to problem solve was investigated. The general predictionswere (1) children who interacted with their mothers throughout a problem-solving task wouldsubsequendy exhibit improved independent performance over practice-control children, who re-ceived corrective feedback from a female experimenter at the end of the task; (2) mothers would bemore responsible for task activities, would more ofl:en regulate their child's task behaviors, and offermore specific verbal content when task demands on child competence increased than when theydecreased. 60 3- and 5-year-olds either worked with their mothers or practiced alone and were givencorrective feedback on a sorting task in which miniature pieces of fiimiture were placed in a dollhouse. As predicted, children who interacted witfi their mothers subsequently created more correct,adult-like groupings independently than children who received corrective feedback. Mothers dis-played more task responsibility and regulation with younger children and when task demands onchildren of both age groups increased. Matemal verbal content became less specific when taskdemands decreased. Child performance was related t» (1) variation in matemal regulation of thechild; and (2) degree of specificity of matemal verbal content.

The development of cognitive self-reg- cognitive self-regulation, one must look to theulation during learning and problem-solving social context in which young children's prob-activities is an important aspect of cognitive lem-solving activities are conductod (Vy-growth and change. Cognitive self-regulation gotsky, 1978; Wertsch, Minick, & Ams, 1984;refers to the use of control processes neces- Wertsch & Rogoff, 1984; Wood, 1980). A ma-sary for successful completion of a task, such jor impetus for this view is Vygotsky's (1962,as planning task activities, monitoring success 1978) theory of cognitive development Thisor failure of actions, staying aware of task theory proposes that cognitive self-regulationgoals, and coordinadng strategies to reach develops through early social interactions be-those goals (Baker fie Brown, 1984; Brown, tween the young child and more competent1978; Wertsch, 1977). Cognitive self-regula- individuals such as parents. During even sim-tion is considered one component of metacog- pie parent-child activities, the parent's socialnidon, a concept referring to two separate interactions consist primarily of regulating theforms of cognition: (1) knowledge about one's child's behavioral responses, such as direct-own cognitive processes, and (2) regulation of ing the child's attention, maintaining goal di-those processes (Baker & Brown, 1984). Thus, rection, reminding the child where he or shecognitive self-reguladon refers to the latter is in the task, and/or evaluating the success orcomponent feilure of the child's task behaviors (Wertsch,

1977). According to Vygotsky's theory, theSeveral investigators have suggested that cognidve regulatory processes the child must

in order to understand the development of master to solve problems independendy are

This article is based on the author's doctoral dissertation approved by the faculty of the Gradu-ate School of the University of Maryland Baltimore County. I thank Linda Baker, my dissertationadvisor, for her guidance throughout this research and her many helpful comments on this article. Iam gratefiil to my dissertation committee, Mary Ann Metzger, Susan Sonnenschein, Douglas Teti,and Lowell Groninger, for their support and suggestions during this research project In addition, Ithank Kim CoUison and Susan Rasmussen for their assistance in coding videotapes. I also thank theanonymous reviewers who provided invaluable critiques of an earlier version of this manuscript.Address reprint requests to the author at the Johns Hopkins University School of Medicine, Depart-ment of Child Psychiatry, the Kennedy Institute, 707 N. Broadway, Room 106, Baltimore, MD21205.

[Child Development, 1990, 61, 113-126. © 1990 by the Society for Research in Child Development, Inc.All rights reserved. 0009-3920/90/6101-0014$01.00]

114 Child Development

similar to the reguladng acts originally im-posed upon the child by the parent duringjoint activities. Cognitive development pro-ceeds when parent regulatory acts become in-ternalized by the child to die degree that thechild can eventually regulate his or her cogni-dve processes independendy.

Vygotsky specifies that adults fiicilitetethe child's intemalization of coîtive self-regulation by the manner in which strate^ctask responsibilities are divided betweenadult and cluld and the manner in which thechild's task behaviors are regulated. Specifi-cally, intemalization is facilitated by threeadult acdons during joint problem solvingwidi a child: (1) t^ ing responsibility fw diosetask components diat are clearly beyond thechild's capabilities; (2) restrictir^ regulationof the child's behavior (i.e., directing atten-tion, maintaining goal direction, guiding,monitoring and evaluating behaviors) to thoseaspects of a task that are just beyond thechild's mastery level; and (3) transferring taskresponsibilities to the child during those as-pects of die task diat have been mastered bythe child. The adult's sensitivity to die "zone"between what the child can do widi assis-tance and what the child has mastered is a keyfector within Vygotsky's theory of cognitivedevelopment.

When Vygotsky refers to adult regulationof the child's problem-solving activities, he isproposing something more than the adult's in-struction of the child (Wertsch, McNamee,McLane, & Budwig, 1980). First, adult regu-lation results in the development of higher-level, general control processes (i.e., plan-ning, monitoring, eto.) and is not confined tothe acquisition of instructed, task-specificskills. Adult regulation aUows the child tolearn how to leam, th^ is, how to solve prob-lems beyond die specificity of instructedskills (Bransford, Stein, Shelton, & Owings,1981; Brown, Bransford, Ferrara, & Cam-pione, 1983). Seanid, in Vygotsky's concep-tion of adult regulation of the child, it is thedivision of problem-solving responsibilitiesin a social conteiA diat will best explain howsuch activities will later be performed by thechild indepetklendy. Alternatively, the tradi-tional concept of instruction s u ^ ^ t s that itis the acquisition of the content taui^t thatpredicts later independent perfomuuice. Ofcourse, adult instruction, divisu)n of task re-sponsibilities, and adult regulation of thechild can coexist during the same joint activ-ity. The distinction among the three, how-ever, allows for a more thorough understand-ing of the child's cognitive development.

Vygotsky's theory leads to the conclusionthat underst^ding how a parent interactswidi the child during joint problem-solvingactivities will provide insight into how thechild will function later while problem solv-ing on his or her own. The theory hasmotivated a small body of literature analyzingparent-child interactions during a variety ofproblem-solving tasks such as puzzle comple-tion (Kontos, 1983; Kontos & Nicholas, 1986;Wertsch et al., 1980), classification (RogoflF,Ellis, & Gardner, 1984), and number corre-sfMmdence (Saxe, Guberman, & Gearhart,1987). A consistent finding has been thatadults vary die nature of their inter^:4ion dur-ing joint problem solving in accordance withchild age. Adults are more directive, offermore guMiunce, and oSer more specific infor-mation to younger children (Kontos, 1983;RogofFet al., 1984; Saxe et al., 1987; Wertschet al., 1980).

A nuyor difficulty with interpreting thisliterature is that most adult-child interactionstudies vary child age as a means of varyingchild competency without actually measuringindependent child task performance at all.Widiout assessing the child's level of mastery,it cannot be determined (1) whether die vjuia-tion in adult regulation is appropriate, or (2)whether differences in adult regulation are re-lated to difiBsrences in actual child compe-tency or odier age-related variables such asdevelofHnentid language level or adult expec-tations. Indeed, analyses of mothers' interac-tions with younger and older children duringclassification tasks led Rogoff et al. (1984) tosuggest diat task regâtion is related more toage-related expectetions of task difficulty thanactual child competence.

Evidence for di£brences in adult regula-tion as a functkin of chao^ng child compe-tence within the same adult-child dj^d wouldeliminate the confounding of age and compe-tence. In general, however, such a finding hasnot lên consistently obtained. For example,Wertsch et al. (19S0) found no evidence thatmothers differed in their degree of child regu-lation during die first and second halves ofdieir interaction during a puzzle-completiontask even thou^ one would expect thechild's task cranpetency to increase duringtiie second half. Saxe et al. (1987) disen-tangled aê and con^wtency variables byassessing 2- ami 4-year-old children's un-assisted performmice before the childreninteracted with their mothers on countingtasks that varied in difficulty. Mothers of lessable and younger children provided morespecific instruction than mothers of more able

and older chUdren. Mothers of children inboth age groups instructed with greater spe-cificity in the more difficult task. Althoughvariations in the specificity of matemal in-struction are an indirect measure of matemalregulation of the child, they do not direcdymeasure the division of task responsibilities.When a mother is more specific in her instruc-tion, she may be guiding the child's actions ortaking over the responsibility of a task actionherself.

A major limitation in the adult-child in-teraction literature has been the lack of evi-dence for a beneficial effect of adult inter-action on the child's independent problemsolving. In general, the benefits of adult in-teraction on child cognitive development orlearning have been assumed or based on cor-relational results. For example. Wood andMiddleton (1975) assessed die success andfailure of child task behaviors and maternalbehaviors contingent to those child behaviorsduring a joint problem-solving task. Childrenof mothers who provided more assistance fol-lowing failures and less during successeswere more likely to demonstrate better inde-pendent task performance following the jointtask interaction. In controlled studies, therehas been no evidence of better child perfor-mance among the adult-child interactiongroup than a practice control group (Kontos,1983; Kontos & Nicholas, 1986). A few stud-ies have shown that children receiving adultguidance adjusted to their level of compe-tency performed better on a problem-solvingtask than children who observed a demonstra-tion (Goncu & Rogoff, 1987; Wood, Wood, &Middleton, 1978). However, the ecologicalvalidity of these studies is compromised be-cause the guiding adult was given veryspecific instructions on how to interact withthe child.

The need for a comprehensive assess-ment of the relation between the adults* in-teraction widi the child and the child's abilityto solve problems independendy, as well asconsideration of the limitations identified inprevious adult-child interaction studies, led tothe design of the present study. There weretwo m^or purposes of the study: (1) to inves-tigate die effect of mother-child interactionduring a problem-solving task on subsequent,independent child performance; (2) to inves-tigate the variability in the division of taskresponsibilities and matemal regulation of thechild as a function of task difficulty, child age,and task component In order to achieve thesegoals, the independent performance of 3- and5-year-oId children who had previously in-

LisaS. Freund 115

teracted with their mothers throughout acategorization task was compared to that ofsame-aged children who had received onlycorrective feedback from an adult experi-menter. This specific control comparison wasimportant for establishing that the full processof social interaction with an adult improveschild performance more than simply correct-ing the performance itself at the end of thetask. The task required the sorting of severalpieces of miniature furniture into a one-story,doll house-like structure. The goal was togroup the furniture to represent rooms foundin a typical home (e.g., bathroom, bedroom,etc.). It was predicted that children who in-teracted with their mothers throughout thecategorization task would subsequendy ex-hibit improved independent performanceover children who received corrective feed-back only.

Easy and difficult versions of the taskwere used to assess mother variation in theallocation of task responsibilities, matemalregulation of the child, and matemal verbalcontent as a function of the child's changingtask competency. Among the mother-childdyads, it was expected that mothers would ex-hibit less task regulation, take less personalresponsibility for task activities, and offermore specific, concrete verbal content duringan easy version of the task than a difficult ver-sion. Because of age-related factors not associ-ated with competence, it was expected thatmothers would more often regulate and offermore specific, concrete verbal content toyounger children than older children regard-less of task difficulty. Mothers of older chil-dren were expected to offer more generalmonitoring and planning verbalizations. Workby other researchers has suggested that cer-tain task components may be more conduciveto adult regulation than others (Ams, 1981;Wertsch et al., 1980). Therefore, task regula-tion and responsibility was assessed by taskcomponent with the expectation that motherswould be more responsible for and more of-ten regulate the task component deemed mostcritical to successful task solution. Finally, therelation of child independent task perfor-mance with (1) matemal verbal content and(2) variation in matemal task responsibilityand regulation of the child was explored.

MethodSubjects

Thirty 3-year-olds and 30 5-year-oIdswere assigned in equal numbers to one of twoconditions: Mother Interaction (Interaction)or Corrective Feedback (Feedback). Mothers


and children were recruited from several pre-schools in middle-class, racially mixed sub-urban communities. Random assignment togroups was not conducted because only 38 ofdie 320 mothers contacted through recruit-ment ^reed to participate in the Interactioncondition. Although random assignment wasnot possible, statistical analyses indicated thatthe groups were comparable on initial per-formance measures and mothers' demo-graphic data. A t test showed no evitfencethat mothers differed in age. Chi-square anal-yses revealed no evidence of differences ineducation level or work status between theInteraction or Feedback mothers.

The 3-year-old Feedback group (sevengirls, e i ^ t boys) had a mean age of 3.5 years(SD = 2.0 months). The 3-year-oid Interac-tion group (six girls, nine boys) had a meanage of 3.4 years (SD = 2.5 months). The 5-year-old Feedback and Interaction groups(eight giiis, seven boys each) had mean agesof 5.3 years (SD = 3.5 months) and 5.4 years(SD = 3.5 months), respectively. The meanage of the 30 participating mothers was 35years. All of these mothers had at least somecollege education; 12 of the mothers had agraduate-level degree. There was a fairlyequal distribution among the participatingmothers of those who did not work outsidethe home (eight mothers), those who workedpart time (10 miners), and those who workedfull time (12 mothers).

MaterialsThe task presented to all participants

utilized scale miniature replicas (1 inch to 1foot) of common household furniture and ap-pliances. The goal of the task was to sort theitems into two to six rooms in a one-level, dollhouse—like structure. The doll house was awooden frame, 78 cm long, 62 cm wide, and 8cm h i ^ . Widiin this frame were placed woodpartitions such that six equal-sized spaceswere created. These spaces were referred toas "rooms." The materials included 36 piecesof the miniature furniture and appliance re-productions found in a typical middle'classhome and six same-scale distractor items thatwould not be found in a typical house (e.g., alamppost, a boat, a deer, etc.). Pilot testingshowed that adults arranged the items intothe same room groupings with a h i ^ degreeof agreement. Agreement among the adultswas 92% or better on all items. For example,all adults placed the bathtub, sink, toilet, andtowel rack together and called the room abathroom. Other room groups included akitchen (e.g., stove, refrigerator, kitchen sink,table, and chairs), a living room (e.g., couch.

matching overstuffed chairs, coffee table, TV,bookshelf), a dining room (e.g., table, chairs,and hutch), a baby nursery (e.g., matchingcrib, rocker, playpen), and an adult bedroomand a child bedroom (each with bed, endtables, and dressers) differentiated by color ofitems. Not all items are listed here; a com-plete listing is available in Freund (19^).Adult agreement was only 65% on two of theitems—the television and bookshelf. Place-ment of these two items was subsequendyignored when assessing children's arrange-ments of the items. There were more possibleroom groupings of items than doll houserooms in order to allow for variation in sortingpresentations. Three-year-olds and 5-year-olds who Efflrticipat«d in pilot work and a sub-sequent study using the room-sorting task(Freund, Baker, & Sonnenshein, 1988) wereable to name or provide an appropriê fiinc-tion for each of the items, including distrac-tors. Distractors were included to make thesorting task more challenging.

The study design required diffioilt andeasy versions of the task, with difficulty beingcompâble for each age group. The difficultand easy levels (^the task for each age groupwere determined by pilot work in which 3-and 5-year-oIds s o r ^ the same miniatureitems used in diis study. The difficult levelwas established first A six-room doll housewith 35 fiimiture itsms and six distractoritems was determined to be of a moderate toh i ^ difficulty level for 5-year-olds; the 5-year-olds in the Freund et al. study attainedan average of 43% correctly ôuped items.The same presentetion given to 3-year-oIdstook too long fbr die children to finish; most 3-year-olds stopped before they had traced halfof the items. It was important that the 3-year-olds have the s^ne opporbmity as 5-year-oldsto display their level of competency fixim startto finish of die task. By reducing the six-roomdoll house to a 28-item, four-room doll house(24 items and four distractors), most 3-year-o\6s could finish die task in approximately thesame amount of time that the 5-year-oldsfinished die six-room doll house. The reducednumber of rooms resulted in conq>arabledifEcuhy for the two age groups. Widi diefour-room doll house, 3-year-oIds in theFreund et al. study achieved an average scoreof 40% conrecdy ©wiped items, which wasnot significandy different from the 5-year-oldperformance.

The easy le\^ls of the task were derivedby cutting in half the number of room group-ings to be sorted by each age group. Pilotingshowed that a two-room, eight-item doll

LisaS. Freund 117

house presented to 3-year-olds and a four-room, 16-item doll house presented to 5-year-olds resulted in an average increase of 35%correcdy grouped items over the mean per-cent correct for the difficult level for both agegroups.

ProcedureAll children in the Feedback condition

and all mother-child Interaction group dyadswere seen individually in sessions lasting 40-55 min. Each session was videotaped with thecamera visible to all participants. A three-phase procedure was used. In phase 1, allchildren were presented a difficult level ofthe task to complete independendy. In phase2, the Interaction groups were presented twolevels of the task (easy and difficult) to com-plete with their mothers. Feedback groupscompleted the same two tasks independendy.In phase 3, all children were presented adifficult level of the task to complete aloneonce again. Different room combinationswere presented in each phase. For example,the difficult version of the task in phase 1 hada different set of possible item groupings thanthe difficult version of the task in phase 3. Foreach phase, participants were seated in frontof the empty doll house, next to which werethe sorting items. Items were randomly ar-ranged in the same array across subjects. Thesame female experimenter ran all phases ofthe experiment with all participants.

In phase 1, children in each group wereintroduced to the task through the context ofhelping a puppet move his furniture into hisnew house. The child was told to pick out allthe things that belonged in a house and putdiem into the rooms where they belonged,just like a regular house. The child was alsotold to leave outside die doll house anythingnot belonging inside a house. Children werenot told the names of the rooms into whichthey were to sort the items, nor were childrenexpected to arrange rooms in any specificflooiplan design. Children were not expect-ing any formal evaluation of their sorts. Whilethe child sorted items, the experimenter re-mained at a distance but in view of the child.If the child's mother was participating, sheremained separated from the child by a largescreen and was not able to observe her childduring this phase.

During phase 2, the procedures for theInteraction and Feedback conditions differedas follows:

Interaction condition.—Children in theInteraction groups were presented two levelsof the task (easy and difficult) with their

mothers while the experimenter left theroom. Each level was presented separatelywith order counterbalanced across subjects.Before phase 2 sorting began, the mother wasshown the task materials while her childplayed with toys unrelated to the sorting taskbehind a large screen. Modiers were in-structed to help their child group items as anadult would for rooms in a typical house.Mothers were not told to teach their child butencouraged to help in any manner with whichthey felt comfortable. They were also told thatwhen they finished, their child would oncemore complete the task alone with differentitems. Mothers were then given a card listingthe names of the rooms into which the itemsshould be grouped. Pilot work had shown thatmothers were not as likely to guide theirchild's behavior toward an adult-like sort ofthe items when they were not informed ofroom categories. Mothers were responsiblefor informing the experimenter when finishedwith a difBculty level.

Feedback condition.—Children in theFeedback group were presented the two tasklevels to solve independendy. The order ofpresentation of the task levels was counterbal-anced among the children. After completingeach difficulty level, the child was praised forworking hard. Most of these children madeerrors in their groupings for each level. Theywere told that they had made mistakes, andthen the experimenter corrected any itemgroupings to conform with the establishedadult groupings. Children whose sorts wereentirely correct were told they were correct.Following a short break after phase 2, chil-dren in both conditions completed a difficultlevel of the task independendy once again inphase 3.

ScoringChild performance.—The criteria for as-

sessing the accuracy of child room groupingswere based on the groupings adults createdwith high agreement in pilot testing. Theseadult groupings were considered the correctgroupings. To determine child accuracy foreach sorting presentation, each room was as-sessed separately for degree of correct group-ing. If a child's room contained at least threeitems that adults had consistendy grouped to-gether, those items were considered correctfor that room grouping. For example, adultsconsistendy grouped the tub, the toilet, thewash basin, and the towel rack together andcalled the grouping a bathroom. If a childgrouped the toilet, tub, towel rack, and akitchen chair in one room (four items placed),then the toilet, tub, and towel rack were con-


sidered correct and a score of 3 was given tothe room. In this way children were givencredit for groupings that were at least partiallyconsistent with adult groupings. A minimumof three correcdy grouped items per room wasused because several of the items to be sortedcame in matehing pairs (e.g., the living roomchairs, the bedroom end tables). Since thepurpose of the correct grouping score was toassess the accuracy of groupings based onfunctional relations and not the ability tomatch items, the three-item minimum pre-vents confounding a functional-groupingstrategy widi a matching-items strategy.

For each sorting presentation, everyroom in the doll house was given a score rep-resenting the number of correcdy groupeditems in that room. If a room contained itemsassociated with two different types of rooms,then the larger number of items was consid-ered correct. In the rare case in which allavailable items were grouped in one room, nocredit for correct groupings was given. A pro-portion-correct score was computed by sum-ming the scores associated with each room ofthe doll house plus the number of distractorsremaining outside the doll house. This sumwas then divided by the total number of items(including distractors) that had been pre-sented to die child.

Matemal task responsibilities and regu-lation of the child.—In order to differentiatethe modier-child activities during the task,the mother-child interaction was brokendown into discrete units or episodes in a man-ner similar to that described in Ams (1981)and Wertsch et al. (1980). An episode wasdefined as a segment of the interaction fo-cused on the selection and placement of oneitem in a room within the doll house. For suc-cessful completion of the sorting task, cer-tain steps were necessary in order to place anitem in an appropriate room. These steps (orepisode components) included: (1) selectionof an item; (2) selection of an appropriateroom category (e.g., kitehen, bathroom, etc.)or functional associsrtion (e.g., where yousleep, where you eat, etc.) to which an itembelonged; and (3) placement of the itemwithin the chosen room. The order of ccmipo-nents 1 and 2 could vary depending on thestrategy chosen to complete die task (i.e., onecan decide upon a room category and look foritems belonging to that room, or one canselect an item and look for the room to whichit belongs). The room-selection componentwas considered the most crucial componentto successful task completion because successhinged upon identification of appropriate

functional relationships (room categories).Task success did not hinge on the order inwhich items were selected or where exacdythey were placed within rooms.

In order to evaluate matemal task activi-ties, it was necessary to differentiate betweenwhen the mother performed an episode com-ponent for the child and when the motherguided or prompted the child to perform acomponent The mother was considered re-sponsible for a component if she physicallyperformed the component fbr the child. Thatis, mother was responsible for item selectionif she picked up an item when the child madeno reference to it, was responsible for roomselection if she placed an item in a roomwhen the child made no reference to aroom or functional association, or was respon-sible for item placement if she placed an itemin a room when the diild made no referenceto a specific location. The mother was consid-ered to be regulating the child's performanceof a component if she directed or guided thechild's action. More specifically, a mother wasconsidered regulating a child action if shepointed to, named, directed, or in some man-ner cued the selection of an item, the selec-tion of a room or functional association, or theplacement of an item in a specific room loca-tion. In addition, it was possible for a child toself-regulate performance of a component. Acomponent was considered self-regulated ifthe child selected an item, selected a room, orplaced an item without any direction or guid-ance from mother. Thus each episode com-ponent was coded for one of three mutuallyexclusive, task-control categories: motherresjKjnsible (MH), mother regulation of child(MRC), or child self-regulated (CSR).

Frequencies for each task-control cate-gory for each of the three episode componentswere determined fbr each dyad. These fre-quencies were converted to proportions bydividing by the total number of episodescompleted by the dyad. Easy and difficultversions of the task were coded separately.The investigator coded all mother-child in-teractions. An independent coder coded arandom selection of 25% of the mother-childdyad interactions. The percentage of agree-ment across episodes was 86%. Broken downby task control category across episode com-ponent, percentage of agreement was 93% forMR, 84% for CSR, and 81% for MRC. Al-though it was necessary to differentiate theCSR category from die MR and MRC catego-ries for coding purposes, die CSR propor-tional frequencies offered no unique informa-tion (i.e., as MR or MRC category proportions

LisaS. Freund 119

increased or decreased, the CSR proportionsdecreased or increased perforce). For this rea-son, only the MR and MRC proportional fre-quencies were retained as dependent vari-ables representing control of task activities,and the CSR proportions were dropped fromanalyses.

Matemal verbal content.—Matemaltask-relevant verbalizations were analyzed inorder to determine how mothers verbally in-structed their children throughout the task in-teraction. In this study, verbal instruction wasdefined as any task-relevant speech. In orderto code mothers' verbal content, transcriptsof mothers' verbalizations were divided intomessage units. A message unit was defined asany single, meaningful statement or question(D^vis & Lange, 1973), including single-wordutterances. Categorization of message unitswas made while viewing the videotape of themother-child interaction.

Five mother-child dyads participated in apilot study used to define the content codingsystem. The coding system differentiatedmothers' task-irrelevant speech ("It's rainingoutside") from task-relevant speech. The con-tent categories of task-relevant speech werebased on Kontos's (1983) system. Kontos dif-ferentiated between "task-variables" content(e.g., references to item attributes, item loca-tions, etc.) and "strategy-variables" content(e.g., references to strategies, goal direction,monitoring success/failure). No specific strat-egies were defined in Kontos's system; there-fore, strategy references could have includedboth specific strategies and more generalplanning and organizing strategies. Thus,"references to strategies" and the other typesof "strategy-variables" content were groupedtogether. From the pilot study for the presentreseareh, matemal references to several spe-cific strategies were identified. In the inter-est of greater specificity, the verbalizationsconsidered "strategy variables" under Kon-tos's system were further differentiated be-tween references to specific strategies andreferences to general planning/organizingstrategies, goal direction, and monitoring.This resulted in three mutually exclusivecategories—"task-specific materials," "strate-gies," and "goals, plans, and monitoring."These categories were considered to be hi-erarchically ananged in degree of specificityof content, with "task-specific materials" themost specific and "goals, plans, and moni-toring" die least specific. Category descrip-tions were as follows:

Task-specific materials.—This categoryincluded all references to item/room descrip-

tions, labels, functions, and/or locations (e.g.,"It's a big bed," or "That stove goes in akitchen").

Strategies.—The strategies included: (1)associations between a specific room andselection of an associated item (e.g., "Findsomething for the living room"); (2) associa-tions between a specific item and selection ofits associated room (e.g., "What room does therefrigerator belong in?"); (3) reference tomatehing properties of items (e.g., "Are thereother red chairs that go with that one?"); (4)reference to child's knowledge of how furni-ture is grouped or used in their home, afriend's home, or a relative's home (e.g.,"Where do we keep our refrigerator athome?"). Only verbalizations related to diesefour strategies were coded in this category.

Plans, goals, and m,onitoring.—This cate-gory included references to plans which wereverbalizations specifying a sequence of activi-ties (e.g., "Let's make the bedroom and thenthe kitehen"). References to goal directionwere included, such as "Let's make akitchen." Also included were monitoringstatements such as, "That's not right. Thatchair doesn't belong there."

Other.—This final category included ut-terances encouraging the child to participate,or offering unelabotated positive feedback OTattention-directing statements (such as "O.K."or "Look"). Verbalizations that could not becoded into one of the three categories definedabove were included.

The investigator was the principal coder.An independent coder coded a random selec-tion of 25% of the Interaction subjects' tran-scripts. Agreement was 96% for designatingtask-relevant from task-irrelevant utterances.The percentage of agreement across task-relevant categories was 91%. The breakdownof agreement by task-relevant verbalizationcontent code was 96% for references to task-specific materials, 81% for strategies, 88% forplans, goals, and monitoring, and 97% forother. Frequencies of each verbal content cat-egory were determined for each dyad. Thesefrequencies were converted to proportions bydividing by the total number of matemal task-relevant verbalizations. Proportions for easyand difficult versions of the task were deter-mined separately.

ResultsChild Performance

The phase 1 and phase 3 independentperformance of the 3- and 5-year-olds in theInteraction and Feedback group is repre-


TABLE 1

MEAN PHASE 1 AND PHASE 3 PROPORTION CORRECT BY CORRECTIVEFEEDBACK AND MOTHER INTERACTION CONDITIONS FOR 3- AND 5-YEAR-OLDS

AGE

3 years:MSD

5 years:MSD

CORRECTIVE FEEDBACK

Phase 1 Phase 3

.41 .36

.14 .15

.51 .64

.15 .18

MOTHER INTERACTION

Phase 1 Phase 3

.46 .70

.17 .19

.52 .94

.16 .08

sented by mean proportions of items coirecdygrouped in Tabie 1. In order to assess dieimpact of mother interaction on subsequentindependent child performance, a 2 (age) x 2(condition) analysis of covarianee was con-ducted with phase 3 proportion-correct scoresas the dependent variable and phase 1 pro-portion correct as the covariate. The use ofproportions necessitated the transformation ofthe scores via an arc sine transformation assuggested by Cohen and Cohen (1975).Transformed scores were used in statisticalanalysis, but untransfomied mean scores arereported. The covariate was significandy re-lated to phase 3 proportion correct, f(56) =3.10, p < .01. Only the two main effects weresignificant. As predicted, children who in-teracted with their mothers had a highermean proportion correct (M = .82) than chil-dren who received corrective feedback (M =.50), F(l,56) = 12.52, p < .01. It is apparentfrom die means in Table 1 that mother in-teraction greatly improved both 3- and 5-year-olds' sorting perfonnance. Indeed, the 5-year-old Interaction group approached ceiling lev-els of sorting performance, and 3-year-olds inthe Interaction group performed similarly to5-year-olds in the Feedback group. Acrossconditions, 5-year-olds had a h i ^ e r meanpropcwtion correct (M = .79) than 3-year-oids(M = .53) in phase 3, F(l,56) = 26.52, p < .01.Thus, the 5-year-oIds improved performancemore than 3-year-olds when experiencingeither corrective feedback or interaction withtheir mothers.

Matemal Task Responsibilities andRegulation of the Child

Variation in matemal task responsibilitiesand regulation of the child were investigatedby comparing the mean proportional frequen-cies of the MR and MRC categories betweenthe easy and difficult levels of the task foreach age group. These comparisons were con-

ducted for the three episode components—item selection, room selection, and itemplacement. Preliminary analyses revealed nosiîficant effect associated with order of taskdifficulty presentation. Thus results are re-ported widi order of presentation eliminated.

Mean proportional frequencies of bothdependent variables for the item-placementcomponent fell below .10 regardless of age ortask difficulty. Since motiiers had so litde todo with item placement, allowing children tobe responsible for item pl^êment about 90%of the time, proportions for the componentwere dropped from further analyses. Themean proportional frequencies of the MR andMRC categories are shown in Table 2 for dieitem-selection and room-selection compo-nents X age group and task difficulty level.

The means indicate that children, regard-less of age or task difficulty, self-regulated theitem-selection and/or room-selection compo-nents in approximately 60% of the bisk epi-sodes on the average. When children werenot self-regulating, mothers regulated theirchild's item and/or room selection in apj»t)xi-mately 31% of the task episodes and were re-sponsible for tJie components themselves inoniy 9% of the task episodes (averaging overdifficulty level and child age group).

An^ysis of the relation between die MRand MRC vambles suêsted that die vari-ab^s were indeed ta^rpiag different behav-iors. MR and MRC variables were arc sinetransfiarmed. CoHapsing proportional fre-quencies over difficulty level, the Pearsonproduct-moment correlation coefficients be-tween MR and MRC were .18 for the itemselection component and — ,19 for the roomselection component, neither of which weresigni£bant. Correlations were also conductedwidi the MR and MRC proportional frequen-cies at each diSBcuky level with phase 1 pro-portiOTi-correct scores. None of these correla-

LisaS. Freund 121

TABLE 2

PROPORTIONS OF TASK EPISODES THAT MOTHERS WERE RESPONSIBLE FORTASK COMPONENTS AND MOTHERS REGULATED CHILD TASK COMPONENT

ACTIVITY BY TASK DIFFICULTY LEVEL AND CHILD ACE

AND AGE

Item selection:3 years:

MSD

5 years:MSD

Room selection:3 years:

MSD

5 years:MSD

MR

.13

.08

.08

.03

.02

.03

.02

.03

Easy

DIFFICULTY LEVEL

MRC

.21

.18

.18

.17

.40

.17

.24

.13

MR

.17

.18

.13

.14

.06

.11

.07

.11

Difficult

MRC

.30

.20

.26

.11

.55

.14

.34

.11

NOTE.—MR = mother responsible for task component; MRC = mother regulationof child's task component activity. Proportions within task difficulty level for each agegroup add to 1.0 when child self-regulation (CS-R) proportions are included.

tions exceeded ± .20 or was significant.There appeared no evidence of a relation be-tween a child's initial independent skill andmatemal regulation of or responsibility for theitem- and room-selection components whenmother and child worked together. Therefore,phase 1 proportion correct was not consideredfor inclusion as a covariate in subsequentanalyses of mean differences.

In order to investigate the mean differ-ences, a mixed multivariate analysis of vari-ance was conducted once the MR and MRCproportional frequencies were arc sine trans-formed. The between-subjects factor waschild age. The within-subjects factors weredifficulty level and task component. Plannedwithin-subjects contrasts were applied to thedependent variables as follows: (1) easy ver-sus diflBcult conditions for item-selectioncomponent; (2) easy versus difficult condi-tions for room-selection component; (3) itemselection versus room selection (across diffi-culty conditions). Multivariate results wereassessed with the Wilks's criterion. All simplemean comparisons were conducted with pro-tected t tests at the .05 level.

The multivariate effect associated withage was not significant. However, the mul-tivariate interaction of the age factor with thewithin-subjects contrasts of the MRC variableindicated that mothers varied how often they

regulated their child's task activity dependingon child age, difficulty level, and die taskcomponent involved, F(3,26) = 3.15, p < .05.Univariate analyses showed a significant ageX difficulty level interaction for the room-selection component only, F(l,28) = 8.47, p< .01. Mean comparisons Indicated thatmothers of both 3- and 5-year-olds regulatedtheir child's room selection proportionatelymore often during the difficult level of thetask than the easy level, but mothers of 3-year-olds increased regulation to a greaterdegree during the more difficult task tlianmothers of 5-year-oIds. Regardless of childage, mothers did not vary the regulation oftheir child's selection of items between diffi-culty levels. A significant interaction of theage factor and the item and room componentcontrast of die MRC variable, F(l,28) = 5.92,p < .05, showed that only mothers of 3-year-olds regulated their child's room selection to agreater degree than item selection. Mothers of5-year-olds did not differ in their regulation ofthese two components. Thus, the analysisof the MRC variable indicates tliat mothers ofchildren firom both age groups more often reg-ulated the crucial task component of roomselection during the difficult version of thetask, and that across diflficulty levels, mothersof younger children regulated that task com-ponent more than the less crucial task compo-nent of item selection.


TABLE 3

MEAN PROPOBTIONS OF MOTHERS' VERBALIZATIONS BY CONTENT CATEGORY, CHILD ACE, AND TASKDIFFICULTY

ACE

3 years:MSD

5 years:MSD

Task

.42

.15

.32. .16

Strategy

.23

.13

.29

.13

Easy

Plans/Coals/Monitor

.29

.13

.34

.17

DIFFICULTY LEVEL

Other

.06,03

,05.05

Task

.43,10

.41

.13

Strategy

.38

.09

.36

.13

Difficult

Plans/Coals/Monitor

.16

.08

.21

.11

Other

.03,05

.02,03

NOTE.—Task = task-specific materials category; Strategy = strategies category; Plans/Goals/Monitor = plans,goals, monitoring category.

Mothers also varied how often they wereresponsible for task activity depending on thetask component and difficulty level, as shownby the significant miiltivariate test of the MRwithin-subjects contrasts, F(3,26) = 7.54, p <.01. Univariate analysis of the MR contrastsrevealed that mothers among bodi child agegroups were more often responsible for theroom-selection component in the difficultcondition than tlie easy condition, F(l,28) =9.^ , p < .05, yet did not vary how often theywere responsible for the item-selection com-ponent between difficulty levels. Motiierswere more often responsible for the item-selection component than the room-selectioncomponent across difficulty levels, F(l,28) =6.34, p < .05. The MR analysis showed thatmothers were more often responsible forroom selection during die difficult version ofthe task than the easy version, even thoughacross difficulty levels, mothers were more of-ten responsible for the less crucial componentof item selection.

The relation between variation in taskregulation and child perfonnance was ex-plored. Mothers who regulated their child'sitem and room selection more during thedifficult task and less during the easy task (n= 17) demonstrated the theoretically ex-pected variation in regulation based on ihetask difficulty manipulation and were there-fore designated task-consistent regul^ors.The renwaning mothers (n = 13), who eitherregulated more during the easy tesk than diedifficult task or did not vary regulation at all,were task-inconsistent regulators. An analysisof covarianee was conducted on children'sphase 3 proportion-correct scores x type ofmother regulation. Phase 1 proportion correctserved as the covariate and was significant.

F(l,27) = 17.9, p < .01. The mean phase 3proportion-correct scores were .86 for chil-dren of task-consistent regulators and .77for children of task-inconsistent regulators,F(l,27) = 5.81, p < .01. Thus, better childperfonnance was associated with task-con-sistent matemal regulation than with task-in-consistent matemal regulation during phase 2.

Matemal Verbal ContentTask-relevant verbalizations accounted

for 92% of all matemal verbalizations. Anamdysis of the mean number of task-relevantverbaliz^ons x child age group and taskdifficulty revealed a significant diflference inthe number of verbalizations made as a func-tion of d^culty level, F(l,28) = 9.38, p <.001. Modiers emitted more verbalizationsduring the difficult task (M = 63.3) dian dieeasy task (M = 25.09). This was expectedgiven that the difficult task contained moreitems and required more episodes to cran-plete for both age groups. There was no evi-dence that modiers of children in one agegroup spoke more than the other modiers,even thcms^ 5-years-oIds had more items tosort than 3-year"0lds.

The mean proportional frequencies foreach content category are shown in Table 3.Because very few "other" verbalizations oc-curred, this variable was dropped from furdieranalyses. Pearson product-moment correla-tions among the categories, collapsed acrossdifficulty level, showed diat the "plans, goals,and monitoring" category was negatively cor-related with the "task-specific materials" cate-gory, r(28) = -.42, p < .05, and the "strate-gies" category, r(28) = - .43, p < .05. Thelatter two categories were not significandy re-lated. Thus, no one content category ac-

Lisa S. Freund 123

counted for a major portion of the variance ofany other category.

In order to investigate the mean diflfer-ences for "task-specific materials," "strate-gies," and the "plans, goals, and monitoring"content categories, proportional frequencieswere transformed via an arc sine transforma-tion. Because the proportional frequencieswere based on widely differing numbers oftotal task-relevant verbalizations, the assump-tions of a standard analysis of variance couldnot be met. A goodness-of-fit statistic, the Q-statistic (Hedges & Olkin, 1985), was used in-stead, which could adjust for the differencesin total task-relevant speech.

The three verbal content categories wereassessed separately by a 2 (age) x 2 (order) x2 (difficulty level) Q-statistic analysis. An ad-justed significance level of p < .016 wasadopted to maintain the overall significancecriterion at .05. Mean comparisons, when nec-essary, were conducted with the Bonferronimethod using a. p < .01 criterion. The orderfactor was not a significant effect and waseliminated frt)m reported results.

The verbal content analysis demon-strated that mothers varied their content ver-balizations as a function of child age and/ortask difficult level depending on the contentcategory. Analysis of the "task-specific ma-terials" category revealed that mothers of3-year-olds offered more concrete, specificcontent to their children than mothers of 5-year-olds, Q(l, N = 30) = 10.25, p < .001.Contrary to expectations, mothers did not in-crease their references to task-specific materi-als during the more difficult version of thetask. Mothers proportionately increased theirreferences to strategies, however, as a func-tion of an interaction of task difficulty leveland child age, Q{3, N = 30) = 27.28, p <.001. Mean comparisons indicated thatmothers of 3-year-olds made references tostrategies more often during the difficult thanthe easy task, but mothers of 5-year-olds didnot vary their strategy references betweendifficulty levels. As predicted, mothers of 5-year-olds made a significandy greater propor-tion of references to plans, goals, and moni-toring than mothers of 3-year-olds, ^ (1 , N =30) = 13.62, p < .001, and modiers of bodiage groups made such references more oftenduring the easy than the difficult version ofdie task, Qil, N = 30) = 38.96, p < .001. Noother effects for the three verbal content cate-gories were significant.

The relation between matemal verbalcontent and child phase 3 performance was

investigated. As a means of simplifying theanalysis, the three categories were dichoto-mized into lower-level, more specific cate-gory and a higher-level, less specific category.The three verbal content categories werereorganized on the basis of Kontos's (1983)two categories, "task variables" and "strat-egy variables," which distinguished betweenlow-level and high-level content, respec-tively. The "task-specific materials" categoryused in this study represents the lower-levelcategory of "task variables." The two catego-ries, "strategies" and "plans, goals, andmonitoring," were representative of Kontos'shigher-level category, "strategy variables."Mothers were designated as offering morehigh-level, strategy-variables content or morelow-level, task-variables content. To computeeach mother's level-of-content score, the fre-quencies for each of the three original codeswere summed across the easy and difficultlevels of the task and divided by total task-relevant utterances across both difficulty lev-els. The resulting proportion for "task-specificmaterials" code represented low-level con-tent; the sum of the proportions for the "strat-egies" and "goals, plans, and monitoring" cat-egories represented high-level content. Theproportional frequency of high-level contentwas subtracted from the proportional fre-quency of low-level content, resulting in afinal level of content proportional frequency.A median split was made on the level of con-tent proportions. Mothers with scores abovethe median were considered providers ofmore low-level verbal content, and mothersbelow the median were considered providersof more high-level content.

An analysis of covarianee was conductedon children's phase 3 proportion-correctscores by matemal level of content and childage. Phase 1 proportion correct was thecovariate and found to be significant, F(l,25)= 17.9, p < .01. The significant age effectfound in the child perfonnance analysis wasduplicated, F(l,25) = 26.26, p < .001. Moreimportant to this analysis, children of motherswho emphasized more specific, low-levelcontent had a lower mean phase 3, propor-tion-correct score (M = .73, SD = .11) thanchildren whose mothers emphasized lessspecific, high-level content (M = .87, SD =.13), F(l,25) = 5.6, p < .02. The interactionbetween age and level of content was notsignificant.

Di8cu88ion

The results of this study supported theprediction that social interaction between an


adult and child during a problem-solving taskwmild lead to improved independent childperformance. Children who interacted withtheir mother throughout the furniture-sortingtask subsequendy created more correct, adult-like groupings on their own than childrenwho practiced the task on their own and wereshown the correct solution at the end of eachtrial. Since mothers offered correct solutionsto die task while diey interacted with theirchild, it was important to include a presenta-tion of correct solution to the practice-controlgroup. This control allowed the inference tobe made that it is the process of fuiult-childinteraction itself, not just die le^miing of a cor-rect solution, that leads to improved perfor-mance.

The beneficial effect of adult-child in-teraction on child performance found in diisstudy has only been suggested by the correla-tional study of Wood and Middleton (1975)and studies using nonspontaneous adult in-teraction (Goncu & Rogoff, 1987; Wood et al.,1978). The present results are contrary tothose reported in Kontos (1^3) and Kontosand Nicholas (1986) in which adult interac-tion proved no more beneficial to subsequent,independent child performance dian self-directed practice. Althou^ there were sev-eral môdological differences between theprior studies and the present one to accountfor the diserepant results, peih£^ the crucialdifference was in the type of task. The puzzletask used in the earlier studies was h i^ lychallenging, unique, and certain to be un-femlliar to die child. The fiimiture-sortingtask was also h i ^ y challenging, but utilized^miliar items within a context that was mean-ingful to both child and modier, that is,mc^her and child shared a common knowl-ed^-base. Mothers in die present study mayhave had not only more to interact about butmore that was meaningful to the child. Thus,more meaning&tl interaction led to greater in-temalization and imiMX)vement in die child'sindependent perfomiance. In accordancewith diis view, Rogoff (1987) has suggestedthat soci d interaction may not benefit chil-dren's cognitive development in all circum-stances; beneficial effects iziay depend on dieease with which the task invokes sharedthinking.

The analysis of the division of task re-sponsibilities and matemal regulation of thechild's task activities generally supported theprediction that mothers would be sensitive tothe increased demands on the child's compe-tency during the difficult version of die task.Mothers were more responsible themselves

and more often regulated the child's activityduring the difficult task but only for the mostcrucial task component, room selection.Mothers offered less general verbal instruc-tion during the d^cu l t task than the easy.When demands on the child's competencywere decreased during the easy version of thetask, modiers were responsible fbr and regu-lated the crucial task component to a lesserdegree and offered more general goal direc-tion, planning, and monitoring infmmation.These modier behaviors allowed the child tomanage the crucial task component mare of-ten and Identify relevant task-speci&; infor-môn and straîes on his or her own whenthe task demand was lessened. Thus, the divi-sion of task responsibilities and variation inmatemal regulatum of the child were indeedimplemented in tfee manner proposed by Vy-gotsky (1962, 1978) to fecilitate cognitivegrowth.

The task responsibility and regulation re-sults are consistent with other studies findingthat mothers vary their regulation of die childand specificity of verbal content when childcompetency is assumed to vary widi age(Kontos, 1983; Rc«(êt al., 1984; Weitsch etal., 19^). The £»«sent study manipulated taskdemands with results similar to those of Saxeet al. (1^7) but, contrary to that study, foundno relation between initial child competenceand mother relation of die child (as shownby die lack c^ rektion between phase 1 per-fomiance and dte motiier responsibility [MR]and mother regulation of child [MRC] vari-ables).

Mothers were also shown to vary theirtask resptmsibillty and regulation of the childon die y^tsis of diild age, beyond viât dieyvaried on the basis of task demands on thechild Mothers of 3-year-olds regulated thecmcial task ccnnponent to a greater degreethan modiers of 5-year-olds. Mo*feers of theyounger chiltfcen used more task-specific,concrete verbalizEôns dian modiers of olderchildren, who used more planning, goal di-recting, and mcHiltoring veri»li2ôns. The{ê-related fectrars that could possibly be im-plicated include language coi^wtence, priorproblem-seeing e:q)erience, the abili^ tostay on-task durmg die interaction, said/or, asproposed by Rcôff et al. (1984), the adultperc^Hions of die child's need in specificproblem-sol viog circumstances.

Individual diSerences in child perfor-mance were explored and yielded significantrelations with maternal task behaviors andverbal content. The greatest improvement in

Lisa S. Freund 125

children's independent problem solving re-sulted when mothers varied their regulationof their children in a manner consistent widitask demands and oflfered more strategy, plan-ning, goal directing, and monitoring contentin their verbalizations. This result is in directaccord with Vygotsky's theory that the moresensitive die adult is to the child's compe-tency (which can vary with changing task de-mands) and the more the adult exposes thechild to processes necessary for successfulproblem solving (e.g., strategies, planning,etc.), the more child performance can im-prove and cognitive development proceed.

Isolating the task components in order toassess matemal task behaviors by componentproved to be a mjyor methodological strengthin this study. It is unlikely that the predictedvariation in mother behavior would havebeen detected if the crucial task componenthad not been identified and analyzed sepa-rately from other task components. A majorlimitation of this research, however, is thatthe analysis of the mother-child interaction isshort term. Longitudinal assessment provid-ing evidence of the child's transition frommother regulation to complete self-regulationover a period of time would offer even stron-ger validation of Vygotsky's social interactiontheory of cognitive development. Anotherlimitation of the study is that the focus of themother-child interaction is one-sided. The ef-fect of the child's behavior on the mother'sbehavior throughout the joint interaction ses-sion is never addressed. Without assessmentof the child's part in the adult-child interac-tion, it is impossible to determine to whatextent the child direcdy elicits matemal reg-ulation or to specify the child's developingunderstanding during the interaction (Saxe etal., 1987).

By addressing some of the limitations inthe emergent adult-child interaction litera-ture, the present study provides a comprehen-sive evaluation of matemal behavior duringmother-child interaction and its effect on thechild's ability to problem solve. The studyprovides support for Vygotsky's social inter-action theory of cognitive development, al-though both child and adult assessment dur-ing joint activities are undoubtedly necessaryto fully characterize the child's cognitive de-velopment In addition, joint problem solvingmay not be the only route to successfril, in-dependent problem solving. Children's self-directed problem-solving activity occurs, canbe successful, and can lead to increased cog-nitive skill (Kontos & Nicholas, 1986). How-ever, the origins of such self-directed activity

may be best explained through the under-standing of the child's early, joint problem-solving activities.

ReferencesAms, P. J. (1981). Joint problem-solving activity in

adult-child dyads; A cross-cultural study (Doc-toral dissertation. Northwestern University,1981). Dissertation Abstracts Intemational,42, 5A, 2112.

Baker, L., & Brown, A. L. (1984). Metacognidveskills and reading. In P. D. Pearson (Ed.),Handbook of reading research (pp. 353-394).New York: Longman.

Bransford, J. D., Stein, B. S., Shelton, T. S., & Ow-ings, R. A. (1981). Cognition and adaptation:The importance of learning to learn. In J. Har-vey (Ed.), Cognition, social behavior, and theenvironment (pp. 93-110). Hillsdaie, NJ: Erl-baum.

Brown, A, L. (1978). Knowing when, where,, andhow to remember: A problem of metacogni-tion. In R. Glaser (Ed.), Advances in instruc-tional psychology (Vol. 1, pp. 99-151). Hills-dale, NJ: Erlbaum.

Brown, A. L., Bransford, J., Ferrara, R., & Cam-pione, J. (1983). Learning, remembering, andunderstanding. In J. Flavell 8c E. Markman(Eds.), P. H. Mussen (Series Ed.), Handbook ofchild psychology: Vol. 3. Cognitive develop-ment (pp. 77-166). New York: Wiley.

Cohen, J., & Cohen, P. (1975). Applied multipleregression/correlation analysis for the behav-ioral sciences. Hillsdale, NJ: Erlbaum.

Davis, A., & Lange, G. (1973). Parent-child com-munication and the development of categoriza-tion styles in preschool children. Child Devel-opment, 44, 624-629.

Freund, L, (1988). The child's transition irommother regulation to self-regulation during aproblem-solving task (Doctoral dissertation,University of Maryland, 1988). DissertationAbstracts Intemational, 49, 0419-4217.

Freund, L., Baker, L., & Sonnenshein, S, (1987,April). Developmental differences in the strat-egies used to classify items two different ways-Paper presented at the Society for Research inChild Development, Baltimore, MD.

Goncu, A., & RogofF, B. (1987, April). Nature ofadult guidance as a moderator of social in-Jiuence on children's cognition. Paper pre-sented at the Society for Research in Child De-velopment, Baltimore, MD.

Hedges, L. V., & Olkin, I. (1985). Statisticalmethods for meta-analysis. New York: Aca-demic Press.

Kontos, S. (1983). Adult-child interaction and theorigin of metacognition. Journal of Educa-tional Research, 77, 43—64.


Kontos, S., & Nicholas, J. G. (1986). Independentproblem solving in the development of meta-cognition. Journal of Genetic Psychology, 147,481-495.

RogoflF, B. (1987). Commentary on social processesin early number development. Monoff'aphs ofthe Society for Research in Child Develop-ment, 52(2, Serial No. 216).

Rogoff, B., Ellis, S., & Gardner, W. (1984). Adjust-ment of adult-child instruction according tochild's age and task. Developmental Psychol-ogy, 20, 193-199.

Saxe, G., Guberman, S., & Gearhart, M. (1987). So-cial processes in early number development.Monographs of the Society for Research inChild Development, 52(2, Serial No. 216).

Vygoteky, L. S. (1962). Thought and language (E.Hanfinann & G. Vakar, Eds. and Trans.). Cam-bridge, MA; MIT Press.

Vygotsky, L. S. (1978). Mind in society (M. Cole, V.John-Steiner, S. Scribner, & E. Soubeiman,Eds. and Trans.). Cambridge, MA: MITPress.

Wertsch, J. (1977, May). Metacognition and adult-child interaction. Paper presented at North-western University annual conference onlearning disabilities, Evanston, IL.

Wertsch, J., McNamee, G. D., McLane, J., & Bud-wig, N. A. (1980). The adult-child dyad as aprobfem-solving system. Child Development,51,1215-1221.

Wertsch, J., Minick, N., & Ams, F. (1984). The cre-ation of context in joint problem solving. In B.RogofiF & J. Lave (Eds.), Everyday cognition:Its development in social context (pp. 151-171). Cambridge, MA; Harvard UniversityPress.

Wertsch, J., 8E Rogoff, B. (1984). Editor's notes. InB. Rogoff & J. Wertsch (Eds.), Children's learn-ing in the "zone of proximal development":New directions for child development research(pp. 1-6). San Francisco: Jossey-Bass.

Wood, D. (1980). Teaching tiie young child—somerelatioDships between social interaction, lan-guage, and tihought. In D. Olson (Ed.), Socialfoundation oflai^uage and cognition (pp. 94-112). New York: Norton.

Wood, D., & Midfieton, D. (1975). A study of as-sisted problem-solving. British Journal of Psy-chology, 66, 181-191.

Wood, D., Wood. H., & Middleton, D. (1978). Anexperimental evaluation of four fece-to-faceteaching str^i^jies. Intemational Journal ofBehavioral Development, 1, 131-147.

matemal regulation of children's problem-solving behavior ... · matemal regulation of...

Documents