Mothers and Fathers Teaching 3-Year-Olds

1 downloads 0 Views 749KB Size Report
Mothers and fathers worked separately with the child on three difficult tasks: a block construction task, a matrix classification task, and a story retelling task.
Copyright 1988 by the American Psychological Association, Inc. 0012-1649/88/S00.75

Developmental Psychology 1988, Vol. 24, No. 6,832-839

Mothers and Fathers Teaching 3-Year-Olds: Authoritative Parenting and Adult Scaffolding of \bung Children's Learning Patricia Kerig, Philip A. Cowan, and Carolyn Pape Cowan

Michael W. Pratt

Wilfrid Laurier University Waterloo, Ontario, Canada

University of California, Berkeley

In this study we investigated variations in parental tutoring of children on the basis of Vygotsky's (1978) theorizing and the concept of parental scaffolding of children's learning. Twenty-four couples and their 3-year-old children participated. Mothers and fathers worked separately with the child on three difficult tasks: a block construction task, a matrix classification task, and a story retelling task. Parental interventions were classified by level of support and were used to define the region of sensitivity to instruction, following the work of Wood (1980). Independent observers rated each parent separately on authoritative/uninvolved and authoritarian/permissive styles, following Baumrind's (1967, 1973) typology. Both mothers and fathers adjusted their support of the child as predicted. Later portions of the tutoring interactions demonstrated more fine tuning of interventions by parents than earlier portions. However, authoritative mothers and fathers were generally more likely than nonauthoritative parents to focus interventions in the region of sensitivity across tasks and to shift their interventions contingent on child success or failure. These patterns were also associated with more dyadic success on task as predicted. We discuss relations between these two levels of parenting style.

children proceeds most effectively when tutoring occurs in what he termed the "zone of proximal development" (ZPD; e.g., Rogoff, Ellis, & Gardner, 1984). These zones are contexts in which the child is as yet unable to perform with success independently, but can accomplish components of the task with direct adult support and guidance. Good tutors will seek out this zone, gradually reducing their support as the child becomes capable of more independent task performance (cf. Wood, 1980). Recent empirical work on this Vygotskian model has already suggested that it provides a reasonable description of early adult-child cognitive and linguistic interactions in several contexts (e.g., Bruner, 1985; McNamee, 1980; Rogoffet al, 1984; Wertsch, 1985). In addition, Greenfield and colleagues have conducted cross-cultural investigations of expert-novice tutoring situations, such as the acquisition of weaving skills by preadolescent children (Childs & Greenfield, 1980), All these studies have indicated that, as a whole, adults follow the pattern expected in adapting their instruction to the child's current level of functioning. A frequently used concept in recent Western research that is closely linked to Vygotsky's ZPD model is the notion of "scaffolding" (e.g., Griffin & Cole, 1984; Reeve, 1987), This concept was introduced by Wood and Bruner (e.g., Wood, 1980) to describe the strategies by which parents support children's learning through interventions that provide task information at different levels of structure, depending on the child's current capabilities. As Griffin and Cole (1984) pointed out, this concept of scaffolding provides only a partial instantiation of Vygotsky's complex notion of the ZPD, However, in appropriate task settings, the scaffolding concept has been a useful metaphor for describing tutoring practices. Most of the research on scaffolding to date has been directed

Recently a good deal of attention has been directed to the work of the Russian developmentalist Vygotsky (e.g., 1978), whose theoretical writings focused explicitly on the role of adult-child interaction in the socialization of early cognitive development. Vygotsky argued that all higher planning and organizing functions in development appear twice, initially on the interpersonal plane of social interaction and subsequently on the intrapersonal plane of individual cognitive functioning, following a process of gradual internalization. Most important in terms of the study reported here, Vygotsky's theory entails a specific model for the organization of adult-child "teaching" interactions (e.g., Vygotsky, 1978; Wertsch, McNamee, McLane, & Budwig, 1980). Initially, complex tasks must be accomplished by such a problem-solving dyad, with the adult assuming the directive and organizing functions. The child's role may be simply to enact the specific behaviors assigned to him or her by the adult "tutor." Gradually, however, as the child becomes more knowledgeable and skilled, he or she takes over more and more of the goal-oriented, planning aspects of the task from the adult. Thus, Vygotsky suggested that learning by

This research was supported by a grant from the Social Sciences and Humanities Research Council of Canada to Michael Pratt, and by National Institute of Mental Health Grant R01MH-31109 to Philip Cowan and Carolyn Pape Cowan. Portions of this article were presented at the Society for Research in Child Development Meetings, Baltimore, April 1987. We thank Sheree Bradford and Bikram DasGupta for their help with data analysis, David Wood for his assistance, and the families who participated in this study for their enthusiastic cooperation. Correspondence concerning this article should be addressed to Michael W. Pratt, Psychology Department, Wilfrid Laurier University, Waterloo, Ontario, Canada N2L 3C5. 832

833

PARENT TUTORING

toward establishing the usefulness of this model of the motherchild dyad as a generic account of cognitive socialization. The success of this program raises several interesting questions. Are there important individual differences in adults' use of scaffolding in tutoring that might have long-term implications for children's cognitive development? Are there differences between mothers and fathers in such tutoring patterns? And what sorts of factors might be linked to individual differences observed in such tutoring skills? Wood's research (e.g., Wood, 1980; Wood & Middleton, 1975) suggests that mothers tutoring of their preschoolers in a simple copying task shows clear variations in use of the "region of sensitivity to instruction" while scaffolding, variations that are predictive of the child's subsequent skills on this task. Furthermore, mothers differed in this study in how well they followed a principle of providing less support after the child succeeded and more support after failure, and use of this "contingent-shift" rule was also predictive of child skill acquisition. It must be noted, however, that Wood's work was based on a very small sample of mothers and on a single task. One purpose of our research was to extend the work of Wood to a larger sample of parents tutoring their 3-year-olds on several different tasks. We sought to compare adult scaffolding patterns with some other tutoring style indices as predictors of child success in task participation. In addition, we wished to examine some of the factors that might be predictive of variations in adult scaffolding patterns. The variable of parenting style was of particular interest to us in this research. Baumrind (e.g., 1967, 1973) developed and validated an "ideal type" model of parent differences in discipline, control, and nurturance practices with children. There were originally three major types of parenting styles in the Baumrind system: authoritarian, authoritative, and permissive. Recent conceptualizations have extended these ideas to include a fourth, uninvolved-neglecting grouping as well (Maccoby & Martin, 1983). Considerable research has linked an authoritative parenting approach with positive child outcomes in the areas of cognitive and social competence (e.g., Baumrind, 1973). In contrast to other parenting types, the authoritative parent is demanding in communicating and firmly enforcing rules, but is supportive and nurturant with the child as well. Baumrind argues that authoritative parents are in effect more sensitive to the child's needs for both direction and support than are other parents. It seems reasonable to predict on this basis that authoritative parents will prove to be better tutors of their children in instructional contexts than will permissive, authoritarian, or uninvolved parents. In particular, we hypothesized that authoritative parents would be more sensitive to the child's ongoing task success and would prove more effective in scaffolding children's learning. To date, however, there has been no research on differences in the Baumrind styles in the tutoring situation specifically (but see Harrington, Block, & Block, 1978). This study provided us with the opportunity to test these hypotheses, because parents were rated separately on parenting style and on patterns of scaffolding and contingent shifting by independent groups of observers. As noted, there have been a number of recent studies of mother-child interaction from the Vygotskian perspective. To date, however, there has been little or no research on fathers'

tutoring from this framework. The general research literature comparing mothers' and fathers1 teaching of their young children suggests that parental differences are fairly small, at least in middle-class populations (e.g., Brody, Pillegrini, & Sigel, 1986; Sigel, 1982). However, some evidence from the language development literature suggests that fathers may be less likely to adapt their speech to young children than mothers (e.g., McLaughlin, White, McDevitt, & Raskin, 1983). Moreover, as the research of Sigel (1982) and his colleagues demonstrates, parent sex often interacts in complex ways with sex of child. In order to explore such interactions, in this study we observed both mothers and fathers teaching either a son or a daughter. In summary, we hypothesized that both mothers and fathers would generally show patterns of effective scaffolding of children's learning in their tutoring on three different tasks. However, we predicted that there would be some individual differences in scaffolding and in parents' contingent shifting of their support following the child's success or failure, and that these differences would be related to child success in task participation. Finally, we hypothesized that parents independently rated as authoritative in their general parenting style would show greater use of the region of sensitivity to instruction in their scaffolding and would be more consistently contingent in response to children's successes and failures in these tutoring sessions.

Method Subjects The subjects for this study were 24 middle-class families. They were randomly selected from a larger group that was being followed from prior to the birth of this first child, with the constraint that they had complete data for all variables used here (see Cowan & Cowan, 1985, for details). At the time of this assessment, 12 of the families had a 3year-old girl, and 12 had a 3-year-old boy. Children's mean ages were 42.0 months for the girls (SD = 2.89 months) and 44.3 months for the boys (SD = 3.58 months) at the time of the assessment, and parental education and income levels were comparable for families of boys and girls (most parents had at least some university-level education).

Tasks, Measures, and Procedure Mother-child and father-child interaction sessions were videotaped during separate laboratory visits. The order of these visits was counterbalanced across the sample. During each, parents were asked to help their child on a block model-copying task and a matrix classification task (adapted from Block & Block, 1980). Each parent also tried to elicit a story from the child that had previously been told to the child by an experimenter while the parent was absent. Two other less structured tasks were also completed during these sessions but are not discussed in detail in this article. Each session lasted about 30 to 40 min. Parallel forms of the various tasks were used for each parent's visit. The story retelling task always came first, usually followed by the block and then the matrix tasks in succession. However, parents occasionally rearranged the order of these last two tasks to maintain children's interest. Block and matrix tasks. One form of the block task required the child to use sets of blocks to build constructions that would be identical with a simple model. The blocks came in a variety of shapes and sizes and were presented in a large cardboard box for the dyad to use as they wished. This task was essentially parallel to the model-copying task used by Wood and Middleton (197 5). Parents were free to have children con-

834

PRATT, KERIG, COWAN, AND COWAN

struct as many examples of the model as they wished, and typically the dyads constructed two or three copies. The parallel form of this task involved constructing large squares to match a model from a variety of smaller, variously colored, flat squares and rectangles. On the average, four of these copies were completed. The matrix task involved a double classification problem. One form required placement of nine small, flat pieces in a size-by-color crossclassification array. The second form involved placement of nine pieces in a color-by-shape array. About half of the dyads managed to complete the matrix at least once overall. In general, these tasks were quite difficult for the children to carry out independently. The videotapes of these tasks were analyzed using a coding system for parent tutoring adapted from the work of Wood (e.g., Wood & Middleton, 1975). The original Wood system distinguished six levels of parent support and structure for child task participation. We found it useful to add one further level (the category of "verbal hints"). Our categorization system for the block and matrix tasks is shown in Table 1. Episodes of parent intervention-child response were established on the basis of the occurrence of an identifiable child response, attempt, or pause with signs of noncomprehension that demarcated the end of each episode. Each separate episode was then characterized by its most structured level in the 7-level coding scheme. There were typically about 25 to 30 episodes per task. We scored child responses as either successful (as a step toward task solution) or as failed, following Wood. Another rater independently coded a randomly chosen sample of eight videotapes for each task. Episodes were reliably distinguished (coefficients of agreement were in the .90s). Interrater agreement on parent intervention level, given that an episode was established, was .87 for the blocks task and .80 for the matrix task. Agreements for success were in the .90s. We further distinguished periods of parental "didactic teaching" separately in the protocols, where parents attempted to give the child an overall perspective on the component features of the task but did not appear to be giving instructions specific to task completion (e.g., "What do we call these shapes?"). These didactic periods could occur both at the beginning of a task or as an interruption to the "tutoring" that we rated using the Wood code. Typically, two or three periods occurred per session. The length of time for these didactic teaching periods was recorded. For the eight videotapes in our reliability sample, the coefficient of reliability for distinguishing these didactic teaching episodes was .85 across both tasks. On the basis of these tutoring codes, a series of measures were derived to characterize each parent-child session (see Table 2). These included two scaffolding measures derived from Wood's (Wood & Middleton, 1975) work: the percentage of tutoring interventions in the "region of sensitivity," and the percentage of appropriately contingent shifting by the parent. We denned the child's region of sensitivity (as did Wood) by examining his or her rates of success for each parental level of intervention. The region for the block and matrix tasks was denned as that level just below the least structured one that showed clear indications of predominant success by the child (see Table 2). For example, a child that usually succeeded with interventions of Level 3 and below but failed at most interventions of Level 2 and above would have his or her region of sensitivity denned as being at Level 2. As in the Wood studies (Wood & Middleton, 1975), success rates by levels were generally orderly and cumulative for each dyad (see the Results section below). In practice, predominant success was denned as over 66% of all episodes successful at any given level. The modal region of sensitivity for these tasks in our sample was at Level 4 in our system (see Table 3). However, some subjects showed shifts later in a tutoring period toward more independent functioning (typically at the region of Level 1; see Table 3). For these subjects, the parent's scaffolding practice was therefore scored on the basis of two distinct levels at different times in the tutoring. For example, Level 4 might be scored as the region of sensitivity to instruction for the early

portion of the interaction and Level 1 for the later portion. Only dyads with at least six scorable episodes were included in the analyses (about 90% for each group). In addition to these two measures that were based on Wood, we also derived (a) the total number of parent-child tutoring episodes per task, (b) the average level of parent intervention across each task session, (c) the amount of time spent in didactic teaching of the child in the session, and (d) the overall percentage of episodes for which the child succeeded in following adult instruction. This last "dyad success" index was used as a type of outcome measure in some of the analyses below. Thus, we conceptualized a high rate of dyadic success in the interaction as indicative of more effective tutoring. We should note here that this "percentage of successful interventions" score was in feet highly correlated with the total amount of construction success for the dyad. For the block task, the number of model copies constructed correlated .74 for fathers and .48 for mothers with this detailed success index. For the matrix tasks, these correlations were .56 and .68 for fathers and mothers, respectively. All correlations were significant at the .05 level. Story task. There were three different stories used for the story retelling task, counterbalanced across parent within each family. Each of the stories contained two episodes and was constructed using the Stein and Glenn (e.g., Glenn, 1978) story grammar, which defines six categories ofstory information per episode. The stories thus contained 11 propositions each (only 1 setting proposition was used for each story). The main characters were either animals or children engaging in familiar, but interesting, activities. For example, in one story a young girl receives an old bike from her neighbor, fixes it, and then wins a bike race at her school. The information in these stories was broken down in a simple item analysis. The number of information items was similar across stories, with a mean of 25.0, or about two items per proposition. Parent levels of intervention for this story retelling task were broader than those for the block and matrix task, but were analyzed in a generally parallel fashion. Different parent elicitation techniques were grouped into three general categories, partly on the basis of the work of McNamee (1980). McNamee identified four possible levels of increasing probe support by adults for maintaining child story telling: (a) repetition of the child's last comment, (b) nonspecific questioning, (c) focused wh- questions regarding the next story information item, and (d) information provided through a tag question. In McNamee's study the adult knew the story, whereas in our task this was not the case. We found that repetitions and clarifications in our task functioned more as comprehension checks by the adult, so we adopted only the last three of McNamee's levels for our purposes. Table I shows the levels of probe support that we identified. General story probes were least supportive and corresponded to Level 1 on the construction tasks. These probes did not focus on particular aspects of the story at all. "Specific information" questions were considered more supportive, roughly paralleling the middle levels of the construction task hierarchy. Finally, "suggests information" question types were the least demanding and most structuring elicitation techniques, paralleling Level 5 of the construction tasks. Although most children could in fact respond to such closed questions (e.g., "Was Smokey a dog?"), it seemed wrong to designate these responses as "successful" in the sense used for the construction tasks. These questions and responses often failed to further the goal of obtaining story information unless the parent happened to guess the specific information item sought. Nevertheless, Table 1 shows the frequency with which children were able to provide an "informative" response to these question levels. A child's response was defined as informative if some new information relevant to the story was produced (either what was or was not in the story). Noninformative responses were failures to answer or displays of uncertainty. As shown in Table 1, Level-3 questions were easiest to respond to (as expected), whereas Level-1 general probes were most difficult. Interrater agreement on these categories for a sample of eight protocols was .84.

835

PARENT TUTORING

Table I Levels ofParental Support for Each Task, Success Scores, and Percentage ofAll Interventions at Each Level Percentage of success

Percentage of total episode

Level

Definition for Block and Matrix

Block

Matrix

Block

Matrix

0 1 2

19 17 17 25

18 18 23 32

17 23 11 17

27 16 14 18

30

26

17

13

76

60

5

8

6

No parent intervention General verbal start ("You do one.") Verbal hints ("It's too big.") Specific verbal instructions ("Needs another little one.") Identifies material or placement ("That one [points] will fit.") Parent specifies both material and placement ("Get that little one and put it here [points].") Parent demonstrates





10

4

Level

Definition for story task

3 4

5

1

2 3

Percentage of child informative response

Total percentage of parent use

12

43

30

31

85

27

General story questions ("Tell me your story.") Specific information questions ("What was the dog doing?") Suggested information questions ("Was the bike broken?')

Because parents were unaware of the correct solution in the story task, our analyses had to be somewhat different from those used for the block and matrix tasks. Most of the stories produced by the children contained only a little information overall (typically, two or three items), and only a few managed to tell coherent story propositions. Thus, it seemed most reasonable to consider that for the majority of children, the region of sensitivity was at the least demanding, suggests-information level, because even that level was typically not producing much actual success in story retrieval overall. However, a minority of the children did produce more story information in their retellings and did so

Table 2 Summary ofMeasures Used Measure Scaffolding Percentage of activity in region of sensitivity* Percentage of correct shiftingb Other types of teaching Total number of interactions Mean level of parental support Total didactic teaching timebyparentb Percentage of successful dyad episodesa

Definition for block and matrix task Find lowest child success level, then level is next below (e.g., 3 for 4) If child succeeds parent provides less support; if child fails parent provides more support Number of parent-child episodes per task Average intervention level across all task interventions Total seconds of explicit didactic instructions on task Percentage of times that child is able to follow parent intervention appropriately

• Defined somewhat differently for story task. b Not defined for story task.

in a more fluent manner (15 of the 48 parent-child story tellings producedfiveor more items of information). We noted that as a group the 7 fathers and 8 mothers in these more successful story retellings showed a significantly greater use of the less structured specific-information question types in their protocols than did other parents. This would be expected from a Vygotskian analysis as a reduction of support for these children's retellings. Accordingly, we treated these children's region of sensitivity for this task as at the intermediate specific-information level (see Table 1) for these 15 story retellings only. The mean story recall for these children was 9.1 items overall, much higher than recall in the less fluent group (M = 1.7 items). Thus, parental scaffolding was analyzed for this task in a fashion parallel to the approach used on the construction tasks. However, dyadic success rate for the story telling task was measured simply using the total amount of story information that the dyad managed to produce overall. Recall that the detailed success measure used for the block and matrix tasks was in fact highly correlated with such a general dyad con-

Table 3 Child Region ofSensitivity Levels by Task and Parent Blocks Level 1

2 3 4

5 4 to 1 shift Othershifts Undefined or not scorable Total

Matrix

Fathers

Mothers

3 0

2

1 10 0 4 2

2 17

4 24

0

Fathers 4 0 1

Mothers 1

0

0

9 0

1 8 2

2

7

9

0

2

2

1 24

1 24

1 24

836

PRATT, KERIG, COWAN, AND COWAN

struclion success measure in these tasks as well. An average level of probe support across the three types of questions was also obtained as an index of typical level of support (see Table 2). No attempt was made, however, to develop a contingency shift measure on the story task, given the difficulties of scoring specific child responses in the story task as successful or as failed in. the sense of direct story-information retrieval. Parenting style measures. The experimenters who conducted the study (a male-female graduate student pair) independently rated mothers and fathers separately on a series of 14 scales at the completion of all the family's sessions. Observations of the tasks described above, of two more tasks, and of another family session using similar tasks all formed the basis for these parenting style measures. Thus, the observational base for these ratings was considerably more extended than {about 1 lh hr per parent), but partially overlapped, the three specific tasks discussed in this article. The five items used to construct the Baumrind typology included structure, limit setting, responsiveness to child, warmth with child, and lack ofanger at child. Interrater reliabilities for these adjectives scales ranged from .70 to .95 for the two raters. Average ratings on these scales were then composited to construct two factor scores for each parent that represented the current fourfold interpretation of the Baumrind typology (Maccoby & Martin, 1983): authoritative/uninvolved and authoritarian/permissive factors. More authoritative parents scored higher than average on all five scales. Uninvolved parents scored lower than average on allfive.More authoritarian parents were high on structure and limit-setting, but low on the responsiveness items, whereas more permissive parents scored high on responsiveness, but low on the structure items. These two composite factors were fairly independent of each other as expected (rs = —.39 and —.30, for authoritative/uninvolved and authoritarian/permissive styles for mothers and fathers, respectively).

Results Construct Validity ofthe Scaffolding Measures Table 1 shows the dyad success ratios for the block, matrix, and story tasks across the sample for each level of parent intervention. As expected, these increased regularly with more supportive interventions, indicating that, although the tasks were difficult, children could generally participate in them successfully at some level of parent support. For both the block and matrix tasks, success ratios for Level 5 significantly exceeded those for Levels 3 and 4, which in turn significantly exceeded Level-1 scores (ps < .05, using chi-square tests). These results generally parallel Wood and Middleton (1975). Table 1 also shows the percentage of interventions characterized at each level of the hierarchy for parents on the tasks. On the block and matrix tasks, interventions at Levels 5 and 6 were relatively uncommon, thus indicating that parents tended to provide more challenging interventions to the children where success rates were generally lower (i.e., in Levels 1 to 4). Level0, child-initiated interventions were somewhat more frequent for the matrix task, but otherwise there were no noteworthy overall task or parent differences in distributions. Parent interventions for the story task also tended to be more frequent at the less supportive question levels. As in the Wood and Middleton (1975) study, greater use of the region of sensitivity in scaffolding was positively related to more consistent shift patterns after child success versus failure for the block and matrix tasks (see Table 4 for means for these variables). Correlations for the region and shift measures were .36 and .56 for mothers and fathers on the block task, and .58

and .66 for mothers and fathers on the matrix task (all ps < .05). This suggests that parents who follow the shift rule more fully are more effective in scaffolding practices. Most of the failures to follow this shift pattern, as in Wood and Middleton (1975), were the result of the adult staying at the same level as before, particularly after child failure.

Temporal Patterns in the Tutoring Performance of Mothers and Fathers We examined temporal patterns in parents' scaffolding practices for all three tasks over the course of each parent-child teaching session. The percentage of use of the region of sensitivity for thefirstversus second halves of each tutoring interaction showed a consistent tendency for this index to increase during the latter part of the interaction across both parents and tasks. Apparently both mothers and fathers, as good diagnosticians, were gradually focusing more precisely on the region of instruction for each task as they worked with the child over time. A four-way, 2 x 2 X 3 x 2 analysis of variance (ANOVA), with child sex as the between factor and parent sex, task, and early/late interaction half as repeated measures, showed only a main effect of early versus late portions of the interactions, F([, 17) = 19.61, p < .001 (Ms = 23.9% vs. 35.7%, for the early vs. late halves). The lack of parent main effects or interactions shows that mothers and fathers were equally capable of identifying and using the region of instruction in scaffolding on these three tasks (see Table 4). It should be noted that examination of the parents' use of specific levels of intervention did not show systematic patterns of change over time. Thus, this result for scaffolding was not accounted for by a general increase in use of one particular level of intervention by parents, for example, the use of Level 4. We also examined patterns in use of contingent shifting over time in the data for the block and matrix tasks. There was no indication of systematic temporal change from the earlier to later portions of the interactions in the percentage of rule-consistent shifting following children's responses. Nor were there any sex differences (see Table 4). However, the frequency of extreme shift "swings," involving moves over three or more levels from one intervention to the next (e.g., from Level 2 to Level 6), did show a systematic decline. Clearly dyads were also modulating their teaching interactions more precisely around the region of instruction with experience. A four-way ANOVA paralleling that described above showed a main effect of early versus late halves of the interaction, F(l, 17) = 15.5, p < .01 {Ms = 24.4% vs. 15.9%, forfirstvs. second halves, respectively). There was also a main effect of parent sex in this analysis, with fathers significantly less likely than mothers to show such extreme shifts in tutoring as well, F{\, 17) = 6.47, p < .05 (Ms = 16.6% vs. 23.8%, for fathers vs. mothers, respectively). There were no other significant effects. Note that the use of contingent shifting could not be examined in the story task because the shift rule was undefined for that task. Finally, in about 30% of the block and matrix sessions, there was evidence of short-term "microgenetic" development, in that later success scores indicated that children's performance level on the task had improved compared with earlier responses at the same level of parent intervention (see Table 3). Such shifts

837

PARENT TUTORING

Table 4 Means Percentage for Scaffolding Measures and Correlates ofDyad Success Scores by Task and Parent Matrix

Blocks Measure

Mothers

Fathers

Mothers

Story Fathers

Mothers

Fathers

5.05 37.3

3.52 29.6

M percentages M dyad success rate Muse of region A/shift rule use

29.3 23.7 56.7

30.3 35.9 55.6

38.5 22.3 52.6

35.8 30.0 59.1

a

Correlation of dyadic success by Percentage use of region Percentage shift use Total episodes Total teaching Mean support

.51* .43* .20 .07 -.15

.66* .73* .29 .15 -.45

.55* .51* -.36 -.32 .17

.70* .53* -.25 .00 .39

.42*

.15 a

-.40

.17

.29

.29

a

' Not defined for story task. * p < .05.

in children's success patterns were accompanied by the provision of significantly less average parental support after the shift, for both the block {n = l,T=2,p< .05, using the Wilcoxon matched-pairs test) and the matrix (n = 20, T - 17, p < .01) tasks. This is predictable from Vygotsky's model of adult tutors as gradual socializers of child independence, and specifically from Wood and Bruner's (e.g., Bruner, 1985) scaffolding concept. Unfortunately, this aspect of temporal patterns of change could not be examined in the story task because of the more global method of success scoring used there.

Correlates ofSuccessful Dyadic Task Performance Our next set of analyses focused on the correlates of tutoring success, using the overall percentage of successful responses as the criterion measure for the block and matrix tasks and the total amount of story information elicited as the criterion for the story retelling task. Table 4 presents correlations of these criterion measures with the various teaching indices, including percentage of appropriate scaffolding in the region of sensitivity, percentage of contingent shifting, total number of episodes per task, total amount of time spent teaching per task, and average level of parent support for each task. Parental scaffolding is clearly a substantial correlate of dyad success in each task, except for story recall for fathers. Parents' consistent use of the shift rule was also a correlate of success in the block and matrix tasks (it could not be scored in the story task). None of the other teaching measures was significantly correlated with percentage of dyad success (see Table 4).

parent that the relation of parenting style with the tutoring indices must be assessed for each task.

Parenting Style and Tutoring Behavior Authoritative parenting style, as rated by independent observers, was studied in relation to three indices of tutoring for each task: percentage of success measures for the dyad, percentage of use of the region of sensitivity by the parent in tutoring, and total percentage of contingent shifting by the parent. In order to compare the role of different parenting styles, correlations were computed between these three indices and the two parenting style factor scores (authoritative/uninvolved and authoritarian/permissive) for mothers and fathers separately. These correlations are shown in Table 6. As predicted, an authoritative parenting style is generally positively related to better scores for the various tutoring indices across tasks. For the dyad success scores, these correlations were significant for mothers for the matrix and story tasks and for fathers for the blocks task. For percentage of use of the region of sensitivity, the correlations were significant as predicted for the matrix and story tasks for mothers and for the blocks and story tasks for fathers. For the shift pattern, the correlations were significant for mothers on the matrix task and for fathers on the blocks task as predicted. Unexpectedly, however, there was a significant inverse correlation for fathers on the matrix task (more authoritative fathers used the contingent shift rule less often). Finally, authoritarian/ permissive ratings were unrelated to these three tutoring measures. Overall then, these patterns suggest a partial, but by no means complete, overlap between the concept of scaffolding and Baumrind's (1967, 1973) concepts of effective parenting style.

Consistency of Tutoring Measures Over Tasks As shown in Table 5, the level of consistency of parent scores on our various tutoring indices across tasks was modest. Dyad success scores, parental scaffolding, and use of contingent shift rule demonstrated little consistency across tasks. Amount of didactic teaching and total number of parental interventions showed more consistency, especially for fathers. Thus, it is ap-

Discussion Thefirstaim of this investigation was to demonstrate the usefulness of the Wood and Bruner (1976) scaffolding concept across three distinct tutoring situations. Both mothers and fathers of preschoolers showed consistent patterns of focusing more on the "region of sensitivity to instruction" as the tasks

838

PRATT, KERIG, COWAN, AND COWAN

Table 5 Consistency Across Tasks for Various Tutoring Measures by Parent Sex Matrix X Story

Block X Story

Block X Matrix Measure

Mothers

Fathers

Mothers

Fathers

Mothers

Fathers

Dyad success Region use Shift rule* Total episodes Total teaching8 Mean support

-.03 .03 .30 .04 .22 .59*

-.16 -.04 -.09 .59* .63* .64*

-.08 .24 — .23 — -.25

-.17 .37* — .45* — .03

.37* .06 — .27 — -.30

.02 -.21 — .47* — .15

' Not scored for story task. * p < .05.

proceeded. As found previously, greater use of the appropriate scaffolding level was consistently correlated with patterns of parent responses to the child that involved less support after success and more support following failure (Wood & Middleton, 1975). Parents also gave evidence of working within a narrowing range of tutoring levels as the tasks proceeded, with fewer extreme swings in sequence. Such extreme swing patterns were previously shown experimentally to be an ineffective tutoring style (Wood, Wood, & Middleton, 1978), so it is interesting to observe a decline in their incidence over the course of the interactions. Parents also responded to the child's improvements over trials on the tasks (in a way that was predictable from the scaffolding model) by systematically reducing their level of task support. Finally, parents' greater use of appropriate scaffolding and of contingent shifting patterns was consistently correlated with the child's successful participation and accomplishment on the tasks. All these results provide clear evidence for the construct validity of a Vygotskian scaffolding framework across three distinct tasks. The second major aim of this study was to observe the relationship between sensitive tutoring as measured by our scaffolding indices and Baumrind's (e.g., 1973) conception of authoritative parenting. Consistent with our hypotheses, more authoritative parents' tutoring was generally higher in dyad success percentages, appropriate scaffolding, and use of the shift rule. However, fathers did show a significant inverse relation between use of shift rule and authoritativeness on the matrix task.

This appeared to result from a tendency of several authoritative fathers to persist at the same level after child failure, but the explanation for such a phenomenon on this single task is obscure. Taken together, our results support the argument that parental use of scaffolding processes shows important individual variability (cf. Reeve, 1987). Furthermore, these findings point to links between detailed parental behaviors at this "microanalytic" level of analysis and the broad parenting styles discussed by Baumrind (1967, 1973). The results are consistent with the hypothesis that more sensitive, authoritative parents in the Baumrind typology may obtain some of their reported success in cognitive skill socialization through the use of a more effective style of tutoring. Nevertheless, the present correlational design cannot rule out the possible role of other variables (e.g., child characteristics) in eliciting both tutoring strategies and parenting styles. In addition, the present study was unable to examine the effects of greater use of the region of sensitivity and of more systematic contingent shifting on the child's independent posttest performance after tutoring. However, previous experimental studies have indicated that clear benefits in child learning result (e.g., Reeve, 1987; Wood etal., 1978). Of course, more experiments are needed to test tutoring and child-rearing effects independently of naturally correlated variables. We want to stress that the association between authoritativeness and successful tutoring observed here cannot involve contamination of ratings. Although the actual observations used

Table 6 Correlations ofParenting Style and Scaffolding Measures by Task and Parent Authoritarian/Permissive

Authoritative/Uninvolved Tutoring measure Mother success3 Father successb Mother region use Father region use Mother shift Father shift a

Block

Matrix

Story

Block

Matrix

Story

.21 .62*

.34* -.28

.47* -.06

-.10 -.17

.30 -.08

-.21 -.26

-.10 .49*

.38* -.18

.42* .40*

.08 .17

-.05 -.14

.03 -.24

.21 .38*

.42* -.58*

c

.10 -.02

-.02 -.24

c

c c

Ns = 23 for block task, 23 for matrix task, 21 for story task. b Ns = 20 for block task, 23 for matrix task, 21 for story task.c Not denned for story task.*/?