developing narrative structure in parent-child reminiscing … · 2016-11-05 · narrative...

Developmental Psychology Copyright 1997 by the American Psychological Association, Inc. 1997, Vol. 33, No. 2, 295-307 0012-1649/97/$3.00

Developing Narrative Structure in Parent-Child Reminiscing Across the Preschool Years

Catherine A. Haden University of North Carolina at Chapel Hill

Rachel A. Haine and Robyn Fivush Emory University

This study is a longitudinal exploration of relations between parents' and children's provision of narrative structure in joint retellings of the past and children's developing personal narrative skills. Fifteen White, middle-class families participated when children were 40 and 70 months old. At both ages, mothers and fathers talked separately with children about shared past events and uninformed experimenters elicited children's personal narratives. Whereas mothers and fathers did not differ in how they structured past narratives, children narrated differently with fathers than with mothers. Further, even at 40 months, girls' narratives were more contexted and evaluative than boys, but parents' provision of narrative structure increased similarly with daughters and sons over time. Children's early abilities to provide evaluative narratives was a strong predictor of their later abilities to provide evaluative narratives; maternal emphasis on evaluations also predicted children's later narrative structure. Parental and child influences on personal narrative skill development are discussed.

Most of the research on the development of personal narrative skills has emerged from a linguistic approach to examining narrative structure first set forth by Labov (1982; Labov & Waletzky, 1967). A basic premise of this approach is that narra- fives are interpersonal: Narratives are told to others in ways that can be understood by others. Thus, recounting personal narratives requires both the ability to recall past experiences and the ability to organize these experiences into culturally con- ventionalized narrative forms. This study is a longitudinal exploration of the ways in which parents and children talk about the past across the preschool years and the possible relations between parents' and children's use of narrative structure and children's developing abilities to tell coherent and meaningful personal stories.

Vygotsky's (1978) zone of proximal development and Ro- goff ' s (1990) guided participation model provide conceptual frameworks for studying the developmental process whereby personal narrative skills are learned in early parent-child interactions. Central to both theories is the idea that cognitive skills have their origins in social interactions with more skilled partners. Essentially, as children begin to participate in social acfivi-

Catherine A. Haden, Center for Developmental Science, University of North Carolina at Chapel Hill; Rachel A. Haine and Robyn Fivush, Department of Psychology, Emory University.

This research was funded by a grant from the Spencer Foundation to Robyn Fivush. Many people were involved in various aspects of the study, but we especially thank Elaine Reese and Liza Dondonan for their help in data collection, Marcella Eppen and Laura Underwood for their help in transcription, and Tamekia Glover for her help in various data- related tasks. We offer special thanks to all the families that gave their time so generously throughout the project.

Correspondence concerning this article should he addressed to Cather- ine A. Haden, Center for Developmental Science, University of North Carolina, 521 South Greensboro Street, C.B.#8115, Chapel Hill, North Carolina 27599-8115. Electronic mail may be sent via Interact to [email protected].

ties that are slightly beyond their competencies, adults scaffold children's performance by providing the necessary structure for accomplishing the task. With increasing skill development, in Rogoff 's terms, a "transfer of responsibility" (p. 100) occurs from adult to child, and the adult-provided scaffold is reduced. Finally, the theories predict that, through their interactions with adults, children will directly learn and internalize all the component skills and will become able to perform the task unscaffolded. The present study involved tracing changes in the degree of parental scaffolding in conversations about the past with their children across the preschool years, together with changes in children's developing abilities to provide more organized and coherent personal narratives over time. Further, we assessed theoretical predictions that children's developing abilities to tell unscaffolded personal stories relates to parental provision of narrative structure in early talk about the past with children. To date, only a few studies have begun to examine the ways in which particular aspects of narrative structure are displayed in parent-child conversations about the past over time, and how these structural elements may become incorporated into children's repertoire of personal narrative skills.

Parents and children begin talking about .the past together virtually as soon as children begin talking at all (Eisenberg, 1985; Hudson, 1990; Miller & Sperry, 1988; Nelson, 1988). Previous research has established that these early conversations about the past are usually heavily scaffolded by parents who provide most of the content and structure of the story. But, not all parents scaffold talk about the past with their young children in structurally similar ways. In conversations in which mothers were explicitly asked to talk with their preschoolers about the past (Engel, 1986; Fivush & Fromhoff, 1988; Reese & Fivush, 1993) and also in more spontaneous past event conversations (McCabe & Peterson, 1991 ), two distinct parental styles have been documented. One style, variously labeled "high elaborative," or "topic-extending," describes parents who construct richly elaborate descriptions of past events, provide a lot of

295

296 HADEN, HAINE, AND FIVUSH

embellished information, and tend to ask many questions. The other style, which has been labeled "repetitive," "low elaborative," or "topic-switching," describes parents who have short conversations about the past, provide little descriptive information, and tend to ask the same questions over and over. Longitudi- nal research focusing on maternal styles of talking about the past indicates clear facilitative influences of a highly elaborative style on children's later abilities to recall memory information. In particular, Reese, Haden, and Fivush (1993) found that mothers who were more highly elaborative in talking about the past with their 40-month-old children had children who were recalling more memory information in mother-child conversations at 58 and 70 months of age. Further, a maternal elaborative or topic-extending style is associated with preschoolers' longer and more elaborative unguided narratives elicited by an experimenter one year later (McCabe & Peterson, 1991; see also Hudson, 1990, for similar results). Thus, there appears to be a good deal of individual variability in the ways parents talk about the past with their young children, and these styles relate differentially to children's developing abilities to recount the past.

Nevertheless, recounting past experiences involves more than simply recalling information. In order to communicate effec- tively with others, one must organize what is remembered into a coherent and meaningful story. In this way, the development of personal narrative skills may be an important prerequisite for other kinds of cognitive and social advances in the preschool and early school-age years. In particular, development of decontextualized language skills for socially sharing personal narratives may provide critical linguistic resources for literacy acqui- sition (e.g., Heath, 1983; Peterson & McCabe, 1994; Reese, 1995; Snow, 1983; Snow & Dickinson, 1990). Also, memories of personally experienced events that can be recounted to others as canonical narratives may be more likely to become integrated into a child's autobiographical life story, and therefore, may more likely be recalled later in development than narratives that are not organized in this way (Fivush, 1988; Fivush & Hamond, 1990; Nelson, 1993; Pillemer & White, 1989).

All personal narratives recount what happened. Thus, all personal narratives include referential information that explicitly states the actions that occurred and describes objective details concerning conditions, persons, or objects involved in the event. But meaningful stories must go beyond referential information. The narrator must begin with orienting information to convey to the listener the spatial-temporal and social context of the past event. Orienting information includes the setting of when and where the event occurred, as well as naming the people present. Perhaps most important, evaluative information must also be included in personal narratives to make explicit why the event was interesting, self-defining, emotional, meaningful, and so on. A variety of evaluative devices intensify ( "I t was really cold." ), emphasize ( " I t n e v e r stopped."), modify ( " I t was a bad movie." ), or provide internal responses ( " I was s a d . " ), marking particular parts of an event as most important or meaningful. Evaluations are an essential linguistic element of personal narratives, conveying the significance of the event for the teller.

There is ample evidence that children as young as two years of age recount information about novel, one-time past experiences, as part of their regular interactions with caregivers, including in their narratives at least some orienting and evaluative

information (Eisenberg, 1985; Miller & Sperry, 1988; Peterson, 1990; Sachs, 1983). Moreover, Fivush, Gray and Fromhoff (1987) found that 29- to 35-month-olds were able to give semi- organized recounts of novel, one-time past experiences in tell- ings elicited by an experimenter. Children's early narratives are, however, rudimentary in structure, including one or two referential statements, little orienting information, and few evaluations that are often expressed nonverbally (Miller & Sperry, 1988; Peterson, 1990; Peterson & McCabe, 1983; Umiker-Seboek, 1979). Not unexpectedly, older children produce longer personal narratives than younger children, but in addition to recounting more actions and details of the event they also provide more and different types of orienting and evaluative information (Fivush, Haden, & Adam, 1995; Hudson & Shapiro, 1991; Keman, 1977; Menig-Peterson & McCabe, 1978; Peterson & McCabe, 1983; Umiker-Seboek, 1979).

That children's narrative skills develop as they develop more sophisticated language skills is not surprising. Moreover, par- eAt-child conarratives about the past are certainly not the only context in which children may be learning personal narrative skills. Several studies indicate that children may learn the kinds of information to include in their personal narratives in myriad types of interactions involving stories told around them and about them to others (e.g., Miller, Potts, Fung, Hoogstra, & Mintz, 1990; Miller, 1994). Yet theories of adult- child scaffolding would lead to the hypothesis that the ways parents and children co-construct personal narratives would be a particularly important context for children to learn the forms for personal narration and the value of this activity of telling stories about themselves to others. As parents model the inclusion of orienting, referential, and evaluative information in early conversations about the past with children, children may directly learn the importance of including such information and do so in their later unscaffolded narratives with others. Whereas previous research has established that more detailed parent-child conversations about past events facilitate children's developing autobiographical memory skills, we still do not know whether children are also learning how to organize their autobiographical recounts into richly contexted and evaluative narratives through participating in specific kinds of narrative interactions.

Three previous studies have provided limited evidence of long-term parental influences on the development of personal narrative skills. Fivush (1991) assessed mothers' provision of orienting, referential, and evaluative information in conversations about unique shared past events with their 2 I/2-year-old children, and children's abilities to tell unscaffolded narratives to an experimenter one year later. Mothers who provided many orienting and evaluative statements in co-constructed narratives with their children had children who were providing frequent orienting and evaluative statements in their unscaffolded narratives a year later. In another longitudinal study, Peterson and McCabe (1992) selected two highly elaborative or topic-extending mothers who tape-recorded past event conversations with their children starting at 27 months of age for an 18-month period. One mother emphasized orienting information, and her child came to incorporate a great deal of orienting information into her later unscaffolded narratives elicited by an experimenter. The other mother, in contrast, elicited more temporally ordered descriptions of actions and occurrences, and her child came to

DEVELOPING NARRATIVE STRUCTURE 297

provide a lot of this information and little orienting information in her subsequent narratives.

Peterson and McCabe (1994) also studied specific types of parental orienting questions and their relations to children's abilities to provide this information in personal narratives with an experimenter. Parents (9 mothers, 1 father, all primary caretak- ers) tape recorded conversations about past events when their children were between 26 and 43 months of age. Parental orientations in the parent-child conversations were categorized as wh- context questions (e.g., "Where did we go? . . . . When did we go? . . . . What did you play?" ) or yes or no context questions (e.g., "Did we go to McDonald's?"). Children's narratives elicited by an experimenter over the same time period were coded for children's provision of when and where information. Results revealed that parents who provided many wh-context and yes or no context questions at the early time points had children who include much when information in their narratives with an experimenter at 38 to 43 months. Also, parents' provision of wh-context and yes or no context questions early in development was positively related to children's provision of where information in experimenter-elicited narratives six months to a year later. Thus, parents who frequently prompted for context orientation had children who came to give much orientation to when and where in their later personal narratives.

There are, however, several limitations of these previous studies. First, all three studies contained small sample sizes (N = 10 or less). Second, two of the studies looked at maternal influences at the first time point and child narrative productions in unscaffolded experimenter interviews at a second time point. This design does not allow an examination of how children's developing narrative skills may contribute to their own development as well as eliciting more sophisticated narrative interactions with their parents. Only Peterson and McCabe (1994) looked at children's narrative productions over time. Over an 18-month time period, they found that the only relation between children's early and later experimenter-elicited narratives was children's when orientations at 26 to 31 months of age and their when orientations 6 months later. Howevel; in none of these three studies were children's utterances in the parent-child conversations coded. Thus, the previous research has not allowed for examination of possible bi-directional influences operating between parents and children in the development of personal narrative skills. One question that obviously arises is what effects, if any, children's participation in the parent-child memory conversations may have on the amount and types of narrative structure provided by parents? And perhaps more critically, how might children's abilities to provide orienting, referential, and evaluative information in the parent-child conversations influence their own developing skills to tell unscaffolded personal narratives? The present study spans a 2 ~/2-year time period, with the first time point when children are beginning to fully participate in conversations about the past, at 40 months of age (Eisen- berg, 1985), and the second time point when children are rea- sonably competent narrators, at 70 months of age (Hudson & Shapiro, 1991; Peterson & McCabe, 1983). 1 A major objective of the present study was to conduct a detailed, longitudinal exploration of child-to-parent as well as parent-to-child influences in the development of personal narrative skills.

Another limitation of the previous research on the develop-

ment of personal narrative skills is the almost exclusive focus on mother-child conversations (see Reese & Fivush, 1993, for similar arguments). As Gleason (1975) has suggested, fathers in their traditional family role may play a different yet facilitative role in children's communicative development. Specifically, fathers may act as a bridge for children in their attempts to make themselves understood by unfamiliar listeners (Mannle & Tomasello, 1987). Fathers may provide less scaffolding in co- constructing personal narratives because they may not be as accustomed to their children's speech styles as primary caregiver mothers, or because fathers may more often be the uninformed recipients of children's stories about events that mothers may have participated in with the child (see Ochs, Taylor, Rudolph, & Smith, 1992). In turn, children may have to work harder to be understood and to perhaps respond to challenges or complica- tions of their stories posed by fathers, and thus may show different skills and abilities than when talking with their mothers. Importantly, because fathers may provide less scaffolding, they may play a particularly critical and different role from mothers in preparing children to structure their personal narratives to less familiar partners. Previous analyses of portions of the present data indicate that mothers and fathers do demonstrate similarly elaborative styles of talking about the past with their 40- month-old children (Reese & Fivush, 1993). But, there is also evidence from this data set that mothers may be differentiating more between daughters and sons in their provision of elaborations and confirmations than are fathers (Reese, Haden, & Fi- vush, 1996).

The present analyses consider possible differences between mothers and fathers in their provision of orienting, referential, and evaluative information with their daughters and sons. Girls and boys may indeed exhibit differences in personal narrative abilities as a function of whether they are talking to mothers, fathers, or someone outside the family sphere. To investigate this question, at each time point in this study children engaged in conversations about past events with their mothers, fathers, and a relatively unfamiliar experimenter who asked only open- ended questions (e.g., Tell me about it, What happened?). Thus, in the experimenter-child interviews, children were asked to tell personal narratives in the absence of child-sensitive adult scaffolding. Mothers and fathers co-narrated events that they had participated in with their children, while the experimenter elicited the children's recall of an unshared event.

In sum, this study focuses on the development of the ability

1 It should be noted that several other analyses have been conducted on various parts of this data set. In particular, Reese et al. (1993) focused on many of the same mother-child conversational transcripts analyzed in the present study in their longitudinal analysis of maternal reminiscing style. Reese, Haden, and Fivush (1996) compared mother-child and father-children conversations for level of elaboration and evaluation, and Fivush et al. (1995) included some of the experimenter-child transcripts used in the present study in their analysis of children's developing narrative structure. In the present study, we recoded and analyzed all the transcripts using a new coding scheme designed to capture the narrative structure of these conversations, and we compared the narrative organi- zation of mother-child, father-child, and experimenter-child conversations about the past. Thus, although we are using many of the same transcripts as previous studies, we are addressing fundamentally different questions, using a new coding scheme and new analyses.


to construct personal narratives that convey both the occurrence and meaning of past experiences. We provide a longitudinal analysis of how the kinds of narrative structure displayed in mother-chi ld , fa ther-chi ld , and exper imenter-chi ld conversations about the past contributes to children's developing abilities to construct personal narratives with and without parental scaffolding over the preschool period.

Method

Participants

As part of a larger longitudinal study of memory and narrative development, families were visited in their homes at two time points during the preschool years. Initially, 24 White, middle-class mothers and their 40-month-old children were contacted through county birth records in the metro Atlanta, Georgia area. Fifteen families (seven with girls, eight with boys) continued to participate when the children were 70 months old and had complete data for the mother-child, father-child, and experimenter-child sessions at both the 40- and 70-month time points. Of the families not included in the present analyses, two families moved out of state, one declined to participate after the 40-month time point, and six families did not complete all three sessions at each of the two time points. Four girls and six boys were firstborns. Nine mothers worked outside the home 40 hours or more per week, four worked 15 to 30 hours per week, and two worked less than 15 hours per week outside the home. All fathers worked full-time. All parents had attended some college, and 83% held college degrees, Children received a small gift at each time point for their participation.

Procedure

At each time point, families were visited in their homes for four sessions. Data were collected by one of three female graduate student experimenters; families were always visited by the same experimenter across the sessions and time points. At the start of the study, experimenters explained that they were interested in how much and what kinds of information children remember about past experiences. Mother-child, father-child, and experimenter-child interviews were conducted during three separate sessions at both of the time points. The mother-child interview always occurred before the experimenter-child interview at each time point, but the order of the mother-child and father-child interviews was counterbalanced. Within each time point, the three memory interviews were conducted at least 48 hours apart, but within a two- week period.

Event selection. At all memory interviews, events were selected during discussions with the parent out of earshot of the child, upon the experimenter first entering the home. At each time point, experimenters helped each parent to select three special, one-time events to discuss with their child. Mothers selected events they had participated in with their child, and fathers selected events that they and their child had shared. Additional events that the children were asked to discuss with the experimenter were generated by the mothers and were events that the mothers had shared with the children. 2 Events such as birthdays or Christmas were excluded because they tend to be routines even by 40 months, with children having trouble recalling a specific instance (see Hudson, Fivush, & Kuebli, 1992). Parents were also asked to exclude movies, plays, or other activities with a story line, to guard against children retelling the story line instead of talking about what they themselves did. Although there was.variability in the kinds of events families participated in, there was also great similarity in the events selected for recall across mothers and fathers, children and time. The most often discussed events included trips to special science museums, aquariums, and amusement parks, entertainment outings, and first time trips on

airplanes, trains, or boats. On average, the events occurred 4 months prior to these conversations. It should also be noted that parents were easily able to generate events for discussion and that the resulting past event conversations were similar to'ones gathered under less restrictive event selection guidelines (e.g., McCabe & Peterson, 1991 ).

Memory interviews. After event selection, the parent and child, or experimenter and child, sat comfortably in a quiet place in the home with a tape recorder recording their conversation. In the case of the parent-child interviews, the parent was asked to discuss the events in a natural and spontaneous way. The experimenter was not present in the room during the parent-child interviews, and no time restrictions were placed on the length of the conversation. In the experimenter-child interviews, experimenters asked general open-ended questions to intro- duce each event (e.g., "Your morn tells me you went to the aquarium. Can you tell me about that?"), and encouraged the child to continue using only general prompts (e.g., "Tell me more about it" or "Anything else?") and repetitions of the children's responses (e.g., "So, you saw ducks." ).

Data Reduction and Coding

Before coding, all memory interviews were transcribed verbatim, and two raters read through each transcript marking the beginning and ending of each event narrative and all off-topic talk. Codeable event narratives were determined in previous analyses of the transcripts (see Reese et al., 1993) as the first three events for which the child remembered at least two unique pieces of information. 3

The coding scheme, developed to capture the narrative structure of the memory conversations, was adapted from Peterson and McCabe (1983) and Fivush et al. (1995). Mothers' and fathers' comments, as well as the children's talk in all three memory interviews, were coded first for the type of narrative structure in one of four mutually exclusive categories: (a) referential actions, (b) referential descriptions, (c) orientations, or (d) evaluations, and then, for orientations and evaluations were coded into subcategories of these codes (see Table 1 ). The instance of the type of talk was used as the coding unit. Because our interest was in narrative structure and not amount of talk, repetitions of information were not coded.

Two raters independently coded 25% of the transcripts at each time point for reliability. For the 40-month time point, the proportion of agreement for the maternal codes was 82% (Cohen's K = .80) for the mother-child transcripts, 81% (K = .80) for the father-child transcripts, and 78% (K = .74) for the experimenter-child transcripts. For the 70- month time point, the proportion of agreement for the maternal codes was 82% (K = .80) for the mother-child transcripts, 82% (K --- .80) for the father-child transcripts, and 83% ( ~ = .80) for the experimenter-

2 Because part of the theoretical rationale for the original design con- cerned the content of children's recall, at each time point, the experimenter asked the child to recall one of the same events that the mother had asked the child to recall in the previous interview. However, other analyses of these data indicate that children do not incorporate the information that mothers recount about an event into the child's subsequent recall of that event (Fivush, 1994). This pattern supports the theoretical position that children are learning autobiographical and narrative skills about events in general in these conversations and not skills for recounting specific events (see also Hudson, 1990).

3 Based on this criterion, one mother-child dyad at the 40-month time point had two codeable events, and the rest of the mother-child dyads had three codeabte events at 40 and 70 months. Two father-child dyads at 40 months and one at 70 months had two codeable events; the rest had three. For the experimenter-child interviews, four children had two events and one had one event at 40 months, and five had two events at 70 months.

Table 1 Narrative Structure Coding

DEVELOPING NARRATIVE STRUCTURE 2 9 9

Narrative code Definition Examples

Actions

Descriptions

Orientations

Spatial-temporal

Person

Evaluations Internal states

Intensifiers

Affect modifiers

Emphasis

Requesting or providing information about actions that occurred during the event.

Requesting or providing objective details about conditions, persons, or objects at the event.

Comments that place the event in spatial-temporal context or provide information to help the listener understand the event, including:

Requesting or providing the location or timing of an event.

Requesting or providing the names of the people present during the event.

Affective or evaluative commentary on the event, including: Comments that reveal the personal feelings or desires of the

participants, including any reference to quoted speech.

Comments that stress or intensify what they modify.

Statements of subjective or emotional response about an object, person, or the event itself.

Comments about something that didn't happen during the event or a forced action or reaction.

P: "Do you remember walking up to the store?" P: "What did we do?" P: "What color was it?" C: "It was my box."

C: "You were in the hospital.'" P: "When did we see Grandma?" P: "'Who was there?" C: "Ginger was there."

P: "You wanted to go on the slide." C: "I was happy." C: "'She said, 'Go to the store'." P: "I t was a huge bowl." C: "It was very, very cold." C: "She was ugly." P: "She got hurt." P: "Mommy was wrong." C: " I had fun." C: "It never stopped." C: "He made me run away from them."

Note. P = parent; C = child•

child transcripts. The remaining transcripts were coded by one or the other of the two raters. After completion of the coding, the mean total number of narrative structure codes per event was calculated, as a measure of total amount of narrative structure provided.

Res u l t s

The results are presented in three major sections. The first section focuses on mothers ' and fathers ' use of elements of narrative structure in conversations about shared past events with their children over time. Specifically, these analyses address whether mothers and fathers differ f rom each other in their provision of narrative forms to their daughters and sons over time. The second section addresses questions of chi ldren 's provision of narrative structure and cohesion devices over time, comparing girls ' and boys ' narratives told to their mothers and fathers and the chi ldren 's unscaffolded narratives told to a relatively unfamil iar experimenter who provided only open-ended prompts. The third section explores relations between the degree of narrative structure displayed in mothe r -ch i ld , fa ther -chi ld , and chi ldren 's experimenter-elicited narratives early in the preschool period and chi ldren 's personal narrative skills 2 1/2-years later. Because a few children did not recall all three events queried, all analyses were computed based on the mean fre- quency of each element per event rather than proport ions of total conversation.

Parental Narrative Structure Over Time

Analyses of parental utterances focused first on mothers ' and fathers ' use of actions, descriptions, orientations, and evaluations by child gender over time. Recall that orientations and

evaluations were subcategorized whereas actions and descriptions were not. Therefore, separate analyses were performed for each element of narrative structure.

An initial analysis considered the total number of parental narrative codes as a measure of total amount of narrative structure provided. A 2 ( t ime point ) x 2 (gender of chi ld) x 2 (parent ) mixed model analysis of variance (ANOVA) was conducted with t ime and parent as within-subjects factors and gender of child as a between-subjects factor. Results revealed a main effect of time, F ( 1 , 13) = 5.49, p < .05. Mothers and fathers used more elements of narrative structure over t ime (mothers Time 1, M = 37.69, SD = 15.05, mothers Time 2, M = 51.47, SD = 29.54; fathers Time 1, M = 36.78, SD = 21.14, fathers Time 2, M = 57.83, SD = 34.10) , but no differences between mothers and fathers or with daughters or sons emerged. Also of note in this analysis was the substantial variability in amount of narrative elements used, indicating marked individual differences in the provision of personal narrative structure.

For referential actions, a 2 ( t ime point ) x 2 (gender of chi ld) x 2 (parent) mixed-model ANOVA was performed with t ime and parent as within-subjects factors and gender of child as a between-subjects factor. Results revealed a marginally significant main effect of time, F ( 1, .13) = 3.65, p = .08. As shown in the upper port ion of Table 2, mothers and fathers tended to increase their comments about the actions that occurred during events over t ime (Time 1, M = 8.84, SD = 3.63; Time 2, M = 12.65, SD = 7.06). Further shown in the upper port ion Table 2, a 2 ( t ime) x 2 (gender of chi ld) x 2 (parent ) mixed-model ANOVA performed for parental referential descriptions revealed a significant main effect of time, F ( 1 , 13) = 5.29, p < .05. Mothers and fathers increased their use of descriptive details in

300 HADEN, HA~NE, AND FIVUSH

Table 2 Parents' and Children's Mean Number and Standard Deviation of Referential Actions and Descriptions Per Event by Parent and Child Gender Over Time

Actions Descriptions

Mothers Fathers Mothers Fathers

Time point M SD M SD M SD M SD

Time 1 Girls 7.95 5.75 11.52 7.48 15.29 9.37 15.52 8.13 Boys 9.69 3.02 6.42 3.70 13.44 4.72 9.88 6.39 M 8.88 4.42 8.80 6.15 14.30 7.05 12.51 7.57

Time 2 Girls 10.19 3.72 13.10 5.13 19.24 7.58 21.86 10.71 Boys 11.21 7.87 15.85 13.34 18.25 12.61 18.85 13.43 M 10.73 6.10 14.57 10.11 18.71 10.22 20.26 11.91

Girls Boys Girls Boys

Time 1 Mothers 1.67 1.37 2.35 2.02 5.71 2.94 5.15 3.24 Fathers 4.36 3.55 3.29 1.74 8.36 4.30 5.00 2.18 Exp. 5.12 3.12 3.58 1.38 5.90 5.19 4.83 2.30 M 3.71 0.76 3.08 0.89 6.66 2.79 4.99 1.83


Note. Exp. = experimenter.

personal narratives with their children over time (Time 1, M = 13.41, SD = 6.19; Time 2, M = 19.48, SD = 9.07).

For orientations, a 2 (time point) x 2 (gender of child) x 2 (parent) x 2 (type of orientation: spatial-temporal or person) mixed-model ANOVA was performed with time, parent, and type of orientation as within-subjects factors and gender of child as a between-subjects factor. Results revealed a significant main effect of orientation type, F (1 , 13) = 83.92, p < .001, and a marginally significant 2-way interaction of Time X Type, F ( 1, 13) = 4.03, p = .07. As shown in the upper portion of Table 3, mothers and fathers were providing more spatial-temporal orientations than person orientations, and tended to increase in their provision of spatial-temporal orientations, but not person orientations, over time (spatial-temporal Time 1, M = 3.64, SD = 1.57, Time 2, M = 4.88, SD = 2.34; person Time 1, M = 3.19, SD = 1.85, Time 2, M = 2.96, SD = 1.44).

Finally, for evaluations, a 2 (time) x 2 (gender of child) x 2 (parent) × 4 (type of evaluation: internal states, intensifiers, affect-modifiers, emphasis) mixed-model ANOVA was performed. Results revealed main effects of time, F ( 1, 13) = 10.13, p < .01, and type, F (3 , 39) = 30.97,p < .001, to be interpreted in light of a significant Time x Type interaction, F (3 , 39) = 4.01, p < .01. As shown in the upper portion of Table 4, mothers and fathers use of all four evaluation types increased significantly over time, although to varying degrees of magnitude of change. At each time point, intensifiers were the most prevalent evaluative device and were used significantly more than affect modifiers and internal reactions; emphasis was used less than any other evaluative device (TUkey post hoc tests, p < .05).

Overall, these analyses indicate mothers and fathers are quite

similar to each other in their provision of elements of narrative structure and cohesion with their children over time. Moreover, parents do not appear to decrease their scaffolding of these conversations over time. Both mothers and fathers increased similarly in their use of orientations, referentials, and evaluations across the preschool period, such that these narratives became more contexted, detailed, and evaluative over time. Importantly, mothers and fathers did not use elements of narrative structure differently in personal narratives with daughters and sons.

Children's Narrative Skill Development

Analyses of the children's narrative skills first focused on girls' and boys' use of actions, descriptions, orientations, and evaluations by partner over time. As with the parental narrative codes, orientations and evaluations were coded into subcategories; thus, analyses of narrative structure for children were performed separately for each element. An initial analysis looked at the total amount of narrative structure children were providing in their conversations with their mothers, fathers, and experimenters. A 2 (t ime) x 2 (gender of child) x 3 (partner: mother, father, or experimenter) mixed-model ANOVA was performed with time and partner as within-subjects factors and gender of child as a between-subjects factor. The analysis revealed a significant main effect of time, F(1 , 13) = 43.25, p < .001, a significant main effect of partner, F (2 , 26) = 3.19, p < .05, and a marginal effect of gender, F (1 , 13) = 3.80, p = .07. Children's provision of narrative structure increased over time (Time 1, M = 14.81, SD = 4.31; Time 2, M = 38.15, SD =


Table 3

Parents' and Children's Mean Number and Standard Deviations of Orientations Per Event by Type, Parent, and ChiM Gender Over Time

Type of orientation

Spatial-temporal Person



Time 1 Girls 3.48 2.04 4.02 2.37 3.05 3.43 5.14 3.81 Boys 3.23 1.51 3.88 2.41 2.02 0.96 2.79 1.78 M 3.34 1.71 3.94 2.31 2.50 2.41 3.04 1.86

Time 2 Girls 5.05 2.22 4.57 3.00 3.48 1.86 3.14 1.81 Boys 4.63 2.47 5.27 3.74 2.67 1.89 2.63 1.68 M 4.82 2.28 4.94 3.31 3.04 1.86 2.87 1.70




Note. Exp = experimenter.

15.97). Also children are providing more narrative structure with their fathers (M = 28.77, SD = 11.56) and to the experimenters (M = 29.96, SD = 14.50) than with their mothers (M = 20.71, SD = 11.49; Tukey post hoc tests, p < .05). Overall, girls tended to use more elements of narrative structure than boys (girls, M = 31.01, SD = 10.02; boys, M = 22.52, SD = 6.75).

For actions, a 2 (t ime) x 2 (gender of child) x 3 (partner) mixed-model ANOVA was performed with time and partner as within-subjects factors and gender of child as a between-subjects factor. The analysis revealed significant main effects of time, F(1 , 13) = 40.04, p < .001, and partner, F (2 , 26) = 5.86, p < .01. As shown in the lower portion of Table 2, children are talking more about actions over time. Children overall are also including more action information in their narratives with their fathers and experimenters than with their mothers (Tukey post hoc tests, p < .05). For descriptions, a 2 (t ime) x 2 (gender of child) x 3 (partner) mixed-model ANOVA revealed a significant main effect of time, F (1 , 13) = 33.61, p < .001. As shown in the lower portion of Table 2, children are also increasing dramatically in their contribution of descriptive information across the 21/2-year period.

For orientations, a 2 (t ime) x 2 (gender of child) x 3 (partner) X 2 (type of orientation: spatial-temporal or person) mixed-model ANOVA was performed with time, partner, and type of orientation as within-subjects factors and gender of child as a between-subjects factor. Analyses revealed a significant 4- way interaction, F (2 , 26) = 3.92, p < .05, a significant 3-way

interaction of Partner X Time × Type, F (2 , 26) = 4.02, p < .05, a significant 2-way interaction of Child Gender x Type, F (1 , 13) = 6.07, p < .05, and marginally significant 2-way interactions of Child Gender × Time, F (1 , 13) = 3.95, p = .07, and Partner × Type, F(2 , 26) = 2.82, p = .08. Significant main effects of time, F ( 1, 13) = 22.78, p < .001, child gender, F (1 , 13) = 7.22, p < .05, and type, F (1 , 13) = 17.08, p < .001, were also obtained. To follow up these effects, further analyses were conducted separately for each orientation subtype by time, partner, and child gender. A 2 (t ime) x 2 (gender of child) x 3 (partner) mixed-model ANOVA on spatial-temporal orientations revealed a significant main effect of time, F ( 1, 13) = 66.79, p < .001, as well as marginal effects of partner, F (2 , 26) = 3.05, p = .06, and Time × Gender, F (1 , 13) = 3.51, p = .08. As shown in the lower portion of Table 3, children's contribution of spatial-temporal orientations increases over time, with girls tending to increase more in their provision of spatial-temporal orientations over time than boys. There was also some suggestion that children were including more spatial- temporal orientations in their narratives with their fathers and the experimenters than with their mothers, although this result did not reach conventional levels of statistical significance (Tu- key post hoc tests, p > .05). A 2 (t ime) x 2 (gender of child) × 3 (partner) mixed-model ANOVA on person orientations revealed significant main effects of time, F(1 , 13) = 4.82, p < .05, and child gender, F (1 , 13) = 8.82,p < .01. As also shown in the lower portion of Table 3, girls were contributing more person orientations in their narratives than were boys, but both


Table 4 Parents ' and Children's Mean Number and Standard Deviations o f Evaluations Per Event

by Type, Parent, and Child Gender Over Time

Type of evaluation

Internal states Intensifiers Affect-modifiers Emphasis

Mothers Fathers Mothers Fathers Mothers Fathers Mothers Fathers

Time point M SD M SD M SD M SD M SD M SD M SD M SD

Time 1 Girls 0.76 0.74 2.26 1.30 3.05 1.38 4.29 3.46 2.00 1.66 2.49 1.63 0.29 0.30 0.43 0.46 Boys 1.58 1.03 0.42 0.39 3.54 2.46 2.12 2.25 1.29 1.53 0.33 0.36 0.58 0.53 0.50 0.40 M 1.20 0.97 1.28 1.31 3.21 1.96 3.13 2.98 1.62 1.57 1.31 1.54 0.44 0.45 0.47 0.41

Time 2 Girls 1.76 2.01 2.57 2.42 5.52 3.67 6.48 4.45 4.52 3.16 3.67 2.89 0.95 0.97 1.14 0.92 Boys 2.29 2.54 2.90 2.71 5.92 6.40 4.75 3.70 3.21 3.73 1.92 1.52 0.83 0.85 1.00 0.85 M 2.04 2.24 2.74 2.49 5.73 5.13 5.56 4.02 3.82 3.42 2.73 2.35 0.89 0.88 1.07 0.86

Girls Boys Girls Boys Girls Boys Girls Boys

Time 1 Mothers 0.24 0.25 0.73 0.73 0.52 0.32 0.79 0.69 0.43 0.60 0.33 0.47 0.33 0.27 0.12 0.25 Fathers 1.59 1.26 0.54 0.77 0.64 0.28 0.79 0.67 0.62 0.34 0.46 0.77 0.79 0.84 0.46 0.35 Exp. 0.55 0.57 0.48 0.91 1.57 1.24 0.94 0.88 0.93 0.65 1.02 0.63 0.86 0.72 0.35 0.36 M 0.79 0.30 0.58 0.43 0.91 0.47 0.82 0.48 0.66 0.26 0.60 0.36 0.66 0.36 0.31 0.27

Time 2 Mothers 2.00 3.43 0.88 0.71 4.00 2.36 1.75 2.01 2.43 1.85 0.46 0.35 0.76 0.53 0.92 1.12 Fathers 3.43 3.33 1.42 0.99 3.43 2.59 2.96 1.97 2.09 1.52 0.94 0.84 1.24 0.60 1.27 1.02 Exp. 2.81 1.71 0.90 1.01 5.38 4.02 4.40 3.68 2.42 1.37 1.30 1.71 1.48 1.14 1.70 1.88 M 2.75 1.40 1.06 0.68 4.27 2.02 3.04 2.26 2.31 1.09 0.90 0.61 1.16 0.61 1.30 1.04

Note. Exp = experimenter.

boys and girls were contributing more of these sorts of orientations over time.

Finally, for Evaluations, a 2 (t ime) × 2 (gender of child) × 3 (partner) × 4 (type of evaluation) mixed model ANOVA revealed a significant 3-way interaction of Time × Child Gender × Type of Evaluation, F (3 , 39) = 3.46, p < .05, 2-way interactions of Time × Type, F (3 , 39) = 14.17,p < .001, and Partner × Type, F(6 , 78) = 4.22, p < .001, and main effects of time, F ( I , 13) = 27.95, p < .001, partner, F (2 , 26) = 3.45, p < .05, and type F(3 , 39) = 23.53, p < .001. A marginally significant main effect of child gender was also obtained, F( 1, 13) = 3.871 p = .07. To further explore these effects, a series of 2 (t ime) × 2 (child gender) mixed-model ANOVAs were performed on the individual evaluation types. Analysis of the internal state subtype of evaluation revealed main effects of time, F (1 , 13) = 20.60, p < .001, child gender, F (1 , 13) = 8.82, p < .01, and an interaction of Time × Child Gender, F ( 1, 13) = 7.54, p < .05. As can be seen in the lower portion of Table 4, girls were contributing more internal state information overall than boys, and girls increased more dramatically in the contribution of this type of evaluative information than boys over time.

For the intensifier subtype, a main effect of time, F (1 , 13) = 25.09, p < .001, was obtained. As shown in Table 4, boys and girls increased in their contribution of intensifiers over time. For the affect-modifier subtype, main effects of time, F ( I, 13) = 22.04, p < .001, and child gender, F (1 , 13) = 7.67, p < .01, were found, as well as an interaction of Time × Child

Gender, F (1 , 13) = 10.75, p < .01. As shown in Table 4, analysis of the simple effects revealed that girls were including more affect-modifiers in their narratives than boys and were showing a more dramatic increase in the use of affect-modifiers over time than were boys (p < .05). Lastly, for the emphasis subtype, a main effect of time, F(1 , 13) = l l . 1 3 , p < .01, was obtained. Girls and boys were increasing in their use of this evaluation type over time.

These results indicate clear developmental changes in the amount of narrative structure children were providing in personal narratives across the preschool years. Over time, both girls and boys included more actions, descriptions, orientations, and evaluations in narrating personal experiences with their mothers, fathers, and in their unscaffolded narratives elicited by an experimenter. However, whereas all children increased in their use of structural elements over time, girls included more person identifications and more internal state information about peo- ples' feelings and desires than boys, even at the earliest time point. Girls also included more affect-modifiers, which relate subjective and emotional responses of event participants, and increased more dramatically in their inclusion of this evaluative device than boys over time. Lastly, it is intriguing that children showed some more advanced narrative skills with their fathers and in narratives with an experimenter than in conversations about past events with their mothers. Girls and boys included more elements of narrative structure in personal narratives with their fathers and experimenters than with their mothers, especially more information about actions.


Longitudinal Relations in Personal Narrat ive Skill Deve lopment

The results indicate that even early in development, parents are engaging in highly contexted and. richly evaluative narratives about the past with their young children. Although parents gener- ally increase in their use of actions, descriptions, orientations, and evaluations over time, they are providing the basic structure to these narratives with their 40-month-old Children. Preschoolers show increasing abilities to include structural elements in "their personal narratives over time as well, such that these conversations are becoming more and more co-constructed over time. A critical question is whether the kinds of narrative structure displayed in mother-child, father-child, and children's unscaffolded narratives early in the preschool period relate to children's later inclusion of structural elements in their narratives. Variability in the degree to which parents and children incorporate elements of narrative structure in personal narratives early in development may relate to differences in children's later personal narrative skills. To explore this issue, hierarchical regression analyses were conducted (using the SPSS statistical package; SPSS, 1995) to assess the relative influence of parents' and children's early provision of kinds of narrative structure on children's unscaffolded narratives at 70 months of age. As we argued in the introduction, from a theoretical perspective, orientations and evaluations are the crux of personal narratives. Moreover; previOus research suggests that maternal use of orientations and evaluations is related to children's developing inclusion of these narrative devices in their unscaffolded recounts. Thus, our analyses focused on these two aspects of narrative structure for both theoretical and empiri- cal reasons. We were particularly interested in examining whether parental emphasis on particular narrative devices facilitated children's use of these devices over time.

In order to examine this, two ratio variables were computed for mothers, fathers, and children at each time point for use in

the regression analyses. The orientation ratio equaled the mean number of orientations (person and spatial-temporal) included in the narratives divided by the mean total of referential actions and descriptions (orientations / [ actions + descriptions ]). The evaluation ratio equaled the mean number of evaluations (internal states, intensifiers, affect-modifiers, emphasis) included in the narratives divided by the mean total of referential actions and descriptions (evaluations/[actions + descriptions]). Thus, these ratios reflected the degree to which parents and children were providing orienting and evaluative information in recounting what happened during a past event; the closer the ratio to 1, the more highly contextualized or richly evaluative the recounting (the Appendix displays the mean orientation and evaluation ratios for the mothers, fathers, and children at each time point by child gender).

As discussed in the introduction, the regression analyses were designed to answer two distinct theoretical questions. First, what are the possible effects of mothers' and fathers' use of orienting and evaluative narrative devices, over and above children's own earlier narrative competencies, in predicting children's use of these sophisticated devices in narratives to those who did not experience the events with them? Second, given that children are showing different narrative skills depending on their conversational partner, what are possible effects of children's own abilities to use orienting and evaluative narrative devices when scaffolded by their mother or father on their developing abilities to use these devices in unscaffolded personal narratives with an experimenter? Thus, two regression models were tested for each of the ratio variables. Due to constraints presented by our sample size, gender was not included as a factor in the regression analyses.

Regression analyses focused first on assessing the relative influence of mothers', fathers', and children's orientation ratio at the 40-month time point on children's orientation ratio with

Table 5 Hierarchial Regression Analyses Predicting Children's Independent Narrative Skills at 70 Months

Variable-forced entry B SE B /3 R 2 change F to enter

Criterion variable: Children's 70-month orientation ratio with experimenter

1.40-month child with experimenter orientation ratio 0.42 0.22 0.52 0.03 0.43 2. 40-month fathers' orientation ratio -0.54 0 .23 -0.64 0.17 2.50 3.40-month mothers' orientation ratio 0.60 0.32 0.50 0.19 3.49* 1.40-month child with experimenter orientation ratio 0.27 0.27 0.33 0.03 0.43 2. 40-month child with father orientation ratio -0.25 0 .30 -0.28 0.06 0.75 3.40-month child with mother orientation ratio 0.04 0.20 0.06 0.003 0.04

Criterion variable: Children's 70-month evaluation ratio with experimenter

1.40-month child with experimenter evaluation ratio 0.85 0.27 0.69 0.36 7.29*** 2. 40-month fathers' evaluation ratio -0.04 0 . 3 4 -0.03 0.0009 0.02 3.40-month mothers' evaluation ratio 0.99 0.38 0.50 0.24 6.80** 1.40-month child with experimenter evaluation ratio 0.79 0.30 0.63 0.36 7.29*** 2. 40-month child with father evaluation ratio 0.21 0.43 0.12 0.02 0.42 3.40-month child with mother evaluation ratio 0.09 0.29 0.08 0.005 0.10

Note. SE B = standard error of B. *p <.10. **p < .05. ***p < .01.


the experimenter at the 70-month time point. To provide the most stringent test of parental influence, in all regression analyses, children's ratios were entered before ratios from the father- child and mother-child interviews. Table 5 presents the results of the regression analyses. In this first analysis focusing on parental effects on children's abilities, children's orientation ratio with the experimenter when they were 40 months old was entered first, followed by the fathers' orientation ratio when children were age 40 months, with the mothers' orientation ratio when children were 40 months entered last. The overall model after entry of all three predictors did not achieve conventional levels of statistical significance, R = .63, F(3, 11) = 2.36, p = .13, although it did account for almost one-quarter of the variance in children's orientation ratio for their experimenter- elicited narratives at 70 months (adjusted R 2 = .23). A second analysis focused on the effect of children's early provision of orientations. This analysis predicting children's orientation ratio with the experimenter at 70 months from children's orientation ratio at 40 months with the experimenter (entered first), children's 40-month orientation ratio in the father-child interview (entered second), and children's 40-month orientation ratio in the mother-child interview (entered last) yielded no significant predictors and a statistically nonsignificant overall model.

Regression analyses were also conducted to assess the relative influence of mothers', fathers', and children' s 40-month evaluation ratio on children's evaluation ratio in the experimenter-child interview at the 70-month time point. In this analysis, children's evaluation ratio with the experimenter when they were 40 months old was entered first, followed by the fathers' evaluation ratio when children were 40 months, with the mothers' evaluation ratio when children were 40 months entered last. The overall model was significant, R = .78, F(3, 11 ) = 5.60, p < .01, accounting for half of the variance in children's evaluation ratio in the experimenter-elicited narratives at 70 months of age (adjusted R 2 = .50). In this model, children's evaluation ratio in the experimenter-elicited narratives at 40 months and the mothers' evaluation ratio at 40 months both contributed uniquely as significant predictors of children's evaluation ratio at 70 months. A second analysis predicting children's evaluation ratio with the experimenter at 70 months from children's evaluation ratio at 40 months with the experimenter (entered first), children's 40-month evaluation ratio in the father-child interview (entered second), and children's 40-month evaluation ratio in the mother-child interview (entered last) indicated that only children's 40-month evaluation ratio with the experimenter was a unique predictor. This model did not achieve conventional levels of statistical significance, R = .39, F(3, 11) = 2.31, p = .13, but accounted for 22% of the overall variance in criterion variable (adjusted R 2 = .22). Thus, the degree to which mothers, but not fathers, provide evaluative information, and the degree to which children include evaluations in unscaffolded, experimenter-elicited narratives early in the preschool period predicts children's later ability to recount evaluative narratives to an unfamiliar listener. Children's contribu- tions of evaluative information in early mother-child and father- child narratives, however, did not predict their later personal narrative skills to a great extent.

Discuss ion

In this study, we examined the developing relations between the ways in which parents and children structure their narratives

about past experiences across the preschool years. Similar to the few previous studies examining this issue, we found evidence that children's abilities to tell unscaffolded personal stories relates to parents' early use of elements of narrative structure in conversations about the past with their children. However, our results extend and expand on previous findings in important ways. This study examines longitudinally the ways mothers and fathers co-construct personal narratives with their children over the preschool years. In this way, we were able to assess changes in parents' and children's inclusion of orienting, referential, and evaluative information in their jointly told stories about past events over time. Further, in this study, we determined relations between parental and child use of narrative structure over the preschool period and children's developing abilities to relate their past experiences in narratives told without child-sensitive, adult scaffolding.

Interestingly, mothers and fathers did not differ systematically from each other in the ways in which they structured past narratives with their preschool children. There is a substantial literature documenting differences in maternal and paternal language (see Leaper, in press, for a meta-analysis and review), and many theorists have argued that mothers and fathers may play different, but complementary roles in facilitating their children's language development (Gleason, 1975; Mannle & Tomasello, 1987). Yet, in the context of talking about past events, both previous research on parental level of elaboration with this same data set (Reese et al., 1996) and the current findings, indicate few, if any, differences. Thus, we must be extremely cautious in drawing general conclusions about parental language differences; just as mothers show different language styles and strate- gies in different conversational contexts (Haden & Fivush, 1996; Hoff-Ginsberg, 1991 ), mothers and fathers may differ from each other in particular conversational contexts but not others. In particular, in tasks in which both mothers and fathers are informed interlocutors, as here when talking with their children about shared past experiences, mothers and fathers appear to scaffold these interactions with their children similarly.

In the context of joint reminiscing, parents displayed quite sophisticated narrative structure even with their young preschool children. In this way, parents may be helping their children to contextualize events, both in terms of space and time, and in terms of personal meaning beginning very early in development. Whereas scaffolding models would predict that parents would provide less structure as children provide more, we found that parents increase in their use of sophisticated narrative devices as their children develop more competent narrative skills. Thus, as we have argued previously (e.g., Fivush, Haden, & Reese, 1996; Reese et al., 1993), for reminiscing, the more appropriate model may be a collaborative "spiral" rather than a scaffold. Parents' underlying goals in talking about the past with their young children may not only be to teach their child to recount their past coherently and unscaffolded. The provision of more and more narrative structure by both participants over time may also reflect social goals of establishing and reestablishing interpersonal bonds through collaboratively co-constructing shared stories of the past.

Whereas parents were sensitive to their children's developmental level, parents did not narrate past experiences differently with daughters than with sons. Yet, again, there is good evidence


that mothers speak differently to girls than boys in a variety of conversational contexts (Leaper, in press; note also that few studies have examined whether fathers speak differently with girls and boys). Moreover, although previous analyses of this same data set indicated few differences between mothers and fathers in their use of elaborations, both mothers and fathers differed in their conversational styles depending on the gender of the child. Parents both elaborated more with daughters than with sons (Reese et al., 1996) and they used more emotion words with daughters than with sons (Adams, Kuebli, Boyle & Fivush, 1995). Clearly, various narrative interactions around, about, and with children may convey gendered messages about the construction of personal narratives; however, in focusing on narrative structure provided by parents of daughters versus parents of sons, no significant differences emerged.

Nevertheless, girls and boys differed in their use of narrative structure even at 40 months of age. Girls' narratives were longer, as indicated by a greater amount of narrative structure, and they were more contexted, in that girls included more orientations and evaluations in their narratives than did boys. Moreover, girls increased in their provision of orientations and evaluations to a greater extent than did boys. These results provide strong evidence that girls are more advanced in their narrative production than are boys. However, the reasons for this gender difference are not immediately apparent. That parents did not differ in their provision of narrative structure with daughters and with sons argues against a simple socialization explanation. However, recall that girls were specifically more likely to include information about internal states and to use affect as a modifier (as in "That was my favorite part. " ) in talking about the past, Thus, it may be that girls are particularly likely to include evaluative information that focuses on the emotional and psychological aspects of events. There is some suggestion in the literature that parents talk more about emotions with daughters than with sons (Dunn, Bretherton & Munn, 1987; Fivush, 1993), and this may account for girls' increased use of this type of evaluative information over boys'. Further, as mentioned earlier, there is evidence that mothers and fathers differ in their language use with daughters and with sons in other contexts, and these differences may indirectly influence girls' narrative skills. However, it is also important to point out that the girls in this study were not more advanced in their overall language skills than were the boys. Because these data came from a larger longitudinal project, we have measures of mean length of utterance (MLU) at the 40- month time point and performance on the Peabody Picture Vo- cabulary Test (PPVT) at the 70-month time point. There were no gender differences on either of these measures. Thus, the gender differences in children's narrative skill seem to have no clear explanation as of yet; certainly this issue deserves additional research attention.

Moreover, although mothers and fathers did not differ in how they narrated the past with their daughters and their sons, children narrated differently with their fathers than with their mothers. Both girls and boys gave longer and more oriented and evaluative narratives when reminiscing with their father and with the unfamiliar experimenter than with their mother. On the one hand, these results may seem curious, given the differences in the task demands presented by the different narrative interactions. Because mothers and fathers were informed co-narrators,

whereas the experimenters were uninformed, we might have expected more similarity in children's performance with mothers and fathers than with fathers and experimenters. On the other hand, Mannle and Tomasello (1987) argued that fathers may provide a bridge to the outside language community in that they demand higher levels of communicative competence than do mothers. In these and previous analyses (Reese et al., 1996), fathers did not differ from mothers in their use of narrative structure, or elaborations and evaluations with their children. However, Reese et al. (1996) did find that fathers were more repetitive in their questioning style than were mothers with their 70-month old children. Perhaps the patterns of children's provision of narrative structure across partners reflect similarities between fathers and experimenters in their use of a more repetitive questioning style than mothers. Other possible factors that might override the task demands include a history of language interactions such that children expect to have to express themselves more fully with their fathers and with strangers than with their mothers. In addition, nonverbal aspects of the interactions may have resulted in more narrative talk in conversations with fathers and strangers than with mothers. Again, whereas we cannot offer a definitive interpretation, the findings support the idea that gender of both parent and child may interact in interesting ways to produce different kinds of narrative interactions.

Important relations between parents' and children's use of narrative structure were also found. Previous research demon- strated correlations between parents' narrative structure at one point in time and children's narrative structure at a later point in time. Our research extends this finding by examining how children's own earlier narrative skills may influence Subsequent development, and what parental structure may add over and above this. We discovered that children's early abilities to provide evaluative unscaffolded narratives is a strong predictor of their later abilities to provide evaluative unscaffolded narratives during the preschool years. This may not be surprising, but it is critical information for understanding individual differences in narrative skill development. Of course, we first assessed children at 40 months of age, and we do not know what has contributed to individual differences by this point in time. Still, our results indicate that when examining models that posit socialization of cognitive skills, we must include measures of the child' s unscaffolded skill development as a baseline.

Yet, even when considering children's earlier skill level as a predictor, what parents (or at least mothers) do in this context matters. Although findings for orientations do not directly repli- cate previous findings (e.g., Peterson & McCabe, 1992, 1994; Fivush, 1991 ), there was some suggestion in the present results that mothers' use of orientations relates to children's later skills (these results were not statistically reliable but in the direction consistent with previous research). Most clearly, over and above the child's previous skill level, those mothers who emphasized evaluations when reminiscing with their child at 40 months of age had children who were emphasizing evaluations in their unscaffolded narratives at 70 months of age. This confirms the results of Fivush (1991), who found correlations between maternal use of evaluations in mother-child reminiscing at 32 to 35 months of age, and children's use of evaluations in unscaffolded narratives a year later.

Moreover, it was only children's focus on evaluation in narra-


tives with a relatively unfamiliar interviewer that predicted later unscaffolded narrative skill. That is, children's ability to incorporate evaluations into their narratives in joint reminiscing with either their mother or their father was not predictive of their later ability to do so with an interviewer. This pattern suggests that maternal facilitation of evaluative devices is not indirect (e.g., Yoder & Kaiser, 1989). That is, it is not the case that maternal emphasis on evaluations leads to concurrent facilitation of the child 's use of evaluations, which in turn, leads to later skills. If this were so, then we would expect children's emphasis on evaluation when in conversation with their mothers to predict their later emphasis on evaluation. Rather, the pattern of results suggest that what mothers do early in development directly influences children's skill development as they mature.

Overall, this study provides compelling new evidence about the ways aspects of narrative structure displayed in mother - child, father-child, and experimenter-chi ld talk about the past

• become a part of children's developing abilities to narrate their past over the preschool years. Although we must be cautious about generalizing our findings beyond a White, middle-class population, the results indicate that early parent-chi ld reminiscing is an important component of children's developing skills. As Labov ( 1982; Labov & Waietzky, 1967) argued, narratives about our past are interpersonal in that they are told to others in ways they can be understood by others. But they are also interpersonal in that they are structured in social interaction. As parents and children co-construct increasingly rich and detailed personal narratives together, mothers' early provision of evaluations of children's experiences influences children's later provision of evaluations when narrating their past experiences to others.

Re fe rences

Adams, S., Kuebli, J., Boyle, P. A., & Fivush, R. (1995). Gender differences in parent-child conversations about past emotions: A longitudinal investigation. Sex Roles, 33, 309-323.

Dunn, J., Bretherton, I., & Munn, P. (1987). Conversations about feeling states between mothers and their young children. Developmental Psy- chology, 23, 132-139.

Eisenberg, A. R. (1985). Learning to describe past experiences in conversation. Discourse Processes, 8, 177-204.

Engel, S. (1986). Learning to reminisce: A developmental study of how young children talk about the past. Unpublished doctoral dissertation, City University of New York.

Fivush, R. (1988). The functions of event memory: Some comments on Nelson and Barsalou. In U. Neisser & E. Winograd (Eds.), Remember- ing reconsidered: Ecological and traditional approaches to memory (pp. 277-282). New York: Cambridge University Press.

Fivush, R. (1991). The social construction of personal narratives. Mer- rill-Palmer Quarterly, 37, 59-82.

Fivush, R. (1993). Emotional content of parent-child talk about the past. In C. A. Nelson (Ed.), The Minnesota symposium of child psychology: Vol. 26, Memory and affect in development (pp. 39-77). Hillsdale, NJ: Erlbanm.

Fivush, R. (1994). Young children's event recall: Are memories constructed through discourse? Consciousness and Cognition, 3, 356- 373.

Fivush, R., & Fromhoff, E (1988). Style and structure in mother-child conversations about the past. Discourse Processes, 11, 337-355.

Fivush, R., Gray, J. T., & Fromhoff, F. A. (1987). Two year olds talk about the past. Cognitive Development, 2, 393-410.

Fivush, R., Haden, C., & Adam, S. (1995). Structure and coherence of preschoolers' personal narratives over time: Implications for childhood amnesia. Journal of Experimental Child Psychology, 60, 32-56.

Fivush, R., Haden, C., & Reese, E. (1996). Remembering, recounting, and reminiscing: The development of autobiographical memory in social context. In D. Rubin (Ed.), Remembering our past: Studies in autobiographical memory (pp. 341-359). New York: Cambridge University Press.

Fivush, R., & Hamond, N. (1990). Autobiographical memory across the preschool years: Towards reconceptualizing childhood amnesia. In R. Fivush & J. A. Hudson (Eds.), Knowing and remembering in young children (pp. 223-248). New York: Cambridge University Press.

Gleason, J. B~ (1975). Fathers and other strangers: Men's speech to young children. In D. P. Dato (Ed.), Georgetown University roundta- ble on language and linguistics (pp. 289-297). Washington, DC: Georgetown University Press.

Haden, C. A., & Fivush, R. (1996). Contextual variation in maternal conversational styles. Merrill-Palmer Quarterly, 42, 24-51.

Heath, S.B. (1983). Ways with words: Language, life and work in communities and classrooms. New York: Cambridge University Press.

Hoff-Ginsberg, E. ( 1991 ). Mother-child conversation in different social classes and communicative settings. Child Development, 62, 782- 796.

Hudson, J.A. (1990). The emergence of autobiographic memory in mother-child conversations. In R. Fivush & J.A. Hudson (Eds.), Knowing and remembering in young children (pp. 166-196). New York: Cambridge University Press.

Hudson, J. A., Fivush, R., & Kuebli, J. (1992). Scripts and episodes: The development of event memory. Applied Cognitive Psychology, 6, 483-505.

Hudson, J.A., & Shapiro, L. (1991). Effects of task and topic on children's narratives. In A. McCabe & C. Peterson (Eds.), New direc- tions in developing narrative structure (pp. 89-136). Hillsdale, NJ: Erlbaum.

Keman, K. T. (1977). Semantic and expressive elaboration in children's narratives. In S. Ervin-Tripp & C. Mitchell-Kernan (Eds.), Child discourse (pp. 91-102). New York: Academic Press.

Labov, U. (1982). Speech actions and reactions in personal narrative. In D. Tannen (Ed.), Analyzing discourse: Text and talk (pp. 219- 247). Washington, DC: Georgetown University Press.

Labov, U., & Waletzky, J. (1967). Narrative analysis: Oral versions of personal experiences. In J. Helm (Ed.), Essays on the verbal and visual arts (pp. 12-44). Seattle: University of Washington Press.

Leaper, C. (in press). M_eta-analyses of parents' talk to their children: Gender, context, and other influences. Developmental Psychology.

Mannle, S., & Tomasello, M. (1987). Fathers, siblings, and the bridge hypothesis. In K. E. Nelson & A. van Kteeck (Eds.), Children's language (Vol. 6, 23-42). Hillsdale, NJ: Erlbaum.

McCabe, A., & Peterson, C. (1991). Getting the story: A longitudinal study of parental styles in eliciting narratives and developing narrative skill. In A. McCabe & C. Peterson (Eds.), Developing narrative structure (pp. 217-253). Hillsdale, NJ: Erlbaum.

Menig-Peterson, C. L., & McCabe, A. (1978). Children's orientation of a listener to the context of their narratives. Developmental Psychology, 14, 582-592.

Miller, P. J. (1994). Narrative practices: Their role in socialization and self-construction. In U. Neisser & R. Fivush (Eds.), The remembering self." Construction and accuracy in the self-narrative (pp. 158 - 179 ). New York: Cambridge University Press.

Miller, P.J., Potts, R., Fung, H., Hoogstra, L., & Mintz, J. (1990). Narrative practices and the social construction of self in childhood. American Ethnologist, 17, 292-311.

Miller, P.J., & Sperry, L.L. (1988). Early talk about the past: The origins of conversational stories of personal experience. Journal of Child Language, 15, 293-315.


Nelson, K. (1988). The ontogeny of memory for real world events. In U. Neisser & E. Winograd (Eds.), Remembering reconsidered: Ecological and traditional approaches to memory (pp. 244-276). New York: Cambridge University Press.

Nelson, K. (1993). The psychological and social origins of autobiographical memory. Psychological Science, 1, 1-8.

Ochs, E., Taylor, C., Rudolph, D., & Smith R. (1992). Story telling as a theory-building activity. Discourse Processes, 15, 37-72.

Peterson, C. (1990). The who, when and where of early narratives. Journal of Child Language, 17, 433-455.

Peterson, C., & McCabe, A. (1983). Developmental psycholinguistics: Three ways of looking at a child's narrative. New York: Plenum Press.

Peterson, C., & McCabe, A. (1992). Parental styles of narrative elicita- tion: Effect on children's narrative structure and content. First Lan- guage, 12, 299-321.

Peterson, C., & McCabe, A. (1994). A social interactionist account of developing decontextualized narrative skill. Developmental Psychol- ogy, 30, 937-948.

Pillemer, D. B., & White, S. H. (1989). Childhood events recalled by children and adults. Advances in Child Development and Behavior, 21, 297-340.

Reese, E. ( 1995 ). Predicting children's literacy from mother-child conversations. Cognitive Development, 10, 381-405.

Reese, E., & Fivush, R. (1993). Parental styles of talking about the past. Developmental Psychology, 29, 596-606.

Reese, E., Haden, C. A., & Fivush, R. (1993). Mother-child conversations about the past: Relationships of style and memory over time. Cognitive Development, 8, 403-430.

Reese, E., Haden, C. A., & Fivush, R. (1996). Mothers, fathers, daughters, sons: Gender differences in autobiographical reminiscing. Re- search on Language and Social Interaction, 29(1), 27-56.

Rogoff, B. (1990). Apprenticeship in thinking: Cognitive development in social context. New York: Oxford University Press.

Sachs, J. (1983). Talking about the there and then: The emergence of displaced reference in parent-child discourse. In K. Nelson (Ed.), Children's language (Vol. 4, pp. 1-28). Hillsdale, NJ: Erlbaum.

Snow, C.E. (1983). Literacy and language: Relationships during the preschool years. Harvard Educational Review, 53, 165-189.

Snow, C. E., & Dickinson, D. K. (1990). Social sources of narrative skills at home and at school. First Language, 10, 87-103 .

SPSS, Inc. (1995). SPSS for Windows (Version 6.1.2) [computer soft- ware]. Chicago: Author.

Umiker-Seboek, D. J. (1979). Preschool children's intraconversational narratives. Journal of Child Language, 6, 91-109.

Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes. (M. Cole, V. John-Steiner, S. Scribner, & E. Souberman, Eds.). Cambridge, MA: Harvard University Press.

Yoder, P.J., & Kaiser, A. P. (1989). Alternative explanations for the relationship between maternal verbal interaction style and child language development. Journal of, Child Language, 16, 141-160.

A p p e n d i x

P a r e n t s ' a n d C h i l d r e n ' s O r i e n t a t i o n a n d E v a l u a t i o n R a t i o a n d S t a n d a r d D e v i a t i o n s

b y C h i l d G e n d e r O v e r T i m e

Ratio variables

Orientation ratio Evaluation ratio



Time 1 Girls 0.31 0.15 0.40 0.22 0.35 0.12 0.39 0.21 Boys 0.25 0.10 0.42 0.16 0.33 0.10 0.22 0.05

Time 2 Girls 0.29 0.08 0.25 0.10 0.43 0.21 0.36 0.18 Boys 0.31 0.24 0.32 0.22 0.40 0.17 0.34 0.11


Time 1 Mothers 0.30 0.29 0.25 0.16 0.28 0.21 0.31 0.17 Fathers 0.28 0.19 0.31 0.17 0.29 0.11 0.25 0.15 Experimenter 0.32 0.23 0.29 0.16 0.40 0.22 0.35 0.14

Time 2 Mothers 0.21 0.12 0.24 0.11 0.39 0.13 0.29 0.13 Fathers 0.76 0.09 0.75 0.09 0.37 0.18 0.29 0.13 Experimenter 0.26 0.16 0.15 0.14 0.45 0.28 0.32 0.13

Received November 27, 1995 Revision received June 15, 1996

Accepted June 15, 1996 •

developing narrative structure in parent-child reminiscing … · 2016-11-05 · narrative...

Documents