Summary and Interpretation of Major Findings

The first major question we addressed was whether there were differences in the amount of information that children ages 3 to 11 years provided about the tornado that devastated their town relative to the amount of information they provided about nontraumatic events that occurred before and after the tornado. Both 4 months and 10 months after the storm, conversations about the tornado were twice as long as conversations about the two nontraumatic events. The children themselves contributed more than twice the number of propositions to the tornado relative to the nontraumatic-event conversations (see Fig. 6.1). Whereas these patterns are consistent with the suggestion that children remembered more about the traumatic relative to the nontraumatic events, we caution against this interpretation. The conclusion is not warranted both because the verbal report and the event memory are not isomorphic, and because the basis for inclusion of the dyads in the study may have biased them to spend more time talking about the tornado than about the events before and after it. We suggest that the better question is the second that we addressed, namely whether the type of information the children provided about the traumatic and nontraumatic events differed.

There was clear evidence that the type of information included in conversations about traumatic and nontraumatic events differed. The differences were not apparent at the level of the details that the children provided about the events at either session, however. That is, per unit of conversation, the children provided the same amount of content about the tornado and the nontraumatic events. The only exception was among the oldest children, whose Session 2 contributions (but not their Session 1 contributions) were more infused with information about internal states (Bauer et al., 2005). Whereas, for the most part, children's contributions to the conversations about the tornado and the nontraumatic events did not differ in level of detail, there were differences between the event types in the breadth of the conversations. Tornado conversations were more likely to include contextual information, information about the causes and consequences of elements of the event, and temporal connections within and between elements of the event (Ackil et al., 2003). Thus the tornado conversations contained more of the features of a "good narrative" relative to the conversations about the nontraumatic events.

The third major question we addressed was whether there was evidence of differential socialization of contributions to conversations about the tornado relative to the nontraumatic events. The significant concurrent and cross-lagged correlations are schematically represented in Figure 6.5. At the first session (4 months after the storm) we observed concurrent correlations between maternal style (measured in terms of the ratio of elaborations to repetitions) and children's participation in both types of conversations. Correlations between maternal behavior and the amount of content that the children provided were strong enough to reach statistical significance only in the context of the tornado a.

Session 1

Session 2

figure 6.5. Schematic representation of statistically significant concurrent (i.e., within session) and cross-lagged (i.e., between sessions, over time) correlations within and between members of the dyads at Sessions 1 and 2. (A) is a representation of the relations for the tornado conversations, and (B) is a representation of the relations for the conversations about the nontraumatic events.

figure 6.5. Schematic representation of statistically significant concurrent (i.e., within session) and cross-lagged (i.e., between sessions, over time) correlations within and between members of the dyads at Sessions 1 and 2. (A) is a representation of the relations for the tornado conversations, and (B) is a representation of the relations for the conversations about the nontraumatic events.

conversations (see also Fig. 6.2). That is, mothers who used a higher ratio of elaborations to repetitions had children who contributed more to the conversations. The ways mothers and their children worked together to co-construct the story of the tornado is nicely illustrated in this excerpt from a tornado conversation 4 months after the storm:

mother: And then we went to Cub Foods and then what did we do?

child: Go to St. Peter home. Tornado hit then. mother: Yeah.

child: And I said there might be a tornado. And you said there is not going to be a tornado. mother: Did I say that? I don't remember that. child: [nods head yes]

mother: And where did we go when we got in St. Peter? child: Jackie and [X] go in the basement and there was not very safe down there in the [X] so we went into the cedar closet. mother: Yep, and we had the radio. child: [nods head yes]

mother: Yep, and do you remember when the radio went off? child: [nods head yes] 'Cause [X] was very very bad 'cause the line got cut.

mother: And what did you think when you came upstairs? child: It looked different. mother: Yep.

child: And it wasn't like it was before.

mother: No, that is right. It looked really different.

Ten months after the storm, maternal behavior was no longer related to children's overall levels of participation in the conversations. The strength of the correlations between maternal elaboration ratios and the amount of content that the children provided about the tornado also had moderated substantially (and was no longer statistically significant). In contrast, the strength of the relation between maternal behavior when talking about the nontraumatic events and the amount of content the children provided about them had increased (see Figs. 6.2 and 6.5).

We suggest that the correlations over time within the participant groups are the key to understanding the finding of stronger concurrent relations between mothers and their children at the first, relative to the second, session (see Bauer et al., 2005, for further discussion). That is, it seems that the lower degree of concordance when talking about the tornado at Session 2 relative to Session 1 is due to changes in the ways that mothers talked about the events at the two sessions, in the face of consistency in the ways that the children talked about the events. For mothers, the degree of concordance between behavior at Session 1 and Session 2 was moderate (see Fig. 6.3) and significant only for the tornado conversations (see Fig. 6.5). In contrast, for children, the correlations were substantial and significant for both event types.

We also asked whether mothers' and children's contributions to the conversations were related over time (i.e., the 6-month interval between sessions). In the context of the conversations about the nontraumatic events, we found that the children's behavior at Session 1 was related to their mothers' behavior at Session 2 (see Fig. 6.5). This type of relation has been observed between mothers and their older children in the context of conversations about largely positive or affectively neutral events: significant cross-lagged relations have been observed between children and their mothers beginning when children are roughly 5 years of age (e.g., Reese et al., 1993). In contrast, in the context of the conversations about the tornado, there was very little evidence of relations between children's behavior at Session 1 and maternal behavior at Session 2. There was, however, evidence of relations between maternal behavior at the earlier session and children's behavior at the later session. Both in terms of general levels of participation and in terms of the amount of content that the children provided, the relations were stronger in the context of the tornado conversations than in the conversations about the nontraumatic events (see Fig. 6.4). It seems then that mothers' approaches to the tornado conversations at the first session were important determinants of the children's approaches 6 months later.

The results from the first session of the tornado study are a replication of the findings obtained by Sales et al. (2003). Specifically, Sales and her colleagues examined patterns of correlation between parental (mostly maternal) and child contributions to conversations about traumatic and non-traumatic events at a single point in time. They found that parents who adopted more elaborative (as opposed to repetitive) questioning styles had children who made more contributions to the conversations about both event types. Together, the two studies indicate that the findings of relations between maternal narrative style and children's contributions to conversations about past events extend beyond positive or neutral event contexts to discussions about traumatic events. Across the two studies, the findings can be said to obtain for children as young as 3 and as old as 11 years of age.

With the tornado study, we also extended the existing literature by testing relations over time as well as concurrent relations. This extension is significant because the strongest test of a socialization model of autobiographical narrative development comes not from observation of relations between maternal and child behavior at a single point in time but from relations over time. In the present research, we observed such relations. Importantly, they were apparent both at the level of overall participation in conversations about past events and at the level of the amount of content that the children contributed to the discussions. This suggests that maternal style operates not only to engage children in conversations about the past but also to focus children's attention on the elements that make for a good story about the experience, including general orienting information, information about the activities in events and how they are connected, and, perhaps in particular, information about the emotional and other reactions of event participants.

In both Sales et al. (2003) and in the present study, in general, the magnitudes of the correlations between maternal conversational style variables and children's contributions were greater in the context of conversations about the traumatic relative to the more positive events. In the present research, this was true at Session 1 and for the cross-lagged relations between sessions; it was not the case at Session 2, however. Why might relations be stronger for negative or traumatic relative to more positive or neutral events? One possibility is that there is something about the nature of the representations of traumatic experiences that makes them especially "susceptible" to socialization influences. Perhaps it is the emotional trauma associated with the event representation that renders them "special" and thus particularly amenable to socialization by others. Two things argue against this suggestion, however. The first is that if stronger relations were a result of special features of representations of such events, then we would have expected to see stronger relations at both time points in the present research. As noted, relations at Session 2 were not especially strong.

The second argument against the notion that memory representations of traumatic experiences have special features that render them particularly good candidates for socialization is that stronger relations are seen not only for events that we recognize as highly stressful and traumatic (medical emergencies and a tornado) but also for less extreme negative experiences, such as occur in the course of everyday activities. In Burch, Austin, and Bauer (2004), we reported an analysis of relations between maternal and child contributions to conversations about positive experiences, such as family vacations and birthdays parties, and to conversations that featured everyday trials and tribulations, such as going on a picnic and being stung by a bee or going on a hike and falling on a hill. By virtually no definition would such experiences be considered "traumatic." Nevertheless, we found stronger relations between maternal style variables and 3-year-old children's participation in conversations about these mildly negative experiences relative to conversations about affectively positive or neutral events.

We suggest that it is not necessary to appeal to special features of memory representations of traumatic events for explanation of the stronger pattern of relations between maternal and child contributions in the context of their discussion. Instead, we suggest that the difference be attributed to the fact that negative experiences are—thankfully—relatively uncommon. As such, children do not have a familiar frame of reference within which to understand and interpret them. Consequently, parental "intervention" in the process of construction of representations of such experiences becomes all the more important. This suggestion is consistent with the patterns of differential relations that have been observed across event types. For instance, in Sales et al. (2003), parents focused more on the causes of events when talking with their children about medical emergencies relative to the nontraumatic events. In Burch et al. (2004), maternal behavior was related to the number of subjective interpretations and suggestions of causal understanding that the children expressed in the context of discussions of negative, but not positive or neutral, experiences. Finally, in the St. Peter tornado study, both concurrent Session 1 and cross-lagged correlations between maternal style and children's behavior were especially strong for the category of evaluations expressed in the context of conversations about the tornado (see Bauer et al., in press, for details). These findings are consistent with the suggestion that patterns of correlation between maternal and child behavior may be tighter for negative or traumatic experiences because it is with these unique events that children require the most assistance in understanding, interpretation, and evaluation (see Fivush & Baker-Ward, 2005, for a similar argument).

The one exception to the general pattern of stronger relations between maternal and child behavior in the context of the conversations about traumatic relative to positive or nontraumatic events is the finding in the present research of generally weaker correlations between the members of the dyads in these conversations at Session 2. Although the interpretation is speculative, we suggest that the explanation for this pattern may be found in the public nature of the traumatic event that the children and their mothers experienced. That is, the tornado—and the devastation wrought by it—was shared and experienced by the entire community. Thus, the event was talked about by mothers and their children, but also by everyone else in the town of St. Peter, Minnesota. Especially over time, both mothers and children would be expected to participate in a number of conversations about the event, with a variety of conversational partners. It is not unreasonable to expect that the strength of relations between any two conversational partners might weaken under the influence of additional conversational partners, each with her or his own unique perspective on the event. Indeed, the different perspective that might be gained by frequent discussions of the event with different individuals may help to account for the relative lack of consistency in mothers' approaches to the conversations at the two sessions. Although these are post hoc interpretations of the findings of the present research, the suggestions could be tested empirically in future research.

Speculation about the public nature of the tornado event and how it might have influenced the pattern of findings in the St. Peter study also suggests an important qualifier on the generalizability of the results of the research. It is possible that the findings—including the pattern of relations between maternal and child behavior—would be different for more private events, such as maltreating experiences. To the extent that such events are talked about at all, they likely are talked about with fewer conversational partners. The partners may or may not lend credence to the child's account, thereby allowing or disallowing the child to own the experience. In addition, partners may attempt to work with the child to reevaluate and reconsider the significance of the experience (see Fivush, 2004, for discussion). Each of these possibilities has implications for the ways in which events are incorporated into the child's autobiography. The public versus private nature of experiences thus should be taken into consideration before extending the conclusions of any particular study beyond their bounds.

0 0

Post a comment