Affective meaning making processes have been seen by various developmental and clinical theorists as being at the heart of mental health and psychopathology, and recent theoretical and methodological advances in developmental studies using narrative approaches hold great promise in examining this idea. Studies using narratives offer ways to study affective meaning making processes, their development within children and between children and parents, and their assessment not only with school-age children, adolescents, and adults but also with preschoolers. At the same time this research has the potential of feeding back to theories regarding both the nature and development of psychopathology as well as possible treatments and interventions. The goal of this paper is to describe narrative methods in developmental research and discuss their implications for developmental psychopathology.
We will first review the notion of affective meaning making processes, and proceed to a discussion of why narratives are particularly suited to examine such processes. The subsequent review of narrative studies and methods will be organized into three sections according to the focus on child, parent, and parent–child narratives. Within each focus we will examine three levels of analysis: (a) narrative organization and coherence, (b) narrative content, and (c) the behavior/interactions of the narrator(s). We will close with a discussion of the implications of this research for developmental psychopathology, with an emphasis on parent–child jointly constructed narratives as the meeting point of individual child and parent narratives.
Affective Meaning Making
At the core of research interest in narratives is an emphasis on affective meaning making, and the claim that narrative approaches and methods are particularly suited to study such meaning making (Wolf, 2003). We are meaning making creatures, organizing our experiences in an ongoing, moment by moment fashion, and attempting to extract meanings from the events that make our lives (Emde, 2003). In the experience of the child, these include the myriad of everyday events, routines, and practices that make up the fabric of life and contribute to the child's sense of predictability and security. They can also include uncommon, highly significant, and emotionally charged moments, and these too have to be emotionally integrated and organized (Stern, 1985). At every developmental level of the child, the capacity to generate coherent and constructive meaning and make sense of emotional events is thought to contribute to emotional regulation and integration, with profound implications for the child's emotional well-being and subsequent development (Cole, Michel, & O'Donnell-Teti, 1994; Main, Kaplan, & Cassidy, 1985; Thompson, 2000).
Meaning making as discussed here does not necessarily imply highly conscious, cognitively complex processes. Rather, it also includes nonconscious appraisals, and these begin early in development and continue throughout life. Consider, for example, the phenomenon of social referencing (Sorce, Emde, Campos, & Klinnert, 1985). Here, the infant is faced with an ambiguous situation, and he uses the expressions and emotional signaling from the face, gesture, and voice of the caregiver to make sense of the experience. The infant's emotional experience will be impacted by such signals, as will be the plan of action that follows (e.g., approach or avoid the situation). The example highlights not only affective meaning making but also how intrapersonal processes are closely tied to interpersonal processes, an important theme of the narrative approach.
The affective component continues to be salient when language emerges, as evidenced in the early appearance of emotion words and their frequent use by parents and young children (Bretherton, Fritz, Zahn-Waxler, & Ridgeway, 1986). Language is a powerful tool for making meaning, allowing not only “on-line” communication about children's immediate experiences but also “off-line” discussions of past events that took place before the discussion, be it moments before or weeks, months, or years earlier (Fivush, Bohanek, Robertson, & Duke, 2004). One form of verbal communication involves narratives, and they play a particularly significant role in shaping the experiences of the child (Emde, 2003; Engel, 1999; Snow, 1990; Wolf, 2003). For example, Fivush et al. (2004) have studied reminiscing between young children and their mothers, and have shown how autobiographical memories, including those of emotionally salient events, are shaped by the dialogue between parent and child and by maternal talking style and are related to the memories of children assessed at a later time point.
Being inherently affective, meaning making has powerful implications for our emotional well-being and, more broadly, for understanding mental health and psychopathology, resilience in the face of stressors or dysregulation and disorganization. The basic premise is that affective meaning making processes are an integral part of normative development (Kaye, 1982) with important implications for the child's view of the self, others, and the broader social environment (Bowlby, 1988; Bretherton & Munholland, 1999). Such processes can promote growth, adaptation, and resilience, or can have a negative developmental impact. In addition, they can act as mediating and moderating variables that determine the short- and long-term impact of stress or trauma on the child (Fivush et al., 2004; Main, 2000; Werner, 2000).
Affective meaning-making processes impact the internal representations children form, and in turn, are impacted by these representations. Such representations have been referred to by different authors as internal working models (Bowlby, 1988), “patterns of being with” (Stern, 1995), or interpersonal scripts (Bretherton, 1990). Various models of psychopathology have implicated the structure and content of internal representations in the development and maintenance of psychopathology (Fonagy et al., 1997; Kernberg, 1987; Lyons-Ruth, Bronfman, & Atwood, 1999). The different terminologies referring to internal representations stem from different models, but all share the emphasis on such representations as key structures crucial for understanding the development of individuals. An important function of internal representations is to make meaning from our experience, interpret our interactions with others, and predict the future course of these interactions (Bowlby, 1982), and therefore, they are central in the process of affective meaning making, and, as will be discussed later, their assessment is a major goal of narrative methods.
As mentioned above, affective meaning making takes place both at the individual level, as a constructive process “inside” the individual, and at the interpersonal level, between individuals. Conversations, discourse, and joint storytelling are part and parcel of everyday life and they powerfully shape the way we experience ourselves and others (Engel, 1999; Fiese et al., 1999). From the emotionally attuned (or mismatched) response of the caregiver to a 3-month-olds' smile (Beebe, 2000; Stern, 1985) to the organizing (or dysregulating) dialogue between a parent and a child about a quarrel in preschool (Koren-Karie, Oppenheim, Haimovich, & Etzion-Carasso, 2003), the interpersonal arena is where meanings are made, constructed, coconstructed, and reconstructed (Peterson & McCabe, 2004; Snow, 1990). Interpersonal affective meaning-making influences intrapersonal processes (including representations), and in turn, is influenced by them in an ongoing, reciprocal manner (Fogel, 1993). As will be reviewed later, this is reflected in narrative methods that have been developed to assess both children and parents individually as well as in joint narration.
Narratives and Affective Meaning Making
As discussed above, affective meaning making between caregiver and child emerges early, at the preverbal level, but the development of language and later, narrative skills introduces a new and powerful medium. Bruner has proposed two modes of thought: the paradigmatic mode that parallels scientific thinking and is concerned with empirical validity, and the narrative mode, which is concerned with a sense of “realness” (Bruner, 1986). For Bruner, we make sense from our experience by organizing it in narrative form. In other words, we are the “authors” of our life experience, and the narrative we compose defines the self, its wishes, goals, fears, and conflicts. The importance of narrative is also emphasized by Stern, who in his developmental sequence regarding self-development postulated a “narrated self” as emerging around the ages of 3 or 4 years on the basis of prior “senses of self” that begin developing early in life (Stern, 1985). For Stern, the narrated self begins to form the autobiographical history that ultimately evolves into a life story, such as the one a patient tells a therapist. Stern's view regarding the centrality of the narrated self led him to postulate in later work that even early, nonverbal experiences of the infant are organized in what he termed “proto-narrative envelopes” (Stern, 1995).
Wolf points out that what transforms an account, a factual report regarding a sequence of events, into a narrative is evaluation (Wolf, 2003). Through linguistic and pragmatic means the storyteller not only tells us what happens but, most importantly, evaluates what happens, highlights the significance of certain aspects over others, and situates the self within the matrix of important people and events. The factual account is transformed into an individual narrative through which the narrator defines the self and others. Wolf further points out that narratives take us from the “here and now” to the “then and there.” As the child becomes more involved in the narrative mode, his horizon expands in both time and space, opening up possible worlds (Bruner, 1986) that can hold great promise but also fear and anxiety. The role of the parent is crucial throughout this developmental process in guiding the child in the acquisition of narrative skills and in determining the emotional direction of the coconstructed narrative. In sum, narratives are postulated by various theoreticians to be powerful organizers of affective meaning, and developmental studies using narrative assessments have begun to demonstrate empirically some of these processes.
Methods for assessing child, parent, and parent–child narratives
The review of narrative methods will be organized according to the three foci of the studies in the field: the child, the parent, and the parent–child dyad (studies of marital and family level narratives such as Fiese et al. (1999) or Oppenheim, Wamboldt, Gavin, Renouf, and Emde (1996) are beyond the scope of this paper). Studies focusing on the child have typically employed narrative methods such as the MacArthur Story Stem Battery (MSSB; Bretherton, Oppenheim, Buchsbaum, Emde, & The MacArthur Narrative Group, 1990) or the Attachment Story Completion Task (ASCT; Bretherton, Ridgeway, & Cassidy, 1990) in which children are asked to complete in words and play-action stories that are elicited using emotionally evocative story beginnings. Studies focusing on parents use narrative methods (e.g., Working Model of the Child Interview: Zeanah, Benoit, Hirshberg, Barton, & Regan, 1995) in which they are interviewed about a range of parenting-related issues including their children's behavior and emotions. These methods emphasize not only general discussions regarding the child but also probe for specific examples, typically in the form of narratives, that focus on emotionally meaningful events from the life of the child. Finally, studies focusing on the parent–child dyad elicit narratives that are jointly told or coconstructed by the parent and the child around significant events from the life of the child (e.g., Fivush, 1994; Koren-Karie et al., 2003) or around fictional emotional themes that are enacted in play action and narrative (e.g., Oppenheim, Nir, Warren, & Emde, 1997).
The three-way division into child, parent, and parent–child narratives reflects the methods used in narrative studies but also an important theoretical point: individual processes (“what this event means to me, how I construct it”) are not only intrapersonal, independently constructed in isolation from the social context, but are also powerfully influenced by the communication patterns between the child and emotionally important dialogic partners (Bowlby, 1988; Fivush, 1994; Peterson & McCabe, 2004). Conversely, dialogues do not start afresh each time they take place. Past dialogues shape the internal representations each of the partners has of the self and the other, and these have a powerful presence in the unfolding of current dialogues and exert important influence on their flow and development.
Narratives are thus fundamentally dialogic (Howard, 1991). For example, when a child completes a story stem involving the theme of the child-character committing a transgression, the story completion can be seen as a product of an internal dialogue between the child's action in defiance of a parental rule and the parental prohibition. In other words, the narrative told by the child reflects an internal drama in which the child and the parent play important roles. Symbolically, therefore, the parent is present in the child's individual narrative construction. Alternatively, when a parent responds to a question about the child's behavior or emotional response, the response is not a product of a static, fixed representation in the mind of the parent but the product of an unfolding of an internal dialogue between the parent and the child as well as other important people in the parents' mind (e.g., spouse, the parent's parent). In sum, the individually constructed narrative is imbued with the presence of the other, and the coconstructed narrative is imbued with the history of each of the two partners.
Child, parental, and parent–child narratives will be examined from three perspectives (a) the organization of the narrative, and more specifically its coherence; (b) the content of the narrative, or more specifically the predominant emotional themes or representations characteristic of the narrative, and (c) the behaviors and interactions of the narrator or narrators. These three dimensions are similar to those proposed by Fiese et al. (1999) in their discussion of family stories: narrative coherence, relationship beliefs, and narrative interaction. Fiese et al. (1999) base their three-dimensional approach on parallel distinctions with regard to language: syntax refers to how language is organized and parallels organization/coherence in the present paper; semantics refers to the meaning of language and parallels predominant emotional themes and representations here; and pragmatics refers to the conversational nature or act of using language (e.g., turn taking, affective tone) and parallels narrative behavior/interactions here.
The three dimensions of narrative analysis: Separate but inseparable
The distinction between the three narratives dimensions described above represents three perspectives on a whole, the whole being the narrative as told by child, parent, or jointly by both. In other words, the dimensions should not be reified, and the holistic nature of narratives should be kept in mind. Furthermore, although the dimensions are somewhat independent, and we can observe, for example, highly coherent narratives with either predominantly positive or negative affective themes, there are sometimes links across these dimensions in which the boundaries between them are more difficult to delineate. For example, highly aggressive content themes in children's story completions are typically not an appropriate response to the story stem, and therefore lower the coherence of the narrative. In this example, content (i.e., aggressive themes) and coherence are not independent, or high levels of behavioral inhibition may limit the child's cooperation with the task, and consequently limit the content themes in the narrative. In this example, the content of the narrative and the interaction of the child with the examiner will be linked. Nonetheless, the distinctions between the three levels of analysis are useful and important.
For example, a reasonable question for proponents of narrative assessments is their added value compared to more conventional assessments. In other words, are there unique features to narrative tasks that make them more revealing than other, nonnarrative tasks? When a child with disruptive behavior constructs a narrative with aggressive themes depicting unavailable parental figures, are we gaining an understanding of the child's negatively biased affective meaning making or are we “merely” seeing an expression of the child's behavior problems? Conversely, when a parent–child dyad have difficulties agreeing on a “good” example for when the child felt scared and the parent claims that the child's example is “not correct,” are we learning about their difficulties in developing a joint narrative about fear themes or is this “merely” another expression of the power struggles experienced by the dyad? Coding of the three dimensions separately allows us to begin to answer such questions. For example, it is possible to control for the child's behavior during the narrative task and examine the unique importance of themes or coherence.
Keeping the three dimensions separate is important because they can be differentially related to correlates or outcome measures. For example, in one study the coherence of children's narratives was found to be more consistently related to their attachment relationships than were parent or child representations (Sher-Censor & Oppenheim, 2004). In another study, the emotional coherence of the child during parent–child narrative coconstructions was related to children's story completions even after controlling for individual differences in children's narrative behavior (in this case, responsiveness to the examiner; Oppenheim, Nir, et al., 1997).
Finally, differentiating between the three dimensions is of theoretical and clinical significance because it can lead to greater specification of the mechanisms involved and the domains that should be targets for intervention. For example, if coherence of child narratives is the dimension most strongly related to early assessments of attachment, this may have important implications regarding both our understanding of the expression of security during the preschool years (e.g., secure attachment as coherent mental organization) and possible targets for intervention (e.g., increasing parental support of coherent child narration).
Child narrative assessments
The most commonly used method for assessing young children's emotion narratives has been the MSSB developed by Bretherton et al. (1990). In this method children are presented with story beginnings, or stems, and are asked to complete the story. The MSSB evolved from three lines of investigation using story-completion approaches with young children. Bretherton et al. (1990) developed the ASCT, a precursor of the MSSB, based on the work of Main, Kaplan, and Cassidy (1985) on internal working models of attachment, but also based on research regarding young children's symbolic play by Wolf, Rygh, and Altshuler (1984), the development of scripts by Nelson and Gruendel (1981), and the development of story schemas or grammars by Mandler (1983). These studies pointed to the early development of children's narrative skills, and suggested that most preschoolers have mastered the building blocks required to complete story stems, particularly if they were emotionally engaging and presented not only verbally but also using dolls and props. Bretherton et al. (1990) developed five story stems depicting attachment-related themes, and asked preschoolers to complete the stories. Results showed that children whose story completions were rated as more secure were also judged to be more secure based on concurrent observations as well as observations conducted at two earlier time points, and their mothers were judged as more insightful/sensitive based on an interview conducted at an earlier time point.
The second precursor to the MSSB was Emde and Buchsbaum's (1990) story-completion procedure designed to examine children's moral internalization including reactions to transgressions, prosocial behavior, conflict resolution, and empathy with peers. Buchsbaum and Emde (1990) described the responses of a group of 3-year-olds, and showed that most articulated coherent stories about moral rules, reciprocity, empathy, and internalized prohibitions. A later study using the same story stems described the development of children's capacity to acknowledge and keep in mind both sides of a dilemma between the ages of 3 and 5 (Oppenheim, Emde, Hasson, & Warren, 1997). The third precursor to the MSSB was Oppenheim's Attachment Doll Play Interview (1997). Here, a story-completion approach was also used around attachment and emotion themes, and children were asked to complete the stories and describe the characters' feelings. Children whose responses received higher ratings of emotional openness, constructive resolutions, and quality of mother–child interaction were rated as having higher self-esteem by their teachers, and showing more exploration of the classroom environment at the first days at preschool. In sum, the encouraging findings of Bretherton, Emde, Oppenheim, and colleagues showed that preschoolers could engage meaningfully in a story completion procedure and develop narratives that revealed important aspects of their emotional development as well as their relationships with their caregivers. These findings provided the impetus to the development of the MSSB by these three researchers and their colleagues.
In the MSSB, story stems are enacted in an emotionally engaging, playful way using a standard set of dolls and props. The stems end at the narrative “high point,” in which there is a problem or dilemma for the child to address and resolve, and at that point the examiner asks the child to “show me and tell me what happens next.” The child enacts the story completion using both words and play action, and the examiner has a predetermined set of follow-up probes that are dependent on the child's response. Examiners are carefully trained to facilitate the child's development of narratives and to refrain from a rigid, question and answer style. The careful attention to a warm-up period before the battery is administered, the pacing and presentation of the stories in a way that is sensitive to the child's emotional state, and the playful “feel” of the story-stem delivery, are all intended to get the child into the narrative “mode.” All of this goes hand in hand with standard administration, which includes careful control of the exact wording of the scripts, the timing and wording of follow-up probes, the specific dolls and props that are used, their arrangement on the table, and so on (Bretherton & Oppenheim, 2003).
The three dimensions of coherence, affective themes/representations, and child's behavior have all been applied to children's MSSB responses, and are represented in the main coding system that has been developed for the MSSB (Robinson & Mantz-Simmons, 2003; Robinson, Mantz-Simmons, Macfie, & The MacArthur Narrative Working Group, 1992). Other coding systems of the MSSB or other story-completion approaches (e.g., the ASCT) also rely on these dimensions in coding children's responses, although the dimensions are not necessarily coded separately. Some global coding approaches have relied on behaviors pertaining to two or three of the dimensions in order to identify overall patterns (e.g., security or insecurity of attachment, Bretherton et al., 1990; Gloger-Tippelt, Gomille, Koenig, & Vetter, 2002).
Coherence has typically been assessed using a scale ranging from incoherent to coherent that is applied to each story completion separately, following which mean scores are formed based on all story completions or on groups of stories that are selected based on common themes (e.g., stories the introduce attachment themes, stories that introduce discipline themes; Sher-Censor & Oppenheim, 2004). A slightly different approach designed to be particularly sensitive to incoherence due to its putative role as a marker of emotion dysregulation has been to dichotomize the scale into a coherent/incoherent split and count the number of incoherent story completions the child has produced (Olds et al., 2004).
Several studies have documented links between lower coherence of children's story completions and more emotional/behavioral problems (e.g., Oppenheim, Nir, et al., 1997; von Klitzing, Kelsay, Emde, Robinson, & Schmitz, 2000). Other studies have shown associations between low coherence of children's narratives and adverse past experiences. For example, Grych, Wachsmuth-Schlaefer, and Klockow (2002) have shown that children with a background of exposure to domestic violence told less coherent narratives than matched controls who were not exposed, and Sher-Censor and Oppenheim (2004) have shown that children who were insecure as infants told narratives at the age of 4.5 that were less coherent than those classified earlier as secure. Finally, attesting to the sensitivity of narrative coherence as a marker of children's socioemotional development, Olds et al. (2004) have shown that 6-year-olds who were part of a nurse home-visitation program as infants told narratives that were more coherent than control participants. As will be evident next, the coherence of children's narratives has been studied less frequently than the themes and representations children show, maybe because narrative coherence is related not only to children's socioemotional development and functioning but also to cognitive and linguistic attainments of the preschool and early school years (Wolf, 2003). Nonetheless, the findings regarding coherence are important given the established importance of coherence in adult narratives (Fiese et al., 1999; Main, 2000) and the (unexamined) important possibility that adults' coherent narration has its roots in childhood.
The second dimension of dominant themes/representations has been frequently assessed in children's story completions. Themes and representations have been coded using various approaches such as identifying specific content themes (e.g., empathy/helping, aggression in Robinson & Mantz-Simmons, 2003; prosocial, limit setting in Steele et al., 2003) and forming overall scores based on the number of such themes within and across story completions. An additional approach involved assessing representations of parental figures or child figures in the story completions with particular emphasis on the emotional tone and relational quality of the representation (e.g., positive, negative, controlling, disciplining; e.g., Oppenheim, Emde, & Warren, 1997; Toth, Cicchetti, Macfie, & Emde, 1997). This approach is particularly relevant to this paper because narrative representations of self and other are presumed to be related to internal representations or working models.
The exact nature of the relationship between the way children portray parental and child figures in their narratives and their internal representations or working models is unclear, however. Internal representations are typically considered to be generalized (e.g., Bretherton, 1990; Stern, 1985), whereas narrative representations are specific, produced in response to a story stem that presents the child with a circumscribed emotional theme or conflict. In addition, internal representations are typically thought of as complex cognitive–affective structures that govern thinking, feeling, and behavior (e.g., Main et al., 1985), and their link with the specific emotional themes in children's narratives or the way children portray parental and self-figures is likely not to be a simple one–one relationship. Children's narrative portrayals sometimes include depictions of specific, single occurrences rather than generalized models. They can also contain playful and fantasy themes, as well as wished-for or feared events. This conceptual complexity notwithstanding, studies that assessed themes and representations of self and other have been fruitful and yielded promising results.
In an early study Oppenheim, Emde, and Warren (1997) showed that children who had more positive and disciplinary maternal representations in their narratives had fewer behavior problems, whereas those who had more negative representations had more behavior problems. Page and Bretherton (2001) found that children who showed more supportive, comforting, and protective portrayals of mothers were rated by their teachers as having fewer conflicts with them, and that authoritative portrayals of the father figure were associated with social competence at the preschool. Children's background of adversity was also reflected in their representations. Children exposed to domestic violence had fewer positive representations and were more inclined to describe interparental conflict as escalating (Grych et al., 2002); those who had been maltreated described fewer parental and child responses to relieve distress and more role reversal (Macfie et al., 1999) and had more negative/constricted representations, less positive/coherent representations (Shields, Ryan, & Cicchetti, 2001), and more negative maternal and self-representations (Toth et al., 1997; see also Toth, Cicchetti, Macfie, Maughan, & Vanmeenen, 2000) than controls.
Finally, we move to the dimension of children's behavior during the narrative tasks. The Robinson et al. (1992) system referred to earlier taps this dimension with codes such as children's responsiveness to, involvement of, or controlling of the examiner. These measures have been useful in studies such as that of Toth et al. (1997), who found that physically and sexually abused children but not neglected children were more controlling and less responsive toward the examiner than controls. Coding these dimensions can also allow one to examine the associations between narrative variables (e.g., coherence, emotional themes) and external correlates (e.g., children's behavior problems, child–parent interaction) while controlling for the child's behavior during the task (e.g., Oppenheim, Nir, et al., 1997). Studies that employ such data-analytical approaches are particularly useful in demonstrating the importance and added value of narrative assessments. Such analyses can show that the associations between narrative variables and behavioral problems are not simply a function of the behavior problems generalizing into the assessment procedure and interfering with narrative production. Rather, they can help elucidate the specific affective meaning making processes that story-completion assessments can reveal and their link to various forms of behavior problems and emotion dysregulation.
Parental narrative assessments
The last 20 years have seen major advances in the development of narrative methods for parents designed to assess the way they construct affective meaning regarding themselves, their personal histories, and their current relationships. An exhaustive review of this work is beyond the scope of this paper. However, because child narration is likely to be powerfully impacted by parental narration (Fivush, 1994; Peterson & McCabe, 2004), and because child narration may be the basis of that child's future narrative capacity as an adult, this literature is briefly summarized. I will begin with the work of Main et al. (1985), because of its strong influence on the development of many of the adult narrative assessments developed later. Following a brief summary of this work the subset of parental narrative assessments that focus on the offspring will be reviewed.
Main's groundbreaking work (Main et al., 1985) regarding adults' narratives about their childhood experiences provided not only a method to elicit narratives, the Adult Attachment Interview (AAI), but also a theoretical and empirical approach for the analysis of these narratives. From the point of view of narrative research it was the latter that was perhaps most important. Specifically, Main (2000) suggested that what is most significant about adults' narratives about their childhoods is not the nature of the experiences they describe (i.e., whether they were positive or negative), but rather the state of mind of the parent in the present with regard to the childhood experiences they describe. Specifically, the parents' autonomy while discussing the past and their free exploration of wide range of memories and emotional experiences is crucial, and it is best indexed by the coherence of the narrative. Thus, the coherence of the narrative emerged from Main's research as a particularly salient dimension and the one most crucial for classifying AAI narratives into secure or insecure groups.
The validity of these classifications has received considerable support in a large number of studies, particularly in terms of predicting the parents' capacity to function as a secure base for the child and to foster the development of a secure child–parent attachment (Hesse, 1999; van IJzendoorn, 1995). These associations were important for attachment theory because they support hypotheses derived from the theory regarding the intergenerational transmission of attachment patterns. These findings are also important from the point of view of narrative research. They provide compelling empirical support to the idea that affective meaning making as revealed in narrative assessments, and particularly in the coherence of the narrative, is highly relevant for the emotional functioning of the individual and their relational capacities, such as the capacity to read emotional signals correctly and become a caregiver who provides security. In other words, narratives are not “merely” verbal constructions but are markers of very important intra- and interpersonal emotional processes with implications to the individual's emotional well being as well as to that of their offspring.
The powerful insights provided by the AAI propelled researchers to develop additional narrative measures for parents that focus on the child (rather than the parents' past, as is the case in the AAI). These were inspired by Bowlby's notion that caregiving, like care seeking, is guided by an internal representational system (Bowlby, 1982). Examples of such measures are the Working Model of the Child Interview (WMCI; Zeanah et al., 1995), and the Insightfulness Assessment (IA; Koren-Karie, Oppenheim, Dolev, Sher, & Etzion-Carasso, 2002). The WMCI interview was developed along the lines of the AAI with some questions similar to those in the AAI (e.g., question asking to describe the child using five adjectives that parallel the AAI question requesting description of the parent using five adjectives) and other questions focusing on the child's behavior and the parent–child relationship. As in the AAI, there is a request to support general statements about the child with specific examples and stories that illustrate the general statements. These are important because when specific examples support general statements the coherence of the narrative is enhanced, as are the completeness and genuineness of the portrayal of the child.
The coding system developed for the WMCI was also influenced by the coding of the AAI. Interviews are first transcribed, then rated on a series of rating scales, and finally classified into groups. Dimensions central to the AAI, particularly those that characterize the way the narrative is organized and the process of constructing the narrative such as objectivity, balanced descriptions, coherence, and reflectiveness also play a role in the coding of this interview. Additional scales were developed in accordance with the issues and themes tapped by these interviews.
The results involving the WMCI support its validity by showing a match between interview classifications of balanced, distorted, or disengaged with attachment classifications of secure, ambivalent, and avoidant respectively (Benoit, Parker, & Zeanah, 1997; Zeanah et al., 1995). Both organization/coherence and content/themes are taken into consideration when coding the transcripts and arriving at an overall classification, but coherence and organization seem to play a particularly important role. For example, the balanced classification is characterized not so much by the content of maternal descriptors (e.g., whether the child is described as easy/rewarding or difficult/challenging), but by the organization and coherence of maternal speech, as reflected in her ability to provide a balanced, well-supported, individualized, and empathic narrative.
Like the WMCI interview, the IA is based on parents' narratives about their children, but the focus of the discussion in this procedure are videotaped segments of the child interacting with the parent (Oppenheim & Koren-Karie, 2002). The IA focuses on the parent's capacity to see things from the child's point of view and show insightfulness into the motives underlying the child's behavior. The parent's narratives are elicited by interviewing them about their children's and their own thoughts and feelings after the parents watch the video segments. As in other parental narrative assessments, parents are asked to support their general descriptions regarding the child's characteristics with examples. Interview transcripts are rated on scales that assess various dimensions such as insightfulness, acceptance, and complexity, and an overall coherence scale. The scales provide the basis for classifying transcripts into one insightful and several noninsightful groups. The validity of these classifications has been supported by studies showing associations between maternal insightfulness, maternal sensitivity, and infant attachment (Dolev, 2006; Koren-Karie et al., 2002; Oppenheim, Koren-Karie, & Sagi, 2001). From the narrative point of view, the IA studies, like prior studies of parental representations, highlight the importance of the way parents construct a narrative regarding their children's inner world, emotional experience, and motives for behavior, and suggest that such narratives provide the meaning structure on which mothers' interactions with their children are based.
Particularly pertinent to this paper's focus on developmental psychopathology is an additional IA study involving mothers and preschoolers who were seen in a therapeutic preschool program for children with emotional and behavioral problems in which both children and mothers were treated (Oppenheim, Goldsmith, & Koren-Karie, 2004). Mothers and children were assessed before and after treatment, and the results showed that mothers who shifted from a noninsightful to an insightful stance had children whose behavior problems diminished, whereas those that remained noninsightful had children whose behavior problems got worse. Although it is not possible to determine whether a positive shift in the mothers led to improvements in the children or vice versa, the study highlights the importance of maternal insightfulness as a possible contributor to children's improved behavioral and emotional regulation. Again, it is important to emphasize that insightfulness was inferred from the narrative of the mothers regarding their children's behavior and internal experience; and, consistent with other studies of parental narratives, this study emphasizes that reflective, emotionally balanced, and coherent parental narratives regarding their children's behavior and internal experience are related and possibly contribute to improved behavioral and emotional regulation in the child.
From the point of view of the three dimensions for analyzing narratives described earlier (coherence, themes/representations, narrative behavior) there is interesting convergence between the various parental narrative measures reviewed above. Although each interview has its distinct theoretical focus, method, and coding system, a common thread across the measures is their emphasis on the coherence of parental narratives as one of the central features, particularly in terms of predicting the parent's capacity to provide security and show empathic and emotionally regulating parenting. Thus, when faced with questions and probes about their children in a broad range of contexts, including some that may be challenging, some parents are able to develop complex, insightful, accepting, well-organized, and believable narratives, whereas others have difficulties in this process. Studies of parental narratives suggest that this parental capacity appears to have important ramifications for children's emotional and behavioral functioning and, more broadly, their mental health. An important question that emerges from these results involves the mechanism linking adult narration to child behavior and emotional development. One such mechanism involves parent–child dialogues and narrative coconstructions, and we describe studies of these next.
Parent–child narrative assessment
As is apparent from the review above, child and parental narratives have received considerable research attention and generated a rather large corpus of empirical findings. Coconstructions of narratives between parents and children, on the other hand, have received less attention, even though they may hold great promise in explaining children's narrative development and the intergenerational transmission of patterns of affective meaning making from parent to child. Because parent–child narratives have been studied less frequently than child or parental narratives, a brief theoretical review is presented first.
Mothers begin talking with their children months before the children can respond verbally, treating them as agents, and laying the groundwork for the cognitive and linguistic skills needed to conduct dialogues. Children's contributions to their dialogues with their parents increase gradually with the cognitive and linguistic gains of the preschool years, although the significance of parents as partners for the coconstruction of personal narratives continues for many years, even after children are relatively autonomous narrators. Particularly important is the way such dialogues shape narratives around the myriad of events and experiences in the life of a child and endow them with affective meaning. Through such dialogues “generic” events become personal and specific with unique meanings, emotional nuances, and implications for the development of the child that are characteristic of a specific parent–child dyad. Examination of such dialogues allows us to go beyond the evaluation of the individual child and parental representational systems, accessed through their individual narratives, and observe the interplay between these systems, the way they influence each other, and the way the particular parent–child dyad negotiates the meaning an event may have for them.
Narrative coconstruction processes may have great significance for the child's emotional development, particularly when we consider the role of the parent as a secure base (Bowlby, 1988; Bretherton, 1990; Waters & Cummings, 2000). The parent does not only support the child's acquisition of narrative skills required to construct a coherent narrative. He or she also provide the emotional security needed to develop narratives around challenging, complex, and affectively charged events, memories, and thoughts. This sense of security facilitates the child's exploration of her internal world knowing that the parent is a reliable and sensitive partner and guide (Bretherton, 1990). Parents support such exploration by following their children's lead, encouraging them to talk about a wide range of emotional topics including those that children may find difficult to discuss, helping them organize their experiences into an emotionally coherent and meaningful narrative, and stressing the child's strengths as well as the adults' protection of the child and commitment to the child's security and emotional well-being (Etzion-Carasso & Oppenheim, 2000; Koren-Karie et al., 2003).
Parent–child narrative coconstructions do not always function to enhance the child's security, however, and under certain circumstances they generate and/or maintain a sense of insecurity, incoherence, and dysregulation (Koren-Karie, Oppenheim, & Getzler-Yosef, 2004). Some parents have difficulties guiding the child's discussion around emotionally difficult topics and join the child in avoiding such discussions. Such dialogues are shallow and side step important emotional issues, leaving the child with little capacity to integrate important emotional experiences into the self and possibly ill equipped to cope with such experiences in the future (Oppenheim & Salatas-Waters, 1995). Other parents have difficulties separating their experiences from those of the child, so that when a difficult topic comes up the focus of the conversation shifts to their own experiences, traumas, and fears. These parents may overwhelm their children with emotional themes or issues that are not related to the child and interfere with the development of a coherent narrative.
Empirical support for the significance of mother–child dialogues and narrative coconstructions comes from several directions, including studies that have shown associations between child–mother attachment and the nature of their emotion dialogues. For example, Strage and Main (1985) showed that mother–child dyads in which children were secure in infancy had fluent and balanced conversations, included expressions of emotions, and elaborated on a variety of topics, whereas dyads whose children were insecure as infants had restricted or dysfluent dialogues, showed little elaboration of topics, and asked rhetorical and repetitive questions. Along similar lines Etzion-Carasso and Oppenheim (2000) showed that 4.5-year-old boys who were securely attached as infants were likely to display open communication with their mothers following a separation, and disorganized children of both genders were likely to display nonopen communication. Gini, Oppenheim, and Sagi-Schwartz (2005) showed that 7-year-olds who were secure as infants were likely to show a mutual/balanced style of joint storytelling around themes of separation, whereas those who were insecure were likely to show a disengaged or overwhelming style. Finally, Oppenheim, Koren-Karie, and Sagi-Schwartz (2005) found that children classified as secure in infancy were more likely than those classified insecure to show at the ages of 4.5 and 7.5 years emotionally matched dialogues about children's emotional experiences, characterized by jointly constructed coherent stories with a clear link between the emotion discussed and the story provided and an emphasis on the child's competent coping.
These studies focused primarily on mother–child interaction during narrative coconstructions, and other studies focused on the content of the dialogues. For example, Farrar, Fasig, and Welch-Ross (1997) observed mother–child memory conversations and found that secure mother–daughter dyads engaged in relatively less negative memory talk but elaborated more on both positive and negative emotional themes than insecure dyads; the latter elaborated primarily on positive themes. In another study Laible and Thompson (2000) assessed dialogues between mothers and their 4-year-olds about an incident in which the child behaved well and an incident in which the child misbehaved. Secure dyads made more references to feelings and had more moral evaluations than insecure dyads.
The coherence/organization of the narratives produced by mother–child dyads was also coded in a few studies. For example, the coding used by Oppenheim et al. (2005) involves a coherence scale as does the Gini et al. (2005) coding system. One reason coherence was coded less often may be the difficulty in conceptualizing and assessing the coherence of a narrative jointly produced by two individuals (rather than a narrative produced by one person), although several studies have been successful in this task (Fiese et al., 1999; Gini et al., 2005; Koren-Karie et al., 2003). Also, certain aspects of the coherence of the joint narrative were captured by the interactional scales. For example, fluid, smooth turn taking (Strage & Main, 1985) and emotionally open (Etzion-Carasso & Oppenheim, 2000) dialogues are likely to contribute to the coherence of the narrative produced by the dyads.
In sum, studies of mother–child dialogues have shown that such dialogues represent an important arena for joint affective meaning making between parents and children. However, what are the implications of such dialogues for children? Research on autobiographical memory suggests that children's memory is shaped in powerful ways by their dialogues with their caregivers (Fivush, 1994; Fivush et al., 2004), but we know less about the implications for children's emotion regulation strategies, behavior problems, and mental health. Theoretically, we expect that dialogues are one important arena where children acquire affective meaning-making skills that they can rely on in coping with stressors and challenging events (Oppenheim, Nir, et al., 1997). Conversely, disrupted, confusing, emotionally unbalanced, or dismissive dialogues (e.g., Koren-Karie et al., 2004) leave children ill equipped to cope with subsequent challenges, and may increase the risk for psychopathology, particularly if other risk factors are present. This hypothesis has only little empirical support to date, and more research is needed on this issue. Particularly interesting can be studies in which mother–child dialogues are examined as moderating the impact of stress or trauma on children's emotion and behavior regulation.
Implications of Narrative Research for Developmental Psychopathology
The relations between narratives and psychopathology
The first and most basic implication is that theories of developmental psychopathology need to take into consideration affective meaning making and narratives to develop a more complete account of the origins and developmental course of psychopathology within the developing individual and also across generations. The research reviewed above indicates that psychopathology is associated with less optimal narratives from the perspective of organization/coherence, themes and representations, and narrative behavior. Generally speaking, there are findings in this direction regarding child, adult, and parent–child narratives. We know less, however, regarding the direction of effects underlying these associations. Are disturbed, incoherent, and dysregulated narratives “merely” epiphenomenal correlates of other disordered processes such as emotion dysregulation or relational disruptions/disorders or do such narratives contribute to emotional and behavioral disorders? From a developmental psychopathology perspective a third possibility would emphasize the bidirectional nature of influences between narrative processes and other developmental processes, and longitudinal as well as intervention studies are needed to understand such bidirectional effects.
A particularly interesting possibility both theoretically and clinically is the role of narratives as moderating variables. In this scenario the impact of stress or trauma on subsequent development is moderated by the affective meaning attributed to it through narrative. Open, empathic, sensitive, and well-organized dialogues between the child and a parent or other adult regarding a difficult or traumatic event can help the child develop a coherent narrative around the trauma with a meaning structure that facilitates coping and further development. Although the memory and experience of the trauma may remain part of the child's experience, their long-term pathological impact may be lessened when the child develops a coherent way to make sense of the trauma.
Unfortunately, when children are traumatized their parents are often traumatized as well (Lieberman, 2004), and therefore their availability as partners for narrative coconstruction may be significantly compromised. Intervention efforts should therefore focus on increasing parental emotional availability, as does child–parent psychotherapy (e.g., Lieberman, 2004; Lieberman, Van Horn, & Ghosh Ippen, 2005). Of interest, although this approach does not target parent–child coconstruction per se, these process are likely to improve as well during treatment due to its emphasis on improving the parent–child relationship and particularly the availability of the parent as a secure base, and may account for the gains children make (Lieberman, Van Horn, & Ozer, 2005). The specific effects of this form of treatment on child narration has been shown in a study by Toth, Maughan, Manly, Spagnola, and Cicchetti (2002), in which child–parent psychotherapy was compared to a psychoeducational approach in the treatment of maltreated children. The children who received child–parent psychotherapy showed in their narratives declines in maladaptive maternal representations, a decrease in negative self-representations, and an increase in positive mother–child relationship expectations, but the children who received the psychoeducational intervention did not show such changes. These results speak not only to the efficacy of the treatment, but also to the sensitivity of child narratives to changes in the dynamics and “webs of meaning” (Lieberman, Van Horn, & Ozer, 2005) coconstructed between parent and child. Child narratives improved only in the group that received the relationship-based treatment that focused on the type of interactions likely to enhance children's capacity to coherently and constructively narrate about emotionally charged themes.
Therapists may also be able to provide guidance in developing a coherent narrative around a traumatic event, and another effective treatment for child trauma, trauma focused cognitive behavioral treatment (Cohen, Deblinger, Mannarino, & Steer, 2004), has at its core the development of “trauma narrative” between the therapist and the child. The trauma narrative is designed to help the child make sense, integrate, and come to terms with the extremely difficult stress the child has experienced. The documented impact of this treatment (Cohen et al., 2004) provides some of the strongest evidence to the point made earlier about narrative processes as moderating the effect of trauma by demonstrating the contribution of coherent and emotionally regulating narration to children's successful coping with traumas.
Up to now, the focus was on child trauma, but what about trauma in the parents' past? How does it impact the parent's ability to facilitate coherent narration with the child? In our own research regarding mothers with a history of trauma (Getzler-Yosef, 2005; Koren-Karie et al., 2004) we have seen how higher levels of maternal resolution of trauma allowed mothers to guide the development of a joint narrative with the child about a range of emotional experiences in a way that was appropriately structuring, attuned, and sensitive. Lower levels of maternal resolution, on the other hand, were associated with over or understructuring, rigid and inflexible interaction, lack of attunement and empathy, and emotional dysregulation. Thus, lack of resolution of parental trauma interfered with the guidance of joint narration, and this may be one mechanism through which traumas are transmitted from one generation to the next. The notion of intergenerational transmission of trauma is not unique to a narrative approach, but narratives, and particularly those coconstructed between parent and child, can shed light on how the internal representations of parent and child get played out, interact with each other, and reinforce each other in ways that can foster or hinder the child's emotional development.
Child narratives as self-reports of psychopathology
One of the motivations for the development of story-stem narratives methods was to develop a “self-report” assessment of psychopathology for young children. Conventional methods in which the respondents report about their symptoms, thoughts, feelings, or behaviors have not been found useful with young children due to limitations in their capacity to reflect upon their experience and provide an accurate report. Clinicians and researchers have therefore relied on parental reports regarding the child's experience, but the need for self-report measures remains. The story-stem approach is designed to be attractive and “user friendly” for young children. Although it does not parallel self-report measures of psychopathology in that children are not asked to report directly about their symptoms, the accumulating data indicate meaningful associations between certain aspects of children's responses to story stems and their psychopathology.
For example, Warren, Oppenheim, and Emde (1996) found that distress and destructive themes were associated with externalizing behavior problems. In another study Warren, Emde, and Sroufe (2000) found the “negative expectations” evidenced in the story stems of 5-year-olds predicted internalizing and anxiety symptoms and maternal reports of separation anxiety 1 year later beyond the prediction offered by parental anxiety and child temperament. An additional line of research focused on dissociation in young maltreated children, and found that physically and sexually abused children showed more dissociation in their narratives than controls, and that over the preschool period dissociation increased in maltreated children but not in control children (Macfie, Cicchetti, & Toth, 2001). Summarizing the research, Warren (2003) described the markers of various forms of childhood psychopathology (e.g., externalizing and internalizing behavior problems) in children's story completions, and offered guidelines regarding the clinical use of the story-stem approach. There are currently several research groups who are advancing the clinical use of the MSSB. This line of research involves developing alternate versions of the MSSB for use in studies in which repeated use of the MSSB is needed, examining the change in children's narration during intervention, and using the MSSB with specific diagnostic groups to identify unique features of story completions characteristic of such groups.
The advantage of narrative assessments with young children is not only in offering a “self-report” measure of child psychopathology, however. Perhaps even more importantly, story completions increase our understanding of the child's experience of the world, of relationships, of availability of support in times of distress, and so on. For example, Hodges, Steele, Hillman, and Henderson (2003) described the use of a narrative approach with young, severely maltreated children, and how the narratives produced by the child can increase understanding and empathy to the child's inner world (e.g., rejection is portrayed whenever the child is in need, catastrophic fantasies are described, no portrayals of realistic, pleasurable, domestic play). Steele (personal communication, February 2005) described how children's narratives can facilitate not only therapists' understanding but also parental understanding. For example, in the case of an adopted child who had difficulties expressing her needs and was overly self-reliant, showing the parents segments of the child's story completion where she did express need helped the parents grasp the defensive nature of the child's self-reliance and respond more sensitively to subtle cues of neediness.
Narratives as ports of entry for intervention
The last example highlights the potential for dual use of narratives in interventions: as assessments and as ports of entry for enhancing parental insightfulness (Oppenheim & Koren-Karie, 2002) or reflectiveness (Fonagy, Gergely, Jurist, & Target, 2002). The use of story-stem methods as assessments was described above, and parent–child coconstruction methods can be used in a similar way. In fact, like story-stem procedures, coconstruction tasks are typically perceived as enjoyable by parents and children and are easy to administer. As described above, such procedures can be highly revealing of the strengths and the difficulties in the way the parent–child dyad processes emotional experiences. One constraint regarding the clinical application of both story-stem narratives and coconstructed narratives is that their coding is complex, labor intensive, and requires specific training. An important challenge for future work, therefore, is adapting research-based coding systems to clinical use. One direction for such adaptation can involve sensitizing clinicians to the processes underlying narrative construction and to specific markers than are particularly significant, rather than attempting to train clinicians to classify the child or parent into a research-based category (see Slade, 2004).
As to the use of narratives for enhancing insightfulness and reflectiveness, here the goal is to identify specific areas of difficulty in parental thinking and feeling about the child's internal experience. One way this goal can be achieved is by using video-replay techniques in which parents view video segments of selected moments of their interactions with their children and discuss them with the therapist. Video replay is used in a variety of clinical approaches (e.g., Marvin, Cooper, Hoffman, & Powell, 2002; McDonough, 2004), most typically with infants and toddlers. Children's responses to story stems and segments from parent–child narrative coconstruction tasks may be particularly interesting for video replay because they can help parents understand not only the interaction between themselves and their children, but, more specifically, the difficulties around representation and coconstruction of specific emotional themes that may underlie the child's difficulties. Because the target of intervention here is the parent's affective meaning making and, more specifically, difficulties in seeing the child's perspective, the research on the IA (Oppenheim & Koren-Karie, 2002) described above can be of great relevance in the identification of important dimensions for intervention and tailoring the intervention to the parent.
For example, in the study of insightfulness of mothers of preschoolers in treatment discussed earlier, Oppenheim et al. (2004) found that almost all mothers lacked insightfulness prior to the intervention, but that those that nonetheless showed relatively higher coherence scores on the IA were more likely to show gains following the intervention. Thus, elevated coherence scores, even among mothers classified as noninsightful, predicted a positive therapeutic outcome. This example illustrates only one of many ways in which a research-based procedure such as the IA can be useful clinically.
Future Research Directions
Several specific directions for future research that can advance the integration of narrative research and developmental psychopathology have been suggested above, and I discuss here two additional, general directions. The first involves examining the interrelations between parental, child, and parent–child narratives. Studies have mostly examined each domain separately, but a fuller understanding of affective meaning making in the parent–child dyad requires that all three domains be assessed. Such studies can provide a more complete account of the way parents' representations and their capacity to narrate coherently about emotionally significant aspects of their lives are also expressed in their dialogues with their children, and how such dialogues contribute to the child's subsequent narration and affective meaning making.
It will not be surprising if the links between parental, parent–child, and child narratives will not always be direct and simple, however. For example, parent–child narratives are not only influenced by the parent but also by the child, and child developmental and temperamental characteristics are likely to play a role in parent–child dialogues much more than in parent narratives. Another example for a disjoint between parent and child narratives can happen in intervention studies. In such studies the pace of change in the parent and child is not always the same. For example, if a parent gains therapeutic progress faster than a child, we may see gaps between parental and child narratives that may appear initially hard to explain. Over time, if the child's progress matches that of the parent, such uncoupling may disappear. These examples point to the need for sophisticated research designs that can illustrate both how parent, parent–child, and child narratives impact one another and are associated and the conditions in which lack of associations can be expected.
Studies of samples selected so that the risk is more clearly associated with the parent or the child can advance our understanding of narratives and psychopathology. For example, if the risk is associated with the parent, such as parental history of abuse (Koren-Karie et al., 2004) or mental illness, studies can examine the conditions under which the trauma or illness are transmitted to the child via dialogue and narrative coconstruction. If the risk status is associated with the child (e.g., the child is experiencing behavior and emotion problems), studies can examine the role of dialogues in the maintenance of the problems experienced by the child or in facilitating improvement in children's coping (e.g., Oppenheim et al., 2004).
A recent impressive example for a study linking parental and child narration in a high-risk context involves a study by Steele, Hodges, and Kaniuk (2005). In this study prospective adoptive parents were interviewed using the AAI prior to adopting children with histories of maltreatment. Children's story-stem narratives were collected following placement and 1 and 2 years later and coded for security, defensive avoidance, intense aggression, and disorganized controlling themes. Results showed an increase in children's security and a decrease in children's defensive avoidance over time, suggesting that children's story completions reflected the consolidation of the attachment relationships between them and their adoptive parents. Particularly significant were the interactions found between parental AAI narratives and children's MSSB narratives, because they illustrate the interplay between parental and child narratives and the need for studies that examine both. Children of mothers whose AAIs were classified secure had significantly less intense aggression scores and less disorganized/controlling themes at all three time points, and the decrease in defensive avoidance initially found for all children was found to be unique to children of mothers whose AAIs were classified as resolved with respect to trauma and loss. From the point of view of narrative research these findings are important because they demonstrate the theoretically predicted links between maternal and child narratives within the high risk context of late-adopted children, when there is no biological tie between parent and child and when parental narratives were obtained before the adoption. The findings point to the significance of mothers' capacity to develop coherent narratives regarding their own childhood histories, including experiences of loss or trauma, for the emotional development of the children they adopt including the children's evolving capacity to develop well-organized, emotionally open, and secure narratives. In a study such as Steele et al., as well as in other studies, measuring parental and child narratives examination of coconstructed narratives between parent and child might be particularly informative to illuminate the mechanism that can account for the links between them.
The second important direction for future research involves tracing the origins of adults' capacity to narrate coherently about their experiences. Although there are many studies, primarily using the AAI, that have documented the importance of coherence, openness, and objectivity in parental narration to the offspring's development, there is surprisingly little research about the developmental roots of such competencies. One source involves the adults' early history, and a few studies have documented continuity from secure attachment in infancy to secure narration in adulthood (Waters, Merrick, Treboux, Crowell, & Albersheim, 2000). The link, however, is certainly not direct (Lewis, Feiring, & Rosenthal, 2000). A narrative perspective should guide us to inquire regarding the coconstructive opportunities children have experienced with their caregivers and with others during their development. These can act as mediating or moderating processes between early childhood experiences and adult narration about these experiences. For example, children who enjoyed secure attachments in infancy are likely to engage in emotionally coherent dialogues with their caregivers later in development, and these can provide the basis for the kind of coherent narration secure speakers show in the AAI. In this case, we will see continuity over time. However, emotionally coherent dialogues can serve as moderating variables leading to lawful discontinuity as well. For example, a child who experienced divorce and resulting doubts about the stability and trustworthiness of close relationships might be able to find over time partners for dialogue that can facilitate the development of an emotionally coherent narrative around the divorce crisis. In such cases early adversity need not lead to an insecure state of mind due to the moderating impact of positive coconstruction experiences. Thus, studies that focus on dialogues and narrative coconstructions between children and significant people in their lives can be very informative in elucidating the pathways to adults' narratives and explaining developmental continuities and discontinuities.
In sum, narrative research has demonstrated the usefulness of a developmental psychopathology approach for bridging the gap between research and application. This line of research integrates knowledge and research on basic developmental processes, particularly regarding affective meaning making through narrative, with a focus on understanding and intervening with high-risk, traumatized children and families. This integration has already produced valuable findings and has raised important new questions and challenges for future work.
