Different approaches assessing mental representations of the mother-child-relationship in borderline personality disorder: how do mothers talk about their children?

Abstract

Mothers with borderline personality disorder (BPD) are assumed to perceive difficulties in the relationship towards their children. However, a detailed and comprehensive investigation on how mothers with BPD reflect and speak about their children as an indicator for mental representations of the mother-child relationship is yet missing. In two articles drawing from a cross-sectional data set of a multi-center study, we examined these aspects in mothers with BPD compared with a healthy and a clinical control group comprising mothers with anxiety disorders, depression, or both. Using the Five Minute Speech Sample, we focused on Expressed Emotion (EE), Narrative Coherence (NC) – both in the primary data analysis – and an extended assessment of Mind-Mindedness (extended MM) in the secondary data analysis. This article mainly addresses the similarities and differences between the three constructs on a conceptual level and regarding the respective study results, and additionally discusses the overall results from both studies. When comparing the three constructs, we identified differences regarding the level of analyzed content (i.e., characteristics of speech vs. characteristics of representation) and the level of orientation (i.e., reference to the mother herself vs. reference to the child). Three aspects are assessed via all operationalizations: (1) reporting negatively about characteristics or behaviors of the child, (2) reporting weaknesses of the child, and (3) descriptions of relationship quality. Mothers with BPD were more likely to express overt criticism (EE) and express more mental attributes with negative valence (extended MM) than both the clinical and healthy control groups. Given that we found particularities in the parent-child relationship in mothers with BPD across all three coding systems, we assume the overlap to be grounded in a tendency towards greater disapproval of child characteristics and reports of challenges in relationship quality. Considering the results of both studies, this article provides the most a comprehensive examination of these relational aspects in mothers with BPD, including comparisons with both healthy and clinical control groups. These insights contribute to a deeper understanding of the complexities underlying mental representations of the mother-child relationship in the context of BPD.

Theoretical background

The Five Minute Speech Sample (FMSS) [1] is a common method to assess Expressed Emotion, and more recently, several other coding systems based on the utterances of the FMSS were developed [2]. In the multi-center study “Preventing maltreatment and promoting mental health in children of mothers with borderline personality disorder” (ProChild), we assessed speech samples of three groups of mothers with small children aged 6 months to 6 years: 1) mothers with borderline personality disorder (BPD), 2) mothers with anxiety or depressive disorder or both (AD/D), and 3) mothers without any mental disorder in the last 7 years (CON). As part of the primary analysis, Jung et al. [3] focused on maternal mental representations of the child and the relationship in general, as aspects contributing to family climate, specifically coding criticism and emotional overinvolvement via Expressed Emotion (EE) [1, 4, 5] and Narrative Coherence (NC), focusing on the organization of the speech sample [6]. In the secondary analysis, Zitzmann et al. [7] focused specifically on mentalization, applying an adapted manual for coding Mind-Mindedness based on the FMSS (extended MM) [8] that considers additional characteristics of mind-related speech according to mentalization theory [9], such as a self- versus child-focus and a not-knowing stance, as well as the emotional valence of mental-state representation according to Demers et al. [10]. A detailed description of EE, NC, and extended MM as well as their assessment and the sample can be found in the respective articles [3, 7].

The three constructs are rooted in different theoretical backgrounds and set a different focus for understanding mental representations of the parent-child relationship. EE draws on the EE model with a focus on criticism and emotional overinvolvement [11], while NC is rooted in attachment theory and relates to the structuring of memory content [2]. MM also draws on attachment and, in addition, social-cognitive theories [12], and extended MM further on mentalization theory [9]. The following discussion will address the differences and similarities between NC, EE and extended MM.

Differences and similarities between Narrative Coherence and extended Mind-Mindedness

While NC focuses on how mothers organize and interpret attachment related information [13], MM refers to a mother’s attunement to mental state content [14] while the extended manual [8] encompasses additional central characteristics of mind-related speech according to mentalization theory [9]. Both coding systems address how the mother talks about herself in relation to her child, in NC indirectly as part of the focus subscale where a loss in focus (i.e., not talking about the child) is coded as incoherent, and in extended MM as part of the self- versus child-related mental attributes indices. This is also reflected in significant correlations between incoherence and the use of self-related as well as child-related mental attributes (see Table 1). However, while NC addresses the overall processing of information and reflect if these processes are flexible, unbiased, and consistent in itself [2], extended MM explicitly focuses on the extent to which a mother reflects on her child and herself in relation to subjective states and mental processes only [8, 15]. In NC coding, more behaviorally oriented or generic descriptions are also used for coding (e.g., “She is a charming girl, a very good girl. Compare to other children that I see around, she listens, […] she is polite. She is really a very good child. We have a strong relationship. That’s it, I have nothing else to add” [13] – coded as incoherent due to a unidimensional portrayal without meaningful examples), while the same descriptions may be no direct part of the extended MM coding (because it does not provide information on mental states of mother or child, such as emotions, desires or preferences). In contrast to other mentalization-based coding systems (e.g., Reflective Functioning) [16], extended MM coding does not assign a ‘low’ category to the absence of mental state language, as only explicitly mind-related attributes are included in the coding process. The absence of such references may only be indirectly reflected in a lower proportional MM score, resulting from an increased verbalization of non-mental aspects. In line with that, references to mental states of the mother herself or of her child (e.g., “He’s scared when I shout around”) are rated according to extended MM, while they may be no part of the NC coding because they lack information on the mental representation of the child in terms of any NC subscale (i.e., no loss in focus, no unelaborated reference, no indication of separateness issues or concern, no clear rejection or acceptance, and it does not indicate an uncomplex or incoherent representation).

Table 1.

Intercorrelations among NC and extended MM indices in the overall sample

	MM	MM (self)	MM (child)	MM (not-knowing)	MM (positive)	MM (negative)	MM (neutral)
NC	0.08	0.25**	− 0.11*	− 0.07	− 0.10	0.33**	− 0.12*

Note. NC: Narrative Coherence, MM: extended Mind-Mindedness. Correlations are calculated with N = 312 (sample size of article of Zitzmann et al.). * indicates p < .05. ** indicates p < .01. Higher scores in NC reflect higher incoherence; higher scores in MM reflect higher mind-mindedness indices

On the other hand, we see some shared variance between NC and extended MM, for example with the highest significant correlations between NC and mental attributes with negative valence, and between NC and self-related mental attributes in the extend MM rating, yielding small to medium effect sizes [17]. This indicates that mothers coded as incoherent use more negative as well as more self-related mental attributes, respectively.

Differences and similarities between Expressed Emotion and extended Mind-Mindedness

Despite their differing theoretical backgrounds, there is also a striking degree of overlap in the assessment of the constructs EE and extended MM. Both coding systems categorize maternal statements as negative, neutral or positive based on emotional valence, assuming that they reflect the mother’s representation of the child in their relationship [1, 8]. As both coding systems focus on valence and frequency of representational content, extended MM differs from other mentalization-based coding systems (e.g., Reflective Functioning). The latter consider the consistency or complexity of descriptions [16], whereas in extended MM, the primary focus is on the valence of mental state speech alone. In extended MM, valence is also coded in regard to the mother’s reflections on own mental states which is not done in the EE rating. This conceptual similarity is reflected in significant correlations between EE criticism (overt and covert) and extended MM negative mental attributes, indicating small effects (see Table 2). Consequently, statements such as “I hate it when my child behaves stubbornly” are coded as a self-related mental attribute (MM_self) with negative valence (MM_negativ) and as overt criticism (EE 1). However, a nearly identical statement (“Stupidly, my daughter hates my partner”) is only included in extended MM as a child-related mental attribute (MM_child) with negative valence (MM_negativ) but not in the EE rating. The latter focuses on the feelings and attitudes of the mother, while extended MM also reflects how the mother represents the mental state references (such as feelings and preferences) of her child [8]. In conclusion, statements regarding a child’s preferences or dislikes are not taken into account in the EE coding, on the condition that they are not accompanied by an indication of the parent’s own rejection. EE aims to provide information about how the mother encounters the child emotionally [18], while the extended MM valence indices attempt to depict how the mother represents her child and herself in relation to her child in terms of mental state references. This also means that negative, critical statements about the child can be included in the EE coding, even if they do not describe mental states (e.g., “My child is impossible to deal with” [18]), while they are not considered in the extended MM coding. In the EE coding, criticism can furthermore be coded due to a critical tone without any critical content or due to a negative relationship statement [1] – aspects that are not included in the extended MM rating if there is no reference to mental states. Accordingly, there are striking differences between both coding systems: (1) the exclusive focus on mental attributes in extended MM (but not in EE), and (2) the exclusive focus on statements that provide information about the mother’s feelings and attitudes towards her child in EE (but not in extended MM).

Table 2.

Intercorrelations among EE and extended MM indices in the overall sample

	MM	MM (self)	MM (child)	MM (not-knowing)	MM (positive)	MM (negative)	MM (neutral)
EE 1	0.01	0.04	− 0.02	− 0.00	− 0.00	0.19**	− 0.13*
EE 2	0.01	0.02	− 0.00	− 0.02	− 0.08	0.23**	− 0.09

Note. EE1: Expressed Emotion (overt criticism), EE2: Expressed Emotion (including covert criticism), MM: extended Mind-Mindedness. Correlations are calculated with N = 312 (sample size of article of Zitzmann et al.). * indicates p < .05. ** indicates p < .01. Higher scores in EE reflect higher criticism and/or emotional overinvolvement; higher scores in MM reflect higher mind-mindedness indices

Extended Mind-Mindedness, narrative coherence, and expressed emotion

Taking the aforementioned similarities and differences between the three coding systems together, one may tentatively organize them around aspects of to whom the speech sample is related to (i.e., if a statement is related to the mother herself, such as describing own feelings or thoughts, or if a statement is related to the child, such as describing the child’s activities) and to the level of analyzed content (i.e., if the information provided is analyzed as characteristics of speech or that of representation). Figure 1 depicts the main operationalizations and subscales of all three coding systems aligned within these overarching levels of orientation and analyzed content. Characteristics of speech are only considered in NC, and EE does not entail exclusively child-orientated aspects. The separation at the level of orientation (i.e., mother- vs. child-orientated) is most evident in extended MM, for example, where the same aspects, such as preferences and mental representations, are considered on both the mother’s and the child’s side. Three aspects are taken into account in all three coding systems (highlighted in bold): (1) reporting negatively about characteristics or behaviors of the child, (2) the report of child weaknesses and (3) descriptions of the relationship in terms of positive, neutral or negative relationship quality. When considering the three common themes across the coding systems, it is important to remember that these systems place great importance on valence (at least EE and extended MM valence indices). We suggest that significant group differences in these systems are based on the identified themes.

Fig. 1.

Differences and similarities between extended Mind-Mindedness, Expressed Emotion, and Narrative Coherence. In blue: characteristics of extended Mind-Mindedness, in red: characteristics of Expressed Emotion, in green: characteristics of Narrative Coherence, in black: characteristics of more than one coding system

The comparison of the coding systems used in the two analyses can be used for future research. Both EE and NC were found to depict a broader picture of the parent-child relationship and maternal mental representations than extended MM, as this construct relates exclusively to mental states. On the other hand, and when specifically interested in mentalization, extended MM allows for the advantage that, in addition to the information processing related to the child, information about the mother herself is also considered (e.g., the frequency of self-oriented vs. child-oriented mental statements can give an indication of how straightforward the respective reflection is for the mother). Furthermore, the extent to which the mother is aware of the opacity of mental states (MM_{not−knowing}) can be captured using extended MM. Therefore, extended MM approaches a more detailed examination of the representation of the child by focusing on mental attributes (in the sense of: “Can the mother imagine what is going on inside her child?” versus “Can the mother describe what the child does?” which is part of the EE and NC ratings).

The three conceptualizations were examined in the above-described studies [3, 7]. The BPD group showed a stronger expression of overt criticism (EE) and a greater use of negative valenced mental attributes (extended MM) than the AD/D and CON groups. However, the results for covert criticism (EE), narrative coherence (NC) and self-related mental state speech (extended MM) tended to indicate transdiagnostic associations, with a significant difference observed between the BPD and CON group, but not between the BPD and AD/D group. As we identified particularities in mental representations of the parent-child relationship in mothers with BPD across different coding systems in both articles – at least in comparison with the healthy control group –, it is arguable that these characteristics are due to one (or all) of these overlapping aspects. It should be noted that the three coding systems tend to reflect general spontaneous maternal tendencies when speaking about their children. Since these systems place a significant focus on the valence of statements, contextual information (at least in EE and extended MM valence indices) is included less strongly, such as explanations for reporting child weaknesses. Therefore, a higher tendency to display negative mental representations in the extended MM coding does not automatically imply low mentalization abilities. However, low attunement to mental state content does. We suggest that mothers with BPD are more likely to report disapproval of their children’s characteristics, perceived weaknesses in their children, and/or challenges in their relationships with their children. These negative mental representations may, but not necessarily, be an indicator of a more negative family climate or dysfunctional parental behavior.

Conclusion

Taken together, both articles provide deep insight into different aspects of parent-child relationships and mental representations in general and in mothers with BPD in particular, and serve to draw more robust conclusions. Considering the results of both articles, we discovered particularities in mental representations of the parent-child relationship in BPD at different aspects, specifically in the form of more incoherence (NC), more self-focus (extended MM), more criticism (EE), as well as more negative mental state descriptions of the child and the mother herself (extended MM) in mothers with BPD compared with mothers without a mental disorder. At the same time, the observation regarding conceptually overlapping aspects of the analyzed constructs becomes more robust, specifically, (1) difficulties in focusing on or representing the child in their speech in the form of more incoherence (NC) and a higher self-focus (extended MM), and (2) an increased negativity of mothers with BPD in the form of more criticism (EE) and negative representations (extended MM). Compared with the clinical control group, particularities for the relationship were also identified, namely more overt criticism (EE), and more negativity in mental state references (extended MM), but we did not see any significant group differences regarding coherence (NC), covert criticism (EE) or self-related mental state speech (extended MM). As we have robustly identified particularities for mothers with BPD across different coding systems, we argue that there may be a common underlying factor in mental representations of the parent-child relationship – at least, when using coding systems that focus heavily on the valence of speech. We propose that this common factor may reflect a tendency towards greater disapproval of child characteristics, a focus on perceived child weaknesses, and/or reports of challenges in relationship quality in BPD – thus suggesting overall more negative expectations and more negative judgments.

According to our results, there is an implication for supporting mothers with BPD in shifting their focus away from themselves and towards their child. Helping parents to gain more insight into their children’s experiences and reasons for their behavior may be achieved by training mentalization capacity or a coherent portrayal of their child. This may be beneficial in order to be able to perceive and classify their child’s behavior and mental states in their entirety, without being too attached to their own interpretations and feelings. In addition, taking the child’s perspective may reduce maladaptive attributions, and critical as well as negative references that may be based on one’s own internal stance without being sufficiently reflected on and differentiated from that of the child. As previous studies revealed an association between stress and difficulties in the mother-child relationship of mothers with BPD [19], affected parents could also benefit from learning skills to deal with stressful situations in order to be insightful to their children’s feelings, even under distress. Mothers with BPD should also be supported in recognizing their children’s strengths and taking notice of their children’s offers of a positive relationship. Establishing a positive relationship quality should be the goal of a parent-child-orientated intervention. If possible, children’s perspective on the relationship should also be included to check whether proposed intervention approaches actually resonate with them.

Abbreviations

BPD

Borderline Personality Disorder

EE

Expressed Emotion

FMSS

Five Minute Speech Sample

MM

Mind-Mindedness

NC

Narrative Coherence

Author contributions

Conceptualization, A.J. and J.Z.; methodology, A.J. and J.Z.; formal analysis, A.J.; investigation, A.J. and J.Z.; writing - original draft preparation, A.J. and J.Z.; writing - review and editing, A.J. and J.Z.; visualization, A.J. and J.Z. Both authors have read and agreed to the final version of the manuscript.

Funding

This study is based on data from a research project investigating a parent training for mothers with borderline personality disorder which was funded by the Federal Ministry of Education and Research (funding codes: 01KR1805A, 01KR1805C). Both authors were employed in this research project.

Data availability

Data sharing is not applicable to this article as no new datasets were generated or analyzed during the current study; we kindly refer to the data availability statements of the articles of Jung et al. [3] and Zitzmann et al. [7].

Declarations

Ethics approval and consent to participate

The procedure for the original study was reviewed by the Ethics Committee of the German Association of Psychology (Deutsche Gesellschaft für Psychologie, DGPs; protocol code RennebergBabette20lg-07-2$VAD, accepted 26th September 2019). Written informed consent was obtained from all subjects involved in the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.