The depression is associated with typical ways of relating to and handling conflict with others. This study empirically examined the most frequent relational demands of female depressive patients, depending on their personality configuration and the characteristics of relevant segments within the session: change and stuck episodes. Ten psychotherapies (n = 230 sessions) were observed in order to identify, delimit and code relevant episodes within session (24 change episodes and 26 stuck episodes) using the CCRT-LU-S Category System (Albani et al., 2002), to determine patients’ relational demands. The results showed that introjective patients verbalized more relational demands on themselves as a relational object, and therefore more demands with a subject-subject direction. These patients expressed their needs and wishes centered on questioning themselves, to the detriment of generating significant interpersonal relationships. Furthermore, it was possible to predict a greater presence of relational demands referred to contents about “loving” during the change episodes, confirming that the most frequent themes in patients’ discourse were a strong wish to feel happy, despite their experiences of helplessness. These findings highlight the need for a more dimensional view of depression that takes into account patients’ personality configurations and their association with little and big outcomes.

Despite differences in its frequency among countries, depression is a universal phenomenon characterized by the presence of depressed or irritable mood, accompanied by somatic and cognitive changes that significantly affect the capacity of individual to function ( Moncrieff, 2018 ; Ormel, Kessler, & Schoevers, 2019). Those symptoms are usually accompanied by anxiety, changes in sleep patterns, increase or decrease in appetite and weight, diminished ability to think or concentrate psychomotor agitation or retardation, feelings of guilt, and suicidality (DSM-5, American Psychiatric Association, 2013). Its worldwide prevalence and social impact produce high levels of disability, as well as individual and societal burden and expenditures in mental health ( Hu, 2006 ; Whiteford et al., 2013).

According to Blatt (2008), the quality of the depressive experience depends on the personality of the patient, and that it develops along the following fundamental developmental lines: relatedness and self-definitional. The first line involves the capacity of an individual to establish mature and mutually satisfying interpersonal relationships, while the second line involves the development of a consolidated, realistic, differentiated, and integrated self-identity ( Wachtel, 2019 ). In normal personality development, these two processes evolve in an interactive, reciprocally balanced, mutually facilitating fashion (Blatt, Shahar, & Zuroff, 2002). However, when there is an overemphasis of one of these developmental lines, a pathological personality configuration develops, as well as the defensive avoidance of the other (Blatt, 2004; Blatt & Shahar, 2004). The patients with an exaggerated and distorted emphasis on interpersonal relatedness developed an anaclitic configuration, which involve disruptions in gratifying interpersonal relationships and preoccupations with interpersonal issues of trust, caring, intimacy, and sexuality. The development of the self is neglected, and defined primarily in terms of the quality of interpersonal experiences; therefore, relatedness refers to feelings of loss, sadness, and loneliness in response to the disruption of relationships, which reflect concerns about the loss of a special person to whom one feels attached. These patients are very dependent and vulnerable to experiences of abandonment, using predominantly avoidant defenses such as denial, repression, and displacement in an effort to maintain interpersonal ties ( Auerbach, 2019 ; Blatt, 2007 ; Blatt, Shahar, & Zuroff, 2001).

By contrast, the patients with an exaggerated and distorted emphasis for establishing and maintaining a viable definition of the self at the expense of establishing meaningful interpersonal relations defines an introjective configuration. These patients distort the quality of interpersonal experiences, which makes them very vulnerable to feelings of failure, criticism, guilt, anger and aggression directed toward the self or others. They use counteractive defenses, including isolation, doing and undoing, rationalization, intellectualization, reaction formation, projection, and overcompensation, because they are more ideational and more concerned with establishing, protecting, and maintaining a viable self-concept ( Auerbach, 2019 ; Blatt, 2004; Blatt et al., 2001).

In addition, each group of patients has a particular way of perceiving their own psychotherapeutic process, precisely because they have different modes of cognition, defense, and adaptation, as well as different experiential modes, behavioral orientations, types of gratification, and most importantly, different relational patterns reflected in their speech (Blatt et al., 2001 ; Luyten, Campbell, & Fonagy, 2019). However, the specific interactional experiences are not only represented in a narrative way (Stapleton & Wilson, 2017); also, a repeated verbalization of such experiences represents the structure of relevant subject-object relationships, as a pattern that transcends the perspective of the individual narrative reconstructed subjectively (Espinosa & Valdés, 2012).

The core conflictual relationship theme method (CCRT) has been used as a reliable quantitative clinical system for identifying relational situations and central relationship patterns in patient narratives, with their respective internal conflicts (Luborsky & Crits-Christoph, 1990). This system uses some psychoanalytic concepts as a starting point because it emerged within the psychodynamic psychotherapy research approach. The central relationship patterns are considered as characteristic ways of relating with others that operate as organizers of experience and patterns which are partially unconscious. In that sense, the CCRT was initially used as a tool to operationalize this concept, starting from the basis that these patterns can be understood as mental representations or as schemes of interpersonal relationships (Barber, Foltz, & Weinryb, 1998), constituting the first kinds of interactions and subsequent relationships. Furthermore, this clinical system makes it possible to extract patients’ relationship repertoire based on the significant interpersonal experiences repeated in his/her discourse (Dahlbender, Albani, Pokorny, & Kaechele, 1994; Van Nieuwenhove, Meganck, Cornelis, & Desmet, 2016). In addition, it allows to identify a unique relational theme in the relational narratives of the patient, as a characteristic factor of personality, but also, to evaluate their presence and transformation throughout the therapeutic process (Espinosa & Valdés, 2012).

The CCRT considers the following three underlying central assumptions: (a) the information used to extract the relational patterns is extracted from the stories of the patient stories with relational contents; (b) the central relationship patterns is inferred from the repetition of each significant interpersonal experiences of the patient; and (c) the extracted patterns can be considered really significant and reliable (Luborsky, Popp, Luborsky, & Mark, 1994). The present study focuses mainly on the patients’ wishes and objects of their relationships. The wish component is referred to the desires, needs, or intentions of the person that could be directed towards others or him/herself (López del Hoyo, Ávila Espada, & Pokorny, 2011), distinguishing two levels of inference: a manifest one, which involves what the patient actually verbalizes, and another with some degree of inference, which, for Luborsky (1998 ), should be moderate to ensure reliability. Nowadays, the wish component can be labeled as explicit (directly expressed) or implicit (deduced from responses or actions expressed in the narrative) (Albani et al., 2002 ). On the other hand, the relational object (the protagonist of the interaction) may be another person/people (parents, partners, friends, coworkers, classmates, friends, etc.), the therapist (only when the therapist can be clearly identified as the other in the interaction), or even narratives about the interactions of the patient with him/herself. There are memories that involve feelings or thoughts about confrontations with him/herself (self-descriptions are not included) (López del Hoyo et al., 2011).

Although the entire therapeutic conversation is about stories narrated by the patients, this study will focus on the explicit narratives of the patient about relationships (Luborsky & Crits-Christoph, 1990). For this purpose, the wish component will be understood as relational demands (hereinafter referred to as RD), since our attention was focused on that which patients request for in the relationships they establish and narrate during the session, either with others (including the therapist) or with themselves.

There are different approaches for studying the relevant moments during psychotherapeutic sessions, which relate to the conceptualization of change, but also, to some problematic moments that hinder the psychotherapeutic change of the patient. Some of the ways used to conceptualize these positive relevant moments are the following: innovative moments, change moments, insight, empowerment events, among others ( Elliott, 2010 ; Gonçalves et al., 2017; Gonçalves, Ribeiro, Silva, Mendes, & Sousa, 2016; Krause & Altimir, 2016; Timulak & Elliott, 2003). Krause et al. (2007 ) propose that change moments are possible to observe from the transformation of the subjective perspectives of the patients about themselves, their symptoms, their problems, and its relationship with the entire context. However, they can also be observed throughout the process, as it occurs in successive phases, starting before therapy and continuing after the end of it, and having important characteristics as result of the combination of intra and extra-therapeutic factors. Each successive phase includes explanations and interpretations patterns with progressively increasing levels of complexity. These authors developed a list of generic change indicators based on the analysis of therapies with different psychotherapeutic modalities and approaches, and hierarchically ordered according to each phase of the process (Krause et al., 2007 ; see Appendix A).

On the other hand, there are relevant segments during the session that may be associated with problematic moments conceptualized as difficult moments, which could include ruptures (Eubanks-Carter, Muran, & Safran, 2015), refusals ( Billow, 2007 ), reactances (Brehm & Brehm, 1981), resistances ( Arkowitz, 2002 ), impasses (Etchegoyen, 1986), and stuck episodes (Fernández et al., 2012 ; Herrera Salinas et al., 2009). The latter type has been considered as moments that halt the change of the patient, which are related to negative results within the process and/or the decrease of advances expected by particular observers. A stuck episode can be clinically expressed in different ways: displeasing dysphoric states (anxiety, lack of motivation, boredom, etc.), hindering the therapeutic process or interventions (avoidance of therapeutic work, thinking and associating, as well as non-compliance with agreements and tasks), the absence of change moments when this would be expected, and finally, relational problems (communication difficulties, understanding or negotiations) (Fernandez et al., 2012; see Appendix B).

The general aim of the study was to identify some similarities and differences of the RD most frequently verbalized by patients within relevant episodes during the session (change and stuck episodes), depending on their personality style (anaclitic or introjective) and symptomatology. For this purpose, the following six hypotheses were put forward: (1) patients’ symptoms do not predict the characteristics of RD; (2) introjective configuration predicts more patients’ RD directed to themselves, whereas anaclitic configuration predicts more RD aimed at others; (3) change episodes predict more patients’ RD directed to themselves, while stuck episodes predict more RD aimed at partners (or former partners); (4) introjective configuration predicts more patients’ RD with a subject-subject direction, whereas anaclitic configuration predicts more RD with an object-subject direction; (5) introjective configuration predicts more patients’ RD with contents about ‘being strong’ and ‘withdrawing into oneself’, whereas anaclitic configuration predicts more RD with contents referencing ‘loving’; and finally, (6) the content of RD verbalized by patients during therapeutic conversation can be predicted from the episode type.

Ten therapies conducted in Chilean private therapeutic centers were analyzed using a mixed methodology (qualitative and quantitative) (see table 1 ). All the therapies are part of the Therapeutic Processes Database provided by the Chilean Millennium Institute for Depression and Personality 1 , which has generated audiovisual recordings over the last 15 years with the purpose of conducting process analysis. The therapies were purposively selected according to the following criteria: (a) therapies with a weekly individual modality; (b) therapists with 10 to 30 years of professional experience; (c) therapies with a significant evolution of change throughout the process; and (d) participants (patients and therapists) who gave their informed consent to participate in the study. However, it is important to make it clear that the sample unit was each RD identified within relevant episodes, and not the therapies analyzed.

The concurrent presence of a depressive disorder with prominent anxiety symptoms or an anxiety disorder is common in clinical practice. Studies have shown that more than 70% of people with depressive disorders also have anxiety symptoms (Dochnal et al., 2019 ; Schafer, Naumann, Holmes, Tuschen-Caffier, & Samson, 2017). Models have been proposed that explain the high correlation between the two: the first model states that they form part of a continuum because both disorders are related in mood; the second model states that the differences between the two disorders are qualitative, since orthogonal factors indicating anxiety and depression have been observed, which only show significant differences if appropriate statistical tests are used; and the third model suggests the coexistence of both syndromes in a group of so-called atypical patients, who tend to show a more chronic course of the disease, compared to depressive patients without anxiety, who tend to show a more chronic course of the disease, compared to depressive patients without anxiety (Adams, Wrath, Mondal, & Asmundson, 2018). In that sense, even when all the patients had depressive symptoms, there were six patients with a predominance of depressive symptoms and four patients with a predominance of anxiety symptoms, aged between 26 and 64 (M = 37, SD = 10.93).

Classification of patients according to their depressive personality styles. For this purpose, an observation guideline was developed (Salvo, Cordes, & Valdés, 2012) to differentiate the predominance of one of the following depressive personality styles: anaclitic, introjective, and mixed. The mixed style was used when there was not a predominant style, but a predominance of features of both styles at the same time. Only the cases with a predominant style were considered as a requirement. Blatt, D’Affitti and Quinlan (1976) proposed these styles because of psychoanalytic theoretical formulations and clinical observation of depressive patients. Some items of the Depressive Experiences Questionnaire (DEQ) were considered at the time of developing the observation guideline, so as to have as much information as possible to identify the patient’s predominant personality style.

As shown in table 2 , concepts of dependency and self-criticism are closely related to these styles: for example, the symptomatology of depressed patients reveals few differences among them, but these depressive styles are much more effective in highlighting variation. A depressed patient with an anaclitic personality style is characterized by deep feelings of loss and loneliness, while a depressed patient with an introjective personality style is characterized by intense feelings of worthlessness (Blatt, 2008; Huprich, Auerbach, Porcerelli, & Bupp, 2016).

Three observers with at least five years of clinical experience conducted a diagnostic examination based on the first two videotaped sessions of all therapies. An inter-rater reliability study was carried out considering the following three successive stages. (a) Two observers individually coded each item of the observation guideline. (b) They discussed their coding in order to reconcile their differences and to make a final decision about the presence or absence of depressive symptoms in each patient. If necessary, they additionally watched a part of the videos or read the transcriptions again to reach a consensus based on the data. And (c) this last coding was compared again with the assessment of a third observer who rated the therapy sessions following the same principles and procedure mentioned above. There was a high degree of agreement between the observers when differentiating the patients’ personality configurations (κ = 0.615, p < 0.05). The total sample was distributed as follows: six anaclitic and four introjective patients.

Demarcation of change and stuck episodes. Expert raters trained in the use of a protocol for detecting and identifying relevant moments during therapeutic sessions (Krause et al., 2007 ) observed the ten-videotaped therapies. All the sessions were listed in chronological order and transcribed, to facilitate the subsequent delimitation of the change episodes and stuck episodes. As shown in figure 1 , the moment of change marks the end of the change episodes. Said moment of change must meet the criteria of theoretical correspondence, novelty, topicality, and consistency; that is, they must match one of the indicators from the Hierarchical List of Change Indicators (GCI, Krause et al., 2007), be new, occur during the session, and persist over time. Afterwards, using a thematic criterion, the beginning of the therapeutic interaction associated with the change moment is tracked in order to define the start of the change episode.

In the case of stuck episodes, it was necessary to identify those periods of the session in which there was a temporary halting of the change process of the patient due to a reissue of the problem, that is, episodes of the session characterized by a lack of progressive construction of new meanings (Herrera Salinas et al., 2009 ). A stuck episode must also match one of the topics from the list of stuck themes, occur during the session, and be nonverbally consistent with the topic of that kind of episode. In addition, a stuck episode must comply with the following methodological criterion: be at least three minutes long and be at least 10 minutes apart from a change episode in the same session. It is important to note here that the identification of stagnation is placed on the patient, regardless of the actions or omissions performed by the therapist that may or may not contribute to such stagnation.

All the sessions of each therapy (n = 230) were transcribed in order to delimit and analyze all the change and stuck episodes identified (see table 3 ). There were identified 50 episodes. Each episode was made up of patient and therapist speaking turns, which began with the verbalization of one participant and ended when another began (Krause, Valdés, & Tomicic, 2009). Therefore, the total sample comprised 1,282 patients’ speaking turns, of which 529 were included in change episodes and 753 were in stuck episodes.

Classification of relational demands of the patients. In the present study, the CCRT-LU-S (López del Hoyo, Ávila Espada, Pokorny, & Albani, 2004) category system was used, the Spanish version of the system developed in Leipzig and Ulm, Germany (Albani et al., 2002 ). Upon the basis of the first studies conducted, Luborsky et al. (1994 ) determined that relational patterns, apart from referring to a relational object, comprised what the patient expected from others or him/herself (labeled wishes [W]), how others responded (object responses [OR]), and how the patient acted (subject responses [SR]).

The CCRT-LU-S category system is not only helpful for studying transference patterns but is more broadly applicable to the field of affective disorders, as it can contribute to our understanding of the interpersonal aspect of these disorders (López del Hoyo et al., 2004 ). Therefore, it provides an operational, clinically valid, and reliable measure of the wishes and needs that are predominant in patients’ interactions. Using the wish component as a starting point, RD were defined as narrations of desires, needs, intentions, aspirations, and/or longings —but only those of the patient—, either in connection with others or with themselves. In addition, only patients’ explicit RD were considered, that is, those verbally expressed by them during the session (Albani et al., 2002 ). The following four dimensions were coded in each RD:

Object. It refers to the protagonist of the RD, who may be other person/people or the patient him/herself.

Direction. Assigned depending on whether the RD is directed towards the object (another person) or the subject (the patient). Three possible directions were established: object-subject, in which the patient desires something from someone else (e.g., “I'd like you to understand me”); subject-object, in which the patient desires something for someone else (e.g., “I'd like to support her”); and subject-subject, in which the patient desires something for him/herself (e.g., “I'd like to feel more confident”).

Temporality. It refers to the time when the RD occurs. A past RD may have occurred at a stage or moment (beyond two weeks) prior to the one currently being experienced. A present RD may be happening in the stage or the present being experienced by the patient (with the ‘present’ including even the last two weeks), while a future RD can be manifested as something that the patient wants to happen at a later stage.

Contents. A tailor-made qualitative label was assigned to each RD, using the particular language of each patient in order to facilitate the subsequent coding using the standard categories of the CCRT-LU-S. For each tailor-made label, one of the standard categories proposed was chosen to make it possible to compare the contents of the patients’ RD. For further guidance on the use of the CCRT-LU-S system, refer to Albani et al. 2002 ) and López del Hoyo et al. (2004 ).

The identification and coding process of the RD was conducted following the Relational Episode Coding Manual, developed by Espinosa and Valdés (2012). A group of six raters was trained to carry out this procedure. During this phase, each researcher individually coded a set of episodes (different from those included in this study), as a form of training and to calculate the degree of agreement between them. The judges had a degree of agreement ranging from significant to almost perfect for coding the relational object, directionality, and temporality (κ = 0.885, p < 0.001). In the case of harmony the degree of agreement was almost perfect (κ = 0.942, p < 0.001), however, it ranged from moderate to significant for coding the content (cluster, super cluster, category, and sub-category), precisely because there were more options to choose from (κ = 0.525 – 0.827; p < 0.001). Afterwards, the group was divided into three pairs of judges for analyzing and coding the episodes considered in this study. Each judge coded individually; and, subsequently, both worked together to reach an agreement about those codes in which they differed.

First, a descriptive analysis of the dimensions present in the RD was conducted (frequencies and percentages); then, the analysis focused on the relationship between said dimensions and the other variables considered (personality configuration, symptomatology, and episode type). Using Hierarchical Linear Modeling (HLM) (Woltman, Feldstain, MacKay, & Rocchi, 2012), this study used a three-level hierarchy: the highest level (Level-3) contains the patient-related variables (symptomatology and personality style); at the middle level (Level-2) is the episode type variable (change and stuck episodes); and at the lowest level (Level-1) are the dimensions of RD (e.g., object, direction, temporality and contents) (see figure 2 ). In addition, the Z-ratio was used to calculate and compare independent proportions (95% confidence intervals when the value of Z could not be estimated).

The results of this study are organized as follows: (a) a description of the characteristics of the RD identified according to the dimensions within them; and (b) the relationship between these dimensions and the patients’ symptoms and personality, as well as their link with episode types.

Within the 50 episodes analyzed (24 change and 26 stuck episodes), 71 RD were identified and coded. The following is a description of the results by dimension and category. Out of the total number of speaking turns in which a RD was present, 32.39% involved oneself as relational object, whereas 28.17% referred to a partner or former partner. The difference observed between these proportions was non-significant. The rest of the RD were related with objects such as the therapist, sons or daughters, and others (relatives or other people). It is noteworthy that the smallest percentage of RD related to the mother or father, while the other objects (friends, coworkers, acquaintances) were not mentioned by any of the patients.

Out of the total number of RD identified, 40.85% were object-subject RD (the patient desires something from someone else), 33.80% were subject-subject RD (the patient wants something for herself), and 25.35% were subject-object RD (the patient desires something for someone else). Among the RD identified, the largest percentage used the present tense (84.51%), followed by those that employed the past (12.68%) and the future (2.81%). Out of the total number of RD identified, 81.69% were harmonious and were 18.31% disharmonious. Among the RD regarded as harmonious (n = 58), 74.41% had contents related with ‘loving’, and mostly referred to: feeling fine, supporting, helping, and protecting. The remaining 27.59% referred to ‘being strong’ (or having self-determination), which mostly concerns being proud and autonomous. Regarding the RD labeled as disharmonious (n = 13), nearly all (84.62%) the contents studied referenced a wish to ‘withdraw into oneself’, which mostly referred to: leaving, distancing oneself, creating a boundary; keeping one’s distance, retreating; being submissive, being compliant, and avoiding conflict.

To calculate these results, low-frequency categories present in Level 1 variables were excluded because the data analysis system (HLM) did not make it possible to use them. Therefore, in general, we considered the two or three categories that displayed the highest frequencies in each variable of the RD throughout all the episodes.

The hypothesis 1 was verified: patients’ symptomatology (depressive or anxious) does not predict any of the dimensions and categories considered in the RD. This means that: (a) patients directed their RD both to themselves and to their partners (or ex partners); (b) the direction of RD were predominantly subject-subject and object-subject; (c) RD were uttered using the past, present, and future tenses; (d) RD were labeled as harmonious and disharmonious; and (e) their contents referenced loving, being strong, and desiring to withdraw into themselves.

In the case of the relational objects, the following two were considered because their frequency was the highest: the patient herself and her partner (or former partner). The hypothesis 2 was partially verified: the probability of verbalizing RD directed at herself —which show the interaction of the patient with herself— was higher when the personality style of the patient was introjective (OR = 0.09, p = .01) (see figure 3 ), however, the verbalization of RD aimed at someone else as relational object, specifically the patient's partner or former partner, was not predicted by the anaclitic configuration; that is, both patients directed their RD to their couple (or ex couple).

The hypothesis 3 was not verified: the verbalization of RD, in which the relational object is oneself or the partner, was not predicted by the episode type. That is, in both change and stuck episodes, the patients verbalized RD directed towards themselves or their couples (or ex couples) as relational objects. Nevertheless, it was possible to demonstrate a statistically significant difference between the proportions of RD, which referred to the patients’ partners depending on the episode type (see figure 4 ). Thus, patients verbalized more relational demand referencing their partners during stuck episodes [CI 95% = 0.0028, 0.3965], p < .05, than during change episodes (SE = 34.6%, CE = 10.5%).

In the case of the directionality, only the following two most frequent alternatives were considered: object-subject and subject-subject RD. The hypothesis 4 was partially verified: the probability of verbalizing RD during therapeutic conversation with a subject-subject direction —in which the patient “desires something for themselves”— was higher when they presented an introjective personality configuration (OR = 0.12, p = .012) (see figure 5 ). However, the verbalization of RD concerning an object-subject direction was not predicted by the anaclitic configuration; that is, both patients desired “something from someone else”.

Regarding the contents of RD, the analysis considered only the following three most frequent ones: ‘loving’ and ‘being strong’, among harmonious contents, and ‘withdrawing into oneself’ among the disharmonious ones. The hypothesis 5 was not verified: both introjective and anaclitic patients verbalized RD with the three type of contents. Concerning episode type, the analysis showed that it only predicted RD with contents referencing ‘loving’. Therefore, the hypothesis 6 was partially verified: the probability of verbalizing this type of RD was higher during change episodes (OR = 5.34, p = .027) (see figure 6 ), however, the verbalization of RD which reference ‘being strong’ and wishing to ‘withdraw into oneself’ was equally likely in both episode types.

The CCRT method measures the capacity of the patient to establish relationships, a capacity that varies with the level of differentiation of the object relations of the patient. The three CCRT components occur in association with each other, thus creating a linked ‘sequence’ in which the wishes and responses of the subject comprise a call-and-answer pattern, a kind of dialogue ( Mitchell, 1995 ). It is true that not every patient links his or her relational components together and places every component in a sequence ( Luborsky, 1998 ; Luborsky et al., 2004). For this reason, we decided to analyze only the RD, which had wishes at the start of the sequence, since in these cases the reactions of other people, and even those of the patient herself, are the result of a desire, need, or intention of hers. This is especially relevant when we consider that patients verbalize many kinds of statements that depict not interaction, but isolated and unconnected behavior by the patient and others. During the therapeutic conversation, many statements that could be considered responses are presented by the patient not as responses belonging to a RD or to a sequence of interactions, but as actions that the patient fails to place in any interactional context.

Concerning the predominant object in the patients’ RD, regardless of personality style and episode type, the most frequently observed one was ‘oneself’, to the detriment of other objects (mother, father, couple, ex-couple, etc.). This result is consistent with research suggesting that self-focused attention also produces negative consequences linked to negative emotion and psychological disorders (Höping & De Jong-Meyer, 2003; Mor & Winquist, 2002; Pyszczynski, Greenberg, Hamilton, & Nix, 1991; Fernández et al., 2012; Pascual-Leone & Kramer, 2016; Timulak, 2015; Valdés, Krause, & Alamo, 2010; Valdés et al., 2012). Language is a fundamental means of communication between patient and therapist, and the achievement of change necessarily requires the coordination of language usage in psychotherapy (Reyes et al., 2008 ). Thus, the establishment of self-focused attention can contribute substantially to change processes by promoting self-regulatory strategies, given that when people express something, they are also shedding light on how they process, organize, or interpret information, and on how they experience their capacity for interaction in relationships, as a way to connect the own demands and actions with those of others (Hegarty, Marceau, Gusset, & Grenyer, 2019).

The predominance of ‘loving’ contents in the RD coded as harmonious (e.g., loving and feeling well) and ‘retreating into oneself’ contents in the RD labeled as disharmonious (e.g., isolating oneself, compromising, or being submissive to avoid conflicts) was a finding that matched the observations of Albani et al. (2002 ), who noted that the most frequent wishes were ‘to be loved and understood’, followed by ‘to be distant’. This also resembles the observations of Vanheule, Desmet, Rosseel, and Meganck (2006), who found that the most frequent themes in patient discourse were a strong wish to feel happy and experiences of helplessness. However, there are also consistent with these studies about the transferential and counter-transferential themes using the CCRT method (Tishby & Vered, 2011; Tishby & Wiseman, 2014).

This study was not able to predict the verbalization of specific RD types based on the patients’ depressive symptomatology; even though it is known that depression can be regarded as a typical way of relating to and handling conflict with others (Vanheule et al., 2006 ). This result can be explained using a fully clinical sample, in which all patients displayed some type of symptomatology, and which could not be compared with a non-clinical population due to design characteristics. If that had been possible, significant differences would have been expectable. In this regard, both theoretically and clinically, it was more predictable for the differences between patient RD to have been determined by personality configuration, especially bearing in mind that some studies show an association between certain relational components and personality organizations (De Roten, Drapeau, Stigler, & Despland, 2004; Descôteaux et al., 2001; Wilczek, Weinryb, Barber, Gustavson, & Åsberg, 2000, 2004). On the one hand, it was only possible to predict RD with the presence of the “oneself” as relational object (which reflects the interaction of the patient with herself), and the subject-subject direction (in which the patient desires something for herself), from introjective patients. This result was expectable considering the notions advanced by Blatt (2008), who states that the main concern of such patients tends to be the establishment and maintenance of self-definition, to the detriment of generating significant interpersonal relationships (Blatt & Shahar, 2004). It must be noted that introjective patients, due to their focus on cognition and their use of rationalization and intellectualization as defense mechanisms, tend to be more willing to verbalize their needs and wishes during the therapy, attending it with motivations centered on questioning themselves and working through the issues that trouble them. On the other hand, in the case of anaclitic patients, in contrast, it was not possible to demonstrate that their configuration predicts a higher frequency of RD aimed at a “romantic partner” as relational object nor an object-subject direction, in which they desire or demand something from another person(s). Anaclitic patients, who are more focused on affects and on establishing interpersonal relationships, tend to end up manifesting their needs and desires through predominantly non-verbal mechanisms, acting rather than narrating their relational demands during the therapeutic sessions (Valdés, Arriagada, & Alamo, 2016).

In connection with this aspect, it must be noted that the present study focused mainly on identifying and coding the RD explicitly verbalized by the patients during the therapeutic conversation, leaving out those more implicit in nature. Therefore, future studies could consider including the analysis of implicit RD and/or those that can be inferred through non-verbal patient behavior, especially in the case of anaclitic patients, who need their relational bond with the therapist to be sufficiently solid before even beginning to work through their conflicts. Including this type of information in future studies can make it possible to identify patient RD associated with the therapeutic bond, in which the therapist emerges as the main relational object. In fact, there are studies that have analyzed the relationship themes of the therapists and their impact on the therapeutic outcome (Hamilton & Kivlighan, 2009; Sommerfeld, Orbach, Zim, & Mikulincer, 2008).

Regarding episode types, even though it was not possible to prove their predictive power with respect to relational objects, it was established that, in proportional terms, patients verbalize more RD referencing their ‘romantic partner’ during stuck episodes than during change episodes. In this respect, it would seem that during periods of the session in which the change process of the patient temporally stops due to a re-emergence of the conflict, their romantic partner is the relational object that most frequently appears in their RD. Verbally, this translated into an argumentative persistence in the discourse of the patient, characterized by contents referencing their current or former partners, which did not ultimately contribute anything to the focus of the change. Previous studies conducted using the SCAT-1.0 classification system (Valdés et al., 2012 ), indicate that, during stuck episodes, patients tend to focus on third parties external to the session (or on their relationship with them), while at the same time focusing less on themselves (Fernández et al., 2012 ; Valdés et al., 2012). This result has been interpreted as a deviation from the focus of the therapeutic work, that is, the halting of the progressive process of construction of new meanings related with the self. For example, patients sometimes spent long periods talking about how their romantic partners affected their well-being, while the therapeutic work remained unable to concentrate on how they themselves participated in maintaining these dynamics.

Although the predictive power of personality configurations could not be proved with respect to the contents of RD, it was possible considering the episode type: contents referring ‘loving’ were more likely during change episodes. Apparently, patients tended to verbalize RD chiefly referred to feeling fine, supporting, helping, and protecting when they were experiencing a transformation in their subjective meanings associated with their own conflicts.

Concerning its main limitations, it should be noted that, although the present study analyzed 2,422 patient and therapist speaking turns within 50 relevant episodes, the number of speaking turns was reduced to almost a half (n = 1,282) when only patients’ RD were considered. If we additionally consider that only explicit RD were coded, the amount of potentially analyzable data was considerably reduced (n = 71). Of course, this had a direct effect on the statistical analyses defined carried out. Therefore, it would be advisable for future studies not only to consider increasing the sample size by adding more therapies in order to demonstrate the stability and reliability of the results reported here, but also to be able to transfer these findings to other people, contexts, situations or environments (Levitt, Motulsky, Wertz, Morrow, & Ponterotto, 2017; Smith, 2018 ). It is also important to code implicit and/or ‘acted’ (non-verbal) RD, which could make it possible to describe the defensive functioning of patients during stuck episodes, for instance. One of the advantages of using this method lies precisely in the possibility of systematically interpreting the patient’s own words during the conversation in the therapeutic context itself, based on the assumption that, during the session, most patients are able to remember and narrate stories about their relationships with meaningful others which could reveal problematic relational themes (Parker & Grenyer, 2007).

Finally, these results suggest certain clinical implications. Although it is true that each patient's internal representations are unique due to being subjective, identifying the predominant ones in each personality style can be a useful clinical strategy for therapists, but not only to characterize the personality styles of their patients in terms of their overriding RD: in addition, they can be used as verbal markers indicating change, stuck or rupture moments experienced during the session (Eubanks, Burckell, & Goldfried, 2018; Valdés et al., 2016). The greater proportion of RD associated with romantic partners as relational objects during stuck episodes may point to the presence of internal representations connected with this meaningful other, which are problematic, and which persistently recur throughout the therapy. For their part, the higher proportion of RD associated with harmonious (loving) needs during change episodes may be regarded as a verbal marker in patient discourse indicating not only how they perceive their interpersonal relationships, but also the extent to which the flexibility of RD increases throughout the therapeutic process as a result of the construction of new meanings. These findings can be used as a guide for therapist training and therapeutic supervision.