From the Multiplicity to the Unfolding of the Self Throughout Time: A Call for New Methodological Approaches

Self-multiplicity has been widely recognized within psychology for quite some time now, regardless of the theoretical orientation. This notion can be traced back at least to William James (2007), and is supported by social-cognitive (e.g. Leary and Allen 2011), psychodynamic (e.g. Bromberg 2004), experiential (e.g. Elliott and Greenberg 1997), social constructionist (e.g. Gergen 1991), narrative (Ribeiro and Gonçalves 2011) and dialogical approaches (Dimaggio et al. 2010; Hermans and Kempen 1993). This multiplicity, nevertheless, needs careful consideration, since the sense of a unitarian ego is a major human feature (Salgado and Hermans 2005; Valsiner 2007).

In the history of psychology, it is possible to distinguish at least two major strategies for dealing with this problem of internal multiplicity. The more classical one is based on assuming an overarching and irreducible subject or ego that encompasses all the apparent fragmentations of psychic life. This has been coined by Dennett (1991) as the “homunculus view”. Therefore, for these classical theories, multiplicity is but an illusion of the mind. Needless to say, these theories were doomed by successive epistemic turns in social and human sciences: the linguistic turn, the interpretative turn, the narrative turn, the cultural turn, even the embodied turn—all kinds of turns cast doubt on the notion of one homuncular ego. As a result of these successive epistemic changes, psychological theories started to concur that the self is multiple: it revolves around multiple images or discursive practices. In this widespread emphasis on the multiplicity of the self, however, the multiple representations of oneself are highlighted and the subjective experience of being an I remains unaccounted for. This occurs in apparently opposing models, such as socio-cognitive (Markus and Nurius 1987) and social-constructionist (Gergen 1991). Thus, the problem of knowing the operating ego was somehow circumvented, but not really solved. As a result, the problem remained of how to account for self-continuity amidst self-multiplicity.

The dialogical approach has been trying to address the problem (Clegg and Salgado 2011; Salgado and Clegg 2011; Salgado and Hermans 2005). Within a dialogical approach, the self is conceived of as a continuously emerging experience of “being a centre” (of being an ego) of the here-and-now experience, and this self-centeredness endures from moment to moment. This creates the feeling of having an ego, the sense of being an “I”. At the same time, this ego (or, simply, I) continuously connects with the material and social world, resulting in a permanent change of the experiential contents. Moreover, this experience is also seen as socially and culturally guided by the use of shared semiotic means (Valsiner 2007). Thus, a dialogical approach that is no longer entrenched in the traditional dualisms enables us to approach the question of self-identity and self-multiplicity. The self is subjectively united but constantly multiple.

This almost consensual recognition of the multiplicity of the self does not mean, however, that major theoretical and methodological challenges no longer need to be addressed. Two inter-related major questions can be highlighted in this context: how are self-identity and stability achieved through time (Valsiner 2002), and how do we explain the emergence of novelty (Valsiner 2007)? The mere statement that the self is multiple and changes through time does not, in itself, shed the necessary light on how the self is changing from moment to moment and how stability is achieved in the midst of those changes.

The potential of this approach calls for new and different methods of studying the self. Specifically, it demands a strategy that enables researchers to track the “movement” of the self on a moment-by-moment basis coherently with the theoretical foundations of a dialogical approach. Methodologically, it represents a serious challenge to our usual methods, since the unfolding of the self through time becomes a central issue, and our usual analytical tools tend to neglect this constant change and the dynamic equilibrium achieved through time. For example, most of the assessment methods of self-multiplicity are static and remote from the actual phenomena they are trying to address (e.g. self-report questionnaires, such as the Self-Description Questionnaire by Marsh 1992).

Positioning microanalysis (PM) is a method which, departing from a dialogical theory about the self, follows a developmental-genetic orientation by studying dialogical processes of the self on a moment-by-moment basis. This method has been evolving in different phases (Cunha 2007a, b; Cunha et al. 2012; Salgado and Cunha 2012), and this work aims to describe its most recent version. The method is designed to study the dialogical processes of the self within the developmental or genetic approach. We start by briefly exposing the dialogical approach to the self and from there we establish the necessary requirements of a method coherent with such theory. Then, we review and compare the main methods available in order to justify the need for different methodological proposals. Finally, we present positioning microanalysis and its main procedures.

The Dialogical Notion of Position and its Relevance for the Study of the Self

A dialogical stance configures an approach to human psychological phenomena in an attempt to integrate three levels simultaneously in a coherent way: the experiential, the socio-relational and the semiotic-linguistic elements of the human mind (see Bento et al. 2012).

The experiential dimension is related to the embodied nature of human beings. At every moment, the person has a self-centered and embodied sense of the current situation, i.e. humans live through a phenomenological psychic field. Within this experience, there is an I emerging at each moment which is associated with the surrounding material and social world (i.e. social and material objects).

This constant interchange with the world is embodied and material, but it is also social and semiotic. Social articulation based on previous historical arrangements and conventions conveys and specifies the possible meanings of contextualized human actions,. Therefore, at the core of a dialogical approach lies the notion of social addressivity. If one hears someone saying “Why don’t we go out?” it immediately invokes a social transaction; one tends to imagine someone inviting a friend to some social activity, and through this symbolic performance one configures the speaker as someone assuming the position of an “inviter” while also co-positioning the other person as an “invitee”. Meaning and social action always go together. Therefore, human actions and meanings imply the assumption of a certain personal perspective about some “objects”, but the meaning and relevance of this perspective is dependable on this social articulation. In other words, at every moment, every person occupies a contextualized “position” regarding the world.

This relation, in turn, is mediated by semiotic (necessarily socialized) tools that organize the ongoing relation with the world, signs that regulate and mediate the possible meanings and possible actions emerging in the current situation (Leiman 2002; Valsiner 2007; Vygotsky 1986). Thus, every action in the world is also a personal experience, with a social and semiotic texture that regulates its meaning and possible development. Following this approach, we suggest that the person at each and every moment enacts or performs a self-position or, in Leiman’s (2002) terminology, a semiotic position. Self-positions are here conceived as “self-events” (Holquist 1990) connected to the present moment of the lived experience (Stern 2004).

Departing from this dialogical perspective, the self can be conceived as “…a dynamic multiplicity of relatively autonomous I positions […] The I has the possibility to move, as in a space, from one position to the other in accordance with changes in situation and time” (Hermans et al. 1992, p.28). This notion of “position” aims to surpass a self-enclosed, solipsistic notion of the mind by making use of the Bakhtinian “law of placement” (Holquist 1990): everything that a human agent says or does is placed within a certain social and historical context (a specific “chronotope”; Raggatt 2010). I-position implies a spatial metaphor then, since every human act is performed according to the social background and surroundings of the person, while it is also directed at someone or something else. At the same time, self-positions are always on the move from one moment to the next. Thus, the self develops as a chain of positions that succeed each other in time. As Hermans et al. (1992, p. 28) observe: “The I fluctuates among different and even opposed positions”.

Methodologically this implies the crafting of tools that enable the study of the psychological holistic/integrated and experiential field that we call “the self” in its dialogical dynamics and development through time.

Dialogical Methods for the Study of the Self

Several methods have been developed to study the self from a dialogical perspective. Most of them are connected with the dialogical self theory (DST; Hermans and Hermans-Konopka 2010), probably the most well-known version of a dialogical approach to the self. Within this theory, two main methods are proposed: the self-confrontation method (Hermans 1999; Hermans and Kempen 1993) and the personal position repertoire (Hermans 2001b). We will briefly outline the main features of these methodological tools and their variations in order to clarify why they do not fully address the dialogical dynamics of the self (for more detailed reviews see Cunha et al. 2012, and Jasper et al. 2011).

Self-Confrontation Method

The self-confrontation method (SCM) is based on an interview in which the person is invited to review his or her life story. Departing from a phenomenological and narrative perspective (Hermans 1987; Hermans and Hermans-Jansen 1995) it emphasizes storytelling of self-narratives as the core element of meaning-making. From this standpoint, by exploring the past, the present and the future, the interviewer invites the person to discriminate different important life stories. The personal meaning of each story is synthesized in a “valuation” (usually a sentence), and afterwards the person assesses the affective profile of each valuation on a scale of affects. In the end, the person obtains a system of valuations designed to synthesize the past life of the individual (Hermans 1987). SCM was subsequently integrated in the dialogical self-theory (Hermans and Kempen 1993), and it was used as a way to explore different I-positions: each position was conceived of as a distinct system of valuations, and consequently each position could be invited to comment and respond to every valuation established by another I-position (Hermans 1999).

Personal Position Repertoire

The personal position repertoire (PPR) was developed by Hermans (2001a, b) as a specific method to explore the multiplicity of self-positions and their inter-relation. One major distinction introduced in this method, which was missing in the SCM, is that between internal I-positions and external I-positions. An internal I-position is considered an internal part of the self (e.g. I as teacher), whereas external positions refer to people and objects of the environment (e.g. my students) (Hermans 2001a). Internal and external I-positions are mutually dependent and trigger each other: for example, an internal I-position such as “I as teacher” is co-relative to the external I-position “My students” and these I-positions establish mutual transactions over time. PPR is devoted to exploring these myriad internal and external I-positions and the kind of dialogical patterns established between them. In terms of procedures, PPR is based on two standard lists, one referring to internal I-positions and the other to external I-positions, previously elaborated (see Hermans 2001b, for access to the standard lists). The person chooses the ones that seem more relevant, but is also given the opportunity to add other relevant positions. Then, the person assesses with a Likert scale the extent to which each internal position is prominent (i.e. invoked or triggered) in relation to the external positions. This step is repeated for every internal I-position, and results in a matrix relating internal and external positions in terms of their prominence.

Variations in the Methods Based on the Dialogical Self Theory

Different research teams have introduced some methodological variations to the methods employed by the DST. Kluger et al. (2008) applied quantitative methods in order to explore PPR results. They showed that the use of bi-plot techniques, such as principal component analysis (PCA), meant the factors underlying positions obtained in a PPR of a single person could be extracted. Therefore, by performing two PCAs (one for the internal positions and another for the external positions), and restricting the analysis to two factors, they found it was possible to create a bi-plot which graphically described the relations between internal and external positions. This analysis is intended to offer a bird’s eye view of the dialogical self, enabling an easy combination of the different internal and external positions into different clusters based on their commonalities.

Raggatt (2000) also makes use of qualitative and quantitative methods in order to portray the dialogical structure of the self. He does not follow the PPR protocol, however. He created a semi-structured interview protocol, named the personality web, which does not make use of standard lists, in order to chart the different I-positions. Moreover, on the assumption that each position is constituted by different attachments to different objects (people, events, objects and our own bodies), the person is invited to elicit and explore the meaning of each of these attachments, and later, to group them into clusters. Albeit this protocol is different from the one used in the PPR, the overall result is quite similar (Jasper et al. 2011): to chart the different I-positions, as well as their mutual relations, see also Raggatt (2011).

Examining Methods Within the Dialogical Self Theory

SCM has several useful features, such as inviting the person to assume multiple perspectives about himself or herself, giving an overview of his or her life (Jasper et al. 2011). It allows the possibility of fostering the dialogue between neglected or downplayed positions by giving them a voice and therefore fostering change. These features have made this method particularly relevant in therapeutic contexts (see Hermans and Hermans-Jansen 1995). In its turn, PPR provides a self-reflective exercise through which the person is guided to chart the organization of internal and external I-positions and their mutual coordination. It may also foster change processes by inviting the person to give voice to particular positions.

There are variations in these methods (for other examples, see Puchalska-Wasyl et al. 2008) but they generally maintain some of the original features: they heavily rely in the explicit recognition of I-positions by the person. They allow interesting exercises of self-reflection by making explicit the usual implicit I-positions, and their clinical merits go without saying: clients may gain new awareness of their own positions, expression of dissociated or suppressed positions is fostered, and the dialogical interchange between positions fuels psychological change and adaptation (see Dimaggio 2011 for an example of the application of SCM as a way of exploring different I-positions).

Nevertheless, there are some drawbacks to the existing methods (see Cunha et al. 2012; Gonçalves and Salgado 2001; Jasper et al. 2011). First, since SCM and PPR rely on the explicit and reflective recognition by the person of his or her positions, these methods are oblivious to the implicit here-and-now experiential I. Then, as we have been arguing (see Gonçalves and Salgado 2001), a position that is not recognized as such by the person goes unnoticed: “Relying almost solely on self-reflexivity and self-report to determine self-positions undermines the possibility of analysing some experiences that are somehow not completely accessible or understandable to the person and, therefore, not captured through one’s self-reflections” (Cunha et al. 2012, p. 66). Gillespie and Cornish (2010) also eloquently argued that if I-positions are not transparent to the speakers, then self-reports are inadequate methods to capture the implicit and experiential nature of the positioning process. In sum: “To base our analysis on the self-reports of participants would be to limit severely our window on the self” (Jasper et al. 2011, p. 329).

Second, dialogue between the positions is a methodological invitation and it does not derive from the naturalistic observation of life events (Cunha et al. 2012; Jasper et al. 2011). As a result, we can only assess change in the self in a relatively large time frame (from one self-investigation to the next, for example), which prevents capture of the ongoing embodied changes taking place from moment to moment (Cunha et al. 2012). Thus, since these methods are oblivious to the implicit constant processes of positioning and the actual dialogical dynamics, they are not suited to capturing the on-going movement of positioning. In sum, they are not fit to capture naturalistic positioning processes in actual social life circumstances. Thus, if we want to study the positions which occur at each moment, other methods need to be developed.

We will proceed by reviewing two other methods with a Bakhtinian influence that seem particularly relevant for the study of the self in movement (but see also Grossen 2010; Grossen and Orvig 2011; Mininni 2010).

Dialogical Sequence Analysis

The dialogical sequence analysis (Leiman 2004, 2012; see also Leiman and Stiles 2001) is a clinical microanalytic method devoted to the study of clients’ utterances in psychotherapy. It is meant to be both a tool for clinical case formulation and a method of process research in psychotherapy. Generally, it assumes that detailed attention to clients’ discourse makes it possible to trace sequences of positions associated with their psychic life and therefore fosters our understanding of their problems.

Essentially, it is a blend between the Vygotskyan approach to semiotic mediation, the Bakhtinian notion of utterance, and the integrative therapeutic approach known as cognitive-analytic therapy (CAT; Ryle 1995). It takes the semiotic position as the unit of analysis in the study of clients’ utterances: “By studying the semiotic positions that individuals adopt or assert in their utterances we can infer how they relate to other people, to things and events in their everyday life, and to themselves” (Leiman 2011, p. 456).

Departing from the Bakhtinian circle tradition (Leiman 2011; see also Salgado and Clegg 2011), it principally assumes that clients’ utterances express their position towards different “objects” of their world and that a position has three main structural elements: an agent, multiple addressees, and a referential object. For example, if someone says, “This is lovely!” while receiving a gift, this person is assuming the semiotic position of someone who is thankful (agent) about a “gift” that is qualified as “lovely” (referential object), and this expression addresses those who gave the gift (addressees). Beyond the obvious or explicit addressees, a client’s utterances may also contain hidden addressees.

At the same time this method also pays special attention to the dynamic aspects of these positions, since it is designed to trace the sequences of action between the agent and the object of activity. It is assumed that the acting subject and the object of activity have a relation of dynamic reciprocity: the subject acts upon the object, constituting his or her position; in its turn, this activity produces effects on the object, composing a reciprocal “counterposition”. The subject may then readapt his or her action towards the object, assuming a new position. DSA uses the term “dialogical pattern” to describe these sequences of reciprocal positions and counterpositions and their unfolding over time (Leiman 2004, 2012).

Based on these notions of “semiotic position” and “dialogical pattern”, DSA is grounded on group discussion of data sessions (transcripts, recordings) and it proceeds by selecting relevant extracts. The next step consists in determining the referential objects in the person’s utterances, and the positions and counterpositions involved, in a way that is coherent with the theoretical principles aforementioned. The dialogical patterns can be described in the form of a diagram (see Leiman and Stiles 2001). This method explicitly opposes procedures that seek to verify reliability or consensual procedures (Leiman 2012).

Compared with the previous methods, DSA represents a significant asset for several reasons: it allows the analysis of the ongoing interaction between client and therapist, without needing explicit self-recognition of the positions by the client; it depicts the process of repositioning between subject and object; and it also enables the study of psychotherapy as a social event as it unfolds. Therefore, it helps to solve some of the previous problems associated with the methods based on the dialogical self theory.

Nevertheless, some points need to be made regarding this method. First, the notion of position and counterposition, and the role assigned to the referential object, in our view, obscure the role of the addressees. In our view (see Salgado and Cunha 2012) the referential object is important insofar as it enables a dialogical relation with others. Therefore, we think that Leiman’s notion of a “dialogical pattern” between positions (occupied by the subject) and counterpositions (occupied by objects) does not entirely capture the dynamic between the subject and the addressees. Thus, DSA assumes the reciprocity between subject and object as the basis of positions/counterpositions, whereas a sound dialogical method would treat the relationship with the multiple addressees as the main element of such an account. Of course, this goes probably unnoticed when the referential object and the addressee overlap (for example, when someone says “You are so nice to me”, the object and the addressee converge). DSA does not seem to have an adequate solution to the problem of how to take addressees into account when the utterance refers to objects, however (for example, “I like this gift of yours so much”).

Moreover, DSA is practiced in a way that makes it difficult to harmonize with more systematic practices. This is not necessarily a problem, but it has certain consequences. For example, some procedures are loosely defined: that criteria for selecting some passages of the cases are not completely clear; the same applies to the rules governing the establishment of a sequence. The final result is that it is difficult for new research teams to apply the method autonomously.

Dialogical Methods for Investigations of Happenings of Change (DMIHC)

A methodological approach known as dialogical methods for investigations of happenings of change (DMIHC) has recently been developed by Seikkula et al. (2012), and it is still at an early stage. Three concepts occupy central stage: voice, positioning, and addressees. The starting-point is a distinction between “egological” and “dialogical” notions of voice. One of the most popular notions of voice comes from the assimilation model of problematic experiences (Osatuke and Stiles 2011; Stiles 2001), an integrative model of therapeutic change, in which voices are seen as “traces of experiences” of the past activated by current events. This definition is seen by Seikkula et al. (2012) as self-centred and egological, and therefore inconsistent with a dialogical perspective (see Salgado and Ferreira 2005 and Salgado et al. 2005, for a similar argument).

Following a Bakhtinian approach, “voice” refers generally to any embodied form of communication. It is “human communication” as it comes alive; it is personal, since it is performed by a human being; and it is also an expression with a direction and a materiality that makes it socially addressed (see Bertau 2008; Josephs 2002). It is the “speaking consciousness” (Bakhtin 1984; Holquist 1990), the consciousness in a communicational or dialogical act of addressing others.

The two other concepts are inherently connected with this dialogical notion of voice. If a voice is a communicational action, there must be social others at the other end of any communicational act. These are called “addressees”, who may be the interlocutors of a dialogue or third parties (i.e. those who participated in discussion of the same issues in the past) (Seikkula et al. 2012). A voice is therefore always a response to something previously said or done, and through this expressive act the person assumes a position towards the social audience. Again, following Bakhtin (1981, 1984), Seikkula et al. (2012) argued that every utterance has an author which position it expresses. Thus, positioning refers to the perspective which the person is assuming in relation to the perspective of the relevant social audience.

DMIHC is designed to study family therapy and the ongoing interaction between the multiple actors involved (family members, therapist) taking these three basic notions as cornerstones. Therefore, it is devoted to the analysis of sessions, and these are made available through verbatim transcripts, but also on video, in order to enable a full observation of the interaction and its subtleties. It involves three successive steps. Step 1 consists in selecting relevant topics in the dialogue which are considered important by the researcher. Step 2 consists in exploring the responses of the different rejoinders in the dialogue. In this step, the researcher pays special attention to the following dimensions: dominance in the dialogue and shifts in this dominance; the content of the utterances (what is being responded to and what is not); and the degree of openness to (and assimilation of) others’ views shown in the dialogue; the style of expression (gazes, gestures, etc.). Step 3 consists of an exploration of the linguistic processes of narration, paying specific attention to the use of indicative versus symbolic forms of expression; and the exploration of the different types of narrative-telling (external, internal, or reflective; Angus et al. 1999; Laitila et al. 2001). In order to foster discussion and broader understanding of the data, the analysis is preferably based on group discussion within a group with different researchers.

This method is clearly rooted in a Bakhtinian perspective and it articulates it in a very refined way. At the same time, it also allows the analysis of “real” social interactions between people and natural movement between positions, since it is applied in family therapy sessions. Nevertheless, since its focus is more on interactional patterns, its use mainly results in an interpersonal understanding of the undergoing processes and the degree of openness to the other. This is evident when we look at the empirical treatment of the main elements of analysis (voice, addressees, and positioning): they are reduced to their interpersonal interactional features (dominance of discourse activity; contents privileged in terms of interaction; openness to the perspective of the other). There is no specific tool to account for the inner dialogical activity of each participant. Consequently, psychological processes are not usually taken into full consideration. As a result, this method is not applicable to individual psychotherapy, for example; more importantly, this difficulty reveals that the method does not create a window on the inner dialogicality of psychological processes. This inner dialogicality is explicitly recognized by the authors, but the method ends up using the notion of dialogicality as something very close to “real dialogue” (and therefore preserves the distinction between “monological” and “dialogical” dialogues—because their focus is only on the dialogue and neglects other forms of human psychological activity). Therefore, even if it is a very valuable resource for the analysis of social interactions, it does not entirely respond to the need for a method that explores the dialogical processes of the self.

Synthesis

The variety of methods for the study of dialogical properties of the self is a sign of the growing interest in the field and its vitality. At the same time, it would be impossible to devise a single method that could account for all the varieties of perspectives and goals that one might have while doing research in this (and any other) field. Nevertheless, given the arguments previously presented, a full dialogical approach to the study of the self implies four main features:

  1. (1)

    a focus on the self, here understood as the subjective and holistic experience of being an agent;

  2. (2)

    discrimination of the dialogical processes involved in the constitution of that experience;

  3. (3)

    a developmental perspective coordinating micro-changes with macro-changes: studying how the constitution of the self unfolds through time involves an articulation of different observational time frames (from more micro-oriented, moment to moment, to more macro-oriented, within larger time frames);

  4. (4)

    accounting for the naturally occurring dialogical processes, which the persons involved may not be aware of, something that calls for an external observer.

Table 1 presents a comparative synthesis of the four methods based on these four elements. As can be seen from the above, we are still in need of a method that accounts for the constantly occurring dialogical processes of psychological activities related to the self. In other words, none of the aforementioned methods is capable of fully meeting these four requirements simultaneously.

Table 1 Comparative analysis of SCM, PPR, DSA and DMIHC
Full size table

Delineating a Research Strategy for Studying the Dialogical Processes of the Self

In order to accomplish the task of creating a research strategy for tracking (1) self-positions, (2) their dialogical dynamics and (3) their development through time by (4) external observers, some conceptual refinements will now be introduced.

Position, Self-Experience and Access to the Subjective State of Mind

In order to be able to study the movement of subjective positions through time, it is necessary to deal with the problem of accessing them. In this section, we will clarify our standpoint regarding this issue in order to clarify how it affects our methodological decisions.

First, it is necessary to explain that by position we mean the subjective experience of the person. The term position implies a spatial metaphor and necessarily requires a relation between the person and others within a social context. Nevertheless, we have suggested that position is not only a concept entailing a material social interaction but is also constitutive of the subjective experience (see Salgado and Ferreira 2005). This distinguishes our notion of position from other notions much more focused on the interpersonal interaction (for example, Harré and Langenhove 1998). In our view, there are two simultaneous, complementary, but contrasting planes of dialogical activity: one is the psychological, experiential plane; the other is a material plane of interaction (communicational and factual) with the socio-material world. Of course, these two planes interact with each other, and we can conceive the psychological plane as a membrane of contact with the surrounding environment (Salgado and Valsiner 2010).

In this sense, self-position is a relational term, but our focus is its experientially subjective element; it is treated here as an emergent psychological experience of (and response to) ongoing contact with the world (social and material) (Cunha and Salgado 2008). Self-positions are conceived, at a microgenetic level, as events connected to the present moment of lived experience (Stern 2004), implying a self-state. Conceptually, we understand this notion as a specific convergence between a self-experience felt here and now (“Here-And-Now-I-System”; Valsiner 2002) and a spatial, temporal and semiotic framework of the subjective experience from which something is said and communicated to others.

Of course, as Leiman (2011) argues, an external observer has no direct access to the internal experience of a person. We do, however, have access to the utterances of the person through which we gain some access to the private realm of personal experience. From a dialogical perspective, embodied human experience only gains meaning if refracted by sign-mediated activities, rendering possible simultaneously,its sense and its communication (Leiman 2011; Valsiner 2002). This applies to private processes (for example, thinking), as well as to public expressions of those processes (talking). Thus, utterances are emergent forms of communicating and regulating, at least partly, the inner realm of personal experience. For example, if someone sobs “I’m terribly alone”, we can start understanding the sorrow and pain of the here-and-now experience of this person at the same time that the person is articulating that same experience. The emergent self-state is a response to actual circumstances, and utterances are constitutive of that self-experience. Thus, we consider contents of utterances as emergent responses through which the person conveys to others his or her personal position and state regarding particular issues (and not only others, but also to himself or herself). Therefore, through the analysis of utterances, we can grasp the inner experience of the person, even if only in a limited extent.

The Inner Dialogical Dynamics of the Self: A Brief Outline

If the self-state is a response, then it is always a response towards someone and something. Thus, a self-position always implies a co-relation between an I and two other necessary elements: others and objects.

This is aligned with the idea that a dialogical perspective implies a triadic account of any human activity, in which an Ego, an Alter, and an Object are simultaneously distinguished and related (Marková 2003). In other words, meaning-making is rendered possible by the activity of an agent, who uses semiotic means to address others (even if only virtual or potential others) about certain objects. Thus, a self-position emerges as a relation between three elements:

  • a communicational agent: it corresponds to the agent, or the author of the action. This agent has a certain ongoing experience, which is partly articulated by verbal and non-verbal expressions.

  • the addressees: Generally speaking, the addressees are those addressed by the agent, and they may differ: they can be the interlocutor, people explicitly invoked by the discourse, or potential or virtual others. In other words, it is the other who is taking part in the utterance.

  • the objects: these correspond to the specific discursive content of what is being addressed to the addressees. They refer to the main object(s) of reference involved, to the objects of discourse.

These elements are essential for the understanding of a position, since it is always through the co-relation between them that a position actually comes alive. Empirically, three major questions need to be answered in order to define a self-position: who is speaking, to whom and about what. The content of the communication is relevant for distinguishing and relating these elements. As Wortham (2001) claimed, we can distinguish referential positioning and interactional/interpersonal positioning. Referential positioning refers to the content of discourse or narrative-telling, in which the agent positions himself or herself towards others. Thus, there is a dialogical activity displayed within the narrative or discursive content, and it is through this referential positioning that interactional/interpersonal positioning is activated. Thus, focus on the activity of the self is related to the dialogicality contained within the utterances. In other words, utterances display a dialogical dynamics involving these three elements. They may be empirically observable, but that is not always the case.

In practice, these elements interact and may overlap: for example, sometimes the object and the addressee may coincide, sometimes the agent may be occupied by the voice of someone else. Therefore, we have developed some finer distinctions in order to have the possibility of empirically defining who is the agent, to whom is speaking, and about what—and how these elements inform each other.

Since there are always multiple addressees, we need to introduce a differentiation between the “second parties” (the you) and third parties (the she/he/they) of the communication. The second party is the one who is being targeted by the agent (what we have called the “other-in-the-self”; Bento et al. 2012; Salgado et al. 2005); the third parties are those who mediate the relation between I and Other (what we have called “inner audiences”). For example, if someone says to a therapist “It is very embarrassing to tell this to someone”, the therapist is the addressee, but there is also an implicit inner audience which is creating this feeling of embarrassment. We will refer to the second parties as “addressees” and use the term “inner audiences” to denote the third parties. Theoretically, we claim that inner audiences necessarily exist, but empirically they will remain, usually, inaccessible, even to the own person (see Linell 2009, for a careful review of third parties and their role).

Sometimes, an observer does not have a clear indication of the other-in-the-self. Occasionally it happens because utterances do not disclose who the target is. For example, insulting graffiti do not (usually) specify for whom the message is intended. Perhaps even more important is the fact that in daily life the addressee is sometimes only virtual or potential. Monologues or personal journals usually do not envisage anyone in particular at the other end of the expressive communicational act. Nevertheless, the other-in-the-self is considered to be a theoretical imperative of a dialogical account. The explanation for this imperative comes from the notion that meaning is brought into being by joint activity between human beings, and therefore there are no private sources of meanings. As a consequence, every form of meaning implies a second element—a “you”—with whom the expression can be shared.

It is also important to distinguish between the addressee and the interlocutor. From an interactional/interpersonal point of view, the interlocutor is the obvious addressee. From a psychological point of view, however, the addressee does not necessarily coincide with the interlocutor. This has to do with the way we conceive the addressee (the other-in-the-self), namely, the one to which the person is directing the content of his or her discourse. Imagine, for example, a woman speaking with emotion about her husband: “He was not there for me… I missed him very much”. In this situation, the therapist is the interlocutor, but operating more as a witness of the woman’s personal drama than as the addressee of her expression (i.e. as a third party to this personal drama). In fact, it is her husband who is being addressed, as it was he whom she missed and feels sorry about. Therefore we would say that her husband is the addressee, and the therapist remains in the background,as the audience mediating the expression. This woman may think her feelings validated by the silent but attentive presence of the therapist.

Regarding the object, it may be varied: social (“my family”) or personal (“my mother”), material (“this table”) or psychological (“extroversion”), concrete (“this car”) or abstract (“the laws of movement”), referring to others or to the self. All kinds of objects can be the focus of the experience, here and now, upon which the person establishes a personal position. When someone is talking simultaneously about and to the “addressed other” the object coincides with the other-in-the-self: for example, when someone says “you are a liar”, the object and the “you” are the same.

In sum, within a dialogical analysis, it seems useful to track the following elements in each utterance: the agent; the addressee (the other in focus in the utterance); the inner audiences (audiences invoked by the utterances); and the referential objects. By tracking the dialogical dynamics established within these elements, we can identify the emergent position of the moment.

From Microgenesis to Macrogenesis

In our view, the observation and the study of the self benefit from a developmental strategy of investigation (Cunha 2007a, b). Valsiner (2007) has proposed a distinction between three different levels of developmental organization: the microgenetic, the mesogenetic, and the ontogenetic. The microgenetic level refers to the primary level of experience taking place at each moment in an irreversible flux of time: there is always a new situation to which the person is adapting. At this level, there is great instability from moment to moment since the person is responding to the current situation by always assuming a new position. This new position may be similar to the previous, but it will never be the same: this is the constant “genetic movement” creating new positions. The mesogenetic level refers to situated activity patterns or social frames, which tend to have a more stable and repetitive occurrence. In terms of positioning, these frames can also be understood as more or less stable patterns of positioning (see Moore et al. 2011). For example, an adolescent doing homework assignments is performing within a social frame of situated activity (mesogenesis). Microgenetic variations certainly happen in the positions occupied (liking, disliking, liking even more, etc.). Instability at the microgenetic level is regulated by social frames that coordinate that variability and warrant some similitude over time (e.g. taking breakfast, jogging, teaching classes, …). Finally, at a more macro level of observation, there is ontogenetic development, referring to the trajectory of one’s life. This is a more general level of organization of a person’s experience: how they are performing in school and at work, their goals in life, and so on. Valsiner (2007) asserts that this level implies the creation of some relatively stable structures that guide the person within their life course, regulating micro and meso levels.

These three levels are not only levels of analysis but also levels of genetic development: they are, in themselves, structures always in the making, always reconstructed, always re-emerging. At the same time, they co-regulate and constrain each other (see Valsiner 2007, for a more detailed account).

These are very important concepts, since they indicate the need to articulate different levels of developmental analysis in empirical observations. Thus, microgenetic development does not necessarily imply mesogenetic or ontogenetic development. There may be small changes of a different sort taking place at a micro level: the “normal” regular movement from one position to the other; or even the emergence of new positions that in fact do not create novelty (see Fogel et al. 2006). Mesogenetic or ontogenetic levels of development always need microgenetic changes, however. In terms of the positioning process, this means that when studying the self, we need to observe these different levels and their mutual coordination at the same time. On the one hand, this implies tracing the movement of positions from moment to moment (micro level); on the other hand, it means extracting the patterns of positioning which occur through time (macro level).

Positioning Microanalysis

As a result of this general strategy, the task we need to pursue is that of building a research tool capable of tracing the subjective process of positioning in terms of its dialogical dynamics while articulating different levels of development. For several years now, we have been developing a method for studying the dialogical processes of the self (Cunha 2007a, b; Salgado and Cunha 2012; see Cunha et al. 2012 for a review of the various phases of development of this method). It is a research tool meant to study the microgenetic development of the self and how this level relates to more macro levels (mesogenetic and/or ontogenetic). Since our research is mainly focused on change processes in psychotherapy, we have been particularly interested in developing a method applicable to verbatim transcripts, whenever possible aided by video recordings. We also think it is adaptable to other kinds of data. We will now present the latest version of this method (the manual is available directly from the authors; Salgado et al. 2012).

Stages of Analysis

Positioning microanalysis is mainly a qualitative method, which involves the following successive phases: (1) dividing the transcripts into units of analysis; (2) aggregating the units in themes; (3) characterization and labelling of self-positions; (4) interpretative developmental analysis. The method is based on teamwork of different kinds: the process of dividing data into units of analysis uses different judges in order to assess inter-rater reliability; all other processes are based on consensual discussion procedures. In its standard format, the method relies on two judges who perform each task on their own and then compare their perspectives, discussing and resolving their disagreements; the product of each phase is presented to another researcher, whose function is to audit the whole process (Hill et al. 1997).

Preparing the Analysis

In order to perform a position microanalysis, there are some requirements which have to be met first: judges must have previous training; data must be selected; and the transcripts need to be read and, whenever possible, videos should be watched.

Training

We have developed a research manual on PM (Salgado et al. 2012). Judges start by reading this manual, in which the main steps in the process are detailed. After becoming familiar with the procedures, they start training for each phase of the process with workbooks in which they have to perform every step, following the sequence of the analysis: unitizing, identification of themes, characterization and labelling of self-positions, interpretative final work. At each stage of the training, results are compared with those of other judges in training, under the supervision of a more experienced researcher. Judges go through 15 h of training before the beginning of data analysis. Training involves both conceptual (e.g. dialogism and dialogical self theory) and methodological components (e.g. consensual qualitative research and analysis of dialogical parameters). This training can be done in small groups in order to create analytic routines that closely follow those that will take place in the analysis of new data.

Selection of Data

This depends on the research problems and questions. We have largely analyzed entire sessions of therapy, but it is also possible to restrict the amount of data, depending on the interests of the team. For example, if the research team is aiming to study the microgenesis of clinical problems, then it is possible to select only the specific points at which those problems arise in the session.

Reading Transcripts and Watching Videos

Researchers are recommended to obtain a global understanding of the data before starting the procedure, since this fosters a better understanding of the material under analysis. Most of the procedures require a written transcript, but it is also advisable to watch videotapes of the same materials whenever possible in order to have a more comprehensive view of the data.

Team Procedures

These procedures vary across the four stages of analysis detailed below. For the first step, at least two trained judges divide the transcript independently into “response units”, which is our basic unit of analysis (see below); after completion, the percentage of agreement between the two segmentations is calculated (see Hill 2009, for further details). Inter-rater agreement is calculated as the percentage of coincidence in the units of analysis independently divided by the judges. A degree of reliability is recommended of at least 90 % of inter-rater agreement in unitizing of the entire transcripts under analysis. Disagreements are solved through consensus. Finally, an auditor also reviews the process of unitizing. Then judges can proceed to the next step.

We use the procedures of consensual analysis defined by Hill et al. (1997) to perform the second, third and fourth steps of the analysis. This process involves a team of at least two trained judges who independently perform the tasks involved at each stage. Then, extensive discussion of the data takes place, based on the questions to be answered in each step of the analysis: which thematic objects are present? what are the dialogical parameters? what would be a good label for the self-position within each unit? and which patterns of positioning can be identified? (see the following steps). The discussion by the team of trained judges will lead eventually to a consensus. The result is sent to an auditor who has to check the interpretations and send his or her feedback to the judges. This process ends when there is a final consensus between judges and auditor.

First Step: Unitizing

The first step of PM consists in dividing the verbatim transcripts into small units of analysis. As previously stated, it is through the analysis of utterances that we gain access to the personal subjective self-positions. The definition of utterance is somewhat loose, however, making it difficult to decide where an utterance starts or ends. In order to solve this problem and to use a reliable process of unitizing transcripts, we adopted Hill’s (2009) procedures for unitizing transcripts. This procedure is based on the notion of a “response unit”. Which is empirically close to the notion of utterance. A response unit is defined as an independent unit of meaning, usually the equivalent of an independent clause in grammatical terms. An independent clause is a complete statement with a subject and a predicate that stands alone. For example, consider the following intervention by the client (see Table 2): “yeah, feelings are um well they recur,/they haven’t gone away,/they’re just more suppressed than all of them”. In this intervention the client is expressing three independent ideas regarding her feelings: 1. feelings recur, 2. feelings have not gone away, and 3. feelings are more suppressed. As these independent ideas stand alone, the client’s intervention would therefore be divided into three different “response units”: 1. “yeah, feelings are um well they recur”, 2. “they haven’t gone away”, and 3. “they’re just more suppressed than all of them”.

Table 2 Excerpt from an analysis of a client’s utterances (Lisa)
Full size table

Second Step: Aggregating the Units into Thematic Objects

In the second step of PM, judges are called upon to aggregate previously defined units of analysis into thematic objects (agent, addressee, inner audience, and relational domain will be coded in the next steps). This step answers the question regarding the “what” (the object) of the self-position, and enables an aggregation of the different parts of the transcript in specific themes. In order to perform this analysis, judges follow some guidelines generally adopted in qualitative research concerning the thematic categorization of discourse. Themes correspond to domains (Hill et al. 1997) or to major thematic units (Stinson et al. 1994, p. 39), i.e. to “macrostructures of meaning expressed in aggregates of one or more of thematic units”, depending on the similarity of the contents that are expressed by them. Focus is on the overarching structures that guide life situations expressed in conversations (e.g. in therapeutic conversations), such as “my marriage” or “my job”. Thus, the thematic object is not equivalent to the specific particular reactions, feelings or thoughts that clients and significant others may have as a result of those situations but refers to the situations themselves.

Third Step: Characterization and Labeling of Self Positions

After the transcripts have been segmented into units of analysis and aggregated into thematic objects, the third step consists in the identification and labeling of self positions and their relational domains. This is done on the basis of the dialogical parameters involved and it involves two complementary sub-steps: the dialogical characterization of self positions, and the labeling of the self positions.

  • Dialogical characterization of positions. As previously outlined, a dialogical account involves a relation between different elements, namely, the agent, the addressee, the inner audiences, and the referential objects. These last were previously identified, and now the judges proceed by identifying, for each unit of analysis, the other dialogical parameters. This analysis also adds another element, the identification of the relational domain. As stated previously, discursive processes refer back to the relational contexts in which they originally took place. In this sense the relational domain renders explicit the relational context to which the content of each unit of analysis refers and it is highlighted in order to facilitate exploration of the patterns of positioning after positions are identified.

    • The agent. The judges code who is speaking. There are two main possibilities: the agent coincides with the speaking person (which is the most frequent situation); or the agent is assuming the voice or perspective of another person (e.g. “At that moment he was very frightened and he said to me: ‘Come on, let’s get out of here, I’m not feeling right’”). It is also possible to find units in which the speaking agent is more than one person, at the same time; in these cases, all the agents should be coded (e.g. “and we all agreed that it was an excellent musical work”).

    • The addressee. The addressee constitutes the focal person or group of persons addressed in each unit. The addressee to be coded is the recipient/s of the symbolic action, which is taking place in the discourse. For example, the addressee to be coded in the case of someone saying in therapy “I am in love with her” will be the loved woman. Therefore, it is distinct from the interlocutor (in the example, it would be the therapist) and from the thematic object (the feelings this client has). There are also situations in which it is vague, undisclosed, or hidden. In these cases, the addressee will remain implicit and cannot be coded. For example, if someone says out loud, when alone, “This is beautiful music”, the coder of such material may remain unclear who the actual addressee of this utterance is. In sum, the addressee varies depending on the referential others contained and addressed in the content of the speech act. It can be the speaker, the interlocutor, but frequently is someone else, including groups, communities, or different forms of generalized others (“my friends”).

    • The inner audiences . When coding these audiences, judges specify all the others involved in the unit. These are usually of two kinds: the interlocutor, whenever she or he is not the addressed other (as in the example “I’m in love with her”); and/or other people, when these are referred to in the discourse. The following exchange is an example of the latter: “[Therapist] Could you tell me a little bit more about that particular situation with your stepfather? [Client] I’m sorry but I’m not comfortable talking about it yet”. In this case, the stepfather acts as an inner audience, and the therapist as the addressee. As this example suggests, the therapist (the interlocutor in the case of a therapy session) is considered as the addressee only when the client is answering a direct and explicit question from the therapist, when he or she is addressing the therapist without mentioning others, or when he or she is discussing aspects of the therapy or the therapeutic relationship. This means that in answer to a direct question from the therapist, if the client refers to others, he or she will be considered an internal audience.

    • Identification of relational domains. Relational domains correspond to the specific relationship that is referred to in each unit of analysis. For example, in the above example, when the client said “I fell in love with her”, the relationship constitutes the relational domain. It may happen that two relational domains are referred to in a single unit: for example, if a woman says “I will leave him, regardless of what his family may say”, there are two simultaneous relation domains involved (the woman’s relationship with her husband, and her relationship with her husband’s family). The unit may refer to a relationship with a person, with a group or institution, social groups, fictional characters, or even fictional relationships with other people. If that is the case, the relational domain should be coded accordingly.

  • Labeling positions. After identifying the dialogical parameters, judges label each self position. This self position is reflective of the internal self state. As a general rule, the label should be as close as possible to the phenomenological content of the utterance. For example, if a person says “I’m feeling sad”, the label would be straightforward: “(I as) sad”. We have developed several rules to facilitate the process of labeling these positions (see Salgado et al. 2012, for a more detailed account):

    • when people use synonyms (for example: frightening and scary) judges should give the same label to these self positions instead of sticking to the person’s words in the different units of analysis.

    • in situations in which the person describes distinct characteristics of their internal state in the same unit of analysis, judges should distinguish the different self-positions. For example, in “It’s sad and scary”, judges should label two self positions (“I as sad” and “I as scared”).

    • in situations in which the person does not explicitly name internal states or contents, whenever possible judges should maintain their focus on the words used by the research participant and be descriptive when labeling the self position. For example: “and I seem to be (sniff) you know, not wanting to move on” becomes “(I as) as not wanting to move on”.

    • in order to distinguish self positions in which the agent is the person who speaks from situations in which the agent “is” someone else (the voice of someone else), we call the latter counterpositions, but the coding follows the same rules.

    • during the process, judges perform a process of systematic comparison as in content analysis (Fassinger 2005), in which some labels are aggregated in wider categorizations. This often implies going back and relabeling previously assigned labels. For example: after labeling a self-position as “I as sad”, if a person says that she has negative recurrent feelings (“I as feeling negatively”) and later on says “in those moments, I’m really depressed”, judges may go back and relabel the previous position as “I as depressed”.

Fourth Step: Finding Patterns of Positioning Through Interpretative Analysis

After concluding the third step, we obtain a very detailed description of positions occurring over time at a very “microscopic” level. The next step involves creating a more generic description of the positioning process by abstracting sequential patterns of self positions. This generates a generic macro description of the process, which can be combined with the detailed micro description previously obtained.

Since the aforementioned three steps result in detailed descriptions across time, the researchers can now observe these in order to find repetitive patterns or sequences of self positions, and their variations. For example, we may have a typical sequence of p1:I as afraid – p2:I as avoidant – p3:I as sad. This sequence, however, can have several variations (e.g. by repetition of some self-positions: p1 – p1 – p2 – p3; oscillations by introducing new self-positions: p1:I as afraid – p4:I as courageous – p2:I as avoidant). Moreover, if we are observing the same person for long periods of time, as may happen in psychotherapy process research, these regularities or patterns of self-positions can be compared, in order to verify if they have changed. For example, if p1 – p2 – p3 was an initial pattern in the therapy, it might have been substituted by another sequence (e.g. p2:I as courageous – p3:I as not afraid anymore – p4:I as free).

Although quantitative analysis can also be used for extracting sequences, in the present phase of PM development qualitative analysis is the norm. This involves reading the transcripts (and the coding) with specific attention to repetitive cycles of self positions. In order to facilitate the detection of these patterns, the following guidelines have proved helpful (Salgado and Cunha 2012).

  1. 1.

    Highly frequent self positions. In terms of self-organization a very frequent self position is a recurrent one, which means that it must be involved in some form of patterned activity. Usually, this is a good starting-point for detection of patterns: focusing on very frequent self positions, and then looking for any regularity in the preceding and proceeding self positions and/or sequences.

  2. 2.

    Themes and their regularity. Themes that are more prevalent in the extracts are more likely to provide regularities than the ones which occur less often. Moreover, patterns of self positions are usually organized around some specific thematic objects.

  3. 3.

    Adjacent occurrences, sudden shifts, and semantic justifications. These can be used as devices to assist the process of detecting and specifying the sequence. Whenever one position occurs after another there is an “adjacent” or contiguous occurrence. A recurrent “adjacency” is usually a good trace of a possible pattern. This includes repeated sudden shifts of positioning and/or theme. The same applies when a person explicitly recognizes a pattern (e.g. the unit “…since I am always thinking pessimistically, I am always anxious” suggests explicitly a small sequence should be taken into account).

  4. 4.

    Disappearance of self positions and emergence of new ones: Variations and alternative patterns. The emergence of new self positions around a particular object is also a potential index of variations on usual patterns, and/or they might indicate a developmental transition towards new ones. The same happens when a self position that was part of a regular pattern appears less often or disappears.

These guidelines cannot disregard the fact that the process is largely “discovery-like”. Following these guidelines, the researchers start by re-reading the transcripts along with the coding (self positions, and dialogical parameters) in order to formulate possible patterns, and their variations. Whenever one sequence of self positions seems to be repeated, the judges produce a description of this pattern. This description operates as a working hypothesis that needs to be refined through successive corrections: the researchers start by identifying this pattern in the transcript, assessing if it fits; if it does not match, corrections need to be made, and a new formulation of the pattern is done. Variations to the pattern are also described and applied in the same way. This process ends when the researchers have a description of patterns and their variations that are involved in the relevant parts of the transcript. The “left-overs” should, to a reasonable extent, be parts that are better characterized as random events without any apparent order.

We call these patterns “cycles of self positions”, and in clinical research it is possible to distinguish between “problematic cycles”—the cycles involving the person’s complaints—and “alternative cycles”—which involve a clinical change regarding the “problematic cycles” (Salgado et al. 2011). This can be represented by narrative-like format or a graphic format, in which case it will appear as a diagram displaying the most relevant positions and their sequencing.

Combining the three previous steps with this one results in a multi-layered description of the self processes taking place: from microscopic observations of the dialogical parameters and their relative emergent self-positions, to patterns or cycles of self-positions and their oscillations, and then to changes in those patterns over time.

A Brief Example of a Microanalysis

In order to illustrate the application of this method, we present a brief illustration. The general context of the passage is a clinical case of emotion focused-therapy, drawn from the York I Depression Project (Greenberg and Watson 1998). The client’s fictional name is “Lisa” and this case has been analyzed by several research teams (see e.g. Angus et al. 2008). Lisa was diagnosed with a major depression disorder. At the beginning of treatment she was 27 years old and unemployed. She was married with two school-aged children. Lisa reported feelings of sadness, guilt, and resentment toward her parents and husband. Her major complaint focused on the gambling problem of her husband and the effects this had on their relationship. Two sessions of Lisa’s therapy (initial and final sessions) were coded by two trained judges and one experienced auditor using PM. Inter-rater agreement for unitizing was 92 % for the two sessions (Salgado et al. 2011). An excerpt from one of the sessions will be used here to illustrate PM. We focus our attention only on Lisa’s utterances. Table 2 includes the results of the first three steps of the analysis, presenting the units of analysis, the coding of their dialogical parameters, and the emerging self positions throughout the interaction.

In this excerpt Lisa is mainly answering the therapist, who therefore serves as the addressee; this dialogue opens up exploration of her own feelings (first object) and then the feelings toward her “family” and her husband who constitute audiences to this interaction. Her “family” and husband become the second object—it will become clear, later in the session, that by family she is referring to her parents, which is only implied so far. Lisa, in this excerpt, is only voicing her own perspective, and therefore, she remains the agent.

The fourth step comprises the detection of a patterned cycle, of which this excerpt represents a good example. There is a transition between two relational domains: from a self-to-self to a self-to-family. In the self-to-self domain, Lisa appears dominantly as depressed and sad; in the self-to-family domain she appears mainly as resentful. The emergent positions appear in this order: I as depressed (four times); I as sad; I as resentful (three times); I as avoiding (hate); I as resentful; I as forgiving; I as empathic; I as resentful. The semantic justification of her depression, however, implies a mutual relation between her depression and her resentment:

Lisa: the sadness… and um I guess resentment still there

T: resentment towards?

Lisa: um - oh it would be my family (Therapist: uh-huh) and my husband

Therefore, her depression is associated with her resentment. The following pattern appears as a hypothetical pattern of coherent or enabling mutual in-feeding (Valsiner 2002; see also Freda 2011; Lyra 2011; Ribeiro and Gonçalves 2011). I as depressed (with I as sad as a possible variation) ←→ I as resentful. The following units suggest another pattern: there are successive movements between I as resentful and self positions that counteract this feeling, namely I as avoiding hate, I as forgiving, and I as empathic (three times). Therefore, these self positions are opposing the enactment of the resentful self position: avoiding hate, (over)empathy and the understanding of others stand out as buffers and blockers of Lisa’s resentment. Thus, we can develop the following working hypothesis: I as resentful ←→ I as empathic/understanding. Whereas the previous sequence was of a coherent nature, this mutual in-feeding between the two self positions is clearly of an opposing nature. Adding together the two sequences we obtain the following working hypothesis to be checked in the rest of the transcript:

$$ \mathrm{I}\kern0.5em \mathrm{as}\kern0.5em \mathrm{depressed}\leftarrow (enabling)\to \mathrm{I}\kern0.5em \mathrm{as}\kern0.5em \mathrm{resentful}\kern0.5em \leftarrow (opposing)\to \kern0.5em \mathrm{I}\kern0.5em \mathrm{as}\kern0.5em \mathrm{understanding}. $$

Conclusion

A dialogical theory accounts for self multiplicity as a matter of constant repositioning and responsiveness to the actual material and social circumstances in which new self states are constantly being formed. At the same time, it assumes that the core of psychological experience is the subjective here-and-now, a central agent that performs this constant action. Thus, selfhood is a matter of constant positioning across time.

From a developmental point of view, the emergence and regulation of self positions is the core question: how are self positions created and regulated within their constant flow over time? PM is a method that tries to answer this kind of question, by enabling an external observer to study the naturally occurring positioning process.

Compared with other available methods for studying the dialogical self, such as SCM or PPR, PM has specificities that need to be pointed out. Its major advantage is that it allows the observation of the naturally occurring positioning processes by external observers, such as therapeutic dialogues. PM, however, also has some limitations. First, it is time-consuming, involving thorough and coordinated teamwork (a regular 50-min session takes about 10 to 15 h to code). Second, it is strongly focused on self states expressed in verbal utterances. Since it was derived from analysis of individual psychotherapy sessions, these are usually easily available in these circumstances. At present, however, the adaptations that would need to be introduced for other kinds of data (see Salvatore 2011, for a more detailed discussion of the necessary connection between psychotherapy research and human communication studies) are unknown.

A final remark should be made about the usefulness of this method. Since it is highly time-consuming, it must add something new to empirical research in order to balance the input required. The question of what to do with the output of the analysis is highly dependent on the questions of the specific research project. Results may be used, for example, to test an idiographic hypothesis. Nevertheless, we believe that it is also possible, and tempting, to use PM as a tool to foster “discovery-oriented” projects.

We need to start by recognizing that, when concluding the analysis, we only have a very detailed description of microscopic units, and a very large-scale description of cycles and their change over time. Thus, we have micro and macro descriptions, but we still have not grasped the genetic movement of the emergence of self positions and cycles over time (see Marukami 2010). Since we still lack a comprehensive theory about the genesis of positions and their flux over time, a whole field of research lies ahead. In order to accomplish this endeavor, a methodological leap needs to be made: an inference of genetic principles or mechanisms governing the emergence of self positions and their flux. For example, the researcher may be interested in understanding the emergence of dominant self positions based on the interplay between the more micro-level dialogical parameters and the processes governing the production of regular cycles of self positions over time. Theoretical and empirical research is scarce, and this represents a big challenge. Valsiner’s (2002) work stands out as an exception to this rule at the theoretical level. It was based on that theory and in a previous version of PM that Cunha (2007a, b) developed research that deduced some new processes of creation of new self positions. That work stands out as an example of how PM can foster our knowledge about the developmental processes of the self.