The origins of voices: links between life history and voice hearing in a survey of 100 cases
Dirk Corstens and Eleanor Longden
From: Psychosis, 2013 Vol. 5, No. 3, 270–285
A data synthesis is presented from 100 clinical cases, 80% with a diagnosis of schizophrenia or other psychotic disorder, in which Romme and Escher’s “construct” method was used to formulate voice-hearing content and characteristics in relation to life events. Across the sample, most participants heard between two and five voices and the average duration of voice hearing was 18 years. At least one adverse childhood experience was reported by 89% of the sample, including family conflict, neglect, physical/sexual/emotional maltreatment, and bullying. In addition, a broad range of acute, precipitating stressors were associated with the onset of voice hearing itself in both childhood and adulthood. In 94% of cases, it was possible to clearly formulate the underlying emotional conflicts embodied by the voices (e.g., low self-worth, anger, shame and guilt). Representations for voice identity (e.g., disowned aspects of self, a family member, a past abuser) were formulated in 78% of cases. It is proposed that many individuals hear voices that make psychological sense in the context of life events, and that this information can be clinically applied in ways that serve personal recovery.
Keywords: auditory hallucinations; psychosis; trauma; formulation; causal explanations; diagnosis; hearing voices
Hearing voices which others cannot hear has traditionally been deemed a perceptual psychopathology, closely associated with schizophrenia, and often accorded limited significance beyond determining diagnostic and prognostic status. More recently however, understandings of this experience have been refined to incorporate the
voice-hearer’s1 psychosocial, emotional, and interpersonal circumstances. In particular, emerging associations between voices and traumatic events particularly, though not exclusively, childhood abuse (e.g., Bentall, Wickham, Shevlin, & Varese, 2012; Longden, Madill, & Waterman, 2012a; Read, van Os, Morrison, & Ross, 2005; Shevlin et al., 2011), have energised an enhanced appreciation of voices as meaningful representations of psychological distress and social adversity. Furthermore, evidence suggests that in addition to voice presence, associations exist between incidences of trauma and loss and the phenomenology of voice hearing itself. For example, voice utterances may directly embody traumatic events (e.g., Hardy et al., 2005) and specific adversities, such as childhood sexual abuse, may influence specific types of voices, such as those that are imperious and commanding
(McCarthy-Jones, 2011). In some circumstances, interactions with one’s voices can reflect a more general, pervasive manner of social relating, including patterns of entrapment, powerlessness, and subordination (e.g., Birchwood et al., 2004); in effect, embodying the social world of the voice-hearer (Mawson, Berry, Murray, & Hayward, 2011). Furthermore, voice content appears to be a better predictor of distress and psychiatric need than more traditional diagnostic variables like Schneiderian structure (e.g., Beavan & Read, 2010). In this respect, traumatic intrusions and appraisals may influence and protract voice maintenance (e.g., Andrew, Gray, & Snowden, 2008), and for many individuals, explicating links between life events and distressing voices provides a valuable framework for advancing recovery (Romme, Escher, Dillon, Corstens, & Morris, 2009). In this regard, Romme and Escher (2000) have advanced a method of psychological formulation, known as “the construct,” which provides a systematic way to place voice phenomenology within a biographical context. In doing so it follows the original tenets of the stress-vulnerability model (Zubin & Spring, 1977) by emphasising the formative influence of environmental factors: in effect, differentiating between the acute stressors that directly precipitate voice-hearing onset, and developmental events that create vulnerability for emotional crisis. In order to relate these factors to voice hearing, the construct utilises specific areas of enquiry (voice identity, characteristics and content, triggers, history of voice hearing, personal history of the voice-hearer) to explore two fundamental questions about representation:
(1) who or what might the voices represent; and (2) what social and/or emotional problems may be embodied by the voices. The method overlaps with best-practice principles of psychological formulation (Johnstone & Dallos, 2013) in that it summarises core areas of difficulty, respects and refers to client views on accuracy and expediency, integrates systemic, social and/or political factors, is amenable to constant revision and re-formulation, and uses an available evidence-base to link theory with practice (e.g., the demonstrated link between voice hearing and psychological stress). The “meaning” of experience is not imposed on the voice-hearer, but coconstructed in a process of collaboration and exploration, then subsequently employed to guide an individualised intervention plan that applies information about voice manifestation and emergence to promote recovery (Corstens, Escher, &
Romme, 2008; Longden, Corstens, Escher, & Romme, 2012b; Romme & Escher, 2000). This is consistent with the premise that construing meaning and narrative from distressing symptoms, including those in the context of psychosis, can help elucidate unresolved emotional conflicts and, in turn, promote hope, understanding, empowerment, reflectivity, and psychological adjustment (e.g., British Psychological Society Division of Clinical Psychology, 2011; Johnstone & Dallos, 2013; Stainsby, Sapochnik, Bledin, & Mason, 2010). The purpose of the current study was to retrospectively analyse constructs from 100 voice-hearers in order to identify common and recurrent themes. Although several investigators have examined voice content and/or characteristics in relation
to life events (e.g., Andrew et al., 2008; Daalman et al., 2012; Hardy et al., 2005; Read, Agar, Argyle, & Aderhold, 2003; Reiff, Castille, Muenzenmaier, & Link, 2011) to our knowledge none have done so in the context of detailed clinical assessment, or provided an extensive, theoretically informed phenomenological survey whilst retaining a semi-structured interview protocol. In addition, the current study will address several issues concerning the construct method: (1) whether it is possible to retrieve enough information to formulate voice representations (including with recipients of long-term psychiatric treatment and/or those with diagnoses of so-called schizophrenia-spectrum disorders); and (2) what psychosocial problems may contribute to voice emergence and maintenance.
At the time of making their constructs, all participants experienced voices to the extent of causing significant emotional distress and/or impairing social or occupational functioning. Participants were self-selecting, to the extent that they requested to develop constructs through clinical contact with the authors. All were in receipt of statutory psychiatric care and had heard voices within at least one week prior to the constructs being made. Participants were recruited from various countries during the course of the authors’ clinical work, including parts of Western Europe (Denmark, the Netherlands, the UK, Sweden), Australia, and Cameroon. The mean age of the sample was 35.94 years (SD = 11.72; range 15–62) with a female to male ratio of 57:43. The majority of participants (80%) had a diagnosis of schizophrenia or other psychotic disorder, and an extensive duration of voice hearing (M =18.21 years; SD = 13.77). The most common experience was to hear between two and five voices, although a small proportion (7%) reported clusters of more than 20 voices. The main demographic and clinical characteristics of the sample are reported in Tables 1 and 2.
The 100 constructs were gathered by the authors either singly or in collaboration (93%) between 2002 and 2012, with an additional seven obtained via two colleagues trained in the construct method. A majority (63%) were devised during teaching sessions facilitated by the authors, in which participants volunteered to talk
Table 1. Main demographic characteristics of the sample.
(N = 100)
South Asian 6
In paid employment 15
Married or co-habiting 16
Age 35.94 (11.72)
Table 2. Main clinical characteristics of the sample.
% endorsed (N = 100)
Other psychotic disorder 12
Borderline personality disorder 8
Affective disorder 5
Diagnosis not disclosed 7
Number of voices
Years of mental health service use 9.48 (12.04)
Duration of voice hearing in years 18.21 (13.77)
about their experiences in front of a mixed group of other voice-hearers and/or mental health professionals. The remainder were obtained during individual clinical contacts. The initial aim in formulating the constructs was as a clinical tool rather than to generate research data. However, permission was sought from individuals to use their information for educational and research purposes, with the understanding that it would only be presented in a numerical or otherwise anonymised form. In group situations, voice-hearers were reassured that the primary purpose of the exercise was to demonstrate ways of discussing and exploring voice-hearing experiences. The formulation aspect was de-emphasised so that participants did not feel pressured to give “the right response” for the benefit of the audience. Information was elicited by questioning around each of the five main construct themes (e.g., identity, characteristics, etc.) and recording summarised responses on flipchart paper so that participants could see what was written and correct it if necessary. The two questions about representation were considered in collaboration with the participant, with the author(s) proposing suggestions and prompts in the respectful and exploratory way stipulated by Romme and Escher (2000). Primacy was always given to voice-hearer interpretations, regardless of whether this was a framework endorsed by the author(s). If constructs were devised in a group situation, the author(s) met with the voice-hearer in private for 30 minutes afterwards to discuss the construct and ascertain its accuracy and usefulness. The notes derived during this process were subsequently used to create fulllength constructs, which were presented in the format described by Corstens et al. (2008) and Longden et al. (2012b). All participants (and, if requested, a nominated mental health worker and/or family member) then received written copies to provide further opportunities for reflection, correction, and feedback. In this respect, all constructs in the present study had been read, verified, and validated by the voicehearer concerned. The criteria for selecting constructs for the current analysis was based on the 100 most recently completed for which voice-hearer consent had been obtained. In addition, constructs were only included if the voice-hearer agreed with the way in which their experiences had been formulated.
As no established method exists for assessing links between life events and voice phenomenology, the coding frame was devised in reference to clinical expertise, existing theoretical and empirical literature, and the pilot work of Romme and Escher (2000) when originally delineating the construct (see below). The study data was derived from coding the contents of completed, participantapproved constructs. The latter all began with (1) brief demographic descriptions, then (2) proceeded to outline the five main variables of enquiry (voice identity, characteristics and content, triggers, history of voice hearing, personal history of the voice-hearer). The final section 3) formulated questions about representation (who or what might the voices represent; and what social and/or emotional problems may be represented by the voices: see Corstens et al., 2008; Longden et al., 2012;
Romme & Escher, 2000). To obtain the necessary frequency data, one author initially coded all constructs according to the criteria described below. To confirm agreement and consistency in this coding system, a subset of 30 constructs were then selected at random and rated on the same criteria by both the second author and an independent rater from the Bradford and Airedale Early Intervention in Psychosis Service. One area of enquiry, voice identity (i.e., numbers of voices, their name, gender and/or age) was not coded in that the information was provided unambiguously by the voice-hearers and was not open to interpretation. The remaining variables were coded for each voice according to the following criteria:
(1) Voice content and characteristics. This query refers to how voices express themselves and how they relate to the voice-hearer and/or each other. Voice characteristics were categorised using the following criteria: verbal or non-verbal utterances; perceived malevolent or benevolent intent; emotional impact; spatial location (internal or external); first-, second-, or third-person articulation; and identification: for example, as a known acquaintance, a family member, religious/ spiritual frameworks (e.g., “God,” “the Devil”), paranormal frameworks (e.g., ghosts), or a voice that is recognised as belonging to oneself, including oneself at a different age. Content was coded in the following way: commanding; criticising; threatening; making premonitions; interacting with one another; making direct references to trauma (e.g., talking about abuse); speaking in a foreign language; providing advice or encouragement.
(2) Triggers. When making a construct, appreciating what aggravates or elicits the voices can be relevant for understanding their emotional dynamics, as well as the underlying problems associated with them. Information was assessed according to whether voices were provoked by the following criteria: specific individuals (e.g., family members, acquaintances); specific circumstances (e.g., social situations, a particular room in one’s home), and/or particular emotions (e.g., guilt, shame, anger). Data were coded as “not identifiable” if there were no obvious triggers for voices.
(3) History of the voices. This variable refers to proximal life circumstances that precipitated voice onset (i.e., events that occurred within six months prior to voices first manifesting). The age of first onset was asked for, and whether new voices had subsequently appeared. Codes were derived from the framework developed by Romme and Escher (2000) around interpersonal stressors that may influence voice emergence and incorporated: childhood maltreatment (sexual, physical, emotional, neglect); serious physical illness (life-threatening and/or necessitating significant educational or social disruption); peer bullying; family conflict (e.g., chronic domestic arguments, rejection of the voice-hearer by other family members); sexual or physical assault post-age 16; bereavement; sudden, unexpected death of significant individuals; relationship breakdown; experiencing excessive criticism and/or high expectations; leaving home for the first time; problems at school; and witnessing
violence. Because of the numerous experiences reported, an “other” category was utilised which incorporated less common stressors associated with voice onset (e.g., workplace stress, giving birth, spousal illness, visiting a psychic, being adopted, starting psychotherapy, moving home).
(4) Childhood history. This question addressed distal, formative experiences and stressful exposures in the voice-hearer’s life prior to voice onset. Codes included: sexual, physical, and emotional maltreatment; neglect; serious physical illness; family conflict; peer bullying or rejection; experiencing excessive criticism and/or high
(5) Who or what do voices represent? When formulating voices, it is often beneficial to identity who or what they resemble (and thus what psychosocial dilemmas, and consequent interventions, are indicated by these representations). Personifications can be metaphorical; for example, a voice deemed “The Devil” may be formulated as representing an abusive perpetrator or, more specifically, as the person’s feelings in relation to the abuse, or the part of the voice-hearer that has “identified” with the perpetrator (Dillon, 2011; Longden et al., 2012a; Romme et al., 2009). Voice representation was coded as: an abusive family member, a nonabusive family member, a known abuser, a non-abusive acquaintance, or aspects of the self. Data was coded as “not identifiable” if the answer to this question was unclear.
(6) What problems do voices represent? This question informs the second half of the construct and explores circumstances at the core of voice hearing experiences; generally a legacy from overwhelming events, emotions, and circumstances that disempowered the person and which have not been properly integrated (Romme & Escher, 2000). As such it provides a “snapshot” of the prevailing areas of difficulty in the person’s life. The codes applied were problems and conflicts relating to: shame and guilt, sexual identity, self-esteem, anger, and attachment and
intimacy. Data were coded as “not identifiable” if the answer to this question was unclear.
Information about the voices and associated life events
The full results of this part of the analysis are reported in Table 3.
Identity of the voices. The most prevalent reported experience was to hear voices that could be clearly personified in terms of age, gender, and name. Adult, male voices were the most common, although a proportion also reported child (19%), adolescent (10%), or female voices (66%). Most voices had either elected names for themselves or been named by the voice-hearer, with only 32% of individuals reporting nameless voices.
Characteristics and content. All participants experienced verbal articulations from their voices, with 17% also reporting non-verbal sounds, such as crying or laughing. The most prevalent experience was hearing solely negative and malicious voices (59%), which either criticised (98%), commanded (73%), or threatened (64%) the hearer.
Table 3. Response characteristics for sections of the construct.
Construct theme Areas of enquiry % endorsed (N = 100)
Identity of voices
Child voices (< 13 years) 19
Adolescent voices (13–18 years) 10
Adult voices 94
Without gender 30
All voices named 46
No voices named 32
Mix of named and unnamed voices 22
Characteristics of voices
Malevolent voices only 59
Benevolent voices only 4
Negative emotional impact only 79
Positive emotional impact only 1
External (outside head) 74
Internal (inside head) 14
Both internal and external 12
Voices speak in 2nd person 97
Voices speak in 3rd person 66
Voices speak in 1st person 42
Combination of 1st, 2nd and 3rd 33
Combination of 2nd and 3rd 33
2nd person only 20
Combination of 1st and 2nd 1
Voice of a family member 47
Voice of a known acquaintance 47
One’s own voice 28
Religious/spiritual framework 14
Paranormal framework 2
Voices criticise the hearer 98
Voices command the hearer 73
Voices criticise others 66
Voices criticise the hearer and others 65
Voices make threats 64
Voices interact with one another 63
Voices give advice/encouragement 35
Voices refer to traumatic events 17
Voices make premonitions 5
Voice speak in foreign language 5
Specific emotions 76
Specific circumstances 50
Specific individuals 15
No obvious trigger 19
History of voices
Age of first onset
Under 10 years 35
11–20 years 24
21–30 years 24
31–40 years 10
>41 years 5
New voices appearing at later ages
Life events associated with initial voice onset
Family conflict 47
Emotional abuse 36
Severe personal criticism 35
Sexual abuse 23
Physical abuse 22
Witnessing violence 17
Problems at school 17
Sudden, unexpected death 16
Relationship breakdown 11
Leaving home 11
Physical assault 9
Physical illness 7
No identifiable event 7
Emotional abuse 72
Family conflict 65
Physical abuse 41
Sexual abuse 30
Domestic violence 23
High expectations 18
Serious physical illness 9
No identified stressors 13
Only four people reported voices that were solely affirming and supportive, although 37% heard a combination of positive and negative voices. The majority of voices were experienced externally (through the ears), and spoke mostly in the second and third person, or a combination of the two. Participants were also more likely to report hearing the voice of a family member or known acquaintance (both 47%) than hearing their own voice (28%). More unusual experiences included voices that made premonitions, that spoke in a foreign language, or were identified
with paranormal or religious/spiritual frameworks.
Triggers. Many individuals could identify precise circumstances or emotions that elicited or exacerbated voice presence. The list of personal triggers nominated by the sample is too extensive to reproduce here, although the two most common themes were emotions, such as guilt, shame, insecurity, sadness, anxiety, and sexual feelings (76%), and discrete circumstances, such as social situations, meeting family members, and being isolated (50%). In this sample, only 19 individuals couldn’t nominate any clear, consistent prompts for the voices.
History of the voices. Almost 60% of the sample reported voices beginning before the age of 20 (35% before the age of ten), although it was also common for new voices to appear at a later age, sometimes many years after initial onset. Only 7% were unable to identify clearly defined precipitating circumstances at the time of voice emergence. The remainder reported a broad range of stressful, interpersonal events directly preceding voice onset; the most common being family conflict, emotional abuse, severe personal criticism, and sexual abuse.
Life events before voices onset. The majority of the sample (87%) had experienced chronic social and interpersonal adversities before their voices started. The most frequently reported were family conflict and four types of childhood abuse (emotional, physical, sexual, and neglect) often in combination. When examining both traumatic precipitating events and antedating events, it is apparent that 89% of the total sample had endured severe stressors at some point over their life-course.
The 100 constructs
The full results of this part of the analysis are reported in Table 4.
Who or what do the voices represent. The substantial majority of the sample (78%) heard voices whose identity could be formulated in terms of lived experience. It was common for voices to be recognised as representations of aspects of self, including the voice-hearer at a particular age, or disowned aspects of the personality. These associations were often closely interlinked with precipitating events; for example, hearing the voice of one’s seven-year old self in adulthood, when one was abused aged seven. However, many voices were personified as significant others, most frequently abusive (45%) or non-abusive (30%) family members who in some way had played an important part in the voice-hearers’ lives. Common representations included abusive parents, a male friend or acquaintance, or a male perpetrator who had victimised the voice-hearer in some way. In this respect, the greater frequency of male representations corresponded to the higher number of male voices.
Table 4. Voice hearing representations.
% endorsed (N = 100)
Who or what do the voices represent
Aspects of self 48
Abusive family member 45
Non-abusive acquaintance 32
Male friend or acquaintance 25
Female friend or acquaintance 20
Non-abusive family member 30
Other female relative 1
Other male relative 1
Other perpetrator 23
Other male perpetrator 23
Other female perpetrator 12
Social and emotional problems represented by the voices
Shame and guilt 60
Attachment and intimacy difficulties 45
Problems with self-esteem 93 Conflict over sexual identity 7
No identifiable problems 6
What problems do the voices represent. The voices heard by the majority of the sample (94%) could be formulated as specific representations of social-emotional conflicts, most often resulting from interpersonal stress. Almost all the sample reported a pervasive lack of self-worth. Other underlying problems embodied by the voices included the expression of anger (60%), shame and guilt (60%), and difficulties with intimacy and attachment (45%).
The current study is consistent with previous research demonstrating associations between voice hearing and traumatic life events (e.g., Read et al., 2005; Shevlin et al., 2011; Varese et al., 2012), and the likelihood of individuals in psychiatric services to report a high volume of negative, persecutory voices (e.g., Andrew et al.,
2008; Beavan & Read, 2010; Honig et al., 1998). Furthermore, it represents a fusion of these perspectives by demonstrating that, in the majority of cases, associations between voices and precipitating life events can be established using principles of psychological formulation. In terms of the study aims, we found that it is possible to retrieve enough information to answer questions considering voice representations with individuals who are actively hearing voices, in receipt of long-term psychiatric treatment and/or have a diagnosis of schizophrenia. Voice emergence was additionally related to a range of social and emotional vulnerabilities, the most common being problems with selfesteem, anger, shame and guilt; which in turn were linked with previous adverse experiences. In this respect, participants were mostly able to provide sufficient] information to formulate a relationship between their life history and their voices. This was still the case when voice content did not contain literal reflections of traumatic events, a finding that is consistent with Romme and Escher’s (2000) experiential,
“making sense” approach to voice hearing, which locates the experience as emotionally significant and psychologically interpretable.
All individuals in the sample reported experiencing voices for at least two years (63% for 10 years or more), and in this respect are representative of patients who have been subjected to voices for long periods and for whom traditional treatment has little impact. Our work with the construct method demonstrates that many voice-hearers can identify clear, precipitating events for the emergence of their voices and, with support, are able to create a personal story about the relationship between their voices and these experiences. For many, these were lived narratives that had previously been obscured behind the label of schizophrenia, the patient role, and the pessimistic implications of a disease-model of voice hearing. Conversely, the construct aims to re-establish associations between voices and the events which precipitated their presence (expressed in terms of the two representational questions: “who or what” and “what problems” do the voices represent?). On the basis of this study, we suggest that psychiatric assessment of voice hearing could reveal a dynamic socio-emotional understanding if specific questions were incorporated into traditional assessment repertoires (see Table 5).
Table 5. Suggested assessment questions, and supplementary prompts, for a socioemotional understanding of voice hearing.
(1) What identity do your voices have? Are they male or female? How old are they? Have they always been that age or have they grown older?
(2) Do your voices have names? Did they name themselves or did you name them? Do you know why they have those names?
(3) What do your voices say to you? Do they interact with each other too? Are some voices more dominant, either with you or amongst themselves?
(4) Do your voices remind you of people you know, or have known? Is their tone familiar? Is their content familiar? Has anyone else ever said those things to you?
(5) Are your voices negative towards you, or other people?
(6) Do you hear comforting or supportive voices?
(7) What triggers your voices, or make them worse? Do the voices respond to particular people/places/emotions? Are there specific people/ places/emotions that make them less intense?
(8) What was happening in your life when the voices started? Have they changed since then?
(9) What was your early life like?
Many distressed voice-hearers have neither a specific awareness of who the voices “are,” nor a coherent, integrated emotional-cognitive model of why their voices appeared. Therefore the representations presented within the construct create a meaning and understanding for the voice-hearer in terms of an individualised, explanatory model for their experiences. By relating voices to overwhelming emotions, events and problems, voice-hearers may acquire a fresh, personal knowledge in which painful feelings and beliefs can be addressed and reconciled with authentic problems from their past (and often in their present). Although voices are often
experienced as “real,” and are associated with genuine people and emotions (which one can learn to cope with) they “are” not the real persons, but represent an internal emotional world. In this way, the construct has the capacity to create both distance (insight, new understanding) and proximity (the relevant emotions can become the focus of future recovery interventions). By deconstructing symptoms into complaints (Bentall, 2006) that emerged in a specific psychosocial context, strategies for relating to voices can therefore be derived from the level of individual experience rather than putative syndrome. In this respect, information on directing the themes of the construct towards treatment and recovery planning can be found in Romme and Escher (2000), Romme et al. (2009), and Corstens et al. (2008). Finally, the high incidence of trauma in the sample supports the contention that
psychiatric staff should receive support and training to facilitate routine enquiries about exposure to abuse and adversity (e.g., NHS Confederation, 2008). This is particularly relevant given the significant under-detection of posttraumatic stress in patients diagnosed with psychosis (Lommen & Restifo, 2009), and that such patients are less likely to receive an appropriate clinical response even when abuse is disclosed (Agar & Read, 2002). As such, accurate assessment is imperative for providing recourse to suitable interventions, and devising treatment plans that encompass relevant cognitive, affective, and psychosocial factors. This may include, for example, ways in which posttraumatic beliefs impact on voice hearing attributions, strategies for coping with dissociation, and narrative techniques to aid memory contextualisation and integrate traumatic material (e.g., Corstens, Longden,
& May, 2012; Ogden, Minton, & Pain, 2006; Ross & Halpern, 2009).
These constructs were created in a systematic, standardised manner with voice-hearers who were broadly representative of a chronic psychiatric population in terms of clinical and demographic characteristics. Nevertheless this was a naturalistic, observational study derived from retrospective clinical data and situated within a broader therapeutic process. Performing a similar enquiry in a more controlled way would necessarily strengthen the reliability of the findings. A main limitation of this study was the self-selecting nature of the sample, and the fact that these individuals were motivated to engage in making sense of their voices may limit the generalisability of the results. An important avenue for future research is to therefore replicate the process with a random sample of voice-hearers in order to clarify indications and contra-indications. In this respect, although we
have received extremely positive anecdotal evidence from both voice-hearers and their family members and/or workers, it would be preferable to derive formal outcome data in order to elucidate the approach’s long-term outcomes in terms of treatment and recovery planning. Furthermore, the descriptive nature of the current research does not permit a more detailed, theoretically driven exploration of how voice representations can arise.
A further limitation is that the study did not permit sufficient exploration of the experiences of participants whose voices could not be formulated in psychosocial terms. For example: whether these links existed but the construct process failed to identify them; or whether voice-hearers were unwilling to disclose particular events (possibly due to lack of report with the researchers), or were not aware of them (i.e., as part of a posttraumatic, dissociative response: Moskowitz, Read, Farrelly, Rudegeair, & Williams, 2009). Some people are also disinclined to work within the confines of a narrative model, and cannot conceive themselves within the framework of a story (Woods, 2011). Another possibility is that these voices were unrelated to social/emotional conflicts; which in turn raises the question of whether they were aetiologically different to those of the other participants. In this respect, all information in the study was reliant on participant disclosure of adverse life events, which it was not possible to independently verify (although retrospective accounts of trauma amongst groups with complex mental health
problems have repeatedly proven sufficiently valid and reliable to justify the use of self-report measures: e.g., Fisher et al., 2011; Goodman et al., 1999; Meyer, Muenzenmaier, Cancienne, & Struening, 1996). In regards the training groups, we found that the atmosphere of acceptance, mutuality, and solidarity usually motivated people to tell their personal stories and share experiences. Generally voice-hearers who didn't want to disclose traumatic content would not volunteer in such a group, or would ask to meet with the trainers and create a construct privately. Nevertheless, it is likely that some voice-hearers didn't feel sufficiently safe to share painful stories, meaning that if this procedure influenced the results of this study, it would probably be an underestimation of associations between trauma and voices.
Our work with the 100 voice-hearers demonstrates that it is both possible and productive to engage in an exploration of traumas and vulnerabilities with individuals who are often strongly identified with a patient role (i.e., schizophrenia). These were people who had generally not responded well to traditional psychiatric treatments and were seeking help in finding a more personal explanation for their distress. These data show that a significant proportion of so-called chronic psychotic patients are responsive to an approach that acknowledges and relates to their socialemotional problems and, by extension, that it is possible to make sense and give meaning to voices; to relate them to overwhelming life experiences; and to understand voices as representations of the events and problems that underlie their emergence. By deconstructing diagnostic classifications and shifting clinical emphasis
towards psychosocial difficulties, we wish to formulate interventions which both provide opportunities to address past adversities in voice-hearers’ lives, and which promote more constructive, healing strategies to deal with the interpersonal and social dilemmas that they are confronted with. In this respect, it is also important to emphasise that such work utilises existing skills already employed by many professionals as part of good clinical practice and, as such, can be developed as the start of a road away from therapeutical nihilism towards personal recovery-oriented goals, wherein emotional vulnerabilities and posttraumatic responses can be addressed in a restorative way.
We wish to acknowledge all the voice-hearers and professionals who have worked with us, without whom it would have been impossible to devise the constructs. We would also like to thank Birgitte Bjerregaard Nielsen, Yolanda van den Broek, Christine Brown, Ron Coleman, Jacqui Dillon, Trevor Eyles, Inge Helle Jul, Matthew Morris, Mervyn Morris, Pernille Norgard Kolk, Melissa Lee, Karen Taylor and Rozemarijn van der Vinne who made it possible to conduct the workshops that enabled us to work with the voice-hearers.
1. We are using the phrase “voice hearing,” because it makes no assumptions about the pathological nature of a subjective experience (Thomas & Longden, in preparation). Although “auditory” or “verbal hallucinations” is a preferred term in psychiatric literature, this is not an expression that patients use to describe their experiences, and has likewise been deemed reductive and depreciatory by authors and service-user organisations (see Dillon & May, 2002; Intervoice, 2010; McCarthy-Jones, 2012)
Agar, K., & Read, J. (2002). What happens when people disclose sexual or physical abuse to staff at a community mental health centre? International Journal of Mental Health Nursing, 11, 70–79. Andrew, E., Gray, N., & Snowden, R. (2008). The relationship between trauma and beliefs about hearing voices: A study of psychiatric and non-psychiatric voice hearers. Psychological Medicine, 38, 1409–1417.
Beavan, V., & Read, J. (2010). Hearing voices and listening to what they say: The importance of voice content in understanding and working with distressing voices. Journal of Nervous and Mental Disease, 198, 201–205.
Bentall, R. (2006). Madness explained: Why we must reject the Kraeplinian paradigm and replace it with a “complaint-orientated” approach to understanding mental illness. Medical Hypothesis, 66, 220–233.
Bentall, R.P., Wickham, S., Shevlin, M., & Varese, F. (2012). Do specific early-life adversities lead to specific symptoms of psychosis? A study from the 2007 The Adult Psychiatric Morbidity Survey. Schizophrenia Bulletin. Advanced online publication. doi:10.1093/schbul/sbs049
Birchwood, M., Gilbert, P., Gilbert, J., Trower, P., Meaden, A., Hay, J., & Miles, J.N. (2004). Interpersonal and role-related schema influence the relationship with the dominant “voice” in schizophrenia: A comparison of three models. Psychological Medicine, 34, 1571–1580.
British Psychological Society Division of Clinical Psychology. (2011). Good practice guidelines on the use of psychological formulation. Leicester: The Author.
Corstens, D., Escher, S., & Romme, M. (2008). Accepting and working with voices: The Maastricht approach. In A. Moskowitz, I. Schafer, & M.J. Dorahy (Eds.), Psychosis, trauma and dissociation: Emerging perspectives on severe psychopathology (pp. 319–331). Oxford: Wiley-Blackwell.
Corstens, D., Longden, E., & May, R. (2012). Talking with voices: Exploring what is expressed by the voices people hear. PsychosisPsychological, Social and Integrative Approaches, 4, 95–104.
Daalman, K., Diederen, K.M.J., Derks, E.M., van Lutterveld, R., Kahn, R.S., & Sommer, I.E.C. (2012). Childhood trauma and auditory verbal hallucinations. Psychological Medicine. Advance online publication. doi:10.1017/S0033291712000761
Dillon, J. (2011). The personal is the political. In M. Rapley, J. Moncrieff, & J. Dillon (Eds.), De-Medicalizing misery: Psychiatry, psychology and the human condition (pp. 141–157). Eastbourne, UK: Palgrave Macmillan.
Dillon, J., & May, R. (2002). Reclaiming experience. Clinical Psychology, 17, 25–27.
Fisher, H.L., Craig, T.K., Fearon, P., Morgan, K., Dazzan, P., Lappin, J., & Morgan, C. (2011). Reliability and comparability of psychosis patients’ retrospective reports of childhood abuse. Schizophrenia Bulletin, 546–553.
Goodman, L., Thompson, K., Weinfurt, K., Corl, S., Acker, P., Mueser, K.T., & Rosenberg, S.D. (1999). Reliability of violent victimization and PTSD among men and women with serious mental illness. Journal of Traumatic Stress, 12, 587–599.
Hardy, A., Fowler, D., Freeman, D., Smith, B., Steel, C., Evans, J., & Dunn, G. (2005). Trauma and hallucinatory experience in psychosis. The Journal of Nervous and Mental Disease, 193, 501–507.
Intervoice. (2010, November 30). About Intervoice: The international network for training, education and research into hearing voices. Retrieved from November 30, 2010 http://www.intervoiceonline.org/about/about-intervoice
Johnstone, L., & Dallos, R. (2013). Formulation in psychology and psychotherapy: Making sense of people’s problems (2nd ed.). London: Brunner-Routledge.
Lommen, M.J.J., & Restifo, K. (2009). Trauma and posttraumatic stress disorder (PTSD) in patients with schizophrenia or schizoaffective disorder. Community Mental Health Journal, 45, 485–496.
Longden, E., Madill, A., & Waterman, M.G. (2012a). Dissociation, trauma, and the role of lived experience: Toward a new conceptualization of voice hearing. Psychological Bulletin, 138, 28–76.
Longden, E., Corstens, D., Escher, S., & Romme, M. (2012b). Voice hearing in biographical context: A model for formulating the relationship between voices and life history. Psychosis: Psychological, Social and Integrative Approaches, 4, 224–234. Mawson, A., Berry, K., Murray, C., & Hayward, M. (2011). Voice hearing within the context of the voice hearers’ social worlds: An interpretative phenomenological analysis. Psychology and Psychotherapy: Theory, Research and Practice, 84, 256–272.
McCarthy-Jones, S. (2011). Voices from the storm: A critical review of quantitative studies of auditory verbal hallucinations and childhood sexual abuse. Clinical Psychology Review, 31, 983–992.
McCarthy-Jones, S. (2012). Hearing voices: The histories, causes and meanings of auditory verbal hallucinations. Cambridge: Cambridge University Press.
Meyer, I., Muenzenmaier, K., Cancienne, J., & Struening, E. (1996). Reliability and validity of a measure of sexual and physical abuse histories among women with serious mental illness. Child Abuse and Neglect, 20, 213–219.
Moskowitz, A., Read, J., Farrelly, S., Rudegeair, T., & Williams, O. (2009). Are psychotic symptoms traumatic in origin and dissociative in kind? In P.F. Dell & J.A. O’Neill (Eds.), Dissociation and the dissociative disorders: DSM-V and beyond (pp. 521–533). New York, NY: Routledge.
Confederation, N.H.S. (2008). Briefing 162: Implementing national policy on violence and abuse. London: National Health Service.
Ogden, P., Minton, K., & Pain, C. (2006). Trauma and the body: A sensori-motor approach to psychotherapy. London: W.W. Norton.
Read, J., Agar, K., Argyle, N., & Aderhold, V. (2003). Sexual and physical abuse during childhood and adulthood as predictors of hallucinations, delusions and thought disorder. Psychology and Psychotherapy: Theory, Research Practice, 76(1), 1–23.
Read, J., van Os, J., Morrison, A., & Ross, C. (2005). Childhood trauma, psychosis and schizophrenia: A literature review with theoretical and clinical implications. Acta Psychiatrica Scandinavica, 112, 330–350.
Reiff, M., Castille, D.M., Muenzenmaier, K., & Link, B. (2011). Childhood abuse and the content of adult psychotic symptoms. Psychological Trauma: Theory, Research, Practice, and Policy. Advance online publication. doi:10.1037/a0024203.
Romme, M., & Escher, S. (2000). Making sense of voices. London: Mind.
Romme, M., Escher, S., Dillon, J., Corstens, D., & Morris, M. (Eds.). (2009). Living with voices: Fifty stories of recovery. Ross-on-Wye: PCCS.
Ross, C.R., & Halpern, N. (2009). Trauma model therapy: A treatment approach for trauma dissociation and complex comorbidity. Richardson, TX: Manitou Communications.
Shevlin, M., Murphy, J., Read, J., Mallett, J., Adamson, G., & Houston, J.E. (2011). Childhood adversity and hallucinations: A community-based study using the National Comorbidity Survey Replication. Social Psychiatry and Psychiatric Epidemiology, 46, 1203–1210.
Stainsby, M., Sapochnik, M., Bledin, K., & Mason, O.J. (2010). Are attitudes and beliefs about symptoms more important than symptom severity in recovery from psychosis? Psychosis: Psychological Social and Integrative Approaches, 2, 41–49.
Thomas, P. & Longden, E. (2013). Madness, childhood adversity and narrative psychiatry: caring and the moral imagination. Medical humanities. Advance online publication. doi:10.1136/medhum-2012-010268
Varese, F., Smeets, F., Drukker, M., Lieverse, R., Lataster, T., Viechtbauer, W., … Bentall, R. (2012). Childhood trauma increases the risk of psychosis: A meta-analysis of patient-control, prospective- and cross sectional cohort studies. Schizophrenia Bulletin, 38, 661–671.
Woods, A. (2011). The limits of narrative: Provocations for the medical humanities. Medical humanities, 37, 73–78.
Zubin, J., & Spring, B. (1977). Vulnerability: A new view of schizophrenia. Journal of Abnormal Psychology, 86, 103–126.