To contents page

To Psychiatric Activism Page

Chapter 8

'Apparently a known schizophrenic':

From confession to surveillance

For the relations of words are in pairs first.

For the relations of words are according to their distance from the pair.

- Christopher Smart, Jubilate Agno, c1760(1)

In the previous chapter I analysed texts cast in the form of spontaneous confidences shared between troubled individuals and an interested listener. In this chapter the focus moves from such confessions to the other side of Foucault's disciplinary diagram - that of surveillance. At this level it is no longer the individual subject which is re-inscribed as truth by discourse under the guise of agentic and unpremeditated speech, but the individual subject created through a system of scientific visibility that enumerates, classifies and defines her in relation to other individuals.

The material I use comes from a psychiatric ward of a general hospital in the same city as Valhalla hospital, and consists of case records entered on a computer data base over a 5 year period in the late 1980s and early 1990s. Not only does this source represent an exemplar of the rapidly proliferating forms of textual data now available in electronic form, as discussed in Chapter 6, but it is also representative of the form of superpanopticism predicted by Poster (1990), in which the technical limitations of Bentham's panopticon are overcome and surveillance becomes truly omnipresent.

The hospital (which I shall refer to as 'Milfield') is a large general hospital in the greater Johannesburg area. I was given access to the records of the psychiatric ward for research purposes by the psychiatrist in charge of the ward, and for some months spent one day per week at the ward working on various research projects for the psychiatrist and some of his colleagues. Unlike at Valhalla, however, I did not attend ward rounds or conduct interviews with patients.

Milfield, like Valhalla, is a state hospital and an academic training facility located in a former white suburb in Gauteng. However, it is very much larger than Valhalla, and the psychiatric ward is a small part of the hospital. Many of the patients accepted into the ward are 'acute' cases, hospitalised in the wake of a suicide attempt, because they had caused a public disturbance, or had suddenly become unmanageable to their families. Patients were also occasionally transferred from other wards in the hospital when their physical ailments were found to be due to or accompanied by psychiatric difficulties. There is no segregation into different classes of mental illness as at Valhalla, and the emphasis is on relatively short-term crisis intervention followed either by discharge or by transfer to medium-term facilities such as Valhalla, private drug and alcohol rehabilitation centres, and the like. In some cases, particularly where there have been several previous admissions, patients are transferred to long-term institutions where they are involuntarily committed.

'Subjects and sampling'

As in the previous two chapters, there is limited utility in thinking of the material used in this chapter in terms of individual subjects or cases, since subjects are the product as much as the source of discourse. Some empirical data on subjects are nevertheless presented to help contextualise the linguistic material analysed. Two groups of individuals could qualify as subjects for the present study: The psychiatric registrars who composed the case histories and the patients about whom the histories were written.

Registrars. In all, 63 registrars contributed case histories, with a mean of 29.84 histories written by each registrar (SD=25.35) and a range of 1 to 129 histories per registrar. Unfortunately no further data are available on the registrars as the database was of course set up with the view that patients rather than medical personnel constitute cases to be studied. From my observations at Milfield it appeared that the registrars were a diverse group in terms of age, gender and cultural background, but that nearly all were white. A list of the registrars' surnames includes one Indian but no African surnames. During their training registrars were placed for periods of six months at a time at various hospitals and clinics forming part of the academic system in and around the city. Placement centres included Valhalla, Milfield, an inner-city outpatients' clinic, a township day clinic and the psychiatric ward at a large township hospital.

Patients. The case histories used for analysis refer to all 1883 psychiatric admissions at Milfield over a 5 year period in the late 1980s and early 1990s. The first 435 of these admissions were fully coded on the computer database, but after this it was apparently decided to code only certain variables. Descriptive statistics for both fully and partially coded biographical variables are presented in Table 8.1 and Table 8.2. These show that patients were typically somewhat older than those at Valhalla, typically spent only about two weeks in the hospital, and were overwhelmingly white. As at Valhalla, the majority were female.

Table 8.1 Age at admission and days spent in the psychiatric ward at Milfield Hospital


Variable N Mean SD Min Max


Age 1858 42.13 17.10 12 85

Days in ward 1859 14.31 16.36 1 241


Table 8.2 Gender and race distribution of patients in the psychiatric ward at Milfield Hospital


N %


Gender Female 247 58.53

Male 175 41.47

Total 422 100


Race White 416 95.63

Black 11 2.53

Coloured 4 0.92

Indian 4 0.92

Total 435 100


Of the first 422 patients in the database 33% (143) were coded as having had previous admissions, and 73.56% were recorded (Table 8.3) as having previously had psychiatric medication or electroconvulsive therapy (ECT) prescribed to them. The kinds of medications prescribed to patients are also listed in Table 8.3. No details regarding prior psychotherapeutic treatments were recorded.

Table 8.3 Somatic treatments given to patients in the psychiatric ward at Milfield Hospital prior to admission


Treatment N %


Benzodiazepines 254 58.39

Antidepressants 191 43.91

Antipsychotics 173 39.77

ECT 64 14.71


No somatic treatment 115 26.44

One type of treatment 94 21.61

Two types of treatment 126 28.97

Three types of treatment 64 14.70

Four types of treatment 36 8.28

Total 435 100


Analytic strategy

As in the previous chapter I started with a quantitative overview of the corpus, and moved from there to qualitative analysis. The quantitative techniques used are the same as for Chapter 7. Also as in the previous chapter, the qualitative analysis proceeded by first locating extracts containing collocational patterns identified from the lexical net, determining how these patterns function in the extracts and then presenting this interpretation together with illustrative extracts. The analysis in some cases continued on to issues thematically but not statistically related to the collocational patterns found.

Getting into the text

The case records in the computer database duplicated some of the information contained in patient files. These files are used at various points during a patient's stay at the hospital, for example by registrars prior to presenting the patient's case to a ward round, by psychotherapists for background information on the patient, to check on the patient's medication history where changes in medication are being considered, and when decisions are being made regarding home leave, transfer or discharge. In each of these cases existing information in the file is scanned and new data added. In cases of readmission old files are retrieved and placed inside the new file.

The most intensive period of activity around the file occurs shortly after admission when an official 'history' is taken from each patient (supplemented by 'collateral' information from family, friends and other medical staff), usually by a psychiatric registrar. This history includes a description of the events leading up to the patient's admission, information about her current symptoms, biographical information and various other psychiatric indispensibilities such as the results of a brief 'mental status examination'. In this analysis I focus in particular on an open ended section of each patient's record headed "EVALUATION/ INITIAL HISTORY: LEADING UP TO THIS ADMISSION" which contains a brief description of the presenting complaint, similar to the following (all case material in this chapter has been edited to ensure anonymity):

[121](2) 29 year old unmarried female, no children, staying with parents in X-city, presenting in a psychotic state with tactile hallucinations, sexual delusions and delusions of misinterpretation, delusions of influence, that started 1 m ago and got progressively worse over last week. Delusions are directed towards father. Father is transmitting his "lust" to her. Hypomanic features. Marked conduct disorder with substance abuse as child. 4 previous admissions - Weskoppies,

Magaliesoord, Phoenix House and Milfield. Defaulted meds.

While the ostensible purpose of a vignette such as this is informational, it is also, as discussed in Chapter 5, a linguistic ritual in which the beliefs and values of the medical world are reinscribed. For example, at least five epistemological strands, explanatory schemas, discourses, can be reconstructed from the short vignette quoted above:

1. The old maid discourse: Having violated the 'normal' developmental path for a woman (getting married and having children), the patient is now sexually frustrated and projects her frustration onto the nearest male figure.

2. The rotten apple discourse. Even as a child the patient was a delinquent who took drugs. Maybe she's fried her brains.

3. The noncompliant patient discourse. She stopped taking her medication. That's why she's gone off the rails again.

4. The schizophrenia discourse. She is deluded and has hallucinations. If I'm pressed for a diagnosis I'd say schizophrenia. But then on the other hand maybe it's mania.

5. The recidivist discourse. She's been here four times before. Are you surprised she's back?

While it is easy to imagine a similar analysis of 10 or even 100 such vignettes, it obviously becomes increasing difficult to sustain as the numbers increase. Given that the data available to me, which I shall call the initial history corpus, consisted of 1880 vignettes (68 939 tokens or 212 pages of single-spaced typescript), the idea of some form of preliminary quantitative parsing seemed even more appealing than for the interview transcripts used in the previous chapter.

Preliminary Quantitative overview

As an initial step the corpus was spell-checked and 783 spurious types removed. There were numerous uncommon types, such as medical terms (surmontil, ativan, emdalin), abbreviations (LOA, ICU, IQ, psych), numerical indicators of various sorts (1, 2, 3, 1st, 25mg, 50mg) and unusual constructions (RHT'd, function++, 4/5). I left these unchanged so as to preserve the tenor of the case histories. Individuals' first names and surnames were replaced with X.

The corpus consisted of 68 939 tokens and 5 916 types, giving a type-token ratio of 1:11.65, which is comparable to the 1:12.33 ratio found for the interview transcripts. However, the initial history corpus is approximately twice as long as the first interview corpus, and (as discussed in Chapter 6) vocabulary richness tends to be lower for longer texts. Taking only the first 33 644 tokens in the initial history corpus (the same length as the first interview corpus), 3753 types were found, giving a type-token ratio of 1:8.96. This higher vocabulary richness is probably in part due to the telegrammatic style of the case histories, which in many cases dispense with high frequency grammatical tokens such as a and the. Table 8.4, which lists the 60 most common types in the initial history text, confirms this interpretation, with lexical types such as history, patient and admitted among the most frequently used types, outranking even the, usually the most frequent type in written or spoken English.

The corpus contains 5768 lexical types and has a lexical density of 69.1%, almost twice that of the first interview corpus. This can again be ascribed to the abbreviated style used in the case histories and to the numerous unusual types mentioned at the start of this section.

Table 8.4 Sixty most common types in the initial history corpus


and 2194 of 2015 to 1589 with 1232

in 1127 for 977 history 866 patient 786

a 732 on 605 was 587 depression 585

year 521 has 499 at 489 admitted 460

had 437 from 432 her 430 2 401

old 398 by 398 no 379 the 366

she 364 years 360 admission 345 previous 345

not 323 suicidal 317 3 316 abuse 305

known 304 depressed 299 been 285 months 284

alcohol 279 he 279 after 268 poor 265

is 260 ago 256 weeks 241 suicide 237

as 235 family 231 very 231 OD 228

behaviour 226 paranoid 225 problems 223 features 220

his 220 disorder 219 treated 211 Valhalla 208

well 208 since 207 ward 202 also 191


Note. Frequencies are to the right of each word type.

Not surprisingly, the two most common lexical types in the initial history corpus (Table 8.5) are history and patient. The remaining types can tentatively be divided into a number of clusters. As with the interview corpus, depression and depression-related words (suicidal, depressed, suicide, OD) are particularly prominent, mirroring the frequency with which patients invoke depression to account for their hospitalisation. However, another finding from the interview corpus, that patients make some use of the discourse of social stress to explain their circumstances, at first glance appears to be absent from this corpus, with no clearly stress-related types included in the table. Types such as cope (N=34), stress (N=34) and stressful (N=88) do occur in the full list of types, but with relatively low frequency. Other kinds of reference to psychiatric signs, symptoms and diagnoses are however plentiful, e.g., paranoid, disorder, hallucinations, ideation, bipolar, psychotic, aggressive, vegetative, and manic, reflecting the more scientific and objective tenor of the histories.

Table 8.5 Sixty most common lexical types in the initial history corpus


history 866 patient 786 depression 585 year 521

admitted 460 2 401 old 398 years 360

previous 345 admission 345 suicidal 317 3 316

abuse 305 known 304 depressed 299 months 284

alcohol 279 poor 265 ago 256 weeks 241

suicide 237 family 231 OD 228 behaviour 226

paranoid 225 problems 223 features 220 disorder 219

treated 211 well 208 Valhalla 208 ward 202

hallucinations 187 delusions 177 ideation 177 admissions 173

bipolar 173 presented 172 psychotic 171 1 159

past 158 4 157 last 155 week 155

previously 154 referred 150 aggressive 147 prior 147

appetite 142 sleep 142 X 141 vegetative 136

6 134 month 133 hospital 132 manic 131

psych 130 treatment 130 auditory 126 home 125


Note. Frequencies are to the right of each word type.

The epistemological types prominent in the interview corpus are absent here, with the exception of the word known - which will be discussed later. In the place of these types there is an array of numerals (1 - 6), again indicative of the more unequivocal and objective genre within which the case histories operate. There are also many time indicators, such as year, years, previous, months, ago, weeks, past, week, previously, prior and month, marking the corpus as a set of psychiatric histories. A final group of words, which may be related to the time cluster, appear to refer to the process of entering psychiatric care - admitted, admission, Valhalla and admissions.

As previously discussed, frequency counts such as that in Table 8.5 provide an indication of the lexical content of texts, but are somewhat limited in their usefulness. As an example, one of the more prominent lexical items, abuse, is a case of homonymy and could (among other possibilities) refer either to substance/alcohol abuse perpetrated by the patient or to child/sexual abuse perpetrated upon the patient.

In order to move beyond such ambiguities, collocation counts were therefore again computed for each pair of types in the text (reproduced in Appendix 5), and a lexical net drawn using this data (Figure 8.1). As before, the z-scores on which the net is based were calculated using a span of 4 words on either side of each target word (truncated at vignette boundaries), but given the large size of the corpus, a larger minimum collocational frequency of 20 was used. The most prominent collocations were plotted down to a z-score of 18, at which point the resultant lexical net threatened to become too unwieldy.

Figure 8.1 Lexical net of the initial history corpus

'Presenting problems'

Reading the lexical net from the top left, one finds a sub-net of terms relating to gender and age (man, woman, lady, year, age and so on). This reflects a stereotypical opening sentence used in many of the histories, as in the following short extracts:

[4] 61 YEAR OLD Jewish divorced MAN


[11] 63 YEAR OLD widow for last 15 years

[720] A 39 YEAR OLD LADY WHO teaches at ... school

[898] 60 YEAR OLD prisoner WHO PRESENTED WITH dehydration

[900] 52 YEAR OLD woman with poor social circumstances

These simple biographical statements appear unremarkable, but clearly have more than an information-giving function. Basic data such as age and gender are strictly superfluous as they are prominently recorded in the patient file and the computer database anyway. Rather, one could argue that they are an effective opening gambit for a history of which it is required that it should appear parsimonious, objective and factual. There are of course variations in the degree to which objective scientific language (male vs man, female vs woman/lady) is considered necessary, and how much additional information is immediately introduced (a Jewish man, an unmarried woman), but what is ubiquitous is the need to signal from the outset that what is being dealt with here is a 'case' - i.e., an entity which is in principle knowable and susceptible to being summarised in a few lines.

Reading further downwards along the lexical net, it is evident that many of the vignettes proceed from an initial statement of biographical information to refer to the difficulty that the patient presented with. This 'presenting problem' can take many different forms:

[34]49 YEAR OLD LADY, divorced two YEARS AGO after 28 YEARS of marriage. PRESENTED WITH symptoms of adjustment disorder with depression.

[74] 54 YEAR OLD Jewish WOMAN WHO PRESENTED WITH depression, poor response to Ludiomil with side-effects and medical problems.

[1169] A 33 YEAR OLD MAN WHO PRESENTED WITH pseudoamnesia after having disappeared from his parents' home.

[1862] 19 YEAR OLD MALE PRESENTED mute WITH intermittent tearfulness.

As is the case with listing biographical information, describing a patient as 'presenting with' a set of problems, symptoms or a diagnosis draws on the idea that she or he can be objectively known as a case. The presenting problem is discursively marked as uncontroversial 'raw data', literally or metaphorically visible to the physician's eye, while at the same time it is signalled that these are preliminary observations and that further investigation may yield further data and more sophisticated interpretations.

In the first instance psychiatric patients thus appear in these vignettes as humanist critiques of psychiatry might have predicted they would - objectified, dehumanised and subjected to a regime of pseudo-scientific visibility. Where they are 'given voice', it is only in the context of professional scepticism, so that they are for example said to complain of various ailments (bottom left of Figure 8.1) rather than to be in any position to diagnose:

[912] COMPLAINS OF slurring, dry tongue, glassy eyes.

[1380] Self referred after several non-suicidal overdoses. COMPLAINS OF a loss of identity.

Other professional distance markers such as claims, denied, unreliable and the use of scare quotes also occur in the corpus, although not with sufficient frequency to be included in the net. Som typical examples:

[174] Has felt a "power" within her which makes her pray and perform religious rituals. Also claims that her husband is confusing her by saying strange things to her.

[798] Burnt all his belongings outside a hotel. Assessed as being psychotic in casualty. Unreliable historian ++. Query history of substance abuse.

[803] Patient denied recollection of interview.

The psychiatric mode of dealing with its subjects, as revealed once again in the opening lines of many of the vignettes, continues to be something of a scandal in the modern, humanised world. However, one should not therefore assume that psychiatric patients are somehow being denied their full individuality. The kind of power that psychiatry now holds (or more accurately - is both a product and a relay of) does not suppress, but fabricates and reinscribes the uniqueness of individuals. Just as the lexical net and similar devices produce the language of psychiatry as a particular kind of discourse, so that discourse is no more and no less than a grid for bringing into visibility the objects it produces - a never-ending series of individual subjects. In Foucault's words:

In a system of discipline (such as that of modernity) the child is more individualized than the adult, the patient more than the healthy man, the madman and the delinquent more than the normal and the non-delinquent. In each case, it is towards the first of these pairs that all the individualising mechanisms are turned in our civilization; and when one wishes to individualise the healthy, normal and law abiding adult, it is by asking him how much of the child he has in him, what secret madness lies within him, what fundamental crime he has dreamt of committing (1979, p. 193).

Histories of the present

The objects created by the discourse of psychiatry, are specifically psychiatric subjects, and part of psychiatry's discursive work in vignettes such as those analysed here is to claim them, with due reference to signs, symptoms and histories, as its own. As can be seen from Figure 8.1, symptoms, problems and diagnoses are most often not introduced directly after the presented/presenting/presents-with construction, but are accessed via the intervening key word history. Not only is this the most frequent lexical type in the corpus (Table 8.5), but it is also at the centre of a web of signification. Patients are portrayed as having a history of anxiety or depression (to the bottom of history), a family history of psychiatric illness (top right), a history of substance, cannabis or alcohol abuse (bottom left) or a history of previous suicide attempts and admissions (bottom right).

In some cases, vignettes run the full sequence from biographical variables and the presenting with construction, via history of to a diagnostic statement, as in the following extracts:


[12] 48 YEAR OLD white FEMALE WHO PRESENTED on 10th Feb WITH HISTORY OF ONE week's manic symptoms

[61] 24 YEAR OLD married WOMAN WHO PRESENTED WITH 6 MONTH HISTORY OF DEPRESSION and changes in her behaviour.


More typically, however, history of is used interchangeably (rather than in conjunction with) presenting with, and appears to have much the same function - to locate initial observations of the patient in the realm of objective fact. Thus patients are said to have a history of deteriorating function [106], depression [164], poor concentration [430], or substance abuse [432]. To further 'factualise' the situation, the time frame of the history is sometimes quantified, as in a one month or one week history (middle left of Figure 8.1). Patients are also sometimes said to have a strong history of one or the other kind:


[394] Past STRONG ALCOHOL HISTORY WITH antisocial personality.

[837] STRONG schizophrenic FAMILY HISTORY.

[976] STRONG HISTORY OF excessive intake of ALCOHOL.

[1424] Recurrent depression...STRONG FAMILY HISTORY.

The kind of warrant being offered here goes a little further than the 'neutral facts' placed on the table by means of biographical variables, the presenting with sequence and some forms of the history of sequence. What is being alluded to is mental illness as hereditary ([117], [837] and [1424] above) and as a recurrent chronic condition ([394] and [976]). Ironically, the strong history construction at the same time somewhat weakens the force of invoking historical data as it implies that history is not simply a matter of fact, but is open to evaluation.

Long, like strong, also functions to intensify history, but has a less obvious element of subjective evaluation and this may be why long history is a more popular construction (see Appendix 5). Long is used like strong to indicate chronicity ([93] Has had a LONG PSYCHIATRIC HISTORY [515]; [1019] LONG HISTORY OF CANNABIS ABUSE; [1603] Patient WITH a LONG HISTORY OF chronic depression), but interestingly is not once used with family to invoke hereditary factors. Family histories, it would appear are per definition long and need no further embellishment along that dimension. Instead, family histories are frequently elaborated on with reference to various diagnoses that family members have had, and as frequently the family is rather vaguely said to have a (positive) psych or psychiatric history, in which case further explanation appears to be optional:

[62] FAMILY PSYCH HISTORY - mother depressed.





[1124] POSITIVE FAMILY HISTORY - brother diagnosed as schizophrenic, currently functioning well.

It is also thought necessary in some cases to indicate that in fact there appears to be no such history ([957] No family psych history; [1111] No family psychiatric history; [1138] Family history nil), with occasionally comical results:

[1155] He denies a FAMILY HISTORY, SUBSTANCE ABUSE or homosexuality.

In a few cases long and strong history is used in tandem to help qualify the nature of the patient's problems:

[37] Patient PRESENTED WITH PARANOID DELUSIONS, concerning black men who wanted TO KILL her. At time of admission was agitated and reported AUDITORY HALLUCINATIONS. LONG prior PSYCHIATRIC HISTORY OF similar problem, STRONG HISTORY OF ALCOHOL ABUSE for 15 years, but had stopped drinking one month previously.

Various of the sub-nets in Figure 8.1 appear to operate in much the same way as a long/strong/family history, namely as a kind of mental crossing off of items on an imaginary checklist. An example is the 'social stressors' sub-net (top left):

[678] Multiple SOCIAL PROBLEMS.

[755] FEELING DEPRESSED for 11 months. FINANCIAL and MARITAL PROBLEMS. Treated with Eglonyl and Lexotan. No improvement.

Another example is the schizophrenic symptoms sub-net, which is connected to a suicide risk sub-net (bottom right):



[908] Recently broke up with girlfriend AND expressed SUICIDAL IDEATION.

[1396] Five day HISTORY OF paranoid ideation and VISUAL HALLUCINATIONS

Perhaps most remarkable for its extreme enmeshment is the sub-net near the top right of Figure 8.1, which can be thought of as a vegetative features net. If the initial history is like a pinball game, the vegetative features net represents an area where the ball bounces rapidly from side to side, setting lights flashing and bells ringing, before finally resuming a more linear, but not necessarily related, trajectory:

[726] Depression for about ONE MONTH after stepson moved into home. UNABLE TO handle situation. Feels neglected and ignored. Feels stepson is encroaching the previous family life pattern.

Doesn't see solution to problem of stepson. Also - increased

SLEEP pattern. No ENERGY or libido. Bulimic. No WEIGHT LOSS.

To the top right of the vegetative features sub-net is an unconnected sub-net relating to similar information, but couched in everyday language. These constructions (.e.g. not eating, not sleeping) are sometimes used interchangeably with the more formal terminology of the vegetative features sub-net, typically in the context of reporting on information supplied by a third party. The term vegetative features itself is not connected to the sub-net, but to the history of depression sequence (bottom left of Figure 8.1), indicating that a global reference to the presence or absence of vegetative features obviates the necessity of referring to individual signs and symptoms. The 'ordinary language' vegetative features sub-net also contains the only reference in Figure 8.1 to the discourse of coping, which is otherwise (as noted earlier) not prominent in the corpus.

By discursively calling on these various sub-nets a sufficient initial case can be made to explain a patient's admission into the psychiatric system. In theory each history should refer to each possible sub-net just as checklist would indicate the presence or absence of each possible complaint or symptom. However, in practice there appears to be a complex (and no doubt inconsistently applied) hierarchy in which for example mention of a 'strong family history' together with 'alcohol abuse' and 'financial problems' obviates the need for referring to the presence or absence of 'vegetative features'.

Histories of the past

History is not used to 'place' the individual as exhibiting a certain subset of psychiatric signs and symptoms, i.e., to account for the patient's illness, but also to give an account of the circumstances which led up to admission. Much of this has to do with the disturbed and disturbing behaviour for which hospitalisation is often seen as a (temporary) solution. In places this information still functions as a psychiatric checklist, but elsewhere it starts to operate as a narrative, as for example in the 'behavioural' sub-net at the bottom of Figure 8.1 :

[201] Chronic HISTORY OF BIZARRE BEHAVIOUR and functional decline.

[632] Change in personality according to husband, including LOSS of WEIGHT (15 kg last year). Apparently neglecting house, INAPPROPRIATE BEHAVIOUR.

[1251] AGGRESSIVE and BIZARRE BEHAVIOUR at a friend's house. Threatened 18 YEAR OLD girl with rape. Became abusive and stripped down to his underpants.

Attempted suicide (to the bottom right of history) and drug overdoses (top right) also sometimes function as checklist items (e.g., [995] Several SUICIDE ATTEMPTS in past. DEPRESSED AND SUICIDAL), but as with the behavioural sub-net a slightly richer history of events leading up to the hospitalisation start in many cases to emerge:

[412] OD previous evening at 8pm of 20 Dormicum tablets. Following minor MVA at 7pm. He had had 2 double brandies PRIOR TO MVA. Spur of moment decision to take OD, because of cost of repair to car, inability to take children on holiday until car repair. No suicide notes. PREVIOUS SUICIDE ATTEMPT 2 and a half YEARS AGO.

[1204] Referred to psychiatry by plastic surgeon following SUICIDE ATTEMPT by slashing his wrists and inhaling organophosphates. He was very delusional, blaming all problems on a pinched nerve in his neck for past 35 years.

Thus even though the grid of what is psychiatrically important and knowable (as reflected in the lexical net) reduces the variability in people's lived experience to a smaller subset of stereotypical possibilities, it at the same time causes individual trials and tribulations to stand out with even greater poignancy:

[888] Brought in to casualty by colleagues. Had gone into Soweto to martyr himself for the black people. Friends report gradual decline in functioning over last few months. NO PREVIOUS PSYCHIATRIC HISTORY.

[1540] COMPLAINS OF DEPRESSION and ANXIETY FOR 3 MONTHS. Had witnessed husband's murder in 1986. Never allowed herself to grieve. Bringing up 4 kids on own, living in poor downtown flat. Recently started recalling husband's death. Sometimes thinks she hears him calling her.

Familiar to the system

An important theme in a great many of the vignettes has to do with the path traced by each patient through the psychiatric system. This is reflected in the net in the sections to the right of history dealing with being admitted to Milfield Hospital, to (or via) a medical ward, to ward P (the psychiatric ward at Milfield Hospital) and having experienced an admission or admissions to Valhalla or Sterkfontein (a psychiatric institution for long-term patients). It is related (via the word admission) to a prior to admission sub-net which makes extensive use of quantified time constructions such as for 3 months and 2 weeks prior to. This sub-net is typically invoked to describe the patient's condition in the period immediately before the current admission ([449] HAD not BEEN talking FOR 2 MONTHS PRIOR TO ADMISSION) or ([1234] physically assaulted by 2 men FEW DAYS PRIOR TO ADMISSION). However, in most cases (multiple) prior/previous admissions and related constructs to the right of history are used not for purposes of story-telling, but to invoke a discourse of chronicity:

[166] 33 YEAR OLD unmarried young MAN WITH a LONG HISTORY OF PSYCHIATRIC problems. Previously admitted at MILFIELD HOSPITAL, Valhalla, Weskoppies, Sterkfontein. PRESENTED WITH unconvincing religious delusions.

[382] Two ADMISSIONS TO VALHALLA 1986/1988. WARD P - 1988. Admission X-Hospital 1987. DIAGNOSED AS a psychotic depression.

[1306] Previous OD 1 MONTH PRIOR ADMISSION, admitted X-Hospital. MULTIPLE ADMISSIONS VALHALLA, last in 1986 with depression.

There is a ritualistic quality about many aspects of the vignettes, but it is particularly evident in the almost ceremonial reference made to previous admissions, with phrases such as the following being very common:

[12] 8 PREVIOUS ADMISSIONS for Bipolar Affective Disorder

[35] MULTIPLE ADMISSIONS to nearly all hospitals in Witwatersrand, Cape Town, Pretoria etc.



These admission histories are typically very sparse, and seem to require little further elaboration. The idea that previous encounters with psychiatry explain her or his current status is quite central to making a case for why any particular person qualifies as a true patient. Reference to previous admissions frequently occur in the middle of vignettes where they appear to play a pivotal role in lending credibility to accounts:

[44] 38 YEAR OLD WOMAN WITH 3 WEEKS HISTORY OF manic symptoms WITH disinhibition, promiscuity and grandiose delusions. MULTIPLE PREVIOUS ADMISSIONS to other psychiatric hospitals and DIAGNOSED AS rapid cycler BIPOLAR DISORDER. SUBSTANCE ABUSE - poppers, CANNABIS, ALCOHOL. Stressors - bad relationship with boyfriend, moved to X, new job, financial, divorce, parents deaths, infertility.

[188] WELL KNOWN BIPOLAR lady admitted from Casualty in a manic state. Not compliant on meds. MULTIPLE ADMISSIONS TO VALHALLA and STERKFONTEIN since 1966. No stressors.

This emphasis on previous encounters with authority is reminiscent of Spencer's (1988) studies of probation officers' reports, in which previous criminal convictions (analogous to previous brushes with institutional psychiatry) are always carefully reported, although much else may be selectively left out.

Another form in which a previous psychiatric history is invoked is by means of the sub-net at the bottom-left of Figure 8.1. The constructions in this sub-net allow for reference to the role of other psychiatric staff earlier in the patient's career. Thus a patient may have been referred by or brought in by Dr X (all surnames were replaced with X), or may previously have been seen by him or her (another similar construction - under Dr X - occurs with some frequency in the corpus, but not sufficiently so as to be included in the net):

[28] HAS BEEN SEEN BY DR X regularly at OPD, treated for depression.

[242] Had phoned DR X twice and was REFERRED by him TO the HOSPITAL.

[650] REFERRED BY DR X after OD of hypnotics.

[1423] SEEN IN the past BY DR X AND DR X.

Interestingly the brought in sequence, although connected to the Dr X construction, does not refer to psychiatric staff, but to a range of other groups and individuals: the police, family, brothers, sisters, parents, work colleagues. Medical personnel 'see' patients and then 'refer' them, lay people simply 'bring them in'. Patients are also frequently (to the bottom of the sub-net) transferred from other hospitals and wards or discharged from such places before being admitted to the psychiatric ward at Milfield.

The only sub-net not yet discussed, at the top right of Figure 8.1 represents another method of accounting for the patient's current situation by positioning it as just another incident in a long psychiatric career. In terms of this sub-net (which is linked by sharing the word disorder to the commonly used phrase personality disorder), patients are said to have been previously diagnosed as having one or the other condition, to be a chronic schizophrenic or a known patient, known schizophrenic or known sufferer from bipolar affective disorder. Significantly, terms usually associated with schizophrenia such as delusions, hallucinations and inappropriate behaviour, are not statistically linked to schizophrenic, suggesting that if a person is already well known as a schizophrenic, further evidence regarding schizophrenic symptoms are considered superfluous, as in the following vignette:

[77] Patient REFERRED FROM X-Clinic, where he HAS BEEN for three and a half years - apparently a KNOWN SCHIZOPHRENIC. Sent to Milfield hospital for treatment and management of aggression which is not possible at X-Clinic.

By contrast, details of symptomatology are required where it is impossible to state categorically that the patient is already known to the medical authorities as a schizophrenic or some other category of mental illness:

[476] Sudden onset, AUDITORY HALLUCINATIONS, PARANOID DELUSIONS, DELUSIONS of control. Inappropriate, disinhibition, insomnia. According to husband she gets these episodes up to three times a year. Includes BIZARRE BEHAVIOUR and disorganised speech.

Again, the language is reminiscent of the criminal justice system, where offenders' accounts and those of their families have to be treated with studied scepticism, while previous convictions instantly qualify a person as a "known house breaker", a "well known child molester" and so on. The contrast between the scepticism accorded lay accounts on the one hand and the certainty provided by previous diagnosis on the other is illustrated in the phrases culled from randomly-selected vignettes in Table 8.6.

Table 8.6 Lay accounts versus previous diagnoses in the initial history text


Lay account Previous diagnosis


According to wife patient has had aggressive outbursts Well known bipolar disorder

According to wife, alcoholic habits in past Known alcoholic

According to parents patient has been totally irrational Well known bipolar patient

Change in personality according to husband Well known schizophrenic


The known patient sequence and related constructs also form an interesting contrast with the epistemological tangle at the centre of the lexical net of the interview corpus (Chapter 7). The uncertainties and equivocations of that net is here replaced with the fixed confidence of something that is already known.

Conclusion: The timeless trajectory

As a distillation of the initial history corpus, the lexical net in Figure 8.1 can be read in many different ways - as a map of the pathways open to the (recidivist) psychiatric patient, as a table of accounting practices used by psychiatric registrars to confirm the patient status of those they have been asked to treat, and so on. However, whatever reading is imposed on the net it is hard to escape the impression that what is shown is a peculiar kind of self-referential epistemology: A patient is a patient because s/he is a patient (has been previously treated/diagnosed/admitted) or because members of her/his family have previously attained patient status. What is revealed is a Kafkaesque world, governed by an insane circularity forever predicated upon itself, where admission explains readmission, where categories of madness are always already known, and where well-rehearsed symptom-checklists are recited over and over again as they ceaselessly confirm the actuality of the objects that are the effects of this psychiatric surveillance technology.

To present the case history as a kind of machine, much as one-on-one interviews were in the previous chapter said to be machine-like, is not however to claim that they are 'great satanic mills' which close upon the primordial innocence of psychiatric patients. Instead psychiatry labours, perhaps often 'in good faith', and always under difficult circumstances, to achieve what would count as a proper understanding of its subject:

The impossibility of getting precise information arises in most instances from the insuperable difficulties under which we are of knowing a person's character and history fully, intimately and exactly. We cannot go through the complex and often tangled web of his whole life, following the manifold changes of it, and seizing the single threads out of which its texture has been woven, unravel the pattern of it (Maudsley, 1899, p. 139).

Although Maudsley's terms are perhaps now somewhat archaic, it applies equally to the difficulties of 'writing up' a case history today. 'Character' (or what we might term 'individuality') is somehow immanent (although difficult to uncover) in the tangled web of epistemological labour, so that the quest for objective apprehension conceals the play of productive discipline at work as it manipulates 'character' as its object and effect.

1. Written while in Mr Potter's madhouse in Bethnal Green (quoted in Ingram, 1991, p. 171-172).

2. Numbers in square brackets refer to patient numbers in the database.

To contents page

To Psychiatric Activism Page