John McLeod has recently reminded us that 'almost
all counselling and psychotherapy research has been carried out from the
discipline of psychology' (1994,190). One consequence has been a focus on
quantitative studies concerned with the attributes of individuals. This
has meant that linguistic and sociological issues, such as language use
and social context, have been downplayed (see Heaton:1979).
Such a psychological focus has also had an impact on research design, leading
to the dominance of experimental and/or statistical methods favoured in
psychology. Of course, no research method is intrinsically better than any
other; everything will depend upon one's research objectives. So it is only
a question of restoring a balance between different ways of conceiving counselling
research.
In this light, perhaps it is time now, as McLeod suggests, to pose the following
strategic question about research on counselling:
'Can useful knowledge be best achieved by accurate, objective measurement
of variables or by respecting the complexity of everyday language' (1994,9)
In my research on HIV counselling, my use of tape-recordings of counselling
sessions was just such an attempt to 'respect the complexity of everyday
language' (see
Silverman:1997). As we shall see, by doing so,
I have been able to generate some unexpected findings. However, my style
of research also involved a particular answer to a further question raised
by McLeod:
'by what criteria are the validity of research findings to be judged?' (ibid).
We can expand (and explain) McLeod's question by relating it to the debate
in social science about the status of any claim to depict 'reality'. The
issue of validity is posed in terms of what constitutes a credible claim
to truth. Thus Denzin and Lincoln discuss the following four criteria applied
by 'conventional positivist social science' (including most psychological
studies of counselling) to 'disciplined inquiry':
'internal validity, the degree to which findings correctly map the phenomenon
in question; external validity, the degree to which findings can be generalized
to other settings similar to the one in which the study occurred; reliability,
the extent to which findings can be replicated, or reproduced, by another
inquirer; and objectivity, the extent to which findings are free from bias'
(Denzin and Lincoln:1994,100)
If we take seriously the demands of validity and reliability, there are
three obvious ways of evaluating counselling:
1. Measuring clients' response to counselling by means of research interviews
which elicit their knowledge and reported behaviour. This would involve
a longitudinal study, following a cohort of patients. The study could have
either an experimental or non-experimental design.
2. Measuring clients' response to counselling by means of objective behavioural
indicators. This also would involve a longitudinal study, following a cohort
of patients.
3. Measuring the degree of fit between actual counselling practice and certain
agreed normative standards of 'good counselling'.
Focussing on HIV/AIDS counselling, I review each strategy below. In doing
so, we will see that each raises both methodological and analytic questions.
I shall suggest that, in terms of either or both these questions, none of
these three strategies is entirely satisfactory.2
1. The Research Interview
As already noted, this might have either an experimental or non-experimental
design.
(a) In the experimental design, we might randomly assign clients to two
groups. In Group 1, clients are counselled, while in Group 2, the control
group, no counselling is provided. Both groups are then interviewed about
their knowledge of AIDS and how they intend to protect themselves against
the disease. This interview is followed up, some months later, with a further
interview examining their present behaviour compared to their reported behaviour
prior to the experiment.
(b) In a non-experimental design, existing counselling procedures are evaluated
by a cohort of patients. Again, we might follow up a cohort some time later.
The advantage of such research designs is that they permit large-scale studies
which generate apparently 'hard' data, seemingly based on unequivocal measures.
However, a number of difficulties present themselves.3 Let me list a few:
a. How seriously are we to take patients' accounts of their behaviour? Isn't
it likely that clients will tend to provide answers which they think the
counsellors and researchers will want to hear (see McLeod:1994,124-6)?
b. Doesn't study (1) ignore the organizational context in which health-care
is delivered (e.g. relations between physicians and other staff, tacit theories
of 'good counselling', resources available, staff turnover, etc)? Such contexts
may shape the nature and effectiveness of counselling in non-laboratory
situations.
c. Even if we can overcome the practical and ethical problems of not providing,
say, pre-test counselling to a control
group, may not the experience of being allocated to a control group affect
the reliability of our measures and the validity of our findings (see McLeod:1994,124).
d. Don't both studies treat subjects as "an aggregation of disparate
individuals" who have no social interaction with one another (Bryman:1988,p.39)?
As such, they give us little hold on how counselling is organized as a local,
step-by-step social process and, consequently, we may suspect that we are
little wiser about how counselling works in practice.
The non-experimental study seeks to identify qualitative research on counselling
with the attempt:
'to enter, in an empathic way, the lived experience of the person or group
being studied' (McLeod:1994,89)
This pursuit of 'lived experience' means that many qualitative researchers
favour the open-ended interview (see Silverman:1993,ix). Even when tape-recording
actual interactions is contemplated, attention is deflected away from analysing
the local organization of talk. For instance, while McLeod (1994) calls
for a study of 'the interior of therapy', he also cites favourably attempts
at 'interpersonal process recall' where participants are played back the
tape 'to restimulate the actual experience the person had during the session'
(1994,147). Thus, in common with many qualitative researchers, what matters
for McLeod is what people think and feel rather than what they do.
Both the 'in-depth' accounts apparently provided by the 'open-ended' interview
and the apparently unequivocal measures of information-retention, attitude
and behaviour that we obtain via laboratory or questionnaire methods have
a tenuous basis in what people may be saying and doing in their everyday
lives. Moreover, if our interest is in the relation of counselling to health-related
behaviour, do such studies tell us how people actually talk with professionals
and with each other as opposed to via responses to researchers' questions?
An example makes the point very well. At a recent meeting of social scientists
working on AIDS, much concern was expressed about the difficulty of recruiting
a sample of the population prepared to answer researchers' questions about
their sexual behaviour. As a result, it was suggested that a subsequent
meeting should be convened at which we could swap tips about how to recruit
such a sample.
Now, of course, this issue of recruiting a sample is basic to survey research.
And, for potentially 'delicate' matters, like the elicitation of accounts
of sexual behaviour, survey researchers are quite properly concerned about
finding willing respondents.
At the same time, it is generally acknowledged that the best chance of limiting
the spread of HIV may be by encouraging people to discuss their sexual practices
with their partners. This implies something about the limits of survey research.
Such interview-based research necessarily focusses on finding people prepared
to talk about their sexuality in an interview. However, it can say nothing
about how talk about sexuality is organized in 'naturally-occurring' environments
such as talk between partners or, indeed, talk about sexuality in the context
of real-time counselling interviews.
2. Behavioural Indicators
This method seeks to elicit behavioural measures which reliably report the
effectiveness of counselling. Its advantage is that, unlike the research
interview, it does not depend upon potentially unreliable client perceptions
and self-reports of behaviour and behavioural change. Moreover, by eliminating
a concern with the information that clients may acquire from counselling,
it takes on board the research that shows that acquired knowledge does not
have any direct link with behavioural change.
In relation to HIV-test counselling, it has been suggested that an appropriate
behavioural indicator is seroconversion.4 Presumably, then, we would need to study a cohort of patients
who test seronegative and are counselled. We would then re-test them after
a further period, say 12 months, to establish what proportions from different
counselling centres and with different counsellors have seroconverted. In
this way, it would be claimed, we could measure the effectiveness of counselling
in relation to promoting safer behaviour.
As already noted, the advantage of this approach is that it generates apparently
objective behavioural measures. However, like the research interview, its
reliability also has serious shortcomings:
a. How do we know that the counselling alone is the variable that has produced
the reported behaviour? Although we may be able to 'control' for some gross
intervening variables (like gender, age, sexual preference, drug use, etc.),
it is likely that some non-measured variables may be associated with the
reported behaviour (e.g. access to other sources of information, availability
of condoms or clean injecting equipment, etc.).
b. How do we know that an initial seronegative test means that the client
is not infected with the HIV-virus? The problem is created by the fact that
there is a 'window period' of up to six months during which the body does
not produce antibodies despite the presence of the virus. Consequently,
if any of the cohort has been exposed to risk of infection during that period,
any negative result would have a dubious status.5
c. A major problem for many counsellors in using behavioural measures is
that such measures do not fit with how the purposes of counselling are usually
defined. As we shall see, 'effective' counselling is often defined in terms
of criteria which are either only indirectly linked to behavioural change
or relate to behavioural changes other than particular physical outcomes.
Hence lack of seroconversion could not be treated as a reliable measure
of effective counselling.
3. Meeting Normative Standards
To many counsellors, the only sensible means of evaluating their work is
to start from their own objectives. These objectives will reflect their
normative theories of 'good practice'.
Like sociology, counselling has been a site for many competing theoretical
orientations (see McLeod:142-164). The post-war dominance of 'client-centred'
theories (Rogers:1957) now seems to have been displaced by 'therapeutic
alliance' theory (Horvath and Greenberg:1986) and theories deriving from
the quasi-sociological 'systemic' approach of the Milan School (Selvini-Palazzoli
et al:1980).
In relation to our data on HIV-counselling, two main normative approaches
were found. In one British centre (see Perakyla:1995) and, to a lesser extent,
in the U.S. centres, counsellors used some variant of the 'systemic' approach.
Indeed, in this British centre, some of the practitioners were authors of
major texts which explained how the approach could be used in HIV-counselling
(Miller & Bor:1988; Bor,Miller & Goldman:1992). In practice, this
commitment to theory-based counselling was seen, among other things, in
the use of 'circular' questions, often based on hypothetical situations
and by the avoidance or delay of direct advice-giving (see Perakyla:1995).
An example of pre-HIV test counselling at this Centre is given in Extract
1 below. Transcript symbols are given in Appendix Three. The counsellor
(C) has just asked her male patient (P) how he would feel about a positive
HIV-test result:
Extract 1 (UK4) [SS/2/25 Tape 1]
P: found out I was [posi- positive I was going die.= C: [Ri:ght. P: =Because I wouldn't lie to (them) because I heard about er oh a thing .h where: i- uh: some Frenchman (0.4) er left a note to-to- to his boyfriend the following morning that .hhh er: wu- welcome to the: to the ai:ds uh:m (0.5) virus. =An[d this chap went ba:ck .h and found= C: [Right. P: =that this chap had infected him. C: Right. P: Now I think that's nasty. That is really is er: C: a So you think you'd probably be fairly k- (.) keen to try and (.) not pass it on to other [people. P: [Oh yeah definite[ly. I couldn't- I couldn't live- C: b [( ) d'you think perhaps (.) if- if- what would happen if an urge came on d'you think possibly you might be able to P: We::[ll C: [(be safe) or something (.) [to try and get= P: [Yeah. C: =rid of the urge. (.) P: I'd try t[o but- C: c [And is that the situa:tion? P: ( ) (0.5) P: Those kinds- those sort of situations it's- it's very har[d to know what you're going= C: [Mm P: =actually do:. (0.7) At that moment in time.= C: =Right. (1.0) P: I could say one thing and maybe (.) I could be- it could be the (.) complete reverse type of thing. C: d Sure. .hh I mean uh say- say if it comes back negative a:nd e (1.1) P: I'd pr[obably C: [You haven't been re-tested so we don't know:: P: Y[eah C: [what's going to happen:, do you think that'll change anything in your relationships with other people. =Do you think you'd carry o:n having sex with them: without protection or:? P: e I think I'll still be carrying o:n (0.5) er sex without protection. (0.5) u-Because I- I (.) because I believe that you can take more ( ) and test and they maybe in three or five- five months or (the same).
Notice here, after P completes his answer (line
12), C offers a candidate summary of what P has been saying (a).
This is immediately confirmed by P (lines 15-16). However, rather than changing
topic or giving further information, C now asks P a hypothetical question
about what P might do (shown as b), using a term ('urge') supplied
earlier by P (data not shown). When P's answers now indicates hesitancy
about his earlier position, C requests more specification (shown as c).
P's answer now casts further doubt on his earlier position and C asks an
additional hypothetical question, requiring P to assume that his test turned
out to be negative (beginning at d). P's answer to this question
(shown at e) confirms that he would be unlikely to engage in safer
sex whatever the result of his HIV-test.
In the other British centres, where counselling was mainly done by nurses
in sexually-transmitted diseases clinics, a primarily 'medical model' dominated.
This meant that, after a brief 'history-taking', the counsellor spent most
of her time delivering information to the client with the avowed aim of
ensuring that the HIV-test only took place with the client's informed consent.
Extract 2 below shows such a counsellor giving advice to a female patient
about safer sex after a negative HIVnegative test result:
Extract 2 (UK1) [SW2-A]
C: .hhhh Now when someo:ne er is tested (.)
and they ha:ve a negative test result .hh
it's obviously ideal uh:m that (.) they
then look after themselves to prevent [any
P: [Mm hm
C: further risk of infection. .hhhh I mean obviously
this is only possible up to a point because if .hhh
you get into a sort of serious relationship
with someone that's long ter:m .hh you can't
obviously continue to use condoms forever. .hh
Uh:m and a point has to come where you make a
sort of decision (0.4) uh:m if you are settling
down about families and things that you know
(0.6) you'd- not to continue safer sex.
[.hhhh Uh:m but obviously: (1.0) you=
P: [Mm:
C: =nee:d to be (.) uh:m (.) take precautions
uhm (0.3) and keep to the safer practices
.hhh if: obviously you want to prevent
infection in the future.
P: [Mm hm
C: [.hhhh The problem at the moment is we've
got it here in {names City} in particular
(.) right across the boar:d you know from
all walks of life.
P: Mm hm
C: Uh::m from you know (.) the sort of
established high r- risk groups (.)
now we're getting heterosexual (.)
[transmission as well. .hh Uhm=
P: [Mm hm
C: =so obviously everyone really needs
to careful. .hhh Now whe- when someone
gets a positive test result er: then
obviously they're going to ke- think
very carefully about things.
Let me make three observations about Extract 2.
First, unlike Extract 1, C delivers advice to P. Second, this advice is
offered without having elicited from P a perceived problem. Reasons of space
do not allow me to include what immediately precedes this extract but it
involves another topic (the meaning of a positive test result) and no attempt
is made to question P about her possible response to this topic i.e. how
she might change her behaviour after a negative test result. Third, C does
not personalise her advice. Instead of using a personal pronoun or the patient's
name, she refers to 'someone' and 'they' (lines 1-4) and 'everyone' (line
28).6
In principle, it should be possible to evaluate HIV-counselling, like that
shown in Extracts 1 and 2, according to how far it meets the normative standards
supported by counsellors. So we might evaluate any given counselling interview,
as appropriate, by whether it followed a 'systemic' method or whether it
properly achieved 'informed consent'.
For instance, in Extract 2, we might note that P only produces minimal uptake
in the form of various utterances of 'mm hmm'. While these may indicate
that P is listening, they do not mark what C is saying as advice. Hence,
at the very least, they do not show P uptake and may also be taken as a
sign of passive resistance (see Heritage & Sefi:1992). Conversely, in
Extract 1, by using hypothetical questioning, C seems to get P to think
through his likely future behaviour and to revise his initial responses.
Although C will later deliver advice (data not shown), it is likely to be
well-grounded in P's own perspective.
We seem on the way here to be developing a firmly-based evaluation of these
extracts. Unfortunately, in practice, it has proved difficult to establish
reliable measures to evaluate counselling. As McLeod (1994,150-164) has
noted, several problems have confronted attempts at evaluation by means
of normative standards. Namely:
a. Observers find it difficult to differentiate different measures (e.g.
'congruence', 'empathy' and 'acceptance') and instead rate counsellors according
to 'their image of a "good therapist"' (McLeod:1994,150).
b. Ad hoc decisions are often made about which part of a counselling interview
should be assessed. The scope extends from a whole counselling interview
(or even several interviews with the same client) down to a micro-segment
of one interview. The latter, narrow approach may gain in precision
but at the cost of understanding the context of the surrounding talk. The
broader approach may find it hard to make detailed assessments of whole
interviews.
c. Even if such measures are reliable and precise, the result 'assesses
only the presence or absence of a mode, and not the skilfulness with which
it is delivered' (ibid,151).
Such problems in attempts to use internal, 'normative' standards of evaluation
look even worse when viewed in the context of studies which seek to relate
such measures to particular outcomes. As McLeod (ibid) notes, one such study
(Hill et al:1988) found that only 1% of variance in client responses were
related to observed measures of counsellor behaviour!
However, the problems we have found in each of these three attempts to evaluate
counselling need not lead to the abandonment of the project. Apart from
anything else, any professional practice, including counselling, in principle
stands to gain from a sympathetic dialogue with social science researchers.
One way out of the impasse is to stand back a little from methodological
details and to examine the assumptions about social research from which
our three methods of evaluating counselling may derive.
I have suggested elsewhere (Silverman:1994) that
there are two potentially dangerous orthodoxies shared by many social scientists
and by policy-makers who commission social research like the three evaluative
studies we have been considering. The first orthodoxy is that people are
puppets of social structures. According to this model, what people do is
defined by 'society'. In practice, this reduces to explaining people's behaviour
as the outcome of certain 'face-sheet' variables (like social class, gender
or ethnicity).
We will call this the Explanatory Orthodoxy. According to it, social scientists
do research to provide explanations of given problems, e.g. why do individuals
engage in unsafe sex? Inevitably, such research will find explanations based
on one or more 'face-sheet' variables.
The second orthodoxy is that people are 'dopes'. Interview respondents'
knowledge is assumed to be imperfect, indeed they may even lie to us. In
the same way, practitioners (like doctors or counsellors) are assumed always
to depart from normative standards of good practice. This is the Divine
Orthodoxy. It makes the social scientist into the philosopher-king (or queen)
who can always see through people's claims and know better than they do.
What is wrong with these two orthodoxies? The Explanatory Orthodoxy is so
concerned to rush to an explanation that it fails to ask serious questions
about what it is explaining.
There is a parallel here with what we must now call a 'post-modern' phenomenon.
It seems that visitors to the Grand Canyon in Arizona are now freed from
the messy business of exploring the Canyon itself. Instead, they can now
spend an enlightening hour or so in an multi-media 'experience' which gives
them all the thrills in a pre-digested way. Then they can be on their way,
secure in the knowledge that they have 'done' the Grand Canyon.
This example is part of something far larger. In contemporary culture, the
environment around phenomena has become more important than the phenomenon
itself. So people are more interested in the lives of movie stars than in
the movies themselves. Equally, on sporting occasions, pre- and post-match
interviews become as exciting (or even more exciting) than the game itself.
Using a phrase to which we shall shortly return, in both cases, the phenomenon
escapes.
This is precisely what the Explanatory Orthodoxy encourages. Because we
rush to offer explanations of all kinds of social phenomena, we rarely spend
enough time trying to understand how the phenomenon works. So, for instance,
we may simply impose an 'operational definition' of 'unsafe sex' or a normative
version of 'good counselling', failing totally to examine how such activities
come to have meaning in what people are actually doing in everyday (naturally-occurring)
situations.
This directly leads to the folly of the Divine Orthodoxy. Its methods preclude
seeing the good sense of what people are doing or understanding their skills
in local contexts. It prefers interviews where people are forced to answer
questions that never arise in their day to day life. Because it rarely looks
at this life, it condemns people to fail without understanding that we are
all cleverer than we can say in so many words. Even when it examines what
people are actually doing, the Divine Orthodoxy measures their activities
by some idealised normative standards, like 'good communication'. So, once
again, like ordinary people, practitioners are condemned to fail.
Both kinds of research are fundamentally concerned with the environment
around the phenomenon rather than the phenomenon
itself. In quantitative studies of 'objective' social structures and qualitative
studies of people's 'subjective' orientations, we may be deflected away
from the phenomenon towards what follows and precedes it (causes and consequences
in the 'objective' approach) or to how people respond to it (the 'subjective'
approach). This can be illustrated in two simple diagrams:
Figure 1: 'Objectivism'
causes > the phenomenon > consequences
Figure 2: 'Subjectivism'
perceptions > the phenomenon > responses
In both approaches, the phenomenon with which ostensibly we are concerned
disappears. In 'Objectivism', it is defined out of existence (by fiat as
Cicourel:1964 calls it). On the other hand, 'Subjectivism' is so attached
to authentic notions of 'experience' that it loses all sight of the socially
organized phenomena to which such experience refers.
At this point, the reader might rightly have two
questions. First, how does this, increasingly abstract, discussion of orthodoxies
in social research relate to the more concrete issues of evaluating counselling
with which we began? Second, since destruction is always easier than construction,
what alternative, if any, is proposed?
Both questions will be answered below. However, it is probably easiest to
attempt to begin with my answer to the first question. In this way, we will
tidy up the 'destructive' part of this chapter in order to be able to concentrate
better on its 'constructive' argument.
In brief, my suggestion, is that counselling research based on measuring
clients' response can fall prey to the Explanatory Orthodoxy, through a
focus on the consequences of counselling. On the other hand, using normative
standards to assess counselling practice derives from the Divine Orthodoxy
because it has a pre-defined sense of the character of counsellors skills
(and of their likely deficiencies).
In all three of the approaches that we considered earlier, the search for
reliable measures of outcome (approaches 1 and 2) or counsellors' behaviour
(approach 3) may create fresh, seemingly intractable methodological difficulties.
Thus, as we have seen, outcome measures may be contaminated by the artificial
situation of the research interview or made problematic by their uncertain
relationship to counsellors own aims. Equally, trying to apply normatively-derived
measures of 'good' counselling practice may result in problematic tabulations
where the skills of counsellors in using a particular method cannot be evaluated.
More to the point, in terms of Figures 1 and 2 above, all three approaches
risk missing the phenomenon of counselling itself. The first approach, using
research interviews, follows what we have called 'Subjectivism'. Its focus
on clients' perception of counselling and response to it gets us no closer
to understanding what has happened in the counselling interview itself.
The second approach, using objective behavioural indicators, reflects an
'Objectivist' methodology. Its concern with the consequences of counselling
does not allow us to comprehend which features of the counselling interview
may be associated with such consequences.
Finally, our third approach has the merit of focussing on actual counselling
practice. But, to some extent, the 'phenomenon' still 'escapes' because
pre-defined measures cannot do justice to the skills in situ that counsellors
deploy.
These assumptions are summarised in Table 1 below:
Table 1: Counselling Research: Measures and Assumptions
| Measure | Orthodoxy | The 'Phenomenon' |
| Interviewing clients | Explanatory | Escapes: only looks at consequenses |
| Normative Standards | Divine | Escapes: predefined version of skills |
It is now time to lay my cards on the table and
to offer the alternative approach on which my research is based -conversation
analysis (henceforth CA). CA, as we shall see, is centrally concerned with
the organization of talk, although its concern with social organization
leads it to describe its subject-matter as 'talk-in-interaction'.
Equally, counsellors, by definition, treat talk as a non-trivial matter.
However, even if we concede the centrality of talk to social life, why should
counselling researchers give priority to recording and transcribing talk?
Given the usefulness of other kinds of data derived, say, from observations
of behavioural change or interviews with clients, what is the special value
of transcripts of tape-recordings of conversation?
One way to start to answer this question is to think about how research
based upon data which arises in subjects' day-to-day activities can seek
to preserve the 'phenomenon' of interactions like counselling interviews.
Although such 'naturally-occurring' data is never uncontaminated (for instance,
it may need to be recorded and transcribed), it usually gives us a very
good clue about what participants usually do outside a research setting.
Conversely, in research interviews, as Heritage puts it:
'the verbal formulations of subjects are treated as an appropriate substitute
for the observation of actual behaviour' (Heritage:1984,236). The temptation
is then to treat respondents' formulations as reflections of some pre-existing
social or psychological world.
However, if we follow this temptation in counselling research, then we deny
something that all counsellors recognize: namely, that talk is itself an
activity. Although this is recognised in many normative versions of counselling,
to base our research on such versions would be to narrow our focus to those
activities which we already know about it.
An alternative is to investigate how counselling interviews actually proceed
without being shackled by normative standards of 'good' communication. In
this way, we might discover previously unnoticed skills of both counsellors
and clients as well as the communicational 'functions' of apparently 'dysfunctional'
counsellor behaviour.
Detailed transcripts of conversation overcome the tendency of transcribers
to 'tidy up' the 'messy' features of natural conversation. Sacks, Schegloff
and Jefferson (1974) offer an Appendix which provides a detailed description
of the notation they use and the interested reader is recommended to study
it. An alternative source is Atkinson and Heritage (1984).
However, it should not be assumed that the preparation of transcripts is
simply a technical detail prior to the main business of the analysis. As
Atkinson and Heritage (1982) point out, the production and use of transcripts
are essentially 'research activities'. They involve close, repeated listenings
to recordings which often reveal previously unnoted recurring features of
the organisation of talk. The convenience of transcripts for presentational
purposes is no more than an added bonus.
Heritage (1984,237) has noted the gains of working with tape-recordings
and transcripts. His observations can be summarised as follows:
1 It is very difficult for an observer working with field notes to record
such detail.
2 The tape-recording and the transcript allows both analyst and reader to
return to the extract either to develop the analysis or to check it out
in detail.
3 What may appear, at first hearing, to be interactionally 'obvious' can
subsequently (via a transcript) be seen to based on precise mechanisms skilfully
used by the participants.
It is worth concluding with Heritage's summary of the advantages of transcripts:
'the use of recorded data is an essential corrective to the limitations
of intuition and recollection. In enabling repeated and detailed examination
of the events of interaction, the use of recordings extends the range and
precision of the observations which can be made. It permits other researchers
to have direct access to the data about which claims are being made, thuis
making analysis subject to detailed public scrutiny and helping to minimise
the influence of personal preconceptions or analytical biases. Finally,
it may be noted that because the data are available in 'raw' form, they
can be re-used in a variety of investigations and can be re-examined in
the context of new findings' (Heritage:1984,238).
This is not the place to formulate the main principles of CA (but see Sacks,
Schegloff and Jefferson:1974 and
Heritage:1984). Suffice it to say that CA uses the practices found in ordinary
conversation as a base-line from which to analyse institutional talk. It
can then examine how particular sequence types found in conversation 'become
specialised, simplified, reduced, or otherwise structurally adapted for
institutional purposes' (Maynard and Clayman:1991,407).
Viewed in this light, one minimal way to describe counselling interviews
is to examine their modified use of certain properties of everyday conversation.
A basic sequence of actions in a recognisable interview is a series of questions
and answers (Silverman:1973). After a question, as Sacks puts it, 'the other
party properly speaks, and properly offers an answer to the question and
says no more than that' (Sacks 1972: 230). However, after the answer has
been given, the questioner can speak again and can choose to ask a further
question. This chaining rule can provide 'for the occurrence of an indefinitely
long conversation of the form Q-A-Q-A-Q-A . . .' (ibid.).
Although question-answer sequences do arise in mundane conversation, they
seem to provide a defining characteristic of many counselling interviews,
as we saw in our discussion of Extract 1. The chaining rule gives a great
deal of space to
the counsellor to shape the flow of topics, while most
clients' questions are positioned after the counsellor has invited them
to ask a question.7
The kind of detailed research on counselling that
CA calls for lays itself open to the charge that it deals 'only' with talk.
The implication is that, because it supposedly refuses to look beyond the
talk, it is unable to offer adequate explanations of its findings.
Of course, I have already offered a critical review of this approach in
our comments on the Explanatory Orthodoxy. Nonetheless, I do not want to
suggest that it is always improper to look beyond talk-in-interaction. Instead,
my position is that we are not faced with either/or choices but with issues
largely of timing.
My assumption is that it is usually necessary to refuse to allow our research
topics to be defined in terms of, say, the 'causes' of 'bad' counselling
or the 'consequences of 'good' counselling. Such topics merely reflect the
conceptions of 'social problems' as recognized by either professional or
community groups. Ironically, by beginning from a clearly defined analytical
perspective, we show how we can later address such social problems with,
we believe, considerable force and persuasiveness.
My argument suggests that one's initial move should be to give close attention
to how participants locally produce contexts for their interaction. By beginning
with this question of 'how', we can then fruitfully move on to 'why' questions
about institutional and cultural constraints. Such constraints reveal the
functions of apparently irrational practices and help us to understand the
possibilities and limits of attempts at social reform.
My aim has been to demonstrate a way of working with data which seeks to
preserve the local production of social phenomena. Take the issue of training
counsellors. The implication is that effective training begins from a close
analysis of the skills of counsellors and their clients revealed in careful
research rather than from normative standards of good practice.
Consider Extracts 1 and 2 discussed earlier. We noted then that the information-driven
agenda found in Extract 2 generated minimal patient uptake. Predictably,
the use of hypothetical-questions and summaries by the counsellor in Extract
1 produced considerable patient response.
Advice-sequences like those found in Extract 2 are very common at three
out of the five Centres examined here. So we have to ask ourselves why counsellors
should use a format which is likely to generate so little patient uptake.
However, since my preference is not to criticise professionals but to understand
the logic of their work, we need to look at the functions as well as the
dysfunctions of this way of proceeding.
The first thing we might note is that topic follows topic with a remarkable
degree of smoothness and at great speed. This might indicate one function
of this style of counselling. With more conventional counselling, based
on extended question-answer sequences like that seen in Extract 1, counselling
interviews average between 40-45 minutes. This compares to the clinic from
which Extract 2 is taken, where pre-test counselling interviews average
been 10-15 minutes. Truncated, non-personalised advice sequences are usually
far shorter - an important consideration for hard-pressed counsellors.
Second, we can observe that, in Extract 2, C avoids referring directly to
P but uses the non-specific term 'someone'. As we argue in Chapter 6 below,
this may allow P to hear what C is saying as information rather than advice.
This has three neat functions for C. First, it allows her to manage possibly
'delicate' topics by formulating them as 'general' information matters not
necessarily relevant to this particular patient. Second, unlike a question-driven
style of counselling, which can involve very complicated strategies to encourage
clients to talk about 'delicate' topics (see Perakyla:1995), in the information-delivery
mode, very little client participation is needed for the introduction and
management of delicate topics. This also mean, thirdly, that counsellors
need not treat P's minimal responses as indicating resistance as they would
if C were offering personalised advice.
Lest readers throw up their hands at my apparent 'support' for what may
appear to them to be, in normative terms, 'bad' counselling practice, let
me speedily point out that I am more than aware of the 'dysfunctions' of
the professional-centred style of counselling seen in Extract 2. Not only
is there no exploration of the patient's perspective but P's minimal responses
mean that C never knows how far P is aligned to her information. This lack
of patient-uptake fails to create an environment in which people might re-examine
their own sexual behaviour (as P does in Extract 1).
Equally, however, apparently 'good' counselling practice, based on extensive
elicitation of clients' perspectives is fraught with interactional traps
(see Perakyla:1995). The point is not to adopt a normative position but
to examine the gains and losses of any method.
Two possible solutions suggest themselves from the data analysed by this
study (see Silverman, Bor, Miller and Goldman:1992). First, avoiding necessarily
'delicate' and unstable advice sequences but encouraging patients to draw
their own conclusions from a particular line of questioning (see Extract
6, Chapter 5). Second, since both this method and step-by-step advice-giving
take considerable time, finding ways of making more time available for more
effective counselling.
Hence I would argue that the question of 'effective' counselling and 'effective'
counsellors can only be addressed in the context of the management of the
interactional and practical constraints on counselling practice. There are
no simple normative solutions to these constraints - although our experience
running workshops for such counsellors suggests the value of these kind
of detailed transcripts in in-service training provision. We take up the
practical issues this raises in a further discussion of 'policy' issues
at the end of this book.
My findings (see Silverman:1996) arose because
of the research methods used. My use of CA led to a focus on the sequential
organization of talk. It also created an embargo on appeals to what participants
are thinking and, instead, the pursuit of what they are doing.
CA has been used for two reasons. First, because it is appropriate to an
understanding of the complexity of our data. Second, because CA has demonstrably
established a fruitful dialogue with practitioners in a range of work settings
(see Drew and Heritage:1992). In this study, CA has encouraged us to look
at the local functions of people's activities in a way which I believe has
a direct bearing upon policy issues.
Of course, this does not mean that other research approaches are a priori
pointless. CA is time-consuming and expensive and cannot directly deliver
the 'outcome measures' beloved by funding agencies. So, in many cases a
more conventional study, based say on interviews or behavioural indicators
may deliver the goods to the mutual satisfaction of all parties.
As I noted at the start of this chapter, I share the belief that there is
no right or wrong method in science, any more than there is an essentially
right or wrong method of counselling (see Feltham:1995). At some level,
everything depends on what you are trying to do.
My aim, then, has not been to argue that more conventional research designs
are entirely 'wrong' or to win converts to a new orthodoxy. I will be entirely
satisfied, therefore, if this treatment of methodological issues maintains
the ongoing debate about appropriate research practices.
Atkinson, J M and Heritage, J (1984) (eds) Structures
of Social Action Cambridge: Cambridge University Press.
Bor, R, Miller, R and Goldman, E (1992) Theory and Practice of HIV Counselling:a
systemic approach, London:Cassell
Bryman, A. (1988) Quantity and Quality in Social Research, London:Unwin
Hyman
Cicourel, A (1964) Method and Measurement in Sociology, New York:Free Press
Drew P and Heritage, J (1992) (editors) Talk at Work, Cambridge:Cambridge
University Press
Feltham, C (1995) What is Counselling?, London:Sage
Heaton, J M (1979) 'Theory in Psychotherapy'. In N Bolton (ed), Philosophical
Problems in Psychology, London:Methuen, 179-198
Heritage, J (1984) Garfinkel and Ethnomethodology, Cambridge:Polity
Heritage, J and Sefi, S (1992) 'Dilemmas of Advice: Aspects of the Delivery
and Reception of Advice in Interactions between Health Visitors and First
Time Mothers'. In P.Drew and J.Heritage (eds.), op.cit.
Horvath, A O and Greenberg, L S (1986) 'Development of the Working Appliance
Inventory'. In L Greenberg and W Pinshof (eds) The Psychotherapeutic Process:
A Research Handbook, New York:Guilford Press, 529-556
McLeod, J (1994) Doing Counselling Research, London:Sage
Miller, R and Bor, R (1988) AIDS - A Guide to Clinical Counselling, Science
Press:London
Perakyla, A (1995) AIDS Counselling: institutional interaction and clinical
practice, Cambridge, Cambridge University Press
Rogers, C R (1957) 'The Necessary and Sufficient Conditions of Therapeutic
Personality Change', Journal of Consulting psychology,21,95-103
Sacks, H. (1972) 'On the analysability of stories by children'. In J Gumperz
and D Hymes (eds), Directions in Sociolinguistics, New York:Holt, Rinehart
& Winston
Sacks, H, Schegloff, E and Jefferson, G (1974) 'A simplest systematics for
the organization of turn-taking in conversation', Language, 50 (4), 696-735
Selvini-Palazzoli, M, Boscolo, L, Cecchin, G et al (1980), 'Hypothesizing,
Circularity,Neutrality:three guidelines for the conductor of a session',
Family Process,19,3-12
Silverman, D (1973) 'Interview Talk: Bringing Off a Research Instrument',
Sociology, 7, 1, 31-48
Silverman, D (1993) Interpreting Qualitative Data:Methods for Analysing
Talk, Text and Interaction, London:Sage
Silverman, D (1994) 'Analyzing Naturally-occurring Data on AIDS Counselling:
some methodological and practical issues'. In M Boulton, editor, Challenge
and Innovation: Methodological Advances in Social Research on HIV/AIDS,
London:Falmer Press
Silverman, D (1997) Discourses of Counselling:HIV counselling as social
interaction, London:Sage
Silverman, D, Bor, R, Miller, R and Goldman, E (1992) 'Advice-Giving and
Advice-Reception in AIDS Counselling. In P Aggleton, P Davies and G Hart
(Eds), AIDS:Rights, Risk
and Reason, London:Falmer Press
1. This is an edited version of Chapter Two of
my book Discourses of Counselling (Silverman, 1997). Order
information.
2. Of course, such an argument may be read
quite polemically. Where this happens, it is a regrettable consequence of
the need to telescope quite complicated issues within a few paragraphs.
However, I hope it will also be clear to the reader that I do not set out
with the belief that there is only one 'right' way of proceeding in either
social science or counselling. Instead, I entirely agree with the conventional
wisdom that, in such matters, everything depends upon what you are trying
to do.
3. Of course, I recognise that these problems
are recognised by researchers who use such research instruments. In turn,
they have ingenious methods for dealing with them.
4. This suggestion was made to me by a senior
physician at an AIDS unit in Sweden.
5. Anssi Perakyla has suggested that the problem
caused by the 'window period' might possibly be overcome by assembling samples
from each centre large enough to carry more or less equal numbers of people
who carry the virus despite having tested seronegative.
6. Extract 2 is discussed at length in Chapter
Five of Silverman (1997).
7. I have discussed the positioning of interviewees'
questions in an early paper on selection interviews (Silverman:1973). However,
as Perakyla (personal correspondence) has pointed out, Sacks' chaining rule
is a too general way to speak about the role of counsellors as questioners.
In fact, counsellors maintain their role as questioners in subtle, locally
variable ways, rather than primarily relying on a general rule (see Perakyla:1995).