Lecture / Seminar 8: The "other half" of CA - categories
Antaki C and Widdicombe, S (1998) Identities In Talk. London: Sage. chapter 1
Hutchby and Wooffitt, rather late in their book (p
213 and after), give a very good account of the sort of analyses that
one can do with the concepts that I shall outline here, but don't lay
the groundwork out as simply as I would like, so use the Antaki and
Widdicombe for preference.
So far I've only talked about conversational structures and the
way people use them to do things with each other.
But there is another part of CA, which is equally insightful into
what talk-in-interaction does. Indeed, it was one of Harvey Sacks's
own starting points in his analyses, and he never gave it up; but
after his death, his co-workers focussed on what has now become
'core' CA, the things we have seen so far about sequential organisation.
This other half of CA is the study of how people use
categories in their talk. That is, how their choice of ways of
describing themselves and others, and what they and others do,
carries with it significant implications.
The basic idea is that the words we use to describe things bring
with them a very heavy set of implications - implications that go a
long way beyond the dictionary, and a long way beyond 'ordinary'
pragmatics. Of course, words in use always have implications
beyond the dictionary; we've seen plenty of evidence for that
already. But what I'm going to talk about in this lecture is a
special kind of implication: namely, an implication about about the
units out of which society is structured. That is a very abstract way
of putting it, so let us turn to an example.
The baby cried...
Sacks' simple example illustrates the theme well. He found the
following two sentences as the opening of a story written by a child:
The baby cried. The mommy picked it up.
There is nothing in logic or pragmatics to tell us that the mommy is
the baby's mommy; still less that she picked up her baby because it
cried. Yet these are two judgements that we make automatically.
This simple observation revealed a very powerful engine that
people used in their talk. The childish author of those two sentences
had (correctly) assumed that her or his readers would unconsciously
make the inferential leap between the words and the social
arrangements they represented. And any other speaker or writer can do
the same. It is obviously economical; you don't have to spell things
out. But it is also subtle, and can go unnoticed. That's why it is so
fertile for analysis.
Sacks invented some terms to help sort out the concepts involved. To
my mind the terms are a bit ugly, but they serve their purpose.
Membership categorisation device
This is the most important one. Words (and other things too, which
I'll come to later) can work as 'devices' that force a set of
otherwise random objects into a 'category' with 'members'.
In the child's story, above, the effect was achieved by the word
'mommy'. That forced the implication that the two people in the story
were joined in a set.
Another example. (An invented one, just for
Suppose I read out a list of two people: Anna, 34, and Ivan, 50.
The very fact that I have put them into a list strongly (but hardly
logically) implies that they have some relationship. But what
relationship? As soon as I say 'Anna, the doctor, is telling Ivan...'
you have fixed them into a framework of doctor and patient. Like
family-member names, job descriptions are among society's most
powerful devices for organising how we see people.
In fact it is so powerful that we need to remember that these
descriptions are always a matter of choice. If I say 'Anna, the
doctor, is telling Ivan...' then the doctor-patient relationship seems the natural
and perhaps even exhaustive one. For all current purposes, speaker
and hearers are seeing Anna interacting with Ivan as doctor to
patient. All other descriptions are off. Indeed it would be an effort
to suddenly talk about (say) Ivan as a Christian. If you did, the
strong assumption would be that this was somehow relevant to his
now-established membership of a doctor-patient category (perhaps they
are talking about euthanasia, or contraception, or something else
where religion and medicine overlap).
This is Sacks' point. What you call them affects hugely influences
the implications your listeners can draw about what sort of scene
they are acting in, and therefore what the rules are of the other
actors, and what sort of plot they are in.
Try some other alternatives. If you identified Anna as 'the
Canadian" then you would be using the Membership Categorisation
Device (MCD) of 'nationality', and implying that Ivan was not
Canadian, and that their nationalities mattered at this point, and
you're ready for a story along those lines.
If you identified Anna as 'white', you would be invoking the MCD of
'race', and suggesting that Ivan was non-white, and you would be
ready for race to be significant. They are always the same people,
and all of those descriptions might be equally true. But their
consequences would be quite different. So the message is: speakers
can cast the people they're talking about (or themselves, of course)
as members of a category with implications for how to see the
other participants in the scene.
Another pedantic expression, but again useful to crystallise the
point that emerges when we think about what happens when we set Anna
and Ivan into a category. We've made the point that each category (be
it 'family', 'medical consultation') has its set of members
(mother-child, doctor-patient and so on). But there is more to it
than that. Each member has a set of behaviour, feelings, rights and
obligations that go along with the role. Mothers are nurturing,
babies are helpless, doctors are expert, patients are in need of
So when you say the words 'Anna, the doctor, is telling Ivan....'
then the listener is going to assume that she is telling him the sort
of things that doctors tell patients, and that he is going to receive
it as a patient as patients receive what doctors say as doctors.
This is a normative assumption (we came across that word at
the start of the lectures on CA). That is, if the speaker has said Anna is a
doctor, that's how people will understand what she is reported to say.
They would think it very odd if the speaker then protested that they never meant
her to be understood that way, and 'just mentioned' that she was a doctor.
Real examples of using category membership devices
Let's move back onto firmer territory with some real examples. We've already seen the way the word "mother" can signal a family
relation with another character in a story. Here is another example,
this time from spoken conversation.
From the Holt corpus Holt:2:3, transcription simplified
|one three five?
|Oh hello it's um Leslie Field he:re:
|Hello, .tch I hope you don't mind me getting in touch
|but uh- we metchor husband little while ago
|at a Liberal meeting.
When introducing oneself to a person one doesn't know well, one
has to choose some description that makes sense of calling, and makes
sense of what the call will be about. Leslie's choice of 'we met your
husband at a Liberal meeting' (the Liberals are a political party in Britain) set up two categories. The "we" sets
her up as a member of some team, presumably husband and wife
(suggested by calling the person they met 'your husband'). The other
device is the one of 'political affiliation', set up through the
explicit nomination of a Liberal meeting.
Leslie need have said neither of these things. The fact that she
did, means that Mary is to understand Leslie as calling on those two bases, and gives Mary a sense of what basis she herself is now
expected to speak - a member of her own husband-and-wife team, and
someone with Liberal sentiments. How these are relevant to the call,
we don't know yet, but they set up a footing for it.
descriptions can be crucial in getting yourself a proper footing in the
interaction when you have very little time - as in the case of calling the
emergency services, for example. If you can establish a 'legitimate' identity
then do so quickly, as this caller does:
From Zimmerman (1998) (CT=
call-taker, C: = caller)
|tch .hh u::h This is u::h Knights of Columbus
|Hall at uh: twenty twenty ni:ne West Broadway
|=Mmhm ((keyboard sounds))
|U::h we had some u::h women's purses u::h stolen
Saying you are "Knights of Columbus Hall" gives you an
institutional identity, and institutions talking to institutions are 'serious'
and 'professional'. You are not 'just anybody' or 'some idiot' etcetera.
Moreover, giving your address without prompting shows that you know the routine,
that you are a co-operative partner in the Call-taker's business. Both those
things together help cement a proper footing for the call to proceed smoothly
(next lecture we'll see an example where it doesn't).
Examples of "trouble"
Claiming a category for yourself is usually trouble-free, but not always. Hutchby and Wooffitt give some useful examples from Wooffitt &
Widdicombe's work on "goth" and "punk" identities. They went to rock
festivals and wandered around the crowd, asking people if they would
care to be interviewed. here are some interesting exchanges:
From Hutchby and Wooffitt, 1998 p 179
|how would you descri:be (.) yourself
|and your appearance and so on
|describe my appearance.
|su- su- slightly longer than average hair
|((goes on to describe appearance))
TroubleNotice the pause at line 3, and R's
choice of a 'clarificatory' question, rather than an answer, at line
4. This is, as we know, a 'dispreferred' response to a question
(which usually expects an answer). Notice also that the R in
repeating the question, edits it; the bit about 'describe yourself'
is dropped and what remains is 'describe your appearance".
The way that Wooffitt and Widdicombe analysed this was to see that
the R was orienting to the implications of answering the question as
put by the interviewer. The R could have self-categorised themselves
as "a Punk" or "a Goth", but this would mean that they would then be
responsible for a whole load of category-implications which they
might not be comfortable with. So a neat move is not to answer the
question in a legitimate way, by asking for a clarification;
and to exercise control over that clarification and steer it into
safer territory - the R's appearance, rather than their
Some general principles.
If we are going to look at identity through the eyes of Conversation Analysis and the idea that people can ascribe
themselves (and each other) to
various categories, some principles will help. I take these from Antaki and
Widdicombe (1998), chapter 1.
"...a person's identity is their display of, or ascription to,
membership of some social category, with consequences for the interaction in
which the display or ascription takes place. Which category, or combination of
categories, and which of the characteristics it affords are matters of
changeable arrangements made locally. Membership of a category is ascribed (and
rejected), avowed (and disavowed) and displayed (and ignored) in local places
and at certain times, and it does these things as part of the interactional work
that constitutes people's lives.
In other words...[it is] not that people passively or silently have
this or that identity, but that they work up and work to this or that identity, for themselves and
others, there and then, either as an end in itself or towards some other end.
If this working-up and working-to of identity happens in interaction, the
argument continues, then the best tools to examine it will be those appropriate
to the medium of interactional business, namely, talk."
That's a very general way of putting it. Here are five more specific
principles, which I'll talk a little about in this lecture, and try to
illustrate in the next one.
- For a person to 'have an identity' - whether he or she is the person
speaking, being spoken to, or being spoken about - is to be cast into a
category with associated characteristics or features (the sort of
thing you'd expect from any member of that category; their actions, beliefs,
feelings, obligations, etcetera)
- Such casting is indexical and occasioned. That is, it only makes
sense in its local setting.
- The casting makes relevant the identity to the interactional
business going on.
- The force of 'having an identity' is its consequentiality in the
interaction - what it allows, prompts or discourages participants to do next
- All these things are visible in people's exploitation of the structures
Again, I've taken these from Antaki and Widdicombe, Chapter 1. More in the
We'll have a look at some data to see if we can find more examples
of membership categorisation devices, and what people do with them.
We'll go on to think about the relationship between 'category
membership' and 'identity'. Carry on with reading Chapter 1 of Antaki
and Widdicombe, and try the chapter by Zimmerman, although it may be