Intended for healthcare professionals

General Practice

Descriptive study of cooperative language in primary care consultations by male and female doctors

BMJ 1999; 318 doi: (Published 27 February 1999) Cite this as: BMJ 1999;318:576
  1. J R Skelton, senior lecturer (j.r.skelton{at},
  2. F D R Hobbs, professor.
  1. Department of Primary Care and General Practice, University of Birmingham, Edgbaston, Birmingham B15 2TT
  1. Correspondence to: Mr Skelton
  • Accepted 10 December 1998


Abstract Objective: To compare the use of some of the characteristics of male and female language by male and female primary care practitioners during consultations.

Design: Doctors' use of the language of dominance and support was explored by using concordancing software. Three areas were examined: mean number of words per consultation; relative frequency of question tags; and use of mitigated directives. The analysis of language associated with cooperative talk examines relevant words or phrases and their immediate context.

Subjects: 26 male and 14female doctors in general practice, in a total of 373consecutive consultations.

Setting: West Midlands.

Results: Doctors spoke significantly more words than patients, but the number of words spoken by male and female doctors did not differ significantly. Question tags were used far more frequently by doctors (P<0.001) than by patients or companions. Frequency of use was similar in male and female doctors, and the speech styles in consultation were similar.

Conclusions: These data show that male and female doctors use a speech style which is not gender specific, contrary to findings elsewhere; doctors consulted in an overtly non-directive, negotiated style, which is realised through suggestions and affective comments. This mode of communication is the core teaching of communication skills courses. These results suggest that men have more to learn to achieve competence as professional communicators.

Key messages

  • Standard teaching on medical communication promotes a cooperative approach to doctor-patient interaction

  • In everyday life, however, cooperative language is more typical of female speech style, and this suggests that male doctors may find it harder to develop appropriate consulting styles; in a sample of 373consultations, male and female general practitioners used examples of cooperative language equally

  • Language based study of doctor-patient interaction can deepen understanding and provide useful insights


The consultation process is at the heart of primary care and is largely realised through language. However, research into the communication that takes place in doctor-patient interaction in primary and secondary care settings has had little to say on language issues. Reviews of doctor-patient communication13are overwhelmingly weighted towards studies in which language features are either not analysed or analysed in a linguistically naive fashion.4 The broadly sociolinguistic tradition of language research is less well known.46 Furthermore, though tools of discourse analysis have revolutionised other areas of language study, 7 8 they have gained little credence within the health professions.9

We used an alternative approach to medical language research, combining quantitative and qualitative techniques, to consider whether male and female primary care practitioners use aspects of “cooperative talk” with similar frequency and in similar fashions in the consultation. Gender language is a complex field, beset with large numbers of contradictory studies. Nevertheless, a major critical review reported a broad consensus that female language in most situations is generally more cooperative.10 As a cooperative consulting style in medicine is viewed as the most appropriate style, there are implications for training if clinicians differ in the type of language they use.

A quantitative inquiry on this topic is unlikely to be sufficient for the understanding of communication, as dealing only with word counts deprives any study of language context. This study therefore also offers an examination of the contextual uses to which ostensibly cooperative language is put, by exploring aspects of what is known of the characteristics of male and female language in the English speaking Western world and comparing the use of some of these characteristics by male and female primary care practitioners during consultations.


Empirical studies of gender speech have concentrated on the language of dominance and support as represented through a wide variety of linguistic realisations.11 We selected three areas to give a balance in the study between the purely quantitative (and therefore reliable but decontextualised) and the qualitative (less reliable, but with context giving validity). These study outcomes were compared in male and female doctors.

(1) Mean number of words per consultation—One objective marker of dominance is verbosity. It is “established unambiguously”10 that men dominate mixed gender talk, even at work when women have superior status,12 and specifically with male patients consulting female doctors.13

(2) Relative frequency of question tags (such as don't you?and isn't it?)—Question tags are grammatical entities and therefore easy to identify with certainty. They may be divided into two main categories, those that are “affective” or “facilitative” (it's sore there, isn't it?), and those that are confirmatory (It's Mr X, isn't it?). We considered only the former, which have been claimed as part of women's language. 14 15This kind of tag can be seen as a marker of support.

(3) Use of “mitigated directives”—Roughly, mitigated directives are suggestions such as and then maybe you could try these…, as opposed to such “aggravated directives” as try these. It has been argued that women offer more mitigated directives and more general markers of tentativeness, specifically that women doctors do so and that this results in greater compliance. 16 17 “Directive” is a functional, rather than a formal, grammatical category, however, and quantitative claims should be treated with caution; this part of the present study is therefore primarily qualitative.


The study is based on the language used by 26male and 14female doctors, based in 17practices, in a total of 373consultations. All these doctors have some involvement in medical education; most have been general practice trainers or undergraduate tutors. Four male doctors were of Asian or mixed ethnic origin, but all were educated in the United Kingdom. The mean number of years since qualification was 15.2(range 5-34).

For each practitioner, all consultations during one standard surgery session were recorded on to a professional tape recorder. All patients consented to their consultations being recorded. A record was kept of patients' and doctors' age and sex. We used standard transcribing conventions,18 adapted in discussion with Cobuild plc, whose concordancing software was used for analysis. Recordings were transcribed by trained staff, with double entry of samples of data to ensure adequate levels of accuracy. All completed transcripts were checked against original tapes by the chief transcriber, and 10% of these data were in turn validated by JRS.

Concordancing analysis

Software programs for language analysis were originally developed for lexicography and subsequently for a wide variety of uses, including forensic linguistics and the study of scientific language.1921 The Cobuild concordancing program, used here, is the best known of these.22

The point of entry is a word or phrase, and the program lists all examples of this in the database, with its immediate context (figure). The length of context on screen may be varied from a few characters, as shown, to the complete text—for this study, a complete general practice consultation. More complex queries are also possible: for instance, the database could be asked for “all examples of either 'isn't it' or 'haven't you' spoken by a female doctor followed within eight words by a patient talking.”


Examples of “you could” for suggestion, spoken by doctors, in Cobuild concordancing program. <D>=doctor, <P>=patient, <C>=companion (for example, a mother with a young child)---these occur at the start of an utterance; <write3> is to be read as “doctor writes for 3 seconds”

The qualitative element of this study looked at the way in which language associated with cooperative talk is used in context. It derives from elements of grounded theory and critical discourse analysis. 23 24


Doctors spoke significantly more words than patients, but the number of words spoken by male and female doctors did not differ significantly (944v 909,respectively; difference 35,P=0.75) (table ). Affective question tags were used far more frequently by doctors than by patients or companions (399v 176; P<0.001). Isn't itis the most common question tag (254examples, of which 243were affective, and therefore relevant); the next most common was don't you, with 52examples (table 2). Frequency of use of question tags was similar in male and female doctors.

Table 1

Words spoken by participants in consultation

View this table:
Table 2

Use of facilitative phrases “isn't it” (243 affective uses) and “don't you” (52 affective uses) by male and female general practitioners in 373 consultations

View this table:

Qualitative results

There was no clear evidence that either male or female doctors used a greater number of mitigated directives, but this was in part because there was no clear distinction between mitigated and other directives. For example, if you don't see any good improvement with it then come back and I'll move you up a scale of sort of erm potency really, give you something a bit stronger. Is this a straightforward directive (it has an imperative), or a mitigated directive (the tentative if and sort ofsoften it), or simply a statement about what might happen? This kind of difficulty in interpretation of language context makes quantitative claims—that male doctors are dogmatic and female doctors cooperative in speech—dubious.

In particular, cooperative talk was associated with clusters, rather than single instances, of suggestions and question tags, and these in turn were further supported (as in the example above) by tentative markers such as if, sort of, I think, etc. A cluster occurred when one marker of cooperation was followed within 25 words by another—for example, I think that what you said about panic feelings I think that it's partially sort of anxiety and stress all the pressure that you're under. Thus, sort ofwithin 25words of sort ofoccurred 71times; if, sort of,and you know65times; and sort of and question tag 36 times. Such language tags are used to transmit not content but mood. Cooperative language is best regarded as a pervasive feature of consultations that are difficult for patients. For example, a woman recently had a child stillborn, and consulted because of vaginal discharge. The doctor raised the question of a further pregnancy, using clusters of cooperative and tentative language:

D: It's just important we get this cleared up before you you sort of you know try again isn't it you know

P: <cry> no <cry> they said it could have been handicapped anyway.

D: Well that's right I mean if you've got a if you've got a erm a situation like that it m[ay] may have been and so that's all right with your rational mind isn't it. It's just that it hurts deep inside

P: yeh

This difficult and upsetting consultation contained a total of 14instances of the doctor using sort of.

Moreover, such clusters are also common when doctors are giving details of management and advice in difficult areas—they are not merely empathic noise but are integral to the conduct of the consultation. This woman cannot have children: the doctor reminds her that her sister may become pregnant again:

P: She wants me to help cos she knows I'm good with children

D: Yeh

P: cos she's seen the way that erm I am with [name] and [name]

D: Well, we talked about this before erm I mean I think if you can accept the role as a loving aunt that would be a very big step you and I think I mean that's a very valuable role to play


These measures of cooperative language do not support the hypothesis that male and female doctors use different, gender specific speech styles in consultation. There is, however, an important qualitative characteristic which may enrich this conclusion. It seems clear that examples of cooperative language tend to occur in clusters, and that these clusters are not merely empathic noise but integral to the conduct of the consultation. These data suggest, nevertheless, that male and female doctors use a speech style that is not gender specific; contrary to findings in other studies, 12 13 the professional role of doctor seems to override the gendered characteristics of speech style.

There are limitations to the generalisability of these data. Though the 40doctors who participated in the study were volunteers and were involved in education, they were representative of practitioners in group teaching practices. The data may not be generalisable beyond an English speaking, Western context, and a different model may be used in the United Kingdom and elsewhere if there is a substantial non-Western population. Differences between men and women might surface if data were controlled for sex or social class of patient, or first presentations were differentiated from consultations for chronic conditions (where doctor and patient might have a less overtly professional relationship).


Communication is often thought of as a skill, as if doing it well were a matter of learning discrete pieces of surface behaviour. Our data show an atmosphere of empathy in these general practice consultations, created partly by clusters of cooperative language. But it is hard to see how this kind of clustering could be formally taught as a piece of behaviour to imitate: rather it is themanifestation of an appropriate attitude, which may be more difficult to teach and learn.

The language features we studied are surface representations of the overtly non-directive, negotiated style—which is realised through suggestions and affective comments—that is taught to medical students in communication skills courses. At least in these respects, medical students are being trained to use what linguists would describe as female speech style in their communication with patients.

Women's language, it has been said, shows a degree of “psychological androgyny”25: that is, women have the ability to approximate male speech styles when necessary—in a male dominated workplace, perhaps. There is also a widely held belief, in medicine and throughout the professions, that women have to “be like men” for career advancement. Our findings suggest that although women may have to adapt to the conventions of medical professional life, it is men who are required to adapt their usual communication style in the consultation. This in turn suggests that men—who have to acquire characteristics of speech that are thought of as representative of female talk—have further to travel than women to achieve competence as professional communicators.


Contributors: JRS and FDRH jointly developed the programme of research of which this is one project. JRS supervised data collection and transcription, and undertook the first analysis. JRS andFDRH developed the analysis and wrote the paper. JRS is guarantor for the paper.

Funding: Programme grant from Sir Siegmund Warburg Voluntary Settlement.

Competing interests: None declared.


View Abstract