7 Solución de problemas
7.1.1 Mensajes de advertencias y alarma
Key concepts
Julia Gillen
In his stimulating introduction to discourse analysis, Gee writes:
Any method always goes with a theory. Method and theory cannot be separated, despite the fact that methods are often taught as if they could stand alone. Any method of research is a way to investigate some particular domain . . . There can be no sensible method to study a domain, unless one also has a theory of what the domain is. (1999: 5) Discourse analysis is concerned with the investigation of language and one might reasonably expect linguis-tics, the discipline concerned with the study of language, to be the root for ‘discourse analysis’. It can be surprising for the student to find that this is not necessarily so, even if branches of linguistics some-times supply significant ‘tools of the trade’. In fact discourse analysis is a term used to embrace many different methods in the investigation of human communication found across the humanities and social sciences. At one end of a continuum one might put the algorithmic approach to language processing that informs computer software such as voice recog-nition and translation software. At another end one might put the poststructuralist disruptions of belief in any notion of ‘transparency’ in language. Yet the glorious characteristic of discourse analysis as it can be encountered from the hybrid interdisciplinary meadow of the social sciences is that any such notion of a continuum fails to work. Any attempt to draw such ends finds them circling back on themselves, like a snake coming to life and swinging round, suddenly
snapping at its tail. In this section I will indicate some strands of thinking in discourse analysis. From the chapter in its entirety you should gain a sense of how discourse analysis might be applied to your own research interests in the social sciences.
Theories, methods and thus disciplinary practices arise from their histories. The traditional linguistics paradigm, with roots in philology, grammar and philosophy, took as data language constructions intu-itively judged as correct by linguists (usually white, Anglo, middle-class men). Discourse analysis is im-mediately different, accepting as data any language as it occurs, whatever the channel or mode. For example, the repairs, hesitations and repetitions characteristic of spontaneous spoken language are approached seriously. Practitioners have examined everything from humour in the workplace to the semiotics of labels on jars of baby food. Another important distinction is that traditional mainstream linguistics works with the sentence as the largest unit of analysis whereas discourse analysis usually considers longer texts.
Of course there has been a relationship between many aspects of linguistic theory and the analysis of actual texts. Brown and Yule (1983) outline ways in which approaches emanating from linguistics can be drawn upon in discourse analysis. Sociolinguistics, the study of individual variation, is a notable field of endeavour. One influential theorist has been Michael Halliday (1985), proposing that all language has a dual function, communicating both ‘ideational’ meaning (regarding ideas and information) and ‘interpersonal’
meaning (furthering social relations in some sense with our interlocutor). In recent decades Mikhail Bakhtin’s emphasis on the dialogicality of language has taken hold. He argues that linguistic meaning
exists ‘neither in the system of language nor in the objective reality surrounding us’ (Bakhtin, 1986: 87).
A specific utterance should be understood as always responsive, in the broadest sense, each element of it being spoken (or written) under the influence of the speaker’s (or writer’s) previous experiences of the words themselves and the discourses in which they are embedded.
Theoretical developments are affected by changes in material practices; technology is always an import-ant influence in fields of interest, methods of data collection and, ultimately, insights generated. An important advance began in sociology at a time when prevailing trends were concerned with abstract, gen-eralizable explanations for human conduct located in identifications of social structures. The advent of the tape recorder, in the hands of fresh thinkers, brought about a new approach to the study of spoken language that continues to proliferate fruitfully today.
In the 1960s a little-known sociology lecturer struggled for recognition of his candidate PhD thesis even within his own faculty. Harvey Sacks, like Erving Goffman, appeared to be turning his back on the ‘big questions’ to focus on seemingly trivial tiny details of how we lead our daily lives. No detail is too small to lead to fascinating insights into human culture and the performance of identity – whether it be the exchange of ‘hellos’ on the telephone or the circumstances in which it is not taboo to ‘talk to oneself’.
As you read the previous sentence did the saying
‘first sign of madness’ come into your head? If so, why is there this association? Also, do you in fact ever talk to yourself: perhaps when driving or carrying out a task regarding fine coordination? Are you always alone or are there any circumstances when you might be ‘caught’
talking to yourself? Goffman argues that we avoid the imputation of madness if in certain circumstances we allow ourselves to be heard talking to ourselves but then ensure those witnesses hear us stop. In fact, there are a lot of sophisticated rules about talking to oneself;
if someone follows them carefully then that person will be judged as avoiding any suggestion of madness yet simultaneously demonstrating awareness of the taboo while communicating aspects of their feelings to others in what is effectively a performance of identity.
A vast amount of analysis has gone into such a commonplace feature of social talk as the exchange of greetings on the telephone. One of Sacks’s notable early exercises was an investigation of calls to a suicide prevention centre. What strategies did the telephone answerer use to endeavour, without directly
asking, to find out the caller’s name in order to establish the beginnings of a rapport? What strategies may be used in evasion? Here is perhaps the central concern of discourse analysis: to establish, or rather reveal, that in any communicative interaction we have not a single goal – to transmit a piece of information – but a multiplicity of concerns. The answerer, at a psychiatric emergency institution, wants to encourage the caller to talk, to remain calm and to give his (in this case) name – as this move towards intimacy makes sudden disengagement less likely. The caller
‘seeks help’ but is wary, unsure how much trust may be warranted. When we are engaged in any more everyday interaction, phoning a colleague, say, we too juggle a multiplicity of concerns, balancing them according to what we hear in return; our orientations may be revealed not just in analysis of the content of what we say, but in the split-second pauses and intakes of breath. The approach pioneered by Sacks (1992, published posthumously), and developed for example by Ten Have (1999) has become known as conversation analysis (CA). CA explicitly rejects the notion that you have to understand the context before you can approach texts.
In opposition are other strands of discourse analy-sis emerging from perspectives in social anthropology and indeed approaches within sociology that have embraced ethnography and an emphasis on reflexiv-ity. So, many practitioners of discourse analysis argue that the more one understands about the socio-historical situation of a text, the more sensitive and insightful will be one’s interpretation. Such an ideal has influenced much discourse analysis in educational settings, for example. In practice the boundary be-tween discourse analysis and qualitative methodolo-gies in general has sometimes become blurred where language data and the construction of meanings is the focus of intensive attention. In many empirical investigations centred on bounded discourse data – transcriptions of interactions, interviews and so on – software tools have been devised to assist in categor-ization and other tasks.
Analytical approaches endeavouring to bring in-sights together from discourse analysis to the study of authentic texts have also been given powerful new dimensions through the collection of computer cor-pora. These are vast databanks of written texts and/or transcripts – for example the International Corpus of English contains both spoken and written texts from Great Britain, East Africa and New Zealand (Meyer, 2002).
1 7 D I S C O U R S E A N A LY S I S
You might wonder, ‘what use is the study of such corpora if I am going to be working with my own texts?’ In fact, investigation of actual usages of a word or phrase, perhaps a key term in one’s research, can lead to some surprising insights. The British National Corpus of over 100 million words tells us that ‘man’ is more than twice as common as ‘woman’ yet the plural
‘women’ is more commonly used than ‘men’ (Leech et al., 2001: x). A few years ago, when engaged in a project investigating teachers’ perceptions of ‘continu-ing professional development’ I found that in the corpus overall this term had more often been applied to occupations such as architects, actuaries and lawyers than to teachers. Could the (then) relatively recent replacement of the term ‘INSET (in-service training)’
by ‘continuing professional development’ be an at-tempt to try to enhance by association the discursive power of the descriptive term for this activity?
Contemporary discourse analysis research is alive to an exciting multiplicity of influences. A significant motivation for many is to use linguistic analysis to unpeel layers of negative evaluation applied, often unconsciously, to the language of some speakers (and writers) positioned as low status or in some way vulnerable within a discourse setting. Hymes (1996), for example, demonstrates with the aid of eth-nopoetics the subtle patternings of a tale told by a young girl in a classroom that was disregarded in a setting ignorant of the cultural practices she was drawing on. Hymes and Gee (1999) are proving particularly influential, concerned as they are with ideologies as reflected by/constructed in discourses.
Their concern to use discourse analysis for critical purposes, particularly in respect of ‘public’ discourses, is shared by a group of contemporary scholars who term their approach critical discourse analysis (CDA) – Ruth Wodak and Norman Fairclough are probably the best known. (See Gotsbachner (2001) for an effective CDA-influenced small-scale study tracing how ‘symbolic representations in xenophobic dis-course . . . sneak under the threshold of awareness’ to disperse themselves in everyday talk — below you will find mention of his methods.) Conversation analysts are sometimes critical of critical discourse analysts for so explicitly bringing ideological positions into their approaches to data; CDA practitioners argue, I think with equal justice, that CA practitioners may be in danger of bringing too little reflexivity into their own approaches to data.
Discursive psychology is one of the most exciting arenas for discourse analysis today, combining incisive
textual analysis with a concern for the construction of identity and use of a broad range of theoretical understandings when probing how ideologies are made manifest. An inspiring example, drawing on feminism, poststructuralism and education, is Bron-wyn Davies’s (1989) study of the play of young children. Poststructuralist/Foucauldian notions of discourse at the same time flood and distil texts, causing us to see them as fluid, against ever-changing backcloths of the conditions in which they were created and are read. Often therefore poststructuralist insights take researchers away from a focus on an
‘authentic text’ viewed as the product of the individual in society towards a sense of language as one of many facets of discourse, and thus, arguably, beyond, or at least apart from, the domain of discourse analysis.
Implications for research design
There are, essentially, two main ways in which discourse analysis may be relevant to your research.
The first concerns your examination of data. Over-whelmingly the majority of research in the social sciences is likely to collect data some of which is in the form of texts that would benefit from close analysis. Barbara Johnstone (2002: 9) suggests that you might begin to construct a plan for such analysis from considering the following heuristic:
1 discourse is shaped by the world, and shapes the world;
2 discourse is shaped by language, and shapes language;
3 discourse is shaped by participants, and shapes participants;
4 discourse is shaped by prior discourse, and shapes the possibilities for future discourse;
5 discourse is shaped by its medium, and shapes the possibilities of its medium;
6 discourse is shaped by purpose, and shapes possible purposes.
I would suggest that in practice different approaches to discourse are operationalized by deciding, in the design, to restrict analysis, at least in the first instance, to a certain number of levels working from ‘6’ up. If your perspective at all times endeavours to operate at level 1 (embracing the rest) then it is likely that contemporary poststructuralism has the most to say to you and you should turn to the works of Foucault, Butler, Kristeva and others interested in language
through cultural theory (see Burke et al., 2000 for a marvellous introduction).
Conversation analysis, to take a contrasting ap-proach, willingly restricts itself to levels 6 and 5, in the process unravelling skeins of talk effectively.
In your practice, you might try first to examine in what ways the participants orient to the particular demands of the channel and any ‘ritualized con-straints’ associated with it – to use Goffman’s helpful (1981) expression. In what ways do the participants reveal their multiplicity of goals and how, if at all, do they acknowledge others’ and negotiate with them?
You might analyse features of grammar and their effect. One telling detail in the analysis by Gotsbach-ner (2001) as mentioned above is his examination of pronouns. Who is included in ‘we’ and referred to by
‘he’ or ‘they’? What characteristics are made or inferred in such generalizations? Another feature to look out for is the active vs. passive voice of verbs.
Norman Blake (1996: 30), a professor of English language at Sheffield, queries the effect if he were to choose to post a notice beginning, ‘I forbid students to . . .’ as opposed to ‘students are forbidden to . . .’
as generally preferred in this genre. The passive voice can be used to disguise agency and impute an authoritative air.
The second major way of using discourse analysis is to make use of resources others have collated in order to investigate an issue, concept or term that is central to your overall research question. For example, the social psychologists Antaki and Naji (1987), interested in the phenomenon of explanations, made an early corpus-based study to reveal that
‘general states of the world’ more often than ‘other people’s single actions’ were employed after the word
‘because’ in general conversations. This empirical finding had consequences for attribution theory.