Introducing Discourse Analysis for Qualitative Research

Qualitative research often focuses on what people say: be that in interviews, focus-groups, diaries, social media or documents. Qualitative researchers often try to understand the world by listening to how people talk, but it can be really revealing to look at not just what people say, but how. Essentially this is the how discourse analysis (DA) can be used to examine qualitative data.

Discourse is the complete system by which people communicate, it’s the widest interpretation of what we call ‘language’. It includes both written, verbal and non-verbal communication, as well as the wider social concepts that underpin what language means, and how it changes. For example, it can be revealing to look at how some people use a particular word, or terms from a particular local dialect. This can show their upbringing and life history, or influences from other people and workplace culture.

It can also be interesting to look at non-verbal communication: people’s facial expressions and hand movements are an important part of the context of what people say. But language is also a dynamic part of culture, and the meanings behind terms change over time. How we understand terms like ‘fake news’ or ‘immigration’ or ‘freedom’ tells us a lot, not just about the times we live in or the people using those terms, but groups that have power to change the discourse on such issues.

We will look at all these as separate types of discourse analysis. But first it’s important to understand why language is so important; it is much more than just a method of communication.

“Language allows us to do things. It allows us to engage in actions and activities. We promise people things, we open committee meetings, we propose to our lovers, we argue over politics, and we “talk to God”…

Language allows us to be things. It allows us to take on different socially significant identities. We can speak as experts—as doctors, lawyers, anime aficionados, or carpenters—or as ‘everyday people’. To take on any identity at a given time and place we have to ‘talk the talk’…”         - Gee 2011

Language is more than a neutral way of communicating, it’s deeply connected with actions and personal identity, and can even shape the way we think about and understand the world. Who we are, what we do, and our beliefs are all shaped by the language we use. This makes it a very rich avenue for analysis.

Types of discourse analysis

Just like so many blanket qualitative terms, there are a lot of different practices and types of analysis called ‘discourse’ analysis, and many different ways of applying them. Hodges et al. (2008) identify 3 meta-types, broadly going from more face-value to conceptual analysis:

     • Formal linguistic (basically looking at words/phrases, grammar or semantics)

     • Empirical (social practice constructed through text)
       
     • Critical (language constructing and limiting thought)

Tannen et al., 2015 categorise three similar broad types of analysis, again becoming increasingly socially conceptual:

• language use

• anything beyond the sentence

• a broader range of social practice that includes non-linguistic and non-specific instances of language

However Gee (2011) only recognises two main categories, essentially those that look at the use of words, and ‘critical discourse analysis’: like the latter of both groupings above, this is analysis of how language is situated in cultural and contextual power dynamics. But before we get there, let’s start with an example of some more obvious linguistic level discourse analysis.

Example

Imagine the following scenario from your favourite fictional medical drama. A patient is wheeled into the ER/casualty unit, conscious but suffering from burns. The doctor attending says three things:

To Patient: “We’re just going to give you a little injection to help with the pain.”

To Nurse: “10cc’s of sodium pentothal, stat!”

To Surgeon: “We’ve got severe second-degree chemical burns, GA administered”

In this situation, the doctor has said essentially the same thing 3 times, but each time using a different response for each recipient. Firstly, when talking to the patient, the doctor doesn’t use any medical terminology, and uses calming and minimising language to comfort the patient. This is a classic type of discourse we are familiar with from medical TV dramas, the ‘good bed-side manner’.

To the nurse, the doctor has a different tone, more commanding and even condescending. It’s a barked command, finished with the term ‘stat!’ - a commonly used medial slang word (actually from the Latin word ‘statum’ meaning immediately, that’s your linguistic analysis!). This is interesting, because it’s not a term you’d hear used in other professional places like a busy kitchen. It shows there is a specific discourse for the setting (a hospital) and for different people in the setting. The ‘10cc of sodium pentothal’ is a commonly used anaesthetic: the same ‘something to help with the pain’ but now with a (trademarked) pharmacological name and dose.

Finally, to the surgeon the same prescription is described by the doctor as an abbreviation (GA for General Anaesthetic). Between senior health professionals, abbreviations might be used more often, in this case actually hiding the specific drug given, perhaps on the basis that the surgeon doesn’t need to know. It could also imply that since only that basic first step has been made, there has been little assessment or intervention so far, telling to an experienced ear what stage of the proceedings they are walking in on. The use of the term ‘we’ might imply the doctor and surgeon are on the same level, as part of the team, a term not used when addressing the nurse.

Even in this small example, there are a lot of different aspects of discourse to unpack. It is very contextually dependent, none of the phrases or manners are likely to be adopted by the doctor in the supermarket or at home. This shows how the identity and performativity of the doctor is connected to their job (and shaped by it, and contextual norms). It also shows differences in discourse between different actors, and power dynamics which are expressed and created through discursive norms.

At a very basic level, we could probably do an interesting study on TV shows and the use of the term ‘stat!’. We could look at how often the term was used, how often it was used by doctors to nurses (often) and by nurses to doctors (rarely). This would probably be more like a basic linguistic analysis, possibly even quantitative. It’s one of the few occasions that a keyword search in a qualitative corpus can be useful – because you are looking at the use of a single, non-replaceable word. If someone says ‘now please’ or ‘as soon as you can’ it has a very different meaning and power dynamic, so we are not interested in synonyms here. However, we probably still want to trawl through the whole text to look at different phrases that are used, and why ‘stat!’ was not the command in all situations. This would be close to the ‘formal linguistic’ approach listed above.

But a more detailed, critical and contextual examination of the discourse might show that nurses struggle with out-moded power dynamics in hospitals (eg Fealy and McNamara 2007, Turner et al 2007). Both of these papers are described as ‘critical’ discourse analysis. However, this term is used in many different ways.

Critical discourse analysis is probably the most often cited, but often used in the most literal sense – that it looks at discourse critically, and takes a comparative and critical analytic stance. It’s another term like ‘grounded theory’ that is used as a catch-all for many different nuanced approaches. But there is another ‘level’ of critical discourse analysis, influenced by Foucault (1972, 1980) and others, that goes beyond reasons for use and local context, to examine how thought processes in society influenced by the control of language and meanings.

Critical discourse analysis (hardcore mode)

“What we commonly accept as objective or obviously true is only so because of negotiated agreement among people” – Gee (2011)

Language and discourse are not absolute. Gee (2011) notes at least three different ways that the positionality of discourse can be shown to be constructed and non-universal: meanings and reality can change over time, between cultures, and finally with ‘discursive construction’ – due to power dynamics in setting language that controls how we understand concepts. Gee uses the term ‘deconstruction’ in the Derridian sense of the word, advocating for the critical examining and dismantling of unquestioned assumptions about what words mean and where they come from.

But ‘deep’ critical discourse analysis also draws heavily from Foucault and an examination of how language is a result of power dynamics, and that the discourse of society heavily regulates what words are understood to mean, as well as who can use them. It also implies that because of these systems of control, discourse is used to actually change and reshape thought and expression. But the key jump is to understand and explain that “what we take to be the truth about the world importantly depends on the social relationships of which we are a part” (Gergen 2015). This is social construction, and a key part of the philosophy behind much critical discourse analysis.

Think of the use of the term ‘freedom’ in mainstream and political discourse in the United States. It is one of the most powerful words used by politicians, and has been for centuries (eg Chanley and Chanley 2015) However, it’s use and meaning have changed over time, and what different people from different parts of the political spectrum understand to be enshrined under this concept can be radically different, and even exclusionary. Those in powerful political and media positions are able to change the rhetoric around words like freedom, and sub-terms like ‘freedom of speech’ and ‘freedom of religion’ are both being shifted in public discourse, even on a daily basis, and taking our own internal concepts and ideas with them. It may be that there has never been an age when so much power to manipulate discourse is concentrated in so few places, and able to shift it so rapidly.

Doing Discourse

So do we ‘do’ discourse analysis? How can we start examining complex qualitative data from many voices from a point of view of discourse? Like so many qualitative analytical techniques, researchers will usually adopt a blend of approaches: doing some elements of linguistic analysis, as well as critical discourse analysis for some parts or research questions. They may also draw on narrative and thematic analysis. But discourse analysis is often comparative, it lends itself to differences in the use of language between individuals, professionals and contexts.

From a practical point of view, it can be started by a close reading of key words and terms, especially if it is not clear from the outset what the important and illustrative ones are going to be. For building a complete picture of discourse, a line-by-line approach can be adopted, but it’s also useful to use ‘codes’ or ‘themes’ to tag every use of some terms, or just significant ones. A qualitative software tool like Quirkos can help you do this.

For critical discourse analysis, examination of primary data is rarely enough – it needs to be deeply contextualised within the wider societal or environmental norms that govern a particular subset of discourse. So policy and document analysis are often entwined and can be analysed in the same project. From here, it’s difficult to describe a single technique further, as it will greatly vary by type of source. It is possible in discourse analysis for a single sentence or word to be the major focus of the study, or it may look widely across many different people and data sources.

The textbooks below are all classic works on discourse analysis, each a rabbit hole in itself to digest (especially the new edition of Gergen (2015) which goes much wider into social construction). However, Hodges et al. (2008) is a nice short, practical overview to start your journey.

If you are looking for a tool to help your qualitative discourse analysis, why not give Quirkos a try? It was designed by qualitative researchers to be the software they wanted to use, and is flexible enough for a whole number of analytical approaches, including discourse analysis. Download a free trial, or read more about it here.

References

Gee, J., P., 2011. An Introduction to Discourse Analysis. Routledge, London.

Gergen, K. J., 2015, An invitation to Social Construction. Sage, London.

Hodges, B. D., Kuper, A., Reeves, S. 2008. Discourse Analysis. BMJ, a879.

Johnstone, B., 2017. Discourse Analysis. Wiley, London.

Paltridge, B., 2012. Discourse Analysis: An Introduction. Bloomsbury.

Tannen, D., Hamilton, H., Schiffrin, D. 2015. The Handbook of Discourse Analysis. Wiley, Chichester.