ARTICLE 90: Research Methods for Ph. D. and Master’s Degree Studies: Data Analysis Part 3 of 7 Parts

Written by Dr. Hannes Nel

I discuss conversation and discourse analysis as data collection methods in this article.

Conversation and discourse analysis

Both conversation and discourse analysis approaches stem from the ethnomethodological tradition, which is the study of the ways in which people produce recognisable social orders and processes. Both of these approaches tend to examine text as an “object of analysis”. Discourse analysis is a rather comprehensive process of evaluating the structures of conversations, negotiations and other forms of discourse as well as how people interact when communicating with one another. The sharing of meaning through discourse always takes place in a particular context so the social construction of such discourse can also be analysed.

Conversation and discourse analysis both study “naturally” occurring language, as opposed to text resulting from more “artificial” contexts, such as formal interviews. The purpose is to identify social and cultural meanings and phenomena from the discourse studied, which is why the process is suitable for almost any culture-related research.

The name “discourse” shows that it is language that is analysed while language is also used to do research. It can be a complex process and is often better suited to those more interested in theorising about life than those who want to research actual life events.

Discourse analysis focuses on the meaning of the spoken and written word, and the reasons why it is the way it is. Discourse refers to expressing oneself using words and to the variety and flexibility of language in the way language is used in ordinary interaction.

When doing research, we often look for answers in places or sources that we can easily reach when the real answers might lie somewhere else. Discourse analysis is one method which allows us to move beyond the obvious to the less obvious, although much more relevant sources of data.

Discourse analysis analyses what people say apart from just picturing facts. Discourses are ever-present ways of knowing, valuing and experiencing the world. Different people have different discourses. Gangs on the Cape Flats, for example, use words and sentences that the ordinary man on the street will find difficult to understand. Discourses are used in everyday texts for building power and knowledge, for regulation and normalisation, for the development of new knowledge and power relations.

As a language-based analytical process, discourse analysis is concerned with studying and analysing written texts and spoken words to reveal any possible relationships between language and social interaction. Language is analysed as a possible source of power, dominance, inequality and bias. Processes that may be the subject of research include how language is initiated, maintained, reproduced and transformed within specific social, economic, political and historical contexts. A wide variety of relationships and context can be investigated and analysed, including ways in which the dominant forces in society construct versions of reality that favour their interests, and to uncover the ideological assumptions that are hidden in the words of our written text or oral speech in order to resist, overcome or even capitalise on various forms of power. Criminals in a correctional facility will, for example, be included or excluded from gangs on account of certain ways of speech and codes that only they know.

Discourse analysis collects, transcribes and analyses ordinary talk and everyday explanations for social actions and interaction. It emphasizes the use of language as a way to construct social reality. Yin[1] defines discourse analysis as follows:

“Discourse analysis focuses on explicit theory formation and analysis of the relationships between the structures of text, talk, language use, verbal interaction or communication, on the one hand, and societal, political, or cultural micro- and macro-structures and cognitive social representations, on the other hand.”

Discourse analysis examines a discourse by looking at patterns of the language used in a communication exchange as well as the social and cultural contexts in which these communications occur. It can include counting terms, words, and themes. The relationship between a given communication exchange and its social context requires an appreciation and understanding of culturally specific ways of speaking and writing and ways of organising thoughts.

Oral communication always fits into a context which lends meaning to it. It always has a double structure, namely the propositional context (ontology) and the performatory content (epistemological meaning). Oral communication can, for example, be used with good effect to understand human behaviour, thought processes and points of view. 

The result of discourse analysis is a form of psychological natural history of the phenomena in which you are interested. To be of value for research purposes oral communication must be legitimate, true, justified, sincere and understandable. It should also be coherent in organisation and content and enable people to construct meaning in social context. Participants in oral communication should do so voluntarily and enjoy equal opportunity to speak.

Discourse analysis is a form of critical theory. You, as the researcher, need to ensure that the discourse and the participants in the discussion meet the requirements for such interaction. It will also be your duty to eliminate or at least reduce any forces or interventions that may disrupt the communication. Such discourse can also be taken further by having other participants in the research process elaborate and further analyse the results of initial communications. For this purpose, you need to be highly sensitive to the nuance of language.

Any qualitative research allows you to make use of coding and structuring of data by means of dedicated research software, such as ATLAS.ti or CAQDAS. This will enable you to discover patterns and broad areas of salient argumentation, intentions, functions, and consequences of the discourse. By seeking alternative explanations and the degree of variability in the discourse, it is possible to rule out rival interpretations and arrive at a fair and accurate comprehension of what took place and what it meant. 

Discourse analysis can also be used to analyse and interpret written communication on condition that the written communication is a written version of communication relevant to the topic being researched. This requires a careful reading and interpretation of textual material.

Discourse analysis has been criticized for its lack of system, its emphasis on the linguistic construction of a social reality, and the impact of the analysis in shifting attention away from what is being analysed and towards the analysis itself. Discourse is in actual fact a text in itself, with the result that it can also be analysed for meaning and inferences, which might lead to the original meaning of oral communication being eroded at the expense of accuracy, authenticity, validity and relevance. 

Conversation analysis is arguably the most immediate and most frequently used form of discourse analysis in the sense that it includes any face-to-face social interaction. Social interaction inevitably includes contact with other people and contact with other people mostly includes communication. People construct meaning through speech and text, and its object of analysis typically goes beyond individual sentences. Data on conversations can be collected through direct communication, which needs to be recorded by taking notes, making a video or electronic recording.

Conversation analysis is the study of talk in interaction and generally attempts to describe the orderliness, structure and sequential patterns of interaction, whether this is universal or a casual conversation. Conversation analysis is a way of analysing data and has its own methodological features. It studies the social organisation of two-way conversation through a detailed inspection of voice recordings and transcriptions made from such recordings, and relies much more on the patterns, structures and language used in speech and the written word than other forms of data analysis.

Conversation analysis assumes that it is fundamentally through interaction that participants build social context. The notion of talk as action is central to its framework. Within a focus group we can see how people tell stories, joke, agree, debate, argue, challenge or attempt to persuade. We can see how they present particular ‘versions’ of themselves and others for particular interactional purposes, for example to impress, flatter, tease, ridicule, complain, criticise or condone.

Participants build the context of their talk in and through the talk while talking. The talk itself, in its interactional context, provides the primary data for analysis. Further, it is possible to harness analytical resources intrinsic to the data: by focusing on participants’ own understanding of the interaction as displayed directly in their talk, through the conversational practices they use. In this way, a conversation analytic approach prioritises the participants’ (rather than the analysts’) analysis of the interaction.

Naturally occurring data, i.e. data produced independent of the researcher, encompass a range of universal contexts (for example classrooms, courtrooms, doctors’ surgeries, etc.), in which talk has been shown both to follow the conversations of ‘every-day’ conversation and systematically to depart from these.

Conversation analysis tends to be more granular than classical discourse analysis, looking at elements such as grammatical structures and concentrating on smaller units of text, such as phrases and sentences. An example of conversation analysis is where a researcher “eavesdrops” on the way in which different convicted criminals talk to other inmates to find a pattern in their cognitive thinking processes.

While conversation and discourse analysis are similar in several ways, there are some key differences. Discourse analysis is generally broader in what it studies, utilising pretty much any naturally occurring text, including written texts, lectures, documents, etc. An example of discourse analysis would be if a researcher were to go through transcripts or listen in on group discussions between convicted serial murderers to examine their patterns of reasoning.

The implications of discourse and conversation analysis for data collection and sampling are twofold. The first pertains to sample sizes and the amount of time and effort that goes into text analysis at such a fine level of detail, relative to thematic analysis. In a standard thematic analysis, the item of analysis may be a few sentences of text, and the analytic action would be to identify themes within that text segment. In contrast, linguistic-oriented approaches, such as conversation and discourse analysis, require intricate dissection of words, phrases, sentences and interaction among speakers. In some cases, tonal inflection is included in the analysis. Linguistic analysis, be it transcripts of conversations, interviews or any other form of communication, often consists of an abundance of material to analyse, which requires detailed analysis. This requires substantial time and effort, with the result that not too many samples can be processed in a reasonable time.

The data source inevitably determines the type and volume of analysis that can be done. Both discourse analysis and conversation analysis are interested in naturally occurring language. In-depth interviews and focus groups can be used to collect data, although they are not ideal if it is important to analyse social communication. Analysis of such data often requires reading and rereading material to identify key themes and other wanted information which would lead to meanings relevant to the purpose of the research. 

Existing documents, for example written statements made by convicted criminals, are excellent sources of data for discourse analysis as well as conversation analysis. In terms of field research, participant observation is ideal for capturing “naturally occurring” discourse. Minutes of meetings, written statements, transcripts of discussions, etc. can be used for this purpose. During participant observation, one can also record naturally occurring conversations between two or more people belonging to the target population for the study, for example two surviving victims of attacks by serial killers, two security guards who had experiences with attempted serial killings, etc. In many cases legal implications might make listening in to conversations difficult to do without running the risk of encountering legal problems.

Text can be any documentation, including personal reflections, books, official documents and many more. In action research this is enhanced with personal experiences, which can also be put on paper so that they often become historical data. In action research the research is given a more relevant cultural “flavour” by engaging participants from the community directly in the data collection and analysis. The emphasis is on open relationships with participants so that they have a direct say in how data is collected and interpreted. If participants decide that technical procedures such as sampling or skilled tasks such as interviewing should be part of the data collection and analysis process, they could draw on expert advice and training supplied by researchers.

Paradigmatic approaches that fit well with discourse and conversation analysis include constructivism, hermeneutics, interpretivism, critical theory, post-structuralism and ethnomethodology.

Summary

Discourse analysis:

  1. Evaluates the structures of conversations, negotiations and other forms of communication.
  2. Is dependent on context.
  3. Analyses and uses language.
  4. Focuses on the meaning of the spoken and written word.
  5. Allows the researcher to move from the obvious to the less obvious.
  6. Is concerned with studying and analysing written texts and spoken words to reveal the relationships between language and social interaction.
  7. Examines a discourse by looking at patterns of the language used.
  8. Delivers a form of psychological natural history of the phenomena being investigated.
  9. Is a form of critical theory.
  10. Is criticised for its lack of system, emphasis on the linguistic construction of social reality and the lack of focus on the research problem.

Conversation analysis:

  1. Is a form of discourse analysis.
  2. Includes face-to-face social interaction.
  3. Attempts to describe the orderliness, structure and sequential patterns of interaction.
  4. Has its own methodological features.
  5. Assumes that it is fundamentally through interaction that participants build social context.

Discourse and conversation analysis:

  1. Stem from the ethnomethodological tradition.
  2. Examine text as the object of analysis.
  3. Study naturally occurring language.
  4. Identify social and cultural meanings and phenomena.
  5. Require intricate dissection of words, phrases, sentences and interaction between people.

Close

The differences between discourse and conversation analysis are subtle.

Discourse analysis is broader than conversation analysis in the range of its analysis.

While conversation analysis tends to go into finer detail than discourse analysis.

Enjoy your studies.

Thank you.


[1] 2016: 69.

Continue Reading

ARTICLE 89: Research Methods for Ph. D. and Master’s Degree Studies: Data Analysis, Part 2 of 7

Written by Dr. Hannes Nel

Hello, I am Hannes Nel and I discuss comparative and content analysis in this article.

Although quite simple, comparative and content analysis are most valuable for your research towards a master’s degree or a Ph. D.

It does not matter what the topic of your research is – you will compare concepts, events or phenomena and you will study the content of existing data sources.

What you need to know, is how to analyse and use such data.

Comparative analysis

Comparative analysis is a means of analysing the causal contribution of different conditions to an outcome of interest. It is especially suitable for analysing situations of causal complexity, that is, situations in which an outcome may result from several different combinations of causal conditions. The diversity, variety and extent of an analysis can be increased, and the significance potential of empirical data can be improved through comparative analysis. The human element plays an important role in comparative research because it is often human activities and manifestations that are compared.

Although theoretical abstractions from reality can be and, in some instances are the only way in which to do valid comparison, the units of analysis can also be whole societies or systems within societies. Comparative research does not simply mean comparing different societies or the same society over time – it might involve searching systematically for similarities and differences between the cases under consideration.

Comparative researchers usually base their research on secondary sources, such as policy papers, historical documents or official statistics, but some degree of interviewing and observation could also be involved. A measure of verification is achieved by consulting more than one source on a particular issue.

Qualitative research approaches are most suitable for the conduct of comparative analysis, with the result that many paradigmatic approaches can be used. Examples include behaviourism, critical race theory, critical theory, ethnomethodology, feminism, hermeneutics and many more.

Content analysis

Content analysis is a systematic approach to qualitative data analysis, making it suitable to serve as the foundation of qualitative research software. It is an objective and systematic way in which to identify and summarise message content. The term ‘content analysis’ refers to the analysis of such things as books, brochures, written or typed documents, transcripts, news reports, visual media as well as the analysis of narratives such as diaries or journals. Although mostly associated with qualitative research approaches, statistical and other numerical data can also be analysed, making content analysis suitable for quantitative research as well. Sampling and coding are ubiquitous elements of content analysis.

The most obvious example of content analysis is the literature study that any researcher needs to do when preparing a research proposal as well as when conducting the actual research for a doctoral or master’s degree.

Especially (but not only) inexperienced students often think that volume is equal to quality, with the result that they include any content in their thesis or dissertations without even asking themselves if it is relevant to the research that they are doing. The information that you include in your thesis or dissertation must be relevant and it must add value to your thesis or dissertation.

We analyse the characteristics of language as communication regarding its content. This means examining words or phrases within a wide range of texts, including books, book chapters, essays, interviews and speeches as well as informal conversation and headlines. By examining the presence or repetition of certain words and phrases in these texts you are able to make inferences about the philosophical assumptions of a writer, a written piece, the audience for which a piece is written, and even the culture and time in which the text is embedded. Due to this wide array of applications, content analysis is used in literature and rhetoric, marketing psychology and cognitive science, etc.

The purpose of content analysis is to identify patterns, themes, biases and meanings. Classical content analysis will look at patterns in terms used, ideas expressed, associations among ideas, justifications, and explanations. It is a process of looking at data from different angles with a view to identifying key arguments, principles or facts in the text that will help us to understand and interpret the raw data. It is an inductive and iterative process where we look for similarities and differences in text that would corroborate or disprove theory or a hypothesis. A typical content analysis would be to evaluate the contents of a newly written academic book to see if it is on a suitable level and aligned with the learning outcomes of a curriculum.

Content analysis can also be used to analyse ethnographic data. Ethnographic data can be used to prove or disprove a hypothesis. However, in this case validity might be suspect, primarily because a hypothesis should be proven or rejected on account of valid evidence. Quantitative analysis is often regarded as more “scientific” and therefore more accurate than qualitative data. This, however, is a perception that only holds true if the quantitative data can be shown to be objective, accurate and authentic. Qualitative data that is sufficiently corroborated is often more valid and accurate than quantitative data based on inaccurate or manipulated statistics.

Content analysis would typically comprise of three stages: stating the research problem, collecting and retrieving the text and employing sampling methods, interpretation and analysis. Stating the problem will typically be done early in the thesis or dissertation. Collecting and retrieving text and employing sampling methods are typically the actual research process, which may include interviewing, literature study, etc.

It is a good idea to code your work as you write. Find one or more key words for every section and keep record of it. In this manner you will be able to find arguments that belong together more easily, and you will be able to avoid duplication of the same content at different places in your thesis or dissertation. Most dedicated computer software enables you to not only keep content with the same code together, but also to access and even print it. This is especially valuable for structuring the contents of your thesis or dissertation in a logical narrative format and to come to conclusions without contradicting yourself.

Content analysis sometimes incorporates a quantitative element. It is based on examining data for recurrent instances, i.e. patterns, of some kind. These instances are then systematically identified across the data set and grouped together. You should first decide on the unit of analysis: this could be the whole group, the group dynamics, the individual participants, or the participant’s utterances. The unit of analysis provides the basis for developing a coding system, and the codes are then applied systematically across a transcript. Once the data have been coded, a further issue is whether to quantify them via counting instances. Counting is an effective way in which to provide a summary or overview of the data set as a whole.

Interviewing is mostly used prior to doing content analysis, although literature study can also be used. Analysing data obtained through interviewing includes analysing data obtained from a focus group. This variation of content analysis usually begins by examining the text of similarly used words, themes, or answers to questions. Analysed data need to be arranged to fit the purpose of the research. This can, for example, be achieved by indexing data under certain topics or subjects or by using dedicated research software. In addition to individual ideas, the flow of ideas throughout the group should also be examined. It is, for example, important to determine which ideas enjoy the most support and agreement.

Paradigmatic approaches that fit well with content analysis include feminism, hermeneutics, interpretivism, modernism, post-colonialism and rationalism.

Summary

Comparative analysis:

  1. Analyses the conditions that lead to an outcome.
  2. Involves searching systematically for similarities and differences.
  3. Mostly uses secondary data sources.
  4. Is mostly used with qualitative research.

Theoretical abstracts can be used for comparative analysis.

Comparative analysis is used:

1.      To increase the diversity, variety and extent of an analysis.

2.      To analyse human activities.

3.      To analyse whole societies and systems within societies.

Content analysis:

  1. Can serve as the foundation for qualitative research.
  2. Can be used with qualitative and quantitative research.
  3. Extensively uses literature as data.
  4. Can also be used to analyse ethnographic data.

The purpose of content analysis is to identify patterns, themes, biases and meanings.

It would typically comprise of three stages: stating the research problem, collecting data, and analysing data.

Coding can be used with good effect in content analysis.

Close

You probably already noticed that the differences between different data analysis methods are just a matter of emphasis.

They share many elements.

For example, both comparative analysis and content analysis use literature as sources of data.

Both fit in better with qualitative research than with quantitative research.

This means that you can use more than one data analysis method to achieve the purpose of your research.

Enjoy your studies.

Thank you.

Continue Reading