I wanted to do a discourse analysis of the text on Twitter, but the data was too large and I had to determine a small range, so I did a random sampling. But the data is still too big. Can I do a word frequency count on the sampled data to find the most frequent words and then randomly select the samples for analysis in details?
Thank you for your answer.
I want to study how people understand the relationship between themselves and the nation and how they define the image of the nation in related posts published in two nationalist incidents. I used keywords to extract posts for a certain period of time, but the number of posts was very large. If I only choose posts that are suitable for my research, I might be questioned as being too subjective. In this case, I want to randomly select 10,000 posts for word frequency analysis, and then select specific posts for discourse analysis. I don't know if this is feasible? Most studies on nationalism use critical discourse analysis.
Any research seeks to answer the three questions: WHAT?, HOW? and the trickiest one WHY? This is applicable to all parts of your investigation. What specifically are you going to study in the (induced / spontaneous) discourse? You should limit yourself to a specific phenomenon in a specific type of discourse. Then limit yourself to the methodology and the amount of material needed for your research will take distinguishable shapes. My experience tells me that 2000-4000 of examples allow to draw reliable academic conclusions. 10 000 will be more than enough.
Random selection will give you a reliable source of material.
You can limit your data to be analysed by specifying your main topic. If your topic focuses on political discourse for example, you choose only tweets on politics to serve your purpose
Dear Wang
Limitation in a resesrch has to look at the topic, type and period of data, and methodology. Too many posts (i.e. 10000 posts) migh be appropriate for a more long- term and wide scale investigation..
Regards
1- 'data was too large': Give citation and reference of the original data in your paper; for example, Youtube link, Google, etc. where complete data is available.
2- Select sentences or small paragraphs related to your topic and critically analyze them one by one.
3- Long sentences and paragraphs will cause plagiarism issues if they are available online, so they should be written as stated below:
For example, the original sentence is: The president used very strong words such as the killer of innocent people, monsters and animals to show his determination to eradicate terrorism in the world.
The sentence in the paper (Italic): The president used very strong words such as the killer of innocent people...to eradicate terrorism in the world.... You can delete some words but you will analyze the complete sentence including the words that you have deleted.
If someone wants to see the original data, he/she can go to the link you provided. I do hope it can help you!
@ Qing Wang
Hope will be having a great day...
The first thing, you have to keep in mind as a researcher, is that you have to be very precise and specific in your research.
Obviously, you had a specific objective at the time, you decided to conduct your research.
If you are confused with data selection, first thing would be to select as much data that will justify the purpose and result of your research. You may select the tweets from a specific context, i.e political, entertainment industry etc...
If you want to draw a comparison between two contexts, try choosing an equal number of context specific tweets, that will justify your stance in the analysis of your research
You must have to focus on the alloted time and your comfort too.
Along with that, in Discource Analysis, the theory you are choosing retains an important point to give worth and value to your work.
Hope it will be helpful to you.
I think that in your case, purposive sampling will be much better but you need to justify all the steps taken with regards to explaining the methodology. Having something specific to look at, either it is gender, or political discourse etc. will lead you to the kind of analysis that will be appropriate to be carried out, especially towards addressing your research questions.
The selection of material for discourse analysis depends on the nature of discourse selected for analysis. For instance, if you aim to analyse humour in sit.coms, you can use purposive sampling and select sit.coms either from different TV channels or from the same channel. The amount of data depends on the research paradigm and also the research questions.
A research of this sort hinging on Discourse analysis would need a holistic approach and methodology to analyse the number or volume of data likely to obtain. Your research objectives and questions will inform the specificities that will be attached to your research and its methodology. I propose the use of CORPUS LINGUISTICS as a method and not a theory to help with the analysis. Corpus linguistic with the help of ANTCONC data analysis will help you work with a large data and identify the specific words, phrases that speak to your research objective.
You should choose a more specified model of analysis that is based on macro discoursal analysis.
Discourse analysis is a qualitative and interpretive method of analyzing texts (in contrast to more systematic methods like content analysis). You make interpretations based on both the details of the material itself and on contextual knowledge.
In order to construct a coherent framework, it is crucial to be aware of the philosophical, theoretical and methodological differences and similarities among the approaches. Obviously, this requires an overview of the field.
- Badges
- Science topic
- Similar topics
- Social Science
- Media
- Social Media