Find the co-occurence of words with a specific word with NVivo?

[Veronique Gosselain]

Hello,

One of our PhD thesis is working with a lots of newspaper articles and would like to find automatically the words co-occuring with a specific word in all these papers. Is it something we can do with NVivo? and how?

Thanks a lot!

Véronique

[QDATRAINING Admin]

Hi Véronique,

Yes, you can do this with NVivo. There are a variety of means all involving queries. Here are three suggestions:

1. Text Search using the operator 'AND'. Find word A AND word B in all or a selection of sources or nodes and specify 'spread coding to' surrounding paragraph for example 
2. Use a compound query to further refine text searches. Use subquery 1 for first word or phrase, use subquery 2 for second word or phrase and choose from a range of operators such as NEAR, PRECEDING, SURROUNDING to define how the return will be made
3. Use a word frequency search to build a series of nodes based on individual words. Spread the finds to surrounding paragraph or specify how many words you want to see on either side of the coded word. Then run a matrix  to see where these these nodes intersect. You can convert the cells in thee matrix to nodes for intersections that meet the search criteria. 

The 'click here' instructions for running these searches can be found in the help files in NVivo: http://help-nv10-en.qsrinternational.com/nv10_help.htm

Basically Véronique, you need to get to know the query tool. It's well worth while investing a little time in this regard. You will get it back many times over as these queries allow you to filter your data in an intelligent way and get just the context you want to see before you. You can interrogate your data and allow the computer to do the administrative tasks as it finds content based on a logic or condition determined by you thereby freeing up your time to interpret the data which is a much better use of you (the researcher) as a resource. The beauty of queries is that you can't hurt your data in any way. The query either works and finds the information you think may be there or it doesn't in which case there is no harm done. Once you learn these skills, you will be able to ask questions of your data very quickly indeed as these queries take no time to run. 

I know that some people don't like searching through help files so If you like, I would be happy to give you  a live on-line demonstration of these tools on the live or tutorial data. Let me know if you would like this and we can easily arrange a day/time to to do it. 

Kind regards,

[Veronique Gosselain]

Hi Ben,

Thank you for your answer!

The problem is that the researcher doesn't know the word associated with the main word whcih she wants to explore the co-occurences. Sorry if my question was not precise enough.

So the question would be: is there a way to find the more frequent word associated (NEAR, SURROUNDING, …) with a single and well identified word?

Thank you in advance!

[QDATRAINING Admin]

Hi Véronique,

Well it's always going to be a little more difficult if you're not sure what you're looking for. So I'm going to suggest a couple of queries combined which should give you what you want:

First run a text search on the specific word you want to test for co-occurrences. Spread the findings to say 'surrounding paragraph' or specify how many words you want to see either side of it. This will help to refine the proximity of associated words for the next search to follow. Create a node with the results.

Next, run a word frequency query limiting the scope of the search to the new node only. The result will be a table of all words co-occurring with your original word and you can decide how close to the original word they should be. For example, if your original text search specified 5 words on either side, then the words displayed in your word frequency search are likely to be in the same sentence. You can decide the context you want to see. The results of the word frequency query will allow you to display the associated words in a tag cloud with the more frequently co-occurring words displayed the largest. Or a Tree Map or Cluster Analysis all available in one click on the tabs at the side of the result.

From here you have a range of options to explore these relationships. You could for example make a set of nodes of each associated word or the more frequently co-occurring ones and intersect them in a matrix. Or use a compound query to further refine the results using more complex conditions such as preceding for example. At this point you will know which words or even phrases are co-occurring so this information will help to direct  further querying based on language use.

I hope this helps!

Kind regards,

 

[Veronique Gosselain]

Hi Ben,

Thanks a lot for your prompt and very interesting answer! I think this time we have a good way to answer the initial question!!

You are right that it will take a little time to explore and test a veriety of possibilities offered by the query tools. I'm new in NVivo and I will need some time and experience …

Thanks for your help!

Véronique

[Author]

Posts