Choose Language Hide Translation Bar
gueriniere
Community Trekker

Analyzing texts

So far I have been using JMP for my statistical calculations in the research, and I have been stuck using different windows-programs for my qualitative research. I wonder if I could use JMP for that too?

 

I have two specific questions concerning text analysis with JMP that I have not found the answer to:

1) is it posible to use JMP for qualitative analysis? i.e. to code the texts in several steps, defining meaningful content, defining codes that should apply for part of the text, cathegorize the codes into groups with new names, etc, etc. (As you do when you perform a qualitative content analysis och thematic analysis or grounded theory analysis)

 

2) is it possible to add other languages that define the texts? (Our research is performed on interviews in Swedish and if I use JMP for ordinary text content analysis I have to choose between German and English, and none of them fit...)

0 Kudos
4 REPLIES 4
Highlighted

Re: Analyzing texts

What is the nature of the texts? The content of the interviews? If so, then see Help > Books > Basic Analysis > Text Explorer. This platform provides many tools for a straight-forward workflow and many opportunities to customize the tokenizing, stopping, and stemming rules. Additionally, the JMP Patterns and Regex functions can be used to extract specific information in the texts to create new features in the data set. The document-term matrix may be saved with weighting for further analysis in other JMP platforms, such as those for multivariate analysis or to be used as features in a predictive model. Three procedures are built in to Text Explorer in the JMP Pro version: Latent Class Analysis, Latent Semantic Analysis, and Discriminant Analysis.

 

You cannot add other languages but you could translate the texts before analysis with JMP.

Learn it once, use it forever!
gueriniere
Community Trekker

Re: Analyzing texts

Actually a qualitative content analysis or thematic analysis is a manual process. It is not possible to use regex or tokenizing or stemming rules since it is based on meanings and not wording or the the occurence of words.

e.g. a sentence like "I feel tingling in my gutt and my heart beat stronger when I see him" and "sitting besides her, just knowing she is there makes med relax and feel that I am home" both might be coded as "In love" in a specific topic. 

 

What you do when you perform qualitative analysis is reading the text, mark the beginning and the end of the specific part you are to code (That could be one or several posts in a table), and then define the code in another field (Where you preferrable would be able to either have a list of codes to select from but just as easily could write a new code.)

 

It is the next steps you need help from a program: You want to get the texts presented according to the codes, you want to be able to add another level of coding where you cathegorize the codes or use them as subcodes, but still be able to see the texts tha are coded similarly, etc.

 

The question is if whether JMP is able to help me in this? I understand JMP can perform a lot of textanalysis close to AI but what I seen of it is that all these are more or less automatic.

 

Regards,

 

0 Kudos

Re: Analyzing texts

Your assessment of JMP is correct. It performs a lot of automated processing at the token, term, and phrase level, but it does not support the kind of analysis that you describe as far as I know.

 

It migjht be possible to write a JMP script that presents a user interface with individual texts and ways to select, delimit, and tag portions of the text with your codes and sub-codes. The resulting data structure (e.g., JSL associative arrays or lists, XML or JSON) could be used to create a normal JMP data table, perhaps one row per tagged portion of the text.

 

That script might be a bit of work, though.

Learn it once, use it forever!

Re: Analyzing texts

Additionally, we developed a two-day course about text analysis.

Learn it once, use it forever!