Skip to content

Latest commit

 

History

History
116 lines (84 loc) · 11.4 KB

README.org

File metadata and controls

116 lines (84 loc) · 11.4 KB

orgqda: Qualitative Data Analysis for Emacs org mode addicts

orgqda defines a minor mode and several commands for making coding, listing of codes, and collection of coded extracts possible in text written in org mode. I created it for coding field notes and interview transcripts in my PhD project.

The idea is to re-use and extend the tagging machinery of org mode, and to keep all information in the source files. For achieving this, headlines can be tagged as usual which is interpreted as “coding” the text below that headline with the tags used. In addition, paragraphs (but not smaller extracts) can be coded by appending a degenerate inlinetask (that is, an inlinetask without a closing line) with tags. Like this:

A paragraph to be coded. It contains several sentences, but the full paragraph
will be considered as coded with the tags below. This is of course a limitation
compared to other QDA systems which usually allow marking sentences and words
easily. I usually use codes for “access” to the relevant section, so this
somewhat coarse coding works ok for me. I also often split up paragraphs into
smaller parts while coding.
​*************** ∈ :codeOne:codeTwo:

Installation

orgqda is not available from any package repository. For installing: download, put it in your load path and load as needed.

If you use use-package, a starting point for making sure the minor mode is autoloaded would be this:

(use-package orgqda-mode
  :commands orgqda-mode)

With package managers like straight, the installation can be done automatically:

(use-package orgqda
  :straight (:host github :repo "andersjohansson/orgqda"))

Getting started with coding

For getting started with coding, an example directory of a text corpus is given in the /examples directory.

Let’s start with a single file, file1.org, containing text (notes or transcripts or whatever) that we want to code. This file doesn’t have a headline structure and we rather want to code paragraphs.

Open the file and while reading through it, with the cursor on a paragraph that you want to code, either call orgqda-insert-inlinetask-coding (C-c C-x n if you activate orgqda-mode) or orgqda-code-headline-or-paragraph (which is a great remapping for org-set-tags-command). This will insert an inlinetask below the paragraph and prompt for tags (codes). Repeat throughout file.

When the file is coded, we can list and count all tags with orgqda-list-tags, and collect extracts for given tags with orgqda-collect-tagged (further details in Listing codes and collecting extracts). A good way for accessing all relevant commands is orgqda-transient.

The most important next step could be to keep data in several files, for which you would want to set orgqda-tag-files which behaves similarly to org-agenda-files and activate orgqda-mode. The best way to to this is usually through a .dir-locals file. See examples/.dir-locals.el for an example. When you have done this tags for listing and completion from are gathered from all files defined via orgqda-tag-files. This is helpful for instance when keeping different field journals and transcripts in different files but still using a common coding scheme for all of them. It could also be good to use a codebook file for describing the codes used.

Listing codes and collecting extracts

“Codes” can be listed with orgqda-list-tags and all coded text for a specific code extracted with orgqda-collect-tagged (or by clicking on a tag or C-c C-o on it). Both these functions create a custom org-mode buffer with links to the original regions . The command orgqda-revert-taglist updates a tag list buffer (key: g), and can also update counts for tag lists in any file with otag-links (this can be useful for writing and updating a codebook, an index with descriptions of used tags).

Codes can also be renamed with orgqda-rename-tag (R in orgqda-list-mode), given a prefix orgqda-prefix-tag (P) (see Hierarchy below), and merged via drag and drop in the code-listing buffer created by orgqda-list-tags. These operations find and replace tags across all files defined in orgqda-tag-files, which means undoing can be messy, make sure to use version control.

Codebook file

When using a file based on a tag list (with otag-links), as a code book, set (dir- or file-locally) orgqda-codebook-file to the path to this file and tag links there will be updated when the rename-functions are called. Such a file can also benefit from the bindings defined in orgqda-codebook-mode which binds the keys of orgqda-list-mode to the prefix C-c (. However, orgqda-transient in orgqda-transient.el also provides a convenient entry point for all relevant commands.

Hierarchy

Codes can be put in a hierarchy if using a delimiter inside tag names (default _, customizable via orgqda-hierarchy-delimiter). The listing of tags is then done according to this implicit hierarchy (unless orgqda-use-tag-hierarchy is set to nil).

Code relations – codes occurring together.

A simple command for displaying code relations is available as orgqda-view-tag-relations. It counts the number of times a pair (or k-tuple with a numeric prefix argument) of tags occur together.

csv-export of coded regions

I have been using the concept mapping software VUE for sorting through coded regions graphically. The csv-export functions is a way of importing regions into VUE, or other programs. The csv-export code is currently not as well-maintained and polished as other parts.

Customizing

Be sure to check out all custom variables in the orgqda group: `M-x customize-group` enter `orgqda`.

Auxilary libraries

The repository includes some extra libraries that can be useful for coding and interacting with orgqda and qualitative text data in general.

Automatically annotate completions of tags with count and information from codebook if defined. Activating the minor mode orgqda-completion-mode will enable annotations, sorting and grouping, with some help from the marginalia library. Additionally it redefines completion to include tags from all files defined in orgqda-tag-files. I recommend activating it with orgqda:

(add-hook 'orgqda-mode-hook #'orgqda-completion-mode)

When completing, tags are sorted according to orgqda-completion-sort , but this can be cycled by calling orgqda-completion-cycle-sorting, which can be bound to a suitable key in a suitable keymap, usually minibuffer-mode-map (With my Swedish keyboard and using vertico for completion I do: (bind-key "C-å" #'orgqda-completion-cycle-sorting 'vertico-map)).

Additionally, if tag hierachy is used, tags can be grouped up to the level given by orgqda-completion-group.

Defines convenient transient keymaps (using the transient library) for accesing the orgqda commands. Bind orgqda-transient to a suitable key in orgqda-mode-map.

A library with some functions for helping in transcribing interviews to structured org files (which can easily be coded with orgqda). It uses mplayer-mode, although it currently and unfortunately depends on my branch with some (kind of incompatible) changes for aligning it with org-mode and other things: org-sessions branch of mplayer-mode (see also a discussion about the future of mplayer-mode).

orgqda-transcript-mode defines a few functions and variables for defining a list of speakers, inserting timestamps, speaker names, switching speakers (if something was misattributed), measuring speaking time, etc. Take a look at the commands and custom variables if you are interested. As everything here, it is of course kind of idiosyncratic and aligned with my current workflow, and also more or less a work in progress.

orgqda-helm-tags.el (Deprecated)

NOTE: I won’t be updating orgqda-helm-tags.el since I’m moving away from helm.

A library which overrides org-set-tags with a custom function that uses helm and displays the tags sorted by use. If you use helm this is recommended! It also allows for quick adding of multiple tags as it calls helm repeatedly until exited with either C-RET or C-g. Tags can also be removed with a secondary action or tagged extracts viewed with a persistent action. If a codebook file is defined, the display fetches the first line (which should be a short description/definition of the code/tag) from this for each tag, to ease correct tagging. Sorting of the completion list can be defined in orgqda-helm-tags-sort, and cycled in the helm session with C-c C-s.

The library defines a minor mode, orgqda-helm-tags-mode, that overrides the C-c C-q binding and that is enabled with orgqda-mode if orgqda-helm-tags-completion is non-nil. This minor mode can be used outside orgqda as well (although it depends on some functionality from orgqda.el).

Note on configuration

I often load orgqda-mode through file or dir local variables, and as activation of the mode can depend on other variables being defined locally (most importantly orgqda-tag-files) loading order is important (i.e. that orgqda-mode gets activated after local variables are set). A solution is to activate it in a locally defined hack-local-variables-hook. So adding this to a .dir-locals.el file is one way of making it work:

((org-mode
   (eval add-hook 'hack-local-variables-hook 'orgqda-mode nil t)))

Note on some possible problems with org mode’s tag-completion

When adding tags and giving completion, org-mode has several mechanisms that determines which tags this should be. Tags that should be available for completion in all buffers can be added to org-persistent-tags-alist, and tags that should be available for a single buffer can be added with the #+TAGS: keyword, but all this interferes with orgqda-mode. If any of these mechanisms (which are activated when org-mode loads) sets org-current-tag-alist, the dynamic fetching of tags that we most certainly want for orgqda-mode is prevented.

One solution for avoiding this is never using the #+TAGS: keyword and setting #+STARTUP: noptags for all files used in orgqda. One measure to prevent problems is taken by orgqda-mode as well, in that it sets org-current-tags-alist to nil when orgqda-mode is activated. So if orgqda-mode gets activated automatically (for instance like detailed above) in all relevant files (even in files defined in orgqda-tag-files from which tags should be fetched) you’re all set. Also, using orgqda-helm-tags, which overrides org-set-tags, avoids all these problems.