Developing a final outline

The problem with developing a final outline for a PhD dissertation chapter is that it is difficult to know when the outline is really final. It will be final of course when you are confident enough that it is of the requisite quality. The difficulty is how to identify the boundary between the conceptual development stage and the stage of writing-up for presentational purposes.

For academic writing it is the former that is the main headache, i.e. coming up with the content, rather than putting it into its final shape. It happens all too often that you think you are in the second stage, only to discover that you have more conceptual development and clarification to do, and you are thrown back into stage 1.

For now I can’t seem to find a better answer than to push on with aiming to reach a final outline, even if it turns out to be an interim stage of conceptual development. Then I just start developing a new “final” outline, hoping that it will really be the final one. As part of this process I am constantly switching between software tools, in search for specific features more suitable for the particular tasks in the subsequent stages of development.

This reflexive and recursive process of dealing with multiple outlines in multiple software tools eventually reveals an arrangement of outlines in an arrangement of software tools, and a shape-shifting process flow that adjusts itself continuously. It is like an experimentally assembled moonshine distillery for the purpose of abstraction to be achieved by routeing ideas (textual content) from one vessel to another, until they acquire sufficient clarity to be bottled and labelled and sold to a discerning public.

Here is a slightly revised graphic (following on my previous post) showing the current distillery and the process flow. Its shape has become slightly clearer to me.

writing-up process flow

On  the left we start with the notes database in ConnectedText. It is the base, the alpha and the omega. It contains the fruits of my research work. However, I now need to extract the spirit and the flavours with my distillery apparatus. Initially I extracted and organised the main conclusions in Natara Bonsai outlines. Bonsai is the fastest tool I have for organising lists into meaningful hierarchies and then drawing ‘final’ conclusions from these conclusions.

Parallel to this I have been developing a chapter outline in Outline 4D. I have also used O4D to summarise some ConnectedText meta topics which were already extractions of important findings (as opposed to the summaries of specific articles that have been analysed in Bonsai). As O4D allows for inline notes in its outlines, I have also pasted in selected quotes (from CT, which came from the original readings) to support particular outline items (my observations).

I also have a 7000-word draft in MS Word that I wrote at an earlier stage partly as an effort to break out of the circular outlining process and force myself to finish this stage (the target is a 10,000-word chapter). However, the draft writing process just turned out to be another stage of the distillery, as I have found some gaps in my conceptual structure. It felt like Word was a laboratory where I tried to assemble the parts but the trial failed and I needed to get back to the drawing board.

VUE was the drawing board, and concept mapping was the process of drawing out the issues for my conceptual apparatus and reorganise my argument. VUE was a kind of an experimental space, a sandbox. It is placed at the top of the diagram because it is an overview of everything that is going on. First I created a reverse outline of my Word draft, to get a better understanding of my argument so far. Then I also developed an outline (well, more of a concept map) for the next stage of writing.

I should mention that I also have an overall outline for the entire thesis kept in a Scrivener project, which contains some relevant material for this current literature review chapter as well. I will need to review it as I embark on constructing my “final” outline.

I have settled on a Freeplane mind map as a way of constructing my “final” outline, which would hopefully lead me through the final – presentational stage – of writing up. In the past I didn’t quite see the point of formal mind maps, as I preferred the freedom of concept maps. But now I see that a mind map – combined with Freeplane’s functionalities – is an excellent way to develop a final outline.

First, a mind map forces you to define a final hierarchical order between sections, ideas and paragraphs, which the final draft needs to have. Secondly, a Freeplane mind map is very economical with screen space, allowing you to cram a lot of information in and be able to navigate it and visualise it easily. Thirdly, it allows you to switch between a mind map and an outline view. Fourthly, it allows you to add notes to items, which can be chosen to be displayed inline, effectively operating as a single-pane outliner with inline notes (of which only a handful exist in the world).

Fifthly, it allows you to export into RTF file format in such a way that the inline notes get displayed below their items, and the items are assigned headings according to their hierarchical level. So for example in Word or LibreOffice, items can be viewed as section headings in the Navigator window, a set-up that can be used as a dual-pane outliner. Also, the headings can be used to create a table of contents.

Finally, (building on the fourth and fifth point), you could construct a complete sentence outline in Freeplane by using items as topic sentences and notes as the evidence to support the topic sentences (e.g. quotes or other details). Even a simple “copy and paste” into Word creates a bullet-pointed hierarchy, which then can be transformed into a draft (although with this export method you don’t get the headings formatting preserved – but that may be preferable in some situations).

I am planning to do the final writing in Outline 4D though (which will require importing the RTF file with the headings and notes), in recognition of the possibility that this “final” Freeplane outline may not turn out to be entirely final, and rather than create yet another reverse outline or concept map to keep track of the underlying and implicit conceptual structure, I could just use O4D’s outlining functionalities to keep track of the evolving implicit outline of the draft. This could also be done in Word, simply by modifying the headings that get displayed in the Navigator (as a dual-pane outliner set-up). However, I find O4D’s single-pane view more conducive for this simultaneous writing and real-time (reverse) outlining, as O4D has many helpful alternative visualisations of the text and the outline.

Do I worry that I get carried away with playing around with software tools instead of doing the writing? You bet. Although there is definitely a ludic aspect to this, in the end the distillation process is real and necessary. I would need to carry out the abstraction process somehow in any case. I am hoping that by constructing a sophisticated distillery I will be able to extract and construct a higher quality final product.

Could I be wrong about that? Yes, that is a distinct possibility. However, academic research is by definition an experimental process and experiments (and experimenters) can fail (and they often do), perhaps even more frequently than they succeed (which is why so many PhD students never complete their theses). All I can do is carry on and hope that my experimental process and set-up produce a satisfactory result.

Analytical process flow for reading notes in ConnectedText

Currently I’m working on my literature review. Here is my analytical process flow for importing data, analysing it, and outputting information using ConnectedText (CT) and a variety of other tools. This process flow is just a snapshot, it’s not set in stone. It keeps changing as my needs change and as I gradually develop approaches that better suit the creative process of analysis, evaluation and synthesis.

  1. Importing reading notes into CT:
    1. I read and annotate academic articles in PDF form using GoodReader (if the text can be highlighted) or PDF Expert (if the PDF is a scanned image) on an iPad.
    2. Then I email the notes (highlighted text + my comments) and the annotated file to myself.
    3. On the PC I replace the original PDF file with the annotated one, re-link my EndNote reference to it, and copy and paste the highlights and the comments from the email into a new CT topic in my “Readings” project (CT database) under the ==Quotes/Comments== heading in my CT reading notes template.
    4. With printed books (i.e. not e-books) I take notes using a pen and a paper notebook, and when I’m finished, I use Dragon NaturallySpeaking to dictate the selected quotes and my comments into DragonPad, from where I copy and paste them into the CT template as described above.
  2. Analysing and evaluating reading notes in CT:
    1. I use CTRL+H to “search and replace” (also available from Search > Replace) the headings inserted by GoodReader/PDF Expert with either ===Quote: === (for highlighted text) or ====Comments: ==== (for my own comments).
    2. I review the text and extract the essence of the quoted text and comments and add it into the headings, such as ===Quote: cognitive functions are socially acquired===, followed by ====Comment: I agree with this====.
    3. I use yellow colour to highlight particularly important quotes and comments.
  3. Organising conclusions:
    1. If the CT topic is short, I can develop and record my conclusions immediately under the heading =Evaluation=. I summarise my main point in a final couple of sentences under =Verdict=, including them in the attribute [[Verdict:=My concluding sentences go here.]], so that they show up in the Infobox at the top of my CT topic.
    2. If the CT topic is long (i.e. it contains a lot of imported quotes and comments), then I need to use additional tools to organise the annotated headings.
      1. I remove the [[$NOTOC:]] markup from my template, so that the Table of Contents (TOC) becomes visible within the topic.
      2. I highlight the contents of the TOC, right-click, copy, and paste it into a blank Natara Bonsai document. In Bonsai I already have the new document template set up, so that different hierarchical levels appear in different colours, to aid the sorting of information.
      3. Using Bonsai’s outlining functionality, I reorganise the imported contents of the CT TOC into a meaningful hierarchy.
  4. Developing an outline for the draft chapter:
    1. The whole purpose of evaluating reading notes is to come up with my own interpretation, supported with evidence. The next (and parallel) stage is to develop an overall outline for the draft thesis chapter. Depending on the complexity of the material, it may require several tools still:
      1. I keep CT open to be able to view given reading note.
      2. I consolidate material (my main points supported by key quotes) in a final Outline 4D outline (which is a single-pane outliner that can have inline notes, in contrast to Bonsai.)
      3. During this whole process I use an overall VUE concept map to work out relationships between concepts and to develop an argument.
      4. I record the very final overall outline in the form of a Freeplane mind map.
  5. Writing up:
    1. to manage the final writing-up process, I use MLO to record to-dos as they develop.
    2. I do the final writing-up in an Outline 4D document. The advantage of using O4D for this is that it obviates the need for another application to do reverse outlining as the text grows, because it is easy to alter headings for multiple hierarchical levels and toggle them on and off, when you only want to see the text. I find this better than dual-pane alternatives such as Word with Navigation Pane or Scrivener.
    3. To add references, I simply type the reference such as (Smith 2008: 35), so as not to be distracted and disrupted by having to switch to EndNote every time (and it is not compatible with O4D anyway).
    4. I export O4D text to Word.
    5. I replace manual references with EndNote references, to build bibliography.

Here is a graphic just showing the latter stages of my writing-up process flow. I embed these types of process flow graphics in the home page (dashboard) of my reading notes wiki, so that they remind me every time I get stuck. Chances are that in a few days it will be replaced with a modified process flow, as I keep tinkering with it.

writing-up process flow

Updated ConnectedText template for reading notes

I’ve added two new headings to my ConnectedText template for my reading notes database.

=Full text=

ConnectedText Project OutlineThis is for pasting in the full text of a publication. I mostly use this for blog posts or other online articles that I don’t have a PDF for, as these could disappear or be modified any time. I could have printed them as PDF ConnectedText topic iconsand linked them to EndNote or added them to my web clippings in Surfulater instead; however, I might forget about them that way. As these are important to read, I can link these topics to my Project Outline in CT, as to-dos. I could even mark the relevant topics with full text with a different icon in the Topic List window to remind myself to read them later.

I have also added the following markup just below the =See also= heading:


==Created on==

==Modified on==

These are so-called “include macros” that display the time and date the topic was created and last modified on. This is what it looks like at the bottom of the topic:

ConnectedText infoBy the way, this information also gets displayed in the Summary pane (if I choose to keep it open), even without the “include macros”:

ConnectedText Summary InfoWhy is it important to know when a reading note was created and last modified? It is not all that important at the time when you are first working on it. But once you have hundreds or even thousands of notes in your database, you may get into a situation when it comes in handy to know when you had read a particular article or book in your academic career or during your research project. For citations of blog posts and other online articles, the “created on” date can also serve as the “date accessed” for your bibliographic reference.

Update (12-Dec-12):

Here is the bare-bones template, which can be copied and pasted into a CT topic and saved as a template (txt file) (a description of the individual items can be found here):

==Cite as==
=Full text=
=See also=
==Created on==
==Modified on==

And this is what it looks like in view mode (keep in mind it’s blank at the moment, so some features such as the Infobox, attributes and categories are not showing yet):

ConnectedText reading notes template

Another example of ConnectedText notes database

Here is another example of a ConnectedText database for academic notes from Jamel Ostwald, although he in the end opted to stay with MS Access. It’s worth comparing the design of the database (or the underlying template for a topic) with some of the other designs mentioned here earlier, such as Brian Lennon’s setup, Steve Zeoli’s example, Glen Coulthard’s library project, or my template.

ConnectedText window across two monitors

After having seen Brian Lennon’s setup of a ConnectedText window stretched across two monitors, of course I couldn’t rest until coming up with my own ideal two-monitor configuration for my current workflow. I would suggest that any CT window configuration is workflow-specific, i.e. different projects require different arrangements of panes to maintain optimal workflow. (CT allows you to save your desktop configurations, so you can have as many variations as you like and switch between them freely.) See my two examples below, one with the browser pane in the second monitor, the other with the Navigator pane. I’m using CT’s help file, “Welcome to ConnectedText,” as my project example.

The rationale behind my arrangement is the following. My current task requires me to work with topics (rather than the reading displayed in the browser), therefore in Monitor 1 I prefer to focus on information that pertains to the topic at hand (such as its “Table of Contents” or its links in the “Summary” pane). The “Topic list” is there for easy browsing and so that I can drag and drop topics into the body of a topic when I need to create a link between two topics (such as adding a reference to a citation). The “History” pane serves as a quick way to alternate between topics that I’m currently working on.

Monitor 2 contains panes that I would use less frequently, so that I can still switch off this monitor when I need to block them from view, such as when I need to focus on writing in the main topic edit pane in Monitor 1. The “Category” pane lists all categories within the currently opened project (database file). The “Search results” pane shows all topics sorted by relevance that contain the search term I’ve recently searched for. To my great delight CT has recognised my default PDF-XChange Viewer plug-in, which means that I can do searches and add annotations (highlighting, notes) to my PDF files directly in XChange Viewer, without having to leave CT.

The only downside of the two-monitors view is that should you need to use an external application for anything else, you can’t work in both the CT window and the other program at the same time. So either CT needs to be minimised or the CT window needs to be dragged to a smaller size, to allow for screen space for the other application. [Update 14-Dec-12: Just found a solution to this. Install Direct Folders (it’s free and it was already on my Favourite Tools list) and by right-clicking on the “Close” button of the non-CT window that you want to stay on top, it will stay on top of the CT window. Brilliant!] This is where alternative saved CT desktop views may come in handy. [Click here and here for full-size images.]

ConnectedText across two monitors, example 1

ConnectedText across two monitors, example 2

If anyone else is willing to share their CT window configuration, just send me an email with your screenshot at docandus [at] and I’ll be happy to post it on this blog.

Docking panes in ConnectedText

ConnectedText (CT) is a highly modular application, with a large number of panes that can give different views of your data more or less simultaneously. Viewing a lot of panes at the same time requires screen real estate, therefore CT can particularly benefit from having a second monitor. Users with two monitors are presented with a decision to make: one could undock some of the panes and move them over to the other monitor, or one could keep them all docked but stretch the CT window across both monitors.

I belong to the first camp and I only use my second monitor for undocked floating CT panes (especially the Navigator pane, occasionally the Outliner pane). My main reason for this is that I only turn on the second monitor when I really have to, as I usually find it too distracting when I need to concentrate on working in the main monitor. However, recently I came across Brian Lennon‘s screenshot of his CT setup, who prefers to work with the CT window stretched out across both monitors.

I find Brian’s setup very interesting. There are clearly some benefits to this approach as well. You can have a more complete and diverse view of your project. You can have a more fluid workflow by staying within the CT environment, rather than using e.g. an external PDF viewer and web browser, as I do, which requires me to switch back and forth. (In fact only now I’ve understood why CT needs an internal browser and file viewer: so that you can create this sort of total CT environment.)

Below is a screenshot of Brian’s CT window (remember, it’s stretched across two monitors). I have added some labels and arrows to show all the different panes that are visible and the relationships between them. As you can see, Brian even had a bit of space left over in the right-hand side monitor, so he could have stretched the CT window out even further and made more of the panes visible. His CT project is also an interesting example of a reading notes database. (If you do have two monitors, you could try to stretch out the image across both of them to see the screenshot in its actual size and full glory.)

ConnectedText panesMany thanks to Brian Lennon for giving me permission to use his image.

P.S. Novices may find the docking process in CT rather tricky at first. Watch the second tutorial video to see how to carry out the docking of panes in ConnectedText.

On hierarchies and networks

I have written a blog post on hierarchical organisation vs. wikis for the Committee on the Anthropology of Science, Technology and Computing (CASTAC) of the American Anthropological Association, in response to their questions:

We all know that robust tools can help facilitate research, but we do not always have the time to test the latest products and processes. Here’s a place to offer advice, suggestions, and ask for help on how to tackle specific problems. What software have you found helpful for capturing data, transcribing interaction, conducting research, or analyzing findings? What problems tend to come up? Are there techniques in conceptualization, mapping, coding or other stages of the research process that you have identified as particularly helpful? Feel free to share information about what worked and what didn’t when using technology to gain insight into your projects.

ConnectedText template for reading notes

Several people have asked me recently how to use ConnectedText (CT) for taking notes for academic research and for writing a literature review. I’m in the middle of writing my own literature review, and unfortunately I don’t have time right now to construct a sample database (my own is confidential) for a tutorial. I will definitely do it one day. In the meantime however here are a few pointers.

  1. Think of CT as a traditional slip-box with index cards. Just drop your notes into CT (each note as a separate “topic”) and add some categories at the bottom or the top of the page, e.g. [[$CATEGORY:politics|Barack Obama]]. That’s it. You now have a reading notes database with two categories (“politics” and “Barack Obama”).
  2. You can create links between the “index cards.”
  3. You can create a bibliographical database in the background by adding “attributes” markup to key data such as [[Author:=Smith, J.]], [[Year:=1989]], [[Publication:=Journal of Sociology]], [[Abstract:=<value>]] etc. Clicking on any of these links will create a dynamic meta-topic that displays a list of topics with those values, such as a list of all the authors with links to the topics that contain those authors. Inserting [[$SUMMARY:]] into a new topic creates a table with your complete bibliography automatically.
  4. You can use the headings tags (=Heading 1=, ==Heading 2== etc.) to annotate your reading notes (e.g. quotes that you have pasted into the topic), which will be displayed dynamically (as you type) in the Table of Contents pane, giving you a quick overview of the contents of your note.

ConnectedText iconsBelow is a template I have created for my “Reading Notes” database in CT. You can copy and paste this into a text file and save it in the folder “My Documents/ConnectedText/Templates.” Then when you create a new topic in CT, select this text file in the “new topic” dialog box from the “Template” pull-down menu, tick the box “Always remember last used  template,” and select an icon to represent your notes (I have imported my own 16x16px icons that I’ve downloaded from the Internet, so notes, categories etc. all have their own icons). Similarly, tick the “Always remember last used icon” box. Icons are useful not only for visual browsing but also because you can filter topics in the Topic List pane by their icons.

You can of course alter this template to suit your own needs by using Notepad. If you haven’t used Categories or Attributes before, you may want to read up on those topics first in CT’s Welcome Project (Help file). By the way, the [[$INFOBOX]] markup will create a neat box that looks like an old style index card (assuming you have marked up the bibliographic data with attributes)! Very satisfying… :) (Obviously you will need to switch from edit mode into view mode to see what these markups actually do.)




### Paste reference from EndNote here

### Add attributes to Author name [[<Attribute name>:=<value>]], e.g. [[Author:=Smith, J.]]
### Add attributes to Year, e.g. [[Year:=1989]]
### Add attributes to Title, e.g. [[Title:=<value>]]
### Add attributes to Publication, e.g. [[Publication:=Journal of Sociology]]
### Add attributes to Publisher, e.g. [[Publisher:=Oxford University Press]]
==Cite as==
### Paste "raw" citation from EndNote here
### Drag & drop PDF or URL here
### Add attributes to Type, e.g. [[Type:=journal article]]

### Paste journal article abstract here and summarise its main point as [[Abstract:=<value>]]

### Paste keywords associated with journal article here
### Paste quotes from readings here
### highlight important information in yellow
==Quote 1: description==
### Annotate quotes with headings
===Comment 1: comment on quote 1===
### Add comments under own headings
### Draw conclusions, summarise key learning points here
### Summarise key contributions and/or shortcomings of article as [[Verdict:=<value>]]

=See also=
### Drag & drop related topic links, external files or URLs here
### Add categories below, e.g [[$CATEGORY:politics]]

This is what a new CT topic with this blank template looks like in view mode. It would need to be filled in with content as discussed above for the attributes to spring into action and produce the Infobox, as well as for the Table of Contents on the left to display the headings as annotations. For now this is just a skeleton for a note (index card):

blank CT topic with template

ConnectedText for me is part of a bigger note-taking system that I don’t have time to describe right now, though it’s similar to the one discussed here. For further examples of how to organise academic reading notes and bibliography see Steve Zeoli’s example here and Glen Coulthard’s tutorial video here.


For me each “topic” represents a specific publication, such as a book or a journal article. I name the topics using the author’s name and the publication date, such as “Smith 1989″ or Smith & Tailor 1988” or “Smith et al. 1977.” If there is more then one publication by the same author, then I use “Smith 1989a,” “Smith 1989b” etc. For historically important books, I also record the original publication date in parentheses: “Smith 1989 (1886).”

This naming convention is helpful because the Topic List pane lists topics in alphabetical order, so it’s easier to find authors this way. Also, if you decide to write up your literature review in CT, you can just drag and drop the topics from the Topic List straight into the body of your draft chapter and they become linked references automatically: “According to some observers (Smith 1989a) …” In edit mode it would look like

According to some observers ([[Smith 1989a]]) ...

This note-taking approach can result in fairly long topics, if they contain summaries of entire books or particularly important journal articles that I have annotated heavily. For me there are benefits to keeping a particular publication in one topic. However, one could just as well pursue a more granular approach and implement the traditional index card format more literally, by breaking down notes about articles and books into much smaller chunks.

Using links, categories and attributes/properties would be much more important under the latter scenario, for threading all of the related notes together. With my approach it’s not a big deal if I forget to add categories or attributes because I can find the relevant publications by their topic names. CT’s powerful search facility comes in handy in both scenarios, as sometimes searching for a single word or phrase might be the quickest way to find something (yes, it’s your own mini-Google :).