Writing a Dissertation in Microsoft Word

A little while back I decided to write my dissertation in Microsoft Word instead of the venerable LaTeX. I’d used LaTeX exclusively for papers and was looking to try something new. I’d seen friends use Word for technical papers, and plenty of people on the Internet were using Word for their dissertations, so why not me?

It turns out both Word and LaTeX are pretty terrible in many ways. I thought I’d share some of my reasoning behind choosing Word (and my experiences living with that decision!) for a sizable technical document: 129 pages, 167 references, 37 figures, a table of contents, a list of figures, lots of sections and cross-references, etc. I’m writing up my experiences as a series of blog posts. For the record, my experience is with Word 2011 for Mac.

I started our pretty positive about Word. Having finished my dissertation, I remain convinced that Word has a few killer features:

WYSIWYG. It was kind of a big deal back in the 80s and I can see why. I find that I write and, especially, edit much more easily with a WYSIWYG interface. Correcting a typo is so easy that it’s almost fun. Fixing widows/orphans and compressing the document to fit into a page limit is pretty painless when I always get to see exactly what the final output looks like. It’s like collapsing the edit/compile/run/debug cycle into a single fluid process: it’s so seamless it’s hard to go back.

With LaTeX I’m always reading the PDF output and then jumping back to the source code, and this switching adds friction to small tasks like correcting typos. There are cool features like SyncTeX to reduce this friction but it never really goes away.

Figure Layout. In Word, figures can be anchored to points in the text (usually what you want) or positioned absolutely to go-where-you-tell-them-to-and-stay-there. Other relatively-positioned figures may float on by and even overlap but those absolute figures are real stubborn and won’t budge. What an amazing feature!

LaTeX’s figure layout is pretty unfathomable at times, and getting things to stay put is pretty much impossible as far as I can tell. Word doesn’t always do the right thing automatically, but at least you can see what it is doing and tell it to do something else if you want. And it actually listens to you. Using a mouse to drag figures around is also a pretty revolutionary UI compared to all those negative \vspace{} commands I was using.

Better Citation Management. Word’s built-in citation manager is pretty limited but I use Zotero and its word processor integration is quite nice. When you want to add a citation, you get a simple search bar that quickly scans my entire Zotero library (a few hundred items at this point), searching all fields for the text I’ve entered. Zotero has nice support for importing publication information from the ACM Digital Library (and lots of other web sites), which lets me import a paper with a single click and then immediately cite it from Word. Nice!

Managing citations in LaTeX is a little clunkier. Adding new entries to the BibTeX file was always a bit of a pain point for me. I also find the presence of BibTeX keys disruptive to the flow of the writing, forcing me to read the PDF version and putting me further away from being able to change the text.

Better Annotating. Word’s annotation tools are nice, letting me or others mark-up specific text, make high-level comments, list TODOs in a non-layout-modifying way, and add or delete text with every edit clearly visible. PDF readers can do most of this too but it’s always one step away from the source code where the changes ultimately need to land.

Pain Points

But it’s not all goodness over in Word-ville. Oh no, there’s lots to dislike, too!

Reliability. Text editors don’t crash much these days, and when they do they generally don’t drag the text file down with them. I experienced two file-corrupting crashes while working on my dissertation, which was not particularly fun. I was lucky enough to have put my Word files in Dropbox so they were continuously backed up (in addition to Word’s backups), so ultimately I didn’t lose any real work. But it sure was exciting for a minute there!

Word also gets sluggish with large documents, which can be quite annoying. The WYSIWYG promise starts to fall down a bit here, but I still found the experience tolerable.

Alternate captions for a List of Figures. An obscure complaint, perhaps, but quite annoying nonetheless. My dissertation formatting requirements necessitated a List of Figures. No problem, Word can do that. It adds the figure numbers, their captions and page numbers in a nice big table. The only problem is that I often wanted to have an abbreviated caption in the List of Figures, and the full caption on the figure itself. Word doesn’t want to let you do that. So I would manually edit the captions in the List of Figures, and that was fine. Until I added a new figure or the page number for a figure changed, and then I had to regenerate the List of Figures which blew away all my abbreviated captions! The “good” news was that, with only 37 figures and only a subset of those needing abbreviated captions, I eventually memorized all the abbreviated captions and could do the changes quite quickly. Success?

Over in LaTeX, there’s almost certainly a package that supports exactly this workflow.

Slow citation adding with >50 references. After you have about 50 references cited in a document, adding new ones starts to take noticeable amounts of time. Once I was over 100 references the delay would be about 30 seconds. This is a known Zotero issue that seems to stem from the very slow interfaces available for accessing Word documents. Perhaps EndNote’s Word integration is faster in this regard? Certainly LaTeX+BibTeX don’t have this problem!

Hyperlinks. I wanted cross-references (\ref{} in LaTeX) to manifest as clickable links in the final PDF version of my dissertation. Shouldn’t be an issue – the cross-refs are clickable in Word itself and I have a Mac so it will just Do The Right Thing! Except that it doesn’t: Word 2011’s PDF generator doesn’t seem to know how to create such links (links to URLs don’t work either, which is kind of amazingly disappointing).

Eventually, I discovered that I could get the links to show up by sending the file over to Word 2010 (in Windows) and doing the PDF generation there. Aside from a few cross-platform rendering issues, the main stumbling block was that I had to export in PDF/A format to get fonts embedded. But export to PDF/A didn’t support figures with transparent parts (which I used a few times). Ultimately, I redid the figures in question to remove the transparency and then everything “worked”, in that it was a ridiculous process to get links to show up but they did in fact show up.

Having citations appear as links would have been nice, too, but Zotero doesn’t support that yet. Hopefully someday!

Update Field. When sections or figures move from one page to another, Word doesn’t automatically update existing cross-refs. You have to manually select the entire document and run the “Update Field” command yourself. Perhaps WYSIWYG should be amended to WYSIWY-used-to-get. Occasionally I’d forget to run Update Field when I should have, and be confused about various small errors in the cross references. Eventually, I created a “pre-export” checklist of all the silly little things I had to remember to do each time I generated a PDF. There’s no reason Update Field needed to be on this list; cross-refs should update automatically. Isn’t that what we have all these extra cores for anyway?

On balance, I think that using Word was still worth it. It made writing and editing a lot easier even though formatting flourishes were generally more difficult or, as with hyperlinked citations, impossible. Perhaps life is also smoother and stabler on the Windows side. I’ve used Word since my dissertation for single-author technical documents (like grants) and, now that I know where the sharp corners are, it’s actually pretty decent!


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s