How do I find spelling inconsistencies when proofreading and editing?

Like so many of my proofreading and editing colleagues, I never rely on my eye alone. I’m human, and my eye sometimes sees what it wants to see rather than what’s there, even when I’m working with clients rather than reading for pleasure.

TextSTAT: Creating a frequency list

One of my favourite tools is TextSTAT. Actually, it wasn’t created with the proofreader or editor in mind. Rather, the program was designed to enable users to analyse texts for word frequency and concordance. However, I use it to generate, very quickly, simple alphabetized word lists.

Time and again, those word lists have flagged up potential problems that I need to check in a proofreading or copyediting project.

If I'm proofreading a PDF, I strip the text from the PDF proof and dump it into a Word file. I remove word breaks from that Word file (using "-^p") so that TextSTAT generates a list of whole words that I can compare, rather than thousands of useless broken words).

If I'm editing in Word, I can obviously bypass the above steps.

Identifying potential problems in text

Here’s a small sample from a word list I generated in TextSTAT. As you can see, there are several possible problems:
​

• -se vs -ze spelling
• US-English vs British-English spelling
• possible misspelling of cited author name
• possible inconsistency with regard to hyphenation
• possible misspelling of non-English words
• inconsistent terminology
• misspelling

analog
analogue
analyse
analyze
beginings
Bronislav
Bronislaw
Bronisław
artifacts
artefacts
behavior
behaviour

Benedict
Benedikt
Caille
Caillé
Davies
Davis
Daviss
decommodified
de-commodified
Defourney
Defourny
deproletariatization
deproletarization

Frans
Franz
Herman
Hermans
Hrdy
Malinovski
Malinowski
mercadante
mercatante
metafisico
metafico

(The colour coding is mine; I've provided it for clarity only. TextSTAT's word lists are in plain text.)

Upon checking the actual proofs, some of these issues turned out to be fine. For example: 

• The US/UK and -s/-z spelling variations did not need amending because within-chapter consistency was acceptable to the client.
• Hrdy wasn’t a typo – it’s someone’s name.
• Herman and Hermans were two different people; both spelling variants were correct.
• The hyphenation inconsistency didn’t exist – the hyphenated version was simply an end-of-line word break in the PDF.

Some issues had to be queried. For example:

• Both “deproletariatization” and “deproletarization” were used several times in close proximity, and the context didn’t make it clear whether the difference was warranted or there should be consistency. If the latter was called for, it wasn't clear which was the preferred term.

Some issues needed further checking and amending. For example:

• Defourny became Defourney
• Malinovski became Malinowski
• Caille became Caillé 
• Bronislav and Bronislaw were both changed to Bronisław
• mercadante became mercantante
• metafico became metafisico

When proofreading hard-copy or PDF proofs, would I have spotted these problems with my eye alone? I'm not confident I'd have got everything, particularly the issues with the names of the less well-known cited authors. And if "beginings" had been in point-9 italic text, my eye might have passed over the missing letter.
​

Where’s the context?

There is no context – that’s the point. When using TextSTAT as a word-list generation tool, we’re just looking at one word and how it compares with words above and below it in our list.

We’re not reading phrases; we’re not paying attention to grammar and syntax. It’s just a long list of words in alphabetical order.

​Later, we can focus on the words in context – TextSTAT’s word lists are just a tiny part of a process that help the proofreader or editor to provide his or her client with a polished piece of work.
​

Fast, free and offline

TextSTAT isn’t the only word-list generation tool available for free. However, I love it because it can handle huge chunks of text without glitching – it will quickly generate word lists for books with hundreds of thousands of words (the sample I gave above was taken from a project of over 150,000 words, but I’ve used the program for larger projects). It’s never crashed on me.

You can download the software to your own computer, so there’s no issue regarding confidentiality. My clients don’t want me to upload their content to third-party browsers without their permission, so when I use a particular proofreading tool to augment my eye, that tool needs to be able to sit offline on my PC.

Furthermore, it costs nothing. Say the creators: “TextSTAT is free software. It may be used free of charge and it may be freely distributed provided the copyright and the contents of all files, including TextSTAT.zip itself, are unmodified. Commercial distribution of the programme is only allowed with permission of the author. Use TextSTAT at your own risk; the author accepts no responsibility whatsoever. The sourcecode version comes with its own license."
​

Is it worth the effort?

Some might think that an hour or so trawling through a simple word list, and cross-checking any potential problems against hard copy or PDF, is a lot of extra time to build into a proofreading project. I think that time improves the quality of my work and increases my productivity.

When I come to the actual reading-in-context stage, I'm confident that some really serious snags have already been attended to. That gives me peace of mind and enables me later to focus on other important issues like the page layout, the sense of the text, and more.

I've found that using this method for dense academic projects has been particularly worthwhile. However, I'll not forget a recent fiction project (a "big name"-authored book that's in its nth edition and was first published over two decades ago) where the main protagonist's name was spelled incorrectly in two places: an easy thing to miss again and again over many years and many proofreads.

​I caught it – not because my eyes are better than those who came before me, or because I'm a better proofreader than those who came before me, but because I used a simple tool that allowed me to concentrate on just the words.
​

Want to try TextSTAT?

If you want to give it a spin, it’s available from NEON - NEDERLANDS ONLINE.

The usual caveat applies: generating word lists as part of the proofreading and editing process isn't the one and only true way. TextSTAT is an example of one tool that I and some of my colleagues utilize to improve the quality of our work. You might utilize different tools and different methods to achieve the same ends. All of which is great!
​

How to use TextStAT 

These instructions are correct as of 24 June 2021.
​

• Go to http://neon.niederlandistik.fu-berlin.de/en/textstat/.
• Download TextSTAT 2.9c for Windows.
• Extract the files.
• Go to TextSTAT-2.9c > TextSTAT-2 > TextSTATE.exe (I’ve pinned this file to my desktop ribbon).
• Select New Corpus.

• Create a file name, then save.

• A window will pop up saying you can add files. Choose OK.

• Select Add Local File.​

• Find your Word doc and left-click on it. Then choose Open.

• Select Show Word Frequencies and your list will appear.
  

• Export the word list. I usually use Excel because I can easily remove all the tiny words (like ‘the’ and ‘and’ that appear hundreds of times).

• Copy and paste the frequency list into a Word doc.

Louise Harnby is a line editor, copyeditor and proofreader who specializes in working with crime, mystery, suspense and thriller writers.

She is an Advanced Professional Member of the Chartered Institute of Editing and Proofreading (CIEP), a member of ACES, a Partner Member of The Alliance of Independent Authors (ALLi), and co-hosts The Editing Podcast.

• Get in touch: Louise Harnby | Fiction Editor & Proofreader
• Connect: Twitter at @LouiseHarnby, Facebook and LinkedIn
• Learn: Books and courses
• Discover: Resources for authors and editors