Grammarly and ProWritingAid are tools that automate many of the duties that would normally fall to a copyeditor or proofreader. But are they up to the task?
What Grammarly and ProWritingAid do
Grammarly invites writers to create ‘bold, clear, mistake-free writing.’ ProWritingAid, meanwhile, is described as a ‘grammar checker, style editor, and writing mentor in one package’.
Both tools work in a similar way.
At the basic level, you can simply paste your text into the online tool. It analyses your text and highlights possible errors with spelling and grammar, along with a little description of the issue.
The tools also integrate with other software like Word, Chrome and Edge. There are a few differences here: for example, ProWritingAid integrates with Scrivener (a tool that’s popular with authors), whereas Grammarly doesn’t.
ProWritingAid also goes beyond checking spelling and grammar. It offers a readability report, amongst other things. I won’t go into detail about these extra functions. That’s because I’m looking at Grammarly and ProWritingAid as possible substitutes for an editor or proofreader, and these jobs don’t typically involve heavy rewriting.
How I tested them
As my test piece, I used a 1,500-word blog post about internet marketing and search engine optimisation (SEO). The author was a good writer but English wasn’t their first language. Several things needed to be addressed to get the article ready for publication.
First, I copyedited the post without using the tools. I did all the things a copyeditor normally does, including:
- correcting grammar and spelling errors
- introducing a consistent editorial style and compiling a style sheet
- highlighting confusing or ambiguous sentences and proposing alternatives
- checking and correcting spellings of names
- using paragraph styles to ensure the document was correctly structured.
Then I pasted the text into ProWritingAid and Grammarly to see:
- how many of the issues I had found were picked up by these tools
- if they found any errors that I’d missed
- if the tools gave any bad advice, like saying something was an error when it wasn’t.
Results
In my initial copyedit, I found 32 potential issues with the document. These either required a change (using track changes) or a comment with a suggestion about how the issue could be improved.
The issues I found included:
- commas used incorrectly, including comma splices
- inconsistent capitalisation in the body text and headings
- incorrect tense used (e.g. ‘will’ instead of ‘would’)
- confusion between quantity and volume
- semicolons used instead of colons
- a zero used instead of an O in an abbreviation
- some sentences without capital letters at the start
- syntactic issues with questions
- ambiguous or otherwise confusing phrases.
These are all fairly common issues for a copyeditor to find.
So, how did Grammarly and ProWritingAid get on with the task? Did they find all the errors I found? Did they find any I’d missed? Did they give any bad advice?
Grammarly
I pasted the original text into Grammarly and within seconds it came back with a score and a list of suggestions.
Overall, Grammarly had 3 suggestions compared to my 32. Surprisingly, there was no crossover between Grammarly’s suggestions and mine. The three things Grammarly suggested were all things that I hadn’t suggested.
Grammarly’s suggestions consisted of:
- a false positive, where Grammarly wrongly assumed that the word ‘search’ was being used as a verb rather than a noun, and therefore flagged a grammatical error that wasn’t actually present
- two suggestions for removing redundancy, of which I agreed with one and disagreed with the other.
Significantly, Grammarly didn’t flag any of these grammar and punctuation errors:
- sentences that didn’t begin with a capital letter
- question marks at the end of phrases that weren’t questions
- incorrect word order in questions (e.g. ‘How you can do…?’ instead of ‘How can you do…?’)
- comma splices, which are generally recognised as grammatical errors
- incorrect tense (‘I decided that I will…’ instead of ‘I decided that I would…’)
- a typo (‘fin’ instead of ‘find’, which wouldn’t have been found by a spell checker because both are valid words)
- an unnecessary comma after an initial adverb
- an open quotation mark with no closing quotation mark
- semicolons used instead of colons
- a non-English phrase, ‘must-to-learn’.
It also didn’t flag any points of editorial style, such as:
- non-standard capitalisation for headings
- inconsistent capitalisation of a key phrase in the article
- inconsistency in the author’s use of abbreviations as opposed to writing out the phrase.
In terms of spelling, Grammarly flagged any word that wasn’t in its dictionary, but only if that word was all lowercase. It seems to assume that any word starting with a capital letter is a name and doesn’t need to be checked. By contrast, a good copyeditor or proofreader would check the spellings of all names used in a text.
Finally, the draft article had a number of sentences that just weren’t particularly clear. Perhaps understandably (because it wasn’t actually reading the text), Grammarly didn’t highlight any of these.
ProWritingAid
ProWritingAid did a slightly better job than Grammarly at finding ‘hard’ errors in grammar, spelling and punctuation.
It found these errors that Grammarly missed:
- comma splice (flagged as ‘possible comma splice’)
- typo (‘fin’ instead of ‘find’)
- unnecessary comma after initial adverb
- no closing quotation mark after an opening quotation mark.
It didn’t pick up:
- semicolon incorrectly used instead of a colon
- inconsistent capitalisation of a key phrase in the article
- inconsistent abbreviation versus writing out
- non-standard capitalisation of headings
- the non-English phrase ‘must-to-learn’
- incorrect tense (‘I decided that I will…’ instead of ‘I decided that I would…’).
Unlike Grammarly, ProWritingAid flagged up any term that wasn’t in its dictionary, regardless of whether or not it started with a capital letter. It was happy with names like Google and Pinterest, but did flag names of less familiar software (like Semrush, for example).
As I mentioned earlier, copyeditors and proofreaders would normally check the spellings of any names used. As such, it was helpful to have ProWritingAid flag these terms as potential spelling errors – it would encourage a conscientious author to double-check.
Although both Grammarly and ProWritingAid gave a score for readability, ProWritingAid flagged a lot more potential readability issues. For example, it highlighted potentially redundant phrases such as ‘you should’, ‘in many ways’ and ‘some of the’.
It also highlighted instances of the passive voice, which Grammarly didn’t do.
ProWritingAid didn’t find any false positives for grammar, but it did make the ridiculous suggestion to replace the phrase ‘high domain authority’ (which is commonly used in the internet marketing/SEO world) with ‘eminent domain authority’.
Discussion – can you rely on Grammarly and ProWritingAid?
In this test, there was a clear winner. Grammarly failed to find any of the 32 copyediting issues I identified in the text. By contrast, ProWritingAid found four.
In terms of detecting spelling and grammar errors, ProWritingAid was undeniably the better tool.
However, both tools had a number of shortcomings.
- They missed several errors of grammar and punctuation.
- Neither tool noticed any stylistic issues such as inconsistent capitalisation.
- Both tools made suggestions that would have made the text worse.
- Several clumsy or confusing sentences passed through undetected.
So while ProWritingAid did a much better job than Grammarly, neither tool really did an adequate job.
I can think of two types of person who might want to use an automated spelling and grammar checker.
Writers might want to use it when self-editing before sending their text to an editor or proofreader. They might even hope to avoid using an editor or proofreader altogether.
The results of my test suggest that ProWritingAid and Grammarly are inadequate for this task.
There’s a huge risk, especially for writers who don’t have a firm grasp of English spelling, punctuation and grammar, of making the content worse when using ProWritingAid or Grammarly.
Editors and proofreaders might want to use it to support their work. They might want to speed up their work or double-check that they haven’t missed any errors.
I can’t see myself using Grammarly or ProWritingAid in my work as an editor. Both tools missed far too many errors and neither tool detected any significant errors that I had missed. A tool like this is beneficial if it makes an editor’s work more efficient by improving the speed at which they work. My test indicated that these tools can waste time rather than saving it.
Moreover, other tools do a better job supporting editors to do their work more efficiently. For example:
- PerfectIT helps editors (and authors) make content stylistically consistent, saving lots of time and improving accuracy when editing longer works.
- There are several Word macros for editors that can speed up the time it takes to analyse and edit a Word document.
Conclusion
If I had to choose between ProWritingAid and Grammarly, I’d definitely say ProWritingAid. But neither tool beats a careful read from a trained editor (or a careful read from someone who isn’t a trained editor) because:
- editors do far more than just checking spelling and grammar, and these tools aren’t very good at those other tasks
- they’re not particularly good at checking grammar either – both tools miss fairly significant errors
- they introduce false positives – things that aren’t actually errors, resulting in suggested changes that would actually make the text worse.
If you’re an editor thinking about using these tools to help you work more efficiently, be wary: they may actually slow you down. And if you’re a writer thinking about using them instead of working with a copyeditor or proofreader, I strongly advise against this. Over time, these tools are likely to get better – so I’ll keep an eye out in case the situation changes.