Editing with Word’s Read Aloud Feature


Editing with Word’s Read Aloud Feature

After a publisher client suggested that all its editors try Word’s Read Aloud feature to help eliminate errors such as missing or repeated words, I decided to give it a shot.

I hadn’t used the feature before and suspected I’d react badly to another voice in my head while editing. But for a last look at a document in the late stages of the editing process, I’m liking it more and more.

(I touched briefly on this feature when writing about the benefits of reading aloud here.)

What Is Read Aloud?

Read Aloud is Microsoft Word’s text-to-speech function in Word 2016 (Office 365). You can access it on the Review ribbon or add it to the Quick Access Toolbar.

When you select this function, the program reads the text to you, starting where the cursor is positioned. Each word is then highlighted as it’s read.

A little control panel will also appear for jumping back a paragraph, jumping forward a paragraph, and playing/pausing the function.

A settings button lets you adjust the speed of the reading, from painfully slow to lightning fast. You can select from three options for readers: Microsoft David, Microsoft Zira, and Microsoft Mark.

I’m partial to Microsoft Zira.

I set a pace at a little over halfway across the speed bar. Without halting the function for edits, this seemed to read through approximately thirty pages of a standard format (Times New Roman, 12 pt., double spaced) romance novel in an hour.

How’s the Quality?

For the most part, Real Aloud wasn’t glitchy. At one point it switched to Microsoft David for no apparent reason (disturbing!), and at another point the synching went off between the reading and the highlighting of the words. For each case, I paused the reading and hit play again, which fixed the problem.

Zira’s voice would also periodically take on a raspy quality for the length of a paragraph, as though the program were encountering difficulty processing what was being read.

I suppose this could have something to do with connectivity or my processing speed or an underlying code for that stretch of text, but this occurred only on certain paragraphs, and each time the reading went back to a fully voiced Zira at the start of the next paragraph.

I was happy with the pace I’d set for the reading, but the program did pause (to my mind) overly long on paragraph marks, while the space between sentences seemed just right.

Zira had little trouble with most words, though she occasionally read Olivier as Oliver, and for some reason she read sun as Sunday in a number of places (though not in all instances).

The function didn’t pause at em dashes or ellipses, reading straight through in a way that a reader never would, and it read abs and expressions such as mmm as individual letters: a-b-s and m-m-m, respectively.

Zira also had comical stumbles over Airbnb (though that’s a tough one) and Liberace (for which she read the last syllable as though it were the playing card).

But the overall quality of the reading was high.

My Process

I usually have my second monitor (the one with the manuscript) turned portrait with the page at 150 percent.

For the Read Aloud pass, I turned my monitor landscape and blew the manuscript up to 200 percent. This might provide “seeing the text anew” benefits both from the larger font and from hearing the text.

As I listened, I “followed the bouncing ball” as Zira read and the words were highlighted. If I were reading as an author and not an editor, I could see a benefit in freeing yourself from the page entirely, but I was too afraid of missing homonyms or weird punctuation.


Doing this kind of reading seemed to require less mental effort (Zira doing the heavy lifting of the read), and this let me move my eye around the text a bit while Zira read. It also may have enabled me to stay fresh for a longer period.

The hope would be that lessening the mental energy of reading during the final pass would result in picking up errors that might otherwise have been missed.

While it was a different editing experience, the internal error-detection alarm that went off when encountering an error was much the same. So the same editorial sensors seem to be at work, and I can easily believe that doing this kind of reading would help prevent my mind from filling in words that I “know” are there or that I expect to be there, but which really aren’t.

An error that seems particularly illustrative is that of a man “siting at the counter,” which of course should have been “sitting at the counter.” The long i in siting jumped out in a way it might not have if reading without the audio.

Without the audio, the context might cause you to read it as “sitting” and not catch the missing t until just after you’ve read the word, whereupon you would then back up and say, “Oh, they actually have ‘siting’ there.” Or the context might cause you to read it as “sitting” and not catch the error at all, which isn’t acceptable.

Another benefit is that Read Aloud moves through the text on its own, so you don’t have to scroll or arrow through the document.

At least for me, I found that doing a read that was markedly different from previous reads on the text renewed my enthusiasm for the read-through. Maybe it’s because it’s still a new process, but I am enjoying using the feature, and those positive vibes can’t hurt.

I’d thereby list these as the benefits of using Read Aloud:

(1) Less mental drain because you’re sharing the reading load

(2) Better identification of missing and repeated words because your mind can’t automatically fill these in or gloss over them

(3) Better identification of errors because of pronunciation clues

(4) No scrolling

(5) Renewed enthusiasm for additional read-throughs


Initial edits on a document require too much hands-on-keyboard time to make it practical for early passes. Constantly pausing the reading to make edits would also grow irritating, so a last look with few expected errors is the only time I can see using the function.

The function also doesn’t seem to work when tracked changes are showing, as it reads deletions along with inserted text.

If you’re tracking changes, you’ll need to show “No Markup” to use the feature effectively.

A final note is that, while the quality of the reading is high, a word mispronounced consistently through a text could set an editor’s teeth on edge.

The following are therefore what I see as drawbacks:

(1) Only practical for final passes

(2) Does not play well with tracked changes

(3) May contain irritating pronunciation errors


While I would not use the feature on initial passes on a document, I’m enjoying Read Aloud and will continue to experiment with it for final passes.

Have you played with this feature?

Leave a Comment