How text-to-speech technology can help journalists avoid copy errors

You’ve run spell-check and closely studied your story. Your editors have done the same and the copy desk — the last line of defense against mistakes — has scrutinized every word and line to ensure error-free copy.

And then the worst happens. You pick up the newspaper or open your online story. A mistake, perhaps several, jump out: misspellings, repeated words, missing ones, sources’ names spelled differently on second reference, any of several embarrassing screw-ups have made their way into publication.

You’re not alone.

Spell and grammar checkers are designed to flag misspellings, dangling modifiers, misshapen clauses and run-on sentences, but they’re far from infallible. Mistakes are easy to ignore on the page, but even more elusive on the screen where everything seems pixelly perfect.

There’s another, much more valuable, tool to cut down on creeping copy errors: Text-to-speech. TTS, which converts text into synthesized speech, adds another sense — hearing — that improves your chances of catching mistakes that your eyes miss. It’s a technological antidote to the mistakes that bedevil writers and editors, and make us look lazy, or worse, stupid. The feature is built into most computers’ operating systems. There are also third-party programs that provide the same function.

Meet “Alex,” a TTS voice that lives inside my Macbook Pro. I just select text I want him to read, hold down the control key and then tap the g key. Alex starts reading what I’ve written, or what I think I have, while I follow along on the screen. I usually plug in ear buds to block distractions.

In the three years that TTS has become part of my editing toolkit, Alex has improved my writing, bolstered accuracy and made my stories more graceful. Text-to-speech lets me hear my stories, simultaneously comparing them with the written version, allowing me to detect flaws of word choice, pacing and grammar that I can change on the fly.

When I listen carefully to Alex, he tells me when “know” should be “now.” He guides me to unnecessary sentences and paragraphs. I still rely on Word’s spell and grammar checker, but Alex always manages to find lingering mistakes. I relied on him for every word in my latest book that already had the benefit of a first-rate copy editor. Alex still found missing words, homonyms, such as “then” and “than,” and things I revised but then neglected to delete my original mistake. These days, I let Alex “edit’ my copy before I even activate spell-check.

Both Macintosh computers and computers running Windows operating software provide text-to-speech, but with varying simplicity. Text-to-speech on Macs requires selecting one checkbox in System Preferences and two keystrokes to make Alex talk.

Computer users running older Windows XP and Vista software need to select multiple options before the feature is ready to work. Fortunately, “Microsoft’s solution has improved significantly with Windows 7 and up,” Omar Schwanzer, a former member of Poynter’s technology staff, said via email.

TTS matters because copy editors are under attack by newsroom cost-cutters who have slashed copy desks and often transferred their crucial duties to editing “hubs” that process copy from multiple news outlets. These losses undermine the commitment to accuracy that news consumers demand.

Even small errors can affect a news organization’s credibility and cause readers to lose trust in us.

But it’s not just journalistic sloppiness at work. The brain conspires to keep us from getting things right. We make unconscious errors based on our kinesthetic memory that preserves motions and explains why we can ride a bicycle for the first time since childhood and, after a few wobbles, confidently pedal away. It stores keystrokes as well, which is why I habitually spell judgment with two e’s.

Procedural memory remembers rules — grammar, style, and punctuation. Writing “it’s” for “its” — those maddening misspellings on signs, menus and supposedly professional copy means that writers and editors don’t know the difference between a contraction and a possessive pronoun.

Inattention is another culprit. When we read, our eyes skip forwards and backwards over words, rapid movement known as visual saccades.

Typically, psychologists say, the brain sees the first and last letters of a word and automatically fills in the blank. That explains “then” instead of “than.” And “though” for “through.”

TTS fans include lawyers, novelist, screenwriters and educators who work with dyslexic children.

“I love the idea,” Vicki Krueger, author of News University’s “Cleaning Your Copy” course, said by phone. She believes that TTS “is especially valuable for those whose primary communication is not writing: photographers and other visual journalists, programmers and the millions who write for social media.”

Carolyn Jewel, a romance novelist, said in a testimonial to Text Aloud that hearing her work read aloud keeps her from “supplying meaning that isn’t really there. … Lots of writers recommend literally reading one’s work aloud because it’s a great way to catch clunky phrases and repetitive bits. I tried that once, but it’s pretty hard on the voice, and it still doesn’t solve the issue of your eyes and brain conspiring to ‘fix’ typos for you.”

More newsrooms would benefit from using TTS. Teresa Schmedding, president of the American Copy Editors Society, said she’s not aware of any newsrooms using it. But it’s time, NewsU’s Krueger and other copy editors say, for them to make friends with my digital editor Alex.

“We already use or should use, a dictionary, stylebook, spell checker and reference books,” Frank Fee, Jr., a veteran copy editor who taught the craft at the University of North Carolina at Chapel Hill and Ohio University, told me when I introduced him recently to Alex. “Why shouldn’t we add another tool to help us when the safety net of copy editing is frayed or vanishing?”

We have made it easy to comment on posts, however we require civility and encourage full names to that end (first initial, last name is OK). Please read our guidelines here before commenting.

  • http://www.facebook.com/chip.scanlan Chip Scanlan

    Great question, Dan. It’s one I’ve thought about a lot.
    First, let’s talk about text-to-speech speed, a crucial issue given your apt
    description of “the fractured workflow of a newsroom.”

    This morning, it took Alex, my TTS voice, 2 minutes, 45
    seconds to read a 555-word news story. Reading the story aloud myself—the
    preferred practice when editing a story—took 5 seconds more, 2 minutes 50
    seconds.

    While that doesn’t include changes that may be required,
    it’s clear that a TTS reader can more than keep up with a human voice. And, I’m
    convinced, Alex does a better job of flagging mistakes since it combines
    hearing and well as sight when editing a story.

    There’s a third issue. The process of news writing and reporting
    involves six individual steps: Idea, Focus, Reporting, Ordering, Drafting and
    Revision. But when I ask reporters, in workshops and coaching sessions, where
    they spend most of their time, they say reporting. Understandable, since we’re
    reporters and without facts a story is impossible. When I ask them when they do
    the other steps, however, they invariably reply, “As I go along.” A story
    without a sharp focus, a clear structure, a first draft, and most important,
    revision, produces “first draft journalism.” These are stories that lack a
    focus, a clear structure, the opportunity for discovery that a draft provides
    and the necessary time to make sure the story is clear, readable and accurate.

    No matter how fractured the workflow, whether it means
    filing updated blog posts or complete stories, TTS can and should play a role.
    It takes the same amount of time to read the story aloud yourself or let a TTS
    reader do so, so why not take advantage of technology that gives you a double
    shot at getting it right? Thanks for asking. I hope this adequately responds to
    your concerns.

    Chip

  • AVSalan

    Really good article and good point. Have you tried the iSpeech voices? These are the most human sounding Text To Speech voices in the industry: http://www.ispeech.org/text.to.speech

  • http://twitter.com/dancow Dan Nguyen

    it’s indisputable that copy editors are an endangered profession. It’s also indisputable that every journalist has less time on each story and more importantly, more variations of each story to file throughout the day. Let’s pretend they have time to listen to a TTL of their own story: how would that sensibly fit in the fractured workflow of a newsroom?