Speaking vs. typing: what the research actually says about writing speed

"Speaking is three times faster than typing" is one of those claims that gets repeated so often it starts to sound like folklore. It isn't. It's backed by real research. But the full picture is more interesting than the headline, and it explains why fast dictation alone has never been enough.

How fast do we actually type?

Some of the best data on typing speed comes from very large online studies.

In 2018, researchers at Aalto University and the University of Cambridge analysed 136 million keystrokes from around 168,000 volunteers typing on physical keyboards (Dhakal et al., Observations on Typing from 136 Million Keystrokes, CHI 2018). The average speed was about 52 words per minute. The fastest participants exceeded 120, but they were rare, and most people were clustered well below that.

A follow-up study of mobile typing with around 37,000 volunteers found an average of about 36 words per minute on smartphones (Palin et al., MobileHCI 2019). Interestingly, the fastest mobile typists used two thumbs and leaned heavily on autocorrect.

So for most people, realistic typing speed is somewhere between 35 and 55 words per minute, depending on the device.

How fast do we speak?

Natural conversational English is commonly measured at somewhere between 120 and 160 words per minute, with many speakers around 150. Professional narration, such as audiobooks, tends to sit in a similar range because it's paced for comfortable listening.

Put those numbers side by side and the gap is obvious: speech runs at roughly three times typical typing speed.

The study that tested it directly

Comparing average speeds from different studies only goes so far. The most cited direct comparison comes from a 2016 collaboration between Stanford University, the University of Washington and Baidu (Ruan et al., published in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies).

Participants entered short messages on a smartphone, either by typing on the built-in keyboard or by dictating with a then-state-of-the-art speech recognition system. The results:

• In English, speech input was about 3× faster than typing.
• In Mandarin Chinese, it was about 2.8× faster.
• Error rates were also lower with speech: roughly 20% lower in English and over 60% lower in Mandarin.

That last point surprised many people at the time. Speech recognition had crossed a threshold: it was no longer just faster but also more accurate than people's thumbs.

Recognition has improved substantially in the years since.

So why isn't everyone dictating?

If speech is faster and more accurate, why do most of us still type? The research hints at the answer: those studies measured short, deliberate messages. Real writing is different.

We don't speak in finished sentences

Spontaneous speech is full of disfluencies: fillers like "um" and "uh", repetitions, restarts and mid-sentence corrections. Linguists who study conversational speech commonly find several of these in every hundred words. In conversation we barely notice them. In a transcript they're glaring.

Compare what someone says:

"so um I think we should, we should probably move the launch to, uh, let's say the second week, because the the design review isn't done"

…with what they meant to write:

"I think we should move the launch to the second week, because the design review isn't finished."

A faithful transcript gives you the first version. You wanted the second.

Editing eats the time you saved

When dictation produces raw transcripts, the speed advantage is real but often temporary. You saved two minutes speaking, then spent three cleaning up punctuation, filler and phrasing. For many people, that trade-off simply isn't worth it.

Speaking takes getting used to

There's a learning curve, too. Most of us have decades of practice composing at a keyboard and very little composing out loud. Early dictation sessions can feel awkward, which is why a few simple habits make such a difference.

Where the real gain comes from

The lesson from the research isn't just "speech is faster". It's that speech is faster to produce, while writing is slower to produce but faster to read. The opportunity is to get the best of both: to speak at the speed of thought and receive text that reads like considered writing.

That's the gap AI cleanup is designed to close. Instead of transcribing exactly what you said, modern language models can preserve what you meant: removing disfluencies, restoring sentence structure, punctuating by meaning, and keeping the names and terminology that matter. We wrote about how that works in Beyond transcription.

When editing time drops close to zero, the threefold speed advantage stops being theoretical.

What this means for you

• Use voice for first drafts. Speed matters most when you're getting ideas down. Speak freely, then refine.
• Expect a short adjustment period. A few days of regular use is usually enough for dictation to stop feeling strange.
• Judge tools by the output, not the transcript. The question isn't "did it hear every word?" but "could I send this without editing?"
• Don't abandon the keyboard. Typing is still great for precise edits, code and short replies. The goal is to use each tool for what it's good at.

Typing has been the default way to write for so long that it's easy to forget it was always a compromise. The research says we can do better, and the technology finally makes it practical.

---

Sources: Dhakal, V., Feit, A. M., Kristensson, P. O., & Oulasvirta, A. (2018). Observations on Typing from 136 Million Keystrokes. CHI '18. · Palin, K., Feit, A. M., Kim, S., Kristensson, P. O., & Oulasvirta, A. (2019). How do People Type on Mobile Devices? Observations from a Study with 37,000 Volunteers. MobileHCI '19. · Ruan, S., Wobbrock, J. O., Liou, K., Ng, A., & Landay, J. A. (2018). Comparing Speech and Keyboard Text Entry for Short Messages in Two Languages on Touchscreen Phones. Proc. ACM IMWUT 1(4).