How to Stop Subvocalizing When You Read (And Whether You Should)

If you pay attention while reading this sentence, you'll probably notice a quiet voice in your head saying the words. That's subvocalization, the internal speech that runs alongside reading, and it's the first thing every speed reading course promises to eliminate. The full story is more interesting, and more useful.

What subvocalization actually is

When you learned to read, you read aloud. Silent reading is a later skill built on top of that, and the speech machinery never fully switches off: studies using electromyography have picked up tiny activations in the larynx and lips during silent reading, and brain imaging shows speech-processing regions staying involved.

This inner speech isn't a bug. It supports comprehension, especially for difficult text, by holding words in your auditory working memory while you assemble meaning. Readers asked to suppress inner speech completely (repeating "la la la" aloud while reading, for instance) show measurably worse comprehension on complex material.

Why it still limits your speed

Speech is slow, and that's the catch. Even a fast talker manages only about 200 words per minute, so if every word you read gets fully "spoken" by your inner voice, your reading speed inherits that ceiling. It's no coincidence that the average adult reading speed (around 200–250 WPM, as we cover in our breakdown of average reading speeds) sits right at speaking pace.

So the goal isn't elimination. You want to stop fully pronouncing every word internally and let your visual system hand meaning to your brain directly. Skilled fast readers still subvocalize, but selectively: key words get a flicker of inner sound, while function words like "the" and "of" get processed visually and skipped by the voice.

What doesn't work

Two classic pieces of advice age badly:

Chewing gum or humming while reading. This ties up your speech system with noise, and comprehension drops more than speed rises.
Forcing your eyes to move faster with a finger or pacer. Pacing does help cut regressions (backtracking), which is genuinely useful, but it doesn't touch the inner voice. You just skim more.

What actually works: outrun the voice

The most reliable way to quiet subvocalization is blunt: show words faster than your inner voice can pronounce them. Your speech system can't keep up much past 300–400 WPM, so at those speeds it drops to keyword-only mode or goes quiet, while your visual word recognition, which is much faster, keeps going.

This is exactly what RSVP (Rapid Serial Visual Presentation) makes easy. Words appear one at a time at a fixed point, at a pace you set. The procedure:

Open the SpeedRead RSVP reader and paste in something easy, like a news article or light fiction.
Start at a comfortable speed, around 250–300 WPM.
Raise the speed in steps of 25–50 WPM. Somewhere between 350 and 450 WPM you'll feel the inner voice thin out: it can't pronounce every word anymore, and the words register anyway.
Stay just above that threshold for several sessions. That's your training zone.

Most people find the sensation strange for the first few minutes. Reading without the soundtrack feels like the words are arriving "directly." It settles quickly.

When you should subvocalize

Keep the inner voice for the reading it's actually good at:

Difficult technical material, where auditory working memory helps you hold a complex sentence together.
Poetry and beautiful prose, where sound is the point.
Anything you need to memorize verbatim, like quotations or definitions.

Fast quiet reading and slow voiced reading are two different tools. Having an inner voice is fine. The trouble starts when it's your only gear.

A realistic expectation

Reducing subvocalization won't quintuple your reading speed, whatever the ads claim. What it realistically buys you, combined with cutting eye-movement overhead through RSVP, is a comfortable move from about 230 WPM to 400–600 WPM on everyday material with comprehension intact. Measure your baseline with the reading speed test, train just above your voice's limit, and re-test after two weeks. The numbers will speak for themselves, quietly.