You’ve got auto captions on a clip and they’re accurate, but they don’t feel right. Maybe a whole sentence lands at once and the viewer is still reading when the speaker has moved on. Maybe single words pop up so fast the clip feels jittery. Most of the time the fix isn’t the font or the colour. It’s how many words each caption shows at once.
The short answer. For most short vertical video, show one to three words at a time, or one short line of about four to six words. Use one word at a time for fast, high-energy edits. Use short phrases for talking to camera. Use two lines only for calm, explanatory videos where people are actually reading along.
Why the number of words matters
On a muted feed, captions are the audio. The viewer has to read each chunk, connect it to the face or the shot, and be ready for the next one, all in the time it takes the speaker to say it.
More words per chunk means each one stays up longer, which sounds easier. But it also means the eye has to travel across a line, and on a phone that line is set big and wraps fast. A chunk of eight words on a vertical frame often becomes two or three lines, and the bottom line can end up behind the platform’s caption text and buttons.
Fewer words per chunk means less to read each time, but more changes per second. Push that too far and the screen flickers rather than reads.
Broadcast subtitles settled on a maximum of two lines. Netflix’s English style guide allows up to 42 characters per line. That works on a wide TV screen, read from a sofa. A phone held upright is a much narrower frame, and short-form captions are set far larger than TV subtitles, so the same 42 characters won’t fit on one line. That’s the main reason short video moved to smaller chunks.
The three common layouts
One word at a time
Each word appears on its own, usually large and centred, and is replaced by the next as it’s spoken. It’s the layout you see most on fast TikTok and Reels edits.
It works because one word is recognised as a shape, almost without reading. The viewer’s eyes stay in one spot, so they can watch the face and still take in the text. It also gives the edit a rhythm that matches the speech.
The cost is context. The viewer can’t glance back at the start of the sentence. With a fast speaker and long words, single words can feel like a strobe. And it puts a lot of weight on accuracy: one wrong word, shown alone and huge, is hard to miss.
Short phrases of two to five words
The caption shows a short group of words that belong together, like “the first thing” or “you need to check”. Often the word being spoken is picked out in another colour inside the phrase.
This is the most forgiving layout. Each chunk carries a bit of meaning on its own, so a viewer who looks away for a moment can pick the sentence back up. It suits talking-head videos, interviews and anything conversational.
Two lines
Two lines of caption on screen at once, with up to about ten or twelve words between them. It’s closest to traditional subtitles.
Two lines give the viewer the most context and the least flicker. They suit slower speakers, tutorials and explainers where the words matter more than the energy. They also take up the most space, which on a vertical frame means covering more of the picture or more of the area where the platform puts its own interface.
Picking by content type
There’s no single right number. It depends on how fast the person talks and what the viewer is meant to do with the words.
| Content | Words per chunk | Why |
|---|---|---|
| Fast talking-head, hooks, hype edits | 1 | Keeps pace with quick speech and holds the eye in one place |
| Everyday talking to camera | 2 to 4 | Readable at a glance, with enough context to follow |
| Interviews and podcast clips | 3 to 5 | Phrases match how people actually speak in conversation |
| Tutorials and explainers | One or two lines | Viewers are reading closely, and the pace is slower |
| Music, lyrics, voiceover over b-roll | 3 to 6 | The picture is the subject, so fewer changes help |
Speech rate matters as much as the content type. A quick speaker can comfortably say three words in under a second. If each of those chunks also has to be read, small chunks win. A slow, careful speaker gives the viewer more time, so a longer phrase stays readable.
A quick test from the earlier post on how long captions should stay on screen: count the characters in a chunk and divide by 20. That’s roughly the shortest time, in seconds, the chunk should be visible. If the speaker gets through it faster than that, the chunk is too big.
Where to break a chunk
How many words is only half of it. Where the break falls changes how a caption reads just as much.
- Break on natural pauses. End a chunk where the speaker breathes or a comma would go, not halfway through a thought.
- Keep small words with what follows. “of”, “the” and “to” read badly alone at the end of a chunk. Move them to the start of the next one.
- Never split a name, number or price. “New” then “York” reads as two words before it reads as a city.
- Start a new chunk on a cut. Carrying half a phrase across a hard cut makes the text feel out of sync even when it isn’t. Auto captions decide these breaks for you, usually from the timing of the speech. They get most of them right and a few wrong. It’s worth one pass through the clip on mute to merge or split the odd chunk before you export. The steps to add captions to a video on iPhone cover where that review fits in the flow.
Setting it up in a caption app
In most caption apps, words per chunk isn’t a number you type. It comes with the layout you pick. Capster follows that pattern: the One Word preset shows a single word at a time, while styles like Yellow Desert set two lines with the spoken word highlighted. The animated caption styles page lists every preset and its layout, so you can match one to the table above.
Placement is the other half of the decision. The more words a chunk holds, the more of the frame it covers, so check it against each platform’s interface. TikTok and Reels both stack buttons down the right edge and put the post text along the bottom, and a two-line caption set low can run straight into them. The guide to add captions to a TikTok video shows where to place them so they stay clear.
A quick checklist
- One to three words per chunk for most short-form video
- One word only for fast speech and punchy edits
- One or two lines for slow, explanatory content
- Breaks on pauses, never inside names or numbers
- Small words moved to the start of the next chunk
- Every chunk visible for at least its character count divided by 20, in seconds