Why subtitles and dubbing do not say exactly the same thing

Viewers who turn on dubbed audio and subtitles in the same language may hear one sentence while reading another. That does not necessarily mean one team translated the line correctly and the other made a mistake. Subtitling and dubbing are different forms of audiovisual translation, produced for constraints that often point in different directions.

A subtitle must be read while the audience follows the image, editing, and performances. A dub replaces the original voice and must work as spoken language: it should sound natural, fit the available time, and follow when an actor opens and closes their mouth. Both versions aim to preserve meaning, intent, and personality, but they are not required to use exactly the same words.

Subtitles compete with the image for attention

A conversation can move faster than a comfortable reading pace. Subtitling therefore involves deciding what is essential, condensing structures, and arranging the text so it remains visible long enough to read. Repetition, hesitation, and information already evident on screen may be reduced when doing so does not change the meaning.

Rules vary by company, language, and type of content. Netflix’s Brazilian Portuguese guide, for example, sets a limit of 42 characters per line. The platform’s general requirements recommend no more than two lines and line breaks at natural syntactic points. Making the sentence fit is only part of the task: a poor break forces viewers to reconstruct the grammar while the scene continues.

Timing matters as well. Accessibility best practices connect each caption block to the moment when the relevant sound occurs and avoid both flashes that disappear too quickly and text that remains after the speech has moved on. In dialogue-heavy scenes, a literal translation may read well on a page and still fail on screen.

Dubbing has to become a performance

A dubbed script faces another set of constraints. The translated line should begin and end near the visible mouth movements, respect pauses and reactions, and match the actor’s energy. Netflix’s dubbing guidelines instruct teams to edit against the image rather than relying only on an audio waveform, while keeping the target-language dialogue natural.

That requirement explains apparently small changes. A brief answer in the source language may need more syllables in English; an exact translation may continue after the character’s mouth closes; a word order that works in writing may sound artificial when spoken. The adapter looks for a solution that carries the same dramatic function within the available time.

Lip synchronization does not mean matching every mouth shape with mathematical precision. Open vowels, highly visible consonants, and close-up shots require special attention, but acting, direction, and mixing also support the illusion. In animation, the necessary degree of adjustment depends on how closely the mouth movements were drawn for specific sounds or syllables.

The two versions can come from separate workflows

Subtitle translators and dubbing adapters may begin with the same script, dialogue list, and reference material, but they deliver different products. The American Translators Association describes adapting a translated script for voice actors as an additional stage in the dubbing workflow. A subtitle track is therefore not necessarily a transcript of the dubbed audio.

It is also important to distinguish subtitles, subtitles for the deaf and hard of hearing, and closed captions. A dialogue subtitle can record only what is said. SDH or CC versions add information needed by viewers who cannot hear the soundtrack, including speaker identification, music, and relevant effects. When a catalog offers more than one option in the same language, its label helps explain what each track is designed to provide.

How to judge whether a difference is acceptable

Comparing isolated words is rarely enough. The main question is whether both versions preserve facts, relationships, humor, tone, and intent. A joke may require a different construction; an honorific may have no direct equivalent; a recurring franchise term may need to follow approved terminology. Those choices are adaptation, not permission to invent material.

An error occurs when a version changes who performed an action, alters a date, turns uncertainty into certainty, weakens a threat, or loses information required to understand the story. Differences caused by the medium’s constraints can preserve the function of the line. Differences that contradict the work are not an inevitable result of subtitling or dubbing.

There is consequently no universally “more faithful” option. Subtitles preserve the original voices but must remain concise. Dubbing frees the viewer from reading and can accommodate more spoken material, but it faces synchronization and performance constraints. Understanding those tradeoffs makes it easier to evaluate each version according to what it actually has to accomplish.