Internationalization vs Localization: Preparing Content, Then Adapting It

The two terms and their abbreviations

The abbreviations are numeronyms: i18n has 18 letters between the i and the n of internationalization, and l10n has 10 between the l and the n of localization. You will also see g11n for globalization, an umbrella term some organizations use for the whole effort, including business decisions about which markets to enter.

Both terms revolve around the idea of a locale: a language combined with regional conventions. Portuguese for Brazil and Portuguese for Portugal are different locales, as are Spanish for Mexico and Spanish for Spain. Each can differ in vocabulary, date formats, currency, measurement units and cultural expectations.

• Internationalization asks: can this be adapted to any locale without redesigning it?
• Localization asks: what does this particular locale need, and how do we deliver it?

Translation is one part of localization, not a synonym for it. The narrower comparison between translating a video and localizing it is covered in video translation vs localization.

Internationalization in software

In software, i18n is mostly engineering work done once, ideally early:

• Separating user-facing text from code, so strings live in resource files that translators can work on.
• Using Unicode throughout, so every script can be stored and displayed.
• Building sentences with placeholders rather than gluing fragments together, since word order differs between languages.
• Handling plural rules properly. English has two forms; some languages have one, others several.
• Formatting dates, times, numbers and currencies through locale-aware libraries instead of hard-coded patterns.
• Leaving room in layouts for longer text and supporting right-to-left scripts.
• Keeping text out of images, or making image text replaceable.

A common test is pseudo-localization: replacing strings with lengthened, accented versions to reveal hard-coded text, truncation and broken layouts before any real translation starts.

Localization in software

With that groundwork in place, localization becomes a repeatable process for each locale: translating the strings, adapting images and examples, setting local formats, adjusting legal text and payment options, and testing the result in the target language. Without the groundwork, every locale exposes the same problems again: text stuck in code, truncated buttons, dates in the wrong order.

What internationalization looks like for video

Video has no resource files, but the same principle applies: build the original so its language-dependent parts can be swapped without remaking the whole thing.

Script

Written for translation: short sentences, no wordplay that only works in one language, names and terms spelled consistently.

Audio

Dialogue recorded cleanly and kept on its own track, separate from music and effects.

On-screen text

Kept in editable graphics layers, with a textless version of the edit exported alongside the final.

Timing

Pauses around key points and room for longer translations, rather than wall-to-wall narration.

Visuals

Fewer culture-specific gestures, jokes and references, or a note of where they occur.

Files

Project files, stems, fonts and a timed source-language subtitle file stored together with clear names.

The script side is covered in depth in making videos translation-ready. Two production habits deserve emphasis because they are hard to fix afterwards.

The first is separating dialogue from music and effects. If the final mix is the only audio you keep, the original voice has to be separated out of it before a translated voice can sit over the original music. AI vocal separation can do that, but not perfectly: in dense mixes faint traces of the voice can remain and the music can sound thinner. Exporting stems at the end of an edit takes minutes and gives you a clean, fully controllable alternative: dub the dialogue stem alone and mix the translated voice with the music-and-effects stem. The guide to background music in translated videos covers both routes.

The second is keeping text out of the picture, or at least keeping a clean version. Burned-in titles, lower thirds and captions must be recreated for every language, and that is far easier with a textless version of the edit.

What localization looks like for video

Localization of a video then means choosing and producing what each audience needs: subtitles or dubbing, translated on-screen text and graphics, adapted examples or footage where cultural fit matters, a suitable voice, and translated titles, descriptions and thumbnails. Translated text often runs longer than the source, which affects subtitle reading speed and dubbing timing; text expansion in translation explains why.

How they relate: sequence, not choice

Internationalization comes first and is done once per product or production process. Localization follows and is done once per locale, and again whenever the source changes. There is no winner because they answer different questions.

The balance shifts by situation:

• A one-off video for a single extra market: light i18n is enough. Clean audio and a textless export cover most of the benefit.
• A product or course that will be localized into several languages and updated regularly: invest in i18n early, because its cost is paid once and its savings repeat for every locale and every update.
• An existing back catalog made without i18n: localize what matters most and accept some retrofit work, while changing the production process for new content.

Pitfalls and trade-offs

• Treating localization as translation only. Formats, images, examples, voice and on-screen text all count.
• Leaving i18n until the first translation request, when retrofitting costs the most.
• Over-engineering i18n for content that will never leave one market.
• Burning captions, titles and graphics into the only master you keep.
• Delivering only a final mix, so every translated version depends on separating the voice from the music instead of using a clean stem.
• Writing scripts with puns, idioms or rapid-fire narration that leave translators and voices no room.

Where mydubly fits in the localization step

mydubly is a localization-stage tool: it takes a finished video and produces translated speech and text. It cannot do your internationalization for you, but good i18n makes its output easier to use. The video localization page describes the overall workflow.

In concrete terms, mydubly's video translator detects the spoken language, translates into one of 21 languages and can dub the video with one of 8 stock voices for the whole video. It gives target-language subtitles as separate SRT and VTT files and transcripts in both languages. The translated video keeps the original picture unchanged and keeps the original music and effects: the original speech is removed by AI vocal separation and the background is mixed under the translated voice. If you exported dialogue and music-and-effects stems during production, dubbing the dialogue stem and mixing the translated voice with the music-and-effects stem in an editor still gives the cleanest, fully controllable result.

mydubly does not translate on-screen text, does not lip-sync, does not use per-speaker voices and does not import existing subtitle files. Graphics, burned-in text and cultural adaptation remain separate localization tasks, which is exactly where a textless master and editable titles pay off. For a series like the onboarding example, the training video translation page walks through a typical setup.

Next step

Before your next production, add three items to the end-of-edit checklist: export a textless version, export dialogue and music-and-effects stems, and save the script with consistent spellings of names and terms. That is the core of internationalization for video, and it makes every later localization simpler.