Multi-Source Localization

Replexica's AI localization platform just got smarter with the introduction of multi-source localization. This feature allows you to leverage additional language information to refine your localizations in specific scenarios.

What is Multi-Source Localization?

Multi-source localization is a technique that uses information from multiple languages to create more accurate and context-aware localizations. It's particularly useful when dealing with languages that have different levels of information density or grammatical structures.

How It Works

The new feature introduces an extraSource field in your i18n.json config:

{
  "version": 1,
  "locale": {
    "source": "en",
    "extraSource": "fr",
    "targets": ["fr", "es", "de", "ja"]
  },
  // ... rest of your config
}

When extraSource is specified, Replexica's AI engine uses both the primary source language and the extra source language to generate localizations for your target languages.

For instance, if you're localizing from English to Spanish with French as an extra source, the AI considers both English and French content when creating the Spanish localization. This can be beneficial because French might provide additional context or nuances that English lacks, potentially leading to more precise Spanish localizations in certain cases.

When to Use Multi-Source Localization

Multi-source localization can be particularly helpful in scenarios where:

1. Your source language lacks certain grammatical features present in your target language.
2. You're dealing with idiomatic expressions that might be better captured through an intermediary language.
3. You want to leverage existing high-quality localizations in a third language to improve results for a new target language.

Remember, this feature is meant to fine-tune localizations in specific scenarios. It's not a must-use for every project, and Replexica's standard localization process already produces high-quality results for most use cases.

How to Use Multi-Source Localization

1. Set the extraSource in your i18n.json config file.

2. Use the Replexica CLI to force-retranslate your content:

   npx replexica@latest i18n --bucket json --force --locale es

   This command force-retranslates Spanish content using English as the primary source and the specified extra source.

   Note: Running npx replexica@latest i18n --bucket json --force without the --locale flag will retranslate all target languages, potentially overriding your multi-source localizations.

Under the Hood

When you use multi-source localization, Replexica's AI engine performs these steps:

1. Analyzes the content in your primary source language (e.g., English)
2. Examines the corresponding content in your extra source language (e.g., French)
3. Combines insights from both sources
4. Generates the localization in your target language (e.g., Spanish)

This process allows the AI to capture nuances that might be present in one language but not the other, potentially leading to more accurate and context-aware localizations in specific scenarios.

Best Practices

1. Choose your extra source language wisely. It should be a language that complements your primary source language in terms of information density or grammatical features.

2. Test the results. Multi-source localization might not always be necessary or beneficial for every language pair or content type.

3. Use version control for your localization files. This allows you to easily revert changes if needed.

4. Remember that multi-source localization is an optional feature. Don't feel pressured to use it if your current localization workflow is meeting your needs.

By leveraging multi-source localization when appropriate, you can further refine Replexica's AI-driven localization capabilities, potentially achieving even more nuanced results in specific scenarios.