Query documents by matching ID prefix for i18n language versions

The issue with your query is that match is designed for full-text search with tokenization, not for pattern matching with wildcards. For your use case of finding documents by ID prefix, you should use the string::startsWith() function instead.

Here's how to fix your query:

*[_type == "project"] | order(_updatedAt desc) {
  _id,
  'lang': __i18n_lang, 
  title,
  'slug': slug.current,
  "translated": *[_type == "project" && string::startsWith(_id, ^._id + "__i18n_")] {
    _id,
    'lang': __i18n_lang,
    title,
    'slug': slug.current
  }
}

The key changes:

1. Use string::startsWith() instead of match - this does exact prefix matching
2. Reference the parent document's ID with ^._id to access the outer document's _id field
3. Build the prefix pattern by concatenating ^._id + "__i18n_" to match your translation suffix pattern
4. Include the _type filter in the subquery to keep it efficient

If you want to get all translations (including other language versions), you could do:

*[_type == "project"] | order(_updatedAt desc) {
  _id,
  'lang': __i18n_lang, 
  title,
  'slug': slug.current,
  "allTranslations": *[_type == "project" && string::startsWith(_id, ^._id)][_id != ^._id] {
    _id,
    'lang': __i18n_lang,
    title,
    'slug': slug.current
  }
}

This filters for documents whose _id starts with the current document's _id, then excludes the current document itself with [_id != ^._id].

The match operator tokenizes text and treats special characters as word separators, which is why it doesn't work for precise ID matching. string::startsWith() is the right tool for prefix-based string matching in GROQ.