stanfordnlp/CoreNLP · error · RuntimeException

Unknown language

Error message

Unknown language <language>

What it means

readResolve() switches on the GrammaticalRelation's language to find the canonical instance for that language. The switch only handles the languages the library knows; any other value falls into the default branch and throws this RuntimeException, meaning the deserialized relation carries a language the code cannot resolve.

Solutions

1. Ensure the relation's language is one supported by this CoreNLP version (e.g. English, Chinese, UniversalEnglish...) before serializing
2. Re-serialize with matching CoreNLP versions on both ends
3. Inspect the stream for corruption if language is unexpectedly null/garbage
4. Use the string-based GrammaticalRelations.valueOf(language, name) API with an explicit valid language instead of raw serialization

Example fix

// before
GrammaticalRelation r = new GrammaticalRelation(someForeignLanguage, "dep", ...); // then serialized
// after
GrammaticalRelation r = new GrammaticalRelation(Language.English, "dep", ...);

if (language == null || !supportedLanguages.contains(language)) {
    throw new IllegalArgumentException("Unsupported language for GrammaticalRelation: " + language);
}

try {
    GrammaticalRelation r = (GrammaticalRelation) in.readObject();
} catch (RuntimeException e) {
    if (e.getMessage().startsWith("Unknown language")) {
        // rebuild relation with an explicitly supported Language constant
    } else throw e;
}

When it happens

Trigger: Deserializing a GrammaticalRelation whose language field is null or a value outside the handled enum cases (e.g. a Language constant from a different library version, or a stream corrupted so the field is wrong).

Common situations: Cross-version deserialization where the Language enum gained/lost constants; manually constructing/serializing a relation with an unhandled language; classloader conflicts yielding a mismatched Language enum.

Understand the failure class

Background: Invalid enum value errors: "Unknown type", "Invalid scope", "must be one of" — when a string is not on the library's allowed list — this error's family across 23 libraries.

Related errors

• CoreMap is actually a CoreLabel
• Could not open temporary feature index file for reading.
• Could not read string buffer fully!
• Could not save model to stream!
• ERROR: Incorrect format for the serialized coref graph

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/97fc16946f2e411c. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/trees/GrammaticalRelation.java:618

          case "nmod":
            return UniversalEnglishGrammaticalRelations.getNmod(specific);
          case "acl":
            return UniversalEnglishGrammaticalRelations.getAcl(specific);
          case "advcl":
            return UniversalEnglishGrammaticalRelations.getAdvcl(specific);
          default:
            // TODO: we need to figure out what to do with relations
            // which were serialized and then deprecated.  Perhaps there
            // is a good way to make them singletons
            return this;
          //throw new RuntimeException("Unknown English relation " + this);
        }
      } else {
        return rel;
      }

    default: {
      throw new RuntimeException("Unknown language " + language);
    }
    }
  }

  /**
   * Returns the parent of this {@code GrammaticalRelation}.
   */
  public GrammaticalRelation getParent() {
    return parent;
  }

  public static void main(String[] args) {
    final String[] names = {"dep", "pred", "prep_to","rcmod"};
    for (String name : names) {
      GrammaticalRelation reln = valueOf(Language.English, name);
      System.out.println("Data for GrammaticalRelation loaded as valueOf(\"" + name + "\"):");
      System.out.println("\tShort name:    " + reln.getShortName());
      System.out.println("\tLong name:     " + reln.getLongName());

View on GitHub (pinned to 1b7edd19c4)