stanfordnlp/CoreNLP · error · IllegalStateException

Bad WordShapeClassifier

Error message

Bad WordShapeClassifier

What it means

WordShapeClassifier.wordShape switches on the named shaper; if the shape classifier name does not map to any known scheme (DIGITS, CHINESE, CLUSTER1, etc.), the default branch throws IllegalStateException. It indicates an unknown/unsupported word-shaping scheme name was passed.

Solutions

1. Check the spelling of the shape classifier name against WordShapeClassifier's documented constants (e.g. "chinese", "cluster1", "digits")
2. Remove or correct the wordShape entry in your properties file
3. Use a known default like "dutch"/"chinese" only as documented; otherwise omit wordShape
4. Upgrade CoreNLP if the shaper name is from a newer version

Example fix

// before
props.setProperty("wordShape", "brad");
// after
props.setProperty("wordShape", "chinese");

Set<String> valid = Set.of("chinese","cluster1","digits");
if (!valid.contains(shapeName)) throw new IllegalArgumentException("unknown wordShape: " + shapeName);

try {
  shape = WordShapeClassifier.wordShape(word, shapeName);
} catch (IllegalStateException e) {
  shape = WordShapeClassifier.wordShape(word, "chinese");
}

When it happens

Trigger: Calling wordShape(str, shapeName) with a name not registered in the classifier lookup, e.g. a typo or a shaper name from a different model/props file (like a feature config specifying wordShape=foo).

Common situations: Typo in the wordShape property of an NER/CRF classifier config; using a shaper name valid in a newer/older Stanford CoreNLP version; passing a raw option string without normalization.

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

• At least two of lang (" + lang + "), openClassTags (length…
• Call next() before calling remove()!
• Could not read Regex mapping
• Either logic is broken or Gabor can't code.
• Error processing :Unknown attribute : from line

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/8b9ac92040362fb9. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/process/WordShapeClassifier.java:194

        return wordShapeJenny1(inStr, knownLCWords);
      case WORDSHAPECHRIS2:
        return wordShapeChris2(inStr, false, knownLCWords);
      case WORDSHAPECHRIS2USELC:
        return wordShapeChris2(inStr, false, knownLCWords);
      case WORDSHAPECHRIS3:
        return wordShapeChris2(inStr, true, knownLCWords);
      case WORDSHAPECHRIS3USELC:
        return wordShapeChris2(inStr, true, knownLCWords);
      case WORDSHAPECHRIS4:
        return wordShapeChris4(inStr, false, knownLCWords);
      case WORDSHAPEDIGITS:
        return wordShapeDigits(inStr);
      case WORDSHAPECHINESE:
        return wordShapeChinese(inStr);
      case WORDSHAPECLUSTER1:
        return wordShapeCluster1(inStr);
      default:
        throw new IllegalStateException("Bad WordShapeClassifier");
    }
  }

  /**
   * A fairly basic 5-way classifier, that notes digits, and upper
   * and lower case, mixed, and non-alphanumeric.
   *
   * @param s String to find word shape of
   * @return Its word shape: a 5 way classification
   */
  private static String wordShapeDan1(String s) {
    boolean digit = true;
    boolean upper = true;
    boolean lower = true;
    boolean mixed = true;
    for (int i = 0; i < s.length(); i++) {
      char c = s.charAt(i);
      if (!Character.isDigit(c)) {

View on GitHub (pinned to 1b7edd19c4)