languagetool-org/languagetool · error

Could not tag and disambiguate '<token>'

Error message

Could not tag and disambiguate '<token>'

What it means

NoDisambiguationIrishPartialPosTagFilter.tag calls the Irish tagger without disambiguation and rethrows any IOException as this RuntimeException identifying the token. Despite the message wording, no disambiguation step runs here — the failure is purely in tagging (dictionary/resource I/O).

Source

Thrown at languagetool-language-modules/ga/src/main/java/org/languagetool/rules/ga/NoDisambiguationIrishPartialPosTagFilter.java:47

import java.util.List;

/**
 * A {@link PartialPosTagFilter} for Irish that does not run the disambiguator.
 * @since 3.2
 * @see IrishPartialPosTagFilter
 */
public class NoDisambiguationIrishPartialPosTagFilter extends PartialPosTagFilter {

  private final Tagger tagger = Irish.getInstance().getTagger();

  @Override
  protected List<AnalyzedTokenReadings> tag(String token) {
    try {
      List<AnalyzedTokenReadings> tags = tagger.tag(Collections.singletonList(token));
      AnalyzedTokenReadings[] atr = tags.toArray(new AnalyzedTokenReadings[tags.size()]);
      return Arrays.asList(atr);
    } catch (IOException e) {
      throw new RuntimeException("Could not tag and disambiguate '" + token + "'", e);
    }
  }
}

View on GitHub (pinned to 2e990059ce)

Solutions

1. Inspect the wrapped IOException cause to find the unreadable/missing resource and restore it.
2. Verify the Irish language module (dictionaries) is fully present on the classpath.
3. Fix filesystem permissions on the LanguageTool resource directory.
4. Consider caching the tagger instance and pre-warming resource loading at startup to surface resource problems early.

Example fix

// before
try { return tagger.tag(...); } catch (IOException e) { throw new RuntimeException("Could not tag...", e); }
// after (fail fast at init)
@FunctionalInterface interface SafeTag { List<AnalyzedTokenReadings> apply(String t); }
// validate resources once in constructor: tagger.tag(List.of("tá")) before accepting traffic

List<AnalyzedTokenReadings> probe = tagger.tag(List.of("tá"));
if (probe == null || probe.isEmpty()) throw new IllegalStateException("Irish tagger not operational");

try { return tag(token); } catch (RuntimeException e) { log.error("Tagging failed for '" + token + "'", e.getCause()); return Collections.emptyList(); }

When it happens

Trigger: tagger.tag(Collections.singletonList(token)) throws IOException — missing or unreadable Irish tagger dictionary resources on the classpath while a POS filter processes a token.

Common situations: Incomplete or custom LanguageTool deployments lacking ga resources; classloader issues in embedded/plugin environments; corrupted resource files after partial upgrades.

Understand the failure class

Background: "failed to read file", EACCES, ENOENT and "could not read <path>" errors: when a program can't read a file from disk — this error's family across 49 libraries.

Related errors

• Could not tag and disambiguate '<token>'
• Could not load coherency data from " + path
• Could not tag and disambiguate '" + token + "'
• Error analyzing sentence: '" + match.getSentence().getText()
• Language was already instantiated, see the cause stacktrace

AI-assisted analysis of languagetool-org/languagetool@2e990059ce (2026-09-06). Data as JSON: /api/errors/88ab02cca8a094ee. Report an issue: GitHub.