Part-of-Speech Tagging Guidelines for the Penn Treebank Project (3rd Revision)

Summary

These are the annotator guidelines behind the Penn Treebank part-of-speech tag set (NN, NNS, JJ, VBD, VBN and the rest). The report defines each tag and, more importantly, gives decision procedures for the hard cases that make tagging inconsistent: adjective versus noun, past tense versus past participle, particle versus preposition, and similar confusions. The tests are mostly substitution and modification tests, for example whether a word can be intensified by "very" or can take a plural ending. Many short examples illustrate each decision. The document became the de facto reference for English PTB-style tagging and is still the authority behind the XPOS column of English UD treebanks.

How Mova uses it

• en/seeds/ — the English grammar seeds (short declarative rule notes with conditions, exceptions, examples and sources) cite the guidelines as the authority for PTB tag decisions; about 43 seed files reference it, mostly under en/seeds/morph/ (noun number, verb forms, adjective degree, pronouns, word formation) plus a few in nominal/ and verbal/.
• Example: Adjective used as a noun: the rich vs the natives uses the "(very) rich" test from §4.1 to separate partial conversion ("the rich" stays JJ/ADJ) from full conversion ("natives" becomes NNS/NOUN).
• The seeds combine the guidelines with classic descriptive grammars, so a tagging rule always carries both a linguistic rationale and the PTB convention it must produce.

Effectiveness in Mova

Reference material for rule writing: it defines the target conventions rather than a technique. Not measured separately.

👨‍🔬💥

Topics: Treebanks · Part-of-speech tagging

Cite

The paper itself, not this card. BibTeX from the library's record of the work.

@book{santorini1990part,
  title = {{Part-of-Speech Tagging Guidelines for the Penn Treebank Project (3rd Revision)}},
  author = {Santorini, Beatrice},
  year = {1990},
  url = {https://catalog.ldc.upenn.edu/LDC99T42}
}

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/santorini-1990-part-of-speech-tagging-guidelines-penn-treebank/