Constants Module

The conllu_tools.constants module provides reference data for Universal Dependencies annotations, including valid tags, relations, features, and mappings between different annotation schemes.

These constants are used internally by the validation and utility functions, but are also available for direct use in custom processing pipelines.

Universal Dependencies Tags

UPOS Tags

conllu_tools.constants.UPOS_TAGS: list[str]

List of valid Universal Part-of-Speech tags.

UPOS_TAGS = [
    'ADJ', 'ADP', 'ADV', 'AUX', 'CCONJ', 'DET', 'INTJ',
    'NOUN', 'NUM', 'PART', 'PRON', 'PROPN', 'PUNCT',
    'SCONJ', 'SYM', 'VERB', 'X'
]

Universal Dependency Relations

conllu_tools.constants.UNIVERSAL_DEPRELS: set[str]

Set of valid Universal Dependency relations.

Includes: acl, advcl, advmod, amod, appos, aux, case, cc, ccomp, clf, compound, conj, cop, csubj, dep, det, discourse, dislocated, expl, fixed, flat, goeswith, iobj, list, mark, nmod, nsubj, nummod, obj, obl, orphan, parataxis, punct, reparandum, root, vocative, xcomp

Universal Features

conllu_tools.constants.UNIVERSAL_FEATURES: set[str]

Set of Universal Dependencies morphological feature names.

Includes: Abbr, Animacy, Aspect, Case, Definite, Degree, Evident, Foreign, Gender, Mood, NumType, Number, Person, Polarity, Polite, Poss, PronType, Reflex, Tense, VerbForm, Voice

Relation Categories

conllu_tools.constants.CONTENT_DEPRELS: set[str]

Set of content (non-functional) dependency relations.

Content relations attach content words that contribute semantic meaning.

conllu_tools.constants.FUNCTIONAL_DEPRELS: set[str]

Set of functional dependency relations.

Functional relations attach function words (aux, cop, mark, det, clf, case, cc).

conllu_tools.constants.LEFT_TO_RIGHT_RELATIONS: set[str]

Relations that must always attach left-to-right in the sentence.

Includes: conj, fixed, flat, goeswith, appos

CoNLL-U Token Structure

conllu_tools.constants.TOKEN_KEYS: dict[str, str]

Mapping of CoNLL-U column names to their expected types.

TOKEN_KEYS = {
    'id': 'int',
    'form': 'str',
    'lemma': 'str',
    'upos': 'str',
    'xpos': 'str',
    'feats': 'dict',
    'head': 'int',
    'deprel': 'str',
    'deps': 'list',
    'misc': 'dict',
}

XPOS Mappings

These mappings are used for converting between morphological features and Perseus-format XPOS tags.

conllu_tools.constants.UPOS_TO_PERSEUS: dict[str, str]

Mapping from UPOS tags to Perseus XPOS first-position codes.

UPOS_TO_PERSEUS = {
    'ADJ': 'a', 'ADP': 'r', 'ADV': 'd', 'AUX': 'v',
    'CCONJ': 'c', 'DET': 'p', 'NOUN': 'n', 'NUM': 'm',
    'PART': 't', 'PRON': 'p', 'PROPN': 'n', 'PUNCT': 'u',
    'SCONJ': 'c', 'VERB': 'v', 'X': '-'
}

conllu_tools.constants.FEATS_TO_XPOS: dict[tuple[str, str], tuple[int, str]]

Mapping from (feature, value) pairs to (position, character) in Perseus XPOS.

See Utils for the complete mapping table.

conllu_tools.constants.XPOS_TO_FEATS: dict[tuple[int, str], tuple[str, str]]

Inverse mapping from (position, character) to (feature, value).

This is the inverse of FEATS_TO_XPOS.

conllu_tools.constants.VALIDITY_BY_POS: dict[int, str]

Defines which XPOS positions are valid for which POS categories.

Used by validate_xpos() to check position validity.

VALIDITY_BY_POS = {
'v',       # Position 2 (person): only valid for verbs
'nvapm',   # Position 3 (number): nouns, verbs, adj, pron, num
'v',       # Position 4 (tense): only verbs
'v',       # Position 5 (mood/verbform): only verbs
'v',       # Position 6 (voice): only verbs
'nvapm',   # Position 7 (gender): nouns, verbs, adj, pron, num
'nvapm',   # Position 8 (case): nouns, verbs, adj, pron, num
'a',       # Position 9 (degree): only adjectives
}

Treebank Concordances

These dictionaries define mappings for converting XPOS formats from specific Latin treebanks to the Perseus standard format.

conllu_tools.constants.ITTB_CONCORDANCES: dict: Mappings for Index Thomisticus Treebank XPOS conversion.

conllu_tools.constants.PROIEL_CONCORDANCES: dict: Mappings for PROIEL Treebank XPOS conversion.

conllu_tools.constants.LLCT_CONCORDANCES: dict: Mappings for Late Latin Charter Treebank XPOS conversion.

MISC Column

conllu_tools.constants.MISC_ATTRIBUTES: set[str]

Known/standard attributes for the MISC column.

Includes: SpaceAfter, Lang, Translit, LTranslit, Gloss, LId, LDeriv