Russian Religious Corpora: Tagset

One sentence in the corpus

The corpus is stored as a vertical file. Each sentence (<s>) contains one line per token with a fixed set of columns. The example below illustrates the exact token-level annotation format used in this corpus.

Example (sentence <s>)

id	word	lemma	tag	feat	head	deprel
1	Ведь	ведь	PART	_	4	advmod
2	когда	когда	SCONJ	_	4	mark
3	человек	человек	NOUN	Animacy=Anim\|Case=Nom\|Gender=Masc\|Number=Sing	4	nsubj
4	приходит	приходить	VERB	Aspect=Imp\|Mood=Ind\|Number=Sing\|Person=3\|Tense=Pres\|VerbForm=Fin\|Voice=Act	12	advcl
5	ко	ко	ADP	_	6	case
6	мне	я	PRON	Case=Dat\|Number=Sing\|Person=1\|PronType=Prs	4	obl
7	со	с	ADP	_	9	case
8	своей	свой	DET	Case=Ins\|Gender=Fem\|Number=Sing\|Poss=Yes\|PronType=Prs\|Reflex=Yes	9	det
9	болью	боль	NOUN	Animacy=Inan\|Case=Ins\|Gender=Fem\|Number=Sing	4	obl
10	,	,	PUNCT	_	4	punct
11	я	я	PRON	Case=Nom\|Number=Sing\|Person=1\|PronType=Prs	12	nsubj
12	должен	должен	ADJ	Degree=Pos\|Gender=Masc\|Number=Sing\|Variant=Short	0	root
13	войти	войти	VERB	Aspect=Perf\|VerbForm=Inf\|Voice=Act	12	xcomp
14	в	в	ADP	_	16	case
15	эту	этот	DET	Case=Acc\|Gender=Fem\|Number=Sing\|PronType=Dem	16	det
16	боль	боль	NOUN	Animacy=Inan\|Case=Acc\|Gender=Fem\|Number=Sing	13	obl
17	,	,	PUNCT	_	19	punct
18	чтобы	чтобы	SCONJ	Mood=Cnd	19	mark
19	понять	понять	VERB	Aspect=Perf\|VerbForm=Inf\|Voice=Act	13	advcl
20	,	,	PUNCT	_	23	punct
21	что	что	PRON	Animacy=Inan\|Case=Acc\|Gender=Neut\|Number=Sing\|PronType=Int,Rel	23	obj
22	он	он	PRON	Case=Nom\|Gender=Masc\|Number=Sing\|Person=3\|PronType=Prs	23	nsubj
23	чувствует	чувствовать	VERB	Aspect=Imp\|Mood=Ind\|Number=Sing\|Person=3\|Tense=Pres\|VerbForm=Fin\|Voice=Act	19	ccomp
24	.	.	PUNCT	_	12	punct

The columns shown above correspond to the token attributes used for search in the corpus. Detailed definitions of these attributes are provided below.

Truncated CoNLL-U format

UD treebanks are normally stored in CoNLL-U format, where each token is represented by multiple columns (ID, FORM, LEMMA, UPOS, XPOS, FEATS, HEAD, DEPREL, etc.).

In this corpus, each <s> sentence uses a truncated CoNLL-U representation: only the fields required for search are retained, and unused columns are omitted.

Mapping to corpus attributes

Corpus attribute	Meaning	CoNLL-U analogue
`id`	Token number within the sentence	ID
`word`	Surface word form	FORM
`lemma`	Lemma	LEMMA
`tag`	POS tag (UPOS)	UPOS
`feat`	Morphological feature bundle	FEATS
`head`	Head token ID (0 = root)	HEAD
`deprel`	Dependency relation to head	DEPREL

Annotation

Morphological annotation:

<tag> — POS tag (UD UPOS; 17 universal POS categories), e.g. NOUN (человек), VERB (приходит), ADP (в).
<feat> — morphological features (UD FEATS) encoded as Feature=Value pairs separated by |. For example: Case=Nom|Gender=Masc|Number=Sing (typical of nouns, adjectives, and pronouns), or Aspect=Imp|Mood=Ind|Tense=Pres|VerbForm=Fin|Voice=Act (typical of finite verbs). If no features are available for a token, the value is _.

<tag>: POS tags

The corpus uses 17 POS tags: ADJ, ADP, ADV, AUX, CCONJ, DET, INTJ, NOUN, NUM, PART, PRON, PROPN, PUNCT, SCONJ, SYM, VERB, X.

ADJ — adjective (e.g. большой).
ADP — adposition (preposition or postposition), e.g. в, с.
ADV — adverb, e.g. быстро.
AUX — auxiliary, e.g. бы, быть (when used as an auxiliary).
CCONJ — coordinating conjunction, e.g. и, но.
DET — determiner (pronominal modifier), e.g. этот, свой.
INTJ — interjection, e.g. увы.
NOUN — common noun, e.g. человек.
NUM — numeral, e.g. три, 10.
PART — particle, e.g. ведь, же, не.
PRON — pronoun, e.g. я, он, что.
PROPN — proper noun, e.g. Москва.
PUNCT — punctuation, e.g. ,, ..
SCONJ — subordinating conjunction, e.g. что, чтобы, когда.
SYM — symbol, e.g. %, §.
VERB — lexical verb, e.g. приходить, понять.
X — other or unknown; foreign fragments, etc.

<feat>: Morphological categories and subcategories

The corpus uses UD-style feature bundles: Feature=Value pairs separated by |, for example: Case=Nom|Gender=Masc|Number=Sing.

In UD Russian SynTagRus, the feature inventory includes, among others, Animacy, Aspect, Case, Degree, Gender, Mood, Number, Person, PronType, Tense, Variant, VerbForm, Voice, etc.

Feature bundles by POS (full lists)

Click on a POS tag to see the morphological categories and values attested in UD Russian SynTagRus.

NOUN

Animacy: Anim, Inan.
Case: Nom, Acc, Gen, Dat, Ins, Loc, Par, Voc.
Gender: Masc, Fem, Neut.
Number: Sing, Plur.

VERB

Aspect: Imp, Perf.
Mood: Ind, Imp, Cnd.
- Ind — indicative. Example: он читает.
- Imp — imperative. Example: читай!.
- Cnd — conditional. Example: он бы читал.
UD: Mood
Tense: Past, Pres, Fut.
VerbForm: Fin, Inf, Part, Conv.
- Fin — finite verb. Example: читает.
- Inf — infinitive. Example: читать.
- Part — participle. Examples: читающий, прочитанный.
- Conv — converb. Examples: читая, прочитав.
UD: VerbForm
Voice: Act, Pass, Mid.
- Act — active. Example: он читает книгу.
- Pass — passive. Examples: книга прочитана, книга была прочитана.
- Mid — middle. Examples: дверь открывается, книга читается легко.
UD: Voice
Number: Sing, Plur.
Person: 1, 2, 3.
Gender (past tense and participles): Masc, Fem, Neut.

PRON

PronType:
- Prs — personal (e.g. я, он).
- Dem — demonstrative (e.g. это, тот).
- Int — interrogative (e.g. кто?, что?).
- Rel — relative.
- Ind — indefinite.
- Neg — negative.
- Tot — total or universal.
- Emp — emphatic.
UD: PronType
Case: Nom, Acc, Gen, Dat, Ins, Loc, Par, Voc.
Number: Sing, Plur.
Gender (where applicable): Masc, Fem, Neut.
Person (personal pronouns): 1, 2, 3.
Animacy (for some pronouns): Anim, Inan.
Reflex (reflexive forms): Yes.

ADJ

Degree: Pos, Cmp, Sup.
Number: Sing, Plur.
Case: Nom, Acc, Gen, Dat, Ins, Loc, Par, Voc.
Gender: Masc, Fem, Neut.
Variant: attested; see the statistics page for the values used in this treebank.
See also: ADJ statistics, Degree, Variant.

ADP

Adpositions (mostly prepositions) are typically not inflected in UD Russian SynTagRus, so they usually have no morphological features (feat=_).

ADV

Degree: Pos, Cmp, Sup.
See also: ADV statistics.

AUX

VerbForm: Fin, Part.
Mood: Ind.
Tense: Past.
Number: Sing, Plur.
Gender: Masc, Fem, Neut.
Voice: Act.

CCONJ

Coordinating conjunctions typically have no morphological features (feat=_).

DET

Case: Nom, Acc, Gen, Dat, Ins, Loc, Par, Voc.
Gender: Masc, Fem, Neut.
Number: Sing, Plur.
PronType: e.g. Dem, Ind, Int, Rel, Tot, etc.
Poss: Yes (possessive determiners).
Reflex: Yes (e.g. reflexive possessives such as свой).

INTJ

Interjections typically have no morphological features (feat=_).

NUM

Case: Nom, Acc, Gen, Dat, Ins, Loc, Par, Voc.
Gender (where applicable): Masc, Fem, Neut.
Number (where applicable): Sing, Plur.
NumType — numeral type. Values: Card, Ord, Frac, Mult, Dist, Range, Sets.
NumForm — numeral form. Values: Word, Digit, Roman.

PART

Particles typically have no morphological features (feat=_).

PROPN

Case: Nom, Acc, Gen, Dat, Ins, Loc, Par, Voc.
Gender: Masc, Fem, Neut.
Number: Sing, Plur.
Animacy (where applicable): Anim, Inan.
NameType — named-entity subtype for proper names. Values: Geo, Prs, Giv, Sur, Com, Pro, Nat, Oth.

PUNCT

Punctuation tokens have no morphological features (feat=_).

SCONJ

Subordinating conjunctions typically have no morphological features (feat=_).

SYM

Symbols typically have no morphological features (feat=_).

The tag X is used for other or unknown tokens; morphological features are typically empty (feat=_).

Syntactic annotation:

<head> — head token ID, i.e. the syntactic governor of the current token within the same sentence. Heads form a single dependency tree for each <s>; the root token has head=0. For example, in the sample sentence, должен is the root (head=0), while войти depends on it (head=12).
<deprel> — dependency relation label to head, describing the syntactic function of the token. For example, nsubj marks subjects, case marks adpositions attached to a noun or pronoun, det marks determiners, and punct attaches punctuation.

<head>: Head index

The head column stores the ID of the syntactic head (governor) of the current token within the same sentence. The dependency structure forms a single rooted tree over the tokens of the sentence.

head is an integer pointing to another token’s id.
head=0 means that the token is the sentence root.
head and deprel must be interpreted together: deprel specifies the relation between the token and its head.

Example from the sample sentence:

должен has head=0 and is therefore the root of the sentence.
войти has head=12, so it depends on token 12 (должен) with relation xcomp.
человек has head=4, so it depends on token 4 (приходит) with relation nsubj.

<deprel>: Dependency relations

Syntactic dependencies are encoded as one dependency tree per sentence using UD relation labels. The examples below are illustrative and use the direction HEAD → DEPENDENT.

root — sentence root (typically the main predicate).
nsubj — nominal subject.
obj — object.
iobj — indirect object.
obl — oblique nominal.
advmod — adverbial modifier.
advcl — adverbial clause modifier.
ccomp — clausal complement.
xcomp — open clausal complement.
mark — marker of a subordinate clause.
case — case marker (adposition).
det — determiner.
amod — adjectival modifier.
nmod — nominal modifier.
conj — conjunct in coordination.
cc — coordinating conjunction.
punct — punctuation attachment.

See UD documentation: https://universaldependencies.org/u/dep/

Source documents

UD Russian SynTagRus treebank description: https://universaldependencies.org/treebanks/ru_syntagrus/index.html

Lyashevskaya, O. et al. (2016), Universal Dependencies for Russian: A New Syntactic Dependencies Tagset (working paper on UD for Russian and CoNLL-U fields).

Contact

name.surname [at] ur.de