This is part of archived UD v1 documentation. See http://universaldependencies.org/ for the current version.
home cs/pos issue tracker

PUNCT: punctuation

Definition

Punctuation marks are non-alphabetical characters and character groups used to delimit linguistic units in printed text.

Punctuation is not taken to include logograms such as $, %, and §, which are instead tagged as SYM.

Examples

Diffs

Prague Dependency Treebank

The PDT texts are from the early 1990s and there are no e-mail addresses. If they were there, the PDT tokenization rules would break them up on all dots and at signs. The same holds for telephone numbers. For example, tel.: (05) 4321 6014  is analyzed as eight tokens (NOUN PUNCT PUNCT PUNCT NUM PUNCT NUM NUM).

References

Treebank Statistics (UD_Czech)

There are 26 PUNCT lemmas (0%), 27 PUNCT types (0%) and 221036 PUNCT tokens (15%). Out of 17 observed tags, the rank of PUNCT is: 14 in number of lemmas, 15 in number of types and 2 in number of tokens.

The 10 most frequent PUNCT lemmas: ,, ., -, “, ), (, :, ?, /, *

The 10 most frequent PUNCT types: ,, ., -, “, ), (, :, ?, /, *

The 10 most frequent ambiguous lemmas: / (PUNCT 700, SYM 18), * (PUNCT 655, SYM 10), o (ADP 10328, PUNCT 100, NOUN 10, ADJ 3, INTJ 2), x (SYM 120, NOUN 70, PUNCT 3)

The 10 most frequent ambiguous types: / (PUNCT 700, SYM 18), * (PUNCT 655, SYM 10), o (ADP 9669, ADJ 110, PUNCT 99, NOUN 4), x (SYM 120, NOUN 41, PUNCT 3)

Morphology

The form / lemma ratio of PUNCT is 1.038462 (the average of all parts of speech is 2.195930).

The 1st highest number of forms (2) was observed with the lemma “-”: -, -103.

The 2nd highest number of forms (1) was observed with the lemma “!”: !.

The 3rd highest number of forms (1) was observed with the lemma “””: ”.

PUNCT does not occur with any features.

Relations

PUNCT nodes are attached to their parents using 3 different relations: cs-dep/punct (220973; 100% instances), cs-dep/root (34; 0% instances), cs-dep/conj (29; 0% instances)

Parents of PUNCT nodes belong to 18 different parts of speech: VERB (126089; 57% instances), NOUN (40365; 18% instances), PROPN (19520; 9% instances), ADJ (17744; 8% instances), NUM (11542; 5% instances), ADV (2977; 1% instances), PRON (1657; 1% instances), PART (515; 0% instances), PUNCT (161; 0% instances), DET (129; 0% instances), CONJ (79; 0% instances), SYM (72; 0% instances), INTJ (68; 0% instances), ADP (40; 0% instances), ROOT (34; 0% instances), SCONJ (31; 0% instances), X (11; 0% instances), AUX (2; 0% instances)

220947 (100%) PUNCT nodes are leaves.

33 (0%) PUNCT nodes have one child.

33 (0%) PUNCT nodes have two children.

23 (0%) PUNCT nodes have three or more children.

The highest child degree of a PUNCT node is 23.

Children of PUNCT nodes are attached using 11 different relations: cs-dep/punct (158; 63% instances), cs-dep/dobj (23; 9% instances), cs-dep/conj (20; 8% instances), cs-dep/nmod (18; 7% instances), cs-dep/nsubj (12; 5% instances), cs-dep/nummod (7; 3% instances), cs-dep/advmod (5; 2% instances), cs-dep/dep (4; 2% instances), cs-dep/cc (3; 1% instances), cs-dep/ccomp (1; 0% instances), cs-dep/parataxis (1; 0% instances)

Children of PUNCT nodes belong to 8 different parts of speech: PUNCT (161; 64% instances), NOUN (48; 19% instances), NUM (26; 10% instances), ADV (5; 2% instances), PROPN (5; 2% instances), CONJ (3; 1% instances), ADJ (2; 1% instances), VERB (2; 1% instances)

Treebank Statistics (UD_Czech-CAC)

There are 10 PUNCT lemmas (0%), 10 PUNCT types (0%) and 60127 PUNCT tokens (12%). Out of 16 observed tags, the rank of PUNCT is: 13 in number of lemmas, 14 in number of types and 3 in number of tokens.

The 10 most frequent PUNCT lemmas: ,, ., -, “, :, (, ), ?, ‘, /

The 10 most frequent PUNCT types: ,, ., -, “, :, (, ), ?, ‘, /

The 10 most frequent ambiguous lemmas:

The 10 most frequent ambiguous types:

Morphology

The form / lemma ratio of PUNCT is 1.000000 (the average of all parts of speech is 2.206260).

The 1st highest number of forms (1) was observed with the lemma “””: ”.

The 2nd highest number of forms (1) was observed with the lemma “’”: ’.

The 3rd highest number of forms (1) was observed with the lemma “(”: (.

PUNCT occurs with 1 features: cs-feat/Abbr (1; 0% instances)

PUNCT occurs with 1 feature-value pairs: Abbr=Yes

PUNCT occurs with 2 feature combinations. The most frequent feature combination is _ (60126 tokens). Examples: ,, ., -, “, :, (, ), ?, ‘, /

Relations

PUNCT nodes are attached to their parents using 2 different relations: cs-dep/punct (60119; 100% instances), cs-dep/conj (8; 0% instances)

Parents of PUNCT nodes belong to 16 different parts of speech: VERB (37540; 62% instances), NOUN (13332; 22% instances), ADJ (5562; 9% instances), PROPN (1440; 2% instances), ADV (925; 2% instances), PRON (625; 1% instances), NUM (323; 1% instances), SYM (200; 0% instances), PART (68; 0% instances), SCONJ (50; 0% instances), DET (28; 0% instances), CONJ (11; 0% instances), AUX (10; 0% instances), INTJ (9; 0% instances), ADP (2; 0% instances), PUNCT (2; 0% instances)

60123 (100%) PUNCT nodes are leaves.

1 (0%) PUNCT nodes have one child.

1 (0%) PUNCT nodes have two children.

2 (0%) PUNCT nodes have three or more children.

The highest child degree of a PUNCT node is 4.

Children of PUNCT nodes are attached using 7 different relations: cs-dep/case (2; 20% instances), cs-dep/dep (2; 20% instances), cs-dep/punct (2; 20% instances), cs-dep/acl (1; 10% instances), cs-dep/advcl (1; 10% instances), cs-dep/mark (1; 10% instances), cs-dep/nmod (1; 10% instances)

Children of PUNCT nodes belong to 6 different parts of speech: ADV (2; 20% instances), CONJ (2; 20% instances), PUNCT (2; 20% instances), VERB (2; 20% instances), SCONJ (1; 10% instances), SYM (1; 10% instances)

Treebank Statistics (UD_Czech-CLTT)

There are 26 PUNCT lemmas (1%), 26 PUNCT types (1%) and 3994 PUNCT tokens (11%). Out of 15 observed tags, the rank of PUNCT is: 8 in number of lemmas, 10 in number of types and 3 in number of tokens.

The 10 most frequent PUNCT lemmas: ,, ., ;, (, ), (1), (2), (3), (4), (5)

The 10 most frequent PUNCT types: ,, ., ;, (, ), (1), (2), (3), (4), (5)

The 10 most frequent ambiguous lemmas: / (SYM 2, PUNCT 2)

The 10 most frequent ambiguous types: / (SYM 2, PUNCT 2)

Morphology

The form / lemma ratio of PUNCT is 1.000000 (the average of all parts of speech is 1.764161).

The 1st highest number of forms (1) was observed with the lemma “”*“”: “*”.

The 2nd highest number of forms (1) was observed with the lemma “(”: (.

The 3rd highest number of forms (1) was observed with the lemma “(1)”: (1).

PUNCT does not occur with any features.

Relations

PUNCT nodes are attached to their parents using 2 different relations: cs-dep/punct (3991; 100% instances), cs-dep/conj (3; 0% instances)

Parents of PUNCT nodes belong to 11 different parts of speech: VERB (2157; 54% instances), NOUN (1273; 32% instances), ADJ (314; 8% instances), X (126; 3% instances), NUM (86; 2% instances), ADV (16; 0% instances), PART (8; 0% instances), PRON (8; 0% instances), SYM (4; 0% instances), DET (1; 0% instances), SCONJ (1; 0% instances)

3994 (100%) PUNCT nodes are leaves.

The highest child degree of a PUNCT node is 0.

PUNCT in other languages: [bg] [cs] [de] [el] [en] [es] [eu] [fa] [fi] [fr] [ga] [he] [hu] [it] [ja] [ko] [sv] [u]