Dataset Viewer
Auto-converted to Parquet Duplicate
corpus
stringclasses
1 value
year
int64
2k
2.03k
slot
stringclasses
4 values
word
stringlengths
1
180
⌀
lemma
stringlengths
1
180
⌀
male_count
int64
0
126k
female_count
int64
0
43.3k
total
int64
1
162k
guardian
1,999
agent
'
'
7
4
11
guardian
1,999
agent
'd
would
7
6
13
guardian
1,999
agent
's
be
63
13
76
guardian
1,999
agent
-
-
38
11
49
guardian
1,999
agent
-along
-along
1
0
1
guardian
1,999
agent
[
[
1
0
1
guardian
1,999
agent
abandon
abandon
19
5
24
guardian
1,999
agent
abandond
abandon
1
0
1
guardian
1,999
agent
abandoned
abandon
166
36
202
guardian
1,999
agent
abandoning
abandon
12
2
14
guardian
1,999
agent
abandons
abandon
12
2
14
guardian
1,999
agent
abashed
abash
1
0
1
guardian
1,999
agent
abdicate
abdicate
2
3
5
guardian
1,999
agent
abdicated
abdicate
1
1
2
guardian
1,999
agent
abdicates
abdicate
3
0
3
guardian
1,999
agent
abdicating
abdicate
0
1
1
guardian
1,999
agent
abducted
abduct
11
1
12
guardian
1,999
agent
abducting
abduct
1
0
1
guardian
1,999
agent
abhor
abhor
1
0
1
guardian
1,999
agent
abhorred
abhor
3
0
3
guardian
1,999
agent
abhors
abhor
3
1
4
guardian
1,999
agent
abide
abide
16
3
19
guardian
1,999
agent
abided
abide
1
0
1
guardian
1,999
agent
abides
abide
1
0
1
guardian
1,999
agent
abjures
abjure
1
0
1
guardian
1,999
agent
abode
abide
1
0
1
guardian
1,999
agent
abolish
abolish
2
0
2
guardian
1,999
agent
abolished
abolish
17
4
21
guardian
1,999
agent
abolishes
abolish
5
0
5
guardian
1,999
agent
abolishing
abolish
2
0
2
guardian
1,999
agent
abolition
abolition
1
0
1
guardian
1,999
agent
abort
abort
0
2
2
guardian
1,999
agent
aborted
abort
4
0
4
guardian
1,999
agent
abote
abote
1
0
1
guardian
1,999
agent
abound
abound
1
0
1
guardian
1,999
agent
abounded
abound
0
1
1
guardian
1,999
agent
about
about
0
1
1
guardian
1,999
agent
absconded
abscond
8
0
8
guardian
1,999
agent
absconding
abscond
1
0
1
guardian
1,999
agent
abseiled
abseil
1
0
1
guardian
1,999
agent
abseiling
abseil
3
0
3
guardian
1,999
agent
absence
absence
1
0
1
guardian
1,999
agent
absented
absent
3
0
3
guardian
1,999
agent
absents
absent
1
0
1
guardian
1,999
agent
absolve
absolve
4
0
4
guardian
1,999
agent
absolved
absolve
4
0
4
guardian
1,999
agent
absolves
absolve
1
0
1
guardian
1,999
agent
absorb
absorb
7
2
9
guardian
1,999
agent
absorbed
absorb
28
4
32
guardian
1,999
agent
absorbing
absorb
3
0
3
guardian
1,999
agent
absorbs
absorb
0
1
1
guardian
1,999
agent
abstain
abstain
2
0
2
guardian
1,999
agent
abstained
abstain
7
0
7
guardian
1,999
agent
abstracts
abstract
1
1
2
guardian
1,999
agent
abuse
abuse
2
0
2
guardian
1,999
agent
abused
abuse
54
6
60
guardian
1,999
agent
abuses
abuse
2
0
2
guardian
1,999
agent
abusing
abuse
11
4
15
guardian
1,999
agent
acccepted
acccept
1
0
1
guardian
1,999
agent
accede
accede
4
0
4
guardian
1,999
agent
acceded
accede
9
0
9
guardian
1,999
agent
acceding
accede
1
0
1
guardian
1,999
agent
accelerate
accelerate
3
0
3
guardian
1,999
agent
accelerated
accelerate
17
0
17
guardian
1,999
agent
accelerates
accelerate
0
1
1
guardian
1,999
agent
accelerating
accelerate
3
1
4
guardian
1,999
agent
accented
accent
1
0
1
guardian
1,999
agent
accentuated
accentuate
2
0
2
guardian
1,999
agent
accept
accept
323
40
363
guardian
1,999
agent
accepted
accept
537
77
614
guardian
1,999
agent
accepting
accept
14
4
18
guardian
1,999
agent
accepts
accept
185
26
211
guardian
1,999
agent
access
access
0
2
2
guardian
1,999
agent
accessed
access
1
1
2
guardian
1,999
agent
accessorised
accessorise
0
1
1
guardian
1,999
agent
accessorises
accessorise
0
2
2
guardian
1,999
agent
acclaimed
acclaim
11
0
11
guardian
1,999
agent
acclaiming
acclaim
2
0
2
guardian
1,999
agent
acclaims
acclaim
1
0
1
guardian
1,999
agent
accommodate
accommodate
2
0
2
guardian
1,999
agent
accommodated
accommodate
1
0
1
guardian
1,999
agent
accommodating
accommodate
1
0
1
guardian
1,999
agent
accommpanied
accommpany
1
0
1
guardian
1,999
agent
accompanied
accompany
57
25
82
guardian
1,999
agent
accompanies
accompany
9
4
13
guardian
1,999
agent
accompany
accompany
11
7
18
guardian
1,999
agent
accompanying
accompany
7
1
8
guardian
1,999
agent
accomplish
accomplish
4
0
4
guardian
1,999
agent
accomplished
accomplish
15
5
20
guardian
1,999
agent
accomplishes
accomplish
3
0
3
guardian
1,999
agent
accorded
accord
6
0
6
guardian
1,999
agent
according
accord
1
1
2
guardian
1,999
agent
accosted
accost
6
0
6
guardian
1,999
agent
accosting
accost
1
0
1
guardian
1,999
agent
accosts
accost
0
1
1
guardian
1,999
agent
account
account
12
2
14
guardian
1,999
agent
accounted
account
15
0
15
guardian
1,999
agent
accounts
account
2
0
2
guardian
1,999
agent
accreted
accrete
2
0
2
guardian
1,999
agent
accrued
accrue
1
0
1
End of preview. Expand in Data Studio

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

guardian_attribute_counts

Word counts attached to male and female characters in The Guardian (1999-2026), per year and BookNLP slot.

attribute_counts_by_year.csv: corpus, year, slot, word, lemma, male_count, female_count, total

  • slot: agent (verbs the character performs), patient (verbs applied to them), mod (modifiers/adjectives), poss (possessed nouns).
  • word: lowercased BookNLP token, as extracted (said, says, wives).
  • lemma: lemma of word, with the slot as part-of-speech hint (agent/patient → verb, mod → adjective then noun, poss → noun). For example said/says → say, wives → wife, happier → happy, ’s → be. Made with LemmInflect's lexicon plus a few hand rules (contractions, decades like 30s, plural-only nouns); see lemmatize.py. Words are lemmatized without context, so true homographs are resolved by the most frequent reading. To work at lemma level, group by year, slot, lemma and sum the counts.
  • Character gender: from pronouns in the coreference chain, honorifics and first names (name table learned from the corpus); any male/female disagreement is dropped (see char_table/gender_en.py, --cues all).
  • Sections are summed out. Same layout as regender/attribute_counts_by_rubrique.csv (French press), minus rubrique and author gender.

fasttext_vectors.npz: fastText cc.en.300 vectors (float16) for every lemma in the data (words, vectors), so the notebook runs without the 7 GB model.

semantic_axis_gender.ipynb: teaching notebook. It builds a fastText semantic axis (or category) from seed words and measures whether it is male- or female-skewed and how that changes over time.

slot distinct tokens distinct lemmas
agent 41,439 22,119
patient 26,142 13,024
mod 34,471 30,382
poss 105,616 91,360
Downloads last month
49