WORD STUDY

data

“Something already there.”

Its grammar and neighbouring words show a material being counted, divided, worked on, governed, and made usable.

94
selected exact probes
1800—2019
fixed book corpus

What looks given enters a workflow.

EXACT BIGRAM1950—2019
DATA ARRIVES AS SOMETHING GIVEN.
1950—691.016
2005—190.263

DATA GIVEN / USES PER MILLION BIGRAMS

EXACT BIGRAM1950—2019
DATA ENTERS A SYSTEM OF COLLECTION.
1950—691.762
2005—192.179

DATA COLLECTED / USES PER MILLION BIGRAMS

EXACT BIGRAMS70 YEARS / SCROLL
DATA ARRIVES AS
SOMETHING GIVEN.
THE SAME 70 YEARS
CHANGE ORDER.
DATA ENTERS A SYSTEM
OF COLLECTION.
1950—69 AVG1.016
2005—19 AVG0.263
DECADES7
YEARS EACH10

COLLECTED DOES NOT MEAN COMPLETE.

2005—19 AVG2.179

DATA GIVENDATA COLLECTED / USES PER MILLION BIGRAMS

A plural becomes
a material.

01 / MASS94 DOTS
1940s / AMOUNT86.5
2010s / AMOUNT70.6%

MASS → CUT

01 / 06From mass grammar to divisible units

The plural form behaves like a mass noun until neighbouring words give it units. Sheets, points, sources, and types turn an apparently continuous substance into things that can be counted and compared.

A material
needs edges.

02 / CUTUSES / MILLION
0.658SHEETS0.017TYPES0.519POINTS0.176SOURCES
1950—69 → 2005—19
0.658SHEETS / MAX

CUT → PACKAGED

02 / 06From divided units to containers

Once data is divided into countable units, the vocabulary shifts toward transfer and storage. Sources and points dominate this selected late-period set, giving the material practical edges for circulation.

The material
gets containers.

03 / PACKAGED4 FORMS / 4 PERIODS
DATABASE84.0%DATABANK4.4%DATASET9.4%METADATA2.3%0%100%
1950—69
0.249FOUR-FORM TOTAL

PACKAGED → NAMED

03 / 06From containers to named qualities

Database, databank, dataset, and metadata do not grow at the same rate. Their changing shares show packaging as an historical choice: the container tells readers what belongs together before they inspect any record.

Three hundred and fifty annual phrase badges interrupt one statement, burst outward, and dissolve into five coloured terms.

5 EXACT BIGRAMS70 YEARS
sufficientavailablerawmissingsensitive
DATA LOOKS LIKE
SOMETHING
ALREADY THERE.
A FACT.
A RECORD.
A RESOURCE
READY TO USE.
AVAILABLE DATA PEAK / 19833.607
SENSITIVE / 20190.440

USES PER MILLION BIGRAMS

5 EXACT BIGRAMS70 YEARS
sufficientavailablerawmissingsensitive
DATA IS CALLED
SUFFICIENT / AVAILABLE
RAW / MISSING
OR SENSITIVE.
THESE ARE NOT
NEUTRAL DESCRIPTIONS.

A quality attached to data changes how it can be trusted, shared, governed, or used. These labels do more than describe a material: they establish expectations, permissions, and limits, helping produce the social condition in which a dataset can circulate and count as evidence for a particular reader or institution.

AVAILABLE DATA PEAK / 19833.607
SENSITIVE / 20190.440

USES PER MILLION BIGRAMS

5 EXACT BIGRAMS70 YEARS / SCROLL
sufficientavailablerawmissingsensitive
DATA LOOKS LIKE
SOMETHING
ALREADY THERE.
A FACT.
A RECORD.
A RESOURCE
READY TO USE.
DATA IS CALLED
SUFFICIENT / AVAILABLE
RAW / MISSING
OR SENSITIVE.
THESE ARE NOT
NEUTRAL DESCRIPTIONS.

A quality attached to data changes how it can be trusted, shared, governed, or used. These labels do more than describe a material: they establish expectations, permissions, and limits, helping produce the social condition in which a dataset can circulate and count as evidence for a particular reader or institution.

AVAILABLE DATA PEAK / 19833.607
SENSITIVE / 20190.440

USES PER MILLION BIGRAMS

Work gives the
material qualities.

04 / NAMED5 SERIES / 70 YEARS
SUFAVLRAWMISSEN195019832019
1950 / CURSOR
1.803AVAILABLE / USES PER M

NAMED → RULED

04 / 06From qualities to institutional rules

Sufficient, available, raw, missing, and sensitive rise along different annual paths. These labels are not decorative adjectives; each establishes an expectation about access, completeness, risk, or the work still required.

Qualities
become rules.

05 / RULEDGB × US / 2005—19
USGBPROTECTION 4.025×PERSONAL 1.845×PRIVACY 0.939×
1× PARITY / DIAGONAL
4.025×MAX GB÷US

RULED → MADE

05 / 06From rules to data work

The Google Books British English (GB) × American English (US) field shows that institutional language does not travel evenly. Distance from the parity line makes regional emphasis visible, turning each qualitative rule into a measurable difference between corpora.

Rules still depend
on work.

06 / MADE4 HISTOGRAMS
ENTRYCLEANLABELANNOT50—6970—8990—0405—19
ROW SCALE
0.093CLEANING · 2005—19

MADE → CLOSE

06 / 06From data work to legibility

Entry, cleaning, labeling, and annotation do not simply prepare data. Their histories rise at different rates, showing that a finished dataset is assembled through repeated operations that decide what can be compared, circulated, and trusted.

CLOSE FINDING

Data is not found whole.
It is made legible.

Units, operations, rules, and labour decide what becomes available for use. Each act of selection also shapes what remains absent, who can interpret it, and how institutions may treat the result as evidence.

SOURCES / RIGHTS / NAV

This study recalculates 94 selected exact-string probes from frozen source captures. The animated field compares two exact bigrams across 70 shared years in Google Books English 2019; printed-book visibility is not a population count or proof of cause.

BACK TO TOP