Observed congressional language, rendered as two cited automated composites.

Automated measurement. Composite prose is generated from claims selected and counted by code.

OnScript reads what members of Congress publish and counts the phrases they say together. Every claim below links to the press releases it came from.

Today on OnScript

Observed congressional language on 2026-08-27, rendered as two cited automated composites.

Democrats

Measurement: 72 publications were observed for the Democratic party on 2026-08-27.

Composite state: deterministic_fallback

Today 72 of us released statements. Our most synchronized phrase, used by 7 of us: to lower costs.

A composite voice, machine-composed from the day’s measured phrases. No member spoke these exact sentences; the quoted spans are verbatim (see receipts).

voice: deterministic template (not a language model) · verifier: passed

No talking point cleared the 3-member threshold today — nothing to cite.

Republicans

Measurement: 47 publications were observed for the Republican party on 2026-08-27.

Composite state: deterministic_fallback

Today 47 of us released statements. 4 of us carried "the men and women". Our most synchronized phrase, used by 5 of us: the next generation.

A composite voice, machine-composed from the day’s measured phrases. No member spoke these exact sentences; the quoted spans are verbatim (see receipts).

voice: deterministic template (not a language model) · verifier: passed
Receipts
4 offices · 4 publications · 4 families carried · 4 of 271 (1.5%)
✓ publication verified✓ message key✓ 4 familiesphrase shown 3/3✓ sourced 3/3
Other
  • first on the men and women who crew our ships
    Brian K. Fitzpatrick (R-PA) · 2026-08-27 · source · archived
  • advocate for the men and women who have sacrificed so
    Juan Ciscomani (R-AZ) · 2026-08-27 · source · archived
  • to serve the men and women who defend our nation
    Zachary Nunn (R-IA) · 2026-08-27 · source · archived
showing 3 of 4 members

Where language spread this window

D Democrats

the supreme court

Peak 45 offices on 2026-06-30.

Adoption curve: distinct offices using this phrase each day, by partyDemocratic offices, drawn as a solid line, peak at 45 on one day; Republican offices, drawn as a dashed line, peak at 15. The horizontal axis runs from 2026-06-01 through 2026-07-09; the vertical axis counts distinct offices, from 0 to 45.02245peak 45 · 2026-06-302026-06-012026-06-092026-06-182026-06-272026-07-09

Democrats (solid line)   Republicans (dashed line)

R Republicans

defense industrial base

Peak 33 offices on 2026-07-22.

Adoption curve: distinct offices using this phrase each day, by partyDemocratic offices, drawn as a solid line, peak at 5 on one day; Republican offices, drawn as a dashed line, peak at 33. The horizontal axis runs from 2026-06-01 through 2026-08-04; the vertical axis counts distinct offices, from 0 to 33.01633peak 33 · 2026-07-222026-06-012026-06-112026-06-262026-07-212026-08-04

Democrats (solid line)   Republicans (dashed line)

How these two were chosen: for each party independently, code takes that party's phrase with the highest single-day office count in the trailing 90 days, counting only phrases that reached at least 15 offices on some day in that window. Ties go to the earlier first appearance, then to the alphabetically first identifier. No model chooses, and neither party's pick constrains the other's.

How this is counted

Participation in eligible message claims

Each measure keeps one count unit from numerator through denominator.

D

  • Office participation: 0 offices of 55 offices. Window: party-day 2026-08-27. Method: participation-measures-v1.
  • Publication participation: 0 publications of 72 publications. Window: party-day 2026-08-27. Method: participation-measures-v1.
  • Document-family participation: 0 document families of 66 document families. Window: party-day 2026-08-27. Method: participation-measures-v1.

R

  • Office participation: 4 offices of 40 offices. Window: party-day 2026-08-27. Method: participation-measures-v1.
  • Publication participation: 4 publications of 47 publications. Window: party-day 2026-08-27. Method: participation-measures-v1.
  • Document-family participation: 4 document families of 47 document families. Window: party-day 2026-08-27. Method: participation-measures-v1.

These two bases differ. Office counts come from the day's eligible caucus offices in the nightly symmetry audit; publication counts come from the day's source publications. A joint release is one publication and can carry several offices, so the two never have to agree.

How phrases are sorted

Classification lanes

The deterministic classifier separates messages, shared names, procedure, and retained raw observations.

Messages

  • to lower costs
  • spend up to 20 million
  • the past year
  • years to come
  • access to care

Shared names

  • No observations in this lane.

Procedure

  • demanding answers from
  • 20 million on

Raw observations

  • No observations in this lane.

Raw observations remain available in the experimental exports.

Nightly symmetry audit: nightly symmetry audit. Every distilled claim above is citation-backed.

Repeated phrase observations

This table lists repeated phrase observations, ranked by how many members used the same exact phrase. It is not filtered to coordinated messages, so it can include official names, procedural formulas, and biographical phrasing, not only messages. The sparkline is the phrase's 14-day trajectory.

Each party’s own most-synchronized phrases, ranked within the party so the larger caucus can’t fill the table — every count travels with its caucus denominator.

D repeated phrases

  1. 7 of 257 (2.7%) to lower costs Phrase trend14 observations; peak 7 members.
  2. 5 of 257 (1.9%) spend up to 20 million Phrase trend2 observations; peak 5 members.
  3. 5 of 257 (1.9%) demanding answers from Phrase trend14 observations; peak 6 members.
  4. 5 of 257 (1.9%) the past year Phrase trend14 observations; peak 5 members.
  5. 5 of 257 (1.9%) years to come Phrase trend14 observations; peak 5 members.
  6. 5 of 257 (1.9%) access to care Phrase trend14 observations; peak 7 members.
  7. 5 of 257 (1.9%) proud to join Phrase trend14 observations; peak 5 members.
  8. 5 of 257 (1.9%) write to express Phrase trend14 observations; peak 6 members.
  9. 4 of 257 (1.6%) plans to spend up to 20 Phrase trend2 observations; peak 4 members.
  10. 4 of 257 (1.6%) excessive use of force Phrase trend14 observations; peak 4 members.

R repeated phrases

  1. 5 of 271 (1.8%) the next generation Phrase trend14 observations; peak 9 members.
  2. 4 of 271 (1.5%) around the world Phrase trend14 observations; peak 4 members.
  3. 4 of 271 (1.5%) the men and women who Phrase trend14 observations; peak 4 members.
  4. 3 of 271 (1.1%) economic and national security Phrase trend14 observations; peak 3 members.
  5. 3 of 271 (1.1%) of war dow Phrase trend14 observations; peak 4 members.
  6. 3 of 271 (1.1%) a unique opportunity Phrase trend14 observations; peak 3 members.
  7. 3 of 271 (1.1%) cannot afford to Phrase trend14 observations; peak 3 members.
  8. 3 of 271 (1.1%) the tax code Phrase trend14 observations; peak 3 members.
  9. 3 of 271 (1.1%) red tape and Phrase trend14 observations; peak 5 members.
  10. 3 of 271 (1.1%) next generation of Phrase trend14 observations; peak 6 members.

Each table is one party measured against that party's own caucus, so row counts are not comparable across the two. Differences in how much each party publishes are measured in the nightly symmetry audit, not here. See the nightly symmetry audit.

Browse all tracked phrases →