Measuring Sovereign Speech

A longitudinal, open-source index mapping how national leaders communicate. Built on the most comprehensive open corpus of global political speech, powered by the Global Institute for National Capability.
Explore Speeches
Browse every leader in the corpus, their full speech library and the shape of their rhetoric across the six dimensions.
Speech Indices
Long-run signal across the corpus. Each dimension exposes a family of indices the user can compare.
AI Connector
Query speech tags and aggregate statistics from any LLM via the GINC MCP connector. Pull live cuts of the corpus into your own analysis.
Dimensions

Sovereign Speech Dimensions

Six lenses on how national leaders speak
Every sentence in the corpus is tagged along six independent dimensions — Time orientation, Capability posture, Agency, Reference, Expression and Stance — so the structural texture of a speech can be compared across leaders, countries and moments. Pick a dimension to see how its share has moved across the corpus and how each option holds its share year on year.

Cross-corpus comparison

10-year moving average share, rebased to 100
For the selected dimension: 10-year moving average of each option’s share of words, rebased to 100 at its earliest year.
Source. Sovereign Speech Index 2026 — Global Institute for National Capability.
Share of Total Words in Sovereign Speeches, by year of speech
100% stacked bar per year. Each segment = share of words tagged to that option, including unclassified.
Source. Sovereign Speech Index 2026 — Global Institute for National Capability.
Featured

Featured Speeches

Recent addresses from world leaders, side by side
A rotating selection of recent major addresses, each rendered as a monospace text block with coloured bars marking sentences that fit the current dimension. Most recent on the left; scroll right for more, click any speech to open its full profile.

Featured Sovereign Speeches →

Each row of bars is one wrapped line of monospace text; coloured blocks mark sentences that fit the chosen dimension.