Diffusion by Proximity: The Emergence and Spatial Spread
of Environmental Discourse in the UN General Debate

Project Overview

Environmental discourse is now a permanent feature of the global debate, voiced by nearly every state at the most visible annual gathering of the world's governments. It was not always so. In the early 1960s the environment was almost absent from the United Nations General Debate. This project traces its rise from the margins to the center across six decades, and asks which structure organized the spread: geographic proximity, in which states adopt the discourse of their neighbors, or economic exchange, in which trade and investment ties condition what states say.

The analysis embeds the full United Nations General Debate Corpus, splitting roughly two hundred eighteen thousand speech windows into nine literature-anchored discourse registers and measuring the environmental share of each country-year. A two-way fixed-effects spatial panel then tests whether a country's environmental emphasis follows the prior environmental emphasis of its geographic neighbors. It does, strongly and consistently. Economic measures of integration leave that spatial effect intact. The spread of environmental discourse was a spatial phenomenon, not an economic one.

Measuring Environmental Discourse

Each speech is divided into fixed windows of about one hundred fifty words and embedded with a sentence transformer. Every window is scored against nine discourse registers, from development and welfare to sovereignty, security, human rights, and the environment, using centered cosine similarity so that no single register dominates by length. The environmental share of a country-year is the count of environmental windows over all classified windows.

The Rise of Environmental Discourse, 1962 to 2024

Over six decades, environmental discourse moved from the margins of the General Debate to its center. The series below traces the environmental share of classified discourse across the full period. Through the 1960s, 1970s, and most of the 1980s the environment had a marginal presence, while security, global inequality, and the politics of sovereignty dominated the agenda. A first spike appears around 1990, coinciding with the lead-up to the 1992 Rio Earth Summit, but environmental discourse recedes through the rest of the decade. The sustained rise begins after 2000, tracking the Kyoto Protocol's entry into force in 2005 and the run-up to the 2009 Copenhagen Conference. By the 2010s, environmental discourse had displaced the security and development concerns that previously dominated.

The Geography of Diffusion

The aggregate rise was produced by a process of spread across countries, and that spread had a distinct geographic structure. The map places a marker at each country that had adopted environmental discourse by each benchmark year. In 1975, environmental discourse was the province of a handful of scattered states. By 1995, clusters had begun to form; by 2015, dense concentrations had emerged across Europe, the Americas, and the island regions of the Pacific and Caribbean; and by 2024, adoption had become nearly universal. The two earliest adopters were Australia in 1969, where environmental discourse emerged from domestic political currents, and The Gambia in 1972, speaking amid the catastrophic Sahelian drought. Use the buttons to step through the benchmark years.

Spatial Panel Results: Inverse-Distance Exposure

The coefficient on spatial exposure, the inverse-distance weighted average of neighbors' prior-year environmental emphasis, is positive and statistically significant throughout. A country's environmental emphasis rises following increases in the emphasis of its geographic neighbors, net of all time-invariant national characteristics and the global trajectory captured by year effects. The effect is essentially unchanged when income and democracy are added, and democracy enters as a positive and independent predictor. Switch between the spatial-lag effect across models and the full specification with all controls.

Robustness: Nearest-Neighbor Exposure

The central result is robust to the formulation of the spatial weights. Replacing inverse-distance exposure with a nearest-neighbor measure leaves the pattern unchanged: the spatial exposure effect remains positive and significant, democracy remains a positive and independent predictor, and the additional economic and vulnerability controls remain null. Whether proximity is operationalized as inverse-distance or nearest-neighbor exposure, a country's environmental emphasis follows its neighbors'.

Key Findings

Environmental discourse spread across nations through geographic proximity. A country's environmental emphasis followed the prior environmental emphasis of its neighbors, net of its own characteristics, its own discursive trajectory, the global trend, and a range of economic controls. The spatial-lag coefficient is positive and significant in every specification, and it survives the inclusion of a lagged dependent variable, which rules out the objection that the result is just national momentum.

The economic-structure account carried no comparable influence. Measures of income, fossil-fuel dependence, economic globalization, and climate vulnerability left the spatial effect intact. This is the paper's pivot: the spread tracked the map, not the trade network. Democratic governance was independently associated with a higher environmental share of speech, framed as a complementary channel rather than a rival to proximity.

Viewed through the words that states speak, the global rise of environmental discourse appears not as a uniform awakening to a shared planetary danger but as a structured social process organized by geography and democratic institutions. The finding connects the world-society tradition, in which models of legitimate state conduct travel between proximate states, to a concrete and well-powered measurement of how that travel actually happened.