Terroir shapes what grapes can become. History, ambition, and culture determine what a region chooses to be.
Every wine region carries two identities. The first is geological: the soils, the climate, the slopes, the latitude — what we call terroir. The second is cultural: the stories a region tells about itself, its relationship to tradition, the ambitions of its winemakers, the weight of its history. The wine world has long assumed these two identities move in lockstep. This study tests that assumption.
We built two independent classification systems for 59 wine regions across 16 countries. The first groups regions by terroir, using natural language processing to analyse factual profiles of climate, soil, and topography. The second groups them by cultural identity, using expert-scored dimensions of anthropological character — how inward or outward a region is, how much it struggles, whether it looks to tradition or reinvention.
The result is unambiguous: the two systems are statistically independent. Knowing a region's terroir type tells you essentially nothing about its cultural identity. Burgundy and Santa Cruz Mountains share an identity cluster — both are deeply interior, obsessively place-defined — yet their terroir could not be more different. Santorini and Hunter Valley are grouped by the same spirit of defiance against harsh conditions, despite sharing no geology, no climate, no grape varieties.
Six identity clusters emerge, each a distinct mode of being a wine region: from the inward devotion of Burgundy and Piedmont, to the joyful extroversion of Beaujolais and Provence, to the constructed ambition of Napa and Swartland. Culture groups regions differently than terroir does. The map is not the soul.
Physical geography creates the conditions for winemaking, but historical depth — time, tradition, institutional memory — is the dimension that most strongly differentiates identity clusters. The Mosel and Finger Lakes share steep, slate-lined river valleys; yet the Mosel's identity is one of timeless poetry, while Finger Lakes burns with the urgency of a region still proving itself. The land sets constraints; the culture interprets them.
Harsh, marginal terroir tends to produce identities of struggle and endurance. Generous terroir tends to produce identities of ease and pleasure. But the exceptions are as revealing as the pattern. Santorini's volcanic extremity produces survival; Dalmatia's equally Mediterranean terroir produces tranquility. Same climate type, opposite cultural souls.
New World regions with favourable growing conditions show the highest individuality and urgency. When the land does not impose an identity, the winemaker constructs one. Napa's ambition, Swartland's rebellion, Mendoza's reinvention — these are identities forged by human will, not geological necessity. Ambition is the mirror image of terroir determinism.
If you have ever found yourself drawn to a region without quite knowing why — or surprised by a connection between two places that share nothing on the map — this study offers a framework for understanding that instinct. The identity clusters reveal affinities that geography alone cannot explain, and they open new paths for the curious wine explorer.
If you love Burgundy's inwardness, the data suggests you may find the same quality in Piedmont, Tokaj, or the Santa Cruz Mountains — regions you might never have grouped together on a map.
Bordeaux and Napa share terroir traits, but their identities diverge completely — one institutional and tradition-facing, the other pioneering and ambition-driven. The six dimensions explain where and why.
The most revealing regions are those that defy their terroir. Santorini's Mediterranean climate should suggest ease; instead, its identity is one of extreme survival. These divergence cases are where the study is most illuminating.
Where the land is generous, the winemaker's ambition shapes identity most strongly. Understanding this pattern reframes how we think about New World wines — not as imitators, but as regions constructing identity from human will.
The finding that physical substrate and cultural identity are independent classification systems is not unique to wine. The same structural pattern — two ways of grouping the same objects that systematically disagree — appears across the sciences and humanities. This study is the first to demonstrate it formally in the wine domain, but it joins a broader intellectual tradition.
World culture mapping studies consistently show that cultural clustering and geographic clustering diverge — nations that share a border often share less culture than nations on different continents.
Genetic classification (language families) and areal classification (contact features) group the same languages differently.
Structural brain connectivity constrains but does not determine functional connectivity — the brain's physical wiring and its emergent cognitive states are partially independent systems.
Radar charts showing the identity profile of each region across six dimensions. Filter by cluster, click any region for its soul metaphor and character summary.
Explore clusters →Regions projected into two-dimensional space, coloured by cluster. The identity map shows clear groupings; the terroir map does not predict them.
View the map →An alluvial diagram showing how regions flow from terroir clusters to identity clusters. The criss-crossing bands are the visual proof: the two systems are independent.
View the map →The full score matrix for all 59 regions across six identity dimensions. Colour-coded, sortable, and filterable by cluster, country, or world.
Open dashboard →The full narrative paper. Hypothesis, related literature across four academic fields, methodology, results, and a guide to the six identity types for the wine-focused reader.
Open PDFMethodology deep-dive: the dual-layer architecture, identity dimensions, clustering pipeline, independence testing, SME review log, and the complete score matrix.
Open PDFThe 59 anthropological identity narratives (280–320 words each) that form the foundation of the study. Organised by country within Old World and New World.
Open PDFThe 59 factual terroir profiles: climate, soils, principal varieties, winemaking, production structure, and historical position. The raw input for the NLP terroir classification.
Open PDFA plain-language guide to the statistical and computational methods used in this study: NLP, TF-IDF, PCA, k-means clustering, and independence testing. No prior statistics knowledge assumed.
Open PDFThe study uses natural language processing and unsupervised machine learning to independently classify 59 wine regions along two axes — terroir and cultural identity — then tests whether the two classifications agree. They do not. The terroir classification is fully algorithmic; the identity classification combines structured expert judgment with machine learning.
A 280–320 word anthropological portrait of each region, written without any reference to geology, climate, or grape varieties. Pure cultural character.
Six factual fields per region — climate, soils, varieties, winemaking, production structure, historical position — processed by NLP independently of identity.
Each region scored on six axes: Interiority–Exteriority, Struggle–Ease, Tradition–Reinvention, Individual–Collective, Urgency–Timelessness, Earthly–Transcendent.
Machine learning clusters regions into identity groups and terroir groups separately. A formal independence test confirms the two groupings share no structure.