Pick two humans at random. Any two, from anywhere on Earth. What are the odds they can have a conversation?
Stop and answer before you read further. The instinct of most people I have asked is somewhere between 30 and 50 percent. We live in what we call a connected world. Trade goes everywhere. Tourists go everywhere. Migration goes everywhere. The internet supposedly goes everywhere. Surely two random humans can find a few hundred words in common.
The answer is 10 percent.
If we restrict to pairs from different countries (which is the case that actually matters for international cooperation, refugee integration, scientific collaboration, and most everything we associate with global anything) the answer drops to 7 percent.
93 percent of cross-border human pairs cannot communicate.
This is not an opinion. It is what the data shows when you measure the actual conversational language repertoires of 8 billion people across 235 countries and territories at A2+ proficiency, which is the international threshold for sustaining a basic conversation. This is the work of LINGACON - a worldwide linguistics research project measuring multilingualism across every country on Earth. I spent the year on the audit, country by country, with my collaborators. The 7 percent is what came out.
Let me give you the second number you need. The average human on Earth speaks 1.6 languages. Not three or four. Not the cosmopolitan-elite figure international magazines like to invoke. 1.6. Most of those second languages are local: a regional vernacular, a heritage tongue, a tribal language. They serve their communities beautifully. They do not connect the continents.
The geography of disconnection
The disconnection is not evenly distributed across the planet. Some parts of the world are linguistically tight. Most of the Americas can talk to each other through Spanish, English, and Portuguese. Pacific island states share English at near-universal A2+ rates. Western Europe, despite a dozen languages, has knit itself together through English and decades of education policy.
Then there is Asia.
Asia is the world’s most populous continent, and its most internally disconnected. Two random Asians have a 13 percent chance of sharing a language. Africa-Asia is the worst cross-continent pair at 4 percent. Asia-Europe is 6 percent. The continents holding 60 percent of humanity barely speak to each other.
When I first computed this, I checked the numbers three times. Asia internally less connected than Africa? Africa-Asia at 4? I expected something like 25, 30. I knew the situation was worse than the popular framing, but I did not expect this.
The reason is structural. Every major Asian language family - Sinitic (Mandarin, Cantonese, Wu), Indic (Hindi, Bengali, Marathi, Tamil, Telugu and a hundred others), Iranic (Persian, Pashto, Kurdish), Semitic (Arabic varieties), Japonic, Koreanic, Turkic, Austronesian (Malay, Indonesian, Tagalog, Javanese) - is mutually unintelligible with the others. The “Asia is connected” narrative is a comfortable story we tell ourselves. The data does not support it.
China and India, together 2.85 billion people, more than a third of humanity, and the only realistic shared language between them is English at modest A2 rates. The two largest national populations on Earth have no native linguistic bridge. None. Their conversation, when it happens, happens in a third country’s tongue.
The thread we are all hanging from
Now the most important finding.
I computed which languages carry the world’s connectivity, weighted by how much each language contributes to the probability that two random humans share something. English contributes 49 percent of all global linguistic connectivity. Half. One language.
If you removed English from the planet tomorrow, half of all the conversations across borders would not happen. Half of the world’s connection rests on a single thread. We’ve been calling this globalization. It’s actually monolingual dependence.
This is not a celebration of English. English carries 49 percent of human connectivity because of three centuries of empire, two centuries of finance, one century of cinema, four decades of internet. It is the lingua franca by accident of history, not by any inherent capacity to bridge. If history had run differently (if French had won, or Spanish, or Mandarin if China had colonized as Britain did), the bridge would be a different language and we would be celebrating that. The point is not which language. The point is that we have only one.
The top 7 languages - English, Mandarin, Hindi, Spanish, French, Bengali, Urdu - carry 91 percent of all global connectivity. The other 7,000+ languages? They serve their communities beautifully and connect the world barely at all.
Read that again. Seven languages out of seven thousand carry nine-tenths of the connection. The asymmetry is staggering. And it is more concentrated than wealth, more concentrated than internet bandwidth, more concentrated than military power.
Why this is felt every day
We have built a world that runs on linguistic connectivity, and most of us do not have it.
A Syrian refugee resettling in Lebanon shares language with most of his neighbors. A Syrian refugee resettling in Germany does not. The integration costs are radically different — measured in years of language learning, in education gaps for children, in lifetime earnings forgone — and the underlying reason is one number: the GLCI between Syria and the host country.
A pharmaceutical company in Switzerland coordinating clinical trials across Indonesia, Nigeria, and Brazil pays a translation-and-coordination tax that does not appear on any balance sheet but absorbs an enormous amount of executive time and creates an enormous amount of error.
A diasporic Armenian family in Argentina — three generations from the homeland — loses the language by generation three. The grandchildren cannot speak with their grandparents on the phone. This loss is not a tragedy of one family. It is happening in every diaspora simultaneously, and the cumulative loss across the planet is uncounted.
International scientific collaboration runs through English. Diplomats negotiate in English. Engineers in Vietnam working with engineers in Brazil work in English. Two researchers in Cairo who could in principle collaborate with their counterparts in Algiers often go through English papers and English conferences instead, because the Arabic-language scientific infrastructure is thin. The disconnection compounds.
What to do
I do not believe the 93 percent is fate. It is the result of policy choices made and unmade by every government on Earth, decade after decade, with very little coordination and almost no measurement until now.
The disconnection is closable. Not to zero, not for free, not soon. But meaningfully closable.
There are three pathways. School systems can introduce a second language early and treat the curriculum as seriously as mathematics — Singapore, the Netherlands, Luxembourg show what this produces. Luxembourg has an ALPS of nearly 4; the global average is 1.6. Motivated adult learners can close their own gaps through self-directed and app-based learning, and the global adult-learning market is now larger than at any time in history. AI-augmented language learning, deployed at scale through partnerships between governments and educational technology companies, has the potential to deliver functional second-language proficiency to populations that no traditional school system could ever reach. These three pathways are not mutually exclusive. The countries that move toward universal bilingualism in the next decade will run all three at once.
What we have lacked, until very recently, is a measurement framework that lets us actually quantify the gap, see where the leverage is, and check whether interventions are working. That is what I built. The ALPS, the GLCI, the country-pair matrix, the policy simulator — these are not academic indulgences. They are the diagnostic instruments we need to close the disconnection deliberately rather than letting it close itself in directions we do not choose.
Back to the airport
Imagine two humans meeting somewhere — a transit lounge in Istanbul, a market in Lagos, a hospital waiting room in São Paulo. Today, 93 percent of the time, the strangers exchange polite nods and nothing more. A whole human life adjacent to another, in absolute silence.
We have organized 8 billion people into a configuration where this is the default outcome of every random encounter. We did this by accident. We can undo it by choice.
The first step is to measure how disconnected we really are.
Now we have the number. 93 percent. Whatever we build next, that is the starting point.
LINGACON is a global linguistics research project measuring how humanity communicates, across every country on Earth