One of the most ambitious—and controversial—attempts to answer this question is the “three-wave” migration theory, famously championed by linguist Joseph Greenberg. It proposes a deceptively simple narrative to explain the peopling of an entire continent, using language as its primary evidence. While the theory is now largely rejected by specialists, it remains a fascinating chapter in the story of historical linguistics and a powerful illustration of the search for deep connections in human history.

The Three Waves: A Grand Classification

In his 1987 book Language in the Americas, Greenberg argued that nearly all Indigenous languages of the Americas could be classified into just three massive families, each corresponding to a distinct wave of migration from Asia across the Bering land bridge.

1. The First Wave: Amerind
   This is the oldest and, by far, the largest and most controversial of Greenberg’s proposed families. He suggested that the first migrants arrived around 11,000 BCE and spread throughout North and South America. Their linguistic descendants, according to Greenberg, form the “Amerind” superfamily. This single group lumps together almost every Indigenous language from Canada to Chile that isn’t in the other two families. This includes well-established and vast families like Uto-Aztecan (which includes Nahuatl and Hopi), Mayan (spoken in Mesoamerica), Quechuan (the language of the Inca Empire), and hundreds more.
2. The Second Wave: Na-Dené
   Greenberg proposed a second, later migration around 7,000 BCE. These migrants brought the ancestor of the Na-Dené languages. This family is much less controversial than Amerind and is widely accepted by linguists (though its internal structure is still debated). It includes Tlingit in Alaska, the extinct Eyak language, and the widespread Athabaskan languages, such as Navajo (Diné Bizaad) and Apache in the American Southwest, and Dene Suline in northern Canada.
3. The Third Wave: Eskimo-Aleut
   The final and most recent migration, according to the theory, occurred around 2,000 BCE. These peoples remained in the Arctic regions of North America and Greenland. The Eskimo-Aleut family is the smallest and least controversial of the three. It consists of two main branches: Aleut, spoken on the Aleutian Islands, and the Eskimo languages, which include the Yupik languages of Siberia and Alaska and the Inuit languages spoken across Arctic Canada and Greenland.

Greenberg’s Method: Mass Comparison

How did Greenberg arrive at such a sweeping conclusion? He used a method called “multilateral comparison”, or “mass lexical comparison”. Instead of the painstaking, slow work of the traditional comparative method—which reconstructs ancestral “proto-languages” by identifying regular sound correspondences between two or three languages at a time—Greenberg looked for broad similarities across hundreds of languages simultaneously.

He focused on core vocabulary (like pronouns, body parts, and simple verbs) and grammatical elements. If he saw a pattern emerge across a wide geographic area, he considered it evidence of a distant genetic relationship.

For example, one of his key pieces of evidence for the Amerind family was the widespread pattern of first-person pronouns (“I”) using an n- sound and second-person pronouns (“you”) using an m- sound. He pointed to words like no (“I”) in Wappo (California) and ma (“you”) in a Cariban language of South America as evidence of a shared heritage, despite the immense distance separating them.

The Controversy: Why Most Linguists Remain Unconvinced

While Greenberg’s proposal was bold and initially seemed to align with some early genetic studies, it was met with immediate and forceful criticism from the majority of historical linguists. The objections primarily center on his methodology.

• Chance and Borrowing: Critics argue that with hundreds of languages and short, common words, some similarities are bound to occur by pure chance. Furthermore, Greenberg’s method doesn’t have a reliable way to distinguish between words that are related by descent (cognates) and words that were simply borrowed from neighboring languages over millennia of contact.
• Lack of Regular Sound Correspondences: This is the cornerstone of the critique. Mainstream historical linguistics is built on the discovery of systematic sound changes. For example, we know English father and Latin pater are related because there’s a regular pattern where Proto-Indo-European *p became *f in Germanic languages. Greenberg largely ignored this principle, accepting resemblances that didn’t fit into any predictable system of sound change. To his critics, this made his work more akin to collecting curiosities than to scientific reconstruction.
• The Amerind Superfamily: The idea of “Amerind” is the most heavily criticized part of the theory. The time depth involved—over 13,000 years—is considered far too great for any current linguistic method to reliably prove a relationship. Most linguists believe that if a single “Proto-Amerind” language ever existed, its descendants have diversified for so long that any traceable signal of their common origin has been erased by time.

Interestingly, subsequent research has lent more credence to a deep connection for the Na-Dené family. The Dené-Yeniseian hypothesis, which proposes a link between the Na-Dené languages of North America and the Ket language of Siberia, is based on rigorous comparisons of morphology and sound systems and has gained significant acceptance. This stands in stark contrast to the Amerind proposal, which remains on the fringe.

A Tapestry, Not a Triptych

So where does this leave us? The three-wave theory, for all its flaws, was a thought-provoking attempt to paint a big picture of American prehistory. It forced linguists to confront the challenges of deep time classification and sparked decades of debate and new research.

Today, the consensus is that the pre-Columbian Americas were not a neat triptych of three language families. They were, and remain, a rich, complex, and ancient tapestry woven from countless distinct linguistic threads. Most linguists currently recognize at least 150-200 separate language families and isolates (languages with no known relatives) in the Americas. The work of understanding their histories is a slow, careful process—one that involves reconstructing smaller, more plausible families like Uto-Aztecan or Mayan, rather than making grand, sweeping claims.

While Greenberg’s simple map has been put aside, the true linguistic map of the Americas is far more intricate and, ultimately, more fascinating. It speaks to millennia of human migration, innovation, and interaction—a story that is still being carefully pieced together, one language at a time.