science tumbled

(pretty pics / longer stories / ask)

What can we learn about the earliest human language by comparing languages known today?

If you ever need to write an academic smackdown, let this document be your guide. Lyle Campbell, an expert in historical linguistics, basically destroys any sort of argument that we can know anything about the earliest human language.

Language, presumably, evolved once. Logically, then, there should be a most recent common ancestor of all natural languages in the world. Linguists have a procedure called the comparative method by which they can figure out which languages are related, how they are related, and by which they can make reasonable reconstructions of the theoretical, unattested proto-languages that form the root of a language family. As you might expect, it hinges on side-by-side comparisons of languages. (Fun fact: one of the founders of historical linguistics and the comparative method is Jacob Grimm, most famous as one half of the folklore-collecting Brothers Grimm. Grimm’s Law, a set of systematic sound changes in the history of Germanic languages, is named after him.)

Most famously, linguists have reconstructed Proto-Indo-European, the unattested ancestor of languages as varied as Sanskrit, English, Latin and Russian. This language is believed to have been spoken some five or six thousand years ago. But humans are believed to have reached Australia by 50,000 years ago, so the common ancestor of all human languages would be at least ten times as old as Proto-Indo-European.

Some people have attempted to use a comparative approach to reconstruct this language, but Campbell shows again and again how this approach yields results that are no better than chance. Basically, what defenders of “Proto-World” do is try to find words that mean roughly the same thing in different language families and which look roughly the same, but both the semantic and the phonetic constraints they use are extremely liberal. One of their defining examples is *tik, a root which supposedly meant “finger”. Although Campbell doesn’t say so in that many words, it’s essentially wishful thinking. You find what you want to find, because, as Campbell demonstrates, it’s extremely easy to find false correlations. Not to mention that if there is a common ancestor of all human languages, from what we know about how languages change over time, we would expect that every single word in every single language would have changed so much as to be unrecognizable over the past fifty thousand years.

In the end, says Campbell, we will never know much of anything about the earliest human languages unless we invent a time machine. There simply isn’t any written material going back far enough, and there isn’t enough information in present-day languages to reconstruct their oldest ancestors. The only thing we can know for sure about the earliest human language is that it would have obeyed certain universals—the particulars of which are a subject of acedemic debate—which all present-day human languages obey, but that is only by definition, because anything that came before that would not be what we today define as human language.

It’s rare that we run into a scientific question that is fundamentally impossible to answer, but this is one of them. The information required to answer it simply doesn’t exist. It’s just one of those things, which is sad, because I, for one, find the emergence of language—one of the defining features which distinguish humans from other animals—to be extremely fascinating.