2018年经济学人 迁徙美洲:检查站(1)（在线收听）

Science and Technology

科技

Peopling the Americas: Checkpoint

迁徙美洲：检查站

The first migrants to the new world had to wait 8000 years to be admitted.

新大陆的第一批移民在8000年后才被公认。

HOW America was originally colonised is a topic of perennial interest at the AAAS. Until recently, the earliest uncontested archaeological evidence of people living in the New World came from Swan Point, in Alaska. This dates back 14,400 years. Linguists, however, maintain that the diversity of native languages in the Americas could not have arisen so quickly. Conventional models of linguistic evolution assume tongues separate in the way populations of organisms do—so that the flow of vowels, words and grammatical structures between groups must cease before new languages can emerge, just as a cessation of gene flow gives rise to new species. This suggests it would take at least 50,000 years for a single population speaking a single language to diversify and spread through the Americas in a way that yielded the pattern heard today. Since Native Americans’ genes do, indeed, indicate they all derive from a single population, this discrepancy in timing is a paradox.

美国最初如何沦为殖民地一直是美国科学促进会（AAAS）非常感兴趣的话题。最近，我们从阿拉斯加天鹅点得到了一些当时居住在新大陆的部落的考古信息，这些信息是无可争议的。可追溯到14400年前，不过，语言学家认为，美洲本地语言的多样性不可能如此迅速的形成。语言进化的传统模式认为，语言以生物种群的方式分开，因此在新语言出现之前，种群之间的元音，词语和语法结构就不再相互的流动，正如基因流动的停止会形成新的物种一样。这表明，说着单一语言的单一民族，要想语言达到多样化并且通过美洲传播达到今天所听到的模式，至少需要5万年。由于美洲印第安人的基因表明它们都来自同一群体，因此这种时间上的差异是一个矛盾。

That paradox may be close to resolution. Recent digs have pushed the physical evidence of America’s settlement back in time. Meanwhile, as the meeting heard from Mark Sicoli, a linguist at the University of Virginia, in Charlottesville, a different model of linguistic evolution brings the common ancestor of Native-American tongues forward. Apply a few error bars to the results and the two estimates touch—at about 25,000 years ago.

这种矛盾可能即将解决。最近的挖掘发现了美国定居点的实物证据。同时，正如Charlottesville的弗吉尼亚大学的语言学家Mark Sicoli在会议上所说的，一种不同的语言进化模式，使得拥有相同祖先的美国本土语言不断的发展。运用一种新的条形图误差分析结果，得出在2500年前，两个估计数值非常接近。

The problem with explaining linguistic evolution in pure Darwinian terms is that words are not genes. Species, once separate, do not exchange genetic information because they do not interbreed. Languages, though, can exchange grammatical and semantic elements when they meet, which can speed up diversification. DrSicoli thus turned to computational phylogenetic analysis, an area of linguistic research that tries to work out whether and how such interaction may have taken place.

完全用达尔文的观点解释语言进化是有问题的，即：语言不同基因。物种一旦分离便不能再进行遗传信息的交流，因为它们无法交配。但不同的语言相遇时，他们的语法和语义可以相互交流，这可以加快语言的多样化的进程。因此，西里科利斯转向利用计算系统分析，一个试图找出语言如何相互交流的研究领域。

From the thousand or so Native-American languages he chose four dozen spoken in Alaska and northern Canada, the part of the Americas closest to humanity’s point of entry from Asia. He and his colleagues created a database that recorded, for each of them, 116 linguistic features such as sounds, parts of words, the functions of these parts and the ways a language combines words into phrases. They then used this to identify the influences of languages on each other. They also added geographical information, plotting the flow of linguistic change along the Pacific coast and through the river valleys. This nearly halved the time needed to give rise to the modern situation if the languages had evolved independently from a single common ancestor. That suggests the process of divergence may have begun as recently as 25,000 years ago.

从成千上万的美洲土著语言中，他选择了四十八种阿拉斯加语和加拿大北部语言，他们是美洲最接近人类进入亚洲的一部分。他和他的同事们创建了一个数据库，记录了每个人的116个语言特征，例如声音，部分单词，这些语言特征所起的作用，以及将单词组合成短语的方式。然后，他们利用数据库来分析语言的相互影响，还增加了地理信息，绘制了沿太平洋海岸和通过河谷的语言变化流动的路径。如果语言是独立的从共同的祖先发展而来，只需要原本的一半时间，就会进化到现在社会的情况。这表明分歧过程的起点可能是在25,000年前开始的。