Гипотетическая макросемья языков Америки: Amerind.

Hypothetical language macrofamily of the Americas

Amerind is a hypothetical higher level language family proposed by Joseph Greenberg in 1960 and elaborated by his student Merritt Ruhlen. Greenberg proposed that all of the indigenous languages of the Americas belong to one of three language families, the previously established Eskimo–Aleut and Na–Dene, and with everything else—otherwise classified by specialists as belonging to dozens of independent families—as Amerind. Due to a large number of methodological flaws in the 1987 book Language in the Americas, the relationships he proposed between these languages have been rejected by the majority of historical linguists as spurious. The term Amerind is also occasionally used to refer broadly to the various indigenous languages of the Americas without necessarily implying that they are a genealogical group. To avoid ambiguity, the term Amerindian is often used for the latter meaning.

The idea that all the languages of the Americas are related goes back to the 19th century when early linguists such as Peter Stephen DuPonceau and Wilhelm von Humboldt noticed that the languages of the Americas seemed to be very different from the better known European languages, yet seemingly also quite similar to each other. When studies of American Indian languages began in earnest in the early 20th century linguists quickly realized that the indigenous languages were in fact not all that similar, but had a diversity much greater than among the languages of Europe. After a period of uncertainty about whether indigenous languages could be described and investigated by the methods applied to European languages, the first linguists began the daunting task of trying to classify the languages of the Americas by using the comparative method. Among the most prolific and gifted linguists of his times was Edward Sapir, who was among the first to apply the comparative method to Native American languages. However, contrary to current practice in historical linguistics, Sapir also often relied on "hunches" and "gut feeling" when proposing new language families. Some of these suggestions have been proven correct while others have not. Sapir entertained the idea that ultimately all languages of the Americas might turn out to be provably related and such a phenomenon as the apparent Pan American tendency to have first person forms with a prefixed n was suggestive for this line of thought. Since Sapir's death in 1939, linguists have spent their time researching his proposals; typically, there have been two opposing camps in this endeavor: the so called "lumpers" who usually look towards notions of genetic relationships, and the "splitters" who are widely critical of such proposals and expect successful family relations to be proven by the most rigorous standards of scholarship. Joseph Greenberg worked in the tradition of "lumpers" and following Sapir, was mindful of evidence not generally acceptable to those who hold that only actual linguistic reconstruction—through the comparative method—can yield reliable proof of genetic relationships between languages. In elaborating his classification of the Amerind languages, Greenberg relied heavily on Sapir's early work on the North American languages and the highly impressionist classification of South American languages by Paul Rivet.

Ruhlen reconstructed a morphological (ablaut) gender system for proto Amerind, with masculine kinship terms containing the vowel *i and feminine the vowel *u, that he claims proves Greenberg's reconstruction. This is based on Greenberg's *t'a'na 'child', to which Ruhlen adds a masculine derivation *t'i'na 'son, boy' and a feminine *t'u'na 'daughter, girl'. Unlike the n /m pattern in the pronouns, an intact i/u gender system is not attested across language families, and the consensus is that the pattern is a spurious one.

The consensus among historical linguists specializing in Native American languages is that the Amerind hypothesis is unsupported by valid evidence, particularly because the basis for the proposal is mass comparison, but also because of many other methodological flaws made by Greenberg in the elaboration of the hypothesis. Critics regard this technique as fundamentally flawed, unable to distinguish chance resemblances from those due to a historical relationship among the languages and providing no means of distinguishing resemblances due to common descent from those due to language contact. In addition, critics have pointed out errors in the citation of data, including erroneous forms, erroneous glosses, unjustified morphological segmentation, attribution to the wrong language, and citation of entirely spurious forms. A further criticism is that, contrary to normal scholarly practice, no source references are given for the data, which in most cases come from languages for which there is no standard, authoritative source. In addition, Greenberg does not normalize the spelling of the data, so it is impossible without knowing the source of each form to know what the notation represents.