very archaic features of Hittite, for example of the Hittite neuter nouns. And he
saw that “even stronger evidence for the Indo-Hittite hypothesis is presented
by innovations common to the Indo-European languages but foreign to Hittite.”^4
A very early bifurcation of a language that, faute de mieux, we must call “Proto-
Indo-Hittite,” eventually produced two language families. One of them was the
Anatolian: this family consisted of Hittite (more correctly “the language of
Kanesh”), Luwian and their cognate languages, all of which had disappeared by
the end of antiquity. The other family was the Indo-European, which has spread
over most of the earth.
The competing view, dominant until almost the end of the twentieth century,
was that the Anatolian languages were just another branch of Indo-European and
had—like the others—evolved from PIE. In this view, there were folk migrations
of PIE speakers from the north into Anatolia: after arriving in Anatolia, the
migrators were much influenced by the native languages, and so their speech soon
lost many of its PIE features. The theory that Anatolian lost many of its PIE features,
known among linguists as the Schwundhypothese(“loss-hypothesis”), is no longer
tenable. The Anatolian languages are archaic, reflecting an evolutionary stage
considerably prior to the stage reached in PIE.
The scholarly consensus on this point is recent. In 1994 Jaan Puhvel professed
his opposition to the Indo-Hittite heresy and his abiding faith in the Schwund -
hypothese: “I have been an instinctive adherent throughout and do not anticipate
any belated Pauline conversion.”^5 More cautious was Craig Melchert. In his
Anatolian Historical Phonology, also published in 1994, Melchert wrote, “I con -
sider it an open question whether the non-Anatolian Indo-European languages
underwent a period of common linguistic development after ‘separation’ of the
Anatolian group.”^6 By 2000 Melchert considered the question closed: “the crucial
point is that there is now limited but compelling evidence that the rest of the Indo-
European languages underwent a set of shared common innovations in which
Anatolian did not share.”^7 Alexander Lehrman also observed that when in 1994
he wrote his “Indo-Hittite Revisited” few linguists accepted the theory, but that
by 2000 many had adopted it “either in name as well as in essence... or in essence
if not in name.”^8
Lehrman’s own work contributed to the success of the Indo-Hittite theory, but
most important were the computational cladistics that a group of scholars at the
University of Pennsylvania began to apply to the Indo-European language family
in the middle 1990s. Borrowing from biologists’ use of the computer to establish
phylogenetic trees, linguists Don Ringe and Ann Taylor and computer scientist
Tandy Warnow concluded that in the original bifurcation of a proto-language one
branch evolved after many centuries into Anatolian, and the other branch led
eventually to all of the “other” Indo-European languages.^9 This conclusion seems
to be accepted by most Indo-European linguists.^10 The plain consequence of all
this, unwelcome as it may be, is that Hittite was not an Indo-European language:
it was a distant relative of the Indo-European languages. What Bedrich Hrozny
actually discovered is that Czech, French, English, and all the other Indo-European
languages belong to something still larger: the Indo-Hittite language family.
2 Origins and spread of Proto-Indo-European