II.

Establishment of the Family

Proof that these highly diverse languages are members of a single family was largely accumulated during a 50-year period around the turn of the 19th century. The extensive Sanskrit and ancient Greek literatures (older than those of any other Indo-European language except the then-undeciphered Hittite) preserved characteristics of the basic Indo-European forms and pointed to the existence of a common parent language. By 1800 the close relationship between Sanskrit, ancient Greek, and Latin had been demonstrated. Hindu grammarians had systematically classified the formative elements of their ancient language. To their studies were added extensive grammatical and phonetic comparisons of European languages. Further studies led to specific conclusions about the sounds and grammar of the assumed parent language (called Proto-Indo-European), the reconstruction of that hypothetical language, and estimates about when it began to break up into separate languages. (By 2000 bc, for example, Greek, Hittite, and Sanskrit were distinct languages, but the differences among them are such that the original tongue must have been fairly unified about a millennium earlier, or about 3000 bc.) The decipherment of Hittite texts (identified as Indo-European in 1915) and the discovery of Tocharian in the 1890s (spoken in medieval Eastern Turkistan, and identified as Indo-European in 1908) added new insights into the development of the family and the probable character of Proto-Indo-European.

The early Indo-European studies established many principles basic to comparative linguistics. One of the most important of these was that the sounds of related languages correspond to one another in predictable ways under specified conditions (see Grimm's Law and Verner's Law for examples). According to one such pattern, in some Indo-European subfamilies—Albanian, Armenian, Indo-Iranian, Slavic, and (partially) Baltic—certain presumed q sounds of Proto-Indo-European became sibilants such as s and ś (an sh sound). The common example of this pattern is the Avestan (ancient Iranian) word satem (“100”), as opposed to the Latin word centum (“100,” pronounced “kentum”). Formerly, the Indo-European languages were routinely characterized as belonging either to a Western (centum) or an Eastern (satem) division. Most linguists, however, no longer automatically divide the family in two in this way, partly because they wish to avoid implying that the family underwent an early split into two major branches, and partly because this trait, although prominent, is only one of several significant patterns that cut across different subfamilies.