Proto-Indo-European Ergativity… Still to be Discussed

Since Uhlenbeck's seminal article ("Agens und Patiens im Kasussystem der indogermanischen Sprachen", 1901) many scholars have accepted the hypothesis of an ergative case in Proto-Indo-European (PIE) given the light it could shed on obscure facts discovered by the comparatist school inside the IE family. The Soviet linguistic school has been particularly active on ergativity in relation with their interests for living languages of the Caucasus and for ancient languages of the Middle East. More recent works on ergativity have shifted the focus to Australian languages. When the theory of language universals took ergativity into consideration, scholars began to seek an explanation of the so-called "split ergativity" in relation with Silverstein's animacy hierarchy. A sequel of this was that the kind of split ergativity demonstrated by PIE seemed contrary to the accepted universals and, consequently, discarded. This paper challenges the way language universals have been used to refute the PIE ergativity hypothesis. Indeed, the influence of the animacy hierarchy is known to be effective in many languages, but more as a tendency than as an absolute universal. Also, PIE is not a fully-fledged language, but rather a field of experimentation. I also present the viewpoint that PIE could have had no split at all, but solely a semantic impossibility to use inanimate noun phrases in an agent role, which seemed backed up by similar "embarrassments" in modern languages and by the so-called "Hittite ergative".