In this paper I do not propose to give you a time depth, or even a range of possible time-depths, for the Indo-European language family. Rather, I propose firstly to examine some of the existing assumptions about the time depth of Indo-European in recent scholarship, and then to state why the whole notion of a time depth for a reconstructed language is problematic.
Most authors of handbooks and survey articles on reconstructed Proto-Indo-European (PIE) are reluctant to come clean on their thinking on how old the Indo-European language family is (a notable exception is Mallory, in Mallory& Adams 1997, 583–7). At the same time, it appears that a number of linguists have a gut-feeling or intuition about the likely age of Indo-European — I am unaware of any scholar working in the field who wishes to place PIE later than 2500 BCE, and most scholars, to judge from reactions to Colin Renfrew’s Archaeology and Language (Renfrew 1987), are reluctant to go even as far back as 7000 BCE without qualifying to some extent the PIE with a further ‘Pre-’ or ‘the earliest stages of’. One particular model for PIE which finds quite widespread favour at the moment is not afraid to stratify PIE and date the strata. This is the so-called ‘Space-Time’ or ‘Zeit-Raum’ model put forward by Meid in the 1970s, and since refined by others. (1) According to Meid’s model, a reconstructed proto-language should be considered not as a monolithic unity, but an entity which exists over time, and through time over ever-increasing space. Hence one can draw a triangular representation of PIE, with the dispersal of the language over space plotted as the horizontal axis, and the time scale plotted on a vertical axis as below:
PIE 1 5000–4000 BCE I
PIE 2 4000 BCE IIIIIIIIIIIIIIIIIII
PIE 3 3000–2500 BCE IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII
How do Indo-Europeanists such as Meid come up with such seemingly accurate stratigraphy with so little explanation of what they are doing? The time scale given in Meid’s model finds widespread acceptance (and not just among those who follow his space–time model) for a number of reasons: age of the attested daughter languages; dating of the culture of the speakers of PIE through the reconstructed lexicon; the Romance model and glottochronology. (2) I shall consider each of these in turn.