Within MIE, the clearest major sub-group is an eastern one characterized
by a series of linguistic innovations and represented by Indo-Iranian,
Armenian, Phrygian, and Greek. It is sometimes called Graeco-Aryan.^12 To
the north, Slavonic and Baltic seem to be closely related to each other, and in
the west Celtic and Italic. But overall one finds a network of multiple over-
lapping links connecting different languages and groups, especially where
they are neighbours, or have been at some time in the past: for example
connecting Slavonic with Iranian, or Germanic with Italic or Celtic on the
one side and with Baltic on the other. Linguistic changes (especially
phonetic changes) frequently cross dialect and language boundaries and so
blur them, and it may come about that a dialect of one language shares
features with neighbour languages that other dialects do not.^13 When one
bears in mind that most peoples have had different neighbours at different
times, and so been exposed successively to different linguistic influences,
it is not surprising if the outcome is a complex layered pattern that resists
instant stemmatic analysis.
An example of an isogloss that once appeared fundamental for language
grouping, but is now seen to be of secondary importance, is the celebrated
satem shift. This is the generalized change of palatal velar consonants to
sibilants, as illustrated by the [s] in Avestan satəm‘a hundred’ corresponding
to the [k] in Latin centum or Greek -κατν. In the nineteenth century Indo-
European languages were routinely divided into centum and satem languages,
and this was taken to be a basic dichotomy. As we now understand, the
absence of the satem shift is not a significant indicator of a relationship
between languages. Its presence does make a link, but only a superficial one,
as the shift was an areal phenomenon which affected a number of languages
that were in contact at the time, cutting across older established and more
basic divisions. The satem languages include Indo-Iranian, Armenian, and
Slavonic, but not Greek or Phrygian. The shift thus affected only a part of the
Graeco-Aryan territories, together with some other lands adjoining them.
The central part of the Indo-European area is represented by little-known
ancient languages such as Illyrian, Thracian, and Dacian, and by modern
Albanian. Some regard Albanian as descended from ancient Illyrian, while
others connect it rather with Dacian. Thracian and Albanian, and probably
the other two, are satem languages. Dacian and Thracian are considered to be
(^12) On the term Aryan, which in modern usage refers to Indo-Iranian, see p. 142. For the
Graeco-Aryan grouping cf. Kretschmer (1896), 168–70; Durante (1976), 18–30; Euler (1979),
18–23 (history of views since 1858); James Clackson, The Linguistic Relationship between Arme-
nian and Greek (Oxford 1994), who contests the belief often encountered that Greek and
Armenian have a specially close relationship within the group.
(^13) Kretschmer (1896), 24 f., 411.
6 Introduction