Albanian (2)

From: Piotr Gasiorowski
Message: 29908
Date: 2004-01-22

2. The development of PIE coronal stops

PIE *t is simply continued in most positions (its post-nasal voicing
will be discussed later). Unlike *sp and *sk, the cluster *st was not
simplified in Proto-Albanian and survives as Mod.Alb. sht /s^t/.

PIE *-tt- (phonetically *[ct], with the first *t affricated) is
reflected as /s/ in Modern Albanian. This can't be directly compared to
the Italic, Celtic and Germanic change of *-tt- > *-ss-, since Mod.Alb.
s reflects an older postalveolar affricate (*c^). The most likely
scenario seems to be as follows: Early Proto-Albanian had two "colours"
of affricates: "light" {*c', *3'}, resulting from the affrication of
pre-Albanian palatal stops, and "dark" (labialised) {*c'W, *3'W},
resulting mostly from the palatalisation of PIE labiovelars. The *W
component of the latter series favoured a "marginal" (apical or laminal)
rather than predorsal realisation of the affricate, and perhaps its
velarisation, enhancing its "dark" character and increasing the contrast
between the two sets; hence the shift of *c'W > *c^(W). The inherited
cluster *ct was phonetically simplified to *c at that stage, and the
hitherto allophonic affrication of the first element became distinctive.
On account of its marginal character it was articulatorily closer to
*c^(W) than to predorsal *c', and became identified with it. The Late
Proto-Albanian fronting of *c' > *c took place after the merger:

*k' > *c' > *c' > *c > T
*kW /+, *k'w > *c'W > *c^(W) > *c^ > s
*tt > *c(t) > *c^(W) (merger with the above)

PIE *d, *dH > pre-Albanian d.

Between Late Proto-Albanian and Old Albanian *d was lenited and lost
between vowels (the loss will be discussed later). The intermediate
stage seems to have been *D, merging with the reflex of Proto-Albanian
*3. /D/ (spelt <dh>) is also the Modern Albanian reflex found after /r/,
both in native words and in loans (e.g. Lat. ordinem --> Geg urdhën,
Tosk urdhër 'order'). We find it also word-initially, where it
alternates with /d/, presumably as a result of sandhi processes (see the
discussion of the reflexes of *g^(H)). For example, we have dhjetë '10'
< *Die(T&)t& < *diäcata < *dek^m.t (+ *-a:), where the fricative may
have originated in composite numerals such as '30' and '50' (Mod.Alb.
tri-dhjetë, pesë-dhjetë) or in 'N-teen' numerals (Mod.Alb.
N-mbë-dhjetë). A stop can be found e.g. in <det> 'sea' < *dHeubeto- or
in <djeg> 'burn' < *dHegWHo:.

The cluster *zd (< PIE *sd(H)) gives Modern Albanian /D/, but the
development is difficult to reconstruct in detail. One could envisage
either of the following:

a. *zd > PAlb. *3 > D
b. *zd > PAlb. *zd > OAlb. z^D > Mod.Alb. D

Both *tj and *dj underwent palatal assimilation in Proto-Albanian,
yielding the affricates *c^ and *3^, which have regularly developed into
Modern Albanian /s/ and /z/.

3. The development of PIE labial stops

PIE *p is simply continued in Albanian. Like the other stops it was lost
early when followed by another obstruent, as in *septm.' (+ *-t-a:) >
*s&táta > shtatë /s^tat(&)/ '7'. This change, like the simplification of
*kt, took place before the period of borrowings from Balkan Latin, in
which *ks, *ps --> *fs > fs^ and *kt, *pt --> ft.

PIE *sp > Mod.Alb. f (presumably via *sf). Like the parallel change of
*sk > x the change must be old, since it didn't affect loans from Latin.

PIE *b, *bH > pre-Albanian *b (its intervocalic loss in Old Albanian
will be discussed later).



Piotr

(Sorry, but the next instalment won't be posted till Monday or so. I'll
be away at the weekend.)