 |
Discussion Examples for
Comparative Reconstruction
(APS Reader 58-61)
|
proto-Bade-Ngizim liquids
Bade (of which there are several distinct dialects) and Ngizim
are closely related languages of the West Chadic-B branch of Chadic,
spoken in northeastern Nigeria. Reconstruct the liquid phoneme
system of proto-Bade-Ngizim and the sound changes that account
for the modern forms. Tones are omitted. Vowel length is marked
by a macron. [r] = a tongue tap or trilled "r"; 
= a lateral flap.
(Data from personal field notes.)
Back to top
proto-Turkic initial stops
The Turkic languages, a subbranch of the Altaic phylum, are
spoken in central Asia. Turkish and Azeri belong
to the Southern Turkic group, Uzbek belongs to the Eastern
Turkic group. Reconstruct the initial stop consonants of proto-Turkic
and the sound changes which account for the modern forms. (Data
adapted from a Linguistics 110 paper by Calvin Mazlumyan, to whom
I am grateful for introducing me to these languages.)
Turkic languages have vowel harmony, i.e. vowels fall into
two sets and words may contain vowels from only one of those sets.
In Turkish, at least, all the vowels in a word fit the pattern
[aback, ßround], i.e. all vowels in a word must agree in
both backness and roundness (though the data here suggests that
/a/ is "neutral" as to the feature [round]). The basic
vowel system of Turkic languages is as follows:
See sound
changes
Correspondence sets (consider only stops in initial position):
See correspondence
sets
Sound changes (click on the language names to see a
more general account of the sound changes):
Generalizations and discussion
Turkish
No changes: Turkish preserves the reconstructed initial
stops of the ancestral language AND the vowels of the anscestral
language following these stops (see discussion
below with respect to *g in Uzbek #13).
Azeri
Problems and comments
- #8 top 'ball': We would expect *dop because
the vowel is a [+back] vowel. This is the only exception in the
data. While it is possible that the formulation of our change
is not quite correct, a more likely possibility is that this
word has been borrowed since the change ran its course in Azeri.
We would like more information on the distribution of this item
in neighboring languages and information on its origin if such
information is available.
Back to top of Turkic problem
Uzbek
Problems and comments
- Relative chronology ("bleeding order"): The
changes had to have applied in this order. Change (1) removes
the original *g before back vowels by changing it into
a fricative. Change (2) affects only stops.
- #13 kalin 'bride': We would expect *ghalin
because the vowel is [+back] and hence change (1) should apply.
Note, however, that in both Turkish and Azeri, the vowel is [-back].
We assume that this item would be reconstructed *kelin,
with a change in the vowel in Uzbek after change (1) had run
its course. This fits with with picture that devoicing of stops
in Uzbek was a fairly late change, i.e. change (1) voiced *g
before back vowels fairly early, later the vowel in *kelin
changed, and later yet change (2) devoicing the stops took place.
Back to top of Turkic problem
Comparing the outcomes in Azeri and Uzbek: "Strength
Hierarchies" again!
Azeri and Uzbek have undergone essentially opposite changes,
with Azeri voicing non-labial stops and Uzbek devoicing them.
- Azeri: The change can be viewed as a weakening of
voiceless to voiced. Note that it has NOT applied to labials,
which are at the "strongest" point of articulation.
Why did voicing only apply before [+back] vowels? Some theories
would claim that [+back] vowels are stronger than [-back] vowels.
This is controversial, but if one accepts it, then one might
propose that the stronger vowel "robs" strength from
relatively weak consonantal onsets.
- Uzbek: The devoicing of voiced stops can be seen as
a strengthening process in relatively strong word initial position.
There are problems with this interpretation, however. First,
the change of *g to velar fricative before back vowels
is a weakening. Why would the same language undergo word
initial weakening, then later word initial strengthening? Second,
devoicing does not apply to the labial b. One would expect
that word initial devoicing (a strengthening process) would apply
first to the strongest segments (the labials), the later
to the weaker segments.
Back to top of Turkic problem
Comment on Turkic p
Initial p is rare in Turkic languages. According to
Nicholas Poppe (Introduction to Altaic Linguistics, Harrassovitz,
1965), the major authority on Altaic languages, Altaic initial
*p was lost in Turkic. (Actually, evidence seems weak that
*p can even be reconstructed to any ancestral stage of
Turkic.) Whatever the reason for the rarity of p in Turkic,
it is difficult (and perhaps meaningless) to put together of set
of cognate words illustrating p in all possible environments.