LINGUIST List 5.1420

Fri 09 Dec 1994

Disc: Comparative method in linguistics

Editor for this issue: <>

Directory

1. "c.a. creider", Re: 5.1404 Comparative method in linguistics
2. , Re: 5.1404 Comparative method in linguistics
3. , Comparative method (shared sound changes)

Message 1: Re: 5.1404 Comparative method in linguistics

) Leo. Karamojong is one of a number of languages which are collectively
) called Jie. They are Nilotic languages. Obviously related languages
) are spoken in Sudan, e.g., Dinka, Nuer, Shilluk, in Uganda, e.g.,
) Achooli, Lango and in Kenya/Tanzania, e.g., Luo. If you saw a
) word list from each of these languages, e.g., the numbers from one to
) ten, you would have little difficulty reconstructingthe parent language
) yourself. Now when it comes to another branch of Nilotic which includes
) Maasai and Kalenjin I think you could also see the genetic relationship.
) All the Nilotic languages were recognised as genetically related long
) before Greenberg. But then the bigshots of pre-Greenberg, esp the
) German super-star Carl Meinhof went further. They wanted to connect
) these languages with"Hamitic"(you know people related to the Egyptians
) with all the implications of the jaded racial arguments about whether
) the Egyptians were black or not -- or rather NOT black, or not not).

One small correction and one comment on Benji Wald's interesting letter
on pre-Greenber classifications of the Nilotic languages. Karimojong is
actually a member of "the other branch of Nilotic which includes Maasai
and Kalenjin". This branch, termed Nilo-Hamitic by some, and Paranilotic
by Tucker and Bryan was divided into two groups by Oswin Koehler in
1955 as part of a tripartite division of the Nilotic language--Western
(Dinka, Nuer, Shilluk, Achooli, Lango, Dholuo), Eastern (Maasai, Karimojong,
Turkana, etc.), Southern (Kalenjin, Datooga). Greenberg saw the
correctness of Koehler's classification and adopted it. It is accepted
by all today.

It is also worth noting that there were scholars, such as Tucker and
Bryan, who, although they did not accept the connection with "Hamitic",
did feel that there was sufficient evidence to connect Southern and
Eastern in a group (the old Nilo-Hamitic or Paranilotic). Tucker
and Bryan believed that these languages were Nilotic (or would
ultimately be shown to be Nilotic, so the question was really whether
the tree was:

 / | \ or / \
 /\

The evidence for the Paranilotic subgrouping was syntactic (VSO word
order), morphological type (particularly the fairly elaborate verbal
derivational morphology of the Eastern and Southern languages vs. the
reduced morphology of the Western ones), and unusual features such
as tonal case (found in the Southern languages and all of the Eastern
ones except Bari). None of this has proved sufficient to sustain
the Paranilotic grouping.

Chet Creider
creidercsd.uwo.ca

Message 2: Re: 5.1404 Comparative method in linguistics

Karl Teeter is mistaken, I think, when he says that you cannot
classify languages on the basis of phonological correspondences
in the lexical items, when he says that you would treat English and
French as related (via the Norman borrowings) if you did this, when
he says that the idea of using this method is "novel", when he says
that you can only classify lamnguages as related if you have written
a comparative grammar, etc.

(a) For many language families, there IS no other basis for classification
available, because they lack the kind of morphological complexity so
beloved of Indo-Europeanists, Algonkianists, and those other -ists who
happen to enjoy its blessings in "their" languages,

(b) The danger of confusing borrowings for cognates is always real,
but it is easy to see that the lexical connaections of English with
French are more transparent phonologically and less central semantically
than those with Dutch, hence more recent, hence borrowings.

(c) There is nothing novel in what I am saying, since it is the method
which, for example, Edward Sapir used to establish that the Uto-Aztecan
languages are really a family (rather than three families).

(d) You cannot possibly realistically expect normal people to
spend time writing comparative grammars of languages which have
not PREVIOUSLY been shown to be related. So classification must
come first! On the other hand, it is perfectly easy to write a
"crazy" comparative grammar for any random group of languages, e.g.,
French and English.

The last is perhaps the most important point: reconstruction of
morphological systems can be done well or badly, so can comparison of
lexical items. Both can yield correct results; both can also
produce garbage.

P.S. I say 'lexical items' deliberately, because there seem to be
language families where there is NO morphology at all to compare.
In most cases, howveer, there are SOME bound morphemes, and one
compares those as well (as did Sapir in the case of Uto-Aztecan).
But this is a far cry from reconstructing the kind of intricate
morphological patterns found in the older Indo-European languages,
for example, but absent from many other language families.

Alice Faber's very sound posting brings up a very important point:
indvidual sound changes, especially those which are known to be
common in languages of the world, cannot be the basis of subclassification.
What you need is either sets of unrelated changes (English has
diphthongized some of the same long high vowels as High German, so
that Eis and ice, Haus and house sound rather similar, but they
do not share other changes), or a chronology of the sound changes
with respect to other changes (e.g., the diphthongization in English
was presumably preceded by the loss of /x/ or /c,/, but there is
no such change in High German). On the other hand, we find that
the Northern Uto-Aztecan languages share the changes of intervocalic
/l/ -) /n/, /n/ -> velar nasal, and /c/ -> /y/. Any one of these
would mean little, but all three together (and the fact that the /l/
-) /n/ does not feed the /n/ -> /ng/ change) make it reasonably
certain that these really are shared innovations, and that Northern
Uto-Aztecan is a genuine classificatory unit (a valid node in the
Uto-Aztecan tree, if you will)--something that has until recently
been generally doubted.