Re: The map of typological features

John Atkinson wrote:
"Darkstar" <darkstar100@xxxxxxxx> wrote in message
news:1148741436.842961.72040@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

John Atkinson wrote:

(4) Noun classes and measure words don't occur in most Austronesian
languages.
Classifiers are used in many Austronesian languages (Indochina,
Sumatra, southern and central Sulawesi, east Indonesia, Micronesia,
etc).

But not, I think, in the numerical majority of Austronesian languages. (I
could easily be wrong, many Oceanic languages do have them).

OTOH, only one branch of AN has any classifier languages at all. The other
10 or so branches (the ones in Taiwan) don't. So, odds are the
proto-language didn't.

Anyway, there's still a considerable incidence of classifiers at least
in Malay, Javanese, some Polynesian languages (Kiribati, Samoan, etc),
Kilivila (east of Papua New Guinea, classifiers highly developed), and
especially in Micronesian. For example, in the Indonesian language
proper we have:

Saya beli empat buah jeruk.
I bought four [fruits] oranges.

As to the extensive use of numeral classifiers in Micronesian, well,
Micronesia is just a large island group closest to Taiwan that had
probably been completely uninhabited until the arrival of
Proto-Austronesians who apparently spread by utilizing kontiki-type
ships with sails (In Egypt/Sumer, the sail was a relatively recent
invention, c. 3000-4000 BC). Similarly, Cabo Verde was uninhabited
until 1450’s despite its proximity to Africa, because the ocean is
below 500 ft there, and the Ice Age bridges could not form. As a
result, there was no substratum to Micronesian, and it might be just as
close to PAN as you may wish. Generally, Oceanic languages conform to
Type B very well (analytic, typically sonoric, many nasals, prefixes,
inclusive/exclusive pronouns, some even have tones, and many have SVO
(!), not VSO), so they are supposed to be more archaic than Indonesian
groups which probably ran into Indo-Pacific substratum and consequently
don’t look so pretty.
I’d also remind of numeral classifiers in Japanese similar in form to
the ones in Micronesian (the numeral and the classifier coalesce into a
single word) – that might indicate the thing is ancient.
Anyway, it’s a whole article there. A special discussion of this
issue would be required.

The idea is that noun classes of Africa are related to numerative
classifiers of Asia,

No, the Niger-Congo classes are genders (i.e., agreement classes, agreeing
with verbs, adjectives, etc), not at all like classifiers (used with
numerals and determiners). According to Sasha Aikhenvald, the only
classifiers in Africa "are found in a few Kegboid languages (Cross River),
Ejagham, and a few Grassfields languages (e.g., NgyembOOn)".

Well, I suppose, it doesn’t matter if they agree or not.
Basically, there are three things in the world of noun: Euroasian
genders, African noun classes and Asian numeral classifiers, all of
which have something in common.
Genders are syntactic with no clear-cut morphological manifestation
(Junge, Maedchen – der, die, das?). Also, genders (and Nakh-Dagestan
noun classes!!) are often hidden and have no explicit markers at all.
Genders classify things only into 2-3 categories. They are rather a
part of the declension paradigm, sometimes showing up only when cases
come into play. Sometimes, they cannot even exist outside synthetic
languages, and that seems to be the reason why they are extinct in
English (even “he or she” is neuter these days, whereas a system
similar to numeral classifiers has arguably arisen – “a block of
ice”, “two bars of chocolate”, “a piece of information”,
"three pairs of jeans", etc)

On the contrary, good and true African noun classes and Asian
classifiers tend to have distinct morphological markers. They are a
whole metaphysical reality classifying objects into many semantic
categories. For this reason I see this second group as
*metaphysically-oriented, classificatory noun morphemes* (and this is
what I mean by B7) – they’re like the many tenses of the verb
system categorizing action in many specific ways -- whereas Eurasian
genders are neither classificatory, nor big-scale, nor even morphemes
(at times).
It's true that numeral classifiers are less developed and more fluid,
but still they’re closer to noun classes than genders, I believe.

Though, basically this talk does not affect the thesis itself, rather
the (dis)similarities in the gender--noun classes--numeral classifiers
triangle, as well as between the Asian and African typological
subtypes.

because they classfy nouns into arbitrary
static/dynamic semantic classes. On the contrary, Type A languages took
a different route of classifying nouns along certain situational
spatial/temporal criteria which led to the development of case systems.

(8) PIE, Afro-Asiatic and Kordofan have noun classes.
PIE seemed to have a rudimentary (?) system of animate and inanimate
classes that gave rise to genders. Though the question is interesting.
Anyone know of historical IE languages with clearly developed noun
classes?

French (2 classes)? Russian (6 classes)?

Again, in Russian, these are not proper semantic classes, but fusional
morphological phenomena observed only because of the complexities of
the declension. They have no semantic meaning of their own.

Noun classes in AA? Never heard. Maybe they have a different name
there. Can anyone help with this?
Kordofan languages are included.

AA has an agreement class system very very similar to those of the Dravidian
languages and the North and NE Caucasian languages, (Some people call them
genders, that's another word for the same thing.)

If it’s the same thing, I’ll call them genders.

(9) A majority of Afro-Asiatic languages have tone.
Not the majority. Chadic/Kushitic have them.

Indeed. And there are 135 Chadic languages. The entire rest of AA consists
of about 100 languages. QED.

Well, let them be. Chinese is also tonal. (And it happens to be the
language with most speakers in the world). Anyway, my system is not
supposed to work along the border. Border-line groups are just supposed
to mix, match and blend, that's only natural.

Maybe an expert in AA can
provide more info. Again, when classifying AA, we should exclude all
southern languages that are in close contact with Nilo-Saharan/NC to
identify the proto-state type.

Or, should we exclude all northern languages that have been in close contact
with Elamite, Hurrian, Hittite, Sumerian, etc, all of which were, arguably,
toneless, to identify the proto-state type?

Chadic are *right* at the border, so the intercultural and ethnical
contact there is extremely intense even more so than with Sino-Tibetic
or Japanese/Ainu/Austronesian. Consider Tibeto-Burman and Japanese –
they definitely changed to resemble other languages in their local
areas. That's absolutely natural and predictable. On the other hand,
Iran, Iraq and Arabia are areas free of any “blue” languages.
There’s nothing but the Indian Ocean down there.

FWIW (not much) the (marginally visible) resemblances between AA and
Nilosaharan suggest to me (oh yeah) that their proto-languages must have
been in contact.

More seriously, there's no reason to think that any currently existing tonal
system in any language family in the world dates back much more than a
couple of millennia -- it's a pretty fluid feature.

There might exist a general mechanism that precludes it from
dissolution. It looks like Chinese preservesd tonality for at least
3300 years judging from the intense monosyllabism and isolation in
Wenyan. The Tao by Laotsu (600 BC) written in verses seems to read
well, too. The extensive use of tones in Khoisan languages that are
seen as prehistoric because of their rare consonantism indicates that
the African tonal system is also very old. Generally, I believe that
analytic languages may encapsulate tonality for very long periods of
time, because once a particular feature is there, it no longer takes
just a minor change to get it out – rather it takes to transform the
whole system of grammar and phonology in the whole geographic area,
because changes inflict other changes. My whole point is probably that
all the features described in the research are partly interdependent,
so they work together as one structural type, forcing the language to
evolutionate along a certain, predetermined trajectory.

(10) Maban is SOV. Mebaan is SVO. Both are Nilo-Saharan, but they're
not
geographically or linguistically close.
John.
Maban is Western Nilotc, and the other one is a branch-off isolate,
which is strategically more important, and it has SVO.

The language called Mabaan in the Ethnologue is called Mebaan by Bender and
classified by him as Nilo-Saharan: East Sudanic: Nilotic: E9a: Northern.
The same sub-sub-sub-group E9a contains 15 million speakers, about half the
total for all NS, and includes Shilluk, Luo, and Dinka. I know everyone has
their own pet way of dividing up NS, and their own opinion whether it is or
isn't a true genetic family, but it's hard to imagine that anyone could call
Mabaan an isolate.

John.

Well, South Burun, or Maban (whatever the spelling) is indeed Western
Nilotic. But *the other* Mabaa (or of similar name) is part of a small
Mabaan top-level group of NS languages.

The two languages are hardly related. Here are the numerals for 1,2,3.

Maban (South Burun) (Nilotic)
1 kielo
2 yio
3 drogo

Mabaa (top-level NS)
1 teen
2 mbаar
3 kongаal

.

Quantcast