Cavalli-Sforza Luigi Luca. Genes, Peoples and Languages

in detail. I

am

not

a linguist,

but

I find Greenberg's arguments most

convincing. Furthermore, Greenberg has

been

through this before.

Many

years ago,

he

proposed a now widely accepted classification

of

.

African languages into only four families: Afroasiatic, including all

Semitic languages and most languages in Ethiopia

and

North Africa;

Nilo-Saharan, comprising languages spoken along

the

upper

Nile

and

the

southern Sahara; Niger-Kordofanian, which includes most

central, southern, and West African languages, especially

the

Bantu

ones;

and

Khoisan languages, spoken by Khoi-Khoi

and

San popula-

tions in southern Africa. Greenberg's classification

endured

a bar-

rage

of

criticism when it

was

first proposed, and it is now widely

accepted. A similar change

of

attitude toward

the

Amerindian classi-

fication will probably come with time.

An

examination

of

some

of

the

objections

that

Greenberg's col-

leagues have raised against his classifications can help us understand

both

the

objective difficulties that affiict studies

of

linguistic evolu-

tion, and also

the

subjective ones, which are typified by Greenberg's

attaclrers. Languages change very rapidly and it is terribly hard to

establish clear connections between distant languages. Significant

phonolOgical and

·semantic changes to all languages occur over time.

The

magnitude

of

these changes complicates

the

reconstruction

and

evaluation

of

linguistiC commonalities. Grammar also evolves,

although usually sufficiently slowly

to

aid tlle recognition

of

more

ancient linguistic connections.

Under

the

pressure

of

phonetiC

and

semantic change, a language rapidly becomes incomprehensible.

Modem

languages derived from Latin would not

be

understandable

to a Roman

of

two thousand years

ago-a

thousand-year separation

is often enough

to

render

a language incomprehensible

to

its first

speakers. After a separation

of

five

to

ten thousand years,

the

rate

of

recognizably similar words can drop to

ten

percent

or

less. FOltu-

nately, certain words and certain parts

of

speecll exhibit a slower rate

of

change and give us a

better

chance

of

discerning more distant lin-

guistic relationships.

As

for

the

problems caused

by

misguided methods, some anti-

Greenberg

linguists believe it

is

impossible to posit a quantitative

relationship between any two languages

..

By

disallOwing reliable

137

measurements,

and

by

limiting

the

relationship

between

two lan-

guages only

to

"related

or

not

related,"

the

American linguists

opposing

Greenberg

have

ruled

out

the

possibility

of

hierarchical

classifications,

an

essential prerequisite

to

taxonomy.

Interestingly,

tins. position completflly contradicts

the

view

of

linguists

who

use

sophisticated methods to

measure

linguistic simi-

larity,

based

on

the

fraction

of

words from a

standard

list

that

have a

detectable

common

origin. This approach was developed by

an

American linguist, Morris Swadesh.

He

suggested

that

the

proba-

bility

that

a

word

will lose its Original meaning stays constant over

time.

When

the

fraction

of

related words

retained

after a known

peliod

of

time

has

been

calculated (by observing

the

change from

Latin

to

the

modern

Romance languages, for example),

we

can

con-

struct a "calibration" curve,

allOwing us to

read

the

time

elapsed

since two living languages shared a

common

ancestor. This

method,

which has

been

called "glottochronology," uses a "linguistic clock"

very similar

to

the

"molecular clock"

we

discussed previously for

genetiCS.

In

biology,

we

have

the

advantage

of

using

many

proteins

or

DNA

sequences to obtain many

independent

estimates

of

the

date

of

separation

between

two species. Unfortunately,

there

is

no

such variety

and

wealth

of

data

in

linguistics to

strengthen

our

con-

clusions. Glottochronology is a less rigorous

method

than

those

used

in

biology,

and

is especially difficult to apply

to

distant com-

parisons

when

the

fraction

of

cognate words

more

than

ten

thou-

sand

years ago is very small.

The

word

lists

cannot

grow,

becanse

only a limited

number

of

words change slowly; moreover, every.

word

has its own rate

of

change, a fact neglected

by

glottochronol-

ogy, which assumes a constant rate

of

change.

Other

linguists insist

that

the

resemblance

between

similar

words in different languages

be

examined

in

light

of

"classical

sound

correspondences," which

are

very strict rules

of

phonolOgical

change.

If

these

rules

are

not

followed exactly,

they

say,

two words

cannot

be

considered

"cognates,"

that

is,

one

cannot

determine

if

they

share

a

common

origin.

Greenberg

has

answered

them

with

an

impressive list

of

exceptions

to

these

rules within

the

Indo-

European

language family

and

others.

He

concluded

that

it

would

138

be

impossible to establish

the

Indo-European language family

if

these rules

were

applied too rigorously. Fortunately,

the

Indo-

European

family was proposed and accepted before

the

theory

of

sound correspondences adopted its most rigid fonnat.

Finally, some linguists believe that

the

parent

language

that

gave rise to a family

or,

in general, a cluster

of

languages must

be

reconstructed in

order

to

demonstrate a phylogenetic relationship

among families.

Here

too, biology provides

an

analogy-when

the

"consensus"

DNA

sequence

is calculated from

the

sequence

of

two

modern species.

The

consensus sequence is

the

best

guess

of

the

ancestral sequence that would require

the

fewest changes to gener-

ate

the

diversity observed in a particular sample.

But

the

search for

consensus is less rigorous in linguistics because linguistic variation

is much wider

than

biolOgical variation, since only four nucleotides

make

up

DNA.

In

"biology,

some proteins are so important to an

organism

that

little

or

no change could

be

tolerated. TIms, many

sequences

of

proteins change extremely slowly,

and

it is possible to

prove

their

relationship without reconstructing ancestral sequences

of

millions

or

even billions

of

years ago. Knowledge

of

a proto-

language may

be

helpful in comparative analyses,

but

imposing this

exercise

on

all linguistic claSSifications is a serious limitation when

so few proto-languages have

been

generated. Furthermore, recon-

structions have a low probability

of

being entirely reliable. Green-

berg's

method

avoids this impasse.

It

may

be

more

subjective

than

is deSirable,

but

it

can go

much

further

than

other

methods.

TI,e classification

of

families by Merritt Ruhlen (a student

of

Greenberg's) appears to

me

to

be

satisfactory for comparing genetic

and linguistic evolutions,

as

we

will do in

the

next section. Defining

a family does

not

appear

to

be

an entirely objective task,

but

the

distinctions

between

families, subfamilies,

and

supetfamilies are

mostly a

matter

of

convenience

and

are unnecessary for certain

purposes.

What

matters is

the

possibility

of

establishing a simple,

logical, and hierarchical relationship. Unfortunately, most modern

classifications stop at

the

level

of

families,

of

which

there

are as

many

as

seventeen in Ruhlen's unifYing system.

There

are some

superfamilies,

but,

as

already noted, modern linguistic methods

139

have

not

yet generated a complete

tree

growing from a single

source.

It

is

of

interest

to

consider some

of

the

proposed superfamilies,

even

if

they are rather Controversial. According

to

Ruhlen, Austric

is

a

superfamily consolidating four families: Miao-

Yao

(spoken in pockets

of

southern China, and northern Vietnam, Laos, and Thailand); Aus-

troasiatic, comprising Munda langnages spoken in northern India

and

Mon-Khmer (spoken mostly throughout Southeast Asia); Daic (spo-

ken widely

in

southern China and

much

of

Southeast Asia), and Aus-

tronesian.

There

are some 1,000 Austronesian languages, spoken by

about

one

hundred eighty million people, including Taiwanese abo-

rigines

and

Malayo-PolynesianS.

The

latter group ranges from Taiwan

to

Polynesia, parts

of

Melanesia,

the

Philippines, Indonesia, Malaysia,

and

as

far west

as

Madagascar.

The

oldest Austronesian langnages are

spoken by Taiwanese abOrigines.

Whether

or

not

this superfamily is

accepted, it ties together a

vel)' wide geographic region, including

Southeast

Asia,

both insular and non-insular areas, and a

large

num-

ber

of

islands in

the

two oceans separated by Southeast

Asia.

Superfamilies extending over Europe

and

Asia are

of

particular

interest.

1Wo linguistic groupings in this region are

current

today

and

are closely related: Nostratic and Eurasiatic. Rejected at first by most

linguists,

they

are slowly being recoguized.

The

Nostratic superfam-

ily,

as

originally described

by

various Russian scientists, includes

the

Indo-European, Uralic (spoken across

the

uraIic mountains), Altaic

(widely spoken in Central Asia), Afroasiatic, which comprises many

North African

and

also Semitic languages, Dravidian (currently spo-

ken almost only in southern India), and

South Caucasian families.

One

Russian linguist, Vitaly Shevoroshkin, showed

that

the

Nostratic

superfamily has strong similarities

to

the

Amerindian group, as

defined by Greenberg.

The

Eurasiatic superfamily proposed by

Greenberg is similar

to

Nostratic,

but

it

is different in

the

extension

given

to

some families like Altaic

and

comprises smaller families like

Eskimo and Chukchi, in addition

to

Japanese. Thus, Eurasiatic

extends further east

than

Nostratic,

but

not

as

far southwest,

as

it

does

not

include Afroasiatic and Ilravidian, which, according

to

Greenberg, have an older origin.

140

One

could continue building

on

this

tree

by

adding an earlier

branch, leading

to

the

Nostratic/Amerindian group

on

·one side,

and to a new, older superfamily

on

the

other, Dene-Caucasian. Sapir

initiated this

new

grouping,

but

Sergei Starostin proposed it offi-

cially only a few

years ago.

The

Dene-Caucasian superfamily includes

essentially

three

families: North Caucasian, Na-Dene,

and

Sino-

Tibetan.

The

last family

of

languages is spoken ·by almost a billion

individuals (in

China; India, Nepal, Burma, and also Southeast Asia,

plus some isolates

hl

Europe

and

West Asia),

and

is therefore

the

most populous one.

Thus,

one

could draw this approximate diagram:

Eurasiatic Afroasiatic Dravidian

V

Nosh-atic Amerindian

v~-

Eurasian

This hierarchy

of

superfamilies includes almost all

of

Europe,

northern Africa, most

of

Asia,

and

all

of

America.

It

lacks only

three

African families, Khoisan, Niger-Kordofanian

and

Nilo-Saharan,

as

well

as

Australian (170 languages), and Indo-Pacific, a group

of700

languages, spoken mostly in New Guinea

but

also in neighboring

islands

and

in

the

Andaman Islands

near

Malaysia.

There

is, however, a small group

of

languages called isolates,

which most linguists

are

unable to clasSify in any

of

the

better-

established families.

The

best

known is Basque. Still spoken by

approximately

12,000

French

and

perhaps a million

and

a half

Spanish, this language is likely a relic

of

a pre-Neolithic period and

is

pOSSibly

the

language spoken by Cro-Magnons,

the

first

modem

humans in Europe.

But

it has certainly changed enough

141

that

modem

Basques

and

Cro-Magnons would

not

be

able

to

communicate

if

they

had

the

chance to meet.

In

fact

they

probably

wouldn't even recognize

their

languages

as

related. Several linguists

suggest a relationship

between

Basque

and

modem

languages

of

the

northern

Caucasus.

It

is thus possible

that

one

or

more

pre-Indo-European

languages were spoken in Paleolithic

Europe.

Other

linguists see even

more

encompassing resemblances

among

Basque, Caucasian, Sino-Tibetan,

and

Na-Dene

languages.

The

last

are

spoken in

the

northwestern region

of

North

America. Another

claim

is

that

Burushaski, an isolate spoken in a high valley in

the

Himalayas, is related to Basque

and

Caucasian. And, according to

other

linguiSts, Sumerian, Etruscan,

and

other

linguistic "fossils"

belong to

the

same ancient

family,

Dene-Caucasian.

If

the

Nos-

tratic/

Amerindian

group

were

added

to

the

Dene-Caucasian

group

to form a hypothetical Eurasian superfamily, it may have

extended

across all

Europe

and

Asia (except

the

southeast),

and

spread

to

the

Americas. This giant superfamily later differentiated into several

branches,

and

local

twigs

of

this

tree

flourished

and

extended

to

regions far

and

wide.

These

are interesting

and

hopeful hypotheses

that

need

to

be

explored further.

If

we want

to

stay

on

absolutely firm ground,

the

situation is worse

than

simply lacking a reliable

tree

to link

all

mod-

em

languages: it is

not

even certain that

all

languages share a com-

mon

origin. Most linguists consider both problems insoluble. It's a

bit like trying

to

determine

if

all

life

on

the

planet

had

a Single ori-

gin. (Many

biolOgists believe in a Single origin, since

there

is

only

one

form

of

the

twenty aminoacids found in proteins.)

Greenberg

noticed

that

there

is at least

one

word all

the

linguistic families

share:

the

root ttk.

It

means finger,

or

the

number

one

(a semantic

shift, which

reqUires

no

explanation).

In

other

languages

we

find

other

semantic changes

of

this root, which also

appear

acceptable,

"hand"

and

"arm," for example,

or

"pOint, indicate."

In

French,

doigt (pronounced "dwa")

and

in Italian, dito (meaning "finger")

come

from

the

Latin root digit.

Extending this example, American linguist John D. Bengtson has

proposed, with Ruhien, about thirty

other

roots

that

are nearly as

142

universal.

But

it will take a long time for

other

linguists to examine

and.accept these newest results.

As

might

be

expected,

there

are

vel)' few roots

that

are common

to

most languages. Most

of

them

designate parts

of

the

body,

or

are

personal pronouns

or

small num-

bers (one,

two,

and

three).

It

is

not

surprising

that

words conserved

since

the

beginning

of

linguistic diversification are among

the

first

words

we

learn: eyes, nose, mouth, and so on.

But

there are others

that were certainly very important in Paleolithic life

and have

been

preserved in many languages; "lice" is

one

example.

Comparison

of

linguistic

Families

with

the

Genetic

Tree

Even

without a comprehensive linguistic tree,

we

can still compare

our

genetic

tree

to existing linguistic trees.

There

are some impres-

sive similarities.

In

figure 12,

the

language families have

been

drawn next

to

the

populations

that

speak

the

respective languages. We see

that

a fam-

ily corresponds

to

one

or

more

branches in

the

genetic tree. Some-

times a language family is represented in

the

joint genetic-linguistic

tree

by

only

one

branch, because

the

populations speaking these

languages

were

grouped together in

the

genetic analysis.

In

effect

they

show great Similarity, either genetic

or

ethnographic, and live

near

each other.

An

example is

the

Bantu subfamily, belonging to

Niger-Kordofanian, which is genetically homogeneous

and

distinct

from

other

African groups. Although

the

word

Bantu deSignates a

language group, it is also useful

as

a biolOgical categol)'.

Other

genetic groupings also have

been

corroborated by linguistic infor-

mation; for example, southern Indians speak Dravidian languages

and

the

Na-Dene

speak Native American ones. Shared language

families often

point

to a common genetic and ethnic background.

The

genetic

tree

shown in figure 12 is composed

of

38 popula-

tions, some

of

which are broadly grouped (e.g., Europeans, Melane-

sians).

There

are only 16 language families (we

had

no genetiC data

on

Caucasian populations when we

drew

this tree). Therefore some

143

Genetic'Ii'ee

Linguistic

Families

Mbnti Pygmy

---

Original language unknown

::~rican

~

Niger-KordofanJan

Nilotic NilQ.·Saharan

San

(Bushmen)

--

IChoisan

Ethiopian~

Berber, N

..

A~

Afro-Asiatic==============11

S.

W.

Asian

::

Iranian

I:

Europe~

::

Sordinia~

Indo-Euoopean

=====il

In.dian

::

S.

E. Indian

___

._

Dravidian

========

-1',1)

t=J--

1I

Z

--i

~

.-::

i

~~~d

=-=---t-:

Uralic-Yukaghir •

.g.

i·=il

~

I

'}

g.

::

~

~~:;O~

Sino-Tibetan

5i::

~

Korean

::

~

Japanese

AJtaic

==

= =

=:oJ:

l

Ainu .

N.

Turkie

Es'khno Esld.mo-Aleut

Chukchi Chukchi-Kamehatkall I

s.Amerlnd3

:

C.Amerind Amerind - -

--

-

--

-

__

I

N.Amerind

N.

W.

Amerilld

--

Na-Dene

S.

Chinese

----

Sino-Tibetan

Mon-Khmer---

Aust~.atie

Thai Daie

Indonesian Austrie

Malaysian

~

Philippine Austronesian

Polynesian

Micronesian

Melanesian

-----,.....,.

I d P

'6

New Guinean

~

11

0-

aCl

Ie

Australian

----

Austnilian

Figure 12.

The

comparison

of

genetic and linguistic trees (Cavalli-Sforza

et

al.

1988. pp. 6002-6).

144

populations

in

the

genetic

tree

must

and indeed usually do belong

together

to

the

same linguistic family.

We

can immediately

note

that

populations

that

are

adjacent

in

the

genetic

tree

usually speak lan-

guages

of

the

same family. Because

of

this

we

can

use

the

genetic

tree

to

help

us

date

the

approximate origin

of

a linguistic family.

Most language families appear to have developed dUling a

brief

period,

between

6,000

and

25,000 years ago.

There

are, however, exceptions to

the

tendency

of

populations

speaking related languages to

be

close

in

the

ge~etic

tree. Ethiopians,

for example,

are

palt

of

the African genetic brancll,

but

they gener-

ally speak languages from

the

Afroasiatic family. whicll

are

wide-

spread

in

North Africa

and

the

Middle East, where

the

people are

generally

Caucasoid. Ethiopians are

in

effect a bit more African than

Caucasoid genetically

and

more· Caucasoid than most Africans lin-

guistically.

The

Saami (Lapps) illush-ate another exception to this

mle: they are genetically Caucasoid

but

speak a Uralic language

whose

other

representatives live mostly in northeastern European

Russia

and

northwestern Siberia

near

ti,e Urals. TI,e Uralic people

of

Asia

are

generally MongolOid,

and

ti,e Lapps are a

mix

of

Caucasoid

(probably from Scandinavia)

and

MongolOid (of Siberian origin),

with a prevalence

of

the former. Even without looking at

their

genes,

we

can

see

this in

the

color

of

their

skin

and

hair,

and

in

the

color

and

shape

of

their

eyes, whicll vary

in

type from MongolOid to Caucasoid.

There

is a simple explanation for

these

disagreements between

genetic

and

linguistic classifications. As

we

have explained, tI,ese

two populations are

the

products

of

relatively recent genetic admix-

ture:

of

Europeans

and

Sibelians for

the

Lapps,

and

of

Africans

and

Arabs for

the

Ethiopians.

In

the

genetic tree,

the

populations are

placed with

ti,e groups that

conhibuted

t1;e

greater proportion

of

genes. Extended admixture may also

put

them

in a more isolated

and

somewhat intermediate position

in

the

tree.

The

genetic effects

of

population admixture

are

mucll Simpler

and

more

predictable

than linguistic

cllange. Genes

of

a mixed population

occur

in

pro-

portions corresponding to those

of

its ancestral parental popula-

tions.

But

a genetically mixed population

tends

to preserve only

one

of

the

two original languages. Sometimes,

the

language

of

a mixed

145

population will

not

change

at

all; more often, however,

we

find a few

words or, sometimes, sounds borrowed from

the

other

language.

Grammar is

more

resistant

to

change

than

vocabulruy. As for

the

ori-

gin

of

the

mixtures

that

generated Ethiopians and Lapps, we know

of

intimate contacts between Arabs and'Mricans in Ethiopia during

the

first millennium

B.C.

An Arab-Ethiopian kingdom established a

capital first in Arabia (in

the

Saba region)

and

later in northern

Ethiopia (at Aksum)

..

But

earlier contacts may have occurred

that

took place too early for history

to

record. We also know

that

the

Lapps have

been

in

their

present

territory for

at

least 2,000 years.

In

both these cases,

the

shortage

of

written records limits

our

ability to

know just how

far back contact originated.

In

each case,

the

degree

of

genetic mixing established can vary

and

depends

on

the

amount

and

mode

of

contact between two populations.

It

is easy to calculate

that

the

genome

of

a people

who

received

a constant genetic flow

of

5

percent

per

generation from its neigh-

bors would keep only

70

percent

of

its original genome after

three

centuries. This is

the

average value

of

admixture for Mrican Ameri-

cans, who have retained

70

percent

of

their original genome

and

received 30

percent

from mostly white settlers.

If

this flux contin-

ues with

the

same speed, Mrican Americans will have little

more

than

10

percent

of

their

original genes after 1,000 years

of

habita-

tion in America.

In

the

cases

of

Lapps and EthiopianS, their

parent

populations may have

been

in reciprocal contact for a long time

(perhaps several thousand years),

and

.degrees

of

admixture are

greater than those observed in African Americans, who were in

contact with whites for a much shorter

pedod,

and

under

condi-

tions

of

strong social inferiority.

We can find still

other

interesting exceptions to exact ·corre-

spondence

of

the

genetic

and

linguistic trees. Genetically, Tibetans

belong to

the

group

of

northern Mongols,

but

they speak a Sino-

TIbetan language like

the

Chinese.

The

Chinese represented in

our

tree, however, originate from southern China and

are

genetically

more similar

to

southern Mongols. History comes

to

our

rescue in

this case as well. According

to

Chinese histOrians, TIbetans are re-

lated

to

the

northern Chinese. Starting from

the

third century

B.C.

146

Cavalli-Sforza Luigi Luca. Genes, Peoples and Languages

Подождите немного. Документ загружается.