Indo-European_Renfrew - Prehistory Of Languages Xpm.pdf - NEW SHIT - havocslasher

Cambridge Archaeological Journal 10:1 (2000), 7–34

At the Edge of Knowability:

Towards a Prehistory of Languages

Colin Renfrew

The issue of ‘knowability’ in relation to the origins and distribution of the language

families of the world is addressed, and recent advances in historical linguistics and

molecular genetics reviewed. While the much-debated problem of the validity of the

concept of the language ‘macrofamily’ cannot yet be resolved, it is argued that a time depth

for the origins of language families greater than the conventional received figure of c. 6000

years may in some cases be appropriate, allowing the possibility of a correlation between

language dispersals and demographic processes following the end of the Pleistocene

period. The effects of these processes may still be visible in the linguistic ‘spread zones’,

here seen as often the result of farming dispersals, contrasting with the linguistic 'mosaic

zones' whose early origins may sometimes go back to initial colonization episodes during

the late Pleistocene period. If further work in historical linguistics as well as in archaeol-

ogy and molecular genetics upholds these correlations a ‘new synthesis’, whose outlines

may already be discerned, is likely to emerge. This would have important consequences for

prehistoric archaeology, and would be of interest also to historical linguists and molecular

geneticists. If, however, the proposed recognition of such patterning proves illusory the

prospects for ‘knowability’ appear to be less favourable.

T his article, based upon the eleventh McDonald Lec-

ture, 1 is about linguistic diversity and the origins of

the 6500 or so languages spoken in the world today.

These origins may be traced back in some cases over

the past 10,000 or so years. Controversial claims have

been made that some features may be traced back

even further. But the more general question of the

origins of language itself, as a typically human ca-

pacity, will not be discussed, although it is one of the

most fascinating issues in contemporary archaeol-

ogy (Mellars 1998; Pinker 1998). There is a general

consensus (Noble & Davidson 1996) that a fully mod-

ern language capacity is a feature of our species

Homo sapiens sapiens , and that this is likely therefore

to have been the case of our sapiens ancestors of

more than 40,000 years ago, and I do not propose to

consider greater time depths than that. The title of

this article is drawn in part from the current inter-

ests of the Alfred P. Sloan Foundation of New York

on the whole issue of knowability . The question is not

so much about the extent of our current knowledge,

but about the limits to knowledge and the ways by

which we can assess what is in principle knowable,

given favourable conditions of research, and what is

in principle unknowable, being beyond the reach of

human investigation. Some of the research described

here has in part been funded by a generous grant

from the Sloan Foundation applied to the project

‘The Prehistory of Languages’ conducted at the

McDonald Institute.

My own concern for this theme dates back to

my first research in the Aegean and the realization

(Renfrew 1964) that earlier inferences about the chro-

nology of the prehistoric place names of the Aegean

(Haley & Blegen 1928) were invalidated by the dis-

covery of Neolithic settlements on the Cycladic Is-

lands (Evans & Renfrew 1968). This line of reasoning,

applied to the origins of the Greek language, led to a

growing scepticism about the current consensus on

the origins of the Indo-European languages of Eu-

rope (Childe 1926; Gimbutas 1973) that their advent

was due to the activities of warlike nomad pastoralists

Colin Renfrew

Figure 1. The early spread of agricultural communities

in Europe, as summarized by Stuart Piggott in 1965.

1) Proto-Sesklo and Star ˇ evo cultures; 2) Linear Pottery

culture; 3) Impressed pottery culture. It is suggested

below that the early spread of Proto-Indo-European

speech accompanied the dispersal of farming in these

areas. (From Piggott 1965, 57.)

Figure 2. Radiocarbon chronology (uncalibrated) as

proposed by Grahame Clark in 1965 for the spread of

farming from Anatolia to Europe. Radiocarbon

determinations are shown for the earliest sites of farming

settlements as dated by 1965. The direction and source of

the farming dispersal is very clear. (From Renfrew

1973a, 71.)

from the Pontic steppes. It invited instead the for-

mulation of an alternative theory.

The quest for some radical change in European

prehistory with more convincing explanatory power

than was offered by the alleged coming of the

mounted warrior nomads was influenced by the clear

picture on farming origins outlined for instance by

Piggott (1965, 57, fig. 26) as seen in Figure 1. This

was supported by the striking patterning for the

chronology of the dispersal of farming in Europe

(seen in Fig. 2) established on the basis of radiocar-

bon determinations by Grahame Clark (1965, 46). I

therefore proposed (Renfrew 1973a) that the spread

of farming to Europe from Anatolia was the princi-

pal agency responsible for the arrival and dispersal

of Proto-Indo-European speech. If this proposal were

accepted it would follow that the ‘homeland’ for the

original Proto-Indo-European language would be

situated somewhere in south-central Anatolia (where

James Mellaart had been making striking Early

Neolithic discoveries) and that the arrival of Proto-

Indo-European speech into Europe from Anatolia

could be dated using the available chronology for

the coming of farming to around 7000 BC .

In developing the farming dispersal thesis for

Indo-European into a more comprehensive statement

in Archaeology and Language (Renfrew 1987), I uti-

lized the elegant device proposed by Ammerman &

Cavalli-Sforza (1973) of the ‘wave of advance’ which

modelled the propagation of the farming economy

by a process of ‘demic diffusion’. Although this was

soon criticized by Zvelebil & Zvelebil (1988; 1990) in

its application to the European case, and with some

justice, as an oversimplification, the basic notion of

the arrival of Proto-Indo-European speech in Europe

with the coming of farming has nonetheless been

widely accepted (Zvelebil 1995). At the same time,

the notion of a single Proto-Indo-European dispersal

was seen by many linguists as too simple a hypoth-

esis to account for the complexity of the relation-

ships among the various sub-families of the Indo-

European language family, well illustrated for

instance by Raimo Anttila’s diagram (Anttila 1989)

of isoglosses. Both objections were valid. They are

discussed further below.

In retrospect, then, Archaeology and Language

suffered from two principal defects (among other

deficiencies). In the first place, it laid too much store

by the ‘demic diffusion’ model of Ammerman &

Cavalli-Sforza and not enough on the phenomenon

of contact-induced language change. Ironically per-

haps it is the application of molecular genetics which

has given new insights into the limited extent to

which demic diffusion took place (see below). And

At the Edge of Knowability

secondly it did not sufficiently deal with the com-

plexity of the further developments of the Indo-Eu-

ropean languages in the long time span after the

initial Proto-Indo-European spread, with local con-

vergence (or advergence) effects responsible for the

formation of some of the sub-families (see now Ren-

frew 1999). But it did, perhaps for the first time in

the context of recent discussions, establish two prin-

ciples or processes which turn out to be applicable

on a much wider canvas than that of prehistoric Eu-

rope. The first is the creation or at least the foundation

of a language family not through some long-distance

tribal migration (like that of the legendary mounted

warrior-nomads) but through a spread phenomenon

which may be seen as the result of an intelligible

economic and demographic process (on the subsist-

ence/demography model). In this case the spread of

farming — a prime case of what Dixon (1997), fol-

lowing Stephen Jay Gould, was later to term a ‘punc-

tuation’, leading to the formation of what Johanna

Nichols (1992) was to name as a linguistic ‘spread zone’.

The second principle is a matter of time depth.

Hitherto, historical linguists have in general come to

employ what the archaeologist would term a ‘short

chronology’, with a time depth for language families

of often just five or six thousand years before the

present. But many of the decisive demographic proc-

esses in world history are climate-related, either the

direct consequence of such phenomena as the end of

the Late Glacial Maximum or the end of the Pleisto-

cene period, or the indirect consequence, dependent

for instance upon the origins of farming, a process

which is now viewed in most parts of the world as

initiated by those climatic events but slower to de-

velop in some areas. By establishing a date as early

as 7000 BC for the first spread of a Proto-Indo-Euro-

pean language (or at least by seeking to do so) it

became possible to view the origins of language fami-

lies within a time-frame linked to the end of the

Pleistocene period which allowed such global

changes to be regarded as relevant to the issue. This

is a principle which, for instance, Nettle (1999a) has

used to very good effect.

It was the publication of Merritt Ruhlen’s very

clear survey A Guide to the World’s Languages (Ruhlen

1991) which opened the way for the employment of

the same principles, using explicit models (Renfrew

1989a) for language change. The wider application

of the language/farming dispersal model, first in

Eurasia (Renfrew 1991) and then much more widely

(Renfrew 1992b; 1996), suggested a global solution

to the problem of the spatial distribution of the

world’s languages, although one which at first

presents a number of problems, particularly linguis-

tic ones. It is noteworthy that over the same period

Peter Bellwood, working in the Pacific, first with the

Polynesian and then with the wider Austronesian

language families, reached similar conclusions on

the relationship between the dispersals of languages

and of agriculture (Bellwood 1989; 1991) in the gen-

esis of language family distributions. He has subse-

quently generalized these to a world level (Bellwood

1996; 1997).

The third strand in this interdisciplinary scene

is the application of molecular genetics, in which

Luca Cavalli-Sforza (Cavalli-Sforza et al. 1988; 1994)

has been a pioneer, and where the pace of new re-

search is extraordinarily rapid. I concluded a review

of these matters in 1991 by predicting the emergence

of a ‘new synthesis’ between historical linguistics,

prehistoric archaeology and molecular genetics (Ren-

frew 1991, 20). These circumstances form the back-

ground to the present enquiry. Matters are made

more interesting by a number of developments in

historical linguistics offering greater insights into

processes of language change. There is also the vexed

question of so-called linguistic phyla or macrofamilies

(such as ‘Nostratic’ and ‘Sino-Caucasian’) which, if

their validity could reliably be established, would

have considerable significance for our understand-

ing of prehistory. For instance, the current distribu-

tions of the languages comprising the constituent

language families of the hypothetical Nostratic

macrofamily (including the Indo-European lan-

guages) could plausibly be explained by the wider

application of the farming/language dispersal model

(Renfrew 1991). But, as we shall see, the current

status of such macrofamilies is problematic. There is

also the potential application of the techniques of

molecular genetics to linguistic problems through

the mediating concept of population demography.

It was in the light of this very complex situation

that it seemed appropriate to use the resources of

the McDonald Institute (whose founder Dr D.M.

McDonald was first led to visit the Department of

Archaeology in Cambridge in 1988 through his in-

terest in the issues raised in Archaeology and Lan-

guage ), and to enlist the support of the Sloan

Foundation to examine some of these problems

through the ‘Prehistory of Languages’ project.

On knowability

At first sight the suggestion of investigating the pre-

history of languages may seem like following a will

o’ the wisp. For if past languages are recorded only

Colin Renfrew

though the medium of writing, first developed in

western Asia and in Egypt in the fourth millennium

BC and considerably later elsewhere, how could one

hope to consider linguistic events prior to that date?

Fortunately the well-established discipline of his-

torical linguistics can supply at least a partial an-

swer to that question. It does so through the concept

of the language family, established already in 1786

by Sir William Jones (1807). He indicated, with refer-

ence to what is now termed the Indo-European lan-

guage family, that there were a number of related

languages (Greek, Latin, Sanskrit) and sub-families

(Celtic, Germanic) which must be ‘sprung from some

common source’, thereby implying both the notion

of an earlier proto-language (such as Proto-Indo-

European) and an original homeland. Such concepts

are now basic to the discipline of historical linguis-

tics and any history of the discipline will indicate

how the Neogrammarians of the nineteenth and early

twentieth centuries, notably Brugmann, established

the principles of linguistic reconstruction (e.g. Durie

& Ross 1996, ch. 1) with an understanding of regu-

larities in sound change (phonology) and systematic

modifications in word forms and syntax (morphol-

ogy), accompanied by word loss and innovation in

vocabulary (lexical change). The reconstruction of

the hypothetical proto-language can thus be under-

taken, and confidence is gained in the method, for

instance when other languages are discovered which

may be recognized as belonging to the same family

and conforming with comparable principles in their

inferred descent from the reconstructed proto-language.

It is therefore not a vain question to enquire,

using the techniques of archaeology, what material

traces may have been left by the speakers of these

various languages both in the regions where they

are recorded as having been spoken, and in the in-

ferred homeland area. There is a long history of such

researches, not least in the field of Indo-European

studies (Mallory 1973). The matter is of considerable

interest to archaeologists as well as to linguists, since

many archaeological conclusions are based on broad

statements about such issues as the ‘coming of the

Celts’ or the arrival of the first Indo-European speak-

ers (Schrader 1890; Childe 1926). Comparable issues

emerge in every part of the world in relation to the

relevant languages (e.g. McConvell & Evans 1997;

Blench & Spriggs 1997; 1998; 1999a,b).

The relationship between archaeological and

linguistic data has, however, always been a difficult

one. The old equation, established by Childe (1929),

between an archaeological culture, a people and a

language is no longer directly acceptable (Renfrew

1989a). Moreover, while material cultural remains

can be dated with considerable precision by such

techniques as radiocarbon dating, linguistic events

are more difficult to date (Renfrew et al. in press).

And when the data from molecular genetics are

brought into play, the situation becomes even more

complicated (Cavalli-Sforza et al. 1994; Renfrew

1992a).

The matter has become more acute recently with

the postulation of such macrofamilies as ‘Nostratic’

(Dolgopolsky 1998) and ‘Amerind’ (Greenberg 1987)

or indeed ‘Sino-Caucasian’ (Starostin 1989). It is fair

to say that these have been met with scepticism by

the majority of historical linguists. But this has

brought more directly to the fore the question as to

how one should evaluate such proposals. How can

one know?

Of course there is nothing inherently unfamil-

iar in the notion of making verifiable statements about

events in the past which can themselves obviously

no longer directly be observed. Historical linguistics

and prehistoric archaeology are not alone in dealing

with processes and events which are now not di-

rectly observable owing to the passage of time. Such

is the case, of course, with palaeontology and with

cosmology. The relevant deciding test, it is widely

felt, is that which applies to all statements in the

domain of the sciences, namely the Popperian crite-

rion of testability or falsifiability (Bell 1994). But to

say this certainly does not solve the problem, when

it is not clear by what means a particular proposition

may be subjected to testing.

In this field it seems at first sight difficult to

formulate generally applicable criteria (McMahon &

McMahon 1995; Sims-Williams 1998). Certainly there

are cases where a probabilistic approach may be

employed, and where it may validly be asked

whether claims of significant patterning used to sup-

port a hypothesis in fact amount to anything more

than configurations which might equally have arisen

by chance (e.g. Ringe 1996; 1999). But this approach

is more difficult to apply than might at first appear

to be the case. For experience shows that in the de-

velopment of any discipline, new configurations be-

gin to emerge which are not at first clear, and which

are sometimes at first contradicted by counter-exam-

ples. Often it is not easy to formulate the precise

proposition which it is desired to test. One wonders,

for example, whether the statistical tests proposed

by Ringe (1996) would admit Albanian, Armenian,

Tocharian and Sinhalese to the same language fam-

ily, while few linguists today would doubt that all

belong to the Indo-European family.

At the Edge of Knowability

Sometimes theoretical criteria do not well sup-

ply the answer, and it may transpire that the proof of

the pudding is in the eating. The criteria may then

need subtly to be modified. The task of the Prehis-

tory of Languages project has therefore been to

address a number of current issues, seeking the sup-

port of a body of scholars with a wide range of

opinions, and to see what emerges. The wider aim,

however, must be to establish a number of general

principles relating language change to demographic,

social and economic change. If such principles could

be established and supported by a number of well-

understood cases, they might be of great value in

elucidating other cases which are not yet clear. It is

therefore not simply the falsification (or validation)

procedure which counts: it is also the undertaking of

proposing regularities and hypotheses which are

open to further investigation. Recent developments

in historical linguistics offer hope that this process is

now under way.

tures (such as the presence of head/dependent mark-

ing, dominant alignment, complexity, word order,

voice, etc.) and evaluates the frequencies of such

features in broad geographic areas, claiming to draw

therefrom conclusions about language dispersals go-

ing back to a very much greater time depth, as far as

30,000 years ago. For non-linguists, such as myself,

her work presents difficulties — for the non-initiate

such concepts as ergativity (see Dixon 1994) are tax-

ing — and it is perhaps for that reason that her work

has not been more widely discussed. There is no

doubt, however, that on occasion she makes sweep-

ing assumptions which are certainly open to ques-

tion (e.g. Nichols 1997). For this reason the sustained

criticism by Nettle (1999b) of her views on the date

of the first colonization of the Americas is of particu-

lar interest. However, while some of the historical

assumptions may be dubious, and her historical con-

clusions perhaps should be treated with caution un-

til there has been a fuller evaluation, there is no

doubt that her pioneering approach of population

typology is one of the most interesting developments

of recent years, and one whose further (and perhaps

more cautious) application is likely to bring very

fruitful results.

From the standpoint of the archaeologist, how-

ever, the most promising development in historical

linguistics in recent years is a new willingness among

some linguists to relate linguistic change to changes

taking place also in the material world (the archae-

ologist is almost tempted to say the ‘real world’),

including economic, social and demographic changes.

For too long, in my view, there has been the assump-

tion in some quarters that language changes by its

own rhythms and according to its own rules, quite

independently of demographic or social factors. Sev-

eral recent works have decisively broken that tradi-

tional mould. Among them is Christopher Ehret’s

An African Classical Age (Ehret 1998) where the lin-

guistic history of Africa is related to the social and

economic transformations which have taken place

there over the past ten thousand years or so (al-

though it should be noted that his view of the ori-

gins of the Afroasiatic language family differs

significantly from that recently advanced by Dia-

konoff (1998)).

Perhaps yet more significant, because framed

in more general terms, are two refreshing works

which will perhaps open a new chapter in historical

linguistics, and in particular in the understanding of

linguistic change in relation to demographic and so-

cial change, and hence potentially to the archaeo-

logical record. The first of these is Daniel Nettle’s

Recent developments in historical linguistics

The enhanced level of self-awareness in the field of

historical linguistics is well documented by the spate

of recent overviews or textbooks to complement such

earlier works as those of Hock (1986) and Anttila

(1989), including such useful texts as those of

McMahon (1994), Fox (1995), Lass (1997) and Camp-

bell (1998) and the valuable collected work edited by

Durie & Ross (1996).

More controversial, but of particular interest to

the archaeologist because of the very great time

depths which she contemplates, is Linguistic Diver-

sity in Space and Time by Johanna Nichols (1992).

Here she does not dispute the widely-held view

among historical linguists that genetic (i.e. familial)

relationships cannot, employing the standard com-

parative method, be observed between languages

which became separated more than about 8000 years

ago. As she puts it (Nichols 1992, 313):

Standard comparative method, then, requires

accidence of grammar and such things as shared

arbitrary lexical categorization as evidence for an

initial assumption of relatedness . . . it reconstructs

not only a sound system and a lexicon but also a

grammatical system which is an individual and

not a type. The diagnostic kinds of accidence ap-

parently persist in genetic groups only for a few

millennia, and dissipate entirely after about 8000

years.

Instead she advocates a population typology ap-

proach using what she terms stable structural fea-

Indo-European_Renfrew - Prehistory Of Languages Xpm.pdf

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: