Accounting for the pattern of Amerindian Accounting for the pattern of Amerindian languageslanguages

Roger Blench

Kay Williamson Educational Foundation

Lyon, France

May 11-14, 2008

OMLL Workshop: New Directions in Historical Linguistics

The settlement of the The settlement of the AmericasAmericas

The settlement of the Americas continues to be a major puzzle to students of prehistory.

To linguists (and increasingly geneticists), the extreme diversity of languages looks as if an extremely old date must be assigned to this, something on a par with Australia or Melanesia.

But archaeology is stubbornly resistant to such retrodiction. Clovis dates (ca. 12,500 BP) are still accepted as the main date for the settlement of the Americas, and even where the Clovis primacy is rejected, ‘Palaeo-Indians’ are still deemed to be of similar date.

The consensus of the linguists who have looked at the classification of Amerindian languages is that by and large they fall into a pattern of isolates and small phyla.

The settlement of the Americas

Even so, there is much to be explained

why are isolates so numerous in comparison with all other continents?

Why are Amerindian languages so phonologically and syntactically diverse (in contrast to Papuan and Australian for example)?

and why are there no very large phyla?

The situation might be explained in a number of ways;

The conservative archaeological dates are correct. Amerindian languages have diversified more rapidly than any other comparable region of the world and produced a highly atypical result Amerindian languages have been faultily classified and fall into a restricted number of large phyla compatible with these dates (a view is associated with the classification of Joseph Greenberg, 1987) Some early archaeological dates are indeed correct and the settlement of the Americas is significantly older than current models allow.

The default hypothesisI believe we have to assume that, twenty years after Greenberg, that his model has not been validated

No major specialist linguist has come out in support of it, despite abundant new material, although it regularly features in archaeological publications

From this we have to assume that the high diversity model is correct and that furthermore, the New World is very distinctive in terms of the type of diversity

If so, then the model of settlement must also be wrong

Amerindian language families in the pre-Columbian era(isolates not shown)

Amerindian languages divide into four categories

1. Isolates. Many languages in ones or twos with no evident relatives.2. Small phyla.3. Large, widely extended families with members scattered over a large area, often close to extinction and even today, often very small populations.4. Large, numerous and territorially broad groups, all of whose members seem to have practised agriculture and would be good candidates for agricultural expansions.

A note on uncertainties in the A note on uncertainties in the datadata

The classification of Amerindian languages is full of uncertainties, and even the major sources disagree with one another quite radically Few specialists agree with the Ethnologue, which is an extreme ‘splitter’ Families such as Hokan are highly controversial, accepted by some specialists and rejected by others. They do appear to be ‘typological’ (= ‘crime’) Greenberg has come in for especial venom Part of this is a data problem; materials are often fragmentary It could be a ‘splitter’ tradition in contrast to African ‘lumpers’. But a sample examination of Colombian isolates convinced me that this is not the problem

Large, geographically dispersed phyla in the New World

AlgicAlgic 4040 US, CanadaUS, Canada

CaribCarib 2929 Caribbean to BrazilCaribbean to Brazil

Eskimo-AleutEskimo-Aleut 1111 Russia, US, Canada, Russia, US, Canada, GreenlandGreenland

HokanHokan 2828 US, MexicoUS, Mexico

IroquoianIroquoian 1010 US, CanadaUS, Canada

Macro-GeMacro-Ge 3232 BrazilBrazil

Mataco-GuaicuruMataco-Guaicuru 1111 Brazil, ParaguayBrazil, Paraguay

Na-Dene Na-Dene 4747 Canada, USCanada, US

PanoanPanoan 3030 Brazil, PeruBrazil, Peru

PenutianPenutian 3131 US, CanadaUS, Canada

SalishanSalishan 2727 US, CanadaUS, Canada

SiouanSiouan 1717 US, CanadaUS, Canada TupianTupian 7676 Brazil, Bolivia, Paraguay Brazil, Bolivia, Paraguay

Large, geographically coherent phyla in the New World

NameName No.No. LocationLocationArawakan = Arawakan = Maipuran Maipuran 6060 Caribbean to Brazil Caribbean to Brazil

Aymaran Aymaran 33 Bolivia Bolivia Caddoan Caddoan 55 US US Chibchan Chibchan 2222 Colombia to Honduras Colombia to Honduras Guahiban Guahiban 55 Colombia Colombia Mayan Mayan 3030 Mexico, Guatemala Mexico, Guatemala Mixe-Zoque Mixe-Zoque 1616 Mexico Mexico Oto-Manguean Oto-Manguean 172172 Mexico, Nicaragua Mexico, Nicaragua Quechuan Quechuan 4646 Peru, Bolivia, Andes Peru, Bolivia, Andes

Tucanoan Tucanoan 2525 Brazil, Ecuador, Brazil, Ecuador, Colombia Colombia

Uto-Aztecan Uto-Aztecan 6262 US, Mexico US, Mexico Witotoan Witotoan 66 Colombia, Peru Colombia, Peru

Eskimo-AleutEskimo-Aleut

Na-Dene

AlgicAlgic

Siouan

Uto-AztecanUto-Aztecan

CaribanCariban

Tu

pian

Incan

Arau

canian

Isolates, small phyla by Isolates, small phyla by continent continent

ContinentContinent No. IsolatesNo. Isolates No. Small No. Small PhylaPhyla

Total Total languageslanguages

AfricaAfrica 66 00 20922092

EurasiaEurasia 66 11 25082508

PacificPacific 1212 44 10791079

AustraliaAustralia 77 1313 263263

New WorldNew World 2020 3636 10021002

The archaeological recordThe archaeological recordArchaeological models of the settlement of the Americas have been dominated by disputes over dating. For a very extended period, Clovis points were held by North American archaeologists to be the earliest evidence for human occupation and these seem to be no earlier than 12,500 BP. In contrast, throughout South-Central America, much earlier dates are part of public discourse, with 30,000 BP commonly featuring in maps of the settlement of the region. The consequence was that any site which appeared to be older was routinely subjected to intensive skepticism, and of course no procedure can be perfect.

The archaeological recordThe archaeological recordThe dating of sites such as Meadowcroft (19,000 BP), Cactus Hill (15,000 BP) and Bluefish Caves (14,000 BP) is commonly questioned. Direct dating of coprolites at 5-Mile-Point caves in Oregon has recently given a date of 12,300 BP (Gilbert et al. 2008). Similar, very early, unfluted lanceolate points have also been found in South America. Lanceolate El Jobo-like points have been recovered at the Monte Verde site, Chile. The Pre-Clovis occupation at Monte Verde has been dated to at least 12,500 BP. (Dillehay 1997; Meltzer 1997). Even fairly sceptical authors such as Roosevelt et al. (2002) admit to earlier dates for Alaska.

‘‘West Coast diversity’West Coast diversity’ The observation that the linguistic diversity of the Americas was somewhat lopsided and that the greatest numbers of languages are found on the west side of the continent goes back to Barton (1797).

Gruhn (1988, 1997) has been a strong proponent of West Coast diversity and its archaeological correlates and the map reproduces her somewhat outdated maps of language isolates, which illustrates the point.

A similar conclusion can be drawn from the maps accompanying Adelaar and Muysken (2004) where the dense language situation in pre-conquest northwest South American is plotted out.

Whatever the explanation, the skewed linguistic geography has struck many authors and it has been related to models of settlement.

Ruth Gruhn’s map of isolates and small phyla

Round the southern edge of the ice-sheetsRound the southern edge of the ice-sheets

The model to explain all this is to assume that Palaeo-Siberian hunters had access to boats of some type as early as 30,000 kya. [which we know is true for early humans in the Pacific]

Siberian languages are today very diverse and were presumably so in the past. If there was a continuous flow of populations rather than a single impulse when the Bering strait land bridge was ‘open’

This population would then have flowed down the west coast, exploiting aquatic resources.

Stone tools from Eel Point, San Clemente, California, ca. 9-8000 BP are similar to those used in historic time for boatbuilding

People had settled San Nicolas island, about 60 miles from the nearest landfall, between 8000 to 8500 years ago. Clearly people were getting around in some kind of watercraft.

Can genetics help us? Genetic analyses of the Can genetics help us? Genetic analyses of the peopling of the New Worldpeopling of the New World

AuthorsAuthors DateDate Time-frameTime-frame MigrationMigration

Torroni et al. Torroni et al. 1992,1991992,19944

not givennot given FourFour

Shields et al.Shields et al. 19931993 > 12 KyBP> 12 KyBP MultipleMultiple

Bonatto and Bonatto and SalzanoSalzano

19971997 ~30-40 KyBP~30-40 KyBP SingleSingle

Stone & StonekingStone & Stoneking 19981998 23,000–37,000 23,000–37,000 BPBP

SingleSingle

Starikovskaya et al.Starikovskaya et al. 19981998 ~34,000 BP~34,000 BP TwoTwo

Karafet et al.Karafet et al. 19991999 not givennot given TwoTwo

Ruiz-Linares et al.Ruiz-Linares et al. 19991999 9,334–11,456 9,334–11,456 BPBP

SingleSingle

Bortolini et al.Bortolini et al. 20002000 ~14KyBP~14KyBP TwoTwo

Lell et al.Lell et al. 20022002 not givennot given TwoTwo

Silva et al.Silva et al. 20022002 ~21 KyBP~21 KyBP SingleSingle

Fuselli et al.Fuselli et al. 20032003 > 13 KyBP> 13 KyBP Single [?]Single [?]

Seielstad et al.Seielstad et al. 20032003 <18 KyBP<18 KyBP not givennot given

Nelson et al. Nelson et al. 20082008 ~23-19 KyBP~23-19 KyBP SingleSingle

Hunters begin to walk and paddle across from Siberia ~30,000 BP. They people the Americas at extremely low population densities and probably diffuse initially down the West Coast (now largely under water and inaccessible to archaeology).

A wide variety of already diverse language groups and physical types continue to cross Beringia, paddling south of the ice when the land bridge is ‘closed’. Low population densities accelerate language differentiation processes.

There are local expansions of hunting-gathering groups, driven principally by minor technological changes perhaps by flow across the Bering Strait, esp. from 12,000 onwards.

Reconstructing a hypothetical demographic Reconstructing a hypothetical demographic history of the New World Ihistory of the New World I

Incipient intensive plant management begins in scattered locales by 10,000 BP for a variety of purposes, including food, but does not initiate major socio-economic change.

5. By 6-5000 BP the domestication of key starch staples causes certain groups to expand significantly and many small groups are assimilated.

6. Possible transoceanic contacts with both the populations of mainland SE Asia and the Austronesians extend cultural and linguistic diversity

Solutrean parallels remain controversial and for chronological reasons I doubt they are relevant

Reconstructing a hypothetical demographic Reconstructing a hypothetical demographic history of the New World IIhistory of the New World II

The pattern of languages in the New World is a consequence of two main factors; a long time-span to allow language differentiation to develop and the continuing arrival of new language groups from an already highly diverse region, Siberia.

Low population densities allowed language barriers to remain and the absence of very large polities meant that language levelling remained an insignificant factor.

Agriculture or ‘intensive plant management’ developed early, but focused on species that made little distinctive change to subsistence strategies.

Only later did cereal and tuber staples make a significant contribution to diet, allowing the spread of small to medium language phyla (Mayan? Otomanguean?).

Hence the pattern in the immediate pre-Columbian era.

Reconstructing a hypothetical demographic Reconstructing a hypothetical demographic history of the New World IIIhistory of the New World III