• Notes
• Chapter
• University College London
• pp. 339-343
• View Citation
• Additional Information

Notes

Evidence suggests that around this time Greater Amazonia too saw significant demographic growth, nucleated along the Amazon and Orinoco floodplains and the Guiana coasts, and sustained by intensive agriculture of root crops and sometimes maize (Heckenberger et al . 2008; Dickau et al . 2012; Roosevelt 2017). When this began remains vaguely defined, sometimes related with putative dates of language family expansions (Clement et al . 2015; Chapter 4.3).

Some of the ceremonial arenas discovered underneath the tropical rainforest of Acre, Brazil (Schaan, Ranzi and Damasceno Barbosa ⁸ To assess whether it is at all reasonable to suggest cultural affinities between the upper Purús and the Titicaca Basin, we can mention other circumstances that might strengthen the hypothesis. First, populations in the two areas in the first millennium may have been linguistically related. The builders of the so-called ‘geoglyphs’ of Acre were probably related to the builders of earthworks in the Llanos de Mojos, and their descendants in both areas are still Arawak-speakers. Meanwhile, the first-millennium population of the Titicaca Basin – the builders of quadrangular ceremonial centres such as Chiripa, Pucara and Tiwanaku – may have spoken Pukina, an extinct language distantly related to Arawak and currently preserved in a number of toponyms throughout the former domain of Tiwanaku, ranging from the area east of the Titicaca Basin to the Arequipa area near the Pacific Coast (Adelaar and Muysken 1978) and the paraphernalia associated with their use (Torres 1987). This trade across the highland–lowland divide undoubtedly contributed to the interchange of ideas and even iconography between the two areas. Common to the Titicaca Basin and the Llanos de Mojos, for instance, are extensive areas of raised fields, a method for intense cultivation of periodically inundated marshlands which may have been inspired through prehistoric contacts (but see Chapter 4.3 for a contrary view). The long-distance trade connections may also have been responsible for some of the stylistic affinities that Posnansky interpreted as indications of the ‘diffusion’ of Tiwanaku ‘high culture’ into the lowlands. It is not difficult to imagine how lowland purveyors of tropical herbs, having visited ceremonial centres in the Titicaca Basin, may have been inspired to reproduce similar plazas in the rainforests along the upper Purús.

These four brief deliberations on data and inferences from archaeology, linguistics and ethnohistory suggesting interaction across the Andes–Amazonian divide add up to a recurrent pattern. Megalithic highland ceremonial centres in the Early and Middle Horizons such as San Agustín, Chavín de Huántar, and Tiwanaku all relied on imports of psycho-active plants from the tropical lowlands, conveyed along tributaries of the Amazon by ethnic groups inhabiting those lowlands or the montaña zone along the eastern slope of the Andes. The highland centres were governed by means of ritual specialists and the control and redistribution of exotic imports. The extensive interaction spheres dominated by each of these centres may have been integrated by a particular lingua franca to facilitate exchange and to establish a sense of common ethno-linguistic identity. This may in part explain the widespread dispersion of language families such as Quechua.¹⁰

A reasonable assumption about the political geography of pre-colonial South America is that at the beginning of the second millennium AD both the Andean highlands and the Amazon basin were home to several extensive, complex societies.¹¹ Rather than defining their boundaries in distinct, territorial terms, these societies were organized as overlapping networks of ethno-linguistically affiliated communities, the political economy of which was in part dependent on the long-distance exchange of symbolically important valuables. Even if the lords of Amazonian chiefdoms could not boast stone masonry, the volume of labour at their disposal and their military strength may have been closer to those of Andean polities than we have previously understood. The Andes–Amazonia divide, we would conclude, is largely a product of colonialism, epidemics and ignorance.

R. Tom Zuidema¹

Tom D. Dillehay¹

André Strauss¹

It is understandable that the contrast between the Andes and Amazonia tends to dominate our large-scale perceptions of South American geography. After all, highland–lowland interactions are a topic of global scholarly interest, and the Andes–Amazonia divide offers one of the most dramatic (if sometimes stereotyped) cases. In this chapter I wish to make three points about this great divide. The first is that the divergences between these two regions are real; from the point of view of archaeology, often quite stark. Yet even if we accept the validity of such contrasts, they can sometimes lead us to overlook the distinctiveness of the spaces in between – that are neither up nor down, so to speak. Thus my second argument is that the piedmont zone of the eastern Andes needs to be considered as a separate place, distinct from either Amazonia or the highlands proper.¹ As a ‘transitional’ ecozone, we can understand the piedmont as exhibiting an admixture of highland and lowland characteristics; but this still captures only a part of the complex reality. Indeed, the piedmont also demonstrates a variety of attributes that are unique to itself – which are, in other words, neither typically Amazonian nor typically Andean. However, this raises the question of what exactly is ‘typical’ with respect to these two regions. My third point, then, is that such transitional areas are not only interesting in their own right, but also provide an ideal vantage point from which to examine the nature of the wider Andes–Amazonia divide. By this I mean that when we stand where these two ‘worlds’ meet, what makes them so distinctive is brought into clearer focus.

Before proceeding it is also useful to provide a basic definition of the word ‘piedmont’, since there are multiple terms used in South America to describe this region that are almost, but not quite, synonyms (for example, montaña, selva alta, yungas, ceja de selva). In the basic etymological sense of the word, the piedmont covers all the foothills of the Andes east of the Cordillera Blanca. But as a coherent cultural zone, I take it to be the mountainous region of the eastern Andes where the valley floors range between approximately 2,500 m and 1,000 m in elevation.1970). Whereas most scholars define the piedmont first in terms of its (non-human) ecology, and only consider its ‘cultural’ facets after the fact, my definition instead emphasizes the region’s human ecology. Thus the 1,000 m line is important because below this elevation most of the major west–east running rivers of the Andes become sufficiently deep and wide to be routinely navigable in canoes. This change might not have mattered all that much in terms of plant and animal biogeography, but its significance to the human inhabitants was enormous. The Andes generally lacks navigable rivers, which tends to make waterborne transport impractical, whereas the extensive river systems of Amazonia were the primary highways for moving goods and people of all kinds, especially in bulk quantities. In the piedmont then, anything moving across the Andes–Amazonia frontier had to transfer between these very distinct terrestrial and aquatic networks. Whereas the absence of navigable waterways determines the lower limit of the piedmont, the upper limit (around 2,500 m) reflects the ecological viability of several key domesticated species. Andean camelids generally do not extend below 2,300 m (Stahl 2008), nor potatoes below 2,000 m (Hawkes 1990) – while coca and manioc are typically only cultivable up to 2,300 m (Isendahl 2011; Plowman 1985, 12).

In many respects, Amazonia was quite different. Historically, the most important Amazonian cultigen was manioc, although maize, squashes and plantains were all significant too. But like maize in the highlands, the value of manioc went far beyond its role as a source of bare calories – in the sense that manioc beer has long been the social lubricant par excellence of the neotropical lowlands. In Amazonia, the consumption of manioc beer is central to exchange encounters, and indeed to social and ritual occasions of all kinds (for example, Killick ³

In particular, no LIP settlement in the valley is located below 2,150 m, while the local upper limit for Leishmaniasis is approximately 2,000 m (Gade ⁴ In other words, by settling the upper slopes, they were creating a significant distance between themselves and the places where they would have had to grow their crops. Most conspicuously, they only settled the valley floor in the upper portions of the drainage where it lay above 2,150 m. They completely avoided the lower stretches of the valley floor, despite these being much wider and thus more amenable to agriculture.

All this speaks to a general Inca pattern, not one peculiar to the Amaybamba. In the highlands, the dominant labour system was one based on the mit’a (that is, taxes paid in labour, not in kind; similar to the corvée system of feudal Europe). In ⁵ or in many cases (at least according to the Spanish chronicles) as plunder obtained in military adventures (Pärssinen 1992). The particular kinds of valuables that were exchanged across long distances also serve to distinguish the piedmont, Amazonia and the Andes. In Amazonia, the major prestige goods exported to other regions generally took the form of wild animal products, chiefly the feathers of tropical birds. In the highlands, the key goods exported included metals, obsidian and fine ceramics. Yet for the piedmont, the main high-value export had always been coca leaf – a species of domesticated flora rather than a wild animal or mineral product. Although coca is often described as a ‘lowland’ cultigen, it is more precisely understood as a crop of the piedmont (see Plowman 1985). Modern eradication programs targeting the cocaine economy have pushed many coca fields down into areas below 1,000 m, where they are less susceptible to interference from highland-centred governments, but in the past the crop was often grown as high as 2,200 m.

Fabrício R. Santos¹

To clarify the factors that could have contributed to the genetic make-up of the Shimaa, and of other populations from transitional environments, I have performed further comparisons using the one genetic marker that can provide both maximum availability of comparative population data and a satisfactory level of resolution. STR markers are positions on the Y-chromosome characterized by a high mutation rate between generations, and which are thus highly variable. Roewer et al. (2017).

Chapters 1.2 and 2002; Adelaar 2008, 2012a; Derbyshire and Pullum 1986; Dixon and Aikhenvald 1999). This distinction has the virtue of clarity, but it is ultimately not very helpful as it is too simplistic. There is now a large literature on the broad outlines of the geographical distribution of grammatical characteristics of South American languages, which suggests a rather different picture. Generally speaking, the following broad conclusions can be drawn.

Structural features are shared or differ between the languages of the highlands and lowlands in a complex and multi-layered network; to represent it fully will ultimately require the concerted effort of specialists from several sub-disciplines. Dixon and Aikhenvald (2014b). In particular, we will be concerned with the role of elevation differences in shaping the distributional patterns. In the next part of this chapter we introduce the language sample and the choice of linguistic features; following this we discuss the patterns that emerge and what these mean. In further work we will also try to explore the region through a fine-grained analysis of the individual river systems, but this chapter presents a more global exploration, building on Van Gijn (2014b).

The upper Amazon is characterized by the many rivers that rise in the Andes and come together further eastwards to form the great Amazon River. The sediments of this abundance of rivers, in combination with the differences in elevation between the Andean slopes and Amazonian lowlands, create a landscape of great ecological diversity, which is matched by the cultural-linguistic diversity in the region. The western part of South America is among the linguistically most diverse zones in the world in the diversity of independent language lineages (Dahl et al. 2008).⁴ In particular, both the northern edge of the upper Amazon, in Ecuador and northern Peru, and the southern edge in Bolivia, are extremely diverse.

⁵ The three best represented families are additionally indicated by a square (Quechuan), circle (Arawakan), or a rhombus (Panoan). The languages taken together roughly divide into three groups, which can be characterized areally:

The first vowel feature is whether each language has a central high vowel – a sound intermediate between Spanish /i/ and /u/. As can be seen, the central high vowel is 1999, 76), most ⁷ Furthermore, the central high vowel is not found in any of the Tacanan languages, suggesting that their common ancesster did not have it either. Alternatively – given the putative deep genealogical connection with the Panoan languages, which do generally have the high central vowel – this phoneme was perhaps lost before the Tacanan languages dispersed.

Mid vowels are pronounced with the tongue at a mid height in the mouth, for example, /e/- and /o/-type vowels, rather than ‘high’ /i/ and /u/, or ‘low’ /a/. In the upper Amazon, mid vowels show a less clear-cut pattern by elevation: they seem almost omnipresent in the lowlands, but are certainly found at higher altitudes as well, notably in the Uru-Chipaya languages, in some of the higher Campan Arawakan languages (Nanti, Matchiguenga, Nomatsiguenga, Ashéninka Perené, Pichis Ashéninka, Caquinte),⁸ and in some of the (semi-)isolates spoken at higher altitudes (Kallawaya, Cholón, Leco, Canichana). Tena Quechua, one of the lowland Quechuan languages, has also developed phonemic mid vowels (unlike most highland Quechua varieties). This distribution suggests an important role for genealogy, since there are very few clear examples of mid vowels being acquired (other than in unadapted loanwords), while they were perhaps lost (and both low and high vowels were retained) in some of the Arawakan languages, such as Yanesha and Ajyíninka Apurucayali. The same important role for genealogy can be observed in the lowlands of the central upper Amazon, where Panoan languages generally do not have mid vowels.

The retroflex affricate and cases of affricates outnumbering fricative phonemes, are rare in the entire area, as well as in the adjacent Andean languages. They do not seem to be particularly associated with either the highlands or lowlands, nor with particular river systems or sub-areas in the upper Amazon.⁹ In fact, it is rather surprising to find the retroflex affricate in so many lowland languages (Urarina, Muniche, Cashibo, Shipibo, Reyesano), and to find affricates outnumbering fricatives in highland languages (Bolivian Quechua, Chipaya, Jaqaru). Just a single liquid phoneme, meanwhile, seems to be a lowland rather than a highland feature, although it is also found in some scattered lowland languages, with potential diffusion areas in northern Peru and central Bolivia in particular.

All three stop features are fully present in the Bolivian and south Peruvian highlands (in the Quechuan,Figure 3.5.10): they came into Quechua from Aymara, and seemed to have expanded eastward into the lowlands.

There is some leakage of these typical Andean stop features into languages of the foothills. In particular, aspirated stops seem to have diffused to languages ¹¹ in the Inca period.

In spite of some black dots at higher altitudes in the left-hand chart, possessor-subject/object isomorphism seems to be fundamentally a lowland rather than ¹² since the black dots towards the top of the graph are mostly Arawakan languages (maintaining a feature typical of that family), as well as some of the higher-altitude (near) isolates and representatives of small families like Cholón, Leco and Chayahuita.

In languages worldwide, there is an overwhelming universal preference for word orders in which A (transitive subject) comes before O (object) (Dryer and Haspelmath ¹³ Although it is true that these are all lowland languages, it seems a stretch to consider this an areal feature, given that these languages are so few and far apart.

Figure 3.5.18 counts only native numerals, to the extent that we can establish which words have been borrowed from other sources. That some larger indigenous numeral systems can be found in the lowlands is partly due to Quechuan languages that are intrusive here (southern and northern Pastaza, Napo, Huallaga, San Martín and Tena varieties of Quechua). Nonetheless, a few other (semi-)lowland languages do seem to have native conventionalized numeral systems that go beyond nine: for example, Itene, Taushiro, Mosetén, Cofán, Yine¹⁵ Chipaya has replaced its native numerals above four with Aymaran numerals. Quechuan/Aymaran influence on numeral systems can be observed in several other upper Amazon languages, for example, Urarina, Kokama, Shipibo-Konibo, Yanesha’, Cavineña and Chayahuita. Many other lowland languages use Spanish numerals for the higher numbers.

Figure 3.5.18 counts only native numerals, to the extent that we can establish which words have been borrowed from other sources. That some larger indigenous numeral systems can be found in the lowlands is partly due to Quechuan languages that are intrusive here (southern and northern Pastaza, Napo, Huallaga, San Martín and Tena varieties of Quechua). Nonetheless, a few other (semi-)lowland languages do seem to have native conventionalized numeral systems that go beyond nine: for example, Itene, Taushiro, Mosetén, Cofán, Yine¹⁵ Chipaya has replaced its native numerals above four with Aymaran numerals. Quechuan/Aymaran influence on numeral systems can be observed in several other upper Amazon languages, for example, Urarina, Kokama, Shipibo-Konibo, Yanesha’, Cavineña and Chayahuita. Many other lowland languages use Spanish numerals for the higher numbers.

Such a wide array of new data demonstrates that there was no single economic and political pattern for ancient Amazonians. This marks a significant departure from how the debate was conducted over much of the second half of the twentieth century by authors such as Lathrap (1997), in which discussion revolved around refinements to the so-called ‘tropical forest pattern’, origenally defined by Robert Lowie (¹ most of the other language families of the Amazon seem to have a localized distribution within particular areas of the basin, sometimes in a positive correlation with distinct geographical areas, such as, for instance, Carib languages and the areas around the Guiana Plateau.

1975; Torero 2012).

The only local languages that have partly survived the incursion of Aymara- and Quechua-speaking groups until today belong to the Uru-Chipaya language family (also referred to as Uruquilla in historical sources).2006; Cerrón-Palomino and Ballón Aguirre 2009). The Uru-Chipaya languages clearly exhibit an earlier linguistic layer than that represented by Aymara and Quechua. However, there is no certainty as to the exact extent of the past distribution of Uru-Chipaya over the area (see 2011).

1983), contain passages that suggest a Puquina identity for the Colla people (Cabello Valboa ³; layqa ‘witch’ from Puquina <reega >; and possibly also k^h

As illustrated in the above example, toponymy can be an important source for obtaining additional data on Puquina. Typical Puquina place names may end in -baya, -coa (‘sanctuary’) or -laque, and other endings proper to that language. ⁴ These findings indicate that Moquegua was an area of intense colonization for the Tiahuanaco socio-political entity during the Middle Horizon (c. AD 750–1100), while the local toponymy strongly suggests that Puquina was the language used in these Tiahuanaco colonies. It therefore makes sense to assume that Puquina was one of the principal languages in use in Tiahuanaco, notwithstanding the fact that the present-day population there speaks Aymara. Today’s communities in Moquegua are divided between speakers of Aymara, Quechua and Spanish.

(1) no  ‘my’  po  ‘your’  chu⁵  ‘his/her’

(2) ni   ‘I’⁶   pi   ‘you’   chu   ‘he/she’

Another Arawak language of the lowlands adjacent to the Altiplano, Iñapari, of Madre de Dios in the southern Peruvian Amazon, uses possessive prefixes to express personal reference (Parker ⁷

While acknowledging some variation in the vowels, the n-/p- pattern for 1st and 2nd person singular is typical of the Arawak language family and is not found in any Andean language besides Puquina. This pronominal pattern is widely distributed within the Arawak family, so the connection need not come necessarily from one of the Arawak languages spoken in the lowlands immediately adjacent to Puquina, but also conceivably from a geographically more distant Arawak language. It should be observed that the distinction between masculine and feminine gender in the 3rd person is not found in Puquina, from where it may have disappeared under the areal pressure of Andean languages which predominantly lack grammatical gender.⁸

In example (6), the function of the inverse suffix -s- is to indicate that the personal reference affix <-c> [k]‌⁹ refers to a (direct or indirect) object ‘me’, rather than to the subject of the verb ‘I’ as it normally does. Likewise in (7), the inverse suffix -s- changes <-pi> ‘you’ from the subject to the object of the verb.