Theorizing ethnolinguistic diversity under globalization: beyond biocultural analogies

ABSTRACT This paper seeks to advance our theoretical understanding of diversifying and homogenizing processes in human societies by exploring the sources of and threats to ethnolinguistic or cultural diversity. Invoking concepts such as ethnogenesis, schismogenesis, and structural transformations, it discusses the parallels as well as the divergences between biological and cultural theory. Models in historical linguistics risk importing unwarranted assumptions about diversification from biological models of speciation. A more pertinent theorization of ethnolinguistic diversity might build on anthropological perspectives such as those of Fredrik Barth, Gregory Bateson, and Claude Lévi-Strauss, all of whom recognized that such diversity reflects interaction rather than isolation. The empirical test of these considerations is the ethnography and history of language use among Indigenous peoples in Amazonia and the Andes. The conclusion is that premodern expansions of language families, in tolerating local cultural autonomy, multilingualism, and diglossia, did not threaten ethnolinguistic identities as has modern globalization.


Introduction
The aim of this paper is to examine the frequently expressed observation that there are analogies between biological and cultural diversity.It seeks to advance our theoretical understanding of diversifying and homogenizing processes in human societies by exploring similarities and differences between the sources of and threats to these two kinds of diversity.Drawing on concepts such as ethnogenesis, schismogenesis, and structural transformations, it discusses the parallels as well as the divergences between biological and cultural theory.The empirical test of these considerations that has been chosen here is the linguistic history of Indigenous societies of South America.A central argument of the paper is that models in historical linguistics risk importing unwarranted assumptions about diversification from biological models of speciation.It will be argued that a more pertinent theorization of ethnolinguistic diversity might build on anthropological perspectives such as those of Fredrik Barth, Gregory Bateson, and Claude Lévi-Strauss, all of whom recognized that such diversity reflects interaction rather than isolation.
It is widely recognized that biological diversity and cultural diversity are similarly being reduced by processes of globalization.A popular narrative of such processes over the past five centuries is Charles C. Mann's (2011) book 1493: Uncovering the New World Columbus Created.Mann vividly explains the homogenizing consequences of the post-Columbian trade in tobacco, silver, rubber, and sugar, the expanding adoption of crops such as potatoes and sweet potatoes, and the entangled histories of diseases, slavery, and the fugitive communities of enslaved Africans.In these processes, reductions of biological and cultural diversity tend to be intertwined.After providing numerous examples of ecological degradation and cultural marginalization that are hallmarks of what he calls the 'Homogenocene', Mann finally reflects on the ambiguities of globalization as deriving from human desires: [People] want to have what everyone else has, but still be aggressively themselvesa contradictory enterprise.… As human desires bring the Homogenocene into existence, billions of people marching through increasingly identical landscapes, that special place becomes ever harder to find.Things feel changed and scary.Some people hunker down into their local dialects or customary clothing or an imagined version of their own history or religion.Others enfold themselves in their homes and gardens.A few pick up weapons.(Mann, 2011, p. 507) Mann's reflections on the contradictions between globalized capitalism and struggles for cultural identity are offered together with the observation that the biodiversity of purportedly 'traditional' vegetable gardens in the Philippines is 'a polyglot, cosmopolitan, thoroughly contemporary artifact' (Mann, 2011, p. 292).As elsewhere, some invasive species have ravaged ecosystems and extinguished several local species, while other novel imports have become symbols of tradition, arguably enriching garden plots and even their capacity to instil a sense of identity.Such ambiguities of globalization are undeniable, but Mann's analysis seems to represent them as inevitable, never venturing beyond human incentives into political economy or other social theory.If we consider the Homogenocene inevitable, it is because we consider globalizing modernity inevitable.Although it is beyond the scope of this paper to elaborate visions of alternatives to globalizing modernity and its inclination to homogenize and reshuffle ecological and cultural systems, the paper argues that we must move beyond biocultural analogies to illuminate the economic, political, and other conditions that tend to promote or reduce cultural diversity.We must ultimately try to envisage a world in which globalization is not a threat to the diversity of cultural identities where the global integration of social interaction does not jeopardize the maintenance of cultural specificity.

Biocultural analogies
Twenty years ago, worries about simultaneously declining cultural and biological diversity were articulated in the UNESCO booklet Sharing a World of Difference (Skutnabb-Kangas et al., 2003).The authors emphasize interlinkages between linguistic, cultural, and biological diversity and how globalization tends to endanger them all through homogenization. 1 Out of the world's almost 7,000 languages, they report, some scholars predict that as many as 90 per cent may have disappeared by the end of this century.Their concept of 'biocultural diversity' (Maffi, 2001) refers not only to the parallel trajectories of biological and cultural diversity but to their common ecological significance: locally contextualized ecological knowledge, codified in endemic languages, is understood as having adaptive value similar to the biological adaptations of endemic species.In this view, traditional ecological knowledge codifies human adaptation to local conditions and simultaneously enhances the complexity of the local ecosystem. 2As Skutnabb-Kangas and Harmon (2017) explain, 'much of the knowledge of how to maintain biodiversity is encoded in the small languages of Indigenous and local peoples, and it disappears when the languages disappear.'Biocultural analogies at times evoke the anthropological school known as 'cultural ecology', which in its more simplistic versions aspired to account for human culture as a functional adaptation to the environment. 3In its more sophisticated versions, the approach might be referred to as 'contextualism' (Hornborg, 1996), as it emphasizes how local knowledge and worldviews tend to be calibrated with specific ecological contexts.Such observations are generally contrasted with the decontextualizing forces of globalization and modernity, as processes of modernization hinge on the fact that ideas, objects, and persons that are most easily moved from one context to another are favoured by the logic of modern cultural diffusion (Hornborg, 2023).Maffi (2005) shows that biocultural perspectives can be traced back to Charles Darwin and recurs among anthropologists such as Boas, Sapir, Whorf, and Kroeber.It was not until the mid-1990s, however, that there emerged 'a focus on the relationships between linguistic, cultural, and biological diversity, their global overlapping distributions, and the common threats they are facing' (Maffi, 2005, p. 601).The significant co-occurrence of linguistic and biological diversity has strengthened proposals about their functional entanglement.Citing a 2005 study by Jonathan Loh and David Harmon, Maffi observes that 'core regions of exceptionally high biocultural diversity' include the Amazon Basin, Central Africa, and Indomalaysia/Melanesia (Maffi, 2005, p. 611).Such correlations between linguistic and biological diversity have been verified in a global survey of 'biodiversity hotspots': 70 per cent of the world's languages are found within the 25 per cent of the Earth's land area that is classified as either biodiversity hotspots or 'high-biodiversity wilderness areas' (Gorenflo et al., 2012).
In their contribution to the Routledge Handbook of Ecolinguistics, Skutnabb-Kangas and Harmon (2017) provide a useful overview over the various parallels between biological and linguistic diversity.They begin by reviewing recent figures on ongoing processes of linguistic homogenization and language endangerment.Just ten (a minuscule fraction of one per cent) of almost 7,000 languages are spoken by around half of the world's population, while 88 per cent of the world's languages are spoken by fewer than one million speakers and around 21 per cent (1,537 languages) by fewer than a thousand speakers.Skutnabb-Kangas and Harmon observe that 'languages are today being killed at a much faster pace than ever before in human history.'It is well known that a parallel process is afflicting biological species: the rate at which vertebrate species have been lost over the last century is at least eightperhaps as much as 100times higher than the so-called background (i.e.natural) rate.Wilson (1992, p. 280) estimates the background extinction rate as 'about one species per one million species a year.' Biologists reckon that somewhere between 5,000 and 150,000 species become extinct every year.The great variation in these assessments depends on differences between various estimates of the total number of species, most of which are undocumented, but pessimistic predictions suggest that around 20 per cent of species may disappear by the end of this century.Given his high estimates of the number of undocumented species, Wilson (1992, p. 280) suggests that 'human activity has increased extinction between 1,000 and 10,000 times over [the background] level in the rain forest by reduction in area alone.' The analogies between the trajectories of biological and linguistic diversity have inspired extensive methodological and even conceptual transfers from biology to linguistics.As Skutnabb-Kangas and Harmon note, both forms of diversity are fundamentally evolutionary, with all living species and languages the result of descent with modification from, respectively, a common genetic or linguistic ancestor.The histories of species and languages are traced by similar taxonomic methods and result in similar phylogenetic trees that reflect an evolutionary branching process.Both kinds of diversity can be classified hierarchically, with the genes/species/ecosystem ladder of biodiversity matched by structural/language/lineage levels of linguistic diversity.(Skutnabb-Kangas & Harmon, 2017) These parallels seem incontrovertible, as is their conviction that 'linguicide' and 'ecocide' share 'the same root cause: neoliberal globalization.' Uniformity and homogenization, the authors conclude, appear to be requisite to the economic growth that is encouraged by 'dominant economic and political forces.' 4  Given the many parallels, however, it appears that not enough attention has been paid to the differences between how biological and cultural/linguistic diversity is generated.As we shall see, there is a sense in which the two processes are diametrical opposites.Although conceding that there are important differences, Skutnabb-Kangas and Harmon (2017) finally assert that 'the main driver of both speciation and language genesis is isolation (reproductive for species, communicative for languages),' and that 'the same biogeographic factors that give rise to species diversification are probably at least partly behind language diversification' (emphasis added).This is an assertion that calls for closer examination in the light of anthropological and linguistic theory, to which we shall turn in section 4. First, however, we shall briefly review the theoretical model of biological diversification synthesized by Edward O. Wilson in his classic study over 30 years ago, and the main reasons why biodiversity is currently being diminished.

Biodiversity in the Anthropocene
Supporting his explanations with myriad examples, Wilson (1992) presents a compelling account of the forces of evolution.'Reproductive isolation between breeding populations,' he writes, 'is the point of no return in the creation of biological diversity' (Wilson, 1992, p. 50).Speciation (that is, the appearance of new species) results from the emergence of some kind of genetic difference between parts of the original population that precludes interbreeding.Such differences generally originate as genetically determined traits that increase adaptive fitness in specific environments. 5 They thus reflect variations in environment, including varying opportunities to generate new ecological niches in co-evolution with other species.Through natural selection over countless generations, minute genetic differences are incrementally enhanced until they become unbreachable.'The two populations have turned into distinct species because they are reproductively isolated where they meet under natural conditions' (Wilson, 1992, p. 62).The genetic variation which is subject to selection is generated by random mutations.Where completely new environments are colonized, such as when a breeding population of organisms reaches an isolated island and adapts to several new environmental niches, the speed of adaptive radiation can be comparatively rapid.
In addition to accounting for biological diversification, Wilson also explains how human activity tends to reduce biodiversity through alteration or destruction of habitats, introduction of invasive species, and overharvesting.These processes of biological homogenization have more recently been illuminated by Elizabeth Kolbert (2014).She explains the vulnerability of tropical rain forest ecosystems by referring to the extreme specialization of their constituent species.Prompted by intense competition, specialization means that more organisms can coexist within the same space.High diversity of species means low population density of each species, which increases distances between organisms, chances of isolation, and further speciationand finally vulnerability, since isolated populations are more susceptible to extinction (Kolbert, 2014, p. 183).Kolbert succinctly captures the contradictory imperatives imposed on the biosphere by the Anthropocene.It compels species to move in response to rising temperatures, yet it creates 'barriersroads, clear-cuts, citiesthat prevent them from doing so' (p.189).Its intercontinental reshuffling of species is unprecedented in the history of life.It has made a 'hash' of geographic distribution as 'highways, clear-cuts, and soybean plantations create islands where none before existed,' while 'global trade and global travel do the reverse: they deny even the remotest islands their remoteness ' (p. 198).The reshuffling of species increases local diversity but reduces the number of endemics: 'For the same reasons that local diversity has, as a general rule, been increasing, global diversitythe total number of different species that can be found worldwidehas dropped' (p.212).
For many anthropologists, this observation is equally applicable to cultural and linguistic diversity, and for some, it is equally alarming.For others, to deplore cultural globalization as homogenizing is a conservative and morally reprehensible approach, evoking essentialism, boundaries, and ethnocentric parochialism. 6Rather than engage in this normative debate, however, in the next section we shall consider how some prominent twentieth-century anthropologists have theorized the social mechanisms of cultural diversification.

Theorizing cultural diversity
Several of the giants of twentieth-century anthropology, like Claude Lévi-Strauss and Gregory Bateson, were deeply inspired by natural science, particularly biology.In his biography of the former, Patrick Wilcken (2010) mentions the significance of Lévi-Strauss' encounter with D'Arcy Wentworth Thompson's (1961Thompson's ( [1917]]) classic On Growth and Form. 7D'Arcy Thompson had applied mathematics to describe morphological differences between species, such as the shapes of skeletal parts, animal horns, and shells.He showed how biological diversity can be understood, or at least mapped, in terms of quantifiable distortions of geometric coordinates.Wilcken (2010, pp. 141-142, 261) writes that D'Arcy Thompson's book 'fit like a key' into Lévi-Strauss' evolving thought, as he was writing his thesis, Les Structures élémentaires de la parenté: 'It was from D'Arcy Thompson that Lévi-Strauss elaborated on one of the keystones of structuralismthe idea of transformations.'In other words, one of anthropology's most prominent inquiries into cultural diversity was significantly inspired by the phenomenon of biological diversity.Lévi-Strauss was convinced that diversification is not an altogether random process but comprehensible in terms of structural regularities amenable to formalization.In addition to his comparative studies on kinship systems (Lévi-Strauss, 1969[1949]), he devoted the four volumes of Mythologiques to tracing structural transformations through a long series of myths recorded among Indigenous peoples from Brazil to Alaska.Common to both these projects is the premise that seemingly diverse series of cultural phenomena can be analytically derived from underlying structures of human thought and classification.Much as landscape surfaces in part derive from geology, the structuralist approach views empirical cultural diversity as generated by invisible substrates that can only be revealed through analysis.
Lévi-Strauss' structuralism was fundamentally inspired by the linguistic theories of Ferdinand de Saussure and Roman Jakobson.He recognized language as a system of differences in which the meanings of individual phonemes derive from their relations to other phonemes.Approaching other cultural phenomena as cognate to language, Lévi-Strauss consistently sought to discover the smallest constituent units, the combinations and recombinations of which generate meaning.In his analyses of kinship systems, he identified relationship terms, marriage rules, and prescribed interpersonal attitudes as constituting such 'atoms of kinship' (Lévi-Strauss, 1963a).In the structural study of myth, he similarly defined the basic 'mythemes' that recur in various transformations (Lévi-Strauss, 1963b).He was convinced that the empirical diversity of kinship systems and myths was generated by structures of human consciousness whose transmutations were ultimately as comprehensible in terms of mathematical formulation as the diversity of biological forms that had preoccupied D'Arcy Thompson.
Lévi-Strauss' intuitions about the affinities between the biological diversity mapped by D'Arcy Thompson and the cultural/linguistic diversity theorized by Saussure and Jakobson prompt us to ask what species and languages have in common?In the previous section, we mentioned several biocultural analogies such as radiative branching, extinction rates, and similar methodological approaches (e.g.phylogenetic analysis, or cladistics).From an evolutionary perspective, it could be argued that the capacity for language emerged as a trait that accelerated and fine-tuned the adaptive advantage of early humans, as in the social collaboration required in hunting big game (see Tomasello, 2008).Both species and languages are expressions of codes (genetic and phonetic, respectively) and can thus be approached as communicative phenomena.In this perspective, genotype is to langue as phenotype is to parole.If genetic codes are transformed by the interaction of random mutations and varying selective pressures, we may ask through which mechanisms languages change?Are the kinds of phonetic shifts that can be traced by historical linguists the counterpart of genetic mutations (Croft, 2000)?What are the pressures that select for the maintenance and reinforcement of certain phonetic or other cultural shifts?In other words, what produces diversity in linguistic and other cultural phenomena?
In his analyses of kinship systems, Lévi-Strauss approached congruities between alternative representations of social organization not as implicating separate historical stagesas had the 'inventor' of kinship studies, Lewis H. Morgan, as well as other evolutionists like Tylor and Frazerbut as coeval transformations.The intriguing redundancies that characterize Indigenous social structures of central and eastern Brazil, which was the focus of much of his research, were undoubtedly conducive to his structuralist methodology: '[The various types of groupings found in these societies are] a series of expressions, each partial and incomplete, of the same underlying structure, which they reproduce in several copies without ever completely exhausting its reality' (Lévi-Strauss, 1963c, p. 130).
A central example of such structural congruity is the relation between exogamous, unilineal moieties (dual organization) and bilateral cross-cousin marriage, two alternative but equifinal representations of regularities in social organization generated by a repeated practice of sister exchange (symmetric alliance) between the consecutive generations of two intermarrying lines of men.Instead of previous interpretations 'riddled with evolutionism, ' Lévi-Strauss (1969[1949], p. 101) argued that 'it is not the hypothetical succession of the two institutions which should be considered, but rather their structure,' as 'cross-cousin marriage and dual organization correspond to different stages in the growing awareness' of 'those completely basic structures on which the very existence of culture rests.' 8 These two representations of symmetric marriage exchange represent but a segment of an extended series of transformations that realize diverse potentialities immanent in the structural logic of a simple two-line ('Dravidian') kin terminology (Hornborg 1988(Hornborg , 1998)).To account for the diversity generated by this linguistic code, we must envisage a continuous dialectic between cognitive processes (Lévi-Strauss' 'structures') and the actual social processes of exchange that they engender.Cultural diversity, in other words, cannot be exhaustively derived from universal properties of the human mind, but results from diverging trajectories of interaction between mind and the varying social and ecological worlds with which it becomes entangled.A group's kinship terminology, for example, may reflect the group's position within an extensive regional system of exchange (Hornborg, 2013).Neither kinship systems nor myths are completely confined to human consciousnessthey are inextricably entwined with the social regularities and natural environments that they reflect.This becomes particularly evident as we follow Lévi-Strauss in deciphering the shifting zoological and other components of common mythological themes that he thought could be traced through countless campfire narrations over the millennia throughout the Americas.The variations derive not only from successive mutations in the replication of myths over time, but also from the introduction of new elements inspired by shifting social and ecological contexts.Yet, the structuralist approach is rarely explicit about the social agency underlying the cultural variation that it aspires to account for.We are finally left with the question of how cultural diversity is produced: what are the incentives for distinguishing a group's identity against another?
Although famous for his references to universal features of the human mind, Lévi-Strauss sensed a pervasive human preoccupation with difference.He has been vilified for maintaining that cultural identity is inextricably connected to ethnocentrism and an emphasis on contrasts vis-à-vis other groups, which Didier Eribon refers to as the idea that 'cultures want to be opposed to one another' (Lévi-Strauss, 1994, p. 422).Lévi-Strauss' position follows from the fundamental linguistic and structuralist observation that meaning, like information in general, derives from difference.It harmonizes well with Gregory Bateson's (1972) theory of schismogenesis, that is, processes through which two individuals or groups reinforce the distance between them by increasingly entrenching and emphasizing their distinctness. 9Paradoxically, it also harmonizes well with Fredrik Barth's (1998Barth's ( [1969]]) interpretation of how ethnic/cultural differences are organized, which arguably offers a foundation for the theorization of ethnogenesis. 10This convergence between Lévi-Strauss and Barth is paradoxical, as the former has been accused of cultural essentialism while the latter considers his argument constructionist (Barth (1998(Barth ( [1969]], p. 6).It appears that Lévi-Strauss' attribution of significance to cultural contrasts was denigrated because it predated the constructionist turnits reception was mired in an anthropology still steeped in essentialism. 11 Barth observes that 'the simplistic view that geographical and social isolation have been the critical factors in sustaining cultural diversity persists' (p.9).As against this view, he affirms that 'ethnic distinctions do not depend on an absence of social interaction and acceptance, but are quite to the contrary often the very foundations on which embracing social systems are built' (p.10).Barth shows that ethnic differences and boundaries may be emphasized precisely where the interdependence of groups is particularly strong, as when they occupy distinct ecological niches or positions in a traditional division of labour, regularly exchanging goods or services.While recognizing the parallels between biological populations and ethnic groups, he emphasizes the quite different role of boundaries in the two contexts.Largely following Barth, there is today a broad consensus among anthropologists that cultural or ethnic identity is not an objective, primordial essence but an actively communicated experience implicating and reinforcing contrasts with other such identities.The significance of such cultural contrasts is, for instance, a recurrent theme in David Graeber's and David Wengrow's (2021) recent narrative of the history of humankind, in which they explain and abundantly illustrate processes of schismogenesis (Graeber & Wengrow, 2021, pp. 56-58, 165, 175, 180-181, 200, 207, 226, 245, 254, 332, 349, 445, 504).
Given that language is a quintessential marker of ethnic identityas reflected in the widespread concept of 'ethnolinguistic' groupsthis anthropological understanding of ethnogenesis as founded on cultural contrasts should have significant implications for historical linguistics.It should be an important consideration in any attempt at explaining linguistic diversity.As we have seen, however, a common assumption is that linguistic diversity, like biodiversity, reflects periods of relative isolation of the adjacent populations (Skutnabb-Kangas & Harmon, 2017).
Increasing social interaction, in this view, will lead to homogenization.In a similar vein, the biologically inspired Punctuated Equilibrium Model developed by R.M.W. Dixon posits that during long periods of equilibrium, 'as a result of linguistic diffusion, all the languages in the geographical area would become more similar in structural profile' (Dixon & Aikhenvald, 1999, p. 17).Only when such periods of equilibrium are punctuated by some 'cataclysmic' natural or social event, the model suggests, there is a diversification of languages, until equilibrium is reestablished and the tendency toward homogenization resumes.Such events might be a flood or drought, a new means of producing food, conquest by an expanding neighbour, or migration into uninhabited territory.The assumption is that, during most of history, the interaction of politically equivalent groups will, as a normal course of affairs, enhance linguistic convergence.This assumption seems difficult to reconcile with the abovementioned perspectives of Lévi-Strauss, Bateson, and Barth.In the latter view, cultural diversification is actively generated by humans seeking to distinguish their own identities from those of othersnot as a result of cataclysmic events or isolation but as a regular aspect of social life.In this view, it is not diversification but homogenization that needs to be explained in terms of historical political processes.
These two perspectives suggest contradictory or at least divergent accounts of the sources of ethnolinguistic diversity.To test their relevance, the next section offers general overviews of the post-Columbian linguistic histories of Amazonia and the Andes, respectively.The final section will suggest some conclusions on the applicability of anthropological models of diversification in the linguistic histories of these two regions, and on the general conditions for cultural diversity.

Ethnolinguistic diversification and homogenization in Indigenous South America
At the time of European contact, there may have been around 700 languages spoken on the South American continent.Of these, only slightly over half have survived until today, 'a number, however, that is rapidly dwindling' (Muysken & Crevels, 2020, p. 254). 12Recent attempts at classification of currently spoken languages have reached varying estimates of the number of 'genealogical units' (language families and isolates), ranging between 107 and 118.Linguists recognize around 50 families and a slightly larger number of language isolates.Pieter Muysken and Mily Crevels observe that 'there is a broad consensus of linguistic diversity rivaled by only a few other places in the world' (p.257).Because of the difficulties of distinguishing effects of language contact from those of common genealogy, they find that 'the linguistic diversity of South America … poses problems similar to that of a 500-piece jigsaw puzzle, of which neither the color nor the shape of the pieces is fully specified and the model is absent' (p.267).Much of the difficulty of reconstructing the historical linguistics of South America derives from the disastrous demographic collapse following European contact: it is frequently estimated that Indigenous populations during the colonial period were reduced by around 90%, compared to what they had been in 1492.

Amazonia
A similar sense of bewilderment is conveyed by the editors of an authoritative volume focusing on the Indigenous languages of Amazonia.'The Amazon basin,' they observe, 'is the least known and least understood linguistic region in the world': maps of language families give an impression of anarchy, as their distribution is more discontinuous than in any other part of the world (Dixon & Aikhenvald, 1999, p. 1).Yet, they note that 'each language family tends to have a characteristic profile in terms of the type of territory it is found in, methods of food procurement, and material culture' (p.4), observing that groups speaking Arawak, Carib, and Tupí languages tend to be found in the rainforest, practice horticulture, and make canoes, hammocks, and pottery, while Gê-speakers generally inhabit grasslands, have little agriculture, and lack canoes, hammocks, and pottery.In between horticultural groups in the rainforest are predominantly foraging groups belonging to smaller language families, some of which engage in symbiotic exchange relations with the cultivators.Dixon and Aikhenvald propose that 'some of these may be the remainder of earlier populations that occupied larger tracts of land before the agricultural expansion' (p.4).
There have been attempts to explain the expansions of at least two of the largest language families in the South American lowlands -Arawak and Tupíin terms of migrations reflecting the competitive advantage gained by populations practicing agriculture (Dixon & Aikhenvald, 1999, p. 4, 17;Lathrap, 1970).However, the distribution of ethnolinguistic groups and the prevalence of multilingualism suggest a more complex history of interaction (Eriksen, 2011;Hornborg, 2005;Hornborg & Hill, 2011).In 1492, the Arawak languages had a wider distribution than any other language family, ranging from Central America to Paraguay.In comparison with other groups, speakers of Arawak languages have been conspicuously focused on long-distance exchange, river navigation, intensive agriculture, ceremonial feasting, social hierarchy, and geographically extended, patrilineal identities (Hill & Santos-Granero, 2002).This cosmopolitan, Arawakan 'ethos' may have been first established in the northwest Amazon, from which it was conveyed by Arawak-speakers along the Orinoco to the Caribbean and along the Río Negro to the central Amazon and beyond, downriver to the mouth of the Amazon and upriver to the Andes.The emphasis on long-distance alliance-making, suppression of endo-warfare, and prestigious ceremonialism generated a cohesive network of densely populated, Arawak-speaking chiefdoms that dominated the fertile floodplains and main rivers of Amazonia during the first millennium CE.The extensive cartographic knowledge of such riverine traders is reflected in a list of ceremonially chanted toponyms recorded among the (Arawak-speaking) Wakuénai in the northwest Amazon, which reaches across the continent from the mouth of the Orinoco over the Río Negro to the mouth of the Amazon (Hill, 2002, p. 229, 235-236).The prestige and affluence of Arawak settlements may have attracted speakers of other languages along the rivers to adopt an Arawak language and identity, as Arawak men frequently married non-Arawak women, and their offspring generally identified with their fathers (Schmidt, 1917).Multilingualism was widespread, while Arawak may have served as a lingua franca that mediated exchange and other interaction along the river systems from the Orinoco to the lower Amazon.Five hundred years before the arrival of Europeans, Arawakan languages may thus have had an integrating function similar to that of Nheengatú (or Língua Geral) during the colonial period. 13Although tentative, this hypothesis would explain the widespread distribution of Arawakan languages in combination with pervasive bi-or multilingualism, which illustrates how homogenization on one level can be compatible with diversity on another.
While Arawak-speakers have an ancient connection to rivers and water transport, the Tupí, Carib, and Gê language families all appear to have originated in upland areas: Rondônia, Guyana, and central Brazil, respectively.Perhaps consonant with Aryon Rodrigues' suggestion that Tupí, Carib, and Gê may share a common linguistic ancestry (the 'Tu-Ca-Gê hypothesis'), it could be speculated that the routes of expansion of Arawakan languages created ethnic wedges that led to the demarcation of these and other, less dispersed families.Indeed, the distribution of Arawak languages is broadly aligned with the barriers that have historically separated other linguistic families: the Orinoco, Río Negro, Amazon, Ucayali, Purús, and Madeira Rivers, the llanos of Venezuela and Bolivia, and the coastal areas of Guyana.Several of the other language families, such as Carib, Pano, and Gê, may at one time have been denied access to the Arawak-dominated floodplains and other resource-rich areas.Variously engaged in warfare and trade, populations on both sides of such socioecological boundaries would have entrenched their markers of ethnolinguistic distinctness over time.Both the expansive, Arawak-speaking groups and other ethnolinguistic categories (such as Tupí, Carib, Gê, Pano, Tucano) that may have emerged in opposition to them were in their own ways ethnogenetic products of the overarching regional system.While displacing some older languages, the expansion of language families through Amazonia would thus also have generated new ethnolinguistic groups, through fragmentation of previously homogenous populations, circumscription of newly isolated groups, and various forms of inter-group boundary maintenance.
Rather than assuming that ethnolinguistic diversity in general is the result of an absence of interaction, the approaches of Barth, Lévi-Strauss, and Bateson instead suggest that it may be a consequence of the intensity of exchange between groups occupying different socioecological niches.Although linguistic and social cohesion at one level unified wide expanses of the Arawak exchange network, integrating riverine populations through much of Amazonia, its expansion encompassed diverse, bi-or multilingual ethnic groups and generated pervasive ethnic differences at its boundaries.Whether assuming the form of trade or hostilities, such cultural differences were actively reinforced by the groups involved.From an anthropological perspective, the general expectation should be that linguistic diversification is enhanced by interaction among groups rather than by their isolation (Hornborg & Hill, 2011, p. 11).
The linguist Patience Epps (2020) has systematically tested this perspective by examining the evidence from six regions in Amazonia with particularly high linguistic diversity: the Upper Río Negro, the Upper Xingú, the Southern Guianas, the Caquetá-Putumayo, the Guaporé-Mamoré, and the Gran Chaco.'Despite common assumptions,' she concludes, 'one explanation that does not appear to hold up to closer scrutiny is that Amazonian linguistic diversity is simply a factor of isolation ' (p. 275).She shows how the various groups in these linguistically diverse areas tend to actively promote their various cultural differences, for example by specializing in the manufacture of different trade goods, even though the materials may be equally available for all.Epps understands the linguistic exogamy of the Upper Río Negro as an expression of a widespread Amazonian equation of language and identity. 14Despite pervasive multilingualism, each person inherits an essential linguistic identitypatrilineally in the Upper Río Negro (cf. Aikhenvald, 2002, p. 21, 27) and is expected to keep that identity distinct, for example by avoiding to mix languages (code-switching) within a single sentence or to borrow words (or even sounds) from other languages.The recurrent pattern throughout the linguistic areas examined by Epps is that there is more resistance to lexical borrowing than to diffusion of grammatical structures, which no doubt reflects the fact that speakers tend to be more aware of the former than the latter.Epps also suggests that there may be 'active processes of [phonological] diversification' at play in these areas (Epps, 2020, p. 286).Such diversification of sounds and accents, generating local dialects and ultimately separate languages, differs fundamentally from biological speciation in being the result of active identity construction rather than ecological specialization.
The maintenance of linguistic diversity in several areas of Amazonia, as Epps suggests, probably reflects the 'absence of profound differences in socio-economic dominance between communities, such that groups were not normally drawn into top-down social structures imposed by other groups ' (p. 287).In this respect, the linguistic history of much of Amazonia has differed from that of the prehispanic Andes, where states and empires have repeatedly subjugated populations over wide areas.Nevertheless, the expansion of Arawak-speaking chiefdoms along the floodplains of Amazonia must at times have imposed highly hierarchical relations between adjacent groups.Such expansions could be expected to create conditions that were conducive to linguistic homogenization, much as has the imposition of European colonial languages and trade languages such as Nheengatú.As Epps observes, the latter contexts have generated a quite different pattern of linguistic borrowing, with many examples of strong lexical diffusion, frequent code-switching within sentences, and complete language shifts (p.289).As we shall see, however, precolonial Indigenous language expansions appear not to have significantly reduced ethnolinguistic diversity.
Based on her detailed case study of recent language change among Arawak-and Tucano-speaking groups in the northwest Amazon, Alexandra Aikhenvald (2002, pp. 274-276) usefully distinguishes three kinds of language contact situations: a 'traditional' areal contact without a relationship of dominance; a 'more recent' contact between two languages where one is dominant (in this case Tucano); and 'current language contact with diglossia,' where Portuguese and Indigenous languages have a complementary distribution (Portuguese being used primarily in schools, written communication, and media). 15In the first case, there is a tendency toward convergence of grammatical structures but a general disapproval of code-switching and lexical borrowings.In the second case, the (Arawak) Tariana language is suffering from attrition and gradual replacement by Tucano.Finally, the diglossic relationship between Indigenous languages and Portuguese can to some extent contribute to keeping them apart (Aikhenvald, 2002, p. 186, 241).As a whole, however, the long-term prospects of Indigenous Amazonian languages are assessed as sombre: Every year the indigenous languages are used less and less, and Spanish and Portuguese more and more.Every year another few languages pass into oblivion.Of the estimated 300 languages now spoken in Amazonia, only a small fraction are likely to be still actively used in 100 years' time.… In lowland Amazonia there is little hope for even medium-term survival of any language.(Dixon & Aikhenvald, 1999, p. 7, 19) A conclusion from the evidence surveyed in this section is that linguistic diversity and biological diversity are promoted by quite different kinds of factors.Rather than reflecting biogeographical isolation, linguistic diversification may be enhanced by regular interaction between groups.A significant consideration, however, is the nature of this interaction: where reciprocity and political equality is the norm, cultural diversity tends to be entrenched, whereas hierarchy and political dominance, unless tempered by diglossia, may be conducive to homogenization.With this observation in mind, we turn now to the linguistic history of the Andes.

The Andes
When the first Europeans arrived in the sixteenth century, most of the Andean area from northern Ecuador to central Chile was under the control of the Inca Empire (Tawantinsuyu).For several millennia prior to the Inca, the area had experienced a series of expansive social systems that dominated parts of the coast and/or highlands.Judging from archaeological evidence, the periods 900-200 BCE and CE 600-1000 were times of extensive cultural, economic, and/or political unification, referred to as the Early and Middle Horizons, respectively (the Inca Empire being the Late Horizon, customarily dated CE 1438-1532).Given the homogenizing influences that we expect of these horizons, the formidable linguistic diversity of the Andes on European arrival suggests that ethnic groups were generally less vulnerable to language loss in prehispanic times than in the centuries thereafter.
In his classic account of Inca culture at the time of the Spanish conquest in 1532, John Rowe (1946, pp. 185-192) lists a total of 86 'tribes and provinces' mentioned in historical sources.He frequently notes in passing that a group in the list 'had their own language' or even that 'a number of local languages were spoken,' but estimates that dozens of languages perished without leaving a trace in Spanish documents.Linguists observe that the 'wealth of mutually unrelated languages' in the Andean area is 'as opaque as ever,' and that 'scores of native languages, including entire families, have disappeared, often without leaving a trace' (Adelaar & Muysken, 2004, p. 2).They estimate that 'maybe more languages became extinct here during the last five centuries than anywhere else on the continent.In a majority of cases, such languages have remained undocumented ' (p. 22).
Areas where a number of local languages disappeared during the colonial period include highland Ecuador and the north Peruvian coast (Adelaar & Muysken, 2004, p. 12, 23).In many cases, such local languages were not displaced by the two major native languages (Quechua and Aymara) until the end of the colonial period or even the twentieth century.The spread of Spanish has further reduced linguistic diversity in the Andes, as exemplified by its replacement of Mochica (spoken on the northern coast around Chiclayo and Lambayeque) and other languages in northern Peru (p. 23).Other examples of languages that appear to have been lost in the colonial period or later include Quingnam (the language of the pre-Inca Chimú polity, spoken on the coast between Trujillo and Lima), Culli (in the highlands between Trujillo and the Marañon), Cholón (in the upper Huallaga valley), and Puquina (the language of Tiwanaku, spoken around Lake Titicaca and some areas in southwestern Peru and the Bolivian highlands).
Quechua has been referred to as the lingua franca of the Inca Empire (D' Altroy, 2002, p. 44;Mannheim, 1991).The dialect of Quechua promoted by Tawantinsuyu was that of Chincha, a trade-oriented polity on the south coast that had once been part of the Wari sphere of influence.Although Quechua was used in official matters throughout the Inca Empire, much of the population was bi-or multilingual, conducting their everyday social life in non-Quechua languages (Mannheim, 1991, pp. 44-46).This diglossic situation preserved smaller languages in a way that Spanish colonialism and recent globalization has not.Given the pervasive equation of language and identity, linguistic diversity was undoubtedly promoted by the Inca enforcement of ethnic identity (Sillar, 2012, p. 307).The Spanish chronicler Bernabé Cobo reported that the men and women of each nation and province had their insignias and emblems by which they could be identified, and they could not go around without this identification or exchange their insignias for those of another nation, or they would be severely punished.They had this insignia on their clothes with different stripes and colors, and the men wore their most distinguishing insignia on their heads; each nation was identified by the headdress.(Cobo, cited in D'Altroy, 2002, p. 295) Archaeologically documented cultural and political expansions through Andean prehistory no doubt fundamentally influenced the distribution of language families in the region.Increasingly concerted efforts at cross-disciplinary collaboration have offered plausible accounts of such processes (Heggarty & Beresford-Jones, 2012).By and large, the archaeological, ethnohistorical, and linguistic evidence appears to support the proposal of David Beresford-Jones and Paul Heggarty (2012) that the first major spread of Quechua was propelled by the Middle Horizon expansion of the Wari Empire, centred in the Mantaro basin, and their suggestion that the spread of Aymara was associated with the Early Horizon influence of Chavín, emanating from the site of Chavín de Huántar in Ancash. 16The many linguistic affinities between Quechua and Aymara may derive from their long coexistence in adjacent ecological zones, the former associated with terraced maize cultivation in the valleys and the latter with camelid pastoralism at higher altitudes. 17 Wari influence (and the distribution of Quechua around CE 1000) did not extend much further south than Cuzco (destined to be the Inca capital); immediately to its south was the domain of Tiwanaku, which dominated the Titicaca basin during the Middle Horizon.The main language spoken in Tiwanaku was Puquina (Cerrón-Palomino, 2012;Torero, 2002). 18Following the collapse of Tiwanaku at the end of the Middle Horizon, the Puquina language declined and was largely replaced by the spread of Aymara from the northwest.Yet, in 1575 it was recognized by the Spaniards as one of the three 'general languages' (along with Quechua and Aymara) of Peru (Torero, 1987, p. 343).By the end of the sixteenth century, Puquina was spoken in an area north and east of Lake Titicaca and was still significant enough to count as one of the Indigenous languages that parish priests in the area should be familiar with (Bouysse-Cassagne, 2010;Domínguez-Faura, 2014).Its fragmented occurrence in other parts of the southern Andes at this time may have reflected other remnants from the Tiwanaku period, prehispanic (Inca or pre-Inca) settlements of Puquina-speakers beyond their altiplano heartland, and/or dislocations due to the Spanish labour tax (Domínguez-Faura, 2014, p. 196).Although, as mentioned, the Inca generally enforced the maintenance of ethnic differences, their extensive mitma resettlement schemes contributed to the decline of Puquina, as did their violent subjugation of the Puquina-speaking Colla north of Lake Titicaca, who eventually shifted to Quechua (Sillar, 2012, pp. 307-308, 314). 19Nevertheless, it was not Inca policy that finally extinguished Puquina as an ethnolinguistic identity, but Spanish colonialism.The only traces of Puquina that survived the nineteenth century are widespread toponyms (Adelaar & Muysken, 2004, p. 175, 350-351;Torero, 2002) and a ceremonial language preserved by Callawaya herbalists in the formerly Puquina-speaking Charazani area, east of Lake Titicaca (Adelaar & Muysken, 2004, p. 351, 356;Stark, 1972). 20 When the Inca Empire emerged in the early fifteenth century, its four subdivisions (Tawantinsuyu = 'the four parts') appear to have been conceived in terms of the dominant ethnolinguistic categories recognized in Cuzco at that time: the Puquina-speakers of Collasuyu to the southeast, the Aymara-speakers of Condesuyu to the southwest, the Quechua-speakers of Chinchasuyo to the northwest, and the speakers of Arawak and other languages of Antisuyo to the northeast (Hornborg, 2014). 21These categories could thus perhaps be understood as sediments of past cultural and linguistic expansions: Tiwanaku (Puquina), Chavín (Aymara), Wari (Quechua), and Arawak, respectively.Of these four, only Wari can with some certainty be characterized as having been largely propelled by political and military means, as was the further promotion of Quechua by Wari's successor, the Inca Empire.The remaining three were primarily associated with the spread of trade relations, ideologies, and prestigious cosmopolitan identities through extensive areas of South America.Although they might be viewed as early precursors to globalization, they clearly did not have the same devastating consequences for linguistic diversity.In the concluding section, we shall consider some possible reasons for this difference.

Conclusions
We have seen that ethnolinguistic diversity in South America was formidable in 1492, regardless of whether populations had been integrated into hierarchical political systems.The equation of language and identity was fundamental not only among groups of horticulturalists and foragers in Amazonia but also among the millions of ethnically diverse peasants subordinated by the Inca Empire.In the five centuries since the European conquest, however, almost half of the around 700 languages spoken in 1492 have been lost, and most of those remaining are falling into disuse and in danger of complete abandonment.Linguists have been prone to think of this loss of ethnolinguistic diversity in terms of concepts and models strongly influenced by established approaches to biological diversity: phylogenesis, genetic relations, endangerment, extinction, and so on.However, to account for the loss of Indigenous languagesbeyond the tangible historical forces of epidemics, colonialism, and genocidewe must transcend biological analogies and acknowledge that the maintenance of ethnic identity is contingent on social-structural conditions that allow for a certain level of local cultural autonomy.As Indigenous peoples are well aware, the continued vitality of their identities and languages today pivots on their capacity to safeguard their cultural specificity vis-à-vis the homogenizing forces of globalizing modernity.Pervasively, such preservation of linguistic diversity can be understood in terms of diglossia: the local language is reserved for social contexts in which ethnic specificity can continue to be cultivated, alongside but delimited from one's engagement in globalized social relations.A corollary of this observation is that the cosmopolitan lingua franca is also restricted to delimited social contexts.
It can be argued that the difficulties of upholding such boundaries between local and global social contexts is diagnostic of colonial and modern conditions.Ever since the arrival of Europeans in South America, there have been fervent efforts to 'convert' Indigenous peoples to globalized identities, whether in the guise of Christianity, modern 'civilisation', European languages, or the capitalist world economy.The colonial exertion of power has thus been markedly different from that of the Inca, who instead enforced the maintenance of ethnic identities among its subjects.Various ethnolinguistic groups were incorporated into Tawantinsuyu without any expectation of cultural assimilation.Both the Inca expansion in the Andes and the extension of Arawakan trade networks in Amazonia illustrate how, in principle, far-flung geographical integration can coexist with ethnolinguistic diversity.In contrast to pre-Columbian patterns, the currently ongoing abandonment of Indigenous languages reflects the loss of social contexts for cultivating ethnic specificity.Complete language shifts can take place over the course of two generations, as when Indigenous elders are finding it difficult to communicate with their grandchildren.The homogenization of public discoursethrough media, schools, employment, commercials, and political administrationcolonizes the entire modern individual, leaving dwindling spaces for cultural resistance.Abstract, national or global identities make local or cultural identities increasingly superfluous by extinguishing the social contexts in which they might be reproduced.This is ultimately a consequence of how the logic of abstract exchange-value currently organizes the world economy, penetrating human lives even into the innermost recesses of their local contexts.
Given the experience of modern globalization, it is understandable that increasing interaction is often assumed to lead to cultural and linguistic homogenization, but this assumption is clearly not applicable to premodern situations, in which the phenomenon of diglossic bilingualism appears to have enhanced ethnolinguistic diversity.As well understood by Barth, Bateson, and Lévi-Strauss, identity pivots on the expression of specificity as defined through contrasts.Arguably, a population that reserves a space for the expression of local specificity is less likely to submit to abstract identity assertion such as potentially devastating nationalism.The frequent observation that linguistic diversity may promote biodiversity is also quite robust, as locally contextualized, ethnobiological classification systems and 'traditional ecological knowledge' may indeed be vital assets for sustainable resource management.To advocate the use of social theoryrather than biological analogyto explain ethnolinguistic diversity is not to deny that cultural diversity may enhance biodiversity.While there are several other good reasons to safeguard the relative autonomy of local social life, such autonomy thus appears to be essential to both cultural and biological diversity, and perhaps ultimately also to world peace.

Notes
1.The prominence of language in these considerations on cultural diversity deserves reflection.Culture clearly includes much more than language, but no other cultural phenomenon is as readily identified, delineated, and quantified as language.This explains why so many scholars have invoked linguistic diversity as a proxy for cultural diversity.It should be kept in mind, however, that other significant but less easily identified aspects of cultural diversity may be equally threatened by globalization.2. See, for instance, Posey & Balée, 1989;Balée, 1994. 3. The most extreme version of this perspective is no doubt Marvin Harris's theory of 'cultural materialism' (Harris, 1980).4. The homogenizing consequences of modernity are as evident in globalized human food habits as in language use: of the 7,000 kinds of plants that have been grown or collected as food in human history, only '20 species provide 90 percent of the world's food and just threewheat, maize, and ricesupply more than half' (Wilson, 1992, pp. 287-288).Commenting on the effects of the Green Revolution on rice cultivation in India, Wilson (1992, p. 301) warns that 'homogeneity means vulnerability.' 5.The classic case of speciation is the divergent evolution of different species of finches in the Galapagos Islands, which stimulated Charles Darwin's thoughts on evolution.6. Claude Lévi-Strauss is a prominent example of an anthropologist who deplored cultural homogenization.As a consequence of his critical view of globalization, he was extensively derogated.7. See also Lévi-Strauss, 1994, p. 427. 8.For Lévi-Strauss, kinship systems, in codifying the incest taboo, represent the pivotal distinction between nature and culture.Like myths, in his view, they domesticate raw nature by transmuting biological processes through a filter of cultural rules and categories.9.For the theory of schismogenesis, see Bateson, 1972, pp. 35-46 andBateson, 1958, pp. 171-197. Bateson also repeatedly refers to 'difference that makes a difference' as fundamental to communication, ecology, and evolution (see Bateson, 1972, p. 457).From this bio-or ecosemiotic perspective, the reduction of diversity through globalization can be regarded as a reversal of the evolutionary trend toward increasing complexity (Emmeche, 2001;Eriksen, 2023;Hornborg, 2023).10.Although discussions of ethnogenesis (e.g., Moore, 1994;Hill, 1996) have shown surprisingly little interest in Barth's (1998) contribution, they similarly focus on the social construction of cultural identity.11.The widespread shift from cultural essentialism to constructionism in anthropology is related to the shift from cladistic to what Moore (1994) calls rhizotic theories of evolutionary processes.Whereas the cladistic approach tends to assume an equivalence between biological populations, languages, and archaeological cultures, rhizotic perspectives emphasize that 'human history has always been characterized by interaction across profound ethnic and cultural boundaries, by the amalgamation of linguistic traits, and by the recurrent hybridization of cultures' (Moore, 1994, p. 937).An apt metaphor for the latter view is that of a braided river, where channels separate and recombine.A rhizotic perspective on linguistic history helps explain the difficulties linguists are having in distinguishing genealogical from areal contact phenomena (Muysken & Crevels, 2020;Eriksen & Danielsen, 2014).12. Another overview gives the figure 574 South American languages in 1492, reduced to around 420 today (Lev, 2021, p. 330).13.Nheengatú is a Tupí-based trade language that spread throughout the Amazon basin in the seventeenth and eighteenth centuries and is still spoken as a first language by some populations on the upper Río Negro (Jensen, 1999, p. 127;Campbell, 1997, p. 23).14.The phenomenon of linguistic exogamy, in materializing cultural fictions of unilineal genealogy, illustrates how problematic it is to assume that languages and genes are aligned over time.15.In diglossic situations, two languages with different social prestige are used in separate contexts.The term diglossia was originally coined by Charles A. Ferguson (1959, p. 325) for situations 'where two varieties of a language exist side by side throughout the community, with each having a definite role to play.'More recently, it has been applied to the similar coexistence of two unrelated but functionally complementary languages, 'wherefor exampleone language, or dialect, may be used in the home and in religious observances, and another in all other circumstances' (Aikhenvald, 2002, p. 10). 16. Richard Burger (2012, pp. 150-151) points out that the identification of the Chavín sphere of interaction with the expansion of Aymara precludes traditional explanations of linguistic expansion, as archaeology shows that the Chavín phenomenon was not associated with agricultural expansion, warfare, or demographic displacement, but with the spread of a cosmopolitan identity and long-distance exchange requiring a prestigious lingua franca.The same observation can be made, for instance, regarding the expansion of Arawak throughout Amazonia and the identification of Tiwanaku with the expansion of Puquina.17.The connection between Chavín, Aymara, and llama pastoralism suggests that the intensification of long-distance exchange that characterized the Early Horizon was made possible by the expanded use of llama caravans.18.Several linguists maintain that Puquina was affiliated to the Arawak language family (Torero, 2002, pp. 488-489;Adelaar & Muysken, 2004, p. 175, 353;Muysken & Crevels, 2020, p. 259).19.Paradoxically, the Inca elite traced their mythological origins to Lake Titicaca and may have spoken Puquina among themselves as a 'secret language' (Cerrón-Palomino, 2012).In other words, even the Inca nobility relied on diglossia to maintain its ethnolinguistic identity.Cobo reports that, 'besides the language of Cuzco, which is the general language introduced by the Incas in their empire and was spoken by all their subjects, they knew another, which they used solely when they were dealing with and conversing with those of their own lineage' (Cobo, cited in Cerrón-Palomino, 2012, p. 266).20.Again, even the survival of these fragments of Puquina can be attributed to diglossia.21.For an identification of Antisuyu with Arawak languages, see the recent argument of Darryl Wilkinson (2022).

Disclosure statement
No potential conflict of interest was reported by the author(s).

Alf
Hornborg is an anthropologist and Professor Emeritus of Human Ecology at Lund University, Sweden.He is the author of The Power of the Machine (2001), Global Ecology and Unequal Exchange (2011), Global Magic (2016), Nature, Society, and Justice in the Anthropocene (2019), and The Magic of Technology (2023).