23
KNAW Conference on Language Dispersal in Africa, Southeast Asia, and South America Abstracts The dynamics of areas and universals Balthasar Bickel University of Zürich In explanations of how linguistic structures are distributed in the world, the pendulum has swung back from an emphasis on universals that has dominated the second half of the 20th century to a renewed emphasis on local developments and areal diffusion. This talk aims at moving the question of areas vs. universals away from ideological debates and to turn it into an empirical issue. To meet this goal, areal and universal hypothesis are reformulated as unified models that estimate probabilities of change over time under given conditions of geography and grammatical structures. Such models allow us to estimate the relative impact of areal diffusion and universals simultaneously. This overcomes the problem that any demonstration of areas requires previous knowledge of universals (because the features that establish an area must be shown not to be universally correlated with each other) and that, vice-versa, any demonstration of universally requires previous knowledge of areas (because universals must be shown to be independent of areas). The reformulation has two requirements: (i) a universally applicable method that allows estimating probabilities of diachronic trends from synchronic data under given conditions; (ii) a way of grounding statistical models in causal theories outside the model. Using case studies on the distribution of case marking in Eurasia I will present and discuss methods for diachrony estimates that allow testing models that are grounded in what is known about the history of language contact (e.g. from population genetics) and in what is known about universal cognitive behavior affecting language change (e.g. from processing). Multidisciplinary Perspectives on the Bantu Expansion: A State of the Art Koen Bostoen KongoKing Research Group, Ghent University & Université libre de Bruxelles The picture of Africa as a homogenous linguistic landscape consisting of a small number of large families and few isolates has been revised over the last decade in favour of higher diversity (cf. Dimmendaal 2008). Nevertheless, the continent remains linguistically much less diverse than certain other continents. One reason for this lower diversity is the large-scale expansion of recent language families. The best-known case is no doubt the ‘Bantu Expansion’. The Bantu languages constitute Africa’s largest language family, both in terms of number of speakers and languages as well as in terms of geographical spread. In spite of this numerical and geographical superiority, the Bantu family is widely recognized as forming only a low-level subgroup of the Niger- Congo phylum, being at most a sub-branch of a branch of the Benue-Congo node (Nurse and Philippson 2003; Schadeberg 2003; Williamson and Blench 2000). The rapid expansion of Bantu speech communities has fascinated generations of scholars from a wide variety of disciplines. The Bantu Expansion has been construed as a textbook case of farming/language dispersal. Diamond and Bellwood (2003: 598) express this vision in the plainest way possible when describing the Bantu Expansion as one of the world’s clearest examples of “expanding farmers bearing their own archaeologically visible culture, domesticates, skeletal types, genes, and languages.” Although such claims contain an element of truth, they are definitely too bold to correctly reflect the present state of knowledge in different disciplines. In this talk, the facts and fictions on the Bantu Expansion are discussed from the angle of different

KNAW Conference on Language Dispersal in Africa, Southeast

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: KNAW Conference on Language Dispersal in Africa, Southeast

KNAW Conference on Language Dispersal in Africa, Southeast Asia, and South America

Abstracts

The dynamics of areas and universals

Balthasar Bickel University of Zürich In explanations of how linguistic structures are distributed in the world, the pendulum has swung back from an emphasis on universals that has dominated the second half of the 20th century to a renewed emphasis on local developments and areal diffusion. This talk aims at moving the question of areas vs. universals away from ideological debates and to turn it into an empirical issue. To meet this goal, areal and universal hypothesis are reformulated as unified models that estimate probabilities of change over time under given conditions of geography and grammatical structures. Such models allow us to estimate the relative impact of areal diffusion and universals simultaneously. This overcomes the problem that any demonstration of areas requires previous knowledge of universals (because the features that establish an area must be shown not to be universally correlated with each other) and that, vice-versa, any demonstration of universally requires previous knowledge of areas (because universals must be shown to be independent of areas).

The reformulation has two requirements: (i) a universally applicable method that allows estimating probabilities of diachronic trends from synchronic data under given conditions; (ii) a way of grounding statistical models in causal theories outside the model. Using case studies on the distribution of case marking in Eurasia I will present and discuss methods for diachrony estimates that allow testing models that are grounded in what is known about the history of language contact (e.g. from population genetics) and in what is known about universal cognitive behavior affecting language change (e.g. from processing).

Multidisciplinary Perspectives on the Bantu Expansion: A State of the Art

Koen Bostoen KongoKing Research Group, Ghent University & Université libre de Bruxelles The picture of Africa as a homogenous linguistic landscape consisting of a small number of large families and few isolates has been revised over the last decade in favour of higher diversity (cf. Dimmendaal 2008). Nevertheless, the continent remains linguistically much less diverse than certain other continents. One reason for this lower diversity is the large-scale expansion of recent language families. The best-known case is no doubt the ‘Bantu Expansion’.

The Bantu languages constitute Africa’s largest language family, both in terms of number of speakers and languages as well as in terms of geographical spread. In spite of this numerical and geographical superiority, the Bantu family is widely recognized as forming only a low-level subgroup of the Niger-Congo phylum, being at most a sub-branch of a branch of the Benue-Congo node (Nurse and Philippson 2003; Schadeberg 2003; Williamson and Blench 2000). The rapid expansion of Bantu speech communities has fascinated generations of scholars from a wide variety of disciplines. The Bantu Expansion has been construed as a textbook case of farming/language dispersal. Diamond and Bellwood (2003: 598) express this vision in the plainest way possible when describing the Bantu Expansion as one of the world’s clearest examples of “expanding farmers bearing their own archaeologically visible culture, domesticates, skeletal types, genes, and languages.” Although such claims contain an element of truth, they are definitely too bold to correctly reflect the present state of knowledge in different disciplines. In this talk, the facts and fictions on the Bantu Expansion are discussed from the angle of different

Page 2: KNAW Conference on Language Dispersal in Africa, Southeast

Page 2 of 23

disciplines: historical linguistics, archaeology, archaeobotany, palynology, molecular anthropology, phyotogeography, etc. with special reference to several recent interdisciplinary research projects in which the author was involved (Barbieri et al. 2012; Bostoen 2005; 2007; forthcoming; Bostoen et al. forthcoming; Bostoen and Sands 2012; de Filippo et al. 2011; de Filippo et al. 2012; Kahlheber et al. 2009; Neumann et al. 2012; Pakendorf et al. 2011).

Languages and Genes: Taking stock

Bernard Comrie Max Planck Institute for Evolutionary Anthropology University of California Santa Barbara Cooperation between linguistics, more specifically comparative-historical linguists, and geneticists, more specifically population geneticists, has increased significantly in recent years in a concerted effort to unravel a range of problems relating to migrations of people and spread of languages, of contact among populations and among languages. While the most extreme optimism has not been vindicated – we are still far from unraveling the full picture – nonetheless considerable progress has been made, both in addressing particular cases and in refining the methodology.

The paper will examine the lessons to be drawn from combined linguistic and genetic studies involving languages/populations of the Caucasus, Siberia, South Asia, Africa, and the Americas, perhaps adding or subtracting individual cases studies as time allows or requires.

Particular methodological issues that will be addressed include: (a) the need to take both linguistic and genetic data seriously and independently of one another; (b) controversies within linguistics regarding broader claims of genealogical relatedness among languages; (c) the importance within genetics of distinguishing between individuals and populations and between recombinant and non-recombinant DNA. More generally, it will be argued that only mutual respect between practitioners of the disciplines involved will lead to real progress in this interdisciplinary enterprise.

Patterns of diversification and contact: An African perspective

Gerrit J. Dimmendaal University of Cologne The linguistic map of Africa as it manifests itself to us today is the result of a range of factors, involving climate changes, technological innovations as well as social factors. In my presentation, I hope to show how climate changes resulted in the expansion as well as reduction of human habitation and thereby of language families. Technological changes (for example, the introduction of bows and arrows, axes, dogs, pastoralism, or agriculture) probably played an important role as well. A quintessential role, however, is to be attached to social factors, in the present author’s view.

These latter factors involve the maintenance or creation of linguistic boundaries as well as the diffusion of linguistic features and language shift. These factors help to explain, for example, why various linguistic isolates are situated at the edge of expansion zones on the African continent.

On a methodological side, the value of interdisciplinary approaches, in particular the interaction with archaeologists, geographers, and geneticists is emphasized. Such interactions help to avoid certain misinterpretations of scientific results from other disciplines. Moreover, as I hope to show, these results may provide independent support for hypotheses on the linguistic history of certain areas.

Language and Population dispersals in Indo-Malaysia and the Pacific

Mark Donohue The Australian National University The spread of modern humans into and across Indo-Malaysia and the Pacific is significant for two reasons: firstly, the spread across Indo-Malaysia represents the earliest confirmed dispersal of humans across a marine environment, and involved numerous associated technologies that indicate sophisticated societies

Page 3: KNAW Conference on Language Dispersal in Africa, Southeast

Page 3 of 23

on the move. Secondly, the (much) later spread of Austronesian over such a wide region shows language replacement on a scale that is more reminiscent of the period of state-sponsored European colonisation than of a social landscape more than three millennia old.

Standard views, still repeated, cast the Austronesian dispersal as involving the collision of a technologically advanced farming society with 'indigenous Mesolithic hunter-gatherer communities': the pre-Austronesian populations are assumed to have disappeared without a lasting impact on the linguistic ecologies that were taken over by the Austronesians. This talk problematises the association of Austronesian languages with 'Austronesian peoples', and points out a number of dimensions in which the dispersal of Austronesian languages, originating in Taiwan, should not be equated with the technological and demographic steamroller that is often portrayed. This will involve discussion of the nature of pre-Austronesian society and language in the south-west Pacific, and the degree to which it has changed following 'Austronesianisation'.

Ideology and perspective: Considering sociocultural correlates of Amazonian linguistic diversity

Patience Epps University of Texas at Austin Amazonia presents a striking puzzle of linguistic diversity, rivaled by few other regions of the world. Some 50 distinct language families/isolates are distributed throughout the lowlands, clustered most heavily on the western periphery. What widespread families do exist tend to non-contiguous distributions, suggesting diaspora rather than wave-like advance (cf. Hornborg 2005, Clastres 1995[1975]). Explaining this diversity is a challenging task. The number and distribution of linguistic units cannot be attributed to time-depth alone, since South America is the most recently settled continent; nor to geographic obstacles (cf. Nichols 1992), since rivers and interfluvial zones can function as conduits or barriers alike. Amazonian linguistic diversity also cannot be explained simply as a factor of social isolation, particularly in light of evidence for demographic density and long-distance trade preceding European conquest (Heckenberger et al. 2003, Roosevelt 1994). Nor does the presence or absence of agriculture provide a clear-cut explanation, especially if we compare Amazonia (where almost all groups practice some degree of horticulture) to Australia, where Pama-Nyungan-speaking hunter-gatherers cover most of the continent. Instead, I argue that we should look to Amazonian social and cultural practices for important clues to our understanding of the region’s linguistic puzzles. Their relevance is evident when we look at particular microcosms of linguistic diversity within Amazonia, where interaction among groups is frequent within regional ‘systems’ but linguistic diversity widely maintained. The Vaupés region of the northwest Amazon is a well-known example; here, the practice of linguistic exogamy (obligatory marriage across language groups) has apparently fostered an explicit constraint against mixing languages, thus leading to remarkably low levels of code-switching, lexical borrowing, and language shift, despite high multilingualism (e.g. Jackson 1983, Aikhenvald 2002, Epps 2007). However, other regions in which high interaction among groups is not accompanied by linguistic exogamy (such as the Upper Putumayo and the Xingu) nevertheless show strikingly similar patterns of low lexical borrowing and explicit norms of language maintenance – even where grammatical categories and structures converge (i.e. features below speakers’ “limits of awareness”; cf. Silverstein 1981). Moreover, a wider survey of Amazonian languages (Bowern et al. 2011) indicates that rates of lexical borrowing are generally low in the region, in comparison to many other areas of the world. These patterns suggest that many Amazonian peoples engage in sociolinguistic practices that promote linguistic purity, and therefore linguistic difference – presumably grounded in ideologies that essentialize the link between language and identity. But how could an ideology of language be so pervasive, long-term, and widespread to constitute even a partial explanation for Amazonian linguistic diversity? I suggest that an ideology of linguistic purity and distinction may be linked to Amazonian cosmological perspectives relating to animacy and alterity (Viveiros de Castro 1998, Londoño Sulkin 2006, Descola 1994). According to indigenous outlooks, different groups of humans, animals, spirits, and even plants are understood to have points of view that are essentially equivalent, but based on distinct vantage points; they may also

Page 4: KNAW Conference on Language Dispersal in Africa, Southeast

Page 4 of 23

have distinct ‘languages’ (Chaumeil 1993, Nuckolls 2010, Basso 1985). Distinctions must be actively maintained by behaving in a manner appropriate to one’s own group.

An enigma under an enigma: Unsolved linguistic paradoxes in a sometime continent of hunter-gatherers

Nicholas Evans The Australian National University The special characteristics of Australia pose numerous challenges and puzzling questions for our attempts to understand ancient patterns of linguistic diversification and contact. First, Australia is often described as the only continent solely occupied by hunter-gatherers, and as the only continent exclusively occupied by languages from a single related group. The advantage of this is the way that studies of Australian languages – where it is unusual for languages to surpass a couple of thousand speakers, even in traditional times – can focus on processes of diversification and change in multilingual, small-scale speech communities that are likely to have been widespread for most of our human past. Processes of change in a social semiotic driven by high levels of mulitlingual metalinguistic awareness have produced unusual types of change, such as ‘correspondence mimicry’ (Alpher & Nash 1999) and the promotion of highly marked forms as identity markers (Evans 1998), while the ability of social units to transcend the single language has enabled processes of synthesis that meldsystems from neighbouring languages into complex integrated systems, such as the emergence of an eight-valued subsection systems from two four-valued systems (McConvell 1985a,b). Studying cases like this can give historical linguists a richer picture of likely processes of change under conditions of egalitarian, multilingual contact. A disadvantage of this is that the great similarity of many languages, and the tendency of contact with similar languages to retard major phonological change, can make it difficult to get enough signal the comparative method. Some authors like Dixon 2002 denying its applicability to Australia at all, while others such as Evans 2005, Koch 2004 argue that it can work if applied carefully, and in the second part of the talk I briefly review these debates. A second disadvantage is that the material and cultural similarity of all groups makes it difficult to discover distinctive archaeological signatures which can be used to date the arrival of particular linguistic groups by Wörter und Sachen methods. However, some studies have succeeded in overcoming these problems, such as McConvell & Smith (2003) looking at the (re)colonisation of the Western Desert aided by acacia-seed grinding technology, and studies linking dates for the settlement of offshore islands like Groote Eylandt (Van Egmond 2012) and Bentinck Island (Ulm, Evans, Memmott & Rosendahl 2010). In the third part of the talk I provide an overview of studies of this type. The claim that Australia is a ‘continent of hunter-gatherers’ is, however, a temporally naive one – it only holds for the last 8-9,000 years or so, since rising seas separated it from New Guinea, the other half of the great once-continent of Sahul. In New Guinea, as is well-known, agriculture has flourished for a lengthy period, sustaining societies of much greater scale (> 100,000 in some cases). Moreover, once we take the whole continent of Sahul into account, another problem arises: given that humans seem to have resided on both sides of Sahul for comparable periods (50-60,000 years), why are there such striking differences in the level of deep linguistic diversity – around 40 maximal clades (unrelatable language families plus isolates) in New Guinea, but just one in Australia. In the fourth part of the talk I consider a number of possible answers to this paradox, none currently satisfactory, but likely to implicate different rates of diversification and disparification under different socio-demographic conditions. (Here diversification refers simply to the proliferation of linguistic varieties, while disparification refers to the evolution of greater typological disparity between them, borrowing the diversity vs disparity contrast from work on biodiversity by Maclaurin & Sterelny (2008)). In the fifth and final part of the talk I return to the topic of time, focusing on three questions: (a) the seeming paradox of a single language family for which archaeology and anthropology suggest a 60,000 year time depth, (b) the nature and timing of the Pama-Nyungan spread, which at some point in the past extended a single subgroup over seven-eighths of the continent through mechanisms that can not be linked to conquest or agricultural intensification, and (c) the possibility of constructing a nested series of time-pegs to help resolve some of the paradoxes outlined in the rest of the talk.

Page 5: KNAW Conference on Language Dispersal in Africa, Southeast

Page 5 of 23

Language Diversity, Geomorphological Change and Population Movements in the Sepik-Ramu Basin of Papua New Guinea

William A. Foley University of Sydney The Sepik-Ramu basin region of Papua New Guinea is undoubtedly the linguistically most diverse area on earth. In a land area of under 90,000 square kilometers are spoken languages of nineteen genetically distinct language families, for a language family to area ration of 4,737 square kilometers, roughly a distinct language family for every area around the size of the smallest state in the US, Rhode Island, or a bit less than twice the size of Luxembourg, To put this further in perspective, in the neighboring nation of Timor Leste are spoken languages of two distinct language families, but the land area of this country is 14,870 square kilometers, for a language family to area ratio of 7, 435. But even these statistics do not tell the full story, for the bulk of this area is occupied by a mere four successfully expanding language families; the other fifteen are squeezed into an area of less than a quarter of these 90,000 square kilometers, a staggering language family to area ration of 1500 square kilometers, or a distinct language family for every area a little larger than Hong Kong! Even other hotbeds of linguistic genetic diversity like aboriginal California, the Caucasus, or Amazonia do not begin to approach this. This paper will investigate what are the likely causes of this stupendous linguistic diversity here. It will look at geomorphological changes in the region in the last 8,000 years which have led to massive population displacements and later remigration of new peoples into reclaimed land. It will argue that it is forces like this, as well as cultural values that devalue language, that have led to extensive language shifting, but that the retention of language in more isolated residual zones less effected by population displacements has also contributed to this patchwork of languages in the region.

Human settlement processes in insular SEA and Oceania: The archaeological perspective

Jean-Christophe Galipaud IRD, UMR PALOC Human origins in mainland SEA is very ancient and is the result of several waves of migrations from the north and the west over the millennia. First crossing of large water gaps is attested by 40,000 BP and probably earlier with the successful colonization of Sahul. During the Holocene, intensification of human movements possibly correlated with new economic development concurring with the settlement of all Melanesian and western Polynesian islands up to Tonga and Samoa. This period of human exploration of the Pacific is known as Lapita. Because of its visibility and rapid expansion, but also because of the obvious link between the Lapita diaspora and the introduction of Austronesian languages into remote Oceania, the Lapita period has often been perceived as representing the Austronesian diaspora. Human movements into and between the southeast Asian islands during the same period are less visible and still not well documented but of similar importance to understand the cultural transformations which led to the contemporary cultures.

Islands as opposed to continental mainlands offer good opportunities to explore the evolution of past society and cultural processes. Natural limits favor relative isolation and help archaeologists understand the dynamics of evolution and the processes of transformation of cultures over time.

I will present a synthetic review of our knowledge about the evolution of past society in ISEA and Oceania based on archaeology. I will discuss the limits of archaeological evidence study and reconstruction for our understanding of the past evolutionary process. I will also try to highlight major archaeological evidences of changes which might help us share with other disciplines the past history of the region.

Page 6: KNAW Conference on Language Dispersal in Africa, Southeast

Page 6 of 23

What drives language diversification?

Russell Gray University of Auckland Evolutionary biologists typically distinguish between two types of change – anagenesis and cladogenesis. Anagenesis is change within a lineage and cladogenesis is the splitting of lineages. In this talk I will use our recent work on the spread and diversification of Austronesian languages to investigate factors that influence rates of anagenesis and cladogenesis in this language family.

Latitudinal spread effects in Africa and beyond

Tom Güldemann Humboldt University, Berlin Harald Hammarström Radboud University Nijmegen & Max Planck Institute for Evolutionary Anthropology In the recent past the work on large-scale linguistic distributions across the globe has been intensified considerably. For Africa in particular, Güldemann (1998, 1999, 2003, 2005, 2008, 2010) and Clements and Rialland (2008) have independently made first generalizations on the macro-areal profile of this continent and have come to largely comparable conclusions. Two interesting observations can be made regarding these proposals. First, some of the linguistic areas have a pronounced latitudinal extension. Second, all areas are overall distributed across the continent according to a horizontally rather than vertically layered pattern. In other words, the linguistic macro-areas currently observable in Africa tend to have a latitudinal axis orientation. This will be outlined in the first part of the talk. Güldemann (2008, 2010) has linked this observation with a central idea put forward by Diamond (1999, chapter 7) regarding the different population histories of continental areas. This author argued that in the long term their historical dynamics are decisively determined by the orientation of their geographical axis, in that spreads are hampered along longitudinal axes, but facilitated along latitudinal axes, due to the fact that climatic-ecological factors and therefore conditions for human subsistence generally remain similar in a west-east direction. Assuming that this phenomenon can also have its effect on large subcontinental areas, provided the size of the territory in question is sufficient for this factor to come to bear, Güldemann (2010: 580) proposes that what can be called the “latitudinal spread potential” is also one of the major geographical factors influencing the formation of large-scale contact-induced aggregations of linguistic features. The latitudinal-spread-potential hypothesis is at present hard to test systematically for linguistic contact areas due to the lack of a large-enough inventory of such areas that would find a sufficient amount of scholarly agreement. However, other potential implications of the general idea can be tested more easily. One other hypothesis regarding the latitudinal-spread-potential relates to genealogically defined language groups. In so far as they can also be modeled in an abstract sense as clustered distributions of linguistic isoglosses, it is worth testing whether (all other things being equal) they themselves show a preference for greater west-east rather than north-south extension (Güldemann 2010: 582, cf. Hammarström 2010). We have this specific hypothesis and concluded that language families indeed tend to have a latitudinal/horizontal rather than longitudinal/vertical axis orientation with growing geospatial size, and that this is compatible with the latitudinal spread potential hypothesis. This topic will constitute the second part of the talk.

The Arawakan Diaspora in Amazonia

Michael J. Heckenberger University of Florida The paper summarizes archaeological, ethnohistoric, ethnographic and linguistic patterns in South America associated with the Arawakan diaspora, referring to the most-widely distributed language family in the Americas in 1492. The primary period of ethno-linguistic expansion occurred between ca. 3000-2000 BP, although proto-language is situated a millennia or two earlier. This distribution and antiquity is

Page 7: KNAW Conference on Language Dispersal in Africa, Southeast

Page 7 of 23

comparable to other major linguistic diaspora in the tropics, notably Niger-Congo and Austronesian, as is critical features diagnostic of these groups, settled agricultural life, based primarily on root and tree crops, hierarchical models of social organization, and regional political integration beyond the local community. This paper explores aspects of material culture, ritual performance and built environment, or body languages, across late pre-Columbian groups, which like spoken language reflects basic elements that were widely shared across the diaspora. It concludes with discussion of interdisciplinary approaches and their practical and theoretical implications across the Global South.

Language isolates and farming expansion in Africa

Jean-Marie Hombert Dynamique du langage (CNRS, University of Lyon) Language isolates are languages which cannot be connected to other languages in the linguistic classification of the world’s languages. Their geographical distribution is very biased with approximately one half of them in South America and a quarter of them in New Guinea. The absence of language isolates in a large number of geographical areas can probably be explained by the influence of languages with large population densities which historically replaced a much more diverse linguistic situation. The progressive domestication of food production (agriculture and animal domestication) played a major role in increasing population densities and consequently, ethnic groups who mastered Neolithic technology imposed their language, thus erasing earlier linguistic diversity.

In linguistic research the study of language isolates is determinant for the understanding of language classification, language typology and human migrations. A proper classification of the languages of the world must take into account language isolates as well as larger and well established language families in order to ascertain an accurate picture of language diversity at the family level.

The purpose of this presentation is to examine the localisation of language isolates in Africa with respect to the development of agriculture.

Grammatical signatures of different dispersal mechanisms: A comparison of Africa, Southeast Asia, and South America

Johanna Nichols University of California at Berkeley There are two ways in which languages and typological variables can disperse. Diffusion of structural traits across languages does not entail language shift, leaves genealogical boundaries intact, and can increase the typological diversity within language populations. Spread of languages is accompanied by shift, causes extinction (local or categorical), and decreases diversity within populations. Over time, repeated diffusion episodes can increase complexity, while the sociolinguistics of spreading decreases complexity. This paper surveys several grammatical properties across our three areas and worldwide, and argues that areality in Southeast Asia and South America is a matter of diffusion while that of Africa (and Australia, a clear case brought in for comparison) is a matter of spread and extinction.

Archaeological approaches to cultural and linguistic diversity: A review and position paper

James Steele AHRC CECD, Institute of Archaeology, University College London In this talk I will review some recent archaeological approaches to cultural diversification. The talk will introduce some of the kinds of patterns most typically studied in the archaeological record: hunter-gatherer and farming dispersals (chronology, population branching patterns); subsequent cultural diversification (the appearance, post-dispersal phase, of regional stylistic zones); long-term regional population (in)stability (steady growth versus boom-and-bust); and the dynamics of large-scale integration into state-level political systems. I will illustrate the talk from our own current work on the early population history of the Americas and on sub-Saharan farming dispersals, where appropriate, but I

Page 8: KNAW Conference on Language Dispersal in Africa, Southeast

Page 8 of 23

will also give examples of relevant recent work by other groups. I will also highlight, at relevant points in the talk, questions about the kinds of processes affecting linguistic diversity that may have been operating alongside these archaeological trends, but which elude archaeological detection.

Sociolinguistic typology and the uniformitarian hypothesis

Peter Trudgill University of Agder, Kristiansand, Norway

One of the fundamental bases of modern historical linguistics has been the uniformitarian principle (Labov 1972). This principle states that knowledge of processes that operated in the past can be inferred by observing ongoing processes in the present: language structures in the past were subject to the same constraints as language structures in the present; and the mechanisms of linguistic change that operate around us today are the same as those which operated even in the remote past. This leads to the methodological principle of using the present to explain the past: we can’t try to explain past changes in language by resorting to explanations that would not work for modern linguistic systems.

But, from the point of view of sociolinguistic typology (Trudgill, 2011), the present is not like the past at all, particularly with respect to demography and, as a consequence, social network structure. Increasing populations and mobility have led to more and more language and dialect contact, and to larger and larger language communities, so that languages and dialects spoken in small, low-contact, isolated communities with tightly-knit social networks and large amounts of communally shared information are becoming less and less common. Labov himself, in his discussion of the uniformitarian principle, warns that we must be “wary of extrapolating backward in time to neolithic preurban societies”: the methodology of using the present to explain the past might be less useful the further back in time we go (Labov 1994: 23). Given that most of the linguistic past took place in pre-neolithic or neolithic societies - human languages were spoken in neolithic or pre-neolithic societies for at least 95% of their history - where does that leave the uniformitarian principle?

Cultural phylogenetics in lowland South America

Robert S. Walker University of Missouri Comparative phylogenetic analyses based on language are useful for reconstructing phylogeographies and cultural evolution of recent expansions of humans around the world. It is an exciting time for phylogenetic comparative studies in lowland South America as more and larger cultural datasets emerge that include, for example, patterns of warfare, postmarital residence, paternity beliefs, kin coresidence, and population dynamics. Phylogenetic methods are now being applied to more lowland language families, and at deeper time depths, that allow us to track the evolutionary histories of cultural variation.

The phylogenetic comparative method applied to cultural evolution is a two-step process first requiring as input some phylogenetic hypothesis about the historical relationships among cultures. For the creation of lowland phylogenies, so far we have available cognate sets in basic vocabulary (Arawak, Carib, and Pano language families), computerized lexical analysis of 40-item basic vocabulary lists (Automated Similarity Judgment Program), and structural language elements (World Atlas of Language Structures). This talk addresses some recent attempts at using these data for investigating internal language family classifications and deeper phylogenetic relationships among the 6 major lowland South American language families, which can then be compared against genetic phylogenies.

With linguistic phylogeny in hand, the second step is to reconstruct the evolution of a cultural trait over the phylogeny to infer ancestral state and transition rate parameters using a model of trait evolution. Several traits can be shown to have deep evolutionary histories that extend back to last common ancestors of several large language families (e.g., a tentative Je/Pano/Carib/Tupi clade), including tribal warfare, swidden horticulture, matrilocal postmarital residence, and partible paternity beliefs (recognition of multiple fathers). Conventional wisdom suggests that cultural change is often fast and innovative, and it has been claimed that rapid rates of cultural adaptation are the most distinctive of all human characteristics. Phylogenetic analyses are useful for quantifying rates of cultural change to make

Page 9: KNAW Conference on Language Dispersal in Africa, Southeast

Page 9 of 23

valid cross-cultural comparisons of cultural dynamics over relatively deep periods of time. Estimated rates of cultural change at the macro-level across those language expansions so far examined are on the order of only a few changes per 10,000 years and indicate a conservative nature to much of human culture. Relatively slow cultural transition rates highlight the importance of using comparative phylogenetic methods in the first place to study human variation for the simple reason that many cross-cultural similarities might often arise from shared common ancestry even over considerable periods of time.

Page 10: KNAW Conference on Language Dispersal in Africa, Southeast

Page 10 of 23

Workshop on Quantitative Approaches to Areal Typology

Abstracts

Areal Diachronies

Balthasar Bickel University of Zürich Some regions in the world are known to have undergone a large number of language contact events over substantial periods of time (e.g. Eurasia). Over time, repeated events of this kind can be expected to differentiate between biases in diachronic development inside vs. outside a geographical region (e.g. different diachronic trends in Eurasia vs. outside Eurasia). I present a method that estimates signals for such diachronic biases from their expected synchronic results: if a structure S outnumbers non-S significantly in a language family, a change towards S in this family was more likely than a change away from it (either because S was there in the protolanguage and then hardly ever got lost, or because S was not there and then it was innovated early or often in the family). If there is no significant synchronic preference, no signal can be inferred because in this case, there was either no diachronic bias towards any structure, or the difference in biases was too small to leave a signal, or the family is too young to allow a signal. Using extrapolation methods, signals for diachronic biases can also be estimated for isolates and small families. This method (which I call the Family Bias Method) allows detection of signals of area formation over time without incurring the independent sampling problems that have hampered typological research in the past.

In search of areal effects in semantic typology: Reference frames in Mesoamerica

Juergen Bohnemeyer1, Jesse Lovegren1, Katharine T. Donelson1, Elena Benedicto2, Alyson Eggleston2,3, Alejandra Capistrán Garza4, Néstor Hernández Green5, María de Jesús Selene, Hernández Gómez6, Carolyn O'Meara6, Enrique Palancar7, Gabriela Pérez Báez8, Gilles Polian9, Rodrigo Romero Méndez6, Randi Tucker1 1=The State University of New York, Buffalo, 2=Purdue University, 3= East Carolina University, 4=Universidad Autónoma Metropolitana, 5=Centro de Investigaciones y Estudios Superiores en Antropología Social, Distrito Federal, 6=Universidad Nacional Autónoma de México, 7=University of Surrey, 8=Smithsonian Institution, 9=Centro de Investigaciones y Estudios Superiores en Antropología Social, Sureste We examine whether practices of using language may be (i) contact-diffused and (ii) areal features. The domain of our study is that of spatial frames of reference (FoRs), conceptual coordinate systems used to locate and orient entities and motion paths. The reference frame types available to a speech community are not generally constrained by the grammar or lexicon of the language, but are a part of the community’s practices of language use. Since these practices vary with language (Majid et al. 2004; Levinson & Wilkins (eds.) 2006; O’Meara & Pérez Báez (eds.) 2011), it stands to reason that they may be contact-diffused, although it has to our knowledge never been directly demonstrated that this is the case. Our test case for areality is the Mesoamerican (MA) sprachbund (Kaufman 1973; Campbell 1979; Campbell, Kaufman, & Smith-Stark 1986; Smith-Stark 1994). We collected data on FoR use in discourse from six indigenous MA languages (Isthmus Zapotec, Mixe, Otomí, Tarascan, Tseltal and Yucatec Maya), two indigenous languages spoken north and south of the MA area – Seri and Sumu-Mayangna – and three varieties of Spanish (Mexican, Nicaraguan and Peninsular Spanish), using five dyads of adult native speakers per variety. Participants matched four sets of twelve pictures featuring a ball and a chair in varying spatial configurations. We coded descriptions of the location of the ball for eight strategies (topological, object-centered intrinsic, egocentric intrinsic, relative, intrinsic-relative ambiguity, geocentric, geocentric-vertical, ambiguous-vertical). In a given description, a participant could use any or all of the eight strategies. We also estimated, on a three-point scale, each participant's education level, level of literacy,

Page 11: KNAW Conference on Language Dispersal in Africa, Southeast

Page 11 of 23

and level of Spanish usage. We computed for each dyad an eight-dimensional vector assigning to each strategy the frequency with which the dyad used it. Interpreting these vectors as points in an eight-dimensional space, we calculated their Manhattan distances as a measure of the similarity between them. Figure 1 shows an MDS plot of the left-triangular distance matrix. The first dimension of this plot correlates strongly with the geocentric scores (Spearman’s Rho[df=53] 0.95; p<.001; cf. Figure 2) and weakly negatively with the relative scores (Spearman’s Rho[df=53] -0.80; p<.001), while the second dimension correlates weakly with the topological scores (Spearman’s Rho[df=53] 0.79; p<.001). We then modeled the geocentric and relative FoR scores of just the speakers of the indigenous languages as a function of education level, literacy level, L2-Spanish usage level, and areal-linguistic affiliation (MA vs. non-MA). The generalized linear mixed-effects model included nested random intercepts for language and participant ID. The fitted geocentric model revealed L2-Spanish use and literacy as significant factors (Wald-p .01 each) and the relative model only L2-Spanish use (Wald-p .02). We take this as evidence that the use of relative FoRs does indeed diffuse through contact with Spanish. At the same time, our GLMMs failed to find evidence of a sprachbund effect (Wald-p .79 for the geocentric and .16 for the relative model). A hierarchical cluster analysis of the distance matrix using an agglomerative clustering method in R confirmed this: clusters showed some coherence for the individual languages, including for the three varieties of Spanish as a group, but not for the MA sprachbund (Figure 3). The coefficient of the cluster analysis was 0.74. A comparison of the mean distances among the participants sharing a particular value of any of the predictor variables shows that membership in the MA area aligns with no more than loose coherence in FoR use (Table 1). We do not interpret this negative finding as evidence against areal diffusion of practices of language use in general. The MA sprachbund is nowadays largely what one might consider a ‘fossilized’ sprachbund, as contact among speakers of indigenous languages has been greatly reduced after the Conquista and now proceeds to a significant extent through Spanish where it occurs. By hypothesis, a fossilized area would only reflect the most time-stable among the original areal features.

Probabilistic generative models of language contact

Will Chang University of California at Berkeley Ever since a model from population genetics was used on typological characters to illuminate the prehistory of Southeast Asia and the Pacific [1], it became clear that such probabilistic models could be useful for a host of other linguistic problems as well. In this talk I will focus on the nuts and bolts of three such models, each an incremental improvement on, or a variant inspired by, the STRUCTURE model [2] that was used in the aforementioned study.

(1) I have applied the STRUCTURE model to data from the SAPhon database [3], consisting of the phonological inventories of 350+ South American languages. I will discuss three enhancements to the model that were necessary for good results: (i) a realistic enough prior for the feature frequencies in each ancestral population; (ii) a hierarchical Dirichlet prior [4] for language ancestries, so that the number of ancestral populations can be inferred automatically from the data, rather than set by the analyst; (iii) two methods for summarizing and visualizing the resulting posterior samples.

(2) The STRUCTURE model presupposes that all states of a character are equally amenable to borrowing. This assumption is awkward when it comes to modeling binary characters that represent the presence or absence of a feature. Lev Michael and I have constructed a rudimentary Relaxed Admixture Model to model features that one language can influence another to gain, but not to lose. When applied to the SAPhon database, this model proved useful for detecting the effects of relatively mild and recent contact.

(3) Rather than construct clusters of languages, as STRUCTURE essentially does, one can instead construct clusters of features — grouping together features that have similar distributions in the languages. This is especially effective when features are numerous and non-homoplastic. I have taken POLLEX, a comparative word list of Polynesian languages with over 4000 etyma [5], and formed clusters with these etyma to induce isogloss bundles which reveal episodes of contact in the prehistory of these languages.

Page 12: KNAW Conference on Language Dispersal in Africa, Southeast

Page 12 of 23

Quantifying and typology

Mark Donohue The Australian National University Typology is a subdiscipline of linguistics with a confused set of aims, but noble purpose. The quantification of typology allows for a more objective assessment of typological variables, but to what end? In this talk I address two questions that arise in discussions of typology, sampling of languages and sampling of linguistic features. I argue that care is needed when constructing a research question involving either of these considerations. When the goal of typological work is to compare _languages_, a large set of features, allowing for multiple dimensions of comparison, is essential, and that comparing along a single dimension does not lead to easy interpretation. When the goal of typological work is to compare _features_ a large sample of languages is required, and consideration of the sample is essential (to avoid Galton's problem), but that the typologically desirable 'balanced sample' is not achievable with a single selection of languages. Again, multidimensional comparison is required, examining different subsets of the sample (/data) to find different kinds of results, is essential. I illustrate the discussion with data largely drawn from the World Phonotactic Database (http://phonotactics.anu.edu.au).

Evolution and areal dynamics: Pronominal politeness in and out of the Carolingian Sprachbund

Michael Dunn and Kate Bellamy Max Planck Research Group Evolutionary Processes in Language and Culture Politeness, in its various guises, is a basic constituent of the maintenance of human social order (Brown and Levinson 1987). A common locus of the linguistic encoding of politeness is in pronominal systems. These systems are particularly widespread – although by no means ubiquitous or homogeneous – across the Indo-European (I-E) language family (Helmbrecht 2011). Within the Indo-European languages of Western Europe there is an area of linguistic and social interaction associated, at least geographically, with the Carolingian Empire. The languages of this area, representing several branches of I-E, are referred to as Standard Average European (SAE; Whorf 1956; Haspelmath 1998, 2001) or the Charlemagne Sprachbund (the SAE core; Van der Auwera 1998). Pronominal politeness systems are attested in many of the languages most closely associated with the Sprachbund, although the structures of these systems vary (Helmbrecht 2003; Mladenova 2010). Some SAE languages lack polite pronouns (e.g. English), others possess binary distinctions (e.g. French) and some tertiary distinctions (e.g. Rumanian). Such distinctions are also found in I-E languages spoken in areas outside of the Sprachbund.

Politeness is a social phenomenon so we can expect pronominal politeness systems to be highly susceptible to areal norms of social interaction. We test whether the SAE linguistic area correlates with evolutionary processes acting within politeness systems. A Bayesian phylogenetic analysis shows that the rates of diversification of politeness systems within the I-E language family can be best explained by modeling two distinct evolutionary regimes, one inside the Carolingian sphere and the other outside it. While it is known that rate of structural change of language differs within families (Dunn et al. 2011), factors determining these rates have never been quantified or tested. This is a novel but realistic approach to the treatment of areal linguistic similarities. Similarities between distantly related languages sharing an environment can be accounted for with an explicit model of parallel evolutionary change, rather than by borrowing. While a word from one language can be taken over directly by another, a linguistic structure has to be recreated in the target language out of the available linguistic resources. Most so-called structural borrowing in languages might be better thought of in this way.

Page 13: KNAW Conference on Language Dispersal in Africa, Southeast

Page 13 of 23

Quantifying Geographical Determinants of Large-Scale Distributions of Linguistic Features

Tom Güldemann Humboldt University, Berlin

Harald Hammarström Radboud University Nijmegen & Max Planck Institute for Evolutionary Anthropology In the recent past the work on large-scale linguistic distributions across the globe has been intensified considerably. For Africa in particular, Güldemann (1998, 1999, 2003, 2005, 2008, 2010) and Clements & Rialland (2008) have independently made first generalizations on the macro-areal profile of this continent and found that linguistic areas are distributed across the continent according to a horizontally rather than vertically layered pattern. In addition to a horizontal bias, other factors that may influence the geographical shape of large linguistic areas are coastlines and mountain ranges. Complete and detailed data is now available on elevation and coastlines (through the SRTM http://dds.cr.usgs.gov/srtm/) and of climatic conditions (through the Köppen-Geiger map http://koeppen-geiger.vu-wien.ac.at/present.htm). We will quantitatively test the hypothesis that coastlines, mountain ranges and a general horizontal bias account for the geospatial shapes of language feature isoglosses, using WALS and other available linguistic datasets.

Automatic Detection of Borrowings in Lexicostatistic Datasets: A Workflow for Automatic Linguistic Reconstruction

Johann-Mattis List, Steven Moran, and Jelena Prokić Philipps University Marburg In historical linguistics, two models are traditionally used to describe language history: the family tree model (Schleicher 1853) and the wave theory (Schmidt 1872). Although both models are often treated as if they were opposing theories of language history, the models are in fact complementary; one emphasizes the temporal aspects of language evolution and the other emphasizes its spatial aspects. Although the idea to combine both models within a single framework have been put forward (Schuchard 1900[1870], Southworth 1964), it was not until recently that phylogenetic networks were proposed as a formal model to capture both the temporal (vertical) and the spatial (horizontal) aspects of language history. A phylogenetic network is hereby understood as a network containing an explicit time dimension and explicit vertical edges. It consists of a reference tree (backbone) and lateral edges indicating the degree of contact between the taxa or also between internal nodes (ancestor languages) of the tree. In recent times, many new methods that automatize major steps of the comparative method have been proposed, ranging from pairwise and multiple phonetic alignment (Kondrak 2000, Prokić et al. 2009, List 2012a), via automatic cognate detection (Turchin et al. 2010, Steiner et al. 2011, List 2012b), up to automatic phylogenetic reconstruction (Gray & Atkinson 2003, Atkinson & Gray 2006). The problems of phylogenetic network reconstruction and automatic borrowing detection, however, have only rarely been addressed (Nelson-Sathi et al. 2011).

In this talk, we present a new, general workflow for automatic linguistic reconstruction. In contrast to previous approaches, we explicitly address the problem of automatic borrowing detection by employing a modified version of the method for the automatic reconstruction of phylogenetic networks by Nelson-Sathi et al. (2011). This method infers borrowings in sets of cognate words proposed for a given set of languages by searching for those cognate words whose distribution is not in concordance with a given reference tree. We modified the original method in such a way that it now is possible to map the lateral connections inferred by the method onto geographical maps. As a result, the areal component of language history, which is largely neglected in all pure tree-based applications, can now be explicitly modeled. It enables the researches to detect linguistic areas of an intensive language contact, thus providing a useful tool for quantitative approaches in areal linguistics.

We illustrate our workflow with the Dogon comparative lexical spreadsheet from the Dogon Languages Project (Heath et al. 2012), a freely available dataset that includes 20 Dogon languages and several dialects. Dogon languages are spoken in Mali and have until recently been poorly described. The current Dogon linguistic situation is unclear, as is the internal structure of the Dogon family at its lower branches.

Page 14: KNAW Conference on Language Dispersal in Africa, Southeast

Page 14 of 23

The position of Dogon relative to other African languages families is also unclear due to its unusual typological characters in a sea of West African languages (e.g. Dogon does not have noun classes or the same word order as languages in this area). The Dogon Languages Project has also begun genealogical studies with the comparative data, thus providing an ‘old-fashioned’ analysis with which to compare our statistical approaches.

Two-place (in)transitives in European languages: A quantitative typological study of verb classes

Sergey Say Institute for Linguistic Research of the Russian Academy of Sciences, St. Peterburg This study is based on an on-going project devoted to the quantitative typological study of verb classes. The major issue of the project is to assess the degree of cross-linguistic (ir)regularity in distributing verbs among polyvalent valency classes. The study is based on a questionnaire that includes 130 predicative meanings (each is given in a particular context in order to avoid polysemy effects). E.g. the Lithuanian sentence in (1) allows one to classify the Lithuanian equivalent of ‘be afraid’ as belonging to the Nominative + Genitive class, along with equivalents of such verbs as ‘wait’, ‘avoid’, ‘look for’ etc.

(1) Lithuanian Petr-as bij-o šun-s. Peter-NOM.SG is.afraid-PRS.3 dog.GEN.SG ‘Peter is afraid of the dog’.

The project focuses on semantic factors that account for attested deviations from canonical transitivity (the 130 verbal meanings were chosen based on a pilot study, in which verbal meanings that are particularly prone to such deviations were obtained). It is explored whether two-place intransitive verb classes are language-specific or cross-linguistically recurrent, whether there is a hierarchy of verbs that are more and less prone to intransitive patterns of particular types, etc. This paper, however, focuses on the areal-typological dimension of the project, with the emphasis put on methodological and quantitative issues. The area under study is Europe in the sense that was exploited in the former Eurotyp project. In my talk I am going to discuss quantitative data from 13 European languages (representing 6 families): Basque, French, German, Lithuanian, Latvian, Russian, Estonian, Ingric Finnish, Kalmyk, Bashkir, Bagwalal, Tsaxur and Lezgian (other European languages for which data have been obtained are Armenian, Albanian, English, Ancient Greek, Modern Greek, Irish, Kalderash Romani and Spanish, but these data need statistical analysis). The talk addresses quantitative techniques and results including the following. 1. The ratio of non-canonical verbs shows a good deal of cross-linguistic variation (cf. Appendix 1) with

some genealogical and especially areal patterning. In particular, SAE languages show higher preponderance of transitive structures than peripheral European languages, which widens some previous findings wrt to individual groups of verbal meanings [Bossong 1998, Haspelmath 2001].

2. Things get slightly more complicated once one tries to establish regularities in the ways in which various languages distribute individual verbs between transitive and intransitive classes. Given the data for 130 verbs in two languages one can count the distance between the two languages with the help of a Hamming-type measure (the distance is, roughly, the ratio of meanings that are expressed by transitive structures in one language and by intransitive structures, in the other). Thusly obtained distances between languages can be plotted in a Neighbor Net dendrogram (cf. Appendix 2, SplitsTree software was used). This dendrogram seems to capture strong genealogically-determined similarities between languages (cf. e.g. Daghestanian languages), as well as cases of areal convergence (cf. the data for Basque and French).

3. A well-known problem in studying two-place intransitive constructions is that they cannot be directly identified across languages, since the very coding properties (case grams, pre- and postpositions, indexing slots, etc.) are language-particular phenomena, even if they bear identical grammatical labels (e.g. ‘dative’ or ‘comitative’). One of the main objectives of the study is to propose a method of measuring cross-language (dis)similarity in their distributions of verbs between various classes without illegitimately identifying the classes as such. An entropy-based mathematical technique is used in order to circumvent this difficulty: it calculates the degree of predictability of coding a given predicative meaning in L1 based on correspondence between valency classes in L1 and L2. This

Page 15: KNAW Conference on Language Dispersal in Africa, Southeast

Page 15 of 23

technique allows one to build an (asymmetric) L1>L2 predictability matrix for the languages under study (see Appendix 3). In my talk I am going to discuss the methodological issues related to measuring similarity between languages with the help of various techniques, as well as to interpret the linguistic findings of the study. In particular, it will be stressed that the analysis that takes into account minor valency classes yields results that are significantly different from those obtained in the analysis where all polyvalent intransitive classes are collapsed together.

A ‘matreshka’-approach to testing the significance of micro-areas

Björn Wiemer Johannes Gutenberg University Mainz This paper takes up a methodological and an empirical issue. We highlight some aspects of data preparation which seem to have so far been neglected in the study of contact-induced convergence in small-scale areas both by dialectologists and by areal linguists. At the same time, the paper re-considers whether it really makes not much sense to compare macro-areas to micro-areas (as suggested in, e.g., Dahl 2001: 1463 and Wälchli 2012: 235). As for methodology, consider first the main goal of typology: which structural properties and their co-variation are possible (and why) in natural languages? On the one hand, insofar as answers to this question could be skewed by areal and genealogical biases, sampling procedures have to avoid such biases. On the other hand, areal linguistics to a large extent arose from the insight that representative samples must account for the fact that features have been clustering in certain areas due to uneven diffusion caused by contact and/or common ‘heritage’ (e.g., Bisang 2006a-b; Wiemer/Wälchli 2012: 6-8). Regardless of how large or small they are, zones of convergence become ‘visible’ only on the background of some larger geographical units (Wiemer 2004). This does not solve the problem of defining areas on the basis of features or other issues that have been raised against the notion of ‘area’ (cf., e.g., Aikhenvald/Dixon 2001; Dahl 2001; Stolz 2002, 2006; Bickel/Nichols 2006; Bisang 2006a-b; Campbell 2006). But it allows us to see parallels with no less notorious problems inherent to dialect geography (cf. Pšeničnova 1996, 2008; Seiler 2005; Szmrecsanyi 2011, to appear; Glaser, to appear). One of them is the issue that, irrespective of how closely the varieties in question are related (in genealogical or geographic terms), diffusion of features does not only cause convergence with previously less similar varieties, it at the same time increases divergence within an originally more homogeneous continuum. It turns out that dialectology has neglected the relation between convergence and divergence to a similar extent as has areal linguistics, although the size of the ‘units’ under analysis differs a lot. The reasons for this neglect are not only empirical problems in establishing and documenting relative chronologies of changes (or their correlation with periods of contact), but also conceptual ones (as for historical-comparative methods cf. Holzer 1998 and Campbell 2006: 18-21, as for dialectology cf. Wiemer/Erker, to appear). Consequently, if our research focuses not on the aforementioned global aim of typology, but on a better understanding of (i) processes of diffusion (Bisang 2004, 2006b: 76; Campbell 2006), (ii) how assumed areas become internally more heterogeneous and (iii) to which extent they crosscut with dialect continua of the languages (or language groups) involved, more cross-fertilization with insights and methods from dialect geography becomes possible and desirable. Moreover, micro-areal research advocated for here has the advantage that (a) one can zoom in onto a small region and concentrate on multifactorial studies of more fine-grained features; this, in turn, (b) allows discerning different hotbeds. The diffusion areas around the latter ones often ‘coalesce’, so that on a typological coarse-grained level (both in terms of feature resolution and of geographical space) one can easily fall victim to the illusion that the given features diffused in a unitary manner. The study of micro-areal processes leading to convergence and divergence has the further advantage that we can better test hypotheses concerning borrowability (cf. Matras 2007; Thomason 2007) of different structural features as well as the (in)stability of common features – arisen either from common heritage or from parallel development – exactly because the involved varieties are often closely related, structurally similar and have other ‘close relatives’ outside a small area chosen. The empirical case on which these assumptions will be demonstrated and quantified is the zone of overlap between Slavic and Baltic (see map below), called ‘Baltic-Slavic Contact Zone’ (BSCZ). It can be captured as part of a larger cline (from NE to SW) on a whole bunch of properties, only partially accounted for in studies devoted to the Circum Baltic Area (Dahl/Koptjevskaja-Tamm 2001). The fact that the BSCZ is defined on the basis of the geographic overlap

Page 16: KNAW Conference on Language Dispersal in Africa, Southeast

Page 16 of 23

of two genealogically close, yet distinct groups within IE allows not to preconceive which properties should be salient with regard either to (North) Slavic and Baltic in general, or to any variety in larger areas encircling this well-defined zone from the perspective of Eurasia. Thus, properties found within the BSCZ are systematically compared with their occurrence in successively larger areas encircling the BSCZ: [[[[[[BSCZ < [Eastern part of CBA] < CBA] < Eastern Europe] < Eurasia ?] < ...] (‘matreshka-principle’). This helps gaining a quantifiable picture of their areal significance, but also of divergence within the area itself. The properties chosen range from suprasegmental phonetics via morphological categories and grams up to the coding of arguments and other syntactic issues. Finally, I will show how the catalogue of properties raised in this way can be made subject to quantificational methods.

Page 17: KNAW Conference on Language Dispersal in Africa, Southeast

Page 17 of 23

Presentation South American Component ERC ‘Traces of Contact’

Abstracts

Introduction to the ERC Traces of Contact Project, some non-South American results and general considerations

Pieter Muysken Radboud University Nijmegen The volume presented results from the ERC (European Research Council) Advanced Grant project ‘Traces of Contact’, coordinated by Pieter Muysken at Radboud University Nijmegen from 2009 to 2013, with additional support from the Royal Netherlands Academy of Sciences (KNAW) and the Netherlands Organization for Research (NWO). It reports on sustained research on a single theme: language change through contact, in a setting so far little explored in this connection.

It aims to establish criteria by which results from language contact studies can be used to strengthen the field of historical linguistics. It does so by applying the scenario model for language contact studies to a number of concrete settings, which differ widely in their level of aggregation and time depth: the languages of the Amazonian fringe in South America, the complex multilingual setting of the Republic of Suriname, the multilingual interaction of immigrant groups in the Netherlands, and two groups of multilingual individuals. New methods from structural phylogenetics are employed, and the same linguistic variables (TMA and evidentiality marking, argument realization) will be studied in the various projects. In the various projects, use will be made from a shared questionnaire, so that comparable data can be gathered. By applying the scenario model at various levels of aggregation, a more principled link between language contact studies and historical linguistics can be established.

Traces of Contact results from South America, with illustrations from the Isthmo-Colombian area

Loretta M. O’Connor Radboud University Nijmegen In this talk I will present the edited volume we are compiling with results of the South American subproject of the larger Traces of Contact project, highlighting main contributions from various authors, and I will then present illustrations from my own research on contact-induced change in the Isthmo-Colombian area.

The book consists of four main sections, as follows:

Page 18: KNAW Conference on Language Dispersal in Africa, Southeast

Page 18 of 23

The Native Languages of South America. Origins, Development, Typology, edited by Loretta O’Connor & Pieter Muysken I. Introduction to South America Introduction Editors Human migrations, dispersals, and contacts Loretta O’Connor, Vishnupraya Kolipakam Lexical relationships in basic vocabulary Harald Hammarström II. Case studies in Contact The Isthmo-Colombian area Loretta O’Connor The Andean foothills Rik van Gijn The Andean matrix Pieter Muysken, Simon van de Kerke The Arawak matrix Love Eriksen, Swintha Danielsen The Tupi expansion Love Eriksen, Ana Vilacy Galucio III. Comparative perspectives on linguistic structures TAME: Desiderative marking Neele Müller Verbal argument marking patterns Joshua Birchall The Noun Phrase: Demonstratives Olga Krasnoukhova Subordination strategies: Nominalizations Rik van Gijn IV. Major Findings and Conclusions Overview and conclusions Joint work My own chapter in the volume examines the role of structural features as indicators of nested levels of social history in a specific geographic region. Languages of the Isthmo-Colombian area are spoken across the land bridge that connects the American continents and along the northwest coast of the southern landmass. Once characterized by scholars as an ‘Intermediate Area’ between powerful civilizations north and south, the region is now recognized as a place where technologies swept through but human populations were relatively sedentary and remained intact. For this study, structural features in fourteen languages of the region were coded as stable or unstable, using a composite ranking of relative stability, and as template or content, using a functional metric. Patterns of similarity indicate that the set of features defined as ‘content’ that involve choosing what to encode in a given structural feature, is more successful than any other set at replicating areal patterns. The analysis suggests that structural features, like lexical items, may be divisible into types which are more and less susceptible to conscious manipulation by speakers, and that their role must be interpreted within a specific socio-historical context.

The Andean Matrix

Pieter Muysken1 and Simon van de Kerke,2,1 with Harald Hammarström1,3 1=Radboud University Nijmegen; 2=Leiden University; 3=Max Planck Institute for Evolutionary Anthropology. Over the last three thousand years the Central Andean area has been the stage for the rise and fall of different civilizations. Periods of centralization of power were followed by periods of regionalization, but on the whole the direction was towards increasing state control over an ever growing territory. The summit was reached when the Inca Empire controlled an area that ran from Ecuador to Bolivia, just before the invasion of the Spaniards in 1532 AD. Inca policy did not intend to suppress local cultural and language complexes, but to overlay them with the state culture and imperial language: Quechua. This process was so successful that 500 years later the Quechua language family, in its many local varieties, is still spoken in this area, next to one other important other language family that was able to resist the pressure of Spanish: Aymara.

The area where both language families emerged lies in central Peru from where they began to spread in first millennium of our era. There are many structural and lexical similarities between the two languages and the question of whether genetic relatedness or language contact may explain this similarity has been the subject of a sometimes heated debate between linguists over the last 40 years. Thorough lexical comparisons within, but also between, varieties of the two languages are suggestive of a language contact scenario, but the analyses based on morphological and syntactic features never reached the same

Page 19: KNAW Conference on Language Dispersal in Africa, Southeast

Page 19 of 23

level of theoretical depth. The current paper aims at a more refined comparison of the two language families in this respect, within a more encompassing analysis of the dialectal variation within the Quechua family and its internal constitution, and taking into account the other relevant languages spoken in the area.

The Andean Foothills

Rik van Gijn University of Zürich The eastern slopes or ‘foothills’ of the Andean mountain range are among the linguistically most diverse areas of the continent. Their position in between three culture areas that have been expanding over the last millennium (Arawakan and Tupian in the Amazon area to the east, Quechuan-Aymaran in the Andean area to the west) raises the question as to how the inhabitants of the foothills have dealt with these cultural expansions and, more particularly, what imprint, if any, this has made on the languages that they speak. Lists of large-scale areal or regional features have been proposed for both the Andean area and the Amazonian area. In this contribution, I look at how these features take shape in the grammars of foothill languages.

Verbal Argument Marking Patterns

Joshua Birchall Radboud University Nijmegen As more descriptive data on South American languages have become available, it has become apparent that the continent hosts an incredible diversity of linguistic structures. Various proposals have been made attempting to identify the typological features characteristic of certain regions of the continent, especially regarding the way that these languages mark arguments within the clause. However, the distribution of these properties over the continent as a whole is rarely taken into account when discussing such characteristic features. This study explores the distribution of verbal argument marking patterns systematically using a large-scale comparative database composed of 65 South American languages. The structural features under consideration include the presence and alignment of argument markers, the locus of marking, and the grammatical categories indexed through marking, as well as other alignment related patterns such as split intransitivity, hierarchical marking and inverse marking. This chapter addresses the question of whether there are specific geographic patterns in the distribution of the linguistic structures used to index core arguments on the verb across these languages, and explores possible explanations of these patterns that take linguistic, historical and cultural factors into account.

Page 20: KNAW Conference on Language Dispersal in Africa, Southeast

Page 20 of 23

Workshop on Ritual and Secret Languages Abstracts

Ritual and secret languages: An overview

Peter Bakker Aarhus University Ritual languages are languages used in religious and spiritual ceremonies. They are often archaic or even extinct languages, and regularly also a foreign language rather than the colloquial language. The ritual may be an older language because a holy text has been written in that language (e.g. Hebrew among Jews, Classical Arabic among Muslim) and the text is considered authoritative as the word of God. In other cases ancestral languages (Kromanti and Papa in Suriname), an older variety (Maroon Spirit Possession language) are used because those are considered languages the spirits can understand. A society may also have shifted to a different language after the establishment of a language for religious services (Coptic among Egyptian Christians), or evolved (Church Sranan in Suriname), or the religious language is imported from elsewhere (Latin in Catholic services). Finally, the language may have been created (Kallawaya). The reason for not using the colloquial language tends to be a combination of tradition, the idea of a sacred or unchangeable language (God’s own word), the necessity of using a language that the ancestors understand, or the protection of sacred knowledge. Often the language may be inaccessible to the participants to the services, bordering secret languages. In some religions people ‘speak in tongues’ (glossolalia): believers use an unknown language to communicate with their God. However, these languages have never been identified as an existing language, and analysis of the patterns of recurrence of all cases studies show that they never can be, on linguistic grounds.

Secret languages proper can be written or oral. Written languages can be created in order to communicate messages undecipherable for uninitiated audiences. I will focus only on oral secret languages. These are used in order to conceal the content of messages from certain interlocutors: children in a family, business competitors, enemies, authorities (e.g. in drug trade or crime) or eavesdroppers (e.g. civilians listening to police or military radio). There are three basic strategies in which secret languages can be created: one can make use of a foreign or different language (e.g. Dutch parents using French with each other in presence of their children, or Breton rather than French in shift periods). Second, people can also import content words, or a selection of content words, into the everyday speech of the society (e.g. Angloromani, where selected Romani words make a message only understandable for some). Finally, people can distort words –usually only content words– of the colloquial language in order to make the language unintelligible, for example by reversing roots, adding syllables and similar procedures. This is not an exhaustive list: upside-down Warlpiri, for instance, works with opposite meanings rather than forms. Such secret languages have an additional function of stimulating a group feeling among those who know the code and those who don’t. Language of the latter type may also be created as a form of recreation, the so-called play languages. Finally I will also discuss some generalizations about functional aspects, including acquisition.

Shouting at masks: sigi so in Dogon culture

Wouter van Beek African Studies Centre & Tilburg University Dogon masquerades are famous, and their masks have become collectors’ items for over almost a century already. As such, they have become the icons of the whole ritual complex that colours Dogon culture, that of the masks and the sigi, i.e. the rituals surrounding life and death. This ritual corpus includes burial, the first funeral, the mask festival of the second funeral and finally, the sigi ritual, performed each 60 years.

Page 21: KNAW Conference on Language Dispersal in Africa, Southeast

Page 21 of 23

One major common feature of these complicated and spectacular rituals is the mask language, the sigi so, the ‘language of sigi, reserved for men who are initiated into the masks. The language is quite distinct from the standard Dogon language, and spoken only at ceremonial occasions.

This contribution zooms in on the ritual position of this language which is not spoken by the masks, but used to address them. The core of the language is a body of texts highlighting the myths surrounding these two major rituals, the masks themselves and the sigi, and the transmission of both the intimate knowledge of the language and the memorization of the texts is part of the rituals themselves.

Kumanti: A ritual/spirit language among Suriname’s Ndyuka maroons

Robert Borges Radboud University Nijmegen This paper will address three main questions with respect to Kumanti, a ritual/spirit language known by maroon groups throughout Suriname and the Caribbean. However, I address Kumanti’s circumstance in only one such group, the Ndyuka (a.k.a. Okanisi) maroons of Eastern Suriname. According to Thoden van Velzen (2010 p.c.), the majority of Ndyuka have at least a passive knowledge of Kumanti, even though this may be limited to just a few words. I will (1) address ways in which Kumanti is acquired as well as reasons why those who choose to learn it will do so. I will (2) discuss the structural, lexical, and pragmatic similarities and differences between Kumanti and the everyday language of the Ndyuka. Lastly, I will (3) refute the idea (Smith 1987:88; Pakosie 2000; Devonish 2005) that Kumanti is a West-African language transplanted to the Caribbean by African born slaves, then transmitted and preserved by subsequent generations of maroons.

Hide and Seek in the Deer’s Trap: Language Concealment and Linguistic Camouflage in Timor Leste

Aone van Engelenhoven Leiden University This contribution discusses the original language of East Timor’s easternmost sub district Tutuala. Its speakers do not have a name for their language nor for their own etnnic identity. Nowadays, all inhabitants speak Fataluku, ‘the correct speech’, a non-Austronesian language of which the introduction in the region was only completed in the 60-ies of the 20th Century.

Their original language became known in the literature through its Fataluku exonym Lovaia epulu ‘language of the Civet Cat’s Oil’ (Lovaia epulu). The location called Lovaia later became known as the ‘Deer’s Trap’ (Porlamanu) of which the inhabitants were referred to in Fataluku as Makuva ‘idiots’. This paper intends to provide an anthropological linguistic insight in the strategies with which speakers attempt to safeguard their language from extinction in a society that only acknowledges one ‘correct speech’.

Kallawaya - A Secret Language of Bolivia

Katja Hannß University of Cologne Kallawaya is a mixed and secret language spoken by professional herbalists in the Bolivian highlands near Lake Titicaca. The main lexifier language of Kallawaya is Pukina (see Stark 1972: 224), which became extinct latest in the 19th century (Adelaar & Van de Kerke 2009: 125). It is as good as undocumented and the only surviving material of Pukina is a collection of Christian texts of about 300 to 400 words (see de Oré 1607; Muysken p.c.). Apart from Pukina, other languages, such as Quechua and Aymara, also contributed to the lexicon of Kallawaya. The grammar is provided mainly by Quechua Ayakuchano or Cuzqueño (see Stark 1972: 226), which is also the native language of the Kallawaya herbalists. Today, Kallawaya is severely endangered (see Adelaar 2007: 19) and the number of speakers as well as their competence are unknown (for an overview of Kallawaya, see also Muysken 2009).

Page 22: KNAW Conference on Language Dispersal in Africa, Southeast

Page 22 of 23

With respect to the origin of Kallawaya, Muysken (2012: 96, 100) suggests a relexification process, in which (semi-)speakers of Pukina partly maintained the lexicon of their old language but otherwise had already shifted to Quechua (see also Muysken 1994: 201). The etymological composition of the Kallawaya lexicon is currently investigated by the author of the present paper.

In the presentation, two main issues will be addressed: (i) the types of lexical manipulations observable in the Kallawaya lexicon; (ii) possible traces of the lexifier language Pukina.

The etymological investigation makes possible to detect and classify lexical manipulations (see Mous 2003) found in Kallawaya. This is achieved by a comparison between the lexical item of the lexifier language and that of Kallawaya. In the presentation, a (preliminary) classification of lexical manipulations in Kallawaya will be attempted.

Another issue to be addressed are traces the lexifier language Pukina might have left in Kallawaya. For this, etymological and phonological evidence will be considered. First, it is assumed that those Kallawaya words without an established etymological origin are likely to come from Pukina. To further corroborate this hypothesis, phonological evidence will also be taken into consideration. We know that Pukina had a five-vowel system with /e/ and /o/ having phonemic value (see Adelaar and Van de Kerke 2009: 128f), while Aymara and Quechua have only three phonemic vowels. Furthermore, word-initial consonant clusters, as in e.g. Kallawaya pfahcha ‘prostitute’ (Soria Lens 1951: 34) are likewise not found in Quechua and Aymara. It is suggested that words containing at least one of these features may be tentatively analysed as being of Pukina origin.

In addition, it has been observed that alleged Pukina words are often characterised by more complex semantic relations than those that come from Quechua or Aymara. Please consider the example of phekelo meaning ‘artificial light, spotlight, candle, wax, wick’ (Oblitas Poblete 1968: 103). The meaning ‘artificial light’ may be taken as the hypernym, to which ‘spotlight’ and ‘candle’ are in a hyponomy relation. The meanings ‘wax’ and ‘wick’, then, are in a meronymy relation ‘candle’, as they form parts of a candle. It will be discussed whether this observation can be meaningfully related to the issue of the origin of Kallawaya.

Social functions and linguistic aspects of Lá≡ì (Gbáyá, C.A.R.)

Yves Moñino CNRS, LLACAN Social functions Masculine initiation. The target is to transform children into men, in two successive common houses built in the bush, after ‘killing’ them symbolically as children. The initiation takes six months among the ’Bòdòè, one year for other Gbaya communities. The neophytes have 12 to 18 years. They learn men’s activities (progressive self-organisation, hunting, fishing, building houses, dancing, making love, acquiring systematic knowledge of environment, speaking a new language). Women also have their initiation, the bànà, but without any special language. Acquisition of Lá≡ì This language is acquired during the first three months of the initiation in the first camp, with pedagogical methods; it is forbidden to say a word in Gbaya, with corporal punishments in case of errors. The neophytes then move to a second camp, where the atmosphere is much more free and pleasant, they speak the language fluently. This short period of learning is due to the fact they just have to learn a vocabulary (see next section). Linguistic nature of Lá≡ì (I invented the term ‘Pseudo-language’ for it, see Moñino 1991) It is an argot (same grammar as the gbaya), but absolutely all the lexicon is encrypted, even the grammatical morphemes. There are only 150 basic terms, the rest (about 2000 words) is obtained by motivated compounds, derivates and ‘concentrated’ words. Examples of semantic organization an properties of lexical terms will be given. There is no trace of an ancient language, but the basic terms mostly come from Sara (Nilo-Saharan family), a language whose speakers are neighbours of the Gbaya, which is a Niger-Congo language). Manipulations of Láßì as a ‘secret’ language

Page 23: KNAW Conference on Language Dispersal in Africa, Southeast

Page 23 of 23

The language is presented as a secret language; three categories of people must not know it: women, non-initiated children and foreigners. If an initiated man teaches the language to them, ‘he will die’; but the reality is very different. They teach Lá≡ì to me (and to other foreigners) without a problem; all the married women have a passive knowledge of it and the old women speak it fluently in public; even small children understand simple orders like ‘Go and bring me water’. The function of Lá≡ì is to share complicity between initiated people while excluding the non-initiated, but in a second time they finally include them. In some ritual public contexts (the day of neophytes ‘death’, the day before moving to the second camp), the language is used with a strong special voice to impress the neophytes and their mothers.

Secret Knowledges: On the construction and transmission of secret languages Anne Storch University of Cologne Secrecy in language is a complex phenomenon that reaches far beyond the use of specific secret languages. This paper argues that secrecy is actually never fully obtained in secret languages, as these only fulfill their social role when they signify a ‘public secret’, in the sense of not completely hidden knowledge. Secret languages may be constructed in various ways, by different manipulative strategies, and so on. This diversity will be illustrated by selected examples from languages of Nigeria, among other parts of Africa, and will be explained in terms of the sociocultural roles these languages play, for example as ritual languages or secret codes of specific groups. The paper will then address the transmission of secret languages and explore how secret knowledge itself is safeguarded and achieved.