Show Summary Details

Page of

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). © Oxford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Handbooks Online for personal use (for details see Privacy Policy and Legal Notice).

date: 28 May 2020

Introduction

Abstract and Keywords

The first chapter of The Oxford Handbook of Polysynthesis introduces the notion of polysynthesis and the related issues that the volume addresses, such as complexity, the definition of the word, the nature of the lexicon, idiomaticity, and typological features such as argument structure and head marking. It also outlines the part structure of the volume: Part I addresses polysynthesis from different perspectives; Part II contains areal studies of those geographical regions of the world where polysynthesis is particularly common, such as the Arctic and Sub-Arctic and northern Australia; Part III examines diachronic topics such as language contact and language obsolescence; Part IV looks at acquisition issues in different polysynthetic languages.

Keywords: polysynthesis, complexity, language contact, diachronic, obsolescence, acquisition

The term polysynthesis is generally understood in linguistics as extreme morphological complexity in the verb. But morphological structures can be complex in a variety of ways. Most would agree that polysynthetic languages are characterized by words consisting of many meaningful parts, in keeping with the etymology of the term (‘many-combining’). Duponceau, who first coined the term in 1816, defined a polysynthetic construction as one “in which the greatest number of ideas are comprised in the least number of words” (1819).

Soon afterward a similar category was integrated into the growing framework of morphological typology by Humboldt in a work first published between 1827 and 1829 (1988), but under the term einverleibend, ‘incorporating’. Presumably, Humboldt was referring to the incorporation into words of various elements whose meanings would be expressed in separate words in more analytic languages. In modern times, Humboldt’s term has sometimes been interpreted to mean that such languages must also display canonical noun incorporation, though many highly synthetic languages lack it.

The term polysynthesis is now generally established: it is commonly encountered in typological characterizations of languages, and occupies a prominent place in certain theoretical models. But the essence of polysynthesis, both the fundamental features which might most usefully define the type, and other features that are often associated with it, have remained a topic of ongoing discussion. A major purpose of this volume is to explore the extent to which polysynthesis constitutes a clear type with specific necessary and sufficient characteristics, and whether it might have predictive value in accounting for clusterings of typological features.

Perhaps the most commonly cited feature of polysynthesis, beyond general morphological complexity, is the identification within the verb of core arguments. Evans and Sasse summarize this as follows: “Essentially, then, a prototypical polysynthetic language is one in which it is possible, in a single word, to use processes of morphological composition to encode information about both the predicate and all its arguments … allowing this word to serve alone as a free-standing utterance without reliance on context” (2002: 3f.). Such morphological structures might have important implications for syntactic structure. If in a language, arguments are fully specified within the verb, then syntactic relations may be defined within the verb word as well, and relations between the verb and external (p. 2) lexical arguments may not be of the familiar hierarchical type. It has accordingly been proposed that polysynthetic morphology tends to be associated with ‘shallow’ or ‘non-configurational’ syntax. The status of the verb as a complete, grammatical clause in itself raises further questions. Are the syntactic bonds between verbs and lexical arguments in such languages necessarily looser than in other kinds of languages? In many languages, first and second persons are overtly represented within the verb, but third persons are not. Is this difference significant, given that the referents most often identified by lexical arguments are third persons? And where third persons are obligatorily represented and the normal way of achieving reference, does this result in bound pronominal affixes having a significantly different function from their free-pronoun counterparts in other languages, since they no longer form part of an anaphorically sensitive contrast set (Evans 2002)? Is polysynthesis strictly a matter of the verb? What of languages with complex nominal structures that may also encode whole propositions? These are the sort of questions that the present volume addresses.

A number of additional features have been observed in various languages characterized as polysynthetic (Fortescue 1994). Complexity may come about through numerous ‘slots’ in the verbal structure, slots filled by markers that alter argument structure, add adverbial information, and more. Or it may come from large inventories of productive affixes, to the extent that some polysynthetic languages have even been observed to have restricted inventories of lexical roots.

Where affixation is highly productive, particularly in the case of affixes that usually correspond to separate words in other languages, it can be asked what kinds of units are learned, stored, and used as chunks by speakers, and what is processed online during speech. Highly polysynthetic languages obviously contain many more ‘words’ than analytic languages, in the sense of distinct inflected or derived word-forms, with the number of inflected, derived, and incorporation-extended forms of a given root potentially running to the tens of thousands. How many of these formations are lexicalized? Is there less idiomaticity than in the words of only mildly synthetic languages, or more idiomaticity than in their phrases? Elaborate morphology can set the stage for complex morphophonemics, raising further questions about memory and processing. If extensive phonological processes render the internal structures of words more opaque, are these words more likely to be processed as single units? Such questions have a direct bearing on the growing interest in complexity within linguistics today (see, for example, Dahl 2004).

Another set of questions concerns the kinds of circumstances which might foster the development and maintenance of polysynthesis. It has been observed that many polysynthetic languages are spoken by hunter-gatherer peoples or other small-scale societies, or those with recent histories as such. Do these languages or structures ever spread to neighboring peoples? Is it likely that the proportion of the world’s languages which were polysynthetic was greater in the past? What determines the rapidly increasing replacement of polysynthetic languages by majority languages today? Is it the case that they make poor lingua francas, and if so why? (We note that Nahuatl at empire scale, and Bininj Gunwok at regional scale, are two counter-examples of polysynthetic languages that function as lingua francas.) How do children learn them? Do such languages become less polysynthetic as they are learned by new generations in post-contact settings (insofar as they are being learned at all)? One thing is clear: they are nearly all disappearing at an alarming rate, and accurate documentation—especially of their acquisition—is a pressing matter.

(p. 3) On closer examination, it turns out that there is far more variety among these languages than lumping them together as ‘polysynthetic’ might suggest. Consider one sample of such variation in a single area, the Arctic and adjacent Sub-Arctic. (See the papers in Mahieu and Tersis 2009 for more detail than we can give here.) Some show multiple types of extensive incorporation (Chukchi and Koryak). Some show extensive ‘lexical affixing,’ with structures similar in many ways to prototypical noun incorporation, but with verbalizing affixes, often with relatively concrete meanings, in place of verb roots (the Wakashan languages). Some show strong ‘templatic’ structures (the Athabaskan languages and Siberian Ket). Some show hierarchical structures with ‘recursive suffixing’ (the Eskimo-Aleut languages).

For the most part, languages with such constructions also offer speakers analytic alternatives: where speakers can use a verb with an incorporated noun, they can also use a verb plus a separate lexical noun, for example. The motivations behind choices between polysynthetic and analytic constructions within a language can be discovered only through examination of extended spontaneous speech, since they are usually correlated with the management of information flow through time.

A closer look at possible combinations of features associated with polysynthetic languages might tell us much about the limits of what constitutes a stable configuration for a human language. In some cases these combinations seem to defy typological expectations derived from more familiar language types. Some typologies have posited an inverse relationship between affix order and word order, for example, citing correlations between predicate-final order and suffixes on the one hand, and predicate-initial order and prefixes on the other. But languages of the structurally homogenous Athabaskan family show clear SOV constituent order along with almost purely prefixing morphology, both of which have apparently remained stable over millennia. Similar stability can be seen in the suffixing structure of the Eskimo-Aleut languages, and the prefixing, suffixing, and incorporating structures of the Iroquoian and Gunwinyguan languages.

Apart from the general features relevant to polysynthesis we have listed, some more specific distinctions can be investigated:

  • Bound core pronominals

    • Are they clitics or affixes?

    • How many core arguments are marked, and which are they?

    • What alignment patterns do they show?

    • (accusative, ergative, agent/patient; inverse or other hierarchical patterns)

  • Noun incorporation

    • Do both intransitive and transitive bases incorporate nouns?

    • Does incorporation yield both intransitive and transitive stems?

    • Are there limitations on the nouns that are incorporated, such as only body-part terms?

    • What semantic relations may exist between the incorporated noun and incorporating verb stem?

    • Do incorporated nouns match their freestanding source exactly, somewhat, or not at all?

    • Are multiple or phrasal incorporates possible?

  • ‘Heavy,’ semantically concrete verbalizing affixes

    • Are there ‘dummy’ noun stems to which they attach? (p. 4)

  • Verb compounding

    • If verb stems are compounded, what semantic relationships can they bear?

  • Means/manner/instrumental affixes

  • Locative/directional affixes

  • Applicatives

  • Derivation and inflection

    • Is there a clear distinction between the two?

    • Do they occupy distinct areas of the morphology?

  • Templatic ordering of derivational morphology

    • How many slots are there in the maximal template?

    • How many morphemes are there in each slot?

    • Are there discontinuous morphemes?

  • Hierarchical structure among derivational morphemes

    • Does morpheme order reflect semantic scope?

    • Are alternative orders possible?

    • Is there recursion of individual affixes?

    • Can multiple instances of a given morpheme be adjacent?

Linguistic types become more interesting when they reveal clusters of associated features. Features may cluster for a variety of reasons. There may be structural interdependencies, such as specification of grammatical relations within the verb and primarily pragmatically based constituent order. One may set the stage for the development of another: predicate-final clause structure (SOV) can give rise to postpositions, where the postpositions are descended from verbs. Circumstances that foster the development of some features might also foster the development of others, such as small communities with dense social networks. But correlations posited on the basis of a few languages have not always been borne out once more languages were considered. Some additional variables that have been discussed in conjunction with polysynthesis have included:

  • complex morphophonological processes, yielding structures that are more fusional than agglutinating

  • predominantly prefixing structures, suffixing structures, or a combination;

  • general head marking beyond the verb, such as possessive affixes on possessed nouns,1 or absence of nominal case marking;2

  • relative elaboration of nominal morphology, such as noun–noun compounding or multiple lexical classes, including gender;

  • absence of noun case;

  • relative elaboration within the verb morphology of relations indicated periphrastically in other languages, such as causatives and desideratives;

  • absence of a separate class of adpositions; (p. 5)

  • constituent order that is based primarily on pragmatic factors rather than grammatical relations (SOV, VSO, etc);

  • internally or externally headed relative clause constructions;

  • clause combining accomplished primarily via verb inflection, auxiliaries, or conjunctions.

This handbook draws together different perspectives on intriguing languages which until now have been scattered through a rather disparate and specialized literature. It covers the majority of language families that would be termed ‘polysynthetic’ by the majority of linguists today, as well as relevant isolates. In addition, some important borderline cases are also discussed, because the border between polysynthetic and merely synthetic is fuzzy. This is despite at least one influential claim to the contrary, namely Baker’s (1996) Principles and Parameters approach to polysynthesis, which restricts the term to a specific subset of incorporating languages: those with complex head-marking morphologies that share certain specific syntactic and morphological features, such as those displayed by Mohawk. At issue here is the opposition between a binary, essentializing typological contrast, conceptualized by Baker in the form of a single macroparameter (Baker’s Morphological Visibility Condition) that predicts a large number of linked characteristics, and a more functionalist approach that sees prototypical polysynthetic languages as a kind of local minimum in the adaptive landscape, in which ‘functional relationships’ (Heath 1975) (or perhaps better ‘functional interdependencies’) tend to co-select, but with the linkage being stochastic rather than necessary.

Polysynthetic languages are not scattered evenly over the globe. Certain areas show a higher propensity for such languages: North and Central America, parts of the Amazon and the southern cone of South America, areas of Siberia from the Bering Strait across to Ket to the west and Ainu to the south, the north-central part of Australia, the Sepik River area of New Guinea, the Northwest Caucasus, and parts of India and Nepal (some Munda and Kiranti languages). They are not entirely absent elsewhere, however. Gumuz, a Koman language of Africa, shows reference to core arguments, noun incorporation, and other complexity within the verb (Ahland 2012). Sketches of languages typifying almost all of these regions are included in this volume, highlighting the principal grammatical features they share. The historical emergence, fading, and acquisition by children of those polysynthetic languages that remain today are discussed, insofar as this is still possible, drawing upon what is known of their relevant socio-cultural and demographic backgrounds. Suggestions, some of them controversial, are adduced as to how the complexity displayed by these languages might have arisen and not been reduced again in the course of time.

In Part I, the search for core features of polysynthesis is addressed from various perspectives. The phenomenon is first considered against the broader issue of complexity by Östen Dahl. Polysynthetic languages, which bear the marks of multiple grammaticalization processes, are quintessential examples of what Dahl terms ‘mature phenomena,’ displaying not only complex but also non-linear morphology.

Marianne Mithun examines whether the essence of polysynthesis is holophrasis in a specialized sense: the expression of essential elements of the clause, both predicate and arguments, within the verb. The identification of core arguments by pronominal affixes might accomplish this, but such marking is not categorical. Languages vary in which arguments are marked, with differences dependent on such features as person, animacy, grammatical role, and definiteness. Argument specification within the verb often co-occurs with certain other (p. 6) structures, such as noun incorporation, applicatives, rich inventories of adverbial affixes, and pragmatically motivated word order, but ultimately none is predictive of any of the others.

Johanna Nichols discusses the principal features characterizing polysynthetic languages in the Greater Pacific Rim area. She concludes that while there is no particular limit on the diversity of these languages, a tighter definition can be approximated in terms of what she calls ‘open head marking’. Polysynthesis is seen as a natural consequence of head marking, and its emergence presupposes a broader area in which head marking already predominates.

Johanna Mattissen proposes a typology of polysynthetic languages. Her affixal type allows only one lexical root per verb complex, and has ‘non-root bound affixes’ for all other morphemes within the word. This type contrasts with her compositional type, which combines more than one lexical root (noun+verb or verb+verb) in complex verb forms in a productive manner. These two types are cross-cut by another distinction: templatic organization, characterized by a fixed number of morpheme slots within the word in a fixed order, and scopal (hierarchical) ordering, whereby successive affixes are chained freely, as conditioned by their semantics, scope relations, and compatibility restrictions. Different areas of the verbal complex, such as prefixes versus suffixes, or inflection versus derivation, may belong to different subtypes.

Jerrold Sadock challenges the Greenbergian measure of word length as criterial for defining a distinct class of languages. He compares Hebrew, which no one would call polysynthetic, with Greenlandic Eskimo, which almost everyone would include in that type. He pinpoints the relevance of morphological processes that are ‘expressively active’, that is, those with functions typically conveyed syntactically in other languages. He also urges a closer consideration of the role of clitics in individual languages as a counterbalance to the raw quantification of word length.

Michael Fortescue proposes a two-part rule of thumb definition of polysynthesis that combines an inflectional and a derivational component, requiring the combination of bound pronominals and the possibility of at least some semantically ‘heavy’ morphemes (roots or affixes) besides the main verb root. This produces a view of polysynthesis in terms of a core vs. periphery rather than a discrete category, some highly synthetic languages falling short of fully meeting both criteria (often for transparent reasons). If the nub of polysynthesis is the packing of much material into single verb forms that would be expressed as independent words in less synthetic languages, the question is then: What exactly is the nature of and limitations on this ‘material’? His chapter investigates the limits of what the term is generally understood to cover. Cognitive constraints on the maximal extent of polysynthesis are discussed.

Louis-Jacques Dorais examines the problematic nature of the polysynthetic lexicon (both written and mental): how many words does the lexicon actually contain, given the relatively open-ended nature of polysynthetic morphology? Do native speakers regard all of their complex words as holistic units? He focuses on the derivational morphology of Inuktitut in addressing these questions, but draws in comparisons with Algonquian and Iroquoian languages. He concludes that it is impossible to define precisely which elements of a speaker’s mental lexicon function as fully lexicalized lexemes, and which are circumstantial derivatives utilizing compositionality.

Fundamental to discussions of polysynthesis is the notion of the word. Balthasar Bickel and Fernando Zúñiga investigate phonological and syntactic challenges for a precise definition of the word across languages. Their approach includes multiple domains within (p. 7) each: between syllables and phrases for the first, and between morphemes and phrases for the second. They provide a set of analytical tools and parameters for determining phonological and syntactic wordhood, then illustrate the wide diversity of possible polysynthetic words these can define across languages, comparing Mapudungun, an unclassified language of Chile and Argentina, and Chintang, a Sino-Tibetan language of Nepal. They conclude that polysynthetic ‘words’ are never unified entities defined by a single domain in which all criteria would converge.

Peter Trudgill delves into the possible role of the anthropological setting of cultures in the rise and maintenance of polysynthesis. It has been observed that many polysynthetic languages are spoken in hunter-gatherer societies. Trudgill refines this observation, pointing out that these languages are often associated with relatively small, isolated, stable communities, with dense social-network structures; they are less characteristic of large, high-contact situations, in urban, colonial, and standard language varieties. He discusses the factors that might allow complex morphologies to persist unreduced or become still more complex through time.

Moving from the speech community to the individual speaker, Sally Rice argues from a usage- and construction-based cognitive perspective that polysynthetic languages such as Dene Sųłiné (Athabaskan) are not learned as complex systems of rules, but rather by exemplars experienced in specific contexts. Speakers can expand the language by analogy, but such extension is arguably secondary. She proposes that the correct level at which such highly compressed, ‘esoteric’ languages should be investigated is the little-explored one of phraseology. This approach has potential practical consequences for the production of pedagogical materials aimed at maintaining these languages.

Contributions in Part II focus on those geographical areas where polysynthetic languages are most concentrated. Some represent genuine linguistic areas where long-standing interaction has occurred, others a looser, more contingent piling up of highly synthetic languages with relatively little contact between them. In each of the areas treated, the major polysynthetic features and their distribution across the languages concerned is sketched.

The Arctic/Sub-Arctic, as described in Michael Fortescue’s chapter, combines three widespread polysynthetic language families of very different types. Between Eskimo-Aleut and Na-Dene, the latter with affinities further south, there has been relatively little contact, at least in recent centuries. Between Eskimoan and the Chukotian branch of the Chukotko-Kamchatkan family, however, there has been considerable contact and interaction.

The second area, discussed by Marianne Mithun, covers the rest of North America, where a substantial proportion of the languages are polysynthetic or at least relatively synthetic. It was some of these which first prompted Duponceau to distinguish his polysynthetic type, a type initially taken to characterize all of the languages of North America, though subsequent discoveries proved that to be an oversimplification. Within this continent are more restricted areas with their own foci of specific polysynthetic traits, such as the Northwest Coast (Salishan, Wakashan, Chimakuan) and the Eastern Woodlands (Iroquoian and Algonquian).

One of these smaller areas is described in Carmen Jany’s chapter on a set of languages sometimes referred to as Northern Hokan, centered in Northern California and Southern Oregon. The term ‘Hokan’ was earlier thought to designate a set of languages genetically related at a very deep level, though subsequent work has shown that many of the structural traits once attributed to common inheritance are in fact the result of longstanding contact. (p. 8) A great many of the traits shared by these languages are found as well in other neighboring languages never considered genetically related.

The Amazonian basin, described by Alexandra Aikhenvald, is home to a complex patchwork of highly synthetic languages, many of which have barely been documented. One of the better known families is Arawakan, which generally comprises the most synthetic languages of the entire area, though its northern members show somewhat fewer polysynthetic traits. Noteworthy is the relative scarcity of bound object marking in many of these languages.

The concentration of polysynthetic languages in northern Australia is well known. Languages of this area, described by Nicholas Evans, show contrasting varieties of polysynthesis, one type with a single verb stem plus canonical noun incorporation, as seen in languages of the Gunwinyguan family, and the other with a more elaborate two-part verb structure deriving from the combination of a light verb and an inflected auxiliary, as seen in languages of the South Daly River region. Evidence for the stability of the Gunwinyguan-type pattern over many millennia comes from the striking similarities in patterning of the verbal template between Gunwinyguan and the isolate Tiwi, cut off from contact with the mainland for over 6,000 years, and between mainland Gunwinyguan and the Enindhilyakwa language of Groote Eylandt, also cut off from the mainland for several millennia: Enindhilyakwa was formerly considered an isolate but now demonstrated to be related to the Gunwinyguan languages (Van Egmond 2012). At the same time, the structural homogeneity of these languages should not be exaggerated, and a comparison of superficially similar languages from within Gunwinyguan reveals interesting variations on such dimensions as the semantic relationships contracted by incorporated nouns with the verb stem, the overt registration of all person/number values by affixal morphology, and the absence vs. elaboration of subordinative morphology. This speaks against any simplistic view that particular typological features necessarily cohere.

The final area considered is the Sepik Valley region of Papua New Guinea, described by William Foley. In this area of extreme genetic and typological diversity, the nature of the polysynthesis reflects the preponderance of extensive verb serialization in the neighboring Papuan languages. In Yimas and Alamblak, for example, comparable sequences are compacted into single word structures. Foley shows that when four parameters are considered, namely head marking, pronominal marking in the verb, noun incorporation, and clause linkage by parataxis, these languages show that polysynthesis is not an all or nothing trait.

In Part III the phenomenon of polysynthesis is examined from a diachronic perspective. Various routes into polysynthesis are described, all involving a drift towards increased head marking and greater elaboration within the verbal morphology. The demise of a polysynthetic language can be gradual or abrupt, especially under competition from more analytic colonial languages and an influx of adult second-language learners. Sometimes the results are unpredictable, as in the case of mixed languages.

The emergence of complex templatic morphology is described by Edward Vajda, with examples from Yeniseian and Athabaskan (Na-Dene) languages. This particular coupling is not fortuitous: Vajda argues that the two families may ultimately be genetically related. If they are, their similarities indicate how durable complex polysynthetic morphology can be, despite major adjustments and reanalyses within the templates involved. In the case of Yeniseian Ket we see how language contact, here with Uralic and other Siberian languages, can induce major structural changes that result in still greater complexity and irregularity.

The emergence of complex verb structure from sequences of verbs is discussed by Tom Givón in his description of Ute. He traces two pathways from syntactic to morphological (p. 9) structure, one via the embedding of complement clauses within main verb clauses, the other via the condensation of clausal chains into serial verb constructions. Both may involve the ‘co-lexicalization’ of a light verb plus a superordinate main verb, but only the former results in full clause-union. For Givón however, polysynthesis is not necessarily an endpoint, but rather one of many transitional phases along recurring pathways of grammaticalization.

The range of effects that language contact may have on polysynthetic languages is addressed by Peter Bakker and Hein van der Voort, who contrast the impact of contact on Algonquian, Eskimo-Aleut, Wakashan, Athabaskan, and Sepik Papuan languages. Most contact situations result in simplification of one sort or another, but they may also result in the increased complexification of the contact language. They cover lexical borrowing and the diffusion of grammatical traits, as well as the emergence of pidgins and trade jargons involving these languages.

The impact of obsolescence on polysynthesis is discussed by Ekaterina Gruzdeva and Nikolai Vakhtin. Such processes may run their course rapidly, as in Tiwi, which moved from highly polysynthetic to nearly isolating within a couple of generations. The catalyst is often language contact, though the nature of the contact may range from rapid shift to a dominant language, as is occurring in so many parts of the world, to more subtle influences of neighboring languages, like that of Chukchi on Siberian Yupik. The degree of synthesis of the dominant language does not appear to be a significant factor: both Chukchi and Siberian Yupik are polysynthetic. Gruzdeva and Vakhtin investigate the order in which polysynthetic traits tend to be lost, comparing noun incorporation, inflection, and allophonic variation.

Part IV is concerned with an area requiring urgent attention: the acquisition of polysynthetic languages. With few exceptions, very few children are currently learning polysynthetic languages. How children acquire such complex structures with apparent ease would appear to be something of a mystery. Numerous factors come into play, in part according to the nature of the polysynthetic structures to be acquired. Phonology can affect acquisition in various ways. Extreme fusion structures should delay the identification of morphological components, but other factors can enter in as well. In Mohawk, for example, children first acquire the stressed syllables of words, which most often align with the root or a part of it, providing a communicative advantage (Mithun 1989a).

In the first of the three chapters in this section Shanley E. M. Allen describes the acquisition of Inuit languages, focusing on Inuit children between the ages of 1 and 16 in Arctic Quebec. The order of emergence of different types of affixes appears to be very similar to that of related West Greenlandic described in Fortescue (1984) and Fortescue and Olsen (1992a). All Inuit dialects lack distinctive stress, which may have some bearing on the strategies children apply; comparison with the acquisition of Yupik languages, which do have it, might be instructive. Among the topics covered in this chapter are nursery vocabulary, word-internal morphological complexity, verb inflection, incorporative structures (sequences of noun stems followed by verbalizing suffixes with relatively concrete meanings), passives, causatives, tense, and aspect. Allen also briefly considers aspects of narrative acquisition.

In the second chapter on acquisition, Bill Forshaw and colleagues describe the acquisition of Murrinh-Patha, a polysynthetic language of Australia. The complex morphology of this language is of a quite different kind from that of Inuktitut: it is templatic, with discontinuous (p. 10) dependencies among components of bipartite verbs. The authors point out that children learning this language make heavy use of a handful of highly productive ‘classifier’ stems which combine with a wider variety of lexical stems as a key to unlocking the bipartite verb system.

In their chapter, Sabine Stoll, Balthasar Bickel, and Jekaterina Mazara investigate the acquisition of Chintang, a Sino-Tibetan language spoken in the Himalayas (Kiranti group). In this language, a verb stem combines with a large number of other morphemes that have an adjacency requirement and range from grammatical affixes to other verb stems, and the expression of these combinations shows considerable variation in exponence. Many of these morphemes take on different shapes, depending on other morphemes in the same string, which presents considerable challenges for the child learner. Such combinations bring with them yet another challenge: while morphemes like English third person -s on a verb stem are compatible with a very large number of contexts, its equivalents in Chintang are far more specific, tailored to specific configurations of the participants in an event. And since apart from person, many other categories are expressed as well, each string is extremely specific to a particular context. This inevitably lowers the chances of finding repeated occurrences of the same string, and thus increases the challenges for the learner even more.

Part V, the final section of the volume, contains grammatical sketches of a set of representative polysynthetic languages, distributed as shown in Map 1.1. These chapters speak for themselves and illustrate many of the general points raised in earlier chapters.

Salient morphosyntactic features across the individual languages can be compared as tabulated in Table 1.1. The sub-type of each of them according to Mattissen’s four-way distinction can be seen in Table 1.2.

Many theoretical challenges remain in the study of polysynthetic languages, above and beyond those presented in this volume. We mention five of them here.

The first concerns the crucial question of how far production and perception are routinized in normal speech, in any language. As Pawley and Snyder (1983) argued three decades ago, the production of fluent speech requires the rehearsed recall of large preassembled chunks into which a typical maximum of one new element per breath-group is inserted. This view locates normal, fluent language production much closer to the holophrastic end of the spectrum than to the generative, compositional end making ‘infinite use of finite means.’ This phraseological view of language use, and language learning, has recently been bolstered by studies such as that of Tomasello (2003) showing the close matching of child language learning with large chunks of adult input. On the one hand, polysynthesis now appears as a natural consequence, given this more phraseological view of language use: polysynthetic languages simply wear their holophrastic hearts on their structural sleeves more clearly than some other languages. On the other, when we turn this view back on actual descriptive and analytical practice to ask how many of the vast arrays of possible word-forms are actually employed by speakers of polysynthetic languages, we may find that descriptions have exaggerated the morphological productivity of the system. This makes corpus-based studies of actual occurrences of different combinations an important challenge for future work, as outlined in Rice’s chapter.

The second challenge concerns the distribution of semantic labor between predicate, arguments, and other expressions in the polysynthetic lexicon. Capell (1969: 14–17) elaborated a distinction between ‘event-dominated’ and ‘object-dominated’ languages, in an (p. 11) (p. 12) (p. 13) attempt to capture whether the burden of expression focuses on the encoding of events or on the entities involved in them.39

Introduction

Map 1.1 Location of languages described in Part V

© Australian National University. CartoGIS CAP 15-221_KP

Table 1.1 Main polysynthetic features of languages in volume sketches

1

2

3

4

5

6

7

8

9

10

11

12

13

Yup’ik

+

+

+

+

(no)3

+4

+

Innu

+

+

+

+

+

+

+

+

+

+5

+

Nuuchahnulth

+6

+

+

+

+

+7

+

+

Sliammon

+

+

+

+

W. Apache

+

+

+

+

+

+

+

+

Caddo

+

+

+

+

+

+

+

+

Purepecha

+8

+

+

+

+

Nahuatl

+

+

+9

+

+

+

+10

+

+

+

Tariana

+11

+12

+

+13

(no)14

+

(no)15

Lakonde

+

+

+

+

+

+

+

Mapudungun

+

+

+16

+

+

+17

+

+

+

Koryak

+

+

+

+18

+

(no)19

(no)20

+

+

Ket

+

+

+21

+22

+23

+

+

+

+24

Nivkh

+25

+

+26

+

+

+27

+28

Ainu

+

+

+29

+

+

+30

+

+

Yimas

+

+31

+

+

+

+32

(no)33

+

Dalabon

+

+

+

+

+

+

(+)

+

+

+

+

Sora

+34

+

+

(no)35

Adyghe

+

+36

+

+

(no)37

+

+

1. Bound core pronominals

2. Noun incorporation

3. Heavy verbalizing affixes

4. Verb incorporation (single word serialization)

5. Adverb incorporation

6. Lexical affixes (locative and/or instrumental)

7. Heavy adverbial affixes (manner, etc.)

8. Classificatory affixes, i.e. verbal affixes of the Na-Dene/ Wakashan/ Algonquian type that indicate the substance/shape, etc., of the principal argument of the verb

9. Recursive derivational morphology, i.e. recursivity of the Eskimo type with the potential for building up complex words through successive derivational affixes38

10. Templatic derivational morphology, i.e. not just inflectional sequences arranged by fixed ‘slot’

11. Discontinuous morphemes

12. Inflectional/derivational entanglement, i.e. within the polysynthetic word whether inflections ever lie inside the derivational morphemes over which they have semantic scope, in violation of what is sometimes known as the mirror principle

13. Applicatives

We know that languages are subject to considerable variation in whether aspects of the encoded world are presented as events and states (typically by verbs) or entities (typically (p. 14) by nouns), whether, for example, tides are depicted as actions (as in Iwadja) or entities (as in English), or whether kinship relations are represented by verbs or nouns (Evans 2000). It has also been hypothesized (Sasse 1988) that the proportion of verbs in the lexicon is higher in polysynthetic languages, but this is not fully borne out by some recent studies (Evans 2012, Wnuk 2016). We now need a three-pronged attack on this problem, first extending studies of the lexicon to a wider range of languages, second comparing token frequencies between (p. 15) corpora of polysynthetic and non-polysynthetic languages, and third examining the mutual frequency adjustments languages make in their coding choices in cases where they are moving towards or away from a polysynthetic high point.

Table 1.2 Assignment of the languages sketched to Mattissen’s sub-types

Yup’ik

Af+Sco

Innu

Af/Com+Tem/Sco

Nuuchahnulth

Af+Sco

Sliammon

Af+Tem

W. Apache

Af+Tem

Caddo

Com+Tem

Purepecha

Af+Tem

Nawatl

Com+Tem

Tariana

Af+Sco

Lakonde

Af+Tem

Mapudungun

Com+Tem/Sco

Koryak

Com+Sco

Ket

Com+Tem

Nivkh

Com+Tem/Sco

Ainu

Com+Tem/Sco

Yimas

Com+Tem/Sco

Dalabon

Com+Tem

Sora

Com+Tem/Sco

Adyghe

Af+Sco

The sub-types according to Mattissen’s four-way distinction (note the mixed types):

Af.  The affixal type allows only one lexical root per verb complex (and has non-root bound affixes for all other morphemes within the word)

Com. The compositional type combines more than one lexical root (whether noun plus verb or verb plus verb) in complex verb forms in an ad hoc manner.

Tem. Templatic organization is characterized by a fixed number of slots in a fixed order.

Sco. By scope ordering, successive affixes are chained freely, as conditioned by their semantics, scope relations, and compatibility restrictions.

The third issue concerns how far the notion of polysynthesis can be applied to sign languages, in particular to their frequent use of ‘classifier construction signs’ (like CL:PLANE-DOWN for ‘planes crash’). The Berkeley Transcription System Manual40 includes explicit instructions for transcribing ‘polycomponential signs,’ and the deaf linguist Lars Wallin has argued for the existence of polysynthetic signs in Swedish Sign Language (Wallin 1994).

The fourth issue concerns the question of what difference it makes whether grammatical encoding is accomplished by the morphology or the syntax. One view of what happens in polysynthetic languages, forcefully espoused by Mark Baker (1996), is that polysynthetic languages essentially exhibit syntax in morphological clothing, motivating his account of such issues as constraints on what nouns can incorporate by appeals to their place within phrase structure relationships. An alternative view espouses the reinstatement of ‘morphology by itself’ (Aronoff 1993) as a distinct level of grammatical organization with its own properties. An obvious difference is the way in which recursion is handled by different languages, either by syntax or by morphology. It can be seen from the sketches presented in this volume that not all polysynthetic languages treat this the same way: while some display a degree of morphological recursion, many—following from their templatic structures—do not. In fact templatic morphology is by far the commonest type in our sample. In others, such as Dalabon or Bininj Gun-wok, there is limited potential for recursion by incorporating one verb inside another, but this appears to be capped at one level of embedding. For other languages, such as Yimas, the descriptions mention the possibility of multiple compounding of stems, but without giving actual statistics on the distribution curve. The most celebrated case of all, of course, are Eskimoan languages, long cited for the recursive nature of their morphology, but even there one encounters limits. We face the very real possibility that no polysynthetic language allows truly unbounded recursion, and moreover that this is not merely an issue of performance but that there are (possibly still undiscovered) limits on productivity. If templatic morphologies can be likened to ‘finite state grammars’ this may be taken to suggest that they involve just a small number of choices, but in principle a finite state grammar can generate many hundreds of thousands of forms for a given lexeme, and figures in the tens of thousands may turn out to be the working stock of memorized forms, at least of common verbs, for speakers of the more exuberant polysynthetic languages. Here again polysynthetic languages are not mere freaks, but have the potential to illuminate normal features of all languages in a particularly revealing way, by confronting us with the fact that human memory may be a much more powerful tool in normal language use than we have assumed.

This brings us to the fifth point, the question of how polysynthetic languages are processed, in both production and perception. Are polysynthetic ‘word-sentences’ broken down into subcomponents, or treated as single units? How does prosody guide the parsing process, particularly because the role of prosodic contrast appears to be absent or virtually so inside the verbs of polysynthetic languages—it appears that no polysynthetic languages represent (p. 16) an incorporated element in one word as contrasting with a structurally equivalent incorporate in another, of the type ‘she-MEAT-ate but he-FISH-ate,’ or ‘he-wood-CHOPPED but he-wood-HACKED.’ What lemmata are accessed during processing? How is the rhythm of processing affected by what is likely to be a different pattern of inter-word transitional probabilities as between polysynthetic and other languages, given the much looser connection between polysynthetic words and the elements around them? At present we have no answers to these questions, a lacuna reflected in the absence of any chapter addressing them in this book, but we hope that the collection of materials assembled here will at least furnish a body of material from which good candidates for research on the processing of polysynthetic languages can be singled out.

In conclusion, we believe that the contents of this volume will provide the general reader—not only linguistic typologists—with a state-of-the-art overview of research into the phenomenon of polysynthesis at a time when the very object of this research, the languages themselves, is rapidly diminishing. We hope that it will encourage more researchers to get out there and study them from all the various facets that the studies in this volume have covered—and more—before it is too late. There are plenty of questions, both theoretical and empirical, that await investigation in greater detail.

Notes:

(1) Predominant head marking is assumed for most polysynthetic languages at the clausal level, though some (like Eskimo and Chukchi, with nominal case) have double-marking patterns here.

(2) Compare the presence of a nominal case system in Eskimo-Aleut with the lack of same in the equally affixing-only Wakashan languages, lacking case.

(39) Capell’s view—philosophically provocative but perhaps naïve—was that in event-dominated languages “[t]‌the whole interest of the utterances lies apparently in what happened, when and how it happened, rather than in the people or object involved or the place of the utterance” (Capell 1969: 14). Interestingly, the recent neo-Whorfian resurgence of interest in testing relationships between language structure and cognitive bias has not, to our knowledge, led to any studies focussed on this type of suggestion.

(3) But there is a small number of ‘verbal postbases with templatic properties’ occurring just before the inflection (otherwise called ‘sentential affixes’).

(4) As regards ‘quasi-incorporation’ of possessor- or case-marked nominals.

(5) As regards root-medial-final shape of primary derivation.

(6) But bound object only in imperative, subjects (non-3rd) clitic.

(7) Minimal (some successive lexical affixes possible).

(8) Fixed 2nd position clitics, except 3pl object (suffix).

(9) Only a few productive ones, a general intransitive (‘become’, etc.) and a general transitive (‘make’, etc.), also an instrumental (‘do with’).

(10) Non-productive.

(11) Subject only.

(12) Limited.

(13) Limited.

(14) But some limited reduplication of the causative/applicative affix.

(15) Causative affixes have applicative use.

(16) Several ‘verbalizing verbs’ in special meanings as verbalizing suffixes, and inchoative (‘become’) sense of reflexive suffix.

(17) Body parts only.

(18) Limited.

(19) But some heavy deverbal-verbal affixes.

(20) But ‘onion skin’ morphological ordering (with circumfixes).

(21) Quasi-verb-incorporation with action nominals.

(22) Closed class directionals and some affix-like bleached nominals.

(23) Non-productive

(24) Small closed class of comitative applicatives made from basic motion verbs.

(25) Object but not subject (except in converbs).

(26) Just a few, general ones (‘make’, become’, ‘comprise’, etc.).

(27) Only on righthand side of main stem.

(28) Modality more peripheral than (future) tense at least.

(29) Just general verbalizers (‘make’ and ‘have’ and copulative ‘be’).

(30) Limited to righthand side of stem.

(31) Non-productive.

(32) Incorporation recursive.

(33) But paucal number may be marked in final position, going with a paucal/plural number pronominal prefix.

(34) Usually only one pronominal argument suffix at a time.

(35) Nominals in instrumental adverb functions only.

(36) Limited to zero verbalization of incorporating nominal structures and body part applicative objects.

(37) But with templatic ordering of the more inflectional affixes.

(38) Of course this may be a matter of degree – e.g. Dalabon allows one layering of recursion of this type, by incorporating up to one verb into another, but not more.