Introduction, Tony Woodbury

My two talks focus on the problem of how small-scale variation, as revealed in fine-grained dialectology, relates to the overall typological diversity found in facets of two relatively shallow language families: Yupik-Inuit, the most diversified subgroup of Yupik-Inuit-Aleut; and Chatino, a subgroup in the Otomanguean language family that is most closely related to the Zapotec languages. The talks focus on typologically unique and unusual suites of properties special to each family and constituting what may fairly be called part of the special 'genius' of those families. In the Yupik-Inuit case this genius involves a grammar whose fully productive units include both whole words, and stem- and suffix-lexemes within words. In the Chatino case, it involves elaborate tonal contrasts that distinguish lexemes; expone inflectional categories; and classify stems according to inflectional class, lexical category, expressive value, and contact provenance. But the two language families differ radically in how small-scale variation affects typological diversity. In Yupik-Inuit, the older and deeper of the two families, small-scale variation leads to replication and reproduction of the family's singular morphosyntactic genius; whereas in Chatino, the younger and shallower family, local phonological and morphological innovations lead to a kaleidoscope of typological outcomes when measured in terms of the comparative literature on tone.

While the two language families offer strongly contrasting outcomes for typological diversity as we are used to measuring it, they also provide an opportunity for reflection about what we measure and how we measure it. To some extent--especially in the case of Chatino--the typology may look more diverse than it really is because tonal typology has focused too much on phonetics and phonology, and not enough on morphology. But even taking that into account, the diifferences between the two families suggest that typological uniqueness or rarity can be either very stable or very unstable, calling into question any generic idea that some configurations should be more 'canonical' than others. They also underscore the relevance of typological approaches that go beyond those approaches that deal only in strict correlation among language properties and the explanation of such correlation in terms of grammar-engineering or universals of interaction or discourse; or those approaches concerned 'merely' with the profiling or checklisting of language properties. Instead, they point to the use of typology to help us understand and characterize adventitious properties of individual language histories, challenging us to understand such suites of properties in terms of the cognitive, expressive, and aesthetic sensibilities and practices that may characterize communities across generations, or that may change in them from time to time.


Introduction, Sally Rice

These two talks focus on phenomena within the Athapaskan or Dene language family, one of the most prolific (in terms of number of individual languages) and recent, yet extensively distributed (in terms of overland, pedestrian migration) Indigenous language families in the Americas. Linguistic properties that, to quote Tony Woodbury, make manifest the “special genius” of the family include a tension between conservative and innovative tendencies, both of which are on display within individual languages and across the family. Among the conservativizing forces are an exceptionally small stem inventory (and a large number of cognates), a somewhat unique prefixing system that places inflectional material inside of derivational morphological within the polysynthetic verb word, and a very limited set of word-formation strategies. Among the innovating factors are a tendency to fully exploit those limited word-formation options, a tolerance for morphemic order variation within the verb word, and very active relexification and grammaticalization processes. In short, there are only so many ways to skin a cat morphosyntactically in a Dene language, yet individual languages seem to exploit those limited devices and meanings to the hilt. The net result is a feeling––by speakers and linguists––of familiarity yet strangeness when comparing languages across the family.

As interesting as these formal (structural) consistencies and inconsistencies are, it would be a wrong to consider them in the absence of the socio-cultural context in which Dene peoples have existed for the past two millennia or so. I will address aspects of Dene group dynamics (or more particularly, language use) still very much in evidence today: oralcy and an insider-orientation. Both factors may help explain for other language groups how small-scale variation can accrue, leading to wholesale typological diversity.



Tony Woodbury

Monday, November 19

TALK 1: Yupik-Inuit morphology and syntax: Small-scale variation while conserving an unusual typological 'genius'

Following the set of general themes described above, this talk will survey typological variation in Yupik-Inuit, the branch of Yupik-Inuit-Aleut that is spoken on both sides of the Bering Sea and whose Inuit sub-group spread, within the last 1000 years, across the Arctic from Alaska to Canada and Greenland. It focuses on what has persisted in the family, as well as on the nature of variation. All of the languages have polysynthetic words that begin with a morphologically simple or complex base lexeme, and recursively derive higher level structures by adding morphologically simple or complex suffix lexemes to derive higher-level constituents, followed finally by inflection for mood and person (verbs) or case, number, and possessor (nominals), giving a pattern that is both head- and dependent-marking and mostly ergative-absolutive in alignment. This basic profile is persistent across the entire family, and it has been stable over time, as shown by the fact that when lexical cognate sets are compared, bases are cognate with bases, suffixes with suffixes, and only in a tiny handful of cases are bases cognate with suffixes. The innovation and variation that has occurred in the family 'works within the system'. For example, the Yupik languages have innovated productive stress and foot-formation rules that pile up, one upon the next, as one moves east from Central Siberian Yupik and then south through the chain of Central Alaskan and Pacific Yupik varieties. But these rules only serve to amplify the same base-plus-suffix polysynthetic word domain that is evident across the whole family, including Inuit. Likewise, moving from one Yupik or Inuit variety to the next, a given productive derivational suffix may be composed of partially or wholly different suffix elements even as the grammar and meanings of the suffix (and of the stock of hundreds of such suffixes) remain remarkably persistent (e.g., all Yupik-Inuit languages have suffixes for 'to be unable to V' and 'to ask someone to V'; none have 'on land' or 'with the foot', as found in Wakashan and many other polysynthetic language families). Likewise, while verbal moods may differ somewhat in their etymological basis, or may be sematically elaborated into subtypes in one variety but not another, the basic syntactico-semantic system of moods is the same, and moods with their associated person inflection always mark the endpoint of the polysynthetic word. A deeper question, which I will raise but cannot yet answer, is what drives this persistence? It cannot only be 'universal grammar' or 'functional exigency', given how uniformly atypical these languages are. A more likely answer is in the expressive and aesthetic canons that form the poetics of everyday discourse in Inuit-Yupik and in discourse that is especially admired.


Sally Rice

Tuesday, November 20

TALK 1: Variation writ small: A corpus-based account of variation within a Dene (Athapaskan) speech community

The wide-spread adoption and application of corpus linguistic techniques over the past 30 years and the interrogation of large and varied corpora of major world languages has revealed a number of general findings at odds with assumptions associated with most formal linguistic theories. Some of these findings (none of which are uncontroversial) include demonstrations that:

  • languages aren’t monolithic (modality-, genre-, and speaker-specific differences abound);
  • languages aren’t very compositional (instead, a large number of fixed expressions and pre-compiled multi-word units lacking any particular structural status or straightforward semantic analyzability make up, rather idiomatically, the bulk of utterances (in spoken language) and sentences (in writing);
  • languages aren’t very static (diachronic corpora of even a time depth of 20 years can reveal appreciable changes in linguistic patterning beyond the merely lexical);
  • frequency trumps well-formedness (frequency is a very important contributor to phonological, morphological, and even syntactic variation within a community across like items, some of which resist leveling);
  • sequential order may be more of a factor in structuring language than constituency (hierarchical) relations;
  • most generalizations hold at only a low-level of patterning.

This talk begins with a brief survey of some of the major lessons that corpus linguistics has provided us based on analysis of English, a “mega” language that functions at all levels of society, is inter-generationally transmitted, and exists robustly as speech, instant-messaging, and writing. I move quickly to a consideration of “micro” oral languages, especially polysynthetic ones such as the Athapaskan (Dene) language, Dene Sųłiné (ISO 639-3: chp), spoken across the three Canadian prairie provinces and the NWT, which prove to be no exception to the findings itemized above. Just possibly, oral languages may help us understand the role that variation plays within a speech community, since the lack of a conservativizing and homogenizing literary tradition in these languages allows linguists to attend to differences that may be otherwise dismissed in written languages as performance quirks or errors. The narrative arc that I wish to convey in this talk is that in oral, polysynthetic languages, variation in form, pronunciation, and meaning is the norm. How successive generations of variability might have accrued, leading to full-blown language differentiation, is the subject of the subsequent talk.


Tony Woodbury

Wednesday, November 21

TALK 2: Chatino tonal phonology and morphology : Small-scale variation leading to extreme typological diversity

The Chatino family, spoken in Oaxaca, Mexico, consists of just three closely related languages, one of which--Eastern Chatino--is divided into about 15 quite distinct but still more or less mutually intelligible single-village-level varieties. Following the set of general themes described in the introduction, the basic nature of Chatino tonal morphology and phonology will be presented via San Marcos Zacatepec Eastern Chatino (ZAC), a variety that in most respects the most tonally conservative within Eastern Chatino as well as for the family as a whole. ZAC has five distinct tones, and these tones form about 15 distinct tonal sequences or melodies consisting of zero (toneless) to four tones in a certain order. Every non-compound word links autosegmentally to one of these melodies, similar to what has been described for West African languages like Tiv or Mende. The ZAC melodies are "morphemes" (morphological formatives) and they have a range of functions, including the marking of person, number, aspect, and inflectional class; indication of lexical category (e.g., some occur only occur on nouns, some only on verbs), provenance (some are particular to Spanish loans); and expressive value (one occurs only with a few nouns that refer to cute or little things like babies or small animals).

Taking ZAC as a starting point, it is evident that relatively minor changes in other Chatino varieties have had outsized typological consequences. In most other Eastern Chatino varieties--more or less mutually intelligible with ZAC--non-word-final morae are lost, leading to monosyllabification of stems, together with a crushing of the 15 tone sequences into paradigmatically-contrasting cognate single tones, which phonetically speaking may distinguish up to 5 level tones as well as variously contrasting rises, falls, and complex contours. Typologically this resembles what has been been described for various East Asian and certain African "contour tone" languages, but it is functionally different in that the tones still retain their cognate morphological value as markers of inflectional and lexical categories. In a subset of Eastern Chatino varieties (including ZAC), the tonal value of one word 'spreads' to the next word or words if those words are toneless, as has been described extensively for Bantu languages; whereas in others this does not happen. In another (orthogonal) subset of Eastern Chatino varieties, a 'floating tone' that is not heard in pronunciations of a word in isolation passes to the next word under certain circumstances, occluding or changing that word's original tone; this same thing has been described for some Chinese languages as well as some Mixtec languages. Thus there are varieties of Eastern Chatino with both spreading and floating tones; those with only one of the two; and those with none, giving entirely sandhi-free, stable phrasal tonology such as has been described for languages like Thai or Triqui. Meanwhile, in yet another monosyllabic Eastern Chatino variety (Panixtlahuaca), the tonal system has simplified somewhat but the tones that remain divide into two macro-categories, one provoking a systematic change of the vowels to a 'high vowel register' and the other provoking a change of the vowels to a 'low vowel register', a phenomenon found in some Chadic languages. If the vowel is historically nasalized, however, the vowel quality is unaffected, but the high register surfaces with strong nasalization and the low one with very light nasalization. As a result, what appears as tonal ablaut in the marking of aspect in conservative Chatino varieties comes out in Panixtlahuaca as Indo-European style vocalic ablaut, and as a more typologically exotic nasal-grade ablaut. Moving outside the mostly mutually intelligible Eastern Chatino varieties to the two outlier languages, Tataltepec and Zenzontepec Chatino, polysyllabic structure is retained, but in one (Tataltepec) the melodies are truncated and in the other (Zenzontepec) the tones composing the melodies are reduced to just two distinct tones, H and M, a typologically highly unusual inventory because there is no L target.

A somewhat conservative thread that runs through this otherwise highly innovative state of affairs in Chatino is morphological function: even as tonal phonology varies sharply, the underlying morphological system is quite similar, marking the same inflectional and lexical categories across the family. As such, Chatino encourages typologists to take firmer measure of the morphological structuring of tone in formulating tonal typology, rather than primarily or exclusively considering phonetic and phonological characteristics. This more nuanced view may allow us to apprecate Chatino as presenting its own adventitious mix both of typological conservatism and typological innovation, just as we saw for Yupik-Inuit.  Finally, a more difficult question raised by Chatino, which I can pose but not answer, is why such exhuberant and diverse phonological change should occur from village to village; as well as how speakers of mutually intelligible but typologically very different varieties should be capable of discerning the common morphological outlines when communicating with speakers of other varieties, despite the very significant phonological and phonetic changes.


Sally Rice

Thursday, November 22

TALK 2: Variation writ larger: How an ethos of esoterogeny may have spawned linguistic innovation and differentiation across a language family

Some basic premises, originating with Sapir, about the Athapaskan (or Dene) language family include the following: (i) that it is exceptionally weird in its polysynthetic nature; (ii) that phonotactic and morphosyntactic considerations may have mitigated against borrowing across the family, thus providing an insulating and stabilizing effect; and (iii) that speakers of the four-to-five dozen daughter languages attested at the time of colonial contact may have been mutually intelligible until quite recently (late 19th century). Indeed, this last claim is all the more striking given the extensive and discontinuous geographic distribution of current Dene speech communities and the pronounced cultural and cosmological differences among distinct Dene peoples today. I will offer a different perspective on point (ii), that typological complexities in the language diminished the opportunities for contact effects to hasten diversification. Instead, I propose that group dynamics among the Dene traditionally––chiefly an insider-orientation or esoteric posture, fostered tolerance towards variation and innovation in the language.

Looking comparatively across the family, one definitely finds evidence of the “dialect continuum” that Sapir (and others) was referencing. Equally impressive, though, and less appreciated are the pronounced and often irregular phonological differences among even neighboring languages, as well as differences in the endolexicon (universal concepts that reference, for example, body parts and kin relations), in basic morphosyntax (TAM, agreement, valency marking, possession), not to mention sentence and discourse structure, including markers of subjectivity and evidentiality. Using a variety of sources, from published material to the author’s own comparative lexicon and field data from several Dene languages, ample evidence of heterogeneity in the family is provided. Whether this diversity (some of which––in terms of inflection and word order––qualifies for distinctiveness at the typological level) has resulted from the incremental accumulation of small-level, idiolectal and dialectal variation or is the result of attrition and language loss in the face of colonialization and globalization is beyond the scope of this presentation. However, because of known historical events in the geological and archaeological record, we can assert that the dispersion of Dene peoples in western North America has a relatively shallow time depth of 1-1.5 ka BP. Thus, diversification in the family could be taken as a test case for what happens when tensions between oralcy, an ethos of sociolinguistic opacity in the face of cultural adaptability, and opportunities afforded by climate change and migration conspire to produce what we see across the family today: exuberant differences built upon a common substrate.

  • Australian Government
  • The University of Queensland
  • Australian National University
  • The University of Melbourne
  • Western Sydney University