Summary and Keywords
Dene-Yeniseian is a proposed genealogical link between the widespread North American language family Na-Dene (Athabaskan, Eyak, Tlingit) and Yeniseian in central Siberia, represented today by the critically endangered Ket and several documented extinct relatives. The Dene-Yeniseian hypothesis is an old idea, but since 2006 new evidence supporting it has been published in the form of shared morphological systems and a modest number of lexical cognates showing interlocking sound correspondences. Recent data from human genetics and folklore studies also increasingly indicate the plausibility of a prehistoric (probably Late Pleistocene) connection between populations in northwestern North America and the traditionally Yeniseian-speaking areas of south-central Siberia. At present, Dene-Yeniseian cannot be accepted as a proven language family until the purported evidence supporting the lexical and morphological correspondences between Yeniseian and Na-Dene is expanded and tested by further critical analysis and their relationship to Old World families such as Sino-Tibetan and Caucasian, as well as the isolate Burushaski (all earlier proposed as relatives of Yeniseian, and sometimes also of Na-Dene), becomes clearer.
1. The Na-Dene family
Na-Dene includes some of the best-known indigenous languages of the Americas, such as Navajo in the American Southwest and Tlingit in the Alaskan Panhandle. The family, which was recognized early in the 19th century based on grammatical and lexical similarities, consists of two primary branches: Tlingit and Athabaskan-Eyak. Eyak, once spoken on the Alaskan coast near Yakutat, north of Tlingit areas, lost its last native speaker in 2009. The Athabaskan subfamily (alternatively spelled Athapaskan or Athabascan and increasingly referred to as “Dene,” after one version of the native word meaning “people”) occupies most of northwestern North America. There are also geographic outliers along or near the Pacific coastline of California, Oregon, and Washington. Finally, there is the Apachean subgroup, which includes Navajo, the language currently with the largest number of speakers, in the American Southwest.
Eyak is related equally to the entire Athabaskan group, which means that the family’s history involves three distinct comparative levels of reconstruction: the oldest is Proto-Na-Dene (PND, or PAET, meaning Proto-Athabaskan-Eyak-Tlingit), followed by Proto-Athabaskan-Eyak (PAE) and, finally, Proto-Athabaskan (PA).
Haida, a severely endangered language isolate spoken on British Columbia’s Queen Charlotte Islands (Haida Gwaii) in three dialectal forms, was earlier regarded as a member of Na-Dene by Boas, Sapir, and Greenberg, but this idea has now been largely abandoned. Lexical similarities between Tlingit and Haida are the result of language contact, and Haida shares with Na-Dene languages no genuine morphological homologies (word-building features that derive from a common ancestor). Many aspects of Proto-Na-Dene (or Proto-Athabaskan-Eyak-Tlingit, as it is sometimes called to underscore the exclusion of Haida) are now fairly well described, including hundreds of cognates in basic vocabulary showing interlocking sound correspondences (Krauss & Leer, 1981; Krauss, 2005; Leer, 2010; Nikolaev, 2014) and a great deal of complex shared morphology, most strikingly manifested in finite verb structure (Leer, 2010; Vajda, 2010a).
Distributed from the Alaskan Arctic (Koyukon, Gwitchin) to northern Mexico (Lipan Apache), the vast Athabaskan sub-branch of Na-Dene shows the broadest geographic spread of any indigenous language group in the Americas. The internal subgrouping of Athabaskan languages is still debated, a problem that might be resolved by tracing morphological innovations in verb structure across the various Athabaskan daughter languages. The time depth of Athabaskan has been estimated from a few thousand years to up to 10,000 (Kari, 2010), with most scholars favoring the more shallow estimate. Based on the rate of Tlingit and Athabaskan-Eyak cognate retention, the time depth of Na-Dene is probably at least 5,000 or 6,000 years old. The ancestors of Na-Dene speakers appear to represent a later, possibly Early Holocene, migration into the Americas in contrast to speakers of the unrelated language families to the south, whose ancestors may have arrived as early at 15,000 years ago. The Na-Dene migration seems to have preceded the arrival of Eskimo-Aleut speakers, who later populated the North American tundra (Scott & O’Rourke, 2010). A plausible Na-Dene homeland, based on the Age-Area Hypothesis, would be somewhere in the area from interior Alaska to interior British Columbia, near Eyak and Tlingit.
Superb descriptions of a number Na-Dene languages have been published, along with a great deal of high quality historical investigation of the family, especially its Athabaskan sub-branch (see Further Reading).
2. Yeniseian languages
The Yeniseian microfamily of central Siberia is today represented by the three surviving Ket dialects (Southern, Central, and Northern Ket), which are mutually intelligible and show only by minor lexical and phonological divergence. The Yugh language, sometimes regarded as the Sym dialect of Ket (named for a river near which it was spoken), is related to the surviving Ket dialects at a slightly deeper time depth. Yugh lost its last fluent speaker in the 1970s after being extensively documented by Heinrich Werner (Werner, 1997, 2012). Ket-Yugh clearly forms a primary branch of the family: its shared innovations include the development of verb–initial subject prefixes (often becoming clitics in Ket) that supplemented the largely eroded original preroot subject prefixes inherited from Proto-Yeniseian (see example set (1) below). A second primary branch contains Kott, for which the great Finnish field linguist M. A. Castrén recorded extensive dictionary materials and grammatical paradigms in the mid-19th century while working with the language’s five last-known native speakers (Castrén, 1858). An assortment of travelers during the 18th century likewise documented vocabulary lists representing Ket, Yugh, and Kott in several dialectal forms, along with the only known attestations of tree other Yeniseian languages: Assan, Arin, and Pumpokol. The complete 18th-century documentation of Yeniseian languages, which is of great significance for Yeniseian comparative linguistics, is readily available in Werner (2005).
Assan is closely related to the Kott dialects and together with Kott forms a second primary branch of the Yeniseian family. Ket-Yugh and Kott-Assan differ in a host of ways, most strikingly by the presence of verb–final subject person and number agreement suffixes in Kott-Assan (2), which follow the animate-class subject plural agreement suffix inherited from Proto-Yeniseian. The abbreviations are: sbj (subject agreement, inan.obj (inanimate-class object agreement), sg (singular), pl (plural), and 1, 2, 3 signifying 1st, 2nd, and 3rd person:
‘I made it.’
‘I made it.’
‘You (sg.) made it.’
‘You (sg.) made it.’
‘He made it.’
‘He made it.’
‘We made it.’
‘We made it.’
‘You (pl.) made it.’
‘You (pl.) made it.’
‘They made it.’
‘They made it.’
The sparse 18th-century word lists documenting Arin and Pumpokol show that these Yeniseian languages differ significantly from one another as well as from Ket-Yugh and Kott-Assan. The two are best regarded as a third and fourth primary branch of the family, though morphological evidence suggests Arin is closer to Ket-Yugh than Kott-Assan. Arin shares the use of the 1st person singular possessive prefix b- with Ket-Yugh (Ket b-qu’s, Arin bi-qus “my tent”), and both branches sometimes use the same morpheme verb-internally as a 1st person singular agreement marker as well (Ket bo-ɣatn, Arin ba-xatum “I go”). This suggests that Arin may have branched off from the ancestor of Ket-Yugh. However, even if it this did occur, the separation was early enough that Arin could just as easily be regarded as a separate primary branch for purposes of reconstructing Proto-Yeniseian lexicon. Pumpokol appears to share no significant innovations with any of the other Yeniseian languages, though the paucity of attested material leaves the question somewhat open. The sole documented Pumpokol conjugated verb paradigm fragment reveals the spread of predicate agreement suffixes to the finite verb, unlike either Ket-Yugh-Arin or Kott-Assan. The abbreviations in (3) include pron (personal pronoun), tc (thematic consonant, here a derivational prefix probably originally meaning “up” or “upright”), pres (present tense), and intrans (intransitive, meaning that there is a subject but not an object).
(3) Pumpokol finite verb paradigm
[i]t-š-a-diŋ-gu [? dissim. *gu > du ?]
you (sg) stand’
‘you (pl) stand’
Predicate agreement suffixes were inherited from Proto-Yeniseian into all the daughter branches (4–5) but appear to have been innovated as a form of finite verb subject agreement marking only in Pumpokol (3).
Ket predicate adjective agreement
Kott predicate adjective agreement
‘I am strong.’
‘I am strong.’
‘You (sg) are strong.’
‘You (sg) are strong.’
‘He is strong.’
‘He is strong.’
‘We are strong.’
‘We are strong.’
‘You (pl) are strong.’
‘You (pl) are strong.’
‘They (anim) are strong.’
‘They (anim) are strong.’
When Russian fur traders penetrated central Siberia in the early 17th century, the various Yeniseian-speaking tribes occupied wide areas of the middle reaches of the Yenisei and its tributaries.
Studies of substrate river names show that the primary branches of Yeniseian (including Arin) were already well differentiated across much of south-central Siberia before Turkic and Samoyedic pastoralists began to take over that territory more than a millennium earlier. Based on the highest dialectal diversity of substrate river names in a compact area (the Age-Area Hypothesis), as well as on recorded Ket folkloric evidence relating ancient migrations northward down the Yenisei to escape rival tribes, the original Proto-Yeniseian homeland is most likely to have been located between the southern tip of Lake Baikal, northern Mongolia, and the headwaters of the Yenisei.
3. Dene-Yeniseian lexicon
Vajda (2010a) proposed over 100 cognates in basic vocabulary between the Proto-Yeniseian forms in S. Starostin (1982, 1995) and the Proto-Na-Dene, Proto-Athabaskan or Proto-Athabaskan-Eyak forms reconstructed by Krauss or Leer (Krauss & Leer, 1981; Krauss, 2005; Leer, 2010). A few of the same comparisons appeared earlier in Trombetti (1923) or Ruhlen (1998).
The proposed cognates include body part terms (rump, foot, finger, head, stomach, belly, liver, kidney, wart), basic concepts (black, light, sun, hot, cloud/obscure/dark), basic actions (hit, poke, chop, rub, stretch, cook, lie down), as well as realia from boreal forest environment (wood, wolverine, wolf, squirrel, birch, willow, fallen snow, conifer pitch, conifer needles, spruce grouse), and possibly reflecting animistic hunter-gatherer lifestyle (shaman/cure by singing, sled runner, boat/vessel, dog), though the last group includes some of the phonologically more questionable matches. Significantly, shared kinship terms are virtually absent (conceivably connected somehow with the fact that the Athabaskans are matriarchal, while the Kets are strongly patriarchal). Numeral words are also not shared, which is not surprising given the lack of shared numerals between Tlingit and Athabaskan-Eyak.
At present, the Dene-Yeniseian lexical corpus remains thin on quantity and some of the individual comparisons have been challenged (Campbell, 2011, 2013; Starostin, 2012); see Vajda (2011, 2013) for a (mostly positive) response to these critiques. Campbell’s rejection of cognates based on alleged onomatopoeia may be overstated, at least in some cases. For example, Campbell (2013) rejects the comparanda for “spruce grouse” (Ket dīˑt ~ dek-, Proto-Athabaskan dex) based on this reasoning although the call of the spruce grouse in fact bears no resemblance to the root in question. Vajda’s (2010a) preliminary search for Dene-Yeniseian cognates was intentionally limited to basic vocabulary items that happened to have been reconstructed in each individual family. Also, almost no semantic latitude was allowed in most of the original comparisons. This technique was followed to reduce the potential for coincidental matches that would skew the initial search for sound correspondences by mixing “substantial chaff into the wheat.” The result is that too few potential cognates have yet been proposed to confidently defend every aspect of the sound correspondence system that has emerged so far from the comparison. However, the technique did achieve a firm basis to build upon. If the Dene-Yeniseian connection is valid, the list of proposed lexical homologies between the two families should substantially increase once more comparanda are analyzed.
4. Dene-Yeniseian sound correspondences
The sound correspondences between Proto-Na-Dene and Proto-Yeniseian proposed in Vajda (2010a) did not follow earlier Sino-Caucasian phonological comparisons (Starostin, 1982) but attempted to discern patterns based solely on cognates in Dene-Yeniseian lexicon. The result is more of a working hypothesis rather than a firm system unlikely to be modified significantly in the future. The comparison generally used the Proto-Yeniseian forms reconstructed by Starostin (1982, 1995). However, in a few instances it veered from that system, citing new insights into Common Yeniseian morphology or adopting a different interpretation of the comparative Yeniseian data. Below are only a few aspects of the proposed correspondences, the full system of which can be found in Vajda (2010a).
Modern Ket and Yugh have four phonemic tonal contrasts in monosyllabic words: high, abrupt rising (checked by laryngealization in the vowel’s second phase), rising-falling on a geminate vowel, and falling.
(6) Tones in Ket
(closed or open)
tense [e, F, o]
(closed or open)
lax [E, √, ç]
(closed or open)
lax [E, √, ç]
lax [E, √, ç]
The origin of the latter two tones is already suggested by Yeniseian-internal lexical comparisons. The Ket-Yugh rising-falling tone often corresponds to a disyllable in Kott where the loss of some sort of intervocalic guttural consonant (probably *k, *q, or *h) has left the geminate-vowel monosyllable in Ket and Yugh.
(7) Yeniseian-internal cognates for “snow sled”
The falling tone in Ket regularly corresponds to a Yugh long vowel with heavy pharyngealization in its second phase. Cognates in Na-Dene, where available, show that the pharyngealized Yugh element corresponds to a fricative or other consonant in an originally complex coda (or reduced second syllable) that has yielded tone in Yeniseian.
(8) Three Ket-Yugh falling tone words and their pre-Proto-Athabaskan cognates
The origin of the two other Yeniseian tonemes—the high and abrupt tones, which account for more than 75% of monosyllables in Modern Ket—cannot be determined by Yeniseian-internal reconstruction. However, external comparison shows that the high tone in Yeniseian corresponds to a Na-Dene syllable ending in a glottalized coda, while the Yeniseian abrupt tone correlates with a Na-Dene nonglottalized coda.
(9) Yeniseian tones and Na-Dene coda glottalization
Yeniseian languages lack a phonemic distinction between glottalized and plain obstruents. The differing tonal reflexes in Yeniseian before original Dene-Yeniseian glottalized vs. plain obstruents help connect the simplified Modern Ket consonant phoneme inventory through systematic sound correspondences with the much richer inventories of Na-Dene languages.
Another simplification that occurred in the Modern Ket dialects involves the merger of earlier affricates and palatals with plain dental stops. Yeniseian-internal comparisons show that the Proto-Yeniseian consonant inventory was larger than Modern Ket.
(10) Dental stop phonemes in Ket (all dialects) compared with Yugh
Ket (all dialects)
When cognates from Kott, Arin, and Pumpokol are considered and a comparison is made with potential Na-Dene cognates, the picture becomes clearer still. Proto-Yeniseian can be shown to have a series of palatal obstruents distinct from affricates or simple dental obstruents.
(11) Reflexes of proto-palatals vs. apical affricates in Yeniseian and Na-Dene
a. Proto-palatal series
b. Proto-apico-affricate series
The proto-forms in (10) follow Leer’s (2010) reconstruction. Nikolaev (2014) reconstructs different phonetic segments for Leer’s palatals and apico-alveolars, but the Dene-Yeniseian sound correspondences themselves still hold. More research needs to be done to reconcile these two phonological systems of Na-Dene reconstruction.
Rather than striving to amass a large quantity of evidence for Dene-Yeniseian, Vajda (2010a) highlighted instances where the external comparison handily explained idiosyncrasies in one or the other family in cases where family-internal analysis provided no insight. The explanatory value of the comparison for understanding such problems as the origin of Yeniseian tones was indicated as supporting the likelihood that the proposed genealogical link was valid. Dene-Yeniseian external comparison has proven similarly useful for understanding morphological idiosyncrasies in each family.
5. Evidence from nominal morphology
Morphological evidence for Dene-Yeniseian includes over two dozen bound morphemes (mostly affixes, but also cognates between body part nouns and postpositions). This material involves several distinct morphological systems (instrument noun creation, possessive constructions, action nominal derivation, finite verb structure) rather than simply being a list of phonologically short lookalikes of the type likely to be coincidental. Some of the data in this section and the next comes from the author’s ongoing work on Yeniseian morphological reconstruction and has not yet been published elsewhere.
Vajda (2013) compared Yeniseian and Na-Dene possessive morphology, which is used not only to express true possession (both alienable and inalienable) but also to build denominal postpositional constructions and adverbial expressions such as “toward the water’s edge,” “away from open space,” etc., that involve a closed set of directional roots. Some of the directional roots and postpositions themselves appear to be cognate between the two families. More of interest here, however, is that possessive, postpositional, and directional constructions in both families are all built using the same basic formula:
(12) Shared Dene-Yeniseian possessive morphology
POSSESSOR noun or pronominal + generic possessive affix ~*ŋw + POSSESSUM
What is key for showing genealogical relatedness is not the shared structure, which could be a typological coincidence, but rather the combination of cognate morphemes that it contains. The mysterious n-class nouns in Athabaskan basic vocabulary (which correspond to l-class nouns in Eyak) can be explained as vestigial retentions of the original possessive connector morpheme before an inalienably possessed noun. Idiosyncratic differences between the subject and object (or possessive) forms of pronouns can also be explained in this way in both families, and some of the phonological incongruities in pronoun correspondences between the two families are also resolvable by understanding the morphonological effects of the generic possessive nasal affix, vestigially present in both families.
Another morphological combination shared by the two families is their creation of instrument nouns by adding the suffix *-ɬ to verb roots. One such noun has been already examined above as part of the discussion on tonogenesis: Ket sùɬ, Yugh sùħɬ “holding hook (for baby’s cradle)” *suk “return, bent 180o, hook shaped” + “instrument suffix” *-ɬ. The same morpheme combination exists in pre-Proto-Athabaskan: *swǝq’ɬ “holding hook” *swǝq’ “hook shaped” + “instrument suffix” *-ɬ. Importantly, both root and suffix also occur separately in other combinations in both families, so that the morpheme division being proposed here is not arbitrary.
The morphological pattern of action nominal (gerund, infinitive) derivation from finite verb stems is also intricately shared across both families. Yeniseian contains a form meaning “VERB-ing,” “to VERB,” which serves parallel syntactic functions to English participles, infinitives, and gerunds. The Eyak language also has such a form, called a gerund, though only a few cognate traces of it are detectible in Athabaskan, and none so far have been found in Tlingit. Both Yeniseian and Eyak build these deverbal forms according to the same formula that includes a cognate prefix and suffix:
(13) Eyak and Yeniseian action nominal/gerund morphology
action nominal formula
cognate example forms
prefix *’is + VERB ROOT + suffix *-ɬ ~ l <*N
Eyak ’is-te:-l ‘lying’
prefix *si + VERB ROOT + suffix *-ǝŋ
Ket si-bagd-eŋ ‘stretching’
As with possessive constructions, the key here is the shared morphological cognates in combination, not the shared structural pattern itself. In finite verbs with derivational prefixes (Eyak qualifiers or Yeniseian thematic consonants), the derivational prefix regularly replaces the sibilant action nominal/gerund prefix. In Yeniseian, this prefix survived mainly in Kott simple forms such as ši-ten “lying,” ši-puk “stretching” (the Ket form si-bagd-eŋ “stretching” shows a rare retention of the prefix in Ket). The nasal suffix survives widely in Ket and Yugh where not eroded into falling tone but is vestigial in Kott-Assan (Assan taŋ-n “seeing” is a rare example of its survival). As with possessive constructions, the external comparison between homologous structures in Yeniseian and Na-Dene languages helps clarify the origin of unusual features in each family.
6. Verb morphology
Lexical cognates showing interlocking sound correspondences in quantities sufficient to garner universal support for Na-Dene (in the narrow sense of Athabaskan-Eyak-Tlingit) were achieved only recently (Campbell, 2011). Earlier indications of the validity of the family came rather from unmistakable homologies observable through casual inspection of the finite verb morphology. The intricate similarity of the three verb templates is striking:
(14) Generalized Proto-Athabaskan template (Vajda, 2010a, p. 38)
(newer derivational or thematic prefixes)
oldest prefix positions
verb stem (root + TAM suffix)
deictic or object pronominal prefixes
lexical ‘qualifier’ prefixes
tense mood aspect *s(ə)-, *ɢə-, *nə-
1st, 2nd person subj agr
pfv-stative prefix ñi-
class-ifier Ø, d- ɬ-, l-
(15) Eyak template, based on Krauss (1965, p. 171)
derivational or thematic prefixes of various sorts
oldest prefix positions
(root + TAM suffix)
deictic or object pron. prefixes
shape or anatomical prefixes, and other elements
tense, mood, aspect ɢə
1sg, 2sg, 2pl subject agr
tense moods s(ə)-
class-ifier ɬ- ~ Ø ɬə- ~ ɬi-də- ~ di-(i < stative prefix)
(16) Tlingit template, based on Leer (1991, pp. 147–149, 159)
deriv-ational or thematic prefixes of various sorts
oldest prefix positions
verb stem (root + tense-mood-aspect suffix)
suffixes and enclitics
deictic or object pro-nominal prefixes
tense mood aspect ɢa-ÿu- pfv
subject agr. (1 or 2 p)
classifier (i < stative prefix) ɬa- ~ ɬi-da- ~ di- sa- ~ si-, etc.
Special terms used traditionally by specialists in Na-Dene verb morphology include disjunct (the leftmost set of prefixes in Athabaskan, thought to be more recent additions to the verb complex), mode prefix or conjugation marker (prefix that changes to express different tense and mood forms), and classifier (an amalgam of three historically distinct prefixes found directly before the verb root, often used to signal various shades of transitivity). Other linguistic terms worth defining in this context are thematic (any part of the verb form that does not change with the grammatical context), perfective (completed action, abbreviated pfv), and imperfective (incomplete action, abbreviated imfv).
Vajda (in press) argues that metathesis and reanalysis have repeatedly affected the innermost layers of the Na-Dene and Yeniseian finite verb structure and suggests that these types of change may be historically more prevalent in complex templatic morphology than previously realized. Nonconcatenative morphological structures may be unusually useful to historical comparative linguistics precisely for this reason, since the most ancient layers change in ways that are still traceable by the comparative method, whereas the complete loss of templatic morphological material—at least in Dene-Yeniseian—has turned out to be surprisingly infrequent.
In Tlingit, Eyak, and Athabaskan, metathesis and reanalysis affected the conjugation prefixes in significant ways. The *s(ǝ)- conjugation prefix remained in its original slot in Athabaskan, while in Eyak it metathesized ahead of the 1st and 2nd person subject markers. Krauss’s (1965, p. 24) original comparison of Na-Dene templates suggested that the *s(ǝ)- conjugation prefix metathesized even farther rightward in Tlingit, giving rise to the s-classifier. Leer (2010), however, identified the Tlingit s-classifier as a completely different morpheme inherited from Proto-Na-Dene as an original member of the classifier complex, regarding the Athabaskan-Eyak conjugation prefix *s(ǝ)- as cognate to the Tlingit perfective prefix ÿu-. However, such a sound correspondence is not attested elsewhere. Contrary to Vajda (2010a, pp. 40–47), which followed Leer’s interpretation, the s-classifier appears to represent the metathesized conjugation prefix *s(ǝ)-, while the Tlingit perfective prefix ÿu- is instead cognate with the nasal element found in perfective forms of Athabaskan verbs and also with Proto-Yeniseian perfective *ŋʷ-. If this identification is valid, it supports the following reconstruction of the Proto-Na-Dene template, ancestral to the chronologically later Proto-Athabaskan, Eyak and Tlingit templates shown above in (14), (15), and (16), respectively:
(17) Hypothetical reconstruction of the Proto-Na-Dene verb template
object marking and various thematic prefixes
oldest prefix positions
aspect suffix *ɬ- impfv or *ŋʲi- pfv, etc.
deictic or object pro-nominal prefixes
tense mood prefix *si- *ɢa-
pfv prefix *ŋʷ-
1, 2 subject agr.
classifier ɬ- ~ ɬi- d- ~ di- (i < stative prefix *ji-)
The general trend in Na-Dene seems to be for tense/mood/aspect prefixes to metathesize rightward (Vajda, in press; Leer, 2006). Classifier elements identifiable as former tense/mood/aspect markers on the basis of Na-Dene internal evidence include the stative prefix (representing I-component of the classifier complex in all three Na-Dene daughter templates) and the S-component of the classifier in Tlingit.
Vajda (in press) reconstructs the following finite verb template of Proto-Yeniseian after taking into account a number of distinct metatheses that took place in Ket-Yugh but not in Kott-Assan:
(18) The Proto-Yeniseian template (Vajda, in press)
aspect l-impv n-pfv
pfv + stative
anim pl sbj agr
Just as observed with Na-Dene templates, the outright addition and loss of morpheme classes is surprisingly uncommon in the development of the Yeniseian daughter languages, and metathesis and reanalysis have instead been the major engines of change. Ket-Yugh and Kott-Assan have not completely lost a single inherited morpheme position, and each branch has innovated only a single new position: the subject affixes illustrated earlier in example sets (1) and (2).
Yeniseian and Na-Dene templates bear more than coincidental resemblance, since they involve combinations of cognate morphemes. A single template ancestral to both Na-Dene and Yeniseian could have arisen from an earlier, more analytical sequence of light and heavy verb, each with its own agreement prefixes and TAM (tense-mood-aspect) suffixes:
(19) Hypothesized genesis of Dene-Yeniseian verb morphology
subject agreement prefix
subject agreement prefix
3 inan. *w- 3 anim. or plural *d-
conjuga-tion prefix *si-, *ɢa-
aspect prefix continuative *ɬ- completive *ŋʷ-stative *j-
1, 2 subject agr prefix, 3 anim. or plural *d-
aspect suffix continuative *-ɬ completive *-ŋʷ stative *-j
This model would explain why aspectual categories are expressed on both sides of the verb root in both families, and why subject agreement, particularly plural marking, is prefixed before the conjugation markers as well as before the verb root. If this model is correct, then the template that evolved into Proto-Yeniseian lost its verb-final continuative suffix *-ɬ. And the Ł-component of the Na-Dene classifier would represent a metathesis rightward of continuative *-ɬ. In its new position after the subject markers, it resembled the homonymous instrumental affix and was reanalyzed as a valence-increase marker. This occurred only in the language ancestral to Na-Dene and provides one piece of evidence that Athabaskan-Eyak-Tlingit represents a unified family that can be related to Yeniseian only at a considerable time depth. Vestiges of the original continuative aspect-related function of *-ɬ can be seen in Athabaskan stative verbs such as Modern Navajo ɬi-zhin “it is black,” ɬi-kan “it tastes good.”
The D-component of the Na-Dene classifier likely originated from a 3rd person agreement subject prefix *du- “someone, people” (Thompson, 1996), the reflex of which in Tlingit is still in complementary distribution with the d-classifier. Vestiges of this morpheme survive in both families as a plural, distributive, or pluractional marker prefixed either before the conjugation markers or directly before the verb root—a dual distribution that becomes understandable in light of the ancestral bipartite structure proposed above. In both Yeniseian and Na-Dene, pronominal d- is involved in 1st person plural subject marking, as well as in expressions of a more generic animate-class plural. Proto-Na-Dene reanalyzed the d-pronominal as a valence-decrease prefix, probably based on its nonfocal subject usage in verb forms like “someone takes it” ~ “it gets taken.” In Yeniseian, pre-root pronominal d- atrophied and did not become a valence-decrease marker, though some archaic verb forms still contain it.
External comparison with Na-Dene also explains otherwise inexplicable irregularities in Modern Ket, Yugh, and Kott tense-mood-aspect marking patterns. The irregular Yugh transitive in (20), the full paradigm of which appears in Werner (1997, pp. 194–195), not only contains the original P-1 perfective aspect suffix but also preserves the original tense/mood alternation between the Dene-Yeniseian conjugation markers *si- and *ɢa-, which in these forms alone remained in position class 3 rather merging with the preceding object pronominals in position 4:
(20) Yugh verb conjugation fragment with archaic morphological features and numbers corresponding to the Proto-Yeniseian position classes shown in (18)
‘you (sg.) killed them’
‘you (pl.) kill them’
‘he kills them’
‘they killed them’
If a Proto-Dene-Yeniseian template did exist as hypothesized here, then the morpheme classes in its inherited core underwent different instances of reduction, metathesis, and reanalysis in each individual daughter branch. Most or all of these processes are still traceable as morphological innovations. For example, metathesis and reanalysis created the Na-Dene classifier system—a key innovation not shared by Yeniseian—though each classifier component does have a recognizable cognate in the Modern Ket verb complex.
All of the other verb morphologies of North and Inner Eurasia, as well as the American Arctic, are strongly suffixing rather than prefixing and offer a striking contrast to that of both Yeniseian and Na-Dene (Vajda, 2010a). The Na-Dene family itself was first detected mainly on the intricate evidence from shared verb structure, with cognate vocabulary only later accruing in sufficient quantity to fully demonstrate the relationship. The same scenario appears to be playing out with Dene-Yeniseian, where homologies in verb morphology are the most salient pieces of evidence, while identification of lexical cognates is proceeding at a slower pace.
The idea that the Ket people and their extinct relatives are somehow linked to Native Americans dates back to at least 1708, when the Dutch Orientalist Adriaan Reland speculated that the Ket and Kott people, whom local Russians called “Siberian Indians” because of their physical appearance and hunter-gatherer lifestyle (all other North Asians outside the Pacific Rim were pastoralists), were lost relatives of Native Americans who had remained in North Asia (Reland, 1708).
The first known proposal linking Ket and Kott specifically with Athabaskan and Tlingit (the core of what today is known as Na-Dene) was by Italian linguist Alfredo Trombetti (1923). Meanwhile, linguists in the Russian Federation began comparing Yeniseian with various Old World families and isolates. In 1982 Sergei Starostin, a comparativist working in Moscow, reconstructed the proto-Yeniseian sound system and linked the family to Sino-Tibetan and Caucasian (Abkhaz-Adyghe and Nakh-Daghestanian) in what is known as Sino-Caucasian (Starostin, 1982). Each of these connections had earlier been proposed separately, based on minimal comparative data: Yeniseian/Sino-Tibetan in 1892, Caucasian/Sino-Tibetan in the mid-19th century, and Caucasian-Yeniseian in the 1950s, if not earlier (for more on this historiography, see Vajda, 2001, 2010b; Bengtson, 2008). Starostin was the first to identify cognates supported by a system of proposed sound correspondences between the three reconstructed proto-families. Starostin’s colleague Sergei Nikolaev soon began comparing Caucasian to Na-Dene in North America (Nikolaev, 1991), expanding the proposed family into what has become known as Dene-Caucasian. In 1998 Merritt Ruhlen, an American collaborator of Joseph Greenberg (who had never examined Yeniseian data in any of his own writings), published 36 cognate sets between Yeniseian and Na-Dene (Ruhlen, 1998). Ruhlen followed Sapir and Greenberg and included Haida in Na-Dene, alongside Athabaskan, Eyak, and Tlingit. Like Trombetti, Ruhlen’s cognates were not supported by systematic sound correspondences. A few of them, without the Haida comparanda, were later supported in Vajda (2010a) as part of evidence offered for what has become known as the Dene-Yeniseian hypothesis. Unlike earlier attempts to link Yeniseian with other families, Vajda’s comparisons included both reconstructed morphological systems along with lexical cognates supported by systematic sound correspondences.
George Starostin, son of the late Sergei Starostin, who has continued his father’s work on Sino-Caucasian, now supports a genealogical relationship between Yeniseian and Na-Dene, but only within the context of a larger Dene-Caucasian family (Starostin, 2015). While Dene-Yeniseian and Dene-Caucasian are not contradictory hypotheses, resolving the issue of whether Dene-Yeniseian is indeed a language family (i.e., related more closely to one another than either is to any other known language) or simply two separate branches of a broader family that includes Basque, Caucasian, Burushaski, Yeniseian, Sino-Tibetan, and Na-Dene without actually forming a valid taxon within that family is of paramount concern for linguists working in this area. George Starostin regards Yeniseian and Burushaski (a language isolate spoken in northern Pakistan) as a valid subgroup (Starostin, 2015), while placing Na-Dene closest to Sino-Tibetan (echoing Edward Sapir’s earlier suggestion). See Vajda (2001, 2010b) for a full historiography of the many attempts to compare Yeniseian genealogically with other language families and isolates.
At present, Vajda’s Dene-Yeniseian proposal has received greater support in the West based mainly on its inclusion of morphological homologies between the two families, even though its lexical evidence is inferior in quantity (and sometimes arguably also in quality) to that so far presented for the original Sino-Caucasian proposal. John Bengtson (2008), an American colleague of George Starostin, has begun identifying possible grammatical homologies between the various proposed members of Dene-Caucasian, but these so far appear atomistic rather than indicative of a shared, inherited system. Part of the dissonance among scholars working in connection with Yeniseian comparative linguistics stems from differing degrees of emphasis placed on lexical as opposed to grammatical evidence. Another challenge is the uneven status of acceptance and reconstruction of the language families involved. Caucasian is not widely accepted as a valid genealogical grouping, being instead treated as two completely separate families (Abkhaz-Adyghe and Nakh-Daghestanian); so far, the proposed connection between these two families is supported solely by lexical evidence (Nikolaev & Starostin, 1994). The historical development of Burushaski, which exists in three closely related forms (Yasin, Hunza, and Nagar) with only a shallow time depth, remains poorly known. The linguistic history of Basque, which most linguists consider an isolate, is better known due to that language’s greater dialectal diversity and better analyzed historical records (Trask, 1997). Proto-Sino-Tibetan is only partly reconstructed (Matisoff, 2003). More is known about Old Chinese, which may represent a primary branch of that family (Starostin, 2015); consequently, external comparisons involving Sino-Tibetan have relied more heavily on comparanda favoring Chinese. Substantial Na-Dene lexical reconstructions became readily available to historical linguists working outside that family only very recently (Leer, 2010), and differing versions of the Proto-Na-Dene sound system still must be reconciled (Leer, 2010; Nikolaev, 2014). The core Yeniseian lexicon has been reconstructed with a high degree of reliability (Starostin, 1982, 1995), but Vajda’s planned monograph on Yeniseian historical morphology has not yet been completed and could add much to existing conceptions of Proto-Yeniseian.
8. Parallels from human genetics and folklore studies
Only comparative language data can demonstrate a linguistic relationship, and there are many examples of apparent mismatches among linguistics, culture, and genetics. However, parallel information from other disciplines may help delimit a language family’s most plausible location in time and geographic space. What is known about human genetics can suggest the likelihood (or extreme unlikelihood) of whether two geographically distant modern populations could have shared some sort of prehistoric connection within a time frame thought to be compatible with the possibility of identifying a genealogical linguistic connection or reconstructing a common ancestral language. In regard to the Dene-Yeniseian hypothesis, data from human genetics and folklore studies suggest three things: (1) the Yeniseians appear to be an ancient Siberian population and not a recent intrusion from southern Eurasia or a back-migration from Beringia; (2) Yeniseian peoples have distinct Late Pleistocene (approx. 12,000–14,000 years ago) connections with Native Americans, though not specifically with Na-Dene speakers; (3) nonlinguistic connections between Yeniseians and populations speaking languages belonging to other putative members of Dene-Caucasian (Burushaski, Caucasian, Sino-Tibetan) have either not been researched, do not appear to exist, or, in the case of human genetics, only suggest common ancestry at a much deeper time level (over 18, 000–20,000 years ago). Once again, none of this actually proves anything about language relatedness itself, but it does suggest something about the plausibility of that relationship and its possible time depth.
Although the time depth of the documented Yeniseian languages is probably less than 3,000 years, based on high rates of cognac in basic vocabulary items, evidence from human genetics suggest that at least the physical ancestors of today’s Ket people have been in Siberia for many thousands of years. The predominant Ket Y-DNA haplogroup (Y-chromosome type, passed on from fathers to sons) is Q1a, shared overwhelmingly with Native American males and rare in the Old World outside the central and south Siberian areas known from substrate place names to have been inhabited by Yeniseian speakers. Also, Ket Mt-DNA haplogroups (genetic type of the mother’s mitochondrial DNA, passed on by the mother to daughters and sons) include A1, shared distantly with Native Americans (the predominant Mt-DNA haplogroup among Na-Dene speakers and certain other indigenous northwestern North Americans is the related A2). But the Kets also show a uniquely high rate of U4 and F—found only in trace amounts among other North Asian populations and not at all in the Americas (Starikovskaya et al., 2005). Genetic studies of remains from two 9,000-year-old hunter-gatherer burial sites near the southern tip of Lake Baikal belonging to the Kitoi Culture have uncovered roughly the same or higher percentages of U4 and F in that ancient population (Schurr et al., 2010). The remaining Ket Mt-DNA haplogroups, H and C, likely derive from more recent intermarriage with west Siberian Uralic and east Siberian Tungusic pastoralist groups, respectively. The totality of their DNA profile strongly suggests that the Kets—or at least their physical ancestors—represent an ancient North Asian population rather than a recent intrusion from outside of Siberia (Flegontov et al., 2015).
Vajda (2012) hypothesized that the Yeniseians derive from an original ancestral population associated with the same Late Pleistocene Microblade Culture area from which the predominant Native American Y-DNA haplogroup Q1a3 and common North American Mt-DNA haplogroup A2 later arose and spread into the Americas. In the Early Holocene, Yeniseian-speaking ancestors in south Siberia seem to have intermarried with the Kitoi fisher folk or a similar population, adding Mt-DNA haplogroups U4 and F to their original genetic profile of Y-DNA Q1a and Mt-DNA A1. The historically documented Kets were exogamous and strongly patriarchal, with distinct summer riverine and upland hunting cultural components. This suggests the preservation of male genes linked with an ancestral language, despite persistent induction of female marriage partners from outside groups, eventually resulting in two stable Ket-speaking moieties: the Fire People and Water People (or Big-Ski-Pole-Ring People). The Fire People may have their ultimate origins in the originally male-oriented, Yeniseian-speaking, upland hunting ancestors of Dene-Yeniseians, while the Water People may echo an earlier female-oriented, linguistically non-Yeniseian, riverine group or groups that were inducted by marriage into ancient patriarchal Yeniseian society. This is speculative, yet nevertheless plausible.
Ket folklore shares motifs with northwestern North American populations, and the formerly Yeniseian-speaking areas of south Siberia share motifs specifically with the Na-Dene; see Berezkin (2015), who suggests that the ancestral Na-Dene were connected with the Late Pleistocene Microblade Dyuktai Culture of North Asia, which spread with them into Alaska and Northwestern North America by the Early Holocene. By contrast, there are no parallels from human genetics to support any special connection between Yeniseian speakers and North Caucasians or speakers of Burushaski within the past 15,000 years, and none known from folklore at all. The possibility that such parallels exist with early Sino-Tibetan speakers, some of whom do share the possibly Dene-Yeniseian haplotype combination of Q + A, as well as a plausible connection with North Asian Microblade cultures, has not yet been sufficiently researched.
Finally, there is also no evidence—either linguistic, folkloric, or from human genetics—to support the idea that Yeniseians represent a reverse migration from Beringia or Alaska back into Siberia. The Kets show certain genetic similarities with all Native Americans but none specifically with Na-Dene speakers (Scott & O’Rourke, 2010), so that a back-migration is unsupported by known evidence from human genetics. If the Ket ancestors were to have migrated back from Alaska to Siberia in the Early Holocene, their DNA would be expected to be closer to that of Na-Dene speakers, which is not the case. The model put forward by Holton and Sicoli (2014) suggesting that Yeniseian is linguistically closer to Eyak-Athabaskan than either is to Tlingit—and used as a basis to suggest that the Yeniseians were once located in or near Beringia alongside the Na-Dene ancestors—is not substantiated by any concrete lexical or grammatical comparanda. In fact, Na-Dene as a whole shows shared innovations that clearly set it apart from Yeniseian, notably in the classifier system of the finite verb.
9. Conclusions and future prospects
The Dene-Yeniseian hypothesis, because it involves families located on separate sides of Bering Strait, has generated considerable interest among linguists as well as from the general public. The combined evidence from shared morphological systems and cognates in basic vocabulary showing interlocking sound correspondences follows what is generally accepted as proof of a common origin. Parallel evidence from human genetics and folklore suggest a deeper time depth than has been demonstrated for most accepted language families. Mainstream historical linguistics, however, tends to support the idea of a temporal limit of less than 10,000 years for the preservation of inherited evidence of genealogical linguistic relationships, roughly corresponding to the oldest universally recognized language families (Campbell & Poser, 2008), while the Moscow School believes that basic vocabulary, though probably not grammatical systems, can be reconstructed to considerably deeper time depths (Starostin, 2015). Work on Dene-Yeniseian and, more broadly, on Dene-Caucasian is interesting methodologically in how its results seem to challenge these basic assumptions and equally interesting for how the assumptions themselves challenge its purported results. The painstaking comparative work of George Starostin and other Moscow School comparativists raises the real possibility that genealogical language relationships older than 10,000 years can be proven satisfactorily based solely on lexical comparisons that identify a sufficient number of cognates in basic vocabulary to posit interlocking sound correspondences. Even without accompanying grammatical evidence, a preponderance of cognates in basic vocabulary (but not in more superficial layers) would seem logically to have one possible explanation: descent from a common proto-language. Perhaps the term “language family” should be reserved for genealogical relationships demonstrated with lexical cognates, interlocking sound correspondences, and shared morphological systems, while the term “language macrofamily” could be used for cases demonstrated only by cognates in basic vocabulary showing interlocking sound correspondences but little or no shared grammatical homologies or morphological patterns. In general, macrofamilies would be expected to be older than families.
Under such a nomenclature, Dene-Yeniseian, with its combination of lexical and morphological evidence, would appear to be a “language family” of the old-fashioned variety, though probably a very old one. However, the work done so far on Dene-Yeniseian is distinctly out of the mainstream in challenging the assumption that complex morphological systems are not traceable over a deep expanse of time. Of all the evidence presented for Dene-Yeniseian in Vajda (2010a), the proposed homologies in finite verb structure are the most striking. However, they also seem to be the least plausible, since it is commonly assumed that complex morphology cannot survive across many thousands of years, and the Dene-Yeniseian connection, if valid, is likely to be at least 12,000 to 14,000 years old. There is clear evidence, however, that nonconcatenative stem structures can be extremely persistent over time. With concatenative morphology, grammatical systems are more linearly separated from basic lexical roots; therefore, grammar can change radically while the basic roots remain. Nonconcatenative stems, by contrast, themselves represent an integral feature of the basic vocabulary. If nonconcatenative stem morphology changes radically, all of the basic vocabulary must necessarily change with it—a fact that probably ensures its conservatism in societies that have it and do not undergo complete language shift. Therefore, it is not surprising if such structures remain over a deep span of time, alongside the many basic vocabulary items of which they form an intrinsic part.
There are plenty of examples where complex forms of nonconcatenative morphology are strikingly persistent within completely established language families, sometimes more obviously so than cognates in basic vocabulary, yet this fact has gone largely unremarked upon in the general literature on historical-comparative linguistics. Tlingit, Eyak, and Athabaskan themselves share intricate homologies in finite verb morphology that must be at least several thousand years old. Other language families with complex templatic verb morphologies evince the same sort of conservatism: Algic (which includes Algonquian and two Pacific Coast languages), Iroquoian, and Bantu offer three salient examples. The Afro-Asiatic family has preserved its distinctive nonconcatenative consonantal stem structure over a great period of time. The Semitic sub-branch of Afro-Asiatic has passed on its hallmark triconsonantal root pattern with 100% fidelity through more than 4,000 years of development despite cases of significant language mixing such as Maltese Arabic. If Dene-Yeniseian becomes accepted as a language family (or as part of a larger Dene-Caucasian family), its most significant contribution to historical-comparative linguistics may not be in the confirmation of a long-suspected language connection, but rather in the more general realization that nonconcatenative stem morphologies tend toward great conservatism and therefore represent a hitherto largely untapped resource for the comparativist.
Campbell, L. (1997). American Indian languages: The historical linguistics of Native America. Oxford: Oxford University Press.Find this resource:
Fortescue, M. (1998). Language relations across Bering Strait. London: Cassell.Find this resource:
Georg, S. (2007). A descriptive grammar of Ket. Part I: Introduction, phonology and morphology. Kent, U.K.: Global Oriental.Find this resource:
Hargus, S. (2007). Witsuwit’en grammar: Phonetics, grammar, morphology. Vancouver: University of British Columbia Press.Find this resource:
Hargus, S., & Rice, K. (Eds.). (2005). Athabaskan prosody. Amsterdam: John Benjamins.Find this resource:
Kari, J., & Potter, B. (Eds.). (2010). The Dene-Yeniseian connection. Fairbanks: Alaska Native Language Center.Find this resource:
Kotorova, E., & Nefedov, A. (Eds.) (2015). Comprehensive dictionary of Ket. Munich: Lincom Europa.Find this resource:
Leer, J. (1996). Comparative Athabaskan lexicon. Scan of handwritten materials available online: https://www.uaf.edu/anla/collections/search/resultDetail.xml?resource=10881&sessionId=&searchId=.Find this resource:
Mithun, M. (1999). The languages of Native North America. Cambridge, U.K.: Cambridge University Press.Find this resource:
Rice, K. (1989). A grammar of Slave. Berlin: Mouton de Gruyter.Find this resource:
Vajda, E. (2004). Ket. Munich: Lincom Europa.Find this resource:
Werner, H. (1997). Die ketische Sprache. Wiesbaden, Germany: Harrassowitz.Find this resource:
Werner, H. (2002). Vergleichendes Wörterbuch der Jenissej-Sprachen. Wiesbaden, Germany: Harrassowitz.Find this resource:
Young, R., & Morgan, W., Sr. (1988). The Navajo language: A grammar and colloquial dictionary. Albuquerque: University of New Mexico Press.Find this resource:
Young, R., & Morgan, W., Sr., with Midgette, S. (1992). Analytical lexicon of Navajo. Albuquerque: University of New Mexico Press.Find this resource:
Young, R. (2000). The Navajo verb. Albuquerque: University of New Mexico Press.Find this resource:
Bengtson, J. (2008). A comparative grammar of Dene-(Sino-)Caucasian languages. Aspects of Comparative Cinguistics, 3, 45–118.Find this resource:
Berezkin, J. (2015). Siberian folklore and Na-Dene origin. Arkheologija, etnografija i antropologija Evrazii, 43(1), 122–144.Find this resource:
Campbell, L. (2011). Review of The Dene-Yeniseian connection. International Journal of American Linguistics, 77(3), 445–451.Find this resource:
Campbell, L. (2013). Historical linguistics: An introduction (3rd ed.). Cambridge, MA: MIT Press.Find this resource:
Campbell, L., & Poser, W. (2008). Language classification: History and method. Cambridge, U.K.: Cambridge University Press.Find this resource:
Castrén, M. A. (1858). Versuch einer jenissei-ostjakischen und kottischen Sprachlehre. St. Petersburg: Imperatorskaja Akademija Nauk.Find this resource:
Flegontov, P. et al. (2015). Genomic study of the Ket: A Paleo-Eskimo-related ethnic group with significant Ancient North Eurasian ancestry. BioRxiv.Find this resource:
Holton, G., & Sicoli, M. (2014). Linguistic phylogenies support back-migration from Beringia to Asia. PLoS ONE, 9(3), e91722.Find this resource:
Kari, J. (2010). The concept of geolinguistic conservatism in Na-Dene prehistory. In The Dene-Yeniseian Connection (pp. 194–222). Fairbanks: Alaska Native Language Center.Find this resource:
Krauss, M. (1965). Eyak: A preliminary report. Canadian Journal of Linguistics, 10(2/3), 167–187.Find this resource:
Krauss, M. (2005). Athabaskan tone. In Athabaskan prosody (pp. 51–136). Amsterdam: John Benjamins.Find this resource:
Krauss, M., & Leer, J. (1981). Athabaskan, Eyak, and Tlingit sonorants. Fairbanks: Alaska Native Language Center.Find this resource:
Leer, J. (1991). The schetic categories of the Tlingit verb (Unpublished doctoral dissertation. University of Chicago.Find this resource:
Leer, J. (2006). Na-Dene languages. In Encyclopedia of Language and Linguistics (pp. 428–430). Oxford: Pergamon.Find this resource:
Leer, J. (2010). The palatal series in Athabascan-Eyak-Tlingit with an overview of the basic sound correspondences. In The Dene-Yeniseian Connection (pp. 168–193). Fairbanks: Alaska Native Language Center.Find this resource:
Matisoff, J. (2003). Handbook of Proto-Tibeto-Burman. Berkeley: University of California Press.Find this resource:
Nikolaev, S. (1991). Sino-Caucasian languages in America. In Dene-Sino-Caucasian languages (pp. 42–66). Bochum, Germany: Brockmeyer.Find this resource:
Nikolaev, S. (2014). Toward the reconstruction of Proto-Na-Dene. Journal of Language Relationship, 11, 103–124.Find this resource:
Nikolaev, S., & Starostin, S. (1994). A North Caucasian etymological dictionary. Moscow: Asterisk Press.Find this resource:
Reland, A. (1708). Dissertationum miscellanearum partes tres (part 3). Trajecti ad Rhenum: G. Broedelet.Find this resource:
Ruhlen, M. (1998). The origin of the Na-Dene. Proceedings of the National Academy of Sciences, 95, 13994–13996.Find this resource:
Scott, G., & O’Rourke, D. (2010). Genes across Bering Strait: A physical anthropological perspective on the Dene-Yeniseian connection. In The Dene-Yeniseian connection (pp. 119–137). Fairbanks: Alaska Native Language Center.Find this resource:
Schurr, T., Osipova, L., Zhadanov, S., & Dulik, M. (2010). Genetic diversity in Native Siberians: Implications for the prehistoric settlement of the Cis-Baikal region, Siberia. In Prehistoric hunter-gatherers of the Baikal region, Siberia (pp. 121–134). Philadelphia: University of Pennsylvania Museum of Archaeology and Anthropology.Find this resource:
Starikovskaya E., et al. (2005). Mitochondrial DNA diversity in indigenous populations of the southern extent of Siberia, and the origins of Native American haplogroups. Annals of Human Genetics, 69, 67–89.Find this resource:
Starostin, G. (2012). Dene-Yeniseian: A critical assessment. Journal of Language Relationship, 8, 115–138.Find this resource:
Starostin, G. (2015). K istokam jazykovogo raznoobrazija. Moscow: Delo.Find this resource:
Starostin, S. (1982). Praenisejskaja rekonstruktsija i vneshnie svjazi enisejskikh jazykov. Ketskij sbornik, 3, 144–237. Leningrad, Nauka.Find this resource:
Starostin, S. (1995). Sravnitel’nyj slovar’ eniseijskikh jazykov. Ketskij sbornik: lingvistika (pp. 176–315). Moscow, Shkola “Jazyki russkoj kul’tury.”Find this resource:
Thompson, C. (1996). The Na-Dene middle voice: An impersonal source of the D-element. International Journal of American Linguistics, 62, 351–378.Find this resource:
Trask, R. (1997). The history of Basque. London: Routledge.Find this resource:
Trombetti, A. (1923). Elementi di Glottologia. Bologna: Nicola Zanichelli.Find this resource:
Vajda, E. (2001). Yeniseian peoples and languages: A history of Yeniseian studies, with an annotated bibliography and a source guide. Surrey: Curzon Press.Find this resource:
Vajda, E. (2010a). A Siberian link with Na-Dene languages. In The Dene-Yeniseian Connection (pp. 33–99). Fairbanks: Alaska Native Language Center.Find this resource:
Vajda, E. (2010b). Yeniseian, Na-Dene, and historical-linguistics. In The Dene-Yeniseian Connection (pp. 100–118). Fairbanks: Alaska Native Language Center.Find this resource:
Vajda, E. (2011). A response to L. Campbell’s “Review of The Dene-Yeniseian connection.” IJAL, 77(3), 451–452.Find this resource:
Vajda, E. (2012). The Dene-Yeniseian connection: A reply to G. Starostin. Journal of Language Relationship, 8, 138–152.Find this resource:
Vajda, E. (2012). Geography, demography and time depth: Explaining how the Dene-Yeniseian connection is possible. (Video recording of talk given in Fairbanks, AK, on March 24, 2012 at the Dene-Yeniseian Symposium.) Retrieved from https://www.youtube.com/watch?v=7M0QnAqQUmw&list=PL32A13526827D90BCFind this resource:
Vajda, E. (2013). Vestigial possessive morphology in Na-Dene and Yeniseian. Working papers in Athabaskan (Dene) languages (pp. 79–91). Fairbanks, Alaska Native Language Center.Find this resource:
Vajda, E. (in press). Patterns of retention and innovation in templatic polysynthesis. In Handbook of polysynthesis. Oxford: Oxford University Press.Find this resource:
Werner, H. (1997). Das Jugische. Wiesbaden, Germany: Harrassowitz.Find this resource:
Werner, H. (2005). Die Jenissej-Sprachen des 18. Jahrhunderts. Wiesbaden, Germany: Harrassowitz.Find this resource:
Werner, H. (2012). Dictionary of the Yughl. Munich: Lincom Europa.Find this resource: