Dené–Caucasian languages

Dené–Caucasian
(controversial)
Geographic distribution:	scattered in Eurasia; northern North America
Linguistic classification:	Dené-Caucasian
Proto-language:	Proto-Dené–Caucasian
Subdivisions:	Na-Dené (sometimes including Haida) Caucasian Sino-Tibetan Yeniseian Burushaski (usually included) Vasconic (usually included) Tyrsenian (sometimes included) Sumerian (sometimes included) Almosan (rarely included)

The Dené–Caucasian language family is a proposed language superfamily containing at least the Caucasian, Yeniseian, Burushaski, Sino-Tibetan, and Na-Dené languages. The relationship among these languages and the existence of a Dené–Caucasian family is disputed or rejected by nearly all linguists, including Lyle Campbell,^[1] Ives Goddard,^[2] and Larry Trask.^[3][4]

Contents 1 History of the hypothesis 2 Evidence for Dené–Caucasian 2.1 Shared pronominal morphemes 2.2 Shared noun class pre- and infixes 2.3 Verb morphology 3 Family tree proposals 3.1 Starostin's view 3.2 Bengtson's view 4 Proposed subbranches 4.1 Macro-Caucasian 4.2 Karasuk 5 Footnotes 6 References 7 External links 8 See also

History of the hypothesis

Ideas similar to the Dené–Caucasian theory were proposed by Robert Bleichsteiner, Karl Bouda, E. J. Furnée, René Lafon, Edward Sapir, Robert Shafer, Morris Swadesh, Olivier Guy Tailleur, Vladimir N. Toporov, Alfredo Trombetti and other scholars of the early 20th century. Morris Swadesh included all the members of Dené–Caucasian as part of his "Vasco-Dene" (named for Basque and Navajo, the geographic extremes) in 1959, but this also included most other Eurasian languages and thus cannot be considered equivalent to Dené–Caucasian.^[5] Mary Haas attributes the Vasco-Dene hypothesis to Edward Sapir.

In the 1980s, Sergei Starostin, using strict linguistic methods (proposing regular phonological correspondences, reconstructions, glottochronology, etc.), became the first to put the idea that the Caucasian, Yeniseian and Sino-Tibetan languages are related on firmer ground.^[6]

In 1991, Sergei L. Nikolayev added the Na-Dené languages.^[7] Their inclusion has been complicated by the ongoing dispute over whether Haida belongs to the family. The proponents of the Dené–Caucasian hypothesis incline towards supporters of Haida's membership in Na-Dené, such as Heinz-Jürgen Pinnow^[8] or, most recently, John Enrico.^[9] Edward J. Vajda, who otherwise rejects the Dené–Caucasian hypothesis, has suggested that Tlingit, Eyak, and the Athabaskan languages are closely related to the Yeniseian languages, but he denies any genetic relationship of the former three to Haida.^[10] Vajda's ideas on the relationship of Athabaskan–Eyak–Tlingit and Yeniseian have found support independently in works of various authors, including Heinrich K. Werner^[11] or Merritt Ruhlen.^[12] DNA analyses have not shown any special connection between the modern Ket population and the modern speakers of the Na-Dené languages.^[13]

In 1996, John D. Bengtson added the Vasconic languages (including Basque, its extinct relative or ancestor Aquitanian, and possibly Iberian), and in 1997 he proposed the inclusion of Burushaski. The same year, in his article for Mother Tongue, Bengtson concluded that Sumerian might have been a remnant of a distinct subgroup of the Dené–Caucasian languages.^[14] However, two other papers on the genetic affinity of Sumerian appeared in the same volume: while Allan R. Bomhard considered Sumerian to be a sister of Nostratic, Igor M. Diakonoff compared it to the Munda languages.^[15]

In 1998, Vitaliy V. Shevoroshkin rejected the Amerind affinity of the Almosan (Algonquian-Wakashan) languages, suggesting instead that they had a relationship with Dené–Caucasian. Several years later, he offered a number of lexical and phonological correspondences between the North Caucasian, the Salishan, and the Wakashan languages, concluding that the latter two might represent a distinct branch of the former and that they must have separated after the break of the Avar-Andi-Tsezian unity in the period about the 2nd-3rd millennia BC.^[16]

Evidence for Dené–Caucasian

The existence of Dené–Caucasian is supported by:

• many words that correspond between some or all of the families referred to Dené–Caucasian;
• the presence in the shared vocabulary of words that are rarely borrowed or otherwise replaced, such as personal pronouns (see below);
• elements of grammar, such as verb prefixes and their positions (see below), noun class prefixes (see below) and case suffixes that are shared between at least some of the component families;
• a reconstruction of the sound system, the basic parts of the grammar, and much of the vocabulary of the superfamily's most recent common ancestor, the so-called Proto-Dené–Caucasian language.

Potential problems include:

• the somewhat heavy reliance on the reconstruction of Proto-(North-)Caucasian by Starostin and Nikolayev.^[17] This reconstruction contains much uncertainty due to the extreme complexity of the sound systems of the Caucasian languages; the sound correspondences between these languages are difficult to trace.
• the use of the reconstruction of Proto-Sino-Tibetan by Peiros and Starostin,^[18] parts of which have been criticized on various grounds,^[19] although Starostin himself has proposed a few revisions.^[17] All reconstructions of Proto-Sino-Tibetan suffer from the facts that many languages of the huge Sino-Tibetan family are underresearched and that the shape of the Sino-Tibetan tree is poorly known and partly controversial;
• the use of Starostin's reconstruction of Proto-Yeniseian rather than the competing one by Vajda or that by Werner;^[11]
• the use of Bengtson's reconstruction of Proto-/Pre-Basque rather than Trask's;
• the slow progress in the reconstruction of Proto-Na-Dené, so that Haida and Athabaskan–Eyak–Tlingit have so far mostly been considered separately.

Shared pronominal morphemes

Several roots can be reconstructed for the 1st and 2nd person singular pronouns. This may indicate that there were pronouns with irregular declension (suppletion) in Proto-Dené–Caucasian, like "I" vs "me" throughout Indo-European. In the presumed daughter languages some of the roots are often affixes (such as verb prefixes or possessive noun prefixes) instead of independent pronouns.

The Algic,^[20] Salishan, Wakashan,^[16] and Sumerian comparisons should be regarded as especially tentative because regular sound correspondences between these families and the more often accepted Dené–Caucasian families have not yet been reconstructed. To a lesser degree this also holds for the Na-Dené comparisons where only a few sound correspondences have yet been published.

/V/ means that the vowel in this position has not been successfully reconstructed, /K/ could have been any velar or uvular plosive?, /S/ could have been any sibilant or assibilate?.

All except Algic, Salishan and Wakashan are taken from Bengtson (2008).^[21]

Meaning	Proto-Dené–Caucasian	Proto- Basque	Proto- Caucasian	Proto- Burushaski	Proto- Sino-Tibetan	Proto- Yeniseian	Na-Dené	Proto- Salishan	Proto- Algic	Sumerian
1st sg.	/ŋV/	/ni/, /n/-	/nɨ/[1]	/a/-	/ŋaː/-	/ŋ/		/nV/	/nˀV/-	/ŋa(e)/[2]
	/d͡zV/	-/da/-, -/t/	/zoː/	/d͡ʑa/		/ʔad͡z/	[3]	-/t͡s(a)/-, -/s/[4]
	/KV/	/gu/[5], /g/-			/ka/-		[6]
2nd sg.	/KwV/	/hi/, /h/-, -/ga/-[7]	/ʁwVː/	/gu/-~/go/-	/Kwa/-	/(V)k(V)/	[8]	/ʔaxʷ/	/k̕V/-
	/u̯Vn/	-/na/-[9]	/u̯oː-n/	/u-n/	/na-(ŋ)/	/ʔaw/	[10]	/wV/
3rd sg.	/w/- or /m/-	/be-ra/	/mV/	/mu/-[11]	/m/-	/wV/	[12]
2nd pl.	/Su/	/su/, /s/-	/ʑwe/							/t͡sa(e)/[13]

Footnotes: ¹ On Caucasian evidence alone, this word cannot be reconstructed for Proto-Caucasian or even Proto-East Caucasian; it is only found in Lak and Dargwa. (Bengtson 2008:94); ² The final /e/ found in Sumerian pronouns is the ergative ending. The Emesal dialect has /ma(e)/; ³ Proto-Athabaskan */ʃ/, Haida dii /dìː/; ⁴ Also in Proto-Southern Wakashan; ⁵ 1st pl.; ⁶ Tlingit xa /χà/, Eyak /x/-, /xʷ/; ⁷ Masculine verb prefix; ⁸ Proto-Athabaskan */χʷ/-, Tlingit ÿi /ɰi/ > yi /ji/ = 2nd pl.; Tlingit i /ʔì/, Eyak /ʔi/ "thou"; ⁹ Feminine verb prefix; ¹⁰ Proto-Athabaskan */ŋ̰ən/-, Haida dang /dàŋ/, Tlingit wa.é /waʔɛ́/, where the hypothesis of a connection between the Proto-Athabaskan and Haida forms on the one hand and the rest on the other hand requires ad hoc assumptions of assimilation and dissimilation (Bengtson 2008: 94); ¹¹ Feminine; ¹² Proto-Athabaskan */wə/-, Eyak /wa/-, Tlingit wé /wɛ́/, Haida 'wa /wˀà/; ¹³ 2nd sg.

Shared noun class pre- and infixes

Noun classification occurs in the North Caucasian languages, Burushaski, Yeniseian, and the Na-Dené languages. In Basque and Sino-Tibetan, only fossilized vestiges of the prefixes remain. One of the prefixes, */s/-, seems to be abundant in Haida, though again fossilized.

The following table with its footnotes, except for Burushaski, is taken from Bengtson (2008).^[21]

Proto-Dené–Caucasian	Proto-Basque [a]	Proto-Caucasian [b]	Burushaski [c]	Proto-Sino-Tibetan [d]	Ket [e]
/u̯/-	/o/-, /u/-	I /u̯/-	/u/-		/a/, /o/
/j/	/e/-, /i/-	II /j/-	/i/-	/g/- (?)	/i/, /id/
/w/	/be/-, /bi/-	III /w/-, /b/- (/m/-)		/b/-, /m/-	/b/
/r/		IV /r/-, /d/-		/r/-, /d/-
/s/	-/s/-	(-/s/-)		/s/-

Footnotes: ^a In Basque, the class prefixes became fossilized. ^b In many Caucasian languages (28), systems of this type more or less persist to this day, especially in the East Caucasian languages, whereas in West Caucasian, only Abkhaz and Abaza preserve a distinction human-nonhuman.^[22] The Roman numbers are those conventionally used for the East Caucasian noun classes. The forms in parentheses are very rare. ^c Burushaski seems to have reversed the first two animate classes,^[23] which may have parallels in some East Caucasian languages, namely Rutul, Tsakhur, or Kryz. ^d As with Basque, the class system was already obsolete by the time the languages were recorded.^{[24] e} Objective verb prefixes; /a/ and /i/ are used in the present tense, /o/ and /id/ in the past.

Verb morphology

In general, many Dené–Caucasian languages (and Sumerian) have polysynthetic verbs with several prefixes in front of the verb stem, but usually few or no suffixes. (The big exceptions are East Caucasian, where there is usually only one prefix and many suffixes, the similarly suffixing Haida, and Sino-Tibetan, for which little morphology can so far be reconstructed at all; Classical Tibetan with its comparatively rich morphology has at most two prefixes and one suffix. In Burushaski, the number of suffixes can surpass the rather large number of prefixes.)

The following is an example of a Kabardian (West Caucasian) verb from Bengtson (2008:98):^[21]

Kabardian orthography	вадыхэзгъэхьамэ
IPA	/waːdəçɐzʁɐħaːmɐ/
Analysis	/w/-	-/aː/-	-/də/-	-/çɐ/-	-/z/-	-/ʁɐ/-	-/ħ/-	-/aː/-	-/mɐ/
Position	–6	–5	–4	–3	–2	–1	0	+1	+2
	direct object	indirect object	comitative	locative	subject	causative	verb stem	tense	conditional
in this case:	2nd singular	3rd plural	"with"	"in"	1st singular	"make"	"enter"	past	"if"
Translation	if I made you go in together with them

Bengtson (2008) suggests correspondences between some of these prefixes (sometimes suffixes) and between their positions.

For example, a preverb /t/- occurs in Yeniseian languages and appears in position –3 (Ket) or –4 (Kott) in the verb template (where the verb stem is in position 0, suffix positions get positive numbers, and prefix positions negative numbers). In Burushaski, a fossilized preverb /d/- appears in position –3. In Basque, an element d- appears in position –3 of auxiliary verbs in the present tense unless a first or second person absolutive agreement marker occupies that position instead. The Na-Dené languages have a "classifier" /d/- (Haida, Tlingit, Eyak) or */də/- (Proto-Athabaskan) that is either fossilized or has a vaguely transitive function (reflexive in Tlingit) and appears in position –3 in Haida. In Sino-Tibetan, Classical Tibetan has a "directive" prefix /d/-, and Nung has a causative prefix /d/- (positions do not apply because Sino-Tibetan verbs have at most two prefixes depending on the language).

A past tense marker /n/ is found in Basque, Caucasian, Burushaski, Yeniseian, and Na-Dené (Haida, Tlingit and Athabaskan); in all of these except Yeniseian, it is a suffix or circumfix, which is noteworthy in these (with the exception of East Caucasian and Haida) suffix-poor language families.

Another prefix /b/ is found in some Sino-Tibetan languages; in Classical Tibetan it marks the past tense and precedes other prefixes (if any). It may correspond to the Tlingit perfect prefix wu-/woo- /wʊ, wu/, which occurs in position –2, and the fossilized Haida wu-/w- /wu, w/ which occurs in verbs with "resultative/perfect" meanings.

"There are also some commonalities in the sequential ordering of verbal affixes: typically the transitive/causative *s- is directly before the verb stem (–1), a pronominal agent or patient in the next position (–2). If both subject/agent and object/patient are referenced in the same verbal chain, the object typically precedes the subject (OSV or OVS [where V is the verb stem]: cf. Basque, West Caucasian [see table above], Burushaski, Yeniseian, Na-Dené, Sumerian templates […]. [Footnote: "Alone in N[a]-D[ene] Eyak allows for subjects and objects in a suffix position."] In Yeniseian (position –5) [...] and Na-Dene (position –5) [...] noun stems or (secondary) verb stems can be incorporated into the verbal chain." (Bengtson 2008:108)

The mentioned "transitive/causative" */s/- is found in Haida, Tlingit, Sino-Tibetan, Burushaski, possibly Yeniseian ("an 'empty' morpheme occupying the position of object in intransitive verbs with an animate subject"; Bengtson 2008:107) and maybe in Basque. A causative suffix *-/s/ is found in many Nostratic languages, too, but its occurrence as a prefix and its position in the prefix chain may nevertheless be innovations of Dené–Caucasian.

Family tree proposals

Starostin's view

The Dené–Caucasian family tree and approximate divergence dates (estimated by modified glottochronology) proposed by S. A. Starostin and his colleagues from the Tower of Babel project:^[25]

1. Dené–Caucasian languages [8,700BCE]

1.1. Na-Dené languages (Athabascan-Eyak-Tlingit)

1.2. Sino-Vasconic languages [7,900BCE]

1.2.1. Vasconic (see below)

1.2.2. Sino-Caucasian languages [6,200BCE]

1.2.2.1. Burushaski

1.2.2.2. Caucaso-Sino-Yeniseian [5,900BCE]

1.2.2.2.1. North Caucasian languages

1.2.2.2.2. Sino-Yeniseian [5,100BCE]

1.2.2.2.2.1. Yeniseian languages

1.2.2.2.2.2. Sino-Tibetan languages

Bengtson's view

John D. Bengtson groups Basque, Caucasian and Burushaski together in a Macro-Caucasian (earlier Vasco-Caucasian) family (see the section on Macro-Caucasian below).^[26] According to him, it is as yet premature to propose other nodes or subgroupings, but he notes that Sumerian seems to share the same number of isoglosses with the (geographically) western branches as with the eastern ones:^[27]

1. Dené–Caucasian

1.1. The Macro-Caucasian family

1.1.1. Basque

1.1.2. North Caucasian

1.1.3. Burushaski

1.2............................................ (Sumerian?)

1.3. Sino-Tibetan

1.4. Yeniseian

1.5. Na-Dené

Proposed subbranches

Macro-Caucasian

John Bengtson (2008)^[21] thinks that, within Dené–Caucasian, the Caucasian languages form a branch together with Basque and Burushaski, based on many shared word roots as well as shared grammar such as:

• the Caucasian plural/collective ending *-/rV/ of nouns, which is preserved in many modern Caucasian languages, as well as sometimes fossilized in singular nouns with collective meaning; many Basque nouns with a collective meaning end in -/rː/, and one of the many Burushaski plural endings for class I and II (masculine and feminine) nouns is -/aro/. However, such a plural ending is also widespread in the Nostratic languages, e.g. Japanese particles#ra
• the consonant -/t/, which is inserted between the components of some Basque compound nouns and can be compared to the East Caucasian element -*/du/ which is inserted between the noun stem and the endings of cases other than the ergative.
• the presence of compound case endings (agglutinated from the suffixes of two different cases) in all three branches.
• the case endings themselves:

Likely cognates of case endings
Basque Case	Basque	Burushaski	‹The template Dab button is being considered for deletion.›  [//toolserver.org/~dispenser/cgi-bin/dab_solver.py?page=Den%C3%A9%E2%80%93Caucasian_languages&editintro=Template:Disambiguation_needed/editintro&client=Dab_button&fixlinks=North+Caucasian Caucasian]	Comments
Absolutive	-0	-0	-0	The absolutive form is generally used for the subjects of intransitive verbs and the direct object of transitive verbs. Special ergative forms are used for the subject of transitive verbs.
Ergative	-k	-k/-ak(1)	-k’ə(2)	(1) instrumental, occurs only with certain nouns and with verbs meaning "strike" or "shoot"; (2) West Caucasian only: Kabardian ergative, Adyghe instrumental
Dative	-i	-e(1)	*-Hi(2)	(1) used as both ergative and genitive, except for feminine nouns which have a different genitive ending; (2) East Caucasian only; manifests as Avar -e (dative), Hunzib -i (dative) etc., shifted to instrumental in Lak, Dargwa, genitive in Khinalug, or ergative in the Tsezian languages, Dargwa and Khinalug; */H/ is any glottal or epiglottal consonant
Instrumental	-z /s/	-as/-áas(1)	*-sː-(2)	(1) cf. parallel infinitive -s in some Lezghian languages; (2) instrumental animate; general attributive, shifted to closely related functions in most modern languages, e.g. ergative animate in Chechen, adjectival and participial attributive suffix in Lak, dative and infinitive in Lezgi, transformative/adverbial case in Abkhaz, etc.
Genitive	-en(1)		*-nV(2)	(1) possibly also the locative/inessive ending -n; (2) attested as genitive in Lezghi, Chechen (also infinitive, adj. and particip. suff.), possessive in Ubykh etc.; in some languages the function has shifted to ablative (Avar), ergative (Udi, Ubykh)
Allative	-ra(1)	-r/-ar(2), -al-(3)	*-ɫV(4)	(1) some northern Basque dialects have the form -rat and/or -la(t); (2) dative/allative; (3) locative; (4) Chechen -l, -lla (translative), Tsez -r (dative, lative), Khinalug -li (general locative) etc.
Comitative	-ekin		*KV(1)	(1) possible cognates among mutually incompatible suffixes, cf. Avar -gu-n, -gi-n (comitative), Andi -lo-gu, Karata -qi-l, Tindi -ka, Akhwakh -qe-na.

As Bengtson (2008) himself notes, an ergative ending -/s/, which may be compared to the ending that has instrumental function in Basque, occurs in some Sino-Tibetan languages, and the Yeniseian language Ket has an instrumental/comitative in -/s/, -/as/, -/aɕ/. This suffix may therefore be shared among a larger group, possibly Dené–Caucasian as a whole. On the other hand, comparison of noun morphology among Dené–Caucasian families other than Basque, Burushaski and Caucasian is usually not possible: little morphology can so far be reconstructed for Proto-Sino-Tibetan at all; "Yeniseian has case marking, but it seems to have little in common with the western DC families" except for the abovementioned suffix (Bengtson 2008:footnote 182, emphasis added); and Na-Dené languages usually express case relations as prefixes on the polysynthetic verb. It can therefore not be excluded that some or all of the noun morphology presented here was present in Proto-Dené–Caucasian and lost in Sino-Tibetan, Yeniseian and Na-Dené; in this case it cannot be considered evidence for the Macro-Caucasian hypothesis. That said, as mentioned above, Basque, Caucasian and Burushaski also share words that do not occur in other families.

A genitive suffix -/nV/ is also widespread among Nostratic languages.

Karasuk

George van Driem has proposed that the Yeniseian languages are the closest known relatives of Burushaski, based on less than a handful of lookalike elements in grammar and lexicon. He does not seem to have considered the other language families that are hypothesized to belong to Dené–Caucasian,^[28] so whether the Karasuk hypothesis is really incompatible with the Macro-Caucasian hypothesis remains to be investigated.

Footnotes

References

External links

• Dené–Caucasian ethno-linguistic map
• The Tower of Babel(Site in English and Russian including The proposed family tree&Word-final Resonants in Sino-Caucasian)
• A Final (?) Response to the Basque Debate in Mother Tongue 1
• Merritt Ruhlen on the Dené–Caucasian hypothesis (PDF)

See also

• Dené–Yeniseian languages
• Language families and languages
• Proto-language
• Borean languages