Saturday, January 24, 2015

Do Glottal Consonants Tell Important Deep Historical Linguistic Tales?

Glottal consonants, also called laryngeal consonants are consonants articulated with the glottis. They come in three subtypes: ejectives, implosive, and glottalized resonants (also explained here).

Were Glottal Consonants Present In Proto-Indo-European?

One of the leading theories regarding the Proto-Indo-European language is that it contained laryngeal consonants which were lost in all of the successor Indo-European language families except the Anatolian languages, but whose loss explains grammatical irregularities in the daughter language families. Further, some loan words from Proto-Indo-European into the Uralic languages may have carried traces of lost laryngeal consonants.
Hittite and the other Anatolian languages are the only Indo-European languages where at least some of them are attested directly and consistently as consonantal sounds. Otherwise, their presence is to be inferred mostly through the effects they have on neighboring sounds, and on patterns of alternation that they participate in. When a laryngeal is attested directly, it is usually as a special type of vowel and not as a consonant.
The simultaneous loss of laryngeal consonants in all of the daughter language families but the Anatolian languages after they broke away from Proto-Indo-European (e.g. Sanskrit, Iranian, Tocharian, Germanic, Balkan, Italic, Celtic and Greek), however, might arguably be less plausible than a substrate or areal influence limited to the Anatolian languages and Eastern Armenian, as the substrate languages of Anatolia were probably part of the modern Caucasian language families.

Substrate influences from a radically different substrate language family than other Indo-European languages may also explain why the Anatolian languages seem so lexically divergent from the other Indo-European languages.

The time depth of Tocharian's connection to Proto-Indo-European (ca. 3500 BCE or earlier) is particularly notable in this regard, since the extinct Anatolian languages are themselves not attested before 2000 BCE when there is historically attested evidence of a major Hittite expansion, around the same time that Indo-European Greek started to replace the pre-Greek languages of the Aegean, and around the same time that archaeological evidence of Sanskrit associated civilizations appears in the Cemetery H culture of Northwest India.

In contrast, the expansion of Indo-European languages into Central and Eastern Europe as part of the Corded Ware culture dates to around 3500 BCE (most of these languages were replaced by Slavic languages in the 1st millennium CE in an event that also had a meaningful demic component) and one or two of the Corded Ware languages, in turn, probably provided a source for the Germanic, Italic and Celtic branches of the Indo-European language family, that started to expand into Europe in the very late Bronze Age and Iron Age (ca. 1300 BCE).

I am personally inclined to think that laryngeal consonants in Proto-Indo-European, if they were present at all, probably derive from and were limited to loan words from neighboring North Caucasian language, that were quickly shed upon being borrowed by the Indo-Europeans who lacked that phoneme and did not manage to reproduce it as the words were adopted in their own language, and in some cases passed these words, in turn, along to the Uralic languages with similar traces of the laryngeal consonants that were once present in the Caucasian sources for the wanderwort.

What Else Do Glottal Consonants Suggest About Historical Linguistics?

In General

1.  Glottal consonants, like click phonemes, appear to be easier for a language to lose than to gain in the absence of strong areal or substrate influences.  Every language family where they are found has members that lack them.  And, there are large entire language families (counting language isolates as language families) that seem to lack them, or lack them with a small number of easily explained outliers:

* Modern Indo-European languages (except Sindhi and Eastern Armenian)
* Tocharian
* Basque
* Uralic
* narrow Altaic (i.e. Turkish and Mongolian languages)
* Siberian languages (with one single exception in one subfamily probably due to areal effects from North America)
* Inuit
* Tibeto-Burmese (with one small exception in Burma probably due to substrate influence)
* Austronesian (with one small exception on Yap island probably due to substrate influence)
* Papuan languages (with one exception)
* Australian aboriginal languages
* Munda languages (which probably lost implosives in its parent language due to substrate influence)
* Semitic (with two small exceptions in Ethiopia probably due to substrate influences and one exception on an island off Arabia due to Ethiopian areal influence)
* Berber
* Coptic
* Dravidian languages
* Many of the Native American language families of Eastern North America.

In Africa

2.  Implosives are found in a large enough share of the Niger-Congo languages and Nilo-Saharan languages to suggest that they were present in the proto-languages of each and may have a common origin.

3. Ejectives are found in seven of the Khoisan languages broadly defined as click consonant languages that are not a part of another major language family (one of which also uses implosives and one of which also uses glottalized resonants), and the Zulu language of South Africa which a Bantu language with a local substrate that used the Khoisan click consonants.  It is fair to guess that ejectives were present in a proto-language that is the parent to all of the click languages of Africa.  This language probably originated in East Africa or Ethiopia and subsequently migrated to South Africa.

4.  Ejectives are found in all of the language families centered in Ethiopia: two Ethio-Semitic languages, four Omotic languages (two of which use implosives as well), and three Cushitic languages, in Ethiopia, Kenya and Tanzania respectively (each of which also use implosives).

The Chadic languages show linguistic and population genetic signs of being derived in substantial part from the Cushitic language.  Two Chadic languages use both ejectives and implosives.

Taking the evidence as a whole, it seems likely that the Proto-Cushitic language, the Proto-Chadic language derived from a Cushitic language, and the Proto-Omotic languages all used both ejectives and implosives (possibly due to a substrate from a sister language family to the Khoisan languages which the click consonants were lost).

5.  The Ethio-Semitic languages that have ejective consonants probably do so due to substrate influence that had either already lost the implosives, or lost them at the time of Ethio-Semitic language shift.  These languages did not exist more than 3500 years ago and replaced languages that were probably similar to Cushitic and/or Omotic.

6. The presence of ejectives and glottalized resonants in the South Arabian Semitic language on the island of Soqotri is probably an areal effect from Ethiopian Cushitic influence (the Omotic languages of Ethiopia, the Chadic languages, and the Khoisan languages of East Africa are all comparatively geographically remote) sometime in the last 3500 years.

7. Implosives and ejectives are found in three Ethiopian Nilo-Saharan languages, with the ejectives probably due to substrate or areal effects since the ejectives are absent in other Nilo-Saharan languages, and Nilo-Saharan languages are a relatively late arrival to Ethiopia compared to the Afro-Asiatic languages found there (apart from the Ethiosemitic languages).

Similarly, the presence of glottal resonants in one Nilo-Saharan language spoken near the trinational boundary of Uganda, the Democratic Republican of Congo and South Sudan, in addition to implosives in that language is probably due to substrate or areal effects.

8. Glottal consonants are not found in the Northern Afro-Asiatic languages: Berber, Coptic (i.e. Ancient Egyptian), or any Semitic languages other than Soqotri and the Ethio-Semitic languages, and hence were probably absent from all of these proto-languages.

But, contrary to this hypothesis, the earliest attested Semitic language, Akkadian, which is well attested in writing, appears to have had an implosive glottal consonant.  There also appears to be temporal and geographic variation in the use of glottal consonants in different dialects of Arabic (e.g. with some dialects losing a previously attested glottal stop).

Likewise, liturgical Coptic, now used mostly by the Ethiopian Orthodox church, has both implosive and ejective glottal consonants, although it is not clear if the fact that the Coptic language is now used liturgically by people who have those consonants in their native living languages means that it is a reliable indicator of the consonants present in the ancient Egyptian language thousands of years before it was used in Ethiopia.  Since the hieroglyphic writing system was not purely phonetic, this is not easy to determine despite the abundance of available written materials in Coptic.

It could be that the Afro-Asiatic languages that became Berber, Coptic and Semitic originally had a full range of glottal consonants but lost them as these languages were eventually adopted by common people learning them as second languages in regions where these consonants did not exist in now lost substrate languages of North Africa and the Near East.

In The New World

9.  The presence of glottal consonants in Na-Dene languages appears to arise from substrate or areal effects of other pre-existing Native American languages.  The Na-Dene are heavily admixed with the founding population of the Americas and have been present in the Americas long enough for this to be plausible.  The lack of glottal consonants in all but one Siberian language very close to North America, also suggests that Yenesian is the parent language family (as opposed to a back migrating language family from North America, a conclusion also supported by Ket and Na-Dene population genetics).

Another possible substrate influence is that the Yenesian Ket language has a sex based grammatical gender system, while the Na-Dene language, like all of the Native American languages in the area where the Na-Dene languages are spoken, does not.

The linguistic distinction between non-sex based gender systems or more than three genders for male, female and neuter (a common feature, for example, of Niger-Congo languages, Papuan languages, abd Australian Aborginal languages), however, and noun cases that are not called genders (which are numerous, for example in Caucasian and Dravidian languages) which are present in many other languages, is a distinction without a difference in my opinion, that obscures possible relationships between languages based merely on regional conventions about how grammatical features are named.

10. The proto-Amerind languages ca. 20,000 years ago probably had ejective glottal consonants and glottalized resonants, both of which are widespread in the Americas, but some of which were lost in subsequent daughter languages or language families.

Implosive glottal consonants are found in nine Native American languages: alone in two Brazilian and one Bolvian language; together with glottalized resonants in one in Northern Californian language, one language on the Yucatan Peninsula in Southern Mexico, and one language over the Southern Mexican border in Guatamala; and with both glottalized resonants and ejectives in three more Native American languages,  one in Washington State, one in Southern Mexico, and one in Brazil near the Bolivian border.

This pattern tends to favor a hypothesis in which there was population structure between the Pacific route subpopulation of the founding population of the Americas which had all three kinds of glottal consonants, which its daughter languages preserved in various combinations, while losing others.  In contrast, the North American Native Americans who took an Eastern route and then migrated West back towards the Pacific across North America, either was from a subpopulation that never had implosive glottal consonants, or lost them early on.

The non-Pacific route Native Americans are also notable for having mtDNA X2a, which is not present in Southern coastal route intermediate populations between West Asia and the Bering Strait, unlike most of the other Y-DNA and mtDNA clades found in Native Americans.  One could imagine a scenario in which the Bering land bridge is home to an mtDNA X2a rich population with stronger North Asian influences that loses the implosive glottal consonants of the Native American founding population, and another one on the Bering land bridge without mtDNA X2a or the same amount of North Asian influences that retains implosive glottal consonants.

In Eurasia

11. Languages with ejectives are found in ten languages of the Caucasus, both North Caucasian and South Caucasian, as well as geographically adjacent Eastern Armenian (an Indo-European language clearly experiencing an areal effect).

This is one of many distinguishing features of the languages of the Caucasus Mountains that point to the Northwest Caucasian languages, the Northeast Caucasian languages (a.k.a. Nakh-Daghestanian), and the South Caucasian languages (a.k.a. Kartvellian) all having a deep common origin.

These languages are the only extant plausible source for glottal consonant influences on Proto-Indo-European of any kind, if there was one.

The Caucasian languages could have developed glottal consonants independently in deep linguistic history (the early Neolithic at least), or these consonants could have been present in an early Out of Africa population which lost implosive glottal consonants and glottal resonants, but unlike other Eurasian languages retained ejective consonants.

12. Implosives are found in the Indo-European Sanskrit derived Sindhi language of Southern Pakistan.  This could be a residual result of linguistic borrowing through maritime trade with Ethiopia (which dates back to the Copper Age, at least) or trade with Southeast Asia (which is probably at least 1000 years old) or both.

13.  Ejective consonants in the Churkotko-Kamachatkan language called Itelmen spoken on the Kamchatkan Penninsula adjacent to the Bering strait are probably an areal effects involving back migration from, or contact with, Alaskan languages with these consonants are common.

14.  The presence of ejective glottal consonants in Korean is probably the hardest to explain of the data points.  Korean is known to probably derive from somewhere to the north of the Korean pennisula and it has been widely hypothesized that Korean's ultimate linguistic roots are in the Altaic language family from a homeland which autosomal genetic evidence suggests is also home to the modern Eurasians most closely related to the Native American founding population.  An independent innovation in the Korean language itself is also possible.

15. While my primary source (WALS) does not identify any Japonic languages with glottal consonants, it appears that the Japanoic North Ryukyuan languages, such as the language of Okinawa, do have them and that these languages, in general, preserve features found in Old Japanese that are absent in modern Japanese.  The fact that the North, rather than the South Ryukyuan languages have these consonants also suggests (in accord with other lines of evidence regarding the prehistory and ancient history of these islands) that glottal consonants in the North Ryukyuan likely derive from the language spoken by the Yaoyi migrants to Japan, rather than an areal influence from the island of Formosa (Taiwan) or Southern China, of some kind.

If both proto-Japanese and Korean both had glottal consonants, then the glottal consonants in these languages probably indicates a shared common origin (reasonably enough since the Japanese language almost certainly arrived in Japan via Korea) that explains these consonants in both languages (and disfavors North American areal effects).

16. Implosives alone area also found in ten languages of Southeast Asia, one of Southeast China, one of Taiwan, one of Indonesia, and one of Papua New Guinea.  Glottalized resonants are found together with implosive glottal consonants in two Mon-Khmer languages of Vietnam (Khmu' and Sedang) and one Tai-Kadai language of Southwest China (Sui),  Glottalized resonants alone are also found in two languages of Southeast Asia, one spoken in Burma (Chin) and one spoken in Malaysia (Semelai).

All of these probably have a common origin, particularly in light of the likelihood that all Southeast Asian, South Chinese and Formosan languages have genetic ties to each other that are distinct from Tibeto-Burmese ties.  The presence of these consonants in Tibeto-Burmese languages is likely due to substrate influences as Tibeto-Burmese languages are comparatively recent arrivals in the region.

17. As discussed below, the use of ejective glottal consonants and glottal resonant consonants in Yapese is probably either an independent innovation arise after thousands of years of pre-Austronesian isolation in the island of Yap, or a deep substrate influence of an extremely conservative language feature lost everywhere else in successive language sweeps from tens of thousands of years ago.

18.  Notably, it is unclear if the extinct Sumerian language had glottal consonants or not.  If it did, then the apparent presence of glottal consonants in the Semitic Akkadian language that replaced Sumerian in Mesopotamia may have been a substrate influence particular to Akkadian, rather than a feature of the proto-Semitic language that withered away over time.

Overlapping Language Feature Distribution

The geographic distribution of languages with tone systems is similar, although not identical, to the geographic distribution of languages with glottal consonants.  Both are most common in sub-Saharan Africa, Southeast Asia, and the Americas.  But, the Chinese dialect family uses tone, while it does not utilize glottal consonants.  As used in this sense:
Tone is the term used to describe the use of pitch patterns to distinguish individual words or the grammatical forms of words, such as the singular and plural forms of nouns or different tenses of verbs.
Post-Script on Yapese.

The Yapese language of the people of the island of Yap in the Pacific Ocean between Papua New Guinea and the Philippines fits clearly within the Oceanic subfamily of the Austronesian languages, but is a language isolate within that family.  The fact that it makes widespread use of ejective glottal consonants and glottal resonants, as well as is used of a VSO word order unlike superficially similar Admiralty Island languages distinguish it.

There are no other Papuan or Austronesian languages with these consonants.  But, like other Admiralty Island languages, it does appear to have a Papuan substrate, perhaps from a population that may have reached this island before the Austronesians arrived, although not with maritime technology sufficient to maintain ongoing ties to their place of origin.

On the whole, this looks like a case where this pre-Austronesian population was so isolated for so long that it either (1) retained highly conservative first wave hominin phonemes lost everywhere else in the region in one or more subsequent wave of migration or loss of language complexity, or (2) innovated independently in phonology in a manner not suppressed by influences from other cultures from whom it was isolated for so long.
[Yapese] belongs to the Austronesian languages, more specifically to the Oceanic languages. It has been suggested that Yapese may be one of the Admiralty Islands languages, though Ethnologue lists it as a language isolate within the Oceanic languages. The glottal stop is a leading feature of Yapese. Words beginning with a vowel letter (with a few grammatical exceptions) begin with a glottal stop. Adjacent vowels have the glottal stop between them. There are many word-final glottal stops.
Yapese is also notable for its VSO word order, in contrast with the SVO word order of most Admiralty Island languages. The same link notes that:
The Oceanic languages were first shown to be a language family by Sidney Herbert Ray in 1896 and, besides Malayo-Polynesian, they are the only established large branch of Austronesian languages. Grammatically, they have been strongly influenced by the Papuan languages of northern New Guinea, but they retain a remarkably large amount of Austronesian vocabulary.
There are about thirty Admiralty Island languages, with Yapenese and Nguluwan as outliers that might or might not fit. Many "of the Manus languages in the Admiralty Island language family have no bilabial trill or prenasalised consonants, but the Baluan-Pam language of that family does not and does have a glottal frictive consonant (h), but it is a very marginal phoneme in that language.

Yap is home to about 11,000 people today and probably far fewer in pre-modern times, and is known for its use of stone coins ranging in size from 1.4 inches to 12 feet in diameter. Per the Wikipedia article on Yap:
Yap was initially settled by ancient migrants from the Malay Peninsula, the Indonesian Archipelago, New Guinea, and the Solomon Islands. The people of Yap's outer islands are descendants of Polynesian settlers, and as such have significant ethnic dissimilarities from the people of Yap proper. Their culture and languages (Ulithian and Woleaian) are closely related to those of the neighboring islands of Chuuk. . . Yapese society is based on a highly complex caste system involving at least seven tiers of rank. Historically, the caste rank of an entire village could rise or fall in comparison to other villages depending on how it fared in inter-village conflicts. Winning villages would rise in rank as a part of a peace settlement, while losing villages would have to accept a decline in comparative rank. In many cases lower ranked villages were required to pay tribute to higher ranked villages. Further, dietary taboos might be imposed on lower ranking villages, i.e., they might be prohibited from harvesting and eating the more desirable fish and animals of the sea. Further, within each village each family had its own rank comparative to the others.

Until the arrival of the German colonizers, the caste ranking system was fluid and the ranks of villages and families changed in response to inter-village intrigues and confrontations. In the late 19th century, however, the German colonial administration "pacified" Yap and enforced a prohibition against violent conflict. The caste ranking of each village in modern Yap thus remains the same as it was when the system was frozen in place by the Germans. The result of the freeze left Yap with three highest ranks of the villages of Teb, Gachpar, and Ngolog. The village of Teb from the municipality of Tomil remains as the highest of the three. The first recorded sighting of Yap by Europeans came during the Spanish expedition of Álvaro de Saavedra in 1528. Its sighting was also recorded by the Spanish expedition of Ruy López de Villalobos on 26 January 1543, who charted them as Los Arrecifes ("the reefs"). At Yap the Villalobos' expedition received the same surprising greeting as previously in Fais Island from the local people approaching the ships in canoes: "Buenos días Matelotes!". Again, "Good day sailors!" in perfect sixteenth century Spanish evidencing previous presence of the Spaniards in the area. The original account of this story is included in the report that the Augustinian Fray Jerónimo de Santisteban, travelling with the Villalobos' expedition, wrote for the Viceroy of New Spain, while in Kochi during the voyage home.[13] Yap also appeared in Spanish charts as Los Garbanzos (The Chickpeas in Spanish) and Gran Carolina (Great Caroline in Spanish).
Nguluwan, per Wikipedia is: "a "mixed" language spoken on an atoll of that name between Yap and Palau. The grammar and lexicon are Yapese, but the phonology has been affected by Ulithian, and speakers are shifting to that language."

If I were looking for unusual genes outside of Africa, the people of the island of Yap would be a particularly fruitful place to look.


Languages with Ejectives

Ejectives Without Other Glottal Consonants

Languages with ejectives are widespread throughout North America and South America.

Languages with ejectives are found in only two Asia languages, Korean and Itelmen, a Chukotko-Kamchatkan language spoken on the Kamchatkan Peninsula, adjacent to the Bering Strait.

Languages with ejectives are found in ten languages of the Caucasus, both North Caucasian and South Caucasian, as well as geographically adjacent Eastern Armenian.

Languages with ejectives are found in two of the many Ethio-Semitic languages (Tigre and Amharic), two of the Ethiopian Omotic languages (Kefa and Dizi), and a Nilo-Saharan Ethiopian language (Berta), as well as two cypto-Khoisan languages of Tanzani (Hadza and Sandewe), three Khoisan languages of Southern Africa.

Ejectives With Glottalized Resonants

Glottalized resonants are found together with ejective glottal consonants in seventeen Native American languages (eleven near the Pacific Coast of the Unites States; one near Canada's Arctic coast (Slave Lake, a Na-Dene language); three of which are in the Southern United States (Acoma, Wichita and Yuchi); and two near the Pacific Coast of South America (Jebero and Wichi).

Glottalized resonants are also found together with ejective glottal consonants in three other languages, one is a Khoisan language of South Africa, one is the South Arabian Semitic Soqotri language of an island off the Arabian coast.  The third is found on the island of Yap in Micronesia called Yapese.

Glottalized Resonants Alone

Glottalized resonants alone are found in found in two Native American languages, one in Southern Mexico (Chinantec) near where Mazahua is spoken, and one in Brazil near the Bolivian border (Wari) near where Nambikuara is spoken.

Glottalized resonants alone are also found in two languages of Southeast Asia, one spoken in Burma (Chin) and one spoken in Malaysia (Semelai).


Implosives Alone

Implosives alone are found in three languages of South America, two in Brazil and one in Boliva.

Implosives alone area also found in ten languages of Southeast Asia, in one of Southeast China, in one of Taiwan, in one of Indonesia, in one of Papua New Guinea, and in one of Southern Pakistan (Sindhi).

Implosives alone are also found in many Niger-Congo of West Africa, Central Africa, and many Nilo-Saharan languages East Africa (as well as a couple of languages attributed to the small Kadu language family sometimes classified as Niger-Congo and sometimes classified as Nilo-Saharan.

Implosives With Glottalized Resonants

Glottalized resonants are found together with implosive glottal consonants in two Mon-Khmer languages of Vietnam (Khmu' and Sedang) and one Tai-Kadai language of Southwest China (Sui), as well as in one Nilo-Saharan language spoken near the trinational boundary of Uganda, the Democratic Republican of Congo and South Sudan (Lugbara).

Implosives With Ejectives

Implosives and Ejectives Without Glottalized Resonants

Implosives and ejectives are found in three Native American languages (one in Northern California, one on the Yucatan Peninsula in Southern Mexico and one just over the Southern Mexican border in Guatamala).

Implosives and ejectives are also found in eleven African languages: two Chadic (Hausa and Kotoko), a Khoisan language in Southern Africa (Deti), a Bantu language in Southern Africa (Zulu), the Cushitic Oromo language in Ethiopia, a Cushitic language in Tanzania (Iraqw), a Cushitic language in Kenya (Dahalo), two Nilo-Saharan languages (Komo and Ik) near the South Sudan-Ethiopian border, two Omotic language of Ethiopia (Kullo and Hamer).

Ejectives With Glottalized Resonants and Implosives:

Ejectives, Glottalized Resonants and Implosives are all found together only in three Native American languages, one in Washington State (Lushootseed), one in Southern Mexico (Mazahua), and one in Brazil near the Bolivian border (Nambikuara).


G Horvat said...

I have not yet quite determined how but Tibetans seem to have something to do with mtDNA N(xR) haplogrou harbouring populations such as the northern Na-Denes, Inuit and Ainu. The latter is also due to the high frequency of Y chromosome haplogroup D. I can't help but think Edward Sapir was on to something when he contemplated a link between Tibetan and Na-Dene languages. Comments on this?

capra internetensis said...

Interesting survey.

The Indo-European laryngeals are not usually (ever?) reconstructed as glottal consonants in this sense, though.

h1 is usually reconstructed as common or garden /h/ or a glottal stop. The others are usually taken to be pharyngeal or uvular sonorants (or even velars), which are fairly widespread.

Laryngeals are reconstructed for all kinds of roots, including many very basic ones, and also for morphological affixes, so piecemeal borrowing from Caucasian-area languages does not seem a likely explanation. Movement outward into other linguistic areas lacking these sorts of sounds would explain their loss in most branches, however.

andrew said...

@G Horvat:

There is not discernable linguistic link, the lack of Y-DNA D in New World populations likely disfavors such a connection. The mtDNA links to Tibet are remote and probably don't fit timewise.