1.1 Indo-European

1.2 Uralic

Three important languages in this family are Finnish, Estonian, and Hungarian. These three are traditionally grouped into a branch called Finno-Ugric. But while Finnish and Estonian are closely related members of the Finnic branch of Uralic, Hungarian’s membership in a sister branch to Finnic is under challenge; Ethnologue has dropped Finno-Ugric from its listing and now casts Hungarian as a separate Uralic entity. See Salminen (2002) for arguments. The remaining languages of Uralic are smaller ones found in northern parts of Europe and Asia.

1.3 Caucasus Area

The area of the Caucasus Mountains and its environs between the Caspian and Black Seas includes Georgia, Armenia, and Azerbaijan and parts of neighboring countries. This relatively small region may have up to around 40 highly diverse languages, falling into three families, Nakho-Dagestanian, Abkhazo-Adyghean, and Kartvelian. The most important Nakho-Dagestanian language is Chechen. Abkhaz-Adyghean is made up of Abkhaz and Adyghe and is best known among linguists for systems with 60 or more contrasting consonants but very few vowels. The major Kartvelian language is Georgian, with four million speakers. Ethnologue combines Nakho-Dagestanian and Abkhaz-Adyghean into a single family, North Caucasian. Nakho-Dagestanian and Abkhaz-Adyghean are also known by the respective names Northwest (or simply West) Caucasian and Northeast (or East) Caucasian.

1.4 Basque

Basque is an isolate spoken in the Western Pyrenees by about half a million, some in France but most in Spain. Its history in this location is widely thought to go back several millennia, antedating the more recent Indo-European migrations to the region. There have attempts to identify Basque with a wide variety of groups, including Kartvelian, Afro-Asiatic, and Iberian, but without attracting much support. Recent DNA evidence reinforces the notion of Basque descent from an ancient population of farmers and hunters (Günther et al., 2015).

1.5 Turkish

Turkish, a language of Europe and Asia, belongs to the Turkic group, described in the section on Asia.

2.1 Afro-Asiatic

This is the northernmost family, with a few hundred languages spanning all of North Africa and the Middle East, as well as two smaller areas of sub-Saharan Africa. The six branches of Afro-Asiatic are Semitic, Berber, Chadic, Cushitic, Omotic, and Egyptian. The Semitic branch has 78 languages, including Arabic, the first language of up to 300 million throughout North Africa and widely spoken in the Middle East. Among the world’s languages, Arabic ranks fourth in the number of speakers. Other important Semitic languages are Hebrew, which shares official status in Israel with Arabic, and several Ethiopic languages. Amharic, the official language of Ethiopia and the first language of 21 million, is a South Ethiopic language. In the North Ethiopic branch is Tigrigna, an official language of Eritrea spoken by 7 million.

The term Afro-Asiatic was used by Joseph Greenberg to replace the designation Hamito-Semitic, which posited a division between the Semitic branch (named for Biblical figure Shem) and a putative branch named for Biblical figure Ham. The notion that Hamitic languages formed a unified branch seemingly reflected factors like speakers’ typical occupations and a lighter skin color than black Africans to the south. Greenberg argued that extraneous factors like these had no place in language classification, which should be based solely on linguistic data. Comparing languages from the different groups classed as Hamitic, Greenberg concluded that the evidence did not support their grouping into a single branch.

The Berber branch of Afro-Asiatic is spoken in the foothills of the Atlas Mountain in Morocco and Algeria and, spottily, in neighboring countries. Cushitic gets its name from Cush, the son of Ham. The several dozen languages of this group are spoken mainly in Ethiopia and Somalia, with a few in Kenya and Tanzania. Chadic languages are mainly spoken in the countries surrounding Lake Chad and are dominant in northern Nigeria, numbering close to 200 in all. By far the most widely spoken is Hausa, with 25 million native speakers. The languages of the Omotic branch, numbering over two dozen, are all spoken in southwestern Ethiopia. The Egyptian branch, thanks to hieroglyphs, can be traced back before 3,000 bce. Ancient Egyptian was the ancestor of Coptic, spoken in Egypt, but over time was replaced by Arabic until Coptic died out, roughly 400 years ago. Since then Coptic has survived as a liturgical language.

2.2 Nilo-Saharan

The approximately 200 languages occupy a band extending from the Nile region to the Sahara desert. For a relatively small family, they are quite diverse typologically, leaving some doubt as to whether the Nilotic and Saharan branches really deserve to be grouped into a family. Reflecting this, Glottolog divides them into two separate families, Nilotic and Saharan.

2.3 Niger-Congo

The great majority of languages in sub-Saharan Africa are members of the Niger-Congo family. Its 1,538 languages make it the world’s largest language family, and only the Indo-European and Sino-Tibetan language families have more speakers than Niger-Congo. Ideas about the respective genetic affiliations of well-known groups within Niger-Congo have changed substantially over the last half-century. This has been the case with Kwa, Mande, Gur, Atlantic, and Benue-Congo, among others. To date, the truly remarkable event in the classification of this family remains Greenberg’s (1955, 1963) demonstration that Bantu—a group of 538 languages covering most of Central and Southern Africa—was, along with other languages called Bantoid, a subgroup within a group now called East Benue-Congo, most of whose other languages are spoken in Nigeria and Cameroon. This discovery—which took ten years before gaining the wide acceptance it has today—not only challenged earlier assumptions about linguistic classification but also opened the door to hypotheses about Bantu origins. The currently accepted view is that Bantu originated in southeastern Nigeria and expanded east and south from there.

2.4 Khoisan

Among the languages of the world, some are poorly studied and go back so far in time that it is hard to trace their genetic origins. This is the case with Khoisan, which is generally not recognized as an established family but as a set of 27 languages—some with just a handful of speakers—that are likely not to belong to the other three established families of African languages. Ermisch (2008) presents what is known, along with the residual problems.

2.5 Austronesian

Off the southeastern coast of Africa is the island of Madagascar, home to Malagasy, a Malayo-Polynesian language brought over by the island’s earliest settlers maybe 1,500 years ago. For more on Malayo-Polynesian, see the subsection on Austronesian in the section on Oceania.

3.1 Indo-European

3.2 Turkic

The approximately 40 languages of this family extend from Macedonia to Siberia, Central Asia, and western China. Despite the vastness of this area, the languages themselves are typologically quite similar: agglutinative, with vowel harmony involving both backness and rounding.

3.3 Mongolic

The Mongolic languages are a group of about a dozen spoken in Mongolia and in adjacent areas of the Russian Federation and China. Mongolian, with over six million speakers, is by far the largest language in the family and the official language both of Mongolia and of the Inner Mongolian Autonomous Region of China.

3.4 Tungusic

The 11 languages of this family are scattered through Siberia, the Far East of Russia, and northwestern China, but most are endangered and some are nearly extinct. That includes Manchu, the language of the founders of the Qing Dynasty, which ruled China for nearly three centuries up to 1912. The 2016 edition of Ethnologue lists only 20 speakers for Manchu, though over ten million are ethnically Manchu.

3.5 Altaic Area

The Altaic area extends from Turkey across the Altai Mountain area of Central and East Asia to Siberia. Altaic has been regarded by some as a family comprising Turkic, Mongolic, and Tungusic, and for a few even including as distant members Japonic and Korean. Versions of the Altaic hypothesis still have adherents, even though this notion has been cast into doubt as criteria have been challenged and evidence has been rejected as based largely on shared typological similarities, a position summarized in Unger (1990). Despite this, adherents continue to make a case, among them Miller (1991), Georg et al. (1999), and Robbeets (2005). The more conservative consensus is that many resemblances among languages in this linguistic area could have come from language contact rather than a shared ancestor. This view is reflected in Ethnologue and Glottolog, among others.

3.6 Dravidian

Dravidian languages are spoken primarily in southern India, though some are also found further north in the Indian subcontinent. The major literary languages are Tamil, Malayalam, Kannada, and Telugu, each one the first language of tens of millions. More is known about the history of Dravidian than about many other language families, thanks to the long literary periods of the four major languages.

Questions have been raised about Dravidian similarities to Uralic and Altaic, among several others. Austerlitz (1971) dismissed these, and Krishnamurti (2003), briefly surveying archeological and DNA literature along with linguistic evidence in his foundational work on Dravidian, seconds the conclusion that the linguistic arguments behind the proposed genetic relationships are tenuous and speculative.

Dravidian morphology is mainly agglutinative but lacks the long strings of affixes found in other agglutinative languages. The typical word order is SOV. Dravidian’s three-way contrast in coronal stops (dental, alveolar, and retroflex) can be traced back to proto-Dravidian. Sanskrit, an Indo-Aryan language, owes its retroflex consonants to Dravidian, from which they are thought to have spread by diffusion.

3.7 Sino-Tibetan

3.8 Austro-Asiatic

The Austro-Asiatic family extends across south Asia from India to Vietnam. The Munda branch is found in northeastern India, surrounded by Indo-European and Dravidian languages that have influenced its languages greatly over the ages. Typologically they are agglutinative, with SOV word order, making them typologically very different from the rest of the family. Austro-Asiatic includes two important national languages, Vietnamese and Khmer (Cambodian). These two languages were grouped, along with many others, into a branch called Mon-Khmer, a grouping still accepted by Ethnologue but challenged by Sidwell (2009).

Vietnamese has borrowed massively from Chinese and was originally written with Chinese characters. Vietnamese and a few others in this family have developed phonological tones, and still others are thought to be in the process of developing them.

3.9 Hmong-Mien (Miao-Yao) and Tai-Kadai

These two families were once regarded as branches of Sino-Tibetan, and the languages of both families show many influences from Chinese. The Hmong-Mien (Miao-Yao) languages are spoken in scattered areas across southern China and nearby countries of Southeast Asia. The Tai-Kadai languages extend from China south to Thailand, Laos, Myanmar, and Vietnam and include the national languages Thai and Lao. Both families share a number of typological traits: most of their languages are SVO with isolating morphology and contrastive tone that is associated with creaky or breathy voice quality.

3.10 Paleosiberian Area

The name Paleosiberian applies to a set of four languages or language groups of Siberia with no established genetic relationship but sharing some typological features—agglutinative word structure and, with exceptions, SOV word order.

One of these is Ket, unrelated to any extant language and reduced to about 200 speakers, but once a member of the Yeniseian family and unlike the rest of Paleosiberian in several respects. It is tonal and has a highly agglutinative verbal system with complex agreement patterns—features that make it look like Na-Dene in North America. The case for a genetic relationship between the two has been made by Vajda (2010, 2011). For arguments pro and con, see Kari and Potter (2010), Campbell (2011), and Kiparsky (2014, pp. 65–67). Implications of this finding for Beringian migrations are pursued by Sicoli and Holton (2014).

Also in the Paleosiberian area are the Chukotko-Kamchatkan and Yukaghir families and Nivkh, a language with perhaps 200 speakers.

3.11 Korean and Japanese

Two of the major languages of East Asia, Korean and Japanese, are widely considered isolates, or nearly so in the case of Japanese, by far the dominant language in Japonic, a family of twelve languages. The remaining 11 languages of Japonic are the Ryukyuan group of the Ryukyu Islands. Some versions of the Altaic hypothesis include Korean and Japanese in a family with Turkic, Mongolic, and Tungusic. Another isolate of Asia is Burushaski (northeastern Pakistan).

4.1 Austronesian

The 1,250+ languages of this family are distributed across Oceania from Madagascar to Easter Island and total well over 350 million speakers. All but 25 of these languages are Malayo-Polynesian; the rest are aboriginal languages of Taiwan.

The dominant category, Central-Eastern Malayo-Polynesian, has well over half of the languages classified as Malayo-Polynesian but only a few million speakers total, and it is not generally accepted as a valid linguistic grouping. The remaining Malayo-Polynesian languages are found in 17 smaller groups, some of whose languages are widely spoken and highly important politically. Among these are:

-

Javanese, the language of nearly 90 million, centered in Java, Indonesia.

-

Filipino, an official language of the Philippines used by close to 50 million, including L2 speakers, as the national language of Philippines. The variety associated with native speakers, who number over 20 million, is called Tagalog.

-

Sundanese, the language of about 34 million in Java.

-

Malay, an official language of Malaysia along with Mandarin and English, is the language of more than 50 million.

-

Malagasy, spoken by 18 million.

Blust (2013) offers a recent and comprehensive account of the linguistic and anthropological aspects of this family, including internal linguistic groupings, the linguistic structure of its languages, sociolinguistic considerations, and archeological evidence backing up the linguistic groupings. Adelaar and Himmelmann (2005) cover a similar range of topics.

4.2 Papuan Languages

Estimates run to as many as a thousand languages in an area about a quarter of the size of India, making New Guinea the most linguistically diverse region in the world (Foley, 2000, p. 357). Major groupings have been proposed by Greenberg (1971), Wurm (1982), and Ross (2005). Greenberg put all the languages into a single family and included some others from outside New Guinea, but the evidence for this has not generally been deemed credible. Wurm (1982) posited 10 Papuan phyla plus isolates, based entirely on lexicostatistic and typological evidence that others found unconvincing (Foley, 1986). A more recent grouping by Ross (2005), based essentially on evidence from pronouns, has also failed to find wide acceptance. One is left for now with Foley’s (1986) classification, with several dozen families and a similar number of isolates. Correlated with this is extreme typological variation across the families, with morphological types ranging from isolating to polysynthetic. Foley’s Papuan families average about 25 members in size, with the exception of Trans-New Guinea, with 482 member languages in Ethnologue, a figure that experts agree is subject to much revision because the family’s boundaries with others remain unclear. The uncertainty is reflected in Glottalog, which lists only Nuclear Trans New Guinea, with 315 languages.

4.3 Australia

This continent has been inhabited for 50,000 years, but the time frame for language classification is limited to the last 5,000 or so. As a result, we know very little about the historical connections among Australia’s languages. Worse, the number of vigorous Aboriginal languages today is a fraction of what it was before Europeans settled there in the 18th century. Of the 250-odd languages of Australia in 1788, more than half are extinct, and of the remainder, fewer than two dozen are used and learned by the youngest generation.

Beginning with Hale (1966), many sources divide the continent’s original languages into two groups, Pama-Nynngan and Non-Pama-Nynngan, but even this rudimentary grouping is complicated by large-scale phonological and grammatical diffusion. Dixon, author of many standard reference works on Australian languages, among them Dixon (2002), diverges markedly from the others by simply dividing the languages into 50 groups representing different areas, though among them some genetic clusters may be found. For Dixon, Pama-Nyungan “cannot be supported as a genetic group. Nor is it a useful typological grouping.” (Dixon, 2002, p. 53). The problem with applying standard methods toward reconstructing a language tree for Australia, as Dixon sees it, is that Australia is unique, in part to due widespread diffusion, whereby a language “will tend to become more like its neighbors” (Dixon, 2002, p. 448). For alternative studies from a vantage point that differs markedly, see Bowern and Koch (2004).

Phonologically, Australian languages tend to be simple in some ways—usually with three-vowel systems—and complex in others, with as many as four contrasting articulations among the coronal consonants. Morphologically, Pama-Nyungan languages have noun class systems and verbal concord prefixes, and some have extensive noun incorporation constructions. Outside Pama-Nyungan, morphology, especially in nouns, is of a more simple agglutinative type, with suffixes but no prefixes. Most Australian languages have split ergativity, a common pattern being ergative-absolutive alignment for nouns but nominative-accusative alignment for pronouns. Word order tends to be very free, but there is evidence that clauses are best analyzed as verb-final; see Mushin and Baker (2008).

5.1 North America

5.2 Mexico and Central America

5.3 South America

7.1 Pidgins

Pidgins are simplified languages that arise out of a need to communicate among speakers lacking a common language, typically in colonial situations where one group is dominant. Members of the dominated group fuse grammatical features, often simplified, of their native language (called the substrate) with vocabulary from the dominant, or superstrate, language. The resulting language serves restricted purposes, such as trade.

There are not many pidgins. Ethnologue lists only 16, six of them in Africa and five in Oceania, if Indonesia is included. Hiri Motu, an official language of Papua New Guinea, is noteworthy because it goes against some typical views of pidgins. This language developed between the Motu and their trading partners nearby before any European contact. After colonization, its use spread, though the colonizers themselves had little if any knowledge of it. More usual are the cases of the original Chinese Pidgin English, once known as Pigeon English, which arose in 17th-century China for trade with the British, and Nigerian Pidgin, which developed in the same era, again due to trade contact with the British, notably the slave trade.

Hiri Motu, Chinese Pidgin English, and Nigerian Pidgin illustrate three different types of situation. Hiri Motu and Chinese Pidgin English exemplify pidgins that originate when trade partners are equal (Hiri Motu) or unequal (Chinese Pidgin English). The two had similar outcomes, eventually fading away—Hiri Motu in favor of Tok Pisin, a widely spoken creole of New Guinea, and Chinese Pidgin English in favor of Standard English, which came to be commonly taught in schools. (Since then, a different language called Chinese Pidgin English has arisen on the Pacific island of Nauru, for communicating with Chinese-speaking merchants and traders.) By contrast, Chinese Pidgin English and Nigerian Pidgin had analogous origins (for communicating with traders in a dominant position), yet different outcomes, since the first has died out, while the second has vastly expanded its uses and its speaking population. Currently Nigerian Pidgin is learned by many children at an early age for communication with peers in virtually any informal situation.

7.2 Creoles

Creoles are first languages of members of speech communities but originate from types of language contact resembling, if not always identical to, situations that give rise to pidgins. Being acquired as a first language gives creoles a stability that pidgins lack, and so it is not surprising that many more creoles are in current use—93 listed in Ethnologue—than pidgins. Thirty-two creoles are spoken around Latin America and the Caribbean, 26 in Oceania, and 22 in Africa. Like pidgins, creoles have a substrate and a superstrate. English is the superstrate for 33 creoles, Malay for 14, Portuguese for 13, and French for 11.

Probably the most vigorously debated topic in current pidgin and creole studies is how creoles form and evolve. Bickerton (1981, 1988) interpreted creolization in terms of what is known as the bioprogram hypothesis. This would see creoles as developing from a pidgin that learners were exposed to at an early age. The hypothesis was that acquisition is guided by an innate bioprogram that supplies structure to complement and modify the pidgin’s substrate and superstrate. This idea excited those who saw its potential to shed light on the human language faculty in general. At the same time, among creolists, the bioprogram hypothesis gave rise to a literature that almost universally sought to disprove it. Viewed more positively, it engendered lots of new thinking on how creoles come about.

Veenstra (2008) surveys some of the progress made during this period. Early commenters found reason to assign a greater role to the superstrate language than would be the case under Bickerton’s hypothesis, which leaned heavily on universal grammar. Another criticism cited the fact that some creoles develop without having a pidgin as a source. Bickerton’s explanation, relying on acquisition by a generation of speakers with no other first language, implied that a creole would always develop in a single generation, yet this has been falsified by Nicaraguan Sign Language, which took two generations (Kegl, Senghas, & Coppola, 1999). For many more counterproposals and refinements, see DeGraff (1999), Mufwene (1996), Singler (1996), and McWhorter (2005). One area of agreement is that neither pidgins nor creoles are homogeneous types, as earlier work seemed to assume. There are many varieties, as is found with the rest of the languages covered in this essay.