World language families and superfamilies (most linguists do not recognize superfamilies as proven genetic units)

1. Eskimo-Aleut family--a group of polysynthetic languages (languages with very complex words).

2. Na-Dene family—Tlingit, Eyak, and the Athabaskan languages of northern Canada plus Navajo and Apache in the southwest US.  Some include the Haida language isolate here.

3. Amerind superfamily--contains all remaining languages spoken in pre-Colombian North and South America; the splitters divide Amerindian into several dozen unrelated families.

4. Australian superfamily-- all languages of pre-European Australia; splitters see five major families rather than one.

5. Indo-Pacific superfamily-- contains at least 70 major families of languages spoken on the island of New Guinea; also contains the extinct language of Tasmania.

6. Austronesian family--the languages of Malaysia, Indonesia, Madagascar, and the islands of the Pacific (including Hawaii, New Zealand, Tahiti).  Even the splitters admit that all these languages are related.

7. Austroasiatic (Vietnamese, Cambodian), Daic (Thai), and Miao-Yao families of southeast Asia--recently lumped together with Austronesian as a single superfamily.

8. Sino-Tibetan family--contains Chinese, Burmese, and Tibetan; spoken by one out of every four humans on Earth.

9. Indo-European family-- most languages of Europe, Iran, and northern India; one out of every two humans on Earth speaks some form of Indo-European. (See next page for a fuller discussion of this family.)

10. Dravidian family--the languages of southern India.

11. Altaic superfamily--Turkish, Mongolian; some add Japanese and Korean, too, although most linguists consider these two languages to be isolates.

12. Uralic family--Finnish, Hungarian, and minor languages of northern Russia.

13. Caucasian superfamily—two small families of language spoken in the Caucasus mountains (Caucasian here is not a racial term).

14. Afroasiatic family--Arabic, Hebrew, most languages of northern Africa, including Amharic (in Ethiopia) and Somali.

15. Nilo-Saharan superfamily--a small group of very distantly related languages spoken mostly in Southern Sudan and Uganda.

16. Niger-Kordofanian family--the major family of sub-Saharan Africa, contains Swahili and 400 other Bantu languages, as well as several other subfamilies in West Africa.

17. Khoisan superfamily--the languages of the Bushmen of the Kalahari Desert in southwest Africa, famous for their clicks.  If all languages do indeed stem from a common source, this group is probably one of the oldest offshoots.

18. Language isolates--several isolated languages that cannot easily be fit into any of the above large families.  Isolates include Basque of northern Spain, Ket of central Siberia, Georgian of the Caucasus mountains, and Burushaski of northern India.  Isolates are thought to be remnants of ancient families once spoken more widely. Some linguists lump together Basque, Ket and Burushaski and a number of extinct languages into an ancient superfamily called Paleo-Eurasiatic.  For linguists who believe that all languages stem from a common source (proto-World, or the Mother Tongue), the term "language isolate" simply means a language without any close relatives.

The Methods of Comparative Linguists

            How do comparative linguists try to prove that languages are genetically related (diverged from a common ancestor)? They look for similarities. 

            However, the existence of similarities between two languages do not necessary prove a genetic relationship.  Similarities could also be due to two other factors:   borrowing through contact, or pure coincidence   Historical linguists tend to be interested only in similarities which come from divergence from a common ancestor.  How do they know how to avoid similarities due to coincidence or borrowing?

1) Certain words in a language tend to resist borrowing.  Least susceptable to change, as a rule, are the most commonly used words in a language, such as words for family members, parts of the body, basic actions, numbers, etc.  A large number of correspondences between words belonging to this core vocabulary are a strong indication of a genetic relationship.  If a group of similar words is due to borrowing, the words will tend to appear in specific areas of the vocabulary which reflect the cultural nature of the contact.  (Cf.  Arabic borrowings of Islamic cultural terms into such unrelated languages as Malay, Swahili, and Tamil; borrowing of technical terms from European languages into the aboriginal languages of other continents.)

2) Chance correspondences among words in the basic vocabulary may also occur.  These are easy to dismiss because they occur in complete isolation:  Didinga (Sudan) badh= Eng. bad, Mbabaram (Austr.) dog= dog.  Since no other basic words in these languages correspond to those of English, it is safe to assume that the languages are unrelated.  

3) Another piece of non-evidence is an isolated strong resemblance in one aspect of the grammar or phonology but a lack of resemblance elswhere: 

      a) the case of Basque and Georgian object marking in the verb.      

      b) English vs. Chinese word order. 

      c) the use of tones in Chinese and other mainland East Asian languages and tones in West Africa.

      d) the presence of retroflex stops in languages of India (due to borrowing)

4) The strongest evidence of a genetic relationship between languages is the presence of systematic sounds correspondences.  This is especially true if the correspondences are found in basic vocabulary words:  Eng th= Ger. d thick/dick, through/durch, occurs in numerous words of similar meaning-- many of them basic vocabulary-- and are highly unlikely to have been the result of chance borrowing.  English and Russian are also related historically-- g/z-- gold/zoloto, grain/zerno.  Words in two languages that come from the same word in an ancient ancestor are called cognates.

            Traditionally, comparative linguists compare only two languages at a time.  This type of comparison is useful to discover language relationships stemming back for several thousand years.  Because languages change, the similarities between two languages diminish the longer in time they are separated.  After 10,000 years of separation, there are no regular sound correspondences and few if any recognizable cognates.

            As a way to extend the usefulness of comparison back farther into time, some comparative linguists (these are the lumpers) have begun to compare a large number of languages simultaneously, rather than just two.  This method detects the few remaining similarities over a wide range of languages (for instance, the use of n in the second person singular of Amerindian languages).  It is this type of mass comparison that is responsible for the groupings on the map you have been given. 

            The problem with mass comparisons of this kind is that it becomes much more difficult to separate out the similarities that are due to borrowing from those that are due to borrowing or pure coincidence.  And so the debate rages.