This section gives a list of the languages that appear on at least one map in the atlas, organized
according to their genealogical classification. The classification here does not include many
intermediate groupings that appear in Ethnologue (Grimes 2000) or
Ruhlen (1987), restricting to only two levels in the majority of cases, that of family,
the highest level widely accepted by specialists, and genus, explained presently. In some cases, we
provide an intermediate level, that of subfamily, where the subfamily is well-known.
The notion genus is explained in Dryer (1989). It is intended as a level of classification which is
comparable across the world, so that a genus in one family is intended to be comparable in time depth
to genera in other parts of the world. The choice of term is intended to match the general idea of
genus in biological classification, where a genus is a set of species that are clearly closely
related to each other (and where words in everyday language often correspond to genera rather than
species). In the genealogical classification of languages, a genus is a group of languages whose
relatedness is fairly obvious without systematic comparative analysis, and which even the most
conservative “splitter” would accept. Genealogical groups deeper than a genus are often less obvious
and in the absence of detailed comparative work are often not universally accepted. If there is
evidence of time depth of groups, the genus would not have a time depth greater than 3500 or 4000
years. A genus may have a time depth much less than this, but if the time of the split of one group
of languages from other languages in the family appears to be greater than 4000 years, then this
constitutes a reason to say that this group of languages is a separate genus. The standard
subfamilies of
Indo-European
(e.g. Germanic, Slavic, Celtic, etc.) are fairly clear examples of
genera, although Celtic is perhaps a clearer example than Germanic or Slavic, both of which have a
time depth considerably less than 3500 years. The decisions as to which groups to treat as genera
here are best described as my own educated guesses. In many instances they are based on conversations
he has had with specialists. However, in the absence of a tradition within the field of attempting
to identify groups of comparable time depth in different parts of the world, they should not be
considered more than educated guesses. Specialists who think the choices of genera here are mistaken
are encouraged to let me know.
The classification here generally follows the classification given in Grimes (2000),
the 14th edition of Ethnologue. The classification deviates most strongly from that in
Ethnologue in the classification of the languages of
New Guinea.
The Ethnologue classification is a standard one, and is similar to that in Ruhlen (1987),
both being largely based on the classification in Wurm (1975) and Wurm
(1982). However, most specialists are now skeptical of various components of the Wurm
classification. The classification used here is considerably more conservative than the Wurm
classification, positing a much larger number of families, and reflects suggestions of William Foley,
though I accept responsibility for any misinterpretation of his ideas.
Names of genera are given in italics. Language isolates are a special case of this: they are
families that consist not only of only one genus, but also of only one language. There are cases
here of what are generally considered language isolates but for which more than one variety is
included in the WALS languages. When these varieties are all mutually intelligible, then the
language can be considered a language isolate, though since we do not attempt to systematically
distinguish languages and dialects, one cannot tell which families
with more than one language here are language isolates in this sense. Examples of language
isolates in this sense are
Basque
and
Yukaghir.
Within each family, the subfamilies and genera
are organized alphabetically, and within each genus, the languages are organized alphabetically.
Two groups here are not genealogical groups. The first of these is Creoles and Pidgins,
whose nature falls outside standard genealogical classification. The last is Sign Languages. Here
the notion of genealogical classification makes sense, but crosslinguistic work on sign languages
is sufficiently new that it would be premature to attempt a genealogical classification of them.
