Microsoft Word L 2-03-Greenberg-paper doc

R
. 
There are certain further defining characteristics that are most conveniently stated in 
terms of a derived relation 
R*
(read “
R
-ancestral”). 
R*
is defined as any power 
R
n
of 
R
, 
by which is meant the repeated application of 
R
n
times. For example, if 
R
is the 
relation of parent to child, 
R
2
is that of grandparent to child and 
R*
that of ancestor to 
descendant. We require that 
R*
be irreflexive and asymmetrical. A beginner in 
R
is a 
member of the set to which no other member has the relation 
R
. If there is a unique 
beginner, then all the other members of the set are in the field of the converse of 
R*
; 
that is, they all have the unique beginner as a common ancestor. In the case of 
language, if all the languages of the world are related this will be the case, and proto-
sapiens will be the unique beginner. 
There are many examples of the family tree model which do not have a historical 
genetic interpretation; e.g., stochastic processes such as the successive throws of a die. 
The most conspicuous instances in which a historical interpretation is generally 
accepted are languages and species in the theory of biological evolution. It is, of course, 
not the only alternative. Before 1859 creationism was the generally accepted theory in 
biology, while the Tower of Babel account was only gradually undermined in 
linguistics; by the early nineteenth century the historical interpretation of differential 
degrees of language difference was generally accepted. In the nineteenth century the 
similarities of evolutionary biology and genetic linguistic classification were 
recognized both by biologists and linguists.
5
Among the more obvious similarities are 
the correspondence of homology and analogy to genetic and typological resemblances. 
Again, the difficulty of distinguishing language from dialect is analogous to the 
difficulty of distinguishing species from variety. 
In both cases there are conventional tests (mutual intelligibility in regard to 
language and the production of fertile offspring in relation to species), but in both 
instances there are borderline cases. This is because both speciation and language 
formation are dynamic processes. At a certain but not easily definable point, we have 
clearly distinguishable languages and separate species, and a point of no return has 
been reached. In language, however, we may have borrowing between separate 
languages. As far as I am aware there is no analogue of this for species, at least under 
natural conditions. 
Examples of tree structures closer to language as objects of investigation are 
manuscript genealogies (stemmas) in which the relation of original to copy plays the 
role of 
R
and the historical relationships of systems of writing. 
5
For a more detailed discussion of the methodology of classification see Greenberg (1957b, 
1963, and 1986).

---

The Methods and Purposes of Linguistic Genetic Classification
133 
Since language is a cultural institution, it seems natural, in discussing cultural 
transmission, to ask if there is a more general cultural analogue to linguistic genetic 
classification. In attempting to answer this, it is useful to note that both in languages 
and in non-linguistic culture there are four basic sources of resemblance at the trait 
level. In language the classification into these four types applies whether we consider 
resemblances in sound only, meaning only, or sound and meaning simultaneously. 
However the illustration of these types will all involve sound and meaning. 
The existence of these four types was apparently first noted in Pott (1855, p.42; 
repeated in greater detail in 1884, p.66f.), and for culture in Tylor (1865, pp.3, 376). 
Using more modern terminology than that employed by Pott, we may call these 
accident, sound symbolism, genetic, and contact (including borrowing). English 
examples of each of these are: English 
bad
= Persian 
bad
(accident); English 
mama
= 
Savo (Indo-Pacific) 
mama
(sound symbolism); English 
foot
= German 
Fuß
(genetic); 
English 
chance
= French 
chance
(contact by borrowing from French into English). 
The general culture analogues of these are what Tylor calls independent invention 
(= accident), psychic unity (= sound symbolism), common inheritance (= genetic), and 
transmission (= contact). Independent invention arises because of the principle of 
limited possibilities. Since there are a finite number of sounds and a finite number of 
meanings, there are bound to be some accidental resemblances in language. Similarly 
matrilineal clans exactly the same in number have arisen in different ethnic groups in 
different parts of the world. Since in such cases the historical antecedents are likely to 
have been different in each case, this is sometimes called convergence by 
anthropologists. An example of psychic unity is the use of the crescent as a symbol for 
the moon in both Egyptian hieroglyphics and the earliest Chinese writing. Common 
inheritance is the likely source of numerous non-linguistic cultural resemblances among 
the indigenous cultures of the Polynesians deriving from the ancestral culture of the 
speakers of Proto-Polynesian. Examples of cultural borrowings are commonplace. A 
well-known anthropological example is the spread of the Ghost Dance religion among 
various groups of native Americans in the Western part of the United States in the latter