Re: Language family trees

Tristan wrote:
> ... "A significant proportion of linguists no longer subscribes to the
> theory of a language tree"
Well, as an amateur I might be wrong about the details of the theory,
but the metaphor is clearly severely flawed.

'tree' suggests a structure where a node can have multiple descendants,
but only one parent (yes, family trees are built upside down, don't
give me grief about the terms). So the theory goes:

  In the early Maldonian period, Old Bostonian was the primary
  language.  Some early settlers to Revere began the use of nasalized
  [r], while the Dorchester residents introduced the infix "fuckin'"
  to show adjectival agreement. Old Dorchester as spoken in Jamaica
  plain turned initial [T] into [D], and medial [T] into [sQuIki].
  Cambridge maintained a close relationship with the mother language,
  via the number 60 bus, reducing all color terms to the suffix
  "-American". A splinter group of Unitarian-Universalists from
  Cambridge later discover they had evolved forty-seven distinct
  terms for 'snow' and therefore had to move to the Artic Circle,
  where the invented their uniquely beautiful form of Nativity
  poetry, (where each strophe begins with 'close the door' and
  finishes with an antiphonal 'were you born in a damn barn?')

See our neat little tree:

  Boston   Modern Revere   Dorchester Jamaica Plain   Cambridge   Northern UU
     |        |                 \        /                 \         /
     |        |                  \      /                   \       /
     |      Revere            Low Dorchester            Ancient Cambrige
     |        |                    /                        /
      \       |     --------------       -------------------
       \      |    /                    /
         Old Boston---------------------

But we know the data doesn't fit this neat little metaphor; languages
can have variable number of parents. Modern English is descended
from French as well as Old English, and is chock-full of Greek
borrowings as well. There must be a hundred pidgins and creoles,
each put together from two or three languages. Esperanto, it could
be argued, has a half-dozen parent languages.

Imagine this background: after Lemuria sank into the sea, its
citizens escaped to South America and Australia. In Australia, they
intermarried with people from Mu, whose island was destroyed by a
volcano, and their children used Lemurian nouns to describe things
in the house and Muuic verbs to discuss actions outside the home.
Refugees from Mu also fled to Europe, where they lent their case
system to refugees from Atlantis. Atlanteans that escaped to South
America found communication with the Lemurians very easy if the
spoke clearly and loudly (much of their language now is derived
from the Lemurian terms for "I'm not deaf" and "Hush, I just put
the baby to sleep").  Could a tree theory predict the existence of
the three actual parents? Language derivation must be a more general
directed graph.

> Is this some weird form of ... creationism applied to language?
Well, creationism does apply to some languages (per my graph remarks,
some languages have zero languages as ancestors) - tlhIngan, Lojban,
and Láadan were indeed all created by intelligent design by independent
agents. I even hear there's a mailing list full of people who
construct languages.

> To say I'm amazed is an understatement (but then, I've been amazed
> before, like when I heard there really were creationists complaining
> about evolution being taught in American schools). Will wonders never
> cease?
   (Did you just say this to start a fight? Did you really want to
   see four hundred messages arguing inanely over what a 'theory'
   is, filled with theological arguments by people who never finished
   the book of Genesis? Why couldn't you just have mentioned Hitler
   and be done with it?)

        Erich

Replies