Refining what trees show

[ECOLING at aol.com]

We often make much too little use of the capabilities of visual diagrams.
Choices in these do have strong implications for our thinking.

Sean Crist responded to Steve Long on questions of parts of one language
changing substantially, while other parts do not.
(still under the title "When a mother becomes a daughter" -
small excerpt quoted below)

I agree completely with Crist that the matter is in many cases purely
a matter of terminological convention.  Italian might still be called Latin,
just as modern Dutch still is called that.  One branch retains the name.
Steve Long, incidentally, also said that the debate may be purely
terminological.

However,
Steve is also pointing to situations which are not purely terminological,
where in standard treatment we would
say that one branch innovated *a lot*, and the other innovated *very little*.
Never mind that "a lot" and "a little" are not absolute or discretely
different.
There can be cases in which they are different enough to be highly
significant.

I am fully aware of the view that the tree diagrams are also purely
conventional,
that even where there is a continuing "stem" shown, there are also innovations
on that stem, so it is equal to the branch.  Differences don't matter.

But taken as a matter of degree, not of absolutes, there is still some
considerable sense in what Steve is pointing to on this issue.

It may sometimes be useful to indicate large differences of degree
by a more subtle use of our family trees and other diagrams.

***

Family trees need not always divorce themselves completely
from real geography, showing the innovating branches off to one side,
and what appears as a continuing smooth flow of the remainder to the
other side (witness Italo-Celtic branching to the East
in the UPenn trees I have seen pictured).
Rather, they can also incorporate some hints to geography,
showing innovating branches in different directions.

This is not merely formal, as it makes it more possible by one degree to
think in terms of both dialect areas and family trees at the same time,
or in terms of migrations and family trees at the same time.
Personally I usually prefer a map of dialect areas on which a tree is
superimposed, as to me more *practically* useful,
there may be quite a number of mixtures which can be useful
in visual diagramming.
Each will of course be conventional and cannot show infinite detail.
As mentioned long ago, noting some salient innovations on branches,
or using convex ovals for innovating dialect areas,
or both, are very useful tools in aiding understanding.

Many devices which incorporate some notations for actual data
into diagrams which show the overall patterning of that data
can be useful,
by allowing us to focus easily on EITHER THE TREES OR THE FOREST,
by showing (some of) both at the same time.

***

It can be useful, in cases of a tree in which there are indeed large
differences
in degree of innovation, many innovations of one branch clearly branching off,
and few innovations on the remaining core, to show a vertical with a branch,
rather than a forking which visually conveys equality.  (And yes, there will
be intermediate cases where we cannot easily choose between these two
representations - that in no way suggests that the distinction is not most
often quite useful.)  Footnotes can be added to anything, to add to, limit,
or clarify what is or is not signaled by a diagram.

That is we CAN do both

     /\
    /  \
   /    \
  /      \

and

    |\
    | \
    |  \
    |   \
   /|    \
  / |     \
 /  |      \

and have that visual distinction aid us.

We tend to greatly UNDERUTILIZE the facilities of even
two-dimensional paper.

Sincerely,
Lloyd Anderson
Ecological Linguistics

>It's apparently a problem for you.  Linguists are well aware that there is no
>clear criterion for looking at two lects and saying "same language" vs.
>"different languages."  It is purely a matter of convention what distinctions
>we want to draw and what names we want to use.

>For example, consider the following two cases:

>   -Earlier Dutch forked into Dutch and Afrikaans.

>   -Latin forked into Italian, Spanish, French, etc.

>In the former case, we continue to use the term for the parent language for
>one of the daughters.  In the latter case, we don't.  What we choose to call
>the daughters is _purely_ a matter of convention.  Linguistically speaking,
>the two cases are examples of exactly the same thing; it doesn't matter
>whether we've arbitrarily chosen to use the name of the parent language as the
>name of one of the daughters.