Reconstructing a future language

Håkan Lindgren <h5@...> wrote
<<Since reconstruction of past languages started, we have learnt a lot about
language development over time and about the way words change according to
sound laws. What if we used this knowledge not to look into the past, but to
take a look into the future?>>

Just as a side-bar:

An ordinary test of a hard science is its predictive power. Chemical
reaction formulas, for example, based on past observed reactions can predict
future reactions. Experimental psychologists evaluate hypotheses about
learning based how well they predict "rates of response" in animals. On a
very practical level, geologists predict the presence of oil in the ground
based on geologic configurations where oil has been found in the past.

Where the factors affecting future results can't be entirely detected or
predicted, e.g., biological evolution, predictability nevertheless manifests
itself in a directional way. For example, selective breeding in animals and
plants do yield predictable results. And induced mutation in plants has
produced the generally intended results, along with quite a bit of unintended
slop.

Linguistics (but not always mainstream linguists) has been successful at
breaking the code of ancient languages -- and that proves its predictive
power. But it should also be pointed out that when Hittite and Mycenaean
were deciphered, neither language fit what was then being predicted they
would be like. A good deal of this was actually due to "extra-linguitic"
assumptions about history. Hardly a linguist predicted that Linear B would
be Greek and eventually some predicted it would be proto-Greek -- which it
was not. These guesses were based on historical assumptions about "the
coming of the Greeks" and the "Indo-Europeans" that were made out-of-date by
the deciphering of Linear B, among other things. It is these extra-linguistic
assumptions that affect not only linguistics predictive power, but also its
power to interpret the past.

Joseph Greenberg once wrote that he would like to see someone honestly try to
reconstruct the proto-language of the modern Romance languages -- without
using any knowledge of Latin -- and then compare it to Latin. The difference
between that "reconstructed proto-Romance" and Latin might not only
illustrate the predictive power of the methodology, it might also tell us how
accurate our reconstructions of PIE, for example, might be.

Steve Long