Developing Algorithms 
for Computational Comparative 
Diachronic Historical Linguistics

Over time, languages change by regular, systematic processes.  It is possible, by looking at the state of a language now and in the past, to deduce the exact changes that occurred, and the order in which they occurred.  These changes also split languages, therefore it is also possible to, by using modern languages as input, induce the probable structure of their parent language.

I have developed algorithms to form a web of languages, similar to that to the side, given a set of modern languages.  The languages would be the leaf nodes on the tree. 
They function recursively.  Given a group of languages, one algorithm forms a proto-language, a sort of average from which all of the others are probably derived. The algorithm then splits the group into different groups of languages more closely related to each other than the proto-language.  Each of these groups is treated in the same fashion until each group contains a single language.

This process can be applied to any abstract structure, not merely languages.  This makes it useful 
outside of my project, but if any use is going to be made of it in practical historical linguistics, it must
be applied to languages.
The ideal way to deal with language historically is purely phonetically, only through sound.  As
traditional alphabets are woefully ill-conceived, it is necessary to create an organized system to store
phonetic data.
I developed a 5-dimensional array structure, generally following the lines of the International 
Phonetic Alphabet (IPA).

Phonetic data can also be used in 
lower-level attempts to compare 
languages to each other, but direct
phonetic comparison and building from
the bottom-up has proved far less
effective than building language
relations from the top down, which 
requires far less intense phonetic 
computation.

Daniel Wright
2005