Re: Computerising Sound Changes (was: Proto-Albanian Timeframes)

From: alexandru_mg3
Message: 42441
Date: 2005-12-07

--- In cybalist@yahoogroups.com, "Richard Wordingham"
<richard.wordingham@...> wrote:
>
> --- In cybalist@yahoogroups.com, "alexandru_mg3"
<alexandru_mg3@...>
> wrote:
>
> > Next I will write a program that "will go down and up" based
on
> > these rules....generating different outputs for one input
(should be
> > only one but because we are not sure about all these rules will
be
> > more than one...) and different inputs for the same output.
> >
> > This program will be of course public on this forum, so there
isn't
> > any private setup here (not to create confusions).
>
> I'm not sure about multiple outputs - that gets very complicated
very
> rapidly - but for applying a single set of rules which one can
change
> manually, there's Mark Rosenfelder's sound change applier at
> http://www.zompist.com . Several people have converted it to a web
> page, which anyone can run without too much worry about viruses.
>
> Anyone using there employer's computer may be a little nervous
about
> downloading executables or using Excel macros! However, there is
no
> problem with the web page. I've used it at lunch time on my
paranoid
> employer's computers.
>
> My effort, which can be run off-line, is available via
>
http://homepage.ntlworld.com/richard.wordingham/sounds/prep5d_rom.htm
> . It's split into HTML which can be customised to present the
data,
> and a javascript file which need can be shared between data sets.
I
> believe Miguel occasionally uses it, though he may have made his
own
> notational enhancements.
>
> Mark Rosenfelder's sound change applier seems to be the closest
there
> is to a simple, practical standard for expressing sound changes.
>
> The big problem with it is that it doesn't handle segment copying
> rules. Italian gemination or West Germanic germination needs a
> separate rule for every consonant! I made an experimental
> generalisation to handle Sean Whalen's proposed sound changes,
which
> did need segment or feature copying, and the experimental version
ran
> much slower.
>
> Richard.
>

Thanks a lot Richard for this update (I will take a look)
The Excel will be only to keep all the rules together (will be only
a document not a program with Macros etc...) in an 'easy to change
way'...

Marius