One of the my lists is Tango-L, on the Argentine Tango.  Recently, many
posts have started appearing in Spanish (English was almost exclusively
the language of the list until then).  Since many of those closest to
the subject matter of the list are in Argentina (or from Argentina), and
have trouble expressing themselves in English (though many can understand
posts written in English), I do not want to discourage this.  But there
are two new problems it creates: (1) Translations and (2) Language-
specific distributions.  Since these must surely be common to other
bilingual lists, I thought I would describe my plans/ambitions here and,
I hope, get some comments and suggestions from others who have been
running bilingual lists for some time.
 
1. Translations
 
My _guess_ is that 98% of list members understand English and perhaps
25% understand Spanish.  Also, perhaps 90% can write fluently in English
and about 15% in Spanish.  (These numbers are likely to move further
towards the Spanish end as the list moves more towards becoming
bilingual, and also as Internet access becomes more widely available in
Latin America.)
 
Some bilingual volunteers from the list have translated some of the more
interesting Spanish articles into English but with the list volume up to
15 per day with perhaps 3 of those in Spanish, they cannot do that for
all articles.  So, I would like to set up an automatic translation
server, which would repost to the list a translation (potentially both
ways) of all articles.  I am well aware of the pitfalls of automatic
translation, and how easy it is to lose and change meaning, or produce
gibberish, but it would still provide some idea of the meaning of an
article and, besides, I am interested in the experiment anyway.
 
Since there is surely much research going in automatic translation, I
would like to set up a collaboration with a University department.
Perhaps they would be willing to set up their software as an e-mailable
server--in return, there are probably several people on the list who
would volunteer to provide high-quality feedback on the quality of the
translations.
 
The alternative is to purchase commercial software but the problems with
that are: (1) I believe I can get better quality results the first way--
especially with my budget, anything I buy is likely to be a "toy"
translator, and (2) anything within my budget (say <~$300) is likely to
be on a PC platform, and I need something that will run on Unix to make
an automatic server out of it (within the platforms and facilities
reasonably available to me).
 
My questions for people are:
a. Do you know of people I can contact for the former (University
   collaboration) approach, or how I might find contacts?
b. What software do you know of that is available for this purpose
   that will run under (a) DOS/Windows and (b) Unix?
c. Has anyone tried something like this, and with what results?
 
2. Language-Specific Distributions
 
This is easier (I think...), other than the well-documented difficulty
of having people conform to the correct Topic names.  My plan is to use
Topics as follows:
 
ENG: For English originals (this would also be the default)
ESP: For Spanish originals
TENG: For translations into English from Spanish
TESP: For translations into Spanish from English
LANG: For languages other than English or Spanish (these are almost non-
      existent so far)
 
English-only members would subscribe to ENG & TENG, Spanish-only to
ESP and TESP, Bilingual to ENG & ESP (to get the originals) and possibly
to TENG/TESP if they are also checking on the translation quality (or
like receiving lots of mail :-).
 
The translation server would also key off these topic names, presumably,
to figure out what the source language was.
 
Any thoughts, anyone, on any of this?
 
Thanks.
 
Shahrukh Merchant