RE: Multilingual Documents [was: HTML forms and UTF-8]

From: Chris Pratley (
Date: Wed Nov 24 1999 - 16:39:26 EST

See >>>

-----Original Message-----
From: Otto Stolz []
Sent: November 24, 1999 5:15 AM
To: Unicode List
Subject: Re: Multilingual Documents [was: HTML forms and UTF-8]

Am 1999-11-23 um 18:04 h hat Chris Pratley geschrieben:
> 4. It is simply easier for people to create monolingual documents. Even if
> the author is fluent in two languages, keeping the versions in sync with
> edits and making sure they say substantially the same thing is a hassle
> people don't want (but many governments are forced to adhere to)

This remark hints to a possible fallacy in the surveys discussed in this
thread: I have seen no definition of "multilingual document", though any
survey (and the resulting statistics) crucially depend on it.
>>>Actually, Otto, I did define multilingual document earlier, although in
point 4 I referred to a subset of them. From my earlier mail:
Just to clarify, please note that I did not mean to limit my definition of
"multilingual document" to side-by side translations or similar "dual
language" documents. The customer research we did showed that true
"multilingual documents" are far in the minority, whether they are side by
side translations or simply include a single word from another language. The
one exception is when the author's name is included, since people's names
are often "foreign" to the language of the document they are authoring. In
fact, the necessity of supporting the author's name in different scripts in
both the document and the meta properties of the document were major
requests in early versions of Word. (e.g. Greek for EC documents)

If only documents containing the very same content in several languages
are deemed "multilingual", then the ratio between those documents and the
other ones will indeed be exceedingly small (as that notorious limerick
says: "it requres good light and a very strong sight in order to see it,
at all").
>>>indeed, which is why we used a broader definition.

However, the software features discussed in this thread are required (or
at least useful) for any document containing text in several languages,
so all of these should have been included in the surveys and statistics;
and the questionairs should have clearly said so. Examples for such multi-
lingual documents comprise (but are not limited to) dictionaries and
glossaries, bibliographies and citations (including foreign product and
company names), quotes (including mottos, proverbs, ...), and proper names
(very common in Europe, e. g.). You will find "LANG=en" spans in virtually
100% of my own German web pages, for instance; and this very note con-
tains a clause in German, as I am too lazy to retype the automatically
generated quote-heading.

> 5. [...] (When I was growing up in Quebec, I did not see a large fraction
> of documents that were bilingual, except for certain government public-
> ations and advertisements)

I guess, in Quebec, more than 50% of all English e-mail items will contain
a person's or place name in French; right?
>>>Not really - email only rarely includes a full snail-mail address,
especially corporate internal email.

> 6. Another point made by a couple of people is that you can measure the
> ratio of documents (as we were discussing), or you can also measure the
> percent of users who *ever* have to create even one multi-language
> This percentage is considerably higher than the percentage of multilingual
> documents, more like 10-30% of users as opposed to <1% of documents. If
> are strictly looking at business cases, then this is a misleading number,
> because a feature that makes a significant percentage of users' work
> is more valuable than one that makes a tiny percent easier. So measuring
> per-document makes more sense.

I rather think that a feature that makes the work of a significant
of users easier, will constitute a business case. So measuring per-user
more sense.
>>>If I have feature A that helps a user get their work done regardless of
document content (maybe a way to make reviewing for errors faster), and
feature B that helps a user get their work done only for certain rare
document types (easy multilingual authoring), then feature A is of more
value. This is true even to a single user who sometimes creates multilingual
documents, even to the user who often makes multilingual documents, since
feature A works for all their documents and feature B does not. Obviously
the issues are more complex and need to include the "absolute" value of
feature A and B to their respective tasks, but it definitely is important to
factor in the percentage of user's total time saved rather than simply the
percentage of users.

> And, just to reiterate, anyone who knows me knows that I (and Microsoft)
> all in favour of supporting multilingual usage in Office

Glad to read this -- and I have seen this already working in Word 2000.

Best wishes,
   Otto Stolz

This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:56 EDT