Message-Id: <9212171756.AA04636@sc.zib-berlin.dbp.de>
Reply-To: LaTeX-L@vm.urz.Uni-Heidelberg.de
Comments:     Resent-From: "Werenfried Spit" <SPIT@vm.ci.uv.es>
Comments:     Originally-From: "Werenfried Spit" <SPIT@vm.ci.uv.es>
Date:         Thu, 17 Dec 92 18:44:21 EST
From: Werenfried Spit <SPIT@VM.CI.UV.ES>
Sender: Mailing list for the LaTeX3 project <LATEX-L@vm.urz.Uni-Heidelberg.de>
To: Multiple Recipients of <LATEX-L@vm.urz.Uni-Heidelberg.de>
Subject:      Some ideas about multilingual use of BibTeX
Status: R


%  Having developed a bibliographystyle for dutch users I have
%  some ideas about the use of BibTeX in a non-english environment.
%  Below is an article that resumes these. I'd appreciate if
%   * developers of LaTeX 3
%   * developers of BibTeX 1
%   * developers of the Babel package
%  would read it and comment on it. I hope it's of some use.
%
%  Warning 1: the closer the subject gets to actually doing something
%             (like writing a style file) the sloppier the ideas are.
%  Warning 2: if I say that something needs to be done it does not mean
%             it cannot be done with LaTeX2+BibTeX0.99, it just means
%             that I do not know how to do it easily
%
\documentstyle[a4,english,sober]{article}
\newcommand{\BibTeX}{{\sc Bib}\TeX}
\newcommand{\BBT}{Babel-\BibTeX}
\begin{document}
\title{Multilingual \BibTeX}
\author{Werenfried Spit \\
        {\sl  Departament de F\'\i sica Te\`orica, Universitat de Val\`encia}\\
        \verb|spit@vm.ci.uv.es| }
\maketitle
\section{Introduction}
The Babel package of style files makes \LaTeX\ (and plain \TeX)
better suited for writing in other languages than english and
for writing multilingual documents. I think a corresponding functionality
is needed in \BibTeX. I will try to specify what we should
demand from a \BBT-package, and maybe even develop some ideas
about how to implement the demands.

I got this not-so-original-idea while I was updating a
dutch bibliography style; the situation is completely the
same as Johannes Braams describes in Tugboat {\bf 12} (1991) for designing
a language specific \LaTeX-style. Just as for \LaTeX-styles
it is better to move some language specific parts to separate
files, it is for \BibTeX.


Babel does, in its present incarnation, three things for two types
of \LaTeX-users:
\begin{itemize}
\item adaptation of text that is input by style files to other languages
      than english (\verb|\chaptername|, \verb|\today| and the like)
\item dynamically loading hyphenation patterns (if possible)
\item making input easier (especially for accented letters)
\end{itemize}
for
\begin{itemize}
\item anyone who writes a document in language other than english
      (or even english)
\item people that include longer parts in another language in their
      documents, or even write essentially multilingual documents.
\end{itemize}

Parallel to this taxonomy I'll try to formulate what \BBT\ might do.

\section{Demands}
There are a few actions of \BibTeX\ that need to be internationalized
by  \BBT, just like Babel internationalized some \LaTeX-functions.
I think we need:
\begin{enumerate}
\item Language-specific versions of strings and sentences. This would
      be things like \verb|\andname| to get
      \verb|J. Smith \andname\ T. Jones|  to produce
      ``J. Smith and T. Jones''\footnote{In fact
         we need two versions of and, since in cases with many authors
         separated by comma's some languages require ``, and'' between
         the penultimate and the ultimate name, while others do without
         the comma.}
      in an english bibliography, or \verb|\partofname{#1}|
      to make \verb|\partofname{2}  Nuttige Verhandelingen| produce
      ``Deel 2 van Nuttige Verhandelingen'' if the language is dutch.
\item Language-dependant sorting algorithms. These are threefold:
      \begin{enumerate}
      \item We need the possiblility to specify an alphabetical ordering, e.g.
            to enable a scandinavian user to get all \aa\ sorted after z; this
            will be especially needed by users using 8-bit input.
      \item We need to be able to change the way names are parsed to the
            sorting algorithm since e.g. English prefer Ludwig van Beethoven
            to appear under `v', whereas Dutch and Germans list him under `b'.
      \item When titles are parsed to a sorting algorithm we need an
            expandable list of `choppable words' like ``A'' and ``The'' for
            english, ``Un'' or ``Sur'' for french, etc.
      \end{enumerate}
\item Making input of names easier. Although with proper bracing
      and permutationing of first `von' and last parts of names
      anything that is wished can be accomplished, it might
      be desirable to facilitate things a little bit for some.
      \BibTeX\ supposes that a name consists of multiple first
      names, a `von' part, one last name and a junior part. This
      is indeed the standard format for many names. In spanish,
      however, names normally consist of a first name and two
      last names, each possibly with its own `von-part'. It would
      be handy if \BibTeX\ could switch to a mode in which names
      are interpreted according to this scheme. Probably there
      exist other languages with other standard naming schemes.
\end{enumerate}
The first two items of this list are the most important, I think, and
fortunately probably also the easiest to implement; maybe even already
in \BibTeX0.99. The last item is trickier, and maybe not as badly needed.

The use of multilingual facilities could be twofold:
\begin{itemize}
\item Bibliographies in documents in other languages than english.
\item Multilingual bibliographies (in which each entry in the
      bibliography is quoted in the language of the {\em cited} document,
      instead of the language of the citing document).
\end{itemize}
At first I thought that the first case would be the common one, and the
second rather rare. Then, I read the AMS-\LaTeX\ manual where it says that
mathematicians need a cyrillic font to be able to cite Russian papers
with their original title; I also realized that in Russian physics journals
Russian papers are cited in Russian, but English in English. It would
certainly be very strange to cite
 \begin{verbatim} \bibitem{1} {\cyr A.B. Popov} and {\cyr V.G.
Popov}\end{verbatim}
instead of
 \begin{verbatim} \bibitem{1} {\cyr A.B. Popov i  V.G. Popov}\end{verbatim}
even in an english paper, but I found that many books extend this
practice to any language, so you would find
\begin{verbatim} \bibitem{1} J. Jansen en K. Jansen,
             {\em Over Nederlandse voornamen}. Amsterdam, 1980.
 \bibitem{2} W. Smith and X. Smith,
             {\em English family names}. London, 1982.
\end{verbatim}

\section{Implementation}
Some of the language-dependant facilities mentioned above
could be implemented by adding them to the Babel \LaTeX-styles,
as others, however, can only be handled inside \BibTeX, I think
it would be best to let \BBT\ exist of a set of bibstyles
(\verb|.bst|-files) that handle everything.

A scheme to use these could be the following:
\begin{itemize}
\item \LaTeX\ parses (via the \verb|.aux|-file) the language
     of the document (or better: the current language at the
     point in the document where the bibliography is to appear)
     to \BibTeX.
\item \BibTeX\  starts by reading from the \verb|.aux|-file not
     only the name of the bibstyle (\verb|.bst|-file) and the
     name of the bibliography(\verb|.bib|-file), but also the
     value of the language. Then it reads the modification
     file \verb|language.bst| to get the values of language
     dependant strings, the sorting scheme, etc.
\end{itemize}

For multilingual bibliographies a new type of \BibTeX-field needs
to be introduced: \verb|language|. This field could be defaulted
to a value given at the top of a biliography file, or to the
language of the citing document. (Thus, if a user has only french
documents in a \verb|.bib|-file she sould not need to put
\verb|language = "french"|  in every entry.) If a user wants a
bibliography to be multilingual the bibliographystyle should
use the value of \verb|language| to determine for each entry
which values of strings like \verb|\andname| to use, how to
format dates, and which font family to use (to be able to
switch to cyrillic, greek, hebrew or any other alphabet).
The language field should not be used to switch completely
to the corresponding \LaTeX- or \BBT-style as reading hyphenation
patterns would be an unnecessary overload, and alphabetization
should not change (to be consistent throughout).

The sorting of names could be performed by defining macros
like \verb|firstlast|, \verb|lastfirst| or \verb|vonlastfirst|,
that are defined in terms of \BibTeX's name formatting
control sequences like \verb|ff| or \verb|l|.
These could than e.g. be defined by the language specific file as:

\begin{tabular}{|l|l|l|} \hline
                 & english                      & dutch              \\ \hline
\verb|firstlast| & \verb|{ff }{vv }{ll}{,jj}|   & \verb|{ff }{vv}{ll}{,jj}|  \\
\verb|lastfirst| & \verb|{vv }{ll}{,jj}{, ff}|  & \verb|{ll}{, ff}{ vv}{,jj}|\\
\verb|citename|  & \verb|{vv }{ll }|            & \verb|{vv }{ll }|          \\
\verb|sortname|  & \verb|vv ll ff jj|           & \verb|ll vv ff jj|  \\ \hline
\end{tabular}

This would than produce:

\begin{tabular}{|l|l|l|} \hline
                 & english                    & dutch              \\ \hline
\verb|firstlast| & Ludwig van Beethoven       &  Ludwig van Beethoven \\
\verb|lastfirst| & van Beethoven, Ludwig      &  Beethoven, Ludwig van \\
\verb|citename|  & van Beethoven              &  Van Beethoven \\
\verb|sortname|  & {\tt van beethoven ludwig} &  {\tt beethoven van ludwig} \\
\hline
\end{tabular}


The biblographystyle would then use \verb|sortname| to sort names,
choose one of \verb|firtlast|, \verb|lastfirst|, or similar to format
the name in the bibliography, and use \verb|citename| to create
a citation key for those bibliographystyles that allow citation
by author (and year) instead of a number.
\end{document}

P.S. I {\em do not} volunteer to actually create \BBT, I do
     volunteer, however, to $\alpha$- $\beta$- and $\gamma$-test anything
     that is written to address these issues.

------------------------------------------------------------------------
Werenfried Spit                         tel: +34-6-386 4551
  Departament de F¥sica Teèrica         spit@vm.ci.uv.es
  Universitat de ValÛncia               spit@evalun11.bitnet