Copyright 2007 TeX Users Group.
You may freely use, modify and/or distribute this file.

-----------------------------------------------------------------------------
	Hyphenating compound words			(compound.txt)
-----------------------------------------------------------------------------

TeX normally will not attempt to hyphenate compound words (i.e. words that
already contain a hyphen).  This can be a problem when using languages 
(s.a. German) in which compound words are common.  

There is a way to deal with this if the font encoding contains `hyphen'
in two places.  This is the case with TeX 'n ANSI encoding, Windows ANSI
encoding, as well as Cork (T1) encoding.  This cannot be dealt with when
using the old `TeX text' (OT1) encoding (as used by CM text fonts).

(1) The first step is to tell TeX that `hyphen' is to be treated just like 
    an ordinary letter, by giving it a valid `lower case code.' E.g.

	\lccode`\-=`\-		% or \lcode45=45

(2) The next step is to tell TeX to use the character code for the *second*
    occurence of `hyphen' in the encoding when inserting a `hyphen'
    (The reason for this will become apparent in the next step).
    Typically `hyphen' will occur at char code 45 in the character layout.  
    In TeX 'n ANSI, and Windows ANSI, it is repeated at char code 173
    (the latest interpretation of Cork encoding has it appear again at 127).

    It is possible to set the default hyphen character:

	\defaulthyphenchar=173

    Note that this will *only* affect fonts read in after this definition
    (So add this *before* \input lcdplain, \input lcdlatex, \input mtplain,
    \input \mtlatex etc.)  Note also that fonts for which metric files have
    already been loaded in the TeX `format' are *not* affected.

    One can also change the \hyphenchar for a specific font: 

	\hyphenchar\myfont=173

(3) With this change, it can happen that TeX decides to hyphenate a word at
    an existing hyphen; thus creating a *double* hyphen.  This problem can be
    dealt with by creating a pseudo ligature that maps `hyphen' followed by
    `hypen' into a single `hyphen'.  

    It should now be clear why we could not let TeX simply use the ordinary
    hyphen when it decides to break a compound word: we would have had a
    conflict with the existing standard pseudo ligature `hyphen' `hyphen' =>
    `endash'.  The new pseudo ligature instead is `hyphen' `sfthyphen' =>
    `sfthyphen' where `sfthyphen' stands here for the character code of the
    second occurence of `hyphen'. 

    AFMtoTFM version 2.6 (or later) automatically creates the required 
    pseudo ligature when the font encoding contains two copies of `hyphen'
    (assuming it has been asked to insert standard ligatures using `-a').
    If there is no second copy of hyphen it will look for `sfthyphen'.

(4) Finally, the hyphenation tables have to be re-created to take into 
    account compound words.  The new hyphenation patterns will contain
    some patterns that involve `hyphen'.  This requires rebuilding the
    hyphenation patterns for a language that has many compound words.
    While this is likely to be quite a bit of work, it only has to be done
    once for a given language (watch CTAN for new hyphenation patterns).  

    A short-cut --- if a full revision of the patterns is not feasable ---
    is to use the file `hypht1.tex' that may be found on CTAN.  It takes
    existing patterns and adds patterns that prevent hyphenation near
    an explicit hyphen (see instructions in the file itself).

    Such new patterns will work for a variety of encodings, since `hyphen' is
    at char code 45 in virtually all resonable encoding for Western languages.