Re: Word hyphenation in LilyPond version 2.18.2.

[Werner LEMBERG]

> Der Soldat ging in die Wach-
> stube und holte seine Wachs-
> tube aus dem Schrank.
> 
> The sentence above demonstrates that probably _no_ current
> implementation of a hyphenation algorithm can be correct without a
> deep understanding of the contents of a text.

For the German hyphenation patterns we disallow hyphenation at such
ambiguous positions.  Of course, this doesn't work properly for lyrics.

> But: Are there better algorithms than the one described by F.M.Liang
> for the task of finding 99.999% of all hyphenation points?

Certainly, for example `SiSiSi' (sichere sinnentsprechende
Silbentrennung).  However, this is (a) mainly for German, (b) not
free, (c) no longer maintained, AFIAK, and (d) not implemented in TeX
or OpenOffice.

Note that hyphenation can be extremely tricky.  Taken from

  http://hunspell.sourceforge.net/tb87nemeth.pdf

here's a table that shows some complicated cases (I've removed German
since with the new spelling reform cases like `Zucker → Zuk-ker' or
`Schiffahrt → Schiff-fahrt' no longer exist).

  Language   Example     Hyphenation       Description
  -------------------------------------------------------------------
  Catalan    paral·lel   paral- lel        digraph l·l represents
                                           long (geminated) l
  Dutch      reëel       re- eel           diaeresis and hyphenation
                                           sign syllable breaks
             omaatje     oma- tje          vowel lengthening with
                                           diminutive -tje
  Greek      Μαΐου       Μα- ίου           diaeresis and hyphenation
                                           sign syllable breaks
  Hungarian  asszonnyal  asz- szony- nyal  simplified double digraphs
                                           (long sz and ny phonemes)
  Norwegian  bussjåfør   buss- sjåfør      triple consonants at
                                           compound word boundary
  Polish     kong-fu     kong- -fu         repeated hyphen at line
                                           begin
  Swedish    tillåta     till- låta        triple consonants at
                                           compound word boundary

I think the trickiest language is Hungarian...

Note that above covers only the most basic level.  In languages that
use compound words, you also need proper weighting of hyphenation
points, but this is beyond what would be needed for lilypond.


    Werner