The Norwegian ispell-dictionary home page

The most important file available here contains a list of 750 000 Norwegian words. Each word is marked with a number indicating the commonness of that word. Compound words are hyphenated at their compound points. Some words are marked as belonging to a specific classes; mathematics, oil, conservative language, 'samnorsk' etc. Words marked with a star are allowed in Nynorsk.

This file is usable to several things:

• Making dictionaries for the Ispell program of different sizes, choosing which words to include in a sensible way.
• Making Norwegian dictionaries for word processors that doesn't have one, again with a sensible subset of words.
• Making new and better hyphenation patterns for TeX.
• Text-recognition (OCR) programs.
• Encourage the e-TeX team to implement multi-level hyphenation in TeX.
• Encourage people to use frequency-information when they write programs making suggestions for replacements for misspelled words.

Routines for the first three items on the above list is included in the Makefiles. The last three was too hard to implement in Make.

Requirements

• Ispell
  For the ispell-related stuff, you need the ispell program, and you can get it from the ispell home-page. You can also find dictionaries for a lot of languages there. You also need the version of the look program in util-linux-2.9. Older versions have a bug which shows up when searching dictionaries with non-English characters. Ispell uses look to complete words (ispell-complete-word). If you don't plan to have a Norwegian words file for lookup, you don't need to worry about the Look program.
• Emacs
  If you want to use ispell from Emacs, i recommend upgrading to the latest version of ispell.el. This version supports Norwegian, and it has become clean to include local dictionary definitions. It is almost like the version included in in Emacs-20.4. There is also an add-on to ispell.el, flyspell.el, written by Manuel Serrano available, offering better `on-the-fly' spell-checking. An old version is included in emacs-20.3, but you would like to have the new version with important speed improvements.
• (La)TeX
  If you want to make your own hyphenation patterns for TeX (you probably don't), you need a version of the patgen program with greater capacities than standard versions, e.g. you have to compile patgen with a different patgen.ch. See the patterns/Makefile for more information. Almost every TeX distribution contains the patgen program. I recommend teTeX. If you want both kinds of hyphenation in the same TeX format, you probably need to recompile TeX due to capacity problems. Again, this is easy with teTeX.

Distribution

The distribution ispell-norsk-2.0.tar.gz (2204k) is free in the GPL sense and contains these files:

• README
  How to make ispell and emacs work with these dictionaries.
• words.norsk.sq
  This file contains the Norwegian words and the indication of their commonness compressed with the sq program.
• norsk.aff.in
  A template for the affix file for the Norwegian language. This file is made for ispell with 64 maskbits that understands HTML. Most pre-made versions of ispell supports only 32 maskbits and don't understand HTML. Use the patch and recompile ispell, or delete the html-related stuff.
• Ispell-3.1.20.no.patch
  A patch for ispell-3.1 that adds the amsmath and breqn environments to the skip-list and fixes a bug in buildhash. It also makes ispell html-aware, and tries to fix `the backslash bug'. In addition it makes ispell suggest "- as a compound word mark when seeing an unknown compound word, but only in TeX mode if the dictionary is named norsk. This is an ugly hack that works for me. It also implement the -r flag which is like the -a flag, but the suggestions are printed even if the word is found in the dictionary.
• norsk.single.tex
  This is a set of hyphenation patterns for TeX that works well on non-compound words. It is used when making the new hyphenation patterns for TeX. This file is basically made from nohyph3.tex, a hyphenation file I released May 1998. But a lot of errors have been removed by comparing its action on the single words by the action of nohyph.tex (standard in teTeX), nohyph2.tex, and the unreleased hyphenation patterns by Simen Gaure used at the Department of Mathematics, University of Oslo. I have tried to follow the rules given in nohyph.tex, at least where I find it reasonable. Bear in mind that there is no authoritative source for hyphenation in Norwegian. Please get in touch if you want to help improving the Norwegian hyphenation patterns.
• norsk.cfg
  An addition to Babel-3.6 for LaTeX that makes the character " active and offers you many `different' hyphen signs. You can say o"ppussing in LaTeX to get correct hyphenation opp-pussing! This functionality will appear in Babel-3.7 for Norwegian. Danish and Swedish have had it for several years.
• inorsk-compwordsmaybe
  Search for words in a file or from standard input that maybe should be written in one word. Like `matematikk lærer' etc.
• inorsk-hyphenmaybe
  Search for words in a file from from standard input that the Norwegian hyphenation patterns from this distribution might not hyphenate properly. Incorrect hyphenation of words not printed is considered to be a bug in the patterns. There is only a finite number of them.
• Makefile
  This file contains rules for making dictionaries for ispell and lists of the most common words for dumb word processors. There is also a Makefile in the patterns directory for making hyphenation patterns.
• nohyphbc.tex, nohyphb.tex
  This is the hyphenation patterns for TeX. The file nohyphbc.tex hyphenates only at compound points. The nohyphb.tex hyphenates each component of a word too, but avoiding to hyphenate 'near' compound points. I think 'bar-nepsykologen' looks really bad. Too bad TeX doesn't support multi-level hyphenation yet.
  The naming of the files follow the paradigm in Babel; if a replacement for a file foo.bar is offered, it is named foob.bar, where the b stands for big. These new patterns easily outpreforms those available before, mostly because of better compound word hyphenation. For reference I have made lists of about 2000 compound word errors made by previous patterns: err.nokyph and err.nokyph2.
  The size of the patterns can be argued over. The patterns are copied into each format file, thus occupying some disk space. They also limit the number of languages one can load hyphenation patterns for on most TeX systems. But size considerations has become less important recent years, so I prefer to focus on getting things right, not small. It is also possible to recompile teTeX such that there is more room for hyphenation patterns, but the patterns take up more memory then. There is surely a lot of unnessesary structure within the hyphenation patterns , but it is very time-consuming to remove. The file patterns/Makefile can be configured, such that one can make smaller sets of patterns, taking only the most common words into concideration. Everyone is invited to play.
• COPYING
  The GNU general public license.

Changes

There has been a lot of changes since version 1.1a. The quality has improved a lot, and the structure of the distribution is completely new. Therefore i choose not to make the previous versions available from this site.

Here is a rough summary of the changes:

• New distribution format
• Support for Nynorsk
• Commonness indicator for each word from Bokmål
• Words are hyphenated at their compound points
• A lot of common words added, especially compound words
• Makefile completly rewritten. It is possible to configure the size of the dictionary for ispell without beaking the munching.
• Makefile to make hyphenation patterns for TeX added
• The pregenerated TeX patterns are included in the distribution.
• Controlled compoundwords support added. This includes affix file updates.
• Some uncommon and misspelled words removed
• Affix file updated for html. This will only work if you use the patch.

Todo list

• Remove/mark uncommon words that are close to common words. If you type 're' you probably meant 'er', even if 're' is a valid word.
• There are too many words with commonness 0. Split this group in two.
• Some words in the basic category belongs in special categories. When making a small dictionary with all words from mathematics, many such words are missing, since they are in the basic category. They should be moved.
• Make ispell sort the suggested replacements for misspelled words by commonness of the suggested words. One (easy) way to do this is to make an external file containing the most common words, and make ispell look into that file each time it has more than one suggestion. Or the file could be read into memory. (I don't think frequency information is representable within the root/affix structure, since one flag can represent multiple words.) This would slow ispell down a little bit, but only when it makes suggestions. If you would like to help with this, please get in touch.

Comments, suggestions and bug-reports to runekl@opoint.com. If you have or want to make a correct dictionary from some field of knowledge, i would like to include it in the next release. See the README file for some suggestions about how to get started. All you need is a large amount of Norwegian text from the field in question and some time to organize the dictionary.