Algorithm for turning Oahpa entries

  1. revert all <l> - <t(f)> pairs, for which the tg HAS a sematics element.
  2. unify for identical instances of former <t(f)>, now <l>
  3. remove all <l> (i.e., <e>) for <l> with no stat="pref" tag
  4. keep the rest of the Norwegian lemmas
  5. remove multiple stat="pref" tags manually (Comment: This can't be the case!)


  1. what about attribute oa?
  2. what about lacking this attribute?
         <l pos="a">aales</l>
         <app name="oahpa">
               <book name="åa5" />
               <sem class="NATURE_A" />
               <sem class="HERDING" />
            <t dict="yes" oa="yes" pos="a" stat="pref" tcomm="no" xml:lang="nob">rovdyrfritt</t>
            <tf dict="no" oa="yes" pos="phrase_a" tcomm="no" xml:lang="nob">fredelig for rovdyr</tf>
            <tf dict="no" oa="yes" pos="phrase_a" tcomm="no" xml:lang="nob">fred for rovdyr</tf>
            <tf dict="no">fritt for rovdyr</tf>
         <tg xml:lang="swe">
            <t dict="yes" oa="yes" pos="a" tcomm="no" xml:lang="swe"></t>

Principle for dictionary conversion.

  1. Turn smanob to nobsma
  2. Thereafter, manually remove superfluous <t> synonyms in smanob tg-s
  3. in the new nobsma, unify according to lemma
  4. for lemma pairs with conservative/radical Bokmål, pict the (e.g.) conservative as the lemma, and make a pointer from the radical one

Cf. the following example:

vatn -> vann  

	1. (væske) tjaetsie
	2. (innsjø) jaevrie
	(væske) vann