The Muenster lexical database holds lexical data for
five languages. For each language, for a certain amount of lexical
entries, both linguistic and statistical information is contained. The
richness of the lexical information varies from language to language.
Ca. 10 000 entries of each languages are connected by a bilingual
equivalent-relation.   
Data in the Muenster (relational) lexical
database are stored in tables and accessible through a relational
query language (SQL). Conversion of (portions of) the data is possible
and carried out by programming languages (C, PROLOG) with embedded SQL
statements.

I. monolingual data

a. English

  86 000 wordform:lemma:pos tuples 
  47 000 different strings serving as entry points for further linguistic
         information 

  lemma:fullform relation for the English data:

    verbs: 1:4
    nouns: 1:2
    adjectives: 1:1 (regular)


b. Spanish

   45 000 different strings serving as entry points for further linguistic
          information 
    6 500 collocations
   55 000 word-form:lemma pairs

   lemma:fullform relation for the Spanish data

      verbs: regular - about 1:20
             irregular - about 1:45
      nouns: regular 1:2 - 1:4 
      adjectives: 1:4
	


c. French

   135 000 word-form:lemma pairs
    30 000 lemmas
   
   lemma:fullform relation for the French data

      verbs: 1:20 - 1:30
      nouns: 1:2 and 1:4
      adjectives: 1:4

d. Italian

  70 000 word-form:lemma pairs
  12 000 different strings serving as entry points for further linguistic
          information
 



Italian ans Spanish lemmatization is actually worked on.
For these languages the part of speech infoormation for every lexical
entry is specified.

Besides that, there exist ordered frequency lists derived from the
corpora.


e. German

for 485 000 word-form:lemma pairs there exist detailed grammatical information
            describing the place(s) of the word-form in the paradigm of
            the lemma. This information is generated by a morphological 
            generator. GEnerated forms are compared with documented 
            full form extracted from the corpora.                

for 130 000 strings serving as lexical entries you find the following
information:

  GRAPH: citation form
         syllable structure
  SYNCAT
  MORPH: for VERBS: perfaux
                    separable prefix
                    perfect prefix  
         for NOUNS: GENDER
                    inflection classes
                    stem alternation

  SYNSEM: for VERBS: SUBCAT


II. bilingual data

For the following language pairs bilingual dictionaries in both directions
are available or worked on.

German - English
German - Spanish
German - French
German - Italian
German - Standard Modern Greek

Each dictionary comprises 10 000 lexical entries. For each entry the
equivalents in the target language are given. For more than one target
equivalent the readings of the source lemma are distinguished by semantic
glosses in the source language. The lexical entries are selected by
weighting the absolute frequency of their occurrence based on the corpora.