SOME PROBLEMS OF LINGUISTIC DATA BASES 
Jurgen Kunze 
Zentralinstitut fur Spraohwisssnschaft, AdW DDR, Berlin, DDR 
1. Data bases, as parts of procedures 
It is of some advantage to compose procedures for auto- 
matic langua6e processing of two parts: 
(i) a logical part (rspresent~u~ the algQrithm proper) as 
the procedural component; 
(ii) an information part (containing the specific items of 
information and depending on concrete languages and 
purposes) as the data base. 
The logical part determines the structure of the items 
in the information part, preferences between them and their 
contribution to the final result. The information part is 
variable. By designing the logical part sufficiently general, 
one can use it even for various natural languages, if one 
combln~s it with suitable information parts. The variability 
of the information part admits easy corrections and complet- 
ions of the whole procedure. This is illustrated by some 
examples taken from wordfo~n analysis. 
2. Modular structure of logical parts. 
Another important aspect is the connection between the 
structures of the logical part and the information part (data 
base). If the logical part consists of some (rather independ- 
ent) sub-procedures, the information part may be composed in 
a similar way: Either a separate information part for every 
- 165 - 
module or a correspondence between modules and informations 
of a certain type within one (or some) information part(s) 
for more than one module. Such principles allow an easy simpl- 
ification of a full version and facilitates the extension 
of a simple version. 
3. Data bases as tools for establishing procedures 
In many cases the elaboration of procedures may be 
essentially facilitated by data bases: the base contains 
either a representative sample or the whole set of objects 
to be processed by the procedure together with the desired 
results. Combined with a certain (mostly rather simple) dial- 
ogue languages this type of data bases admits an automati~ 
look up of specific types and combinations, which is very 
advantageous for the elaboration, if a manual look up is im- 
possible (at least in a reasonable time). One special aspect 
is the compensation of lack of la~Euage competence. Further- 
more the same data base allows an automatic test of the fi~l 
version of the procedure. This is illustrated by works on 
automatic hyphenation. 
4. Data bases as tools for linguistic research 
The data bases contain again linguistic objects together 
with additional informations (e.g. structural descriptions, 
classifications). Using a (more sophisticated) dialogue 
language one can put a lot of questions %o the data base and 
condense the results to frequencies, correlations,mutual 
exclusions and so on. This is illustrated by applications of 
data bases to research on syntax of na%~ral languages. 
5- Other applications 
Here one may consider terminological data-banks, data 
bases as dictionaries ox -- as a mean for the elaboration of 
(usual) dictionaries etc. 
- 166 - 
