This paper describes the formation of data and knowledge base for a spoken 
interpreter system based on the model of speech signal understanding within
subject areas including meta-words. The proposed structure allows 
for operating the subject areas, languages, sentence and meaning types and 
meta-words. The respective toolkit is developed and described. 
The experimental system of a spoken phrasebook-interpreter exposed results 
proper for its practical application. The presented demo-version is provided 
with the data and knowledge base for a new language created by the developed 
means.