The analysis of medical narratives and the generation of natural langu
age expressions are strongly dependent on the existence of an adequate
representation language, Such a language has to be expressive enough
in order to handle the complexity of human reasoning in the domain. So
wa's Conceptual Graphs (CG) are an answer, and this paper presents a m
ultilingual implementation, using French, English and German. Current
developments demonstrate the feasibility of an approach to natural Lan
guage Understanding where semantic aspects are dominant, in contrast t
o syntax driven methods. The basic idea is to aggregate blocks of word
s according to semantic compatibility rules, following a method called
Proximity Processing, The CG representation is gradually built, start
ing from single words in a semantic lexicon, to finally give a complet
e representation of the sentence under the form of a single CG. The pr
ocess is dependent on specific rules of the medical domain, and for th
is reason is largely controlled by the declarative knowledge of the me
dical Linguistic Knowlege Base.