Abstract
The present article attempts to provide some basic information on, and insight into, the topic of semi-automatic detection of neologisms. The paper is divided into two major parts, developed mainly in sections 2 and 3; the former contrasts automatic and semi-automatic detection of neologisms with the more traditional method of searching for neologisms manually ('reading and marking').
The latter explores the principles underlying the workings of neologism detection tools, while providing examples of such typically exclusion-list based tools, some of them deemed as prototypes. Section 3 furthermore comments on the additional algorithms and functions of some of the tools, viz. the exclusion of proper names (pertaining to named entity recognition), or of abbreviations, and the automatic filtering out of typos and errors.
The paper pinpoints the weaknesses and drawbacks of some of these functions as well as of the semi-automatic detection of neologisms per se.