Lexicography and Generative Artificial Intelligence: A Review in the Framework of Article Heading Definitions

Abstract

In this study, to what extent and how artificial intelligence can be utilised in the definitions of words in the ‘analytical definition’ class is emphasised. In this context, the words greyfurt, mandalina, portakal and turunç, which are defined by analytical definition from TDK Turkish Dictionary, but which are observed to have problems in their definitions and which are species of the same genus, are taken as examples. For this purpose, ‘ChatGPT 4o’ and ‘Gemini 2.0 Thinking Experimental’, which are widely known and used artificial intelligence models, were selected. Four different commands were given to the models in order to understand whether these models could give the desired results with the correct instructions. As a result of these commands, it was observed that the models provided accurate descriptions in terms of form and produced descriptions that could be used with some corrections in terms of information. In addition, it was determined that the models produced semantic trees that can be used in lexicography thanks to the sample semantic tree (semagram) given to the models. As a result, in this study, in which Turkish commands were given and Turkish answers were requested, it was understood that artificial intelligence was effective in the process of analytical definition creation with correct guidance. It has been concluded that artificial intelligence can be used as an informative and accelerating tool in dictionary preparation processes, thus saving both time and cost.