This study explores the application of the NooJ tool in detecting and analyzing Croatian words that contain affixoids, namely, prefixoids and suffixoids, primarily derived from Latin and Greek. Known as neoclassical compounds, these words significantly influence modern Croatian, notably in scientific, technical, and cultural registers. Although previous studies have addressed classical word-formation processes, comprehensive lists and consistent classifications of these affixoids remain scarce. To address this gap, two morphological grammars were developed to recognize prefixoids and suffixoids, respectively, and were tested across six diverse corpora: children’s literature, adult literature, parliamentary speeches, medical texts, and web content. The grammars were refined through manual verification and further analyzed using visualization tools such as Tableau and Voyant. The results reveal clear domain-based variation in affixoid use and segmentation accuracy, highlight challenges in suffixoid detection, and lead to an expanded dataset comprising 75 suffixoids and 224 prefixoids. By integrating linguistic theory with computational implementation, this study offers a robust framework for affixoid recognition and contributes to the development of scalable tools for morphological analysis in Croatian and other morphologically rich languages.