Assessing Dutch Syllabification Algorithms and Improving Accuracy by Combining Phonetic and Orthographic Information through Deep Learning

In three linesComparative assessment of four Dutch syllabification algorithms (Brandt Corstius, Liang, Trogkanis-Elkan CRF, and a novel deep learning model). The deep learning model combining phonetic and orthographic information achieves 99.65% word accuracy (+0.14% improvement over literature). Data-driven algorithms outperform knowledge-based approaches.

Summary generated by Claude — human-verified