The paper evaluates five measures of lexical semantic relatedness that use WordNet as their central resource, focusing on their performance in detecting and correcting real-word spelling errors. The Jiang and Conrath measure is found to be superior to those proposed by Hirst and St-Onge, Leacock and Chodorow, Lin, and Resnik. The authors also explain why distributional similarity is not an adequate proxy for lexical semantic relatedness. The paper discusses various approaches to measuring semantic relatedness, including dictionary-based methods, Roget-structured thesauri, and approaches using WordNet and other semantic networks. It highlights the limitations of edge-counting methods and introduces scaled metrics such as Sussna’s depth-relative scaling, Wu and Palmer’s Conceptual Similarity, and Leacock and Chodorow’s Normalized Path Length. The paper also presents information-based and integrated approaches, such as Resnik’s information-based approach and Jiang and Conrath’s combined approach. Finally, the authors compare the five measures using human ratings of semantic relatedness and find that they generally perform well, with Jiang and Conrath’s measure showing the highest correlation with human judgments.The paper evaluates five measures of lexical semantic relatedness that use WordNet as their central resource, focusing on their performance in detecting and correcting real-word spelling errors. The Jiang and Conrath measure is found to be superior to those proposed by Hirst and St-Onge, Leacock and Chodorow, Lin, and Resnik. The authors also explain why distributional similarity is not an adequate proxy for lexical semantic relatedness. The paper discusses various approaches to measuring semantic relatedness, including dictionary-based methods, Roget-structured thesauri, and approaches using WordNet and other semantic networks. It highlights the limitations of edge-counting methods and introduces scaled metrics such as Sussna’s depth-relative scaling, Wu and Palmer’s Conceptual Similarity, and Leacock and Chodorow’s Normalized Path Length. The paper also presents information-based and integrated approaches, such as Resnik’s information-based approach and Jiang and Conrath’s combined approach. Finally, the authors compare the five measures using human ratings of semantic relatedness and find that they generally perform well, with Jiang and Conrath’s measure showing the highest correlation with human judgments.