Text mining technology holds considerable potential for augmenting research in the humanities, offering a novel approach to overcome the limitations inherent in traditional introspective methodologies and introducing innovative perspectives for literary analysis. This study explores the application of text mining within the humanities and social sciences, employing the Term Frequency-Inverse Document Frequency (TF-IDF) method for the feature vector representation of textual information. It constructs a linguistic feature analysis model using three computational techniques: support vector machine, logistic regression, and the naive Bayesian model. Philip Roth’s “Goodbye, Columbus” serves as the focal point of this research. The study involves preprocessing the text corpus and subsequently comparing the performance indices of the three linguistic feature analysis models to delve deeper into the relevance of syntactic and semantic feature analysis. The evaluation criteria used to delineate the linguistic characteristics of Jewish literature are identified and employed to conclude. Among the models tested, the support vector machine model demonstrates superior performance, evidenced by its higher accuracy (0.839), precision (0.866), and F-value (0.874) relative to the other models. The analysis identifies the most influential features for capturing the linguistic essence of Jewish literature as the proportion of dialogue, sentence disjunction, average word length, and word disjunction. The selected sample of Jewish literature exhibited notable characteristics, including dialogue richness (0.369), sentence rhythm (0.234), and linguistic richness (0.266). The findings affirm that the text-mining approach utilized in this study significantly enhances the linguistic characterization of Jewish literature. This method proves effective in assisting the linguistic analysis and research of Jewish literary works, thereby reinforcing the value of integrating advanced computational techniques in humanities research.