The Importance and Predictive Power of Linguistic Description: Evidence from Korean P2P Lending

Purpose: This study aims to analyze the impact of borrowers' linguistic traits on investors' funding decisions and loan repayment outcomes in P2P lending. Design/methodology/approach: Logistic regression, relative weight analysis, and 10-fold cross-validation are used to compare models with and without linguistic variables. Wilcoxon signed-rank tests are used for statistical validation, and predictive performance is evaluated through area under the curve (AUC), accuracy, and Cohen's Kappa metrics. Findings: Borrowers' linguistic traits, including text length and word usage related to pronouns, numerals, adjectives, emotion, and financial condition, significantly enhance predictive power for investors' funding decisions and loan repayment outcomes, outperforming traditional non-linguistic factors such as loan maturity, income, and work experience. Research limitations/implications: This study highlights the predictive power of borrowers' linguistic traits in P2P lending but identifies several limitations. First, while linguistic traits significantly predict funding decisions, their impact on repayment outcomes is not statistically robust, which necessitates further research. Second, reliance on the Korean Linguistic Inquiry Word Counts (KLIWC) tool limits linguistic analysis depth, which suggests the need for advanced natural language processing (NLP) methodologies for richer insights. Third, cultural and linguistic differences may restrict the generalizability of the findings to non-Korean contexts, underscoring the importance of cross-cultural studies. Future research should address these limitations to enhance the understanding of linguistic traits in financial decision-making. Originality/value: This study uniquely quantifies the predictive power of borrowers' linguistic traits in P2P lending using statistical models, offering insights into the interplay between linguistic and traditional financial factors in predicting funding and repayment outcomes.