Based on corporate social responsibility (CSR) reports publicly released by listed companies from 2008 to 2022, the textual content was analyzed using Named Entity Recognition (NER) and Term Frequency-Back Inverse Document Frequency (TF-BIDF) techniques, and the mechanism explored through which the signals transmitted by CSR disclosure affect stock price synchronicity. The analysis showed that two types of signals play important roles. Heterogeneous information content transmits firm-specific information, and the intertemporal consistency index reflects consistent communication. Both signals reduce information asymmetry between firms and the market, thereby decreasing stock price synchronicity. Further analysis indicated that signaling heterogeneity is more significant in non-state-owned enterprises. Higher levels of social responsibility fulfillment strengthen the effects of heterogeneous content and consistency index, while higher disclosure levels may weaken the signaling effect. Robustness tests were conducted by replacing samples and applying the PSM-DID and Heckman models. From the perspective of signal characteristics, internal content analysis and external longitudinal comparison were employed. The findings provide a new perspective on the impact of CSR reports on market performance. The results also offer theoretical support for improving information efficiency in China's capital market and policy suggestions for enhancing the CSR disclosure system.