Due to the rapid growth of the Internet and content development services, digital text has become the most extensively used type of media in digital communication. However, digital text is at risk of being illegally tampered with, copied or redistributed, which could lead to many security threats related to privacy and ownership protection. Over the last few years, digital watermarking has been used to authenticate and protect the ownership of digital media such as images, audio, and videos. Compared to other digital media, digital text faces greater challenges in regard to watermarking due to its low capacity for information hiding and its high sensitivity to modification. While limited research has been conducted on text watermarking, the preliminary results indicate that the available schemes do not achieve an acceptable trade-off between imperceptibility, robustness, capacity, and security. In this paper, we propose a novel text watermarking method called Bloom Filter and Text Similarity Watermarking (BFTSW). A Bloom filter is employed in our proposed BFTSW method to reduce the size of the watermark that is generated based on a vector space model (VSM) representation of a text document. To verify the watermark, the text similarity between the VSM representation of the original text (which is recovered from the Bloom filter) and the VSM representation of a suspect text is computed and compared against a predefined threshold value to determine whether the watermark is present in the suspect text. Experimental results obtained with a prototype implementation show that the proposed BFTSW method is effective in terms of its robustness against malicious attacks, information capacity, and text quality preservation.