Steganography has introduced a new layer of data protection, offering a higher level of confidentiality and robustness compared to traditional encryption methods. While text-based hiding, as in images, is a method of hiding information, text-based approaches remain weak and easily decodable. In this research, a new steganography method is presented within Arabic text that relies on language features, specifically the diacritic marks. The proposed method dynamically alters certain diacritics according to a specific pattern without the need for prior agreement or a map between the sending and receiving parties. Through an investigation into the proposed method, the effectiveness and superiority over other steganography methods is established, particularly in terms of non-detection, performance, and stability, which are crucial metrics for evaluating steganography techniques. Furthermore, the presented case study demonstrates the ease of use of the proposed method to protect highly confidential data when transmitted through social media applications between individuals.