<p>Recognizing multilingual handwritten text is challenging because of the inherent variability of handwriting styles and the need to learn new character sets over time. Traditional learning methods struggle with this, potentially retraining on ineffective data and becoming less accurate as new classes are added. This can lead to error propagation, further hindering performance. To overcome these issues, an incremental learning strategy is suggested for integration into a deep learning model. However, this process may be time-consuming and inaccurate. The proposed principal sample-based incremental learning approach for multilingual handwritten text recognition addresses these challenges by selectively learning new classes incrementally using only a limited number of strategically chosen principal samples. This novel selection process utilizes the softmax response of the primed network to learn new, unknown classes incrementally. In addition, this principal sample-based approach allows the model to adapt to new languages and character sets without full retraining, reducing computational costs and retaining high accuracy. Compared to traditional methods, this approach significantly reduces the amount of data needed for training, leading to faster learning. The effectiveness of this principal sample-based incremental learning approach is demonstrated through evaluation on various multilingual handwritten datasets. The achieved accuracy surpasses that of state-of-the-art methods, highlighting its potential for multilingual handwritten text recognition.</p>