Writer identification plays a crucial role in forensic analysis and document examination, where texture-based descriptors like Local Phase Quantization (LPQ) and Local Binary Patterns (LBP) have demonstrated strong performance. This study investigates how encoding strategies can further enhance the discriminative power of these descriptors. We apply three advanced encoding methods—Vector of Locally Aggregated Descriptors (VLAD), Triangulation Embedding, and Bag of Words—to LBP and LPQ features extracted from handwritten documents. Experiments are conducted on four benchmark datasets: CERUG-CH (Chinese), CERUG-EN (English), IAM (English), and BFL (Portuguese), representing diverse languages and handwriting styles. The results show that LBP combined with VLAD achieves superior accuracy across all datasets, with 98.1% on CERUG-CH, 99.0% on CERUG-EN, 96.3% on IAM, and 100% on BFL.