Named entity recognition (NER) is a critical task in natural language processing (NLP) that focuses on identifying and classifying entities such as names, locations, and organizations within text. However, for under-resourced and linguistically complex languages like Algerian dialects (Alg-D), achieving high NER performance remains challenging, particularly in bridging the gap between MSA (modern standard Arabic) and Alg-D. This research investigates whether integrating part-of-speech (POS) tagging features can improve NER performance in AD, hypothesizing that the syntactic context provided by POS tags can enhance entity boundary detection and classification. To validate this hypothesis, we evaluate the performance of several Arabic and multilingual transformer-based models on two AD datasets (NERDz and DzNER) and one MSA dataset (CLEANANERCorp). Experimental results demonstrate that incorporating POS tagging achieves a significant improvement, with the best F1 scores of 78.76% on NERDz, 84.64% on DzNER, and 88.7% on CLEANANERCorp, highlighting the potential of POS tagging to address challenges in NER for low-resource languages like Alg-D.