Ontologies are widely recognized as a crucial framework for managing knowledge, enabling sharing, reutilization, and automatic interpretation. They form part of most artificial intelligence applications, including intelligent information retrieval, knowledge management, and contextual organization. However, the exponential growth of textual data across domains has turned ontology acquisition and enrichment into a labor-intensive, time-consuming, and costly process. Consequently, the need for automated approaches, referred to as ontology learning is on the rise. New trends in deep learning have significantly contributed to this field, since they can extract concepts from a large corpus and detect semantic relationships across different datasets. This paper reviews the ontology learning process, discussing key challenges, and methodologies ranging from traditional linguistic-based techniques to recent deep learning models. We further highlight emerging trends and future directions, identifying areas that require further exploration to advance ontology learning.