Transformation approaches for automatically constructing analysis models from textual requirements are critical to software development, as they can bring forward the use of precise formal languages from the coding phase to the requirement analysis phase in the software development life-cycle. Over the decades, numerous transformation approaches have been developed in an attempt to fully or partially automate this initial phase. This systematic review examines transformation approaches in the early stages of software development, examining 25 studies on early-stage requirements transformation documented between 2000 and 2014. The review highlights the widespread use of natural language processing techniques, with tools like the Stanford parser and WordNet being essential. Intermediate models are often used in the transformation process to bridge the gap between textual requirements and analysis models. Significant advancements have been made in early-stage requirements transformation approaches; however, several areas require attention to enhance their effectiveness and reliability. A challenge identified is the lack of robust evaluation methods, with most approaches using simple case studies and running examples for evaluation. This makes it difficult to compare and evaluate the performance these approaches. Although most approaches can generate structural models from textual requirements, many generate incomplete models with missing elements. Furthermore, requirements traceability is largely neglected, with only two approaches addressing it and lacking explicit detail on how traceability links are maintained during the transformation process. This review emphasize the need for formalized evaluation techniques and greater transparency and accessibility of approaches used in the early-stage requirements transformation.