With the increasing digitization of education and large-scale standardized testing, the use of natural language processing (NLP) for the automatic grading of assessments has surged, leading to the widespread availability and adoption of commercial solutions. Automatic grading is not limited to multiple-choice questions, but also includes essays and short answers, highlighting its potential to automate rapid assessment and feedback in higher education. This study explores various NLP techniques used in automatic grading, investigates research outcomes, identifies recommended NLP approaches, and examines future directions for automated assessment systems. A systematic literature review following PRISMA guidelines was conducted using Scopus data, identifying 32 relevant studies from a pool of 279 peer-reviewed articles. Findings indicate that automatic grading techniques vary based on assessment type and available data. In summary, this study provides an updated overview of NLP applications in assessments, enhancing the understanding of current research and suggesting directions for future research.