Automated essay scoring (AES) of student essays aims to train an AES model by applying machine learning (ML) techniques to predict grades on essays written by students in, i.e. a primary school setting. There are many published studies of AES on English student essays, however, there are no known AES systems developed for Norwegian student essays in a primary school setting. This master's thesis explores the possibilities for AES of Norwegian primary school student essays. The KAL corpus provides a collection of Norwegian student essays from two learning plans: M87 and L97, and consists of essays paired with grades given by human examiners. The master's thesis investigates the outcomes from ML experiments for AES on the KAL corpus, implementing traditional ML (TML) models and deep learning (DL) models. The ML techniques are supervised, meaning that essay grades can be predicted based on essay input and grade output example pairs. Findings are that DL models produce unstable results for AES on the KAL corpus. A DL model may deliver classification results that can fluctuate from weak, equal, to strong when classifying the KAL corpus. TML models prove to be more stable models for an AES task on the KAL corpus, not experiencing the same ups and downs in performance as DL models. The results from this master's thesis prove that ML models potentially can be trained to classify Norwegian student essays. TML models generally achieve stronger results classifying the KAL essays, however, results indicate that if provided a larger essay corpus, DL techniques can be a promising approach to solve an AES problem for Norwegian primary school essays.