This study proposes a system that automatically generates assessment reports scripted by natural language from the drawing evaluation index of the landscape montage technique. For this system, we configured an evaluation index for assessing drawings that use the landscape montage technique, built up a knowledge base for indicators, analyzed the client’s emotions through natural language processing, and generated spontaneous sentences. The contribution of this study is as follows. First, a great quantity of data was cumulated by using the system. Second, objectivity about the drawing assessment was resolved by building a knowledge based on the recent research paper and book. Third, as the system automatically generates assessment reports for the drawing assessment, therapist’s workload can be dramatically reduced. In this study, an assessment report automatically generated by the system is compared to an expert’s report. The results of the comparison show that the system-generated assessment report demonstrates a higher number of objectivity on the suggests that the system can accumulate a great number of information and provide credibility for drawing evaluation index of the landscape montage technique.