This thesis explores the role and potential of computational tools in the analysis of an existing corpus of work and synthesis of new designs. The research would like to demonstrate that, the basic grammar rules underlying the composition can be described by analyzing a set of similar designs, and new designs can be derived based on the extracted rules. Examples of Anatolian madrasahs from Anatolian region of Turkey have been chosen as a research corpus. The body of research is limited to Anatolian madrasahs that were built in the period of XII and XIII centuries having morphological similarities. After an initial evaluation of the material gathered from Kuran (1969) and Sozen (1970, 1972), in the first step, common features in plan composition are described within a research body. A classification for the plan types of the madrasahs is established. The location of main components like court, iwan(s), and other rooms are a major factor at the classification stage. The next step is the introduction of a shape grammar system for generating the plan layouts of Anatolian madrasahs through a number of rules by using the main plan components and their spatial relations. In the final step, a simple interpreter is developed by using the programming language AutoLisp for the representation of the shape grammar system for Anatolian madrasahs. The shape grammar system is realized in computer-aided design (CAD) environment to present an automated mechanism for generating different designs of Anatolian madrasahs through these rules. Such computational tools provide easy and flexible manipulation of objects so that many compositions can be created.