As an interdisciplinary field, computational linguistics has its sources in several areas of science, each with its own goals, methods, and historical background. Thereby, it has remained unclear which components fit together and which do not. This suggests three possible approaches to designing a computational linguistics textbook. The first approach proceeds from one's own school of thought, usually determined of study, rather than by a well-informed, delib by chance, such as one's initial place erate choice. The goal is to extend the inherited theoretical framework or method to as many aspects of language analysis as possible. As a consequence, the issue of com pat ibility with other approaches in the field need not be addressed and one's assumptions are questioned at best in connection with 'puzzling problems. ' The second approach takes the viewpoint of an objective observer and aims to survey the field as completely as possible. However, the large number of different schools, methods, and tasks necessitates a subjective selection. Furthermore, the pre sumed neutrality provides no incentive to investigate the compatibility between the elements selected. The third approach aims at solving a comprehensive functional task, with the differ To arrive at the desired solution, suitability ent approaches being ordered relative to it. and compatibility of the different elements adopted must be investigated with regard to the task at hand.