"When linguists describe the structure of a sentence, they typically use a rather restricted set of concepts, such as ‘accusative’, ‘subject’, ‘passive’, ‘verb phrase’, ‘goal’, and ‘adjunct’. These are quite abstract categories, that therefore have the advantage of being generally applicable. Similarly, teaching a system of linguistic analysis to students and teaching them how to apply it, is also typically restricted to abstract categories. The implicit assumption is that the structure of an object of linguistic analysis can be characterized both exhaustively and insightfully in terms of (combinations of) properties that this object as a whole as well as its parts share with many other elements in the language; notions with a limited range of applicability (e.g., going from still relatively abstract to quite concrete: ‘indirect object’, ‘benefactive dative’, ‘addressee (of a verb of communication)’, ‘promisee’) are considered less fundamental and derivative, so not really required for an adequate characterization of the grammatical structure of utterances in the language. This line of thinking follows the well-known – and in principle quite legitimate – idea that a scientific explanation of the properties of concrete, complex entities should be based on insights about the properties of the relatively simple (i.e. abstract) component parts of the complex entities, and the way they are put together"