Verbal communication involves more than the interchange of what we know of the world. The information expressed by natural language sentences is almost always qualified. When we communicate, we constantly provide clues about the origin, certainty and level of precision of what we assert. There is an emerging recognition that this is a fundamental property of language. Utterances are inherently multidimensional: Apart from providing information, they impart information about the provided information. Predication is language's most essential vehicle for communication and it comes with one of the most common sources of qualification: gradability. Predicates like "tall" are gradable in the sense that they are not simply true or false of an individual, but can be said to hold to a degree. Apart from being plain tall, one can be "very tall", "extremely tall", etc. This is usually thought to be an exceptional property, associated with a specific class of adjectives and a small set of specialised degree operators. However, this project is motivated by the idea that gradability is a general feature of predication. As a result, the proposal is to study gradability from the broadest possible perspective by investigating all forms of grading, including cases that lack the use of specialised degree vocabulary, as e.g. (1). (1) Jasper is an unbelievable nerd. The broad scope of gradability necessitates a move away from a purely semantic view on degree modification. In fact, this project takes it that degree is best studied as a phenomenon with both a semantic and a pragmatic base and that, as such, it should be connected to the question of how and why people communicate non-absolute information. As a result the project provides the opportunity to connect semantic theories of predication to theories of the use of vague and qualified language.