SUMMARY
Modern phonological theory is confronted with a wealth of new data from many different sources. This paper gives a summary and taxonomy of the kinds of evidence we currently have at our disposal. For each type it briefly discusses how it has been used, and gives some of the advantages and disadvantages. The main distinction made is between ‘existing’ data and ‘invented’ data, even though it is shown that such distinctions should be considered very carefully. I argue that each type of data can have a function in phonological argumentation, but also that each type of data has some very specific problems. The best evidence is converging evidence from different sources.