For the purpose of grapheme-to-phoneme conversion in text-to-speech systems, in many applications rule based systems are used. The rule format used is often taken from the area of linguistics. In these so-called re-write rules only concatenation and alternative operators are usually provided for. With these the frequently occurring exceptions in pronunciation cannot be handled elegantly and insightfully. If a complement operator were also available, this drawback would be overcome. The introduction of a complement operator, however, gives rise to some serious interpretation problems. This article describes these problems, and offers a solution to them in the form of an insightful and clear definition.