A process model to explain bugs produced by novices early in a programming course are presented. (The term "bug" refers to an error in a person's behavior, particularly an error in a computer program they have written.) The model is motivated by interviews of novice programmers solving simple programming problems. The key idea is that many programming bugs can can be explained as novices inappropriately using their knowledge of step-by-step procedural specifications in natural language. (The term "natural language" refers to the language in which step-by-step procedures are written.) Programming bugs are viewed as patches generated in response to an impasse reached by the novice while developing a program; such patching strategies are called bug generators. Several bug generators are used to describe how natural language pre-programming knowledge is used by novices to create patches. Other kinds of bug generators are also discussed. A representation both for novice natural language pre-programming knowledge and novice fragmentary programming knowledge is described. Using these representations and the bug generators, the model is evaluated by analyzing four interviews with novice programmers. (JN)