Using the doze procedure, the present study examines two kinds of contextual effects; one is phrase priming within a sentence (a garden-path and a non-garden-path sentence), and the other is the preceding sentence context (positive, negative and neutral). First the garden-path sentences produced longer response times. Secondly, there was only facilitation by the positive preceding; context and no inhibiton by the negative context. Two possible explanations were presented. Finally, there was no interaction between the two kinds of context effects, suggesting that contextual effects are additive and that the two kinds of context affect sentence processing independently.