This study investigated the forming process of the sentence semantic representation. Fujiki and Chujo (2005) consider it the integration process of schemata. Their model assumed that the integration occurred when encountering syntactic head words of phrases, and that the schemata of the constituents except the head were integrated into the head schema. However, their model could explain only two types of integration, so this study extended the model including the process of comparison and alignment for schemata. To test the validity of this extended model, we compared sentences including acceptable noun phrases with sentences including unacceptable noun phrase. As the result of the measuring the reading time of those noun phrases and judgment time of the acceptability of those sentences, the semantic processes of these sentences changed depending on the load of sentence processing. We proposed that the model would change its behavior as the amount of the available resources changed