quadratic kappa and cross validation

Question

I am trying to classify text based on their quality (4 classes). The training material contains texts that have been manually scored.
I am using ten-fold cross validation in order to find the best parameters for the classifier (svm/rbf and pol). I am going to use the quadratic kappa measurement to measure the classification results(agreement between manual scores and class results).One approach is to calculate the corresponding quadratic kappa for each cross validation test step. At the end of 10-fold cv, there are 10 quadratic kappa values. The final result for this cv is calculated by averaging the quadratic kappa values. I wonder if this approach to calculate an average quadratic kappa using a 10-fold cv procedure is valid?