As both variables are ordered categorical, you can use a non-parametric test for association. Most of these are available under Crosstabs. For Ordinal data, you can use Gamma, Somer's D, Kendall's Tau b & c (all are described in Siegel 'Non-parametric Statistics'). Gamma and Kendall's Tau are non-directional, but Somer's D is directional. With Somer's D you can see how good/strong the relation for the dependent variable can be predicted.

The use of these statistics assume the ordering is not violated by the use respondents make of the items. This can happen when the middle category is chosen wrongly. For instance, 'neither agree, nor disagree' might also mean 'indifferent' and this is not in the same line of reasoning as the other categories.

Peter Moorer thanks for your prompt response and highly appreciated. this will help me a lot. If I use linear regression analysis, will it be also ok in this case?

I would not use a normal regression analysis. Normality assumption will be violated; the data will be skewed. In SPSS, you can also use Ordinal Regression Analysis instead of normal regression analysis. Beside the normality issue, you will also have restriction in range issues. A normal regression analysis might have predicted values outside the admissible values (< lowest or >highest value for the category scores).

In Ordinal Regression Analysis, you can have ordinal independent variables as well, that will not be a problem.

An alternative might be datamining (regression trees), but that will depend on your sample size. A small sample might not be very usefull to detect interaction effects.