Unless the answers to the open questions can be scored or rated in some kind of quantitative-numerical way, you will probably have to exclude them from your factor analysis and focus on the remaining questions.
As suggested, you must separate your open-ended data for separate analysis. For the quantitatively scored items, I recommend starting with a Principal Axis or Maximum Likelihood extraction followed by an oblique rotation such as Oblimin. This will allow you to treat the factors as correlated variables, which is usually the most reasonable assumption.