Comparing Data from Value-Based Scales with Frequency-Based Scales

I was recently asked to correlate scores between two surveys that supposedly measure similar constructs. One has a value-based scale (very like me, somewhat like me, neither like or unlike me, somewhat Unlike me, very unlike me). The other has a frequency-based scale (never, rarely, sometimes, often, consistency, don't know).

Can anyone comments on potential problems/solutions/suggestions?