KPR preferences are given by
$$ U(c, l) = \frac{\left(cv(l)\right)^{1-\sigma}-1}{1-\sigma}$$
with concave increasing $v$ and $c$, $l$ denoting consumption and leisure. In the limiting case of $\sigma\to 1$, we receive the standard additively-separable preferences
$$ U(c,l) = \log c + v(l)$$
In the latter case, if I want to scale the relevance of leisure in total preferences, I can rewrite the preferences as $\log c + Av(l)$, and use $A$ for this matter.
How do I do that - scale leisure - in the general case?
If I rewrite the nominator as $(c A v(l))^{1-\sigma}$, it is not clear to me whether $A$ is scaling $c$ or $l$ (probably neither).