Likelihood function confusion over definition

#1
When we have a likelihood, say L(theta| x) would we say that L is the likelihood of (x1,x2) or (X1,X2) or neither?

Where X1,X2 are the unobserved random variables and x1,x2 the observed values.