The context
Im trying to show that the reduct to the luanguage $\frak{L}$ of any model of the complete diagram $D(\frak{M})$ of an $\frak{L}$-structure $\frak{M}$ is an elementary extension of some isomorphic copy of $\frak{M}$.
recall that $D(\frak{M})=\mathrm{Th}(\frak{M}^*)$, where $\frak{M}^*$ is the expansion of $\frak{M}$ to the luanguage $\frak{L}_\mathrm{M}$(we add a constant symbol $c_m$ for each $m$ and we interpret it as $m$)
In the proof. they consider $\frak{N}'\models D(\frak{M})$, a model for the complete diagram and they claim that the map $m\mapsto c_m^{\frak{N}'}$ provides the isomorphism!!!
The Question
But for this to work we mustn't have two constant symbol with the same interpretation, right??
Do we always assume, when we say that $\frak{M}$ is an $\frak{L}-$structure, that $M$ has a cardinality greater or equal to the cardinality of the constant symbols? and that we don't interpret two different symbols as the same element?
Or does $\frak{N}'$ being a model of the complete diagram ensures somehow the injectivity of that map??