I am not quite sure what you mean by algebraic. But if you mean by just using the definition in terms of recognizing morphisms in such a manner that the proof would easily generalize to other monoids, then this would not be possible.
The reason is that recognizable languages in general are not closed under Kleene star. Or more specifically, in general we have rational sets that are not recognizable. See this thread for lots of examples. One particular simple example for Kleene star would be $(1,1)^* \subseteq \mathbb N^2$. This set is not recognizable in $\mathbb N^2$.
But, interestingly, the recognizable sets are closed under concatenation. You can proof this with Schützenberger products. A definition of this product could be found in S. Eilenberg, Automata, Languages and Machines, Volume B.