This is why you should always provide more context and the actual workflow from the beginning. The vocabulary you used implied one thing, but your workflow says something else entirely.
From what I can tell, you tried used the Table to R node to concatenate the cluster names. As far as KNIME is concerned, the result is a string, not a list of strings.
If you compare the icons for string variables
and list variables
you’ll see that they’re completely different. This should give you a clue that this node is not accomplishing what you think it is accomplishing.
I don’t understand the role of the Nominal Row Filter node.
I agree with @ipazin that a different approach is warranted.
Without knowing more about what you’re trying to accomplish, the approach I would personally take, is to use a Duplicate Row Filter node to remove duplicate cluster names, then a Transpose node to convert the Cluster column to a row, then a Create Collection Column node to create the list of cluster names. At that point the list can be converted to a flow variable using a Table Row to Variable node and used in whatever widget you like: