Take a dictionary. Each word is defined using other words. Take all the words of the dictionary. The words that appear in their definitions is a subset of words of this dictionary (and not the whole dictionary because some words do not enter in the definition of any other). Now, take the words that appear in the definitions of this subset, this is likely a smaller subset. Iterate this procedure: we will likely converge to a subset of "essential" words, i.e. a list of words from which all other words can be defined. We can further reduce this list of essential words to its core by removing words that appear in a limited number of definitions, say less than 10.
I guess I am not the first to come up with this concept: are there some studies out there that calculate this list of essential words, give its size, compare the lists of different languages?