Lower and higher estimates of the number of “true analogies” between sentences contained in a large multilingual corpus