Nice and clever!
Stepping up a little bit, the input is
"xyz abc pqr", i use a pipelined function to convert row to columns :

SELECT
   COLUMN_VALUE
FROM
   TABLE(convert_string_to_table(p_string => 'xyz abc pqr', delimiter => ' '))

i get:
----

xyz
abc
pqr

always (?) displayed in this order, how the pipelined function processed the input.
I can easily add a "rownum" for this set, so i get:

SELECT
   COLUMN_VALUE word, rownum word_priority
FROM
   TABLE(convert_string_to_table(p_string => 'xyz abc pqr', delimiter => ' '))

and get:
-----

xyz 1
abc 2
pqr 3

But.. how to do a "MINUS", as the second set would not have the same rownum

SELECT
   COLUMN_VALUE word, rownum word_priority
FROM
   TABLE(convert_string_to_table(p_string => 'xyz abc pqr', delimiter => ' '))
MINUS
SELECT 'pqr' , rownum --this # is always dynamic, so it won't get eliminated from the first set .. ?
FROM dual;

Thanks,

Silly me, it's obvious:

SELECT word
        FROM
        (
            SELECT
                COLUMN_VALUE word, rownum word_priority
            FROM
                TABLE(convert_list_to_table(p_list => 'xyz abc pqr', p_delimiter => ' '))

        )
        where word not in (SELECT 'pqr' FROM dual)
        order by word_priority
        ;

output
-----

xyz
abc

Cheers!