Narrowing targeting and effect parameters with additional adjectives and doing likewise with effects and adverbs is simply, though most modifiers should get expensive fast if you stack them up. Making them cost multipliers would account for this.
--but there comes a time when you want to specify a second target, in a slightly different way. Glue Goblin Foot To Floor, is different from Glue Goblin Foot Floor. You're liable to glue the whole floor, the whole goblin and your own feet, without some grammar to codify the relationships between the words.
However, grammar means there will be some combinations that simply don't work. Glue Goblin To Floor Foot? Green Goblin Hair? Putrified Wiggling Vermin Heads? And Self Lack Spoon? You broke the universe, tearing a hole that causes the source of the paradox to be sucked out. Have fun with the formless horrors, out there in the twisting nether.
We can account for some of these. Someone suggested dropping subtargets that make no sense. However, sometimes you're going to end up in a situation where none of the targets is viable. Kill Wall, for instance. Do non-targeted spells simply fail to cast, waste energy but do nothing, or strike the first viable target(usually the caster)?
A lot of people hate mana, either because it's quantifiable magic("Twenty-eight thousand thaums!", "It's over Nine thousand!" "She is a class five psychic" etc.) which sucks some of the mystery out of mystics, or because they saw MP in Final Fantasy, which clearly makes it a product of Game Design Satan, as a part of his evil conspiracy to make all games identical and boring (OMG! They're after my brain! Where's the tinfoil?!).
I don't want to see the idea go away, because of all the cool ways it could interact with spheres, but I'd like to see what can be done to keep it palatable, to those who like their magic arcane and ookey.