Magic-users, Vancian casting, spell memorization effects

Over at Tenkar's Tavern, Erik makes some good points against the 5e design of mixing Vancian magic with 4e at-will spells for magic-users. Having played a 4e wizard for a few months, I came to have a love-hate relationship with at-will and encounter spells. Being able to something 'magical' each turn was nice, for a while at least. Soon, however, it did not become much different than me saying I attack again, except my attacks could effect multiple individuals and give some minor debuff. In other words, the at-will spells lost their 'magic'.

The great thing about Vancian magic is the 'magical' factor, as excerpted from Turjan of Miir by Al at Beyond the Black Gate. In the comments section of the post at Tenkar's Tavern linked above, David mentions minor at-will effects from spells memorized. This reminded me of one of Philotomy's musings regarding spell special effects, which while very flavorful, is not very useful. While some may disagree that magic-users should always have useful magic available, there is something nice about being able to perform magic more than once a day for a first-level magic-user. So, I decided to jot down some of my ideas for spell memorization effects, which seems to have done already by a couple of others: James M. here and here and trollsmyth.

While there are many great ideas in those posts and the comments, here's my take. In accordance with the excerpt from Turjan of Miir linked above, spells are entities of their own with their own will and agenda. The spells want to be released, and will do what they can to be released. Thus, for a magic-user to use a special effect for a memorized spell, he must try to release just a portion of the spell but resist releasing the entire spell. Since first-level spells are the least 'cogent' spells, they take the least amount of mental energy to 'hold onto'. So first-level spells are the only spells that will yield special effects. Second-level spells and higher will simply be too 'cogent' to release in a partial fashion. Of course, the spell will want to be released at the earliest opportunity and would not be happy with being only partially released. So each time a magic-user attempts to partially release the spell, there is a chance the spell would be fully released against the magic-users will. Thus, the spell caster must decide if partially releasing the spell is a worthwhile risk. I haven't figured out if this should be a simple 1- or 2-in-6, an INT check, or something that scales with level, ideas are appreciated.

Here is a list of the spells from Moldvay's edit of Dungeon and Dragons Basic and my list of partial effects (subject to change, of course, ideas appreciated):

Charm Person: The affected creature (save vs. spells allowed) is dumbstruck for one round and cannot act or move.
Detect Magic: Has a 2-in-6 chance of instantly discovering magic in range.
Floating Disc: As the full spell, but can only carry 50 pounds.
Hold Portal: Will hold shut any portal for one round.
Light: Gives off the equivalent of candlelight from an item held by the caster.
Magic Missile: magic dart - attack roll for 1d4 damage (3 darts at 6th level?)
Protection from Evil: As the full spell, but only lasts one round.
Read Languages: Has a 2-in-6 chance of reading a short passage of any unknown language.
Read Magic: Has a 2-in-6 chance of reading a short passage of magical words or runes.
Shield: As the full spell, but only lasts one round.
Sleep: The affected creatures bodies fall asleep (go numb), -1 to attack rolls, AC and saves for one round.
Ventriloquism: As the full spell, but only lasts one round.

Note: tavisallison at the Mule Abides has a wonderful post regarding Vancian spellcasting that does not address spell memorization effects, but is a great resource for thinking about Vancian spells.