Key:

Names:
a: Bender
b: Fry
c: Leila
d: Sasquatch
e: Morbo

Predicates:
Fxy: x robs y.
Gx: x drinks Slurm.
Hx: x knows what Slurm really is.
Ixy: x disintegrates y
Jx: x eats Bachelor Chow.
Kx: x is a head in a jar.
Lx: x is a mutant.
Mx: x sucks.

---

1. Bender robs Fry.
   Fab
2. Fry drinks Slurm only if he doesn't know what Slurm really is.
   (Gb → ¬Hb)
3. Morbo disintegrates Fry only if Leila robs Sasquatch.
   (Ieb → Fcd)
4. All who drink Slurm will rob Fry.
   ∀x(Gx → Fxb)
5. Someone robs Fry, and it isn't Morbo.
   ∃x(Fxb ^ ~Feb)
6. All who eat Bachelor Chow drink Slurm.
   ∀x(Jx → Gx)
7. Some who drink Slurm do not eat Bachelor Chow.
   ∃x(Gx ^ ¬Jx)
8. No one who either eats Bachelor Chow or drinks Slurm is a head in a jar.
   ¬∃x((Jx v Gx) ^ Kx), or ∀x((Jx v Gx) → ~Kx)
9. Morbo is not a head in a jar.
   ¬Ke
10. Some head in a jar did not rob Fry.
    ∃x(Kx ^ ¬Fxb)
11. Some head in a jar did rob Fry.
    ∃x(Kx ^ Fxb)
12. Mutants suck.
    ∀x(Lx → Mx)

---

Notes:

We could have considered "drinks" a relation of two things, but the problem with that is we'd need special tools to indicate a single particular drink of Slurm, because "Slurm" is a mass noun; this would be difficult, so we took the easy route. The same is true of knowing what Slurm really is and of eating Bachelor chow: in these cases its easier not to treat knowing and eating as arity two predicates since the sentences talk about knowing and eating kinds of things.

I assumed "Sasquatch" was the name of a particular being, rather than a kind of being.

A better translation of 5 would be: ∃x(Fxa ^ ¬x=e), if we had introduced the "=" into our language. Given that we haven't, the translation above works pretty well.