You talk about this stuff in high school? Nice :)
I'm not a materials guy, so I can only speak about some general issues. Here's what I make of it, basing on what I learned in a strength of materials course.
Well, as far as "absorbing the impact force" goes, the amount of so-called strain energy absorbed depends on stiffness (e.g. Young's modulus) of material. In fact, it's inversely proportional to Young's modulus. So the stiffer the material gets (E goes up), the less energy it will absorb for a given load, because it'll also deform less. And strain energy is another name for deformation energy.
I'm talking about bulk phenomena, but it also applies to small chunks of material. Generally, the softer grains will deform a lot, while the hard ones will deform less, and the ratio of deformations will be the inverse of ratios of Young's moduli -- this is a very broad approximation of course. I.e. if hard grains have E=300GPa, and soft ones have E=100GPa, the hard ones will deform to about 30% of the soft ones.
The reality is much more complex, though. E.g. since the soft grains may undergo local yielding and even some plastic flow, the hard grains will usually deform much less than expected by ratio of Young's moduli. In fact, the grains are likely to be anisotropic themselves, so they will have more than two independent elastic constants/moduli, i.e. they may have up to 3 Young's moduli, etc.
Now, as far as "breaking" or so called failure goes, there are essentially two modes: one is due to fatigue, and one is called rupture.
In general, elastic materials can be classified as brittle or ductile, depending on how much they elongate if you pull on them (tension test) until failure occurs. Imagine a knife, mounted by its handle and its tip to a machine which will try to "stretch" it. Such machines are huge beasts, as the loads needed are very significant (tens, hundreds, thousands tons!). As you increase the force acting along the length of the knife, the knife will also stretch. For a while the stretch will be proportional to the force, and this is the elastic region of the stress (force/area) / strain (change in length/length) curve. As a certain force is reached, two things will happen: either the material will behave plastically, or it will just abruptly break.
Now, as far as knifes go, my experience tells me that the typical silverware-variety (not e.g. butcher's knives) are quite brittle and you can break them into two pieces by dropping them "properly" on a hard surface. So you have brittle failure.
As far as ductile rupture goes, a nice example would be e.g. copper wire. You pull on it, and after a very short elastic regime it will just stretch plastically, and it can stretch quite a lot before it actually break.
Fatigue is a different beast. In fatigue you have repeated load/unload cycles. Due to internal dislocations that each such load/unload cycle presents, there may be regions where dislocations have self-organized to give an area of material with lower stiffness and/or lower strength than adjacent areas. Such area will most likely become a micro-crack, whose tip will slowly move, elongating the crack, during each load/unload cycle. When the crack is long enough it doesn't need repeated cycles to keep growing, the last static load condition will just make the crack grow through full width of material.
Now as far as dislocations go, you're right in that if there was negligible amount of soft material -- only large, hard grains, the material would be very hard and brittle. OTOH, if there's enough soft grains, they will absorb most of strain energy, and thus there won't be enough deformation demanded of hard grains to even start them to dislocate. This is a very rough sketch of what might be going on, and I might even be wrong. I hope there are some material specialists lurking here who would know more.
