English has such a rich, full, ever-expanding vocabulary because of the wide variety of available derivational morphemes. Like inflectional morphemes, derivational morphemes are usually bound, but some, like "able," just happen to be free. Others do not exist outside of a single morpheme, but are not considered bound morphemes because they have no meaning on their own. This is a relatively rare phenomenon, the most oft-cited example being the "cran" in "cranberry."

As children, we learn how to manipulate derivational morphemes to change the meaning of words. If you say, for example, "I snarfed the cake," snarf being a new word meaning "to eat something small very quickly" (sorry for the stretch here), then a child might, upon seeing the size of the cake, say, "But I didn't think that cake was snarfable." Without even knowing the meaning of the root morpheme "snarf," we can modify it to express the concept in new ways.

Derivational morphemes may also change the grammatical class of a word. Much like adding "able" to the verb "snarf" to form an adjective, many derivational morphemes exist that change verbs to nouns, nouns to adverbs, and so on. Here are some examples:

picture (N) + esque = picturesque (ADJ)
sing    (V) + er    = singer      (N)
quiet (ADJ) + ly    = quietly     (ADV)
vaccine (N) + ate   = vaccinate   (V)
tall  (ADJ) + ness  = tallness    (N)
migrate (V) + ory   = migratory   (ADJ)
Of course, not all derivational morphemes trigger a change in grammatical class. Many prefixes are like this, though it is not limited to prefixes.
friend  (N) + ship    = friendship  (N)
un          + do (V)  = undo        (V)
pink  (ADJ) + ish     = pinkish   (ADJ)
Additionally, there is something called the zero-form morpheme. Although it is not restricted to derivational morphemes, it is very common among noun and verb pairs. The zero-form has no phonological representation, but it changes the grammatical class of a word anyway. "Access" is both a verb and a noun. One can assume that there is an invisible derivational morpheme at the end that distinguishes one from the other. (The zero-form is really just an effort to expand the theory of derivational morphology. You cannot change the grammatical class of a morpheme without a deriviational morpheme. "Access" is both a noun and a verb, so, obviously, there is an invisible suffix or prefix working there.)

There is another sticky part to this topic: Some derivational morphemes are not always, for lack of a better term, themselves. A singer sings; a dancer dances; a butcher does not butch. -er turns the root morpheme from a verb to a noun, but it's presence does not always mean that the root is a verb. Words with derivational morphemes that are not affected by them are called monomorphemes. Using the -er example, such words are "water," butcher," "finger," etc.

Though creating new words with derivational morphemes may seem simple, it is a process governed by expansive, comprehensive rules. Take, for example, the word "nationalism." Why is it not "nationism?" -ism requires the stem morpheme to be an adjective, and though we do not consciously think this, it happens automatically. The rules governing productive derivational morphology (applying derivational morphemes to create new, legimitimate morphemes) in English are far too numerous to list here, and it is quite possible that no linguist has ever bothered to list all of them.

Creating new words is all fine and good, but not all words enter the popular vocabulary or even our own personal Lexicon. Generally, the form that most closely follows the rules for word coinage is accepted, and the rest ignored. Consider the case of "communist." It could just as easily been "communite," except that -ite is usually only applied to proper names ending in a vowel, which is why we have "communist" but "Trotskyite."