Morphemes (minimal units of meaning) are of two basic kinds: roots and affixes . While there is not an absolutely sharp dividing line between them, due to the natural, gradual historical progression from root to affix, there are various properties that typically cluster together, thus allowing us to distinguish the two types. For most morphemes, it is clear which class they belong in.
Properties of roots:
Properties of affixes:
A third type of linguistic element is a function word, which occurs in certain languages like English, which don't have much bound morphology -- that is, languages with lots of free morphemes, instead of mostly words with roots and attached bound morphemes.
In such languages, many grammatical functions are served by function words: small units that have some independence, occuring with more freedom of position than affixes (thus they are somewhat root-like), but which have grammar-like meaning rather than concrete lexical content (which makes them more affix-like). Some function words in English are the, a, he, she, it, if, although, etc.
Function words can be thought of as right in between roots and affixes. Prepositions (like English over, in, through) are sometimes classed as function words and sometimes as roots--because they are, again, intermediate. In form, they are free morphemes. In terms of function, they have (especially in their spatial meanings) more concrete lexical content than most grammatical elements, but their meaning is still rather abstract and relational. (Note that in Greek and Latin, the elements corresponding to the English prepositions are bound morphemes rather than free function words. These are the spatial prefixes such as circum-, meta-, sub-, etc.)
ROOT > PREPOSITION > FUNCTION WORD > AFFIX