In native morphology, the distinction between words - lexical items that are potentially free and affixes that are obligatorily bound, is relatively easy. Not so in neo-classical word formation (Bauer 2013: 18). In words like auto.mobiel, neuro.logie and bio.scoop, we have morphs that are bound. The morphs neuro- and -(o)logie do not occur independently, and they do recur in a number of words, as affixes do. Similar to affixes, some of them typically occur in initial position (e.g. neuro-) and others in final position (e.g. -(o)logie). Still, most morphologists would hesitate to call them affixes. For one thing, if we were to term them affixes, we would be left with the possibility of a word that consists entirely of affixes (for example, neurology), which goes against the definition of affix given above. For another, there are some items in this category which can occur either initially or finally (e.g. derm in dermatitisinflammation of the skin or endodermone of the three primary germ cell layers in the very early embryo), which is never the case for affixes in English. Following Bauer et al., we will make a distinction between obligatorily bound roots (or bound roots for short) and affixes. Obligatorily bound roots can serve as bases for affixes or other obligatorily bound roots.

The distinction between obligatorily bound roots and affixes is usually based on the type of semantic information that the morph carries: bound roots are generally said to have more substantial lexical content than affixes or, as (Bauer 1998: 407) terms it, a higher degree of lexical density. In practice, it is not always easy to determine which morphs have sufficient lexical content to be considered bound roots, and which fall below the threshold. Most morphologists would agree that nomin-, neuro-, and -(i)cide are bound roots, and that in-, pre-, -ize, and -ness are affixes. But items like mini-, mega-, and super- seem to live in the grey area between affixes and bound roots. They convey notions of size (and also evaluation) that are often in other languages conveyed by affixes. Inevitably the amount of lexical content is a gradient matter, and different morphologists might draw the dividing line between bound roots and affixes in a different place.

Additional information on the subject can be found in Van den Toorn (1987),Bauer (1998), Beelen (2004), Meesters (2004), Lüdeling (2006), Smessaert (2013: 78ff), Carstens (2014: 183ff) and Hamans (2014). See also this section on strata in the lexicon and on non-native derivation.