Affixation in the item-and-process and item-and-arrangement models
Let’s look at a simple English example of a lexical function within the item-and-process model, which we happen to be most comfortable with. The following function creates agent nouns from verbs:
(14) X]V er]N
Examples: think]V er]N , runn]V er]N , fli]V er]N , hunt]V er]N
We generally think of lexeme-formation functions as having a phonological, a syntactic, and a semantic component. Phonologically, the function in (14) takes a pre-existing string of segments and adds the suffix /ɚ/. Syntactically, it produces a noun from a verb. Semantically, it produces an agent of the verb.
In the function in (14), the phonology, syntax, and semantics are additive. When we derive an agent noun from a verb via the suffix /ɚ/, we add phonological information in the form of an additional segment, syntactic information in terms of the lexical category noun, and semantic information (the fact that the new noun represents an agent). Furthermore, the old information is preserved, not lost: worker includes both the form and the meaning of the verb work, to which the suffix /ɚ/ has been added, as well as the fact that work is a verb, not a member of some other lexical category. The phonology, syntax, and semantics of most derivational functions are additive in this sense.
Additive functions like this one are easily recast in the item-and-arrangement model. To express (14) in item-and-arrangement, we also need to break agent nouns into two parts. Then we put them back together. One way we can show this is through a tree structure. There will be more on morphological trees in Derivation and the Lexicon:
This tree illustrates how the affix -er attaches to the stem work to form the agent noun worker.
