More on the semantics of affixation: English agent nouns in -er
We next give you an example of another derivation that works in the same way as zero-derived verbs but is a little simpler. Marchand gives examples of English agent nouns in -er. He points out that they fall into four basic categories, which can be broken down even further into two separate sets. The four basic categories are listed below:
These nouns can be further divided into their habitual and non-habitual uses. If we say:
(16) He is a gambler
we usually mean that he gambles regularly. You can also use agent nouns non-habitually:
(17) All ticket-holders may enter
If you are trying to get the meaning of agent nouns in English, you have to say that they fall into the four categories listed in (15), multiplied by the two categories habitual vs. non-habitual. We need to do this because even a word that is normally understood to be habitual, like blotter, can be used in a non-habitual sense. If you use something that’s not a blotter as a blotter, then it must be a non-habitual blotter, because it is only being used as such on this particular occasion.
There are two possible analyses of agent nouns, both of which are reasonable. We won’t try to choose between them here. One is the strategy that we used above with zero-derived verbs, to assign the derivation a sparse semantic rule. We can follow Marchand in saying that an agent noun is ‘someone or something connected with what the base denotes’, or alternatively, ‘somebody or something whose function or characteristic is to perform a particular act’. For now we assume the latter. It permits the categories person, animal, material object, and immaterial object, as well as a habitual or non-habitual interpretation.
The other method which linguists might use to account for possible meanings of X-er agent nouns involves prototypes, also called archetypes. The idea is that not all members of a given category are equal. You can have prototypical, or typical, members of a category, as well as more marginal members. In this case, it is probably reasonable to say that the prototypical agentive is a person who habitually performs a particular type of action. So the prototypical agentive is a word like baker, dancer, gambler, or driver, in the habitual sense. These reflect the core meaning of this particular formation. Other forms, like retriever, blotter, or clincher, involve relaxation of the core meaning. We use them to distinguish one specimen from other members of its class. A pointer is a kind of dog that has the characteristic of pointing; specifically, it has been bred to stand still and point out the prey to the hunter. A retriever is a kind of dog bred to retrieve the prey once it has been killed.
Even within a specific class like that of retrievers, you can get central and marginal exemplars. Most people don’t know that poodles (the big ones, anyway) are retrievers. The fact that we can make a sentence like “You can use a poodle as a retriever,” tells you that retrieving is not a salient, central characteristic of poodles – they are best known for being fuzzy.
When we claim that persons are the prototypical members of the category of agentive nouns, we can also argue that some other members of the category – material objects like screwdrivers, for example – aren’t agents at all. Instead, they are instruments, because they don’t have will. Of course, if there were a special profession for people who drive in screws, then one might say that such a person was a screwdriver, and then screwdrivers could be agents. In short, the second method of analyzing agentives is to establish a central case, the prototype, and to work out from that to get the others. When it comes to denominal verbs, we do not have the option of applying this analysis because there is no central case, no prototype.
