Unlike the problems of affix order and bracketing paradoxes, which have received a great deal of attention from morphologists, the problem we will take, has received little attention, but it is no less interesting. we briefly touched upon the issue of affixal polysemy. To refresh your memory, affixal polysemy is the tendency for affixes to have several closely related meanings. For example, we pointed out that it is a curious fact that the affix that is used for making agent nouns in languages is frequently also used for making instrument nouns. As I pointed out, this is the case in English, and Dutch – not surprising, as they are closely related languages – but also in Yoruba, a Niger-Congo language, Turkish (Lewis 1967: 225), Kannada (Sridhar 1990: 273), and many other languages. It cannot be an accident that agent nouns and instrument nouns are so often created by the same affixes. So the theoretical question that arises is why this should be. To answer this question, let’s again look a bit more closely at English.

It appears that in English the suffix -er forms not only agent and instrument nouns, but also nouns that denote the experience of an action, or even the patient or theme of an action. Even more curious, the suffix -ant covers the same range of meanings. The theoretical question we must raise in light of these data is why -er and -ant nouns cover just this range of meanings and not some others.

The first step in explaining this affixal polysemy is to figure out just what -er and -ant mean. One suggestion that has been made (Lieber 2004) is that these affixes don’t actually mean ‘agent’ or ‘instrument’, but something much more abstract – something like ‘concrete noun concerned with a process or event’. In this, they are semantically analogous to simple nouns like poet or awl that denote people or things defined by what they do. One piece of evidence for this claim is that -er can attach to nouns as well as to verbs, and when it does, it always adds an active or eventive element of meaning to its noun base. So, for example, a villager is someone who lives in a village, and a freighter is something that carries freight. No verb is necessary for the active part of the meaning – this comes directly from the suffix.

The second part of the answer to our question has to do with under-standing the argument structures of verbs. You’ll recall that the argument structure of a verb consists of those arguments that are semantically necessary to the verb. What is most interesting for our purposes is that in English there is a range of semantic roles that the subject of a verb can play:

In (31a), the subject of the sentence is the agent or ‘doer’ of the action. In (31b), the subject is called an instrument rather than an agent, because it is an inanimate noun that does something. In (31c), we call the semantic role that the subject plays the experiencer rather than agent, because one can hear something without doing anything at all – to be an agent, one must act intentionally. Finally, the subject in (31d) is not the agent, but the theme, in other words, the noun that undergoes or is moved by the action.

What is interesting is that -er and -ant nouns denote exactly the semantic role conveyed by the subject of their base verb. So an eater is an agent, just as the subject of the verb eat is an agent, an opener is an instrument, just as the subject of open is an instrument, a hearer is an experiencer, just as the subject of hear is an experiencer, and a sinker is a theme, just as the subject of sink (intransitive) is a theme. The reason that -er and -ant display exactly the range of meanings that they do is that they are linked to the subject argument of the verb. They can mean whatever the subject of a verb can mean.