Lexical Meaning

Our examination of morphological processes has thus far concentrated on morphological structure and lexical rules. It is now time to consider the interpretation of lexical items. Most linguists agree that an entry for a lexical item must include information about the item's meaning. Indeed, it may seem reasonable to suppose that in

this regard, at least, a standard dictionary definition is close to what we have in mind. Unfortunately, there are a number of flaws in this approach; and we once again use the dictionary as a foil to help develop our own ideas about meaning.

Our first worry is that dictionary definitions are often much more detailed than speakers' meanings. Consider, for example, the case of beech and elm, two names for trees. Do you know these words? Most people (botanists and serious gardeners aside) recognize that these are English words that stand for different kinds of trees, but they cannot tell a beech from an elm. Dictionary definitions, on the other hand, can be quite specific. For example, our dictionary defines beech as "any of a number of related trees with smooth, gray bark, hard wood, dark-green leaves, and edible three-cornered nuts." Clearly, this is a poor account of what a typical speaker of English takes to be the meaning of beech.

A second concern, paradoxically, is that there also a sense in which dictionary entries represent far less than what a native speaker knows about the interpretation of a lexical item. There is a vast amount of information that speakers associate with many lexical items that is far too general to make it into any dictionary. Suppose, for example, that you tell someone that you propose to paint a house white. It is clear that you mean that the outside of the house will be painted. Similarly, if you describe a house as white, you mean a house that is white on the outside. On the other hand, if you complain of a dirty house, you mean a house that is dirty inside; whereas when you speak of an expensive house, it seems to involve the whole house as an aggregate. There is reason to think that this kind of semantic information may not be simply gleaned from experience with houses, but that it derives, in part, from the basic assumptions we associate with what we might call "container" words (that is, words for things that have insides, outsides, and capacity). White cars, white boxes, white vases, and white cups are also white on the outside. This suggests that certain properties of container words ought to be captured in our theory of the lexicon. Furthermore, it is easy to show that this type of lexical knowledge is productive: If we introduce a new container term say, a drix, you automatically know that a white drix is one that has a white exterior.

A third complication with the idea that lexical meanings are like dictionary definitions concerns the distinction between content words and function words. It just is not clear that function words have meanings in the usual sense of the word. For example, the

Do the meanings of the words beech and elm seem more elusive than those of other kinds of things? Would you have done better if we had asked you to define (exactly) robin, Texas, or impeachment? Discuss.

entry for to in our dictionary treats the to of / want to leave as "a sign of the infinitive..." and has no entry at all for the to of I gave the car to Mary. Clearly, we need to provide a more systematic account of the lexical information associated with function words.

A fourth consideration is redundancy. The dictionary treats items that are obviously related as if they were independent of one another. We are told that a baker is "one who bakes," and that a singer is "one who sings," missing the generalization that the -er suffix typically contributes the meaning 'one who Xs.' A more general manifestation of the tendency of dictionaries to treat related words as unrelated is their failure to capture other kinds of meaning-relations among lexical items. For example, English-speakers know that dogs and presidents stand for animate objects, veal and bliss are mass terms, and the meaning of leg is related to the meaning of arm in that both are inalienable body parts. We must record this type of lexical knowledge in each lexical entry.

The fifth objection pertains to the productive character of morphological processes. We have already noted the fact that many morphological

rules can be applied creatively to a large range of input forms. Compounds, in particular, can be freely formed by speakers "on the fly." Suppose we invent the compound elephant icebox to describe an invention. Although native speakers are able to interpret this item without difficulty, they cannot look up the word in their lexicon, since it has been newly created. Speakers can calculate the meaning of a compound on the basis of the meanings of its free morphemes. This meaning is compositional, and speakers can compose the interpretations of compounds in a way that is not explained by adopting static, dictionary-type entries for lexical meanings.

Finally, we must remind ourselves of a thorny methodological problem: the dictionary defines words in terms of other words, but meanings must ultimately be something other than words