section banner
5.1  Attributes and attribution

5.1 Attributes and attribution

Attributes of things

Apples come in various colors, sizes, and flavors. If we want to distinguish two apples from one another, we need to be able to talk about these dimensions. One possibility is more nouns, one for each subcategory of apple. What is the advantage of doing this instead of using the noun we already have, apple, along with words like red, big, and sour?

Let's review what our Lexies have so far. They have a set of nouns that allow them to refer to important individuals and to members of categories of things in their world. And each of these nouns is represented in their mental lexicons as an association between a form and a meaning. Specifically, within the class of common nouns, the most important type of word we've seen so far, each word form is associated with, or designates, a category of things. Note that we will need to distinguish the notion of designation from the notion of reference. A Speaker produces an expression like the apple or that apple to refer to a particular apple in the world or in the imagination of the Speaker or Hearer, But within this expression the noun apple itself cannot be said to refer to a particular apple; rather it designates a whole category, the category apple, which includes many possible individual apples. It is only in combination with a word like the or that that it is used for reference.

Let's place some new communicative demands on the Lexies and see what they come up with. The common nouns that they have are very powerful because they allow them to refer to a potentially infinite set of things in the world, that is, all of the members of the categories that are the meanings of the nouns. But what if they need to distinguish one member of a category from another, one apple from another, one tree from another, one tiger from another? This kind of need comes up in two kinds of situations. If you want to announce to the rest of your tribe that you've found a tree, it may be important for them to know that it's tall (perhaps because tall trees are a useful source of wood). That is, in one kind of referring situation, the Speaker introduces a new referent to the Hearer. But say the others already know about a group of trees and you want them to focus on the one that stands out as tallest. That is, in another kind of referring situation, the Speaker calls the Hearer's attention to a referent that they are already aware of.

Nouns by themselves are limited in what things they allow us to distinguish.

The problem is that the nouns you have don't allow you to distinguish the members within each of the categories. One way to deal with this problem would be more semantic categories and more nouns. So the Lexies could have different nouns for apples of different colors or flavors, for trees of different heights or widths, for tigers of different sizes or ferocity. Let's look again at the categories that nouns designate and see if this makes sense.

Recall that each of the things in the world can be thought of as having a value on each of a set of dimensions such as size, color, shape, taste, and consistency. Each of the categories that we have common nouns for represents a strong tendency for particular sets of values on different dimensions to co-occur. An apple isn't just an object of a particular shape; it has a characteristic range of sizes, tastes, consistencies, and locations. In other words, the category apple is a whole cluster of co-occurring features, and it useful to know that something belongs to the category because you can predict a lot about it. Even if you don't know what it tastes like or what it looks like inside the peel, you can predict these things.

But even though apples tend to have particular sizes, they do vary in size. What can you say given that something is a relatively small apple? If being small doesn't imply having a particular taste or consistency or size, then the category small_apple wouldn't be useful in the same way the category apple is. If being unusually ferocious for a tiger doesn't tell us much of anything about its size, color, or location that we don't already know, then having a category ferocious_tiger won't be worth much.

It's not that the size of an apple and the ferocity of a tiger don't matter for the Lexies; they very likely do. A big apple means more to eat. A ferocious tiger means something to stay away from. It's that size for apples may work roughly the same way for raspberries and rocks and that ferocity for tigers may work roughly the same way for bears and crocodiles.

The point is that people are not as likely to form coherent categories for things like small apples, ferocious tigers, and tall trees as they are for things like apples, tigers, and trees. And because they don't have coherent categories like small_apple, they are not likely to have nouns for such things. In any case, having a different noun for all the distinguishable combinations of size, color, sourness, and texture in apples would greatly increase the number of nouns that the Lexies would have to learn and remember.

A major advance: referring with more than one word

You probably can tell where we're headed. There is an alternative to having separate nouns for large and small apples, for large and small raspberries, and for large and small rocks. We can divide the work of referring to something like a small apple into two parts, one part dedicated to the category, apple, the other part dedicated to the dimension along which members of the category vary, size in this case. That is, we can refer to an apple with two words: small apple.

So now we are seeing each of the members of a category as having particular properties that I'll call attributes. Each attribute is really a value on some dimension such as size, color, or consistency. But seeing things this way is not just a convenient way of talking about language. The fact that people can refer to things with expressions like small, sour apple means that people have the capacity, when they see or think about an apple, to separate out its attributes from its category. A small, sour apple isn't just an undifferentiated object for us. It's apple, it's small, and it's sour. Being able to keep these aspects of an separate, at least for the purposes of talking about the object, is an impressive ability, one that is so intimately tied up with the language that it's hard to say which came (or comes in each developing child) first.


If you hear somebody say, please hand me the bleg apple, what can you infer about the function of the unfamiliar word bleg in the sentence? What is it that tells you this?

An expression like red apple consisting of one or more words that make up a unit in the sense that they designate a unified concept is called a phrase. I'll have a lot more to say about phrases and how you can know whether a group of words is a phrase in the next chapter. In this chapter we'll only be concerned with phrases consisting of two words, one of them a common noun such as apple that designates a category of things. But what about the other word in a phrase such as red apple?

Form and meaning (again): syntax and semantics

With such a phrase we can speak of two relations, the relation between the words in the phrase and the relation between the concepts that the words designate. The first kind of relation is syntactic; it is a relation between linguistic forms. The second kind of relation is semantic; it is a relation between concepts designated by linguistic forms. Let's start with the semantic relation. One of these is a category of objects, apple. The other is an attribute that characterizes some members of this category, red, one possible value (more precisely, range of values) on a conceptual dimension, color. The syntactic relation between the words involves at least two components. First, it has an order: the word designating the attribute comes before the word designating the conceptual category. Second, the words belong to their own categories. The word designating the conceptual category, apple in the example, is one of the set of common nouns. The word designating the attribute, red in the example, is one of a set of words called adjectives.

I will refer to both the semantic and syntactic relation in this case as attribution, a subtype of the more general relation called modification. In other words, we can say that the adjective has an attributive function in the phrase and that the adjective modifies the noun. And we can say that the phrase attributes redness to the apple that is being referred to. Recall again that the main theme of this book is the intimate association between form on the one hand and meaning and function on the other. In fact, most of the rest of the book will be about the association between syntactic relations or patterns and semantic relations or patterns. Attribution is our first example. The close association between the syntactic attribution pattern and the semantic attribution pattern allows a Hearer to infer something about the meaning of a novel word, for example, bleg, in the bleg apple. Bleg appears to be an adjective, designating some sort of apple attribute.

Back to our Lexies. Because they realize that coming up with common nouns for all of the things that need to be distinguished is unreasonable, they decide instead to come up with a set of adjectives to designate attributes of things. These include words similar to the English words red, small, shallow, wet, dark, smooth, sour, young, pregnant, angry, kind, and dead. Notice how greatly this extends the capacity of their language. With 50 nouns and 10 adjectives, they can produce 500 different combinations of adjective and noun. Of course some of these are not likely to be produced or to make much sense if they are produced: shallow apple, sour cat, pregnant river. But many of the adjectives will apply to large subsets of nouns: red apple, red raspberry, red stone, red skin; wet apple, wet stone, wet tree, wet skin.

But the increased power goes beyond just the number of combinations that are possible. Speakers and Hearers who know the meanings of the adjectives and the nouns can produce and understand combinations of the words that they've never heard before, in fact, that have never been uttered by anyone before. This is one of the senses in which language is productive; that is, it permits new expressions to be produced and understood. The possibility of combining words into phrases was a revolutionary leap for our ancestors, probably the single most important feature of human language. I will have more to say about productivity later in this chapter and later in the book, especially in Chapter 8.

Before we go on, we should keep in mind the difference between describing of a property of language and explaining how people actually make use of this property. While everyone agrees that people can produce and understand (in some sense of "understand") combinations of words they haven't heard before, there is a lot of disagreement about how this works in the mind (and the brain). In particular, there is the question of how explicit the knowledge is that permits this ability. Do English speakers have an explicit rule that tells them how combinations of adjectives and nouns are to be interpreted? Or do they remember many examples of of these combinations, along with their contexts, and then somehow combine all of the relevant examples when they're faced with producing or understanding a new combination? For the moment we'll be focusing mainly on the perspective of linguistics, that is, describing what's going on from the perspective of the language rather than the language user. But we'll need to return to Speakers, Hearers, and Learners later on.

Scalar adjectives

Compare the meaning of big in big elephant and big mosquito. What do these examples tell you about the meaning of adjectives such as big, soft, and sour?

Let's consider in more detail how adjectives designate attributes. The simpler cases are the ones where the interpretation of the adjective seems to be almost completely independent of the noun that it modifies. Pregnant is pregnant, whether we are talking about a pregnant woman or a pregnant tiger. But there are not many such adjectives, either in English or in other languages. More often the interpretation of the adjective depends in one way or another on the context: on the modified noun, on other words that occur before or after the phrase, or on the situation in which the phrase is uttered.

Most adjectives have relative, rather than absolute, meanings.

Consider a continuous dimension, such as size, darkness, age, or crispness. There are many, potentially an infinite number of, possible values on such dimensions. To designate an attribute for a thing on dimensions like these, we could have words for particular ranges of values, for example, 10-20 cubic meters on the size dimension and 2-3 years on the age dimension. That is, attribute words could have an absolute interpretation. But this isn't the way adjectives for these dimensions usually work. In English, the crispness dimension has adjectives for the two poles of the dimension, mushy (or soft) and crisp. But neither of these adjectives has an absolute meaning; the precise meanings of mushy and crisp depend on how they are used.

Compare the meanings of a mushy apple and a mushy orange. In both cases, mushy means something like 'closer to the mushy end of the crispness dimension than some standard of comparison'. But the standard differs. In the first case, the standard is either the crispness of a prototypical apple or some set of apples familiar to the Hearer or present in the utterance context. In the second case, the standard is the crispness of a prototypical orange or some familiar set of oranges. Mushy in the second example probably describes a lesser degree of crispness than in the first case.

Adjectives such as big, little, crisp, mushy, dark, and light that designate values on continuous dimensions are called scalar adjectives. Scalar adjectives do not normally designate absolute values or ranges of values. Rather their meanings are relative to a standard provided by the context. The Hearer is faced with the problem of determining what this standard is. The relative nature of scalar adjectives allows us to use the same adjectives for things with all sorts of values on the relevant dimensions. More importantly, scalar adjectives are probably relative because it is the relative value of things that matters. Everyone who knows elephants knows that they are large creatures. To distinguish one elephant from another, however, we need a word that designates not absolute size but elephant size. Big and little do this for us.

Consider how adjectives are like and unlike nouns. Just as a noun like raspberry designates a category of things, an adjective like pregnant designates a category of things (all of the animals that are currently pregnant). But most adjectives designate strange sorts of categories. While knowing that something is a raspberry tells us many things about it, knowing that something is red or crisp tells us very little; we know about only one of its properties. At first glance, it may seem that adjectives are simpler than nouns because we have only one feature to worry about. But the fact that each adjective lumps together so many things that are so different from one another makes them more complex than nouns in another way. It may be for this reason that adjectives seem generally harder for children to learn than nouns (this is true in English in any case; in many languages it's hard to compare nouns and adjectives because there are so few adjectives). It may just be hard for children to focus in on the single dimension that matters for a word like thin, red, or smooth. For more on this possibility, see this paper that I wrote with Linda B. Smith.

TOC Email author Printer Friendly Next