In Chapter 2 we introduced the idea that signs can “mean” in two ways. First, they can be associated with a particular concept. This is what we mean when we say “What does tree mean?” We expect an answer that describes the associated concept, e.g. “a perennial woody plant of considerable size with a single thick stem”. We call this sense. Second, they can be used to refer to a particular entity. This is what we mean when we say “Which tree do you mean (when you say the tree)?” We expect an answer that points out a particular tree. We call this reference.
Reference is the simpler of the two notions: in order to determine whether a particular noun N refers to an entity x, we can simply ask whether the statement “x is (an) N” is true. If it is, the noun n refers to the entity x, if it is not, it doesn’t. Put simply, understanding reference is understanding how language users are able to determine the truth of such statements.
This is best understood by looking at nouns, whose typical function is to refer to entities (but we will come back to other word classes later in the chapter). Look at example (1):
This sentence describes an event involving two entities — a person and a tree —, and it contains two nouns referring to these two entities — the proper noun Zoe and the common noun tree. These two types of noun differ fundamentally in the way they refer.
Let us begin with the proper noun Zoe. We use proper nouns to talk about individual entities — the most typical case is that of proper names like Zoe in (1), Spain or Venus, but proper nouns can also be fixed expressions consisting of common nouns, like Terminator, United States Virgin Islands or morning star, or of a mix of common nouns and proper names, like Agent Scully, German Democratic Republic or Barnard’s Star.
So, how are language users able to determine whether the proper noun Zoe refers to a particular individual entity? On what basis do we determine the truth of a statement like “this person (x) is Zoe (N)?”
Some philosophers have suggested that we do so by associating proper nouns with descriptions of their referents. For example, the name Zoe might be associated with descriptions like “my best friend” (in Aylin’s mind), “the young woman with green hair who lives next door” (in the mind of her neigbour), “my youngest daughter” (in the mind of her mother), etc. If Aylin had to decide whether the statement “this person is Zoe” is true, she would have to check whether the person in question is, indeed, her best friend, her mother would have to check whether the person in question is her youngest daughter, her neighbour would have to check whether she has green hair and lives next door from him, etc.
The idea seems straightforward, but it cannot work: language users would have to update their descriptions continually — Aylin and Zoe could have a fight and no longer be best friends, Zoe could move away and no longer live next door from the neigbour, Zoe’s mother could have another daughter so that Zoe would no longer be the youngest. The problem is not that language users are unable to update such descriptions — they are easily changed to “my former best friend”, “the young woman with green hair who used to live next door”, “my second youngest daughter”, etc. The problem is that, in order to know that they have to update their descriptions, Aylin, Zoe’s neighbour and her mother have to have a way of knowing who the name Zoe refers to that is independent of these descriptions. Otherwise, they would automatically transfer the name to a different entity (Zoe’s mother would call her new-born daughter Zoe, Aylin would call her new best friend Zoe, the neighbour would call the new tenant Zoe if she has green hair, or think that Zoe no longer exists, if she does not. In sum: while language users do associate descriptions with proper nouns, these descriptions do not tell them who or what those proper nouns refer to.
Others have suggested that that names refer to a particular entity because a naming event occurred at some point in the past — for example, Zoe’s parents entered this name on her birth certificate, Spain adopts a constitution that stipulates its name, astronomers get together and vote to name a star after the astronomer Barnard, etc. In order to determine whether “x is N” is true in a particular case, language users would have to trace N back to such an initial naming event.
If we take the idea literally, it is highly implausible: it is simply not the case that language users trace the names of entities to specific naming events in the past — we’ve never see most people’s birth certificates, we’ve never read most countries’ constitutions, most of us never attend astronomers’ meetings or even read about them after the fact.
But interpreted a little more generously, the idea does make sense. We learn the names of entities in a variety of ways (people tell us, we read them on a sign at the border, we read them in a magazine or book, etc.). But we accept them based on an assumption they are the result of an initial naming event at some point in the past. It follows that the name of an individual entity is whatever language users agree on: “x is Zoe” is true if “people regularly call x Zoe” is true.
In the case of proper nouns, that agreement is all there is to reference. The members of the category “entities called Zoe” (some of which are shown in Figure 3.1.1) share this one property that distinguishes them from entities outside of the category: the language community (or a relevant part of it) calls them Zoe. Apart from that, they do not have to have anything in common — they can be people, pets, houseplants, ships, hurricanes, and many other things.
In order for a proper noun to have reference, then, language users have to associate it directly with the their concept of one or more individual entity.
Note that names are not usually unique. Zoe is a popular name so there are millions of people (and other entities) called Zoe, and many language users are likely to know more than one of them. This raises the question how speakers determine which Zoe is referred to in any given utterance. The answer is, roughly speaking, that Zoe always refers to the Zoe that is most salient in the current discourse (for example, because she is present, because it is the only Zoe that all language users involved in the communicative situation know, because she has been mentioned earlier, etc.).
Common nouns are different, and much more interesting to linguists. Look at Figure 3.1.3, which shows some members of the category “entities called tree”.
Images of different types of trees, including large and small trees, coniferous and deciduous trees, large and small trees, trees with and without blossoms
Unlike entities called Zoe, entities called tree share a number of properties that, taken together, distinguish them from entities that are not in the category:
This means that in order to determine whether the word tree refers to a particular individual entity, language users do not have to check the truth of the statment “people call x tree”; instead, they check the truth of the statement “x is a plant, grows for more than one season, is woody, has a single large trunk, …”.
More generally, in the case of common nouns, language users do not associate a form directly with a concept of one or more individual entities, but with a category defined by a set of properties. The members of such a category are called the extension of that word — the extension of the word tree includes all trees (past, present and future). The properties that define the category are called the intension of that word — the intension of the word tree includes the properties listed above.
In other words, a proper noun has to be associated with the individual entities in its extension. In contrast, a common noun can be associated with a set of properties that makes up its intension — its extension then follows from this intension.
Thinking about meaning in terms of extensions and intensions allows us to take a more differentiated view on words like unicorn. In Section 2.3, we claimed that this word does not have reference. This seems right: there are no entities of which we could truthfully say “this entity (x) is a unicorn (N)”. We know this, because (a) we know the intension of the word unicorn, which includes the properties ‘is an animal’, ‘has four legs’, ‘has hooves’, ‘has a mane’, ‘has one horn on its forehead’ and (b) we know that the category defined by these properties is empty.
But note that the category is only empty in one particular world — the world we call “reality”. But human languages are not restricted to talking about reality! We can imagine, and talk about, other possible worlds, including worlds in which there are entities of which we could truthfully say “x is a unicorn”. Some of these worlds exist as something we call fiction — for example, Ponyland from the TV series My Little Pony, Centopia from the TV series Mia and Me, or the unnamed world of Pokémon. This list can easily be extended by conjuring up additional fictional worlds privately or collectively. Other worlds exist as hypothetically real worlds — a subgroup of horses could evolve a horn (like narwhals or rhinoceroses did), we might find fossil remains of unicorns, there might be an alternative timeline to ours in which all horses are unicorns.
As far as language is concerned, there is no fundamental difference between the world that happens to be our reality and the other worlds mentioned here. We use language in the same way when talking about reality, when talking about fiction, when talking about alternative realities, etc. This does not mean that we cannot distinguish between these worlds — we know that one of them is real and the others are not. But it means that the question of extension can only be answered with respect to a given possible world.
Note that the “possible” in “possible world” does not mean “possible in terms of what we know about reality” — horses could evolve horns in reality, but even if they could not, there would be possible worlds in which unicorns exist; everything we know about the aerodynamics of flight tells us that pegasuses (winged horses) would not be able to fly, but we can imagine and talk about winged horses just like we can talk about unicorns. Instead, possible means “possible to imagine in a contradiction-free way”. Thus, there is a possible world in which we could truthfully say Onchao is a unicorn or This unicorn has wings, but there is no possible world in which we could say This circle is square or Celestia and Ponyta are each taller than the other.
