Introduction to the Philosophy of Language

Why is Language of philosophical importance?

For the last century or so language has been one of the most important subjects of philosophy. There are a several reasons particular to the period which encouraged

1. Verificationism and meaning

In the early part of the 20th century some European philosophers became frustrated with the style of philosophizing that was then popular. It seemed to them that much of what was then being written was unintelligible gibberish. You can see what they mean by opening almost any work of Fichte, Schlegel, Hegel, Husserl, Heidegger, etc. at any page and trying to understand what is being said. This stuff, they thought, is just meaningles, and should be disallowed. But how to justify ruling so much stuff out of bounds? They had to come up with a criterion of meaningfulness. This they did by declaring that a sentence was meaningful if it contained within itself the means of its own verification.

Now, this criterion turned out to be unsatisfactory for several reasons – not least because it seemed that the statement of the criterion itself would fail the test that it laid down for meaningfulness. Nevertheless, the idea that some criterion of meaingfulness existed – or the question, rather, of whether any such criterion could exist – encouraged philosophers to look closely at meaning itself as a proper object of study to see what its characteristics were.

2. Sapir-Whorf

Also early in the century the increased anthropological awareness of other cultures that was a consequence of the easy access to the world that was itself a consequence of the European conquest of most of the world issued in a theory of epistemological relativism that was based upon observations about the languages of other peoples. A theory was proposed, which we know under the name of the Sapir-Whorf hypothesis, that the way that people see the world is determined by the way that their language describes the world – and not, as might have been naively thought, vice versa. Accordingly, the Hopi (an American Indian tribe) were said to have a peculiar conception of time that did not involve an awareness of time passing, or of the succession of events, and this was because their language had a particular type of tense structure that did not make the appropriate distinctions. Probably the best known of the claims related to this hypothesis is the one about the 40/60/100/10,000 different terms for snow that the Eskimoes have (though it is never quite clear how this factoid would support that theory.)

This theory is no longer so popular – at least not in such a crude form. The relationship between thought and language is now thought to be somewhat less direct and it is once again generally believed that the facts of the world are more important for our conception of the world than the facts about the way that language describes the world. Note also that neither of the examples I’ve just mentioned are credited any longer. Nevertheless, the fact that meaning and language could possible have such huge consequences for epistemology, ethics, metaphysics, and many other essential studies was another encouragement for the philosophical study of meaning and language.

3. Logic

Advances in logic at that time gave an impetus to studies when it started to become possible to treat in a rigorous fashion the relationship between the truth of statements and the references of the parts of those statements. And the one thing that everyone is sure of is that if you know the meaning of a word or phrase then you know what it is referring to, so perhaps there are clues to the nature of meaning to be had from a better knowledge of how reference behaves.

What are the characteristics of a language that we're interested in?

So much for the reasons for the renewal of philosophical interest in language. But whatever the initial reasons this may have been, it has turned out that the philosophy of language is of great interest in its own right, and may have much to say about the solutions to – or the dissolution of – problems in other areas of philosophy. Therefore theories of language have proliferated, all of them trying to solve the problems that arise when one considers in detail how a language can do the things it does. These problems will be presented in detail as we go on, but it may be worthwhile in this introductory section to have a closer look at the characteristics of language, which set the parameters for our plausible theories of it.

There is a nice list in Sterelny Language and Reality. Language is a means of communication, at least, that has the following properties:

1. Stimulus independence

There used to be a philosophical programme called behaviourism that claimed that every behaviour of a creature could be explained in terms of stimulus and response. That included human verbal behaviour – in fact BF Skinner wrote a book of that title making that specific claim. This programme was eventually ended, partly at least by a scathing review of that book by Noam Chomsky. It is quite clear that no matter how detailed our description of the purely physical environment of a language user we will not be able to predict the verbal behaviour of that person. This is quite different from the case of most animals – although I’m willing to make an exception for some primates.

Most animals that have any form of communication at all seem to have no more than a very restricted set of signals. Each signal has a specific function and is never used for any other reason. The peacock spreads his tail and signals a readiness to mate; the robin sings in the morning and signals his readiness to defend his territory, the monkey screeches in the tree and signals a warning to his fellows about a danger. A dog points and signals the direction in which his master’s game may be located. And so on. Such signals can have some layers of complexity, as for example, in the case of the bee’s dance that signals the direction and the distance to a place where she’s just found food. Under a certain description, there are an infinite number of such dances; but there is this difference from human speech, that if we know the distance and the direction we can predict with complete certainty exactly what the dance of the bee will be. There is only one such dance that will be danced. This is very different from what can be expected from a person who has just returned from a restaurant. Who could predict what one of you would say after having been to a restaurant? Your flexibility of behaviour forbids it, and the variety of language would disallow a prediction even if we knew that you were going to say something about it – indeed even if we knew exactly what you had to say about it we couldn’t reliably predict just how you were going to say it.

2. Abstractness

With language we can communicate certain facts about a situation without having to communicate any other facts at all. Thus if I want to say that Cain killed Abel I can do that by saying ‘Cain killed Abel’ without telling you anything about how Cain killed Abel. Did he stone him, or strangle him, or stab him? Who can say? You can’t be so continent so easily with any other form of communication. A picture of Cain killing Abel must tell you a great deal about the act of killing besides that it took place. A picture of Cain killing Abel that shows him being stabbed tells you how it was done, and the relative positions of Cain and Abel when it was done, and what they were wearing, and whether Cain had a longer beard than Abel – because he was a hairy man, and so on.

3. Arbitrariness

This is about the most obvious characteristic of language and has been much remarked upon. The connection between words and objects is, these days, widely accepted as being purely arbitrary. Whether ‘foot’ should mean this object at the end of my leg – as it does in English – or should mean the sexual act – as it does in French (to the amusement of Shakespeare’s audience) seems to us now to be a matter of pure convention. And yet it was not always so. There are tales, possibly apocryphal, of Pharaoh Psammetichus of Egypt or the Prussian King Frederick who wished to discover whether there was a true language that calls things by their true names. Thus children were raised without being corrupted by the language of their guardians to see what they would naturally speak. In the case of the Egyptians I believe the true language turned out to be Phrygian (Herodotus, II), whereas the German, naturally enough, expected the child to speak Hebrew. Hebrew being the language in which God spoke to the Jews it must be a language worthy of God, so it must give the true names of things. Moslems believe that the Koran is written in Arabic in heaven and for a long time were much opposed to its translation. Much magic is involved with discovering these true names because that knowledge will give one power over the things themselves. Etc.

Sterelny quotes Lewis Carroll who makes an amusing and famous joke about this in Through the Looking Glass: Humpty Dumpty says

“… there’s glory for you!”
“I don’t know what you mean by ‘glory’,” Alice said.
Humpty Dumpty smiled contemptuously. “Of course you don’t know – till I tell you. I meant ‘there’s a nice knock-down argument for you!’ ”
“But ‘glory’ doesn’t mean ‘a nice knock-down argument’,” Alice objected.
“When I use a word,” Humpty Dumpty said, in rather a scornful tone, “it means just what I choose it to mean – neither more nor less.”
“The question is,” said Alice, “whether you can make words mean so many different things.”
“The question is,” said Humpty Dumpty, “which is to be master – that’s all.”

4. Medium independence

When a bird sings it uses its voice, and when a bee dances it uses only its own bodily movements. It is impossible for the bird or the bee to comprehend the same messages by their being presented in any other way. For ourselves, by contrast, any message can be transmitted in speech, in writing, or in Braille. Those codes can be on metal, stone, clay, bark, leaves, skin, paper, etc. And it is not beyond our ability to imagine a smell or taste code that would accomplish the same informative function.

5. Power

The bee can only talk about where her last meal came from, and the calls of other animals are similarly restricted in the topics that they can deal with. There is no such limitation that we know of in human language.

6. Productivity

Almost all the signals that we use (either in speaking or in hearing them) are sentences that we have never used before. This indicates that we should not treat these signals as the fundamental objects to which meaning is attached. It cannot be that we learn the meaning of each sentence independently of all the other possible sentences because their infinite number and their undoubted novelty would make that impossible. (Taking this fact seriously is one of the reasons that we can’t take the behaviourist account of verbal behaviour seriously.) Instead, we say that the meaning of a sentence must be somehow derived from the way that the sentence is built up from the parts of which it is composed – which is to say, of the words that occur in it, and the order in which they occur. The number of words is finite and the number of rules that are applied in putting them together in a sentence is conceivably finite; and therefore they are within the power of a finite mind to master.

There are many possible general theories about how words go together to make up sentences, and we’ll have a look at a trivialised form of one of these theories later on, but we won’t go into any details of them. This is not at all to minimise their significance, for the current linguistic theories are some of the most sophisticated scientific theories that we have of any of our cognitive functions, and they have had a considerable impact on some parts of the philosophy of language. Any serious student will have to make himself familiar with them at some stage.

7. Systematicity

There is one further property of language that should be mentioned here, which we generally call by the name of systematicity. By the systematicity of language we mean the fact that if we can say that the cat is on the mat, then we can say that the mat is on the cat. Now, note that I don’t mean that if I have looked up in a phrasebook the phrase for ‘dog tastes better than cat’ in Cambodian that I thereby know how to say ‘cat tastes better than dog’ in that language. There are all sorts of grammatical rules that might be important to that phrase and that the phrasebook has nothing to say about. But, if I know a language that has the resources to say ‘the cat is on the mat,’ then that language will have the resources to say ‘the mat is on the cat.’

Apart from these characteristics that we think must be true of anything that we call language, we should note that there are a few things that are particularly true of human language which may or may not be found to be of philosophical significance.

8. Universality

There is no reliable record that any group of humans have been discovered anywhere who do not possess language. But what is the significance of this? It certainly doesn’t follow that language is a necessary accomplishment of humans for it may well be that it is only a contingent truth of humans that they all speak a language. If we found that there was no reliable record that any group of humans has been discovered anywhere who do not possess the art of body decoration, what could that tell us about the necessity of body decoration? No more than that it is an obvious art, and one that gives pleasure of some sort to people, and within every normal human’s power to indulge. The fact that language is so overwhelmingly useful would ensure that it was spread from any group that had it to all those who did not have it, and it would never die out of a society as long as any parent wished to communicate with any child.

There is a clue in this universality that does suggest that language is more than just such a cultural product, and that is the very uniform degree of sophistication of all languages that we have made any study of. In almost all other cultural forms there is a wide variation in the obvious sophistication of these forms, and it is in those variations that we find our justifications for calling some societies relatively primitive and other societies relatively advanced. Generally speaking too, the society that is primitive in one area is primitive in other areas. But this is not the case with language. There is no defensible intuition that allows us to distinguish the power and complexity and so on of the languages of the Australian aboriginal or the !Kung San or the wild people of Patagonia from languages of Europe, India, China, etc. Even stranger, there is no discernible difference in sophistication between the languages spoken by the most esteemed and the least esteemed members of a single society. (The extent of vocabulary in all cases is not relevant here.) But, again, this could be explained without much implausibility as due to the equal powers of human intellect in all cases, and the similarities of grammatical power, where they exist, are due merely to the necessity of communicating similar phenomena in all cases.

9. Innateness

Universality and uniformity could be explained if language were an innate ability of humans, but as evidence for it they are by themselves insufficient to justify the hypothesis. They are not the only evidence, however, and the idea of language’s innateness is now well accepted. But perhaps we should first ask just what is meant by saying that language is innate. Well, to begin with, we mean more than that people have a predisposition to learn languages or that they all can learn languages. That much is obvious and tells us nothing interesting. What seems, rather to be meant is that there is a specific mechanism in the brain that is designed for language acquisition and which determines the types of languages that can be acquired.

What further evidence, then, has convinced us of this thesis? There are three main considerations.

i. In the first place, there is the evidence that Chomsky and others have noted of the poverty of stimulus for language learners.

We have a fairly well-accepted model of how our general purpose learning device works which goes something like this: if we are learning to write, for example, we are shown examples of writing; we form an hypothesis about how we should write something; and we try it out. If we are correct we are rewarded by the teacher or by being legible to others, but if we fail we have our errors pointed out to us, and we use that information to improve our performance. Similar processes are in play when we learn to fish, play music, ride a bike, and so on.

In that model of learning there is always enough data in the stimulus to reasonably justify an induction to the general rule being sought. There are no great leaps. By contrast the child learning language seems to do nothing but make leaps. The child does not seem to require that there be the information in the stimulus provided in order to form correct hypotheses about how language works. Indeed, given the way that (Western) adults (supposedly competent speakers) actually speak to children, it is as if the information is being deliberately falsified. [We have an implicit theory that we have to teach children to speak using baby-talk or ‘motherese’, but not all cultures do this, and the effects are no different between the two. Note that the Bushmen mothers similarly feel that they have to prop their children up so that they learn to sit, whereas we are comfortable in letting this be a matter of chance.]

Kids learn that if they’re given a sentence like

             John is happy

They can turn it into a question by moving up the first ‘is’, thus

             Is John happy?

So, they should do they same with

             John, who is tall, is happy.

To get

             *Is John, who tall, is happy?

But this never happens. Why not? According to the innateness hypothesis it’s because they already have a model of how languages work in general due to a language-specific cognitive device. This tells them that sentences are built of subordinate elements that can only be transformed in certain ways, and the transformation that led to the grammatical error is just impossible. If the kids were going to learn this by a general learning procedure they’d need to have seen quite a few examples of subordinate clause constructions and transformations. My parents never used subordinate clause constructions in front of their children.

ii. The phenomenon of language learning is quite unlike other types of learning.

a. 

There seems to be a period of full plasticity up to about the age of 6 when languages can be easily learnt, and after that age the facility with which languages are acquired declines to a minimum that occurs just about puberty. After this point languages can not be learnt in the same way that they are learnt during the plastic period.

We’ve all seen this with immigrants: if immigrants begin to learn English as children then they learn to speak it unaccented and with native competence. Their poor parents, by contrast, tend to retain heavy accents and to have grammatical difficulties which never disappear. Children who are isolated from any language at all – like wild children, or children kept locked in cupboards by lunatic parents, or children who are deaf but misdiagnosed as morons – if they are not exposed to language in the critical period, will never acquire facility in even one language. Compare two cases[1]: Isabelle was kept in a silent environment until she escaped at age 6½, within a year and a half she had 1500 words and could say things like ‘What did Miss Mason say when you old her I cleaned my classroom?’ Chelsea received hearing aids at the age of 31 and subjected to intensive language training by specialists. She can now say things like ‘The girl is cone the ice cream shopping buying the man.’ 

One other interesting fact about this language learning period is that there seems to be no limit in principle to the number of languages that can be learnt simultaneously. If a child is being brought up in a house where there are several languages being spoken to it it will acquire these languages at the same time and it will not confuse the lessons that it learns in one language with the lessons it learns in any other.

b. 

The order in which language elements are acquired is the same for all children and all languages.

Children start to understand and produce words at about 1 year old. At 18 months they’re learning one every two hours, and will continue to do so till the critical period closes, and they start forming minimal grammatical fragments of two words. Around three the grammatical sophistication begins to increase at such a rate that exact stages are very difficult to isolate.

c. 

The level of achievement is not as variable amongst language learners as it is for learners of other skills.

iii. There appear to be very specific areas of the brain related to language.

We don’t need to go into any detail about this. I’m sure it’s pretty well known generally that there are areas such as Wernicke’s and Broca’s that are implicated in linguistic behaviour.

The idea that there’s an innate language-specific cognitive device that has information about languages that we can learn brings up once again the idea that there are language universals and perhaps a real or a one true grammar. You’ll recall that when we previously wanted to test this hypothesis, kings Psammetichus and Frederick tried to bring up children without letting their language learning be contaminated by languages being spoken to them. We’ve seen that the type of experiments that those kings ran couldn’t be successful, but we have records of the next best thing when children learn to speak from pidgin speakers. Pidgins are created when people without a common language are forced to live together. They tend to be bits and pieces of vocabulary from several languages and to have very little grammar so that even subjects and objects are indistinguishable. When children are brought up in a pidgin environment they adopt the vocabulary but they fit it into a grammatical structure that they appear to generate all by themselves. Pidgin is not a language, but the creole that the children spontaneously invent is a language, and creoles from many places have such similarities that show that more than mere coincidence is involved.

Probably the most interesting creoles are those created by deaf children who are taught sign language by their non-deaf guardians. The children generate a grammar which they can pass on to other children who learn it later. When the original sign language is developed spontaneously as a result of bringing together a bunch of deaf people (older than about ten) it is a pidgin. This is what happened in Nicaragua. But when children learn it as their first language it is modified to become a fully grammatical language. The special interest here is in two factors: first, the language is in a quite unusual medium which humans don’t typically use for speech; and secondly, the process of language formation can be observed in the timescale of a single researcher.