On Language and Meaning

So many philosophical discussions seem to end up in confusion over words, that I wonder whether that is all that such discussions are: fancy piles of words, with no meaning, that we continually rearrange to see whether they look better when lined up this way than that. Yet many philosophical discussions, particularly some of those those about ethics, politics, religions, logic and science, are very valuable, and can help people live more successful, fulfilling, happier lives. So it seems worth trying to find a way to decide which discussions are meaningful and which are not. This essay sets out the results of my attempts to do that.

Finding meaning via a tree of definitions

My first observation is that we cannot find the meaning of a statement by looking up all the words of the statement in a dictionary. A dictionary defines every word in terms of other words. If we look up each of those other words, and so on, it won’t be long at all before we find ourselves going round in circles, looking up words that we have already looked up.

Digression: here’s a fun parlour game. Everybody picks a word. You look it up in the dictionary, choose the seventh word in the definition (or the last word if there are less than seven words), look that up, and so on until you are sent back to a word you have already looked up. The person that has to look up the most words before closing a loop wins. Or you can reverse the game and make the winner the one that closes a loop in the fewest lookups.

Now think about what happens if we look up all the words in each definition, rather than just the seventh one. Then we will have created a branching tree of definitions. We start with one word, then have to look up maybe twelve different words used in its definition. For each of those words we need to look up each of the different words used in their definition, and so on. The trunk of the tree is the first word. Every definition we look up is a place where the tree splits into several branches.

What we need, to terminate this process and arrive at a final understanding, are leaves. Because the leaf is where the recursive branching process of a tree finally stops.

We are not going to find leaves in a dictionary (yeah I know, pages are also called leaves, ha ha), nor yet in Google or Wikipedia, which are just different forms of dictionary. So where could we find them? The most natural way to find them is to look at the methods employed by a group of people that learn language without the aid of dictionaries – toddlers. A toddler learns a new word by observing others use the word in connection with a certain phenomenon. They learn ‘jump’ by observing people say sentences containing the word ‘jump’ when there is jumping going on. They learn ‘dog’ by observing people say sentences containing the word ‘dog’ in the presence of dogs, maybe accompanied by the speaker helpfully pointing to the dog, or a picture of a dog. They learn blue by being shown blue things while being told ‘blue!’ They learn ‘above’ by hearing that word while attention is drawn to one thing that is above another.

Ludwig Wittgenstein called this method of learning language ‘ostension’, and regarded an ‘ostensive definition’ as a naming of a phenomenon while pointing to it. Ostensive learning is most obvious with toddlers, but it is not confined to them. Most people tend to pick up new words by ostension throughout their life. Things like local slang, or the latest buzzword or jargon used within a specific trade are mostly learned ostensively. That is how I learned, in the early 2000s, that we can say ‘my team is going to verse Little Boggling High School in rugby‘ rather than ‘we are going to play Little Boggling High School in rugby‘. When explorers encounter people speaking languages they do not know, they learn to communicate by ostension. That is less common now, but one imagines it was a frequent occurrence in the 18th and 19th centuries.

Even ostension is not enough to completely terminate the search for meaning though. As Wittgenstein observed, when somebody points at something and names it, we still need the mental equipment necessary to associate the sound of the name with the thing. It appears that that equipment is built-in (‘hard-wired’) rather than learned, as are a range of other mental processes we instinctively perform, such as learning from experience (the ‘principle of induction’) and our ability to interpret our sensations in terms of a continuum with three spatial dimensions and one time dimension (an innate ability that Immanuel Kant calls ‘the Transcendental Aesthetic’). We evolved to have those abilities hard-wired, because ancestors that did not have them quickly perished.

Although we have those learning capabilities built-in, I do not think we have any actual words built in to our brains at birth. So far as I know, there is no word that is universal across all Earthly languages – No, not even ‘No!’ or a shake of the head. Hence, I can only think of two ways that we learn words, or other parts of language: by definitions that we invent (eg for new words or new uses of words we invent) or that are provided to us, that use only words we already know; and by ostension. My theory is that every part of language we learn is acquired by one of those two methods. If somebody can identify another way humans can learn language, I will have to revise my theory, but for the remainder of this essay I will assume it to be accurate.

A consequence of this theory is that, if there is any word that I cannot trace through a tree of definitions, without circularity, to a full set of ostensive leaves, then I do not know what that word means! Even a single non-terminating branch in the tree is enough to make the word meaningless.

One can think of the ‘depth‘ of the tree as the maximum number of definitions one has to pass through to get to the farthest leaf, and the ‘breadth‘ of the tree as the number of leaves. For ostensively defined words, like the above examples of ‘dog’, ‘blue’ and ‘jump’, both breadth and depth are only 1. We know what they mean without having to think of other words. Some words, while not ostensively defined themselves, will be explicable solely in terms of ostensively defined words, and have a tree depth of 2. ‘Panther’ might be an example, defined as ‘black leopard’, if we have access to pictures of leopards but no pictures of panthers.

For abstract words such as ‘evaluate’ or ‘pretend’, the tree of definitions will be quite deep and it can be daunting to trace back to the ostensive leaves from which the meaning is ultimately derived. Yet it seems that it must be possible, because we have come to understand those words and, according to my theory, we could not have done that unless we had traversed such a tree in the course of our learning our language.

It’s worth noting here that for many words we will have more than one tree of definitions available to us. As long as they ultimately convey the same meaning, that is no problem. Indeed, it is to be expected, given the rambling, nonlinear way in which we learn language. Some observers feel that there are so many intersecting trees of definition that a person’s vocabulary is more like a web than a set of trees (I think Quine’s view, from his essay ‘Two Dogmas of Empiricism’ is something like that). But that doesn’t change the fact that ultimately that web must terminate in – be surrounded or supported by a frame of – leaves of words that we have learned by ostension.

Some occasionally meaningless words

That’s enough theory. Let’s get down to case studies of word uses that I think are meaningless. Here are some of the main culprits: ‘be’, ‘true’, ‘possible’, ‘random’ and ’cause’.

Now I don’t mean to say that any sentence containing any of those words is meaningless. Otherwise most of this essay would be meaningless, given that it contains some of those words. Indeed, the verb ‘to be’ is present in almost every piece of English ever written. I am referring only to cases where one of those words is used as if it were fundamentally understood and, like a word that is learned ostensively, needs no definition – an activity that takes place most often in philosophical discussions, especially in the field of Ontology (the study of ‘what there is‘) or Metaphysics more broadly.

There is an easy test for whether a use of a word is meaningless. Let us call a word whose meaning I am calling into question a ‘challenged word‘. A word whose meaning, as used in the sentence, can be indicated by ostension cannot be a challenged word (note that none of the four words listed above can be taught by ostension). For other words, if a sentence containing the challenged word can be translated to a sentence, or sequence of sentences, that does not contain that word, or any synonym thereof, without any loss of meaning, then the use of the word in that sentence may be able to be considered meaningful. Otherwise it cannot. I say ‘may’ rather than ‘can’ because it is possible that the translation will throw the burden of definition onto a related but not synonymous word, and that challenging that word leads, possibly after a small number of additional challenges, back to the word we started with. So this ‘translation test’ for meaninglessness can give false negatives (ie fail to identify meaningless uses of words), but it will not give false negatives (accusations of meaninglessness when there is meaning there).

The translation test, applied to ‘to be’

Here’s how this ‘translation test’ works, applied to several different uses of the verb ‘to be’. There are a number of different ways in which this verb can be used. For each case I label the type of use (bold text), show the archetypal grammatical form of its use (plain text. The word ‘copula’ denotes an instance of ‘to be’), then provide a sample sentence and a translation that avoids use of the challenged word.

  1. identity, of the form “noun copula definite-noun”:
    1. This is Freya;
    2. Translation: People call this person Freya.
  2. class membership, of the form “noun copula noun”:
    1. The cat is an animal;
    2. Translation: The class of animals includes this cat.
  3. predication, of the form “noun copula adjective”:
    1. The cat is furry;
    2. Translation: Fur covers most of the cat’s body.
  4. auxiliary active, of the form “noun copula verb”:
    1. The cat is sleeping;
    2. Translation: The cat sleeps.
  5. auxiliary passive, of the form “noun copula verb”:
    1. The cat is bitten by the dog;
    2. Translation: The dog bites the cat.
  6. existence, of the form “there copula noun”:
    1. There is a cat
    2. Translation: not always possible – see below.
  7. location, of the form “noun copula place-phrase”:
    1. The cat is on the mat;
    2. Translation: The mat has a cat on it.

All of these have satisfactory translations except possibly 6 – existence. Whether 6 can be translated depends on context. If I were pointing at the cat, a suitable translation would be ‘Look! do you see that cat?‘ Alternatively, I might be telling you about a house you are going to visit, whose occupant has a house cat. Then it could translate as ‘You may come across the occupant’s cat when you visit‘. Even if the statement were a bare assertion of existence, such as a claim that, contrary to rumours of extinction, there is still at least one live cat in the world, we could still translate it. In that case an acceptable translation would be: ‘If you could line up all the animals in the world today and inspect them one by one, you would find that at least one of them belonged to the cat species‘.

Notice that most of these translations remove the challenged word by introducing references to experienced observations. That is the easiest way of planning a tree of definitions that can be traced to ostensive leaves. Since all ostensive leaves come from observed experiences, using words that connect to experiences – whether past, future or hypothetical – makes the tree of definition easier to imagine.

It is only when the speaker tries to divorce their meaning from experience that the translation test is failed, and meaning consequently evaporates. Examples of this are:

  1. I think, therefore I am
  2. The consecrated host is the actual body and blood of Jesus Christ, even though it doesn’t look, feel, smell or taste like it
  3. The Father, Son and Holy Ghost are three separate persons but they are one essence, substance and nature
  4. Electrons are there all the time, even when we are not observing them
  5. If Captain Kirk’s body were duplicated atom for atom by a malfunction of the Star Trek teleporter, one of those duplicates would be him, and one would not.
  6. The future already exists [note that ‘exist’ is just a synonym for ‘be’]
  7. Notions in physics like ‘spacetime’ and the ‘quantum wave function’ are not just mathematical abstractions. They really exist.

These sentences, each of them the subject of endless philosophical debates over the ages, are incapable of being translated into sentences that do not contain the challenged word (in bold), because they are shut off from any possible connection with actual or potential experience. According to my theory, they are thus meaningless, and the associated philosophical debates amount to no more than shovelling piles of words around and arranging them in pretty ways.

That may seem like a big claim, and on little evidence. But if the claim is wrong, it should be easy enough to demonstrate that. All that is needed is to explain what those statements mean in non-circular terms. Nobody has managed to do that in the history of philosophy, and it’s not for want of asking.

It’s worth noting that all of those untranslatable examples are cases of either the ‘identity’ or the ‘existence’ use of ‘to be’, that we identified in the above list of different uses. Items 8, 11, 13 and 14 are ‘existence’, and items 9, 10 and 12 are ‘identity’. In my experience all untranslatable, and hence meaningless, uses of ‘to be’ fall into those categories. Not all uses in those categories are meaningless though. Above I described some cases of the ‘Existence’ use that were translatable. Some cases if the ‘Identity’ use are also translatable. I will deal with one that is a famous example that is often used in philosophy, known as ‘Frege’s Puzzle‘ because it was created in the 1890s by the German logician Gottlob Frege. It is as follows:

The early Ancient Greeks thought that Hesperus, the Evening Star, and Phosphorus, the Morning Star were distinct celestial bodies. Later in Ancient Greek civilisation they came to the conclusion that they were the same celestial body – the planet we now call Venus. Frege says that no reasonable person would deny the truth of the sentence ‘Hesperus is Hesperus‘, but somebody that didn’t know that the Morning and Evening Stars were sights of the same physical object at different times could reasonably reject the suggestion that ‘Hesperus is Phosphorus‘. This spurred the creation of a whole genre of philosophical literature discussing the meanings of the phrases ‘Hesperus is Hesperus‘ and ‘Hesperus is Phosphorus‘.

Here is my translation of ‘Hesperus is Phosphorus‘.

The sight of a light in the sky just after sunset, which people call Hesperus, and the sight of a light in the sky just before dawn, which people call Phosphorus, both arise from light waves from the Sun reflected off a single large rocky object that orbits the Sun‘.

Nothing mystical and metaphysical happens in this sentence. Here the ‘identity’ use of ‘is’ just serves to tell us that two different things we see at different times come from a common physical object.

On the other hand the sentence ‘Hesperus is Hesperus‘ cannot be translated, because it is meaningless. It is a sentence that we are well rid of.

Not all sentences of the form ‘X is X‘ are meaningless though. Some of them are capable of being translated. Recently an Australian politician known to be a loose cannon, let’s call him Bruce, made an outrageous public statement about something or other. The leader of Bruce’s party was asked by journalists what he thought of Bruce’s statement. That put the leader in a tricky position, caught between being seen to condone an outrageous opinion on the one hand or being seen as disloyal to a colleague on the other. The leader’s reply was ‘Oh well, you know, Bruce is Bruce.‘ The meaning of this statement was clear to all those that heard it. It was as follows:

As you all know, the politician Bruce that made that statement has displayed a habit of saying silly, offensive things from time to time in the past, although he acts acceptably most of the time. When he says obviously silly things like that, we don’t take them seriously and neither should you.

That’s a whole lot of meaning to pack into three little words. But language can be very expressive, if used with skill.

Other questionable words

Four other words I identified as major trouble-makers are ‘true’, ‘possible’, ‘random’ and ’cause’.

I won’t spend long on them here. I dealt with the problems of making sense of the words ‘true’, ‘random’ and ’cause’ in my essays ‘Replacing Truth with Reason‘ (October 2013), ‘Some random thoughts on whether the world is random‘ (July 2013) and ‘What is a cause – trying to distill clarity from a very muddy concept‘ (June 2013).

Hmmm. Now I look back on it, I see that quite a few of my essays have been about trying to work out what certain words mean. That really accentuates the point of this essay, which is that so many apparent problems in philosophy are really nothing but confusion over words. Once we think hard enough about what the words mean, and whether they even mean anything in the context in which they are used, we see the apparently deep and insoluble problems dissolve into emptiness.

Each of those words have perfectly reasonable, practical meanings that we use every day, and in those essays I try to identify what those meanings are. It is only when people try to claim that the words stand for something much deeper than that that they lose touch with reality (and reason).

Let me say just a few words about ‘possible’ though, as the only word on which I have not yet written a dedicated essay. What do we mean when we say

It is possible that an atom in the lump of uranium in that box will undergo radioactive decay and emit a neutron within the next ten seconds‘?

As with all the other words, I have a practical interpretation that matches how we use the word in everyday life. My translation is:

Neutron emission by an atom in the lump of uranium in that box within the next ten seconds is consistent with my theory of the world and the state of my current knowledge.

Or, slightly more concisely:

I am not convinced that there will be no neutron emission by any atom in the lump of uranium in that box within the next ten seconds‘.

The crucial aspect of this translation is that ‘possibility’ is about my knowledge. Something is ‘possible’ if I know nothing that convinces me it will not happen. It is ‘impossible’ if I am convinced that it cannot happen.

Some types of philosopher – metaphysicians, and most particularly ontologists – are not satisfied with this. They dismiss that interpretation as ‘merely epistemology‘ (epistemology is the philosophy of what we can know) and insist that they are asking about whether it is ‘possible‘ for the atom to decay, irrespective of what I know. The word ‘possible’ is usually heavily emphasised, as if writing it in italics and underlining it somehow made its meaning plain. But when asked what they mean by saying ‘it is possible‘, no coherent answer can be obtained. The most we ever get is the use of a synonym, such as ‘By possible I mean that it can happen‘. But that just shifts the burden of definition to the synonym. What does ‘can’ mean here?

I don’t believe that the word ‘possible’ makes any sense at all if its meaning is not couched in terms of what we know. The absence of any tree of definition for the word is a strong sign that it is meaningless. But beyond that, there is the problem of our simple understanding that either the atom will decay or it will not. When we say it is ‘possible’ we are not saying anything about the world, because the world will unfold in a certain definite way. All we are saying is that we do not yet know how it will unfold (and stronger than that, it is usually impossible for us to know how it will unfold. But that’s the subject of another essay, yet to be written).

That’s enough about ‘possible’. More on that topic can can be found in my essay ‘Some random thoughts on whether the world is random‘, which deals with the very closely related issue of randomness.

There are plenty of other words that are meaningful in the sentences of everyday life yet meaningless in the sentences of metaphysicians. I’ll mention just a few more without discussing them. Often they occur in pairs, where the metaphysician (or theologian) is trying to make some distinction but is unable to articulate it. ‘Subjective’ vs ‘Objective’ and ‘Natural’ (or ‘Physical’) vs ‘Supernatural’ are good examples. Then there is the idea of ‘direct’ vs ‘indirect’ perception of an object. I could go on for pages about how either all perception is direct or all is indirect, but I won’t. In fact ‘direct’ and ‘indirect’ are problem words in other situations as well, such as ‘direct’ vs ‘indirect’ cause (I am confident that for any purported ‘direct’ cause and effect pair I can identify a step between the cause and the effect).


The lesson I have learned from all this deliberation is pretty simple: When confusion or misunderstanding arises over meaning, we can restore clarity by insisting on using only words whose meaning can be traced to ostension. The easiest way to test whether that is possible is to try to translate the passage into one that does not contain the challenged words – the ‘translation test’. Many passages written by ontologists, metaphysicians, theologians and others that try hard to be deep, fail this test. If we apply the test to a particularly confusing bit of writing and it fails, we would do best to consign the writing to the bin rather than waste further time on it. To apply the test fairly, we should ask the person who wrote the passage to translate it or, if they are dead, far away or otherwise unavailable, ask somebody that thinks they know what it means. But if they cannot do so, into the bin it goes!

I do not however want to create the impression that we should never talk about things we do not understand, or find impossible to clearly express. That is what the arts are for. When TS Eliot says

while the evening is stretched out against the sky, like a patient etherized upon a table‘,

we would be silly indeed to try to enumerate the ways in which an evening is really like an anaesthetised patient. The wonder of the phrase is in the emotions and images it conjures up, not the propositional information it conveys.

So by all means let us celebrate and revel in poetry, music, literature and the other arts. They form a central and essential part of my world, without which life itself would seem unbearable. But we should avoid confusing poetic speech with analytic speech. Poetry evokes imagery and emotion, but does not make logical propositions. If we start to mistake poetry for logic, we will end up in a terrible muddle.

Andrew Kirk

Bondi Junction, August 2014


The ideas in this essay are my own, and do not seem to match exactly with those of anybody I know of. However, there have been very strong influences.

I get the impression that the writer to whose ideas this essay approximates most closely is JL Austin, of Oxford University, who is seen as the intellectual father of ‘Ordinary Language Philosophy’. I may be wrong about that however, as my exposure to Austin is only through secondary literature.

Ludwig Wittgenstein’s writings in ‘Philosophical Investigations’ and Bertrand Russell’s essays ‘On Denoting’ and ‘Knowledge by Acquaintance and Knowledge by Description’ are also very influential on these thoughts. I should note that Russell had a lifelong belief in Truth and Material Reality that I do not share. Wittgenstein seemed to share those beliefs when writing the Tractatus, but perhaps discarded them in later life.

Willard van Orman Quine’s essay ‘Two Dogmas of Empiricism’ seems very relevant, but I still don’t know what to make of it. It is a paradoxical essay, in that Quine was a brilliant logician, capable of scintillating clarity, yet that particular essay is written in loose language that makes it hard to understand what he is really saying in some parts.

There is a similarity between some of these ideas and those of the American Pragmatist philosophers Charles Peirce, William James and John Dewey.

Especially interesting to me are Alfred Korzybski and David Bourland, who promoted the use of the language ‘E-Prime’, which is English with all instances of the verb ‘to be’ removed. They saw that verb as a never-ending source of confusion, and maybe even socially harmful as well. Bourland wrote a number of books in E-Prime to show how well we could do without that verb. I understand that the author and thinker ‘Robert Anton Wilson’ wrote a number of novels in E-Prime. Most illustrious of all, Albert Ellis, the founder of Cognitive Behaviour Therapy, by far the most effective form of psychological treatment of depression and anxiety (and which I see as modernised Stoicism), favoured E-Prime and saw the use of ‘to be’ as a frequent source of blame and judgement that was psychologically harmful (‘He is wicked’ vs ‘He harmed many people when he did that’). The wikipedia article on E-Prime is a rewarding read.