The Semantics of Nothing

0.   Introduction

The word ‘nothing’ has interesting semantic features. It is a ‘negative existential’, in the sense that it refers to a non-existing thing. This is perplexing, because if ‘nothing’ is a simple referring term, then the semantic role that it plays in contributing to the meaning of a sentence it features in is to point to its referent. As it has no referent, how can it play this role successfully? There are two general strategies for dealing with this puzzle; one is to treat the idea of nothing as a sort of thing, and the other is to treat it as a case of failure to refer at all.

1.   Creation from nothing

The term ‘nothing’ is deployed as part of one of the supports for the Kalam cosmological argument. The first premise of that argument is: ‘whatever begins to exist has a cause’. One of the lines of support for this premise is the familiar dictum ‘nihilo ex nihilo fit’, or ‘nothing comes from nothing’. When pressed on why this is true, a typical line of defense is that ‘nothing has no causal powers’. I say that this sentence is ambiguous, due to the word ‘nothing’. On one account the sentence treats ‘nothing’ as a referring term; something like ‘the complete lack of any object’. On the other account, the term expresses a failure to refer to any thing. The first reading (which I shall call the ‘referential sense’) is the intended sense, but it strikes me as ad hoc (and I will explain this more below). The second sense (which I shall call the ‘denotative sense’) expresses a different proposition altogether – one that fails to support the premise in any way.

2.   A Toy Example

The ambiguity can be brought to the surface if we consider the two semantic accounts of the word in more detail. Before we look at the sentence ‘nothing has no causal powers’, I want to first play with a less controversial example, to get the distinction clear. So my toy sentence is:

1) ‘Nothing will stop me getting to work on time’

First, let’s look at the referential sense of ‘nothing’, as it applies to this sentence. On this account, ‘nothing’ is just another referring term, like ‘John’, or ‘Paris’, or ‘my favourite type of ice cream’, etc. The referent of ‘nothing’ is ‘the complete absence of any things’, or something along those lines. It’s like an empty void with no contents whatsoever.

The sentence is essentially of the form ‘x will stop me getting to work on time’, where ‘x’ is an empty variable waiting to be filled by any constant (or referring term), like ‘John’ or ‘my favourite type of ice cream’, or ‘nothing’ etc. Let ‘Wx’ be a predicate for ‘x will stop be getting to work on time’. If ‘a’ is a constant that refers to my friend Adam, then the proposition ‘Wa’ means that Adam will stop me getting to work on time. I will not get to work on time, because I will be stopped by Adam from doing so. Something that Adam will do, such as physically restraining me, or hiding my keys, or just distracting me with an interesting philosophical discussion, etc, will prevent me from getting to work on time. That’s what Wa is saying.

Let ‘n’ be a constant that refers to ‘the complete absence of anything’. We could put the logical form of 1) as follows:

Ref)   Wn

Ref says that I wont get to work on time because ‘nothing’ is going to stop me. This mirrors the logical form of the sentence above where Adam prevented me from getting to work on time. But this seems wrong, as 1) doesn’t seem to say that I won’t get to work on time because of nothing (i.e. the complete absence of any thing) getting in my way. We don’t seem to be expressing the idea that ‘nothingness’ is going to hide my keys, or engage me in a philosophical discussion, etc. We are not expressing that I will not get to work on time. Rather, we are expressing something close to the opposite of that; the sentiment expressed by ‘nothing is going to stop me getting to work on time’ is that I will be on time to work, come what may. So the referential way of reading the term ‘nothing’ is not appropriate here.

Let’s look at the second account, the denotative account. On this reading 1 gets analysed out as the following (note that we still have the predicate Wx, but use a quantifier and a bound variable and so don’t need the constant ‘n’):

Den)    ~(∃x)(Wx)

On this reading, we are saying that it is not the case that there is a thing such that it will stop me getting to work on time. We could re-write Den as follows:

Den’) (∀x)~(Wx)

Den’ says that for every x, it is not the case that x will stop me getting to work on time. This captures very well the sentiment that come what may we will not let anything prevent us from getting to work on time. We would say that the denotative proposition is true in this situation, and that seems right.

Thus the two analyses are very different. They render propositions with a different logical forms and different truth-values in this case. In the referential case, we are referring to an entity, and saying of that thing that it will succeed in preventing me from getting to work on time. So the logical form of the proposition, when analysed referentially, is wrong. In addition to this semantic or logical issue, we also have a metaphysical or ontological worry. We may feel that the entity referred to in Ref is of dubious ontological status. Nothing doesn’t exist; it isn’t a thing as such. Successful reference seems to have as a presupposition that the referent exists in some sense or other. If that is right, then when we successfully refer to ‘nothing’ then there is something which is the referent for the term ‘nothing’. But if there is some referent, then ‘nothing’ doesn’t mean the complete absence of any thing. It may be that the combination of this model of reference with the insistence of ‘nothing’ meaning the complete absence of any thing is incoherent. So we can feel dissatisfied with Ref here for both ontological and logical reasons.

We may want to avoid this problem by postulating that ‘nothing’ refers to an entity, yet what it refers to is not an existing thing. Nothing is, even though it doesn’t exist. It is a something, just not an existing something. I find this way of talking almost unintelligible. It seems to me as a bedrock metaphysical principle that there are no non-existing things. There is not two types of existence; rather there is only one type of existence. If ‘nothing’ is, then it exists. The terms ‘is’ and ‘exists’ are synonymous. In this regard, I find Russell (On Denoting) and Quine (On What There Is) to be instructive.

Den, on the other hand, does not refer to any thing of dubious ontological status. When recast in the form of Den’ it clearly and explicitly quantifies over all the things that there are and says of those things that none of them are going to stop me getting to work. So it has going for it that it captures the intention behind the sentence, in that it captures that I will not be stopped. Den doesn’t require postulating two types of existence. We don’t have to say that ‘nothing’ is yet does not exist. We do not directly refer to ‘nothing’, we just refer to what there is (and say that it is none of those things).

The difference between Ref and Den could be put like this: the former is a successful reference to something that does not exist, the latter is a failure to refer to anything which does exist.

3.   The Main Case

Let’s apply this to our example of ‘nothing has no causal powers’. Let’s rewrite having no causal powers as being ‘causally inert’, and represent that as a predicate, ‘Ix’. On the referential reading, the sentence has the form:

Ref2)   In

This says ‘nothing is causally inert’. As we have seen, the model of reference used here treats nothing as a referent of the term n, which means it is the thing referred to by n. The proposition is true only if the referent of n, i.e. nothingness, is actually causally inert. And nothingness, as conceived as an empty void with no contents whatsoever, is plausibly causally inert. So the claim seems to capture well the intention behind the apologist’s assertion here. The reason that the universe couldn’t have ‘popped into being from nothing’ is that ‘nothingness’ has no abilities to make things pop into existence. It cannot do anything; it is causally inert.

The denotative reading would be as follows:

Den2) ~(∃x)(Ix)

This says that it is not the case that there is a thing such that it is causally inert. Recast in universal terms, it says:

Den2’) (∀x)~(Ix)

This says that everything is such that it is not causally inert; everything has causal powers. On this reading, we are effectively saying that abstract objects, and similar proposed causally inert entities, do not exist; there are no abstract objects, etc. This is because abstract objects are causally inert, and Den2 says that there is no causally inert thing.

One would suppose, looking at this that in the case of nothing having no causal powers, we should take the referential reading, as this makes sense of the apologist’s claims about how the universe had to have a cause. It is clearly not their intention to assert that causally inert objects don’t exist; they mean to assert that the complete absence of anything cannot itself cause something.

In the toy example, when we distinguish the referential and denotative sense of ‘nothing’, it is clear that the referential sense is incorrect. It entails something which is clearly not intended by the speaker, that I will not get to work on time, when we meant to express that come what may I will get to work on time. In the apologetical example, the analysis seems to go the other way; the denotative sense seems to entail a proposition which clearly isn’t what the apologist intends. So, while the toy example is denotative, the apologetical example is referential.

I have two worries with this conclusion:

a) If we take the referential reading of ‘nothing’ in the phrase ‘nothing has no causal powers’, then we are referring to an entity that is of questionable ontological status. It is the referent of the term n, yet it is the complete absence of any thing. So it is a thing that does not exist. We might want to follow Russell in On Denoting, and Quine in On What There Is and disallow such talk of non-existing things. Indeed, we may consider such talk of nothing as a dubious case of reification. Nothing is not a thing of any type whatsoever.

b) This is my main worry. It seems to me that most cases of the word ‘nothing’ are denotative, and almost none are referential.

Here are a few examples:

• ‘There is nothing to split the two candidates with only days before the election.’
• ‘There is nothing I like better than ice cream’
• ‘Nothing pisses me off more than ice cream’
• ‘You mean nothing to me’
• ‘There is nothing in the fridge’

The first four cases are clearly denotative (just plug in the different readings of ‘nothing’ and see for yourself in each case). Possibly in the last example, we may want to use the referential sense, but the denotative sense seems at least as plausible. Are we expressing that there is an absence of any thing in the fridge, or that there is not any existing thing in the fridge? Neither seems preferable.

My question is: can there be an example of a sentence that uses the word ‘nothing’, and isn’t the clearly apologetical ‘nothing has no causal powers’ etc, or some other esoteric metaphysical example, for which the referential reading is clearly the correct one (and not the denotative one)?

Are there ever cases where the referential sense is the correct one, apart from the use in things like supporting the Kalam? If the answer to this question is ‘no’, then the use by the apologist is ad hoc in the support for the Kalam case. This is an open question (feel free to suggest candidate sentences in the comments section). If there is a plausible looking case, then the charge of ad hoc-ness can be deflated.