Does the Greek word for “destroy” – apollumi – really mean destroy in the strong sense that annihilationists think it does? Short answer: yes.
One of the key arguments for annihilationism is the fact that the biblical writers frequently claim that those who are not saved in the end will be destroyed. Why this appears to support annihilationism is fairly self-evident. It’s important to stress that this argument does not only rest on the fact that the word “destruction” or “destroy” is used. The biblical writers, like Jesus, sometimes describe destruction without using that specific word. Images of weeds burned up in a furnace or the cities of Sodom and Gomorrah being utterly consumed by fire also serve this purpose. But just now let’s look specifically at the term “destroy.”
It’s common to see those who deny annihilationism claiming that the word “destroy,” when used to refer to the fate of the lost, does not mean destroy in the strong sense of literally kill or wipe out. It means, they claim, something less specific, like “perish,” become “lost,” become “ruined” or something else. Hence, the argument goes, in those texts that refer to final punishment using this term, the meaning is not literal destruction at all, but rather ruination; a state of conscious but miserable existence.
Although the argument ultimately fails, it’s important to note the grain of truth on which it is based. The relevant Greek word for destroy – apollumi – is part of the apoleia word group. It is true that, as Don Carson pointed out, “the apoleia word group has a range of meanings, depending on context.” It can be used passively, for example, to refer to a “lost” son or a “lost” coin. It can be used of non-living things like burst wineskins that are “ruined” (although this may not be a good counterexample, because if I was burst like a wineskin I would obviously die in a literal sense). But we should grant the point: The wider word group does have some range of meaning and does not always imply exactly the same thing.
However it is also important to be wary of what James Barr calls the “illegitimate totality transfer,” interestingly, a fallacy that Carson also warns us about. This is the error of assuming that whenever a word appears, it carries with it its complete semantic range in a way warrants us picking from any of its possible meanings, as though all of those meanings were equally likely in any given context.
An obvious example of this fallacy in English would be to say something like: “The interrogating officer said he was going to grill the suspect. In this cookbook, grilling is defined as cooking something with direct heat, so that poor suspect is about to be cooked!” Obviously this is not what the interrogator meant.
Our effort to determine what a word actually means in any given context must be informed by all the relevant contextual information. One such piece of information is how we know an author uses the same word in similar contexts elsewhere. If one of the Gospel writers, for example, always uses a word in a specific way in a certain kind of context, we would be unwise to pick an example of them using that same word in the same sort of context, but insist that on this one special occasion they mean something quite different – because the word is technically capable of meaning more than one thing. For example if a man frequently refers to his mother in law as “the dragon,” and is in the habit of telling people, before she arrives, that “the dragon is coming to visit,” it would be unjustified to hear him say that for the hundredth time and suppose that this time, unlike all other times, he means that a scaled monster is about to swoop down from the sky. This would be a case of special pleading, because we know that he didn’t mean that any other time.
The fallacious reasoning that I have in mind falls under the general category of fallacies known as “word study fallacies.” This is a common error of those new to biblical studies, where one supposes (incorrectly) that the way to figure out what a word means when an author uses it is to grab your Greek or Hebrew concordance and do a word study – to look up all the occurrence of that word or lexically related words – and to write up a list of all the meanings that it has across all the different contexts in which it appears, and to imagine that in the example that you’re looking at, all of these nuances of meaning are available as possible explanations of what the author means. But this is simply not the case.
So much for setting the scene, now let’s get to the meat of the issue (and no, I do not mean that we’re about to have dinner – to think that would be to make use of the fallacy that I’m discussing here). The biblical writers often say that God will destroy the lost one day. A good example of this is Jesus’ teaching in Matthew 10:28, which reads: “Do not fear those who can kill the body but not the soul. Rather, fear the one who can destroy both body and soul in hell.” In avoiding the annihilationist implications of this saying, the standard argument is to say that since the apoleia word group includes meanings other that destroy in the literal sense, we are justified in believing that apollumi may have one of those other meanings here. It is true, the word can indicate “ruin,” or when used in the passive voice it can mean “perish” (where a person is the object of the action rather than the subject). A similar thing is true of the middle voice, where a person is the object of a verb, and yet there is no other person who is clearly the subject. A good example is in Matthew 8:25 where the disciples are on a boat in a storm and they cry out to Jesus, “we perish” (apollumetha). This sort of counterexample would not necessarily work against annihilationism, since in many cases the perishing does indeed refer to literal death, which is precisely what the disciples feared.
However, one of our goals as interpreters of Matthew 10:28 should be to find out whether or not there is a clear pattern of how the term apollumi in relevantly similar contexts – the more similar the better. If we limit ourselves to the use of this term in the active voice (as it appears in Matthew 10:28), in the Synoptic Gospels (so that we do not stray into writings where significantly different rends of usage might appear), where the word in context clearly refers to the action of one person or agent against another, we will be able to see if there is a clear pattern that should inform our understanding of the term as it appears in Matthew 10:28, and we will hopefully be filtering out cases that have important differences from this instance of the word, which will help to safeguard us against the temptation to engage in illegitimate totality transfer. It may not be a perfect safeguard, and it may be that we will find examples that show a range of meaning even within this strictly limited sample that will require further work, but it is a start.
However, once we apply such limiting criteria, what we immediately find is that the range of meaning that was present in the entire apoleia word group is now filtered out entirely, and one clear emphasis of meaning remains. This is because in every single instance of the word apollumi where these criteria are met – The example is in the Synoptic Gospels, the active voice is used and the word clearly refers to the actions of one person or agent against another, the term apollumi – setting aside Matthew 10:28 – always refers to the literal killing of a person, with not a single exception. I will list just seven representative examples, but the reader is encouraged to check this for themselves:
- In Matthew 2:13, Herod wants to kill the baby Jesus.
- In Matthew 12:14 the Pharisees conspired together about how they might kill Jesus.
- In Matthew 21:41 (story of the wicked tenants) the vineyard owner kills the wicked tenants.
- In Matthew 27:20, the elders and chief priests urge the people to have Barabbas released and Jesus killed.
- In Mark 3:6, the Pharisees plot to kill Jesus.
- In Mark 9:22, the parents of a boy with an unclean spirit tell Jesus that the spirit often throws the boy into water or into a fire, trying to kill him.
- In Luke 6:9, Jesus asks if it is lawful on the Sabbath to save life or kill.
In each and every other instance where all these criteria are met, the meaning is the same. There literally is no semantic range in these cases. Some claims in biblical interpretation are matters of opinion and open to question, but this is not one of them. This is a feature of the raw data itself – what we think it implies however may be questioned. But at minimum, it is clear that to take a meaning that arises from a significantly different usage of apollumi – a different voice, or a different body of literature, or a different context (e.g. where we are no longer looking at the actions of one person or agent against another), and to insist that we should attribute that meaning to a use of the word that conforms to the pattern described here, at very least requires a very robust defence. The mere fact that the wider apoleia word group is capable of expressing such meanings under different conditions (e.g. ruin, lose etc) cannot be the reason that we should find that meaning in Matthew 10:28, for this would be a perfect example of the illegitimate totality transfer. However theologically inconvenient it may be for defenders of the traditional doctrine of the eternal torments of hell, this is an instance where the exegetical evidence is very heavily against them, and there is no apparent escape route via an appeal to semantics.
The meaning of “apollumi” in the Synoptic Gospels
One of the key arguments for annihilationism is the fact that the biblical writers frequently claim that those who are not saved in the end will be destroyed. Why this appears to support annihilationism is fairly self-evident. It’s important to stress that this argument does not only rest on the fact that the word “destruction” or “destroy” is used. The biblical writers, like Jesus, sometimes describe destruction without using that specific word. Images of weeds burned up in a furnace or the cities of Sodom and Gomorrah being utterly consumed by fire also serve this purpose. But just now let’s look specifically at the term “destroy.”
It’s common to see those who deny annihilationism claiming that the word “destroy,” when used to refer to the fate of the lost, does not mean destroy in the strong sense of literally kill or wipe out. It means, they claim, something less specific, like “perish,” become “lost,” become “ruined” or something else. Hence, the argument goes, in those texts that refer to final punishment using this term, the meaning is not literal destruction at all, but rather ruination; a state of conscious but miserable existence.
Although the argument ultimately fails, it’s important to note the grain of truth on which it is based. The relevant Greek word for destroy – apollumi – is part of the apoleia word group. It is true that, as Don Carson pointed out, “the apoleia word group has a range of meanings, depending on context.”1Don Carson, The Gagging of God: Christianity Confronts Pluralism (Leicester: Apollos, 1996), 522. It can be used passively, for example, to refer to a “lost” son or a “lost” coin. It can be used of non-living things like burst wineskins that are “ruined” (although this may not be a good counterexample, because if I was burst like a wineskin I would obviously die in a literal sense). But we should grant the point: The wider word group does have some range of meaning and does not always imply exactly the same thing.
However it is also important to be wary of what James Barr calls the “illegitimate totality transfer,” interestingly, a fallacy that Carson also warns us about.2James Barr, The Semantics of Biblical Language (Glasgow: Oxford University Press, 1961), 218. See also Carson, Exegetical Fallacies (Grand Rapids: Baker, 1984), 62 This is the error of assuming that whenever a word appears, it carries with it its complete semantic range in a way warrants us picking from any of its possible meanings, as though all of those meanings were equally likely in any given context.
An obvious example of this fallacy in English would be to say something like: “The interrogating officer said he was going to grill the suspect. In this cookbook, grilling is defined as cooking something with direct heat, so that poor suspect is about to be cooked!” Obviously this is not what the interrogator meant.
Our effort to determine what a word actually means in any given context must be informed by all the relevant contextual information. One such piece of information is how we know an author uses the same word in similar contexts elsewhere. If one of the Gospel writers, for example, always uses a word in a specific way in a certain kind of context, we would be unwise to pick an example of them using that same word in the same sort of context, but insist that on this one special occasion they mean something quite different – because the word is technically capable of meaning more than one thing. For example if a man frequently refers to his mother in law as “the dragon,” and is in the habit of telling people, before she arrives, that “the dragon is coming to visit,” it would be unjustified to hear him say that for the hundredth time and suppose that this time, unlike all other times, he means that a scaled monster is about to swoop down from the sky. This would be a case of special pleading, because we know that he didn’t mean that any other time.
The fallacious reasoning that I have in mind falls under the general category of fallacies known as “word study fallacies.” This is a common error of those new to biblical studies, where one supposes (incorrectly) that the way to figure out what a word means when an author uses it is to grab your Greek or Hebrew concordance and do a word study – to look up all the occurrence of that word or lexically related words – and to write up a list of all the meanings that it has across all the different contexts in which it appears, and to imagine that in the example that you’re looking at, all of these nuances of meaning are available as possible explanations of what the author means. But this is simply not the case.
So much for setting the scene, now let’s get to the meat of the issue (and no, I do not mean that we’re about to have dinner – to think that would be to make use of the fallacy that I’m discussing here). The biblical writers often say that God will destroy the lost one day. A good example of this is Jesus’ teaching in Matthew 10:28, which reads: “Do not fear those who can kill the body but not the soul. Rather, fear the one who can destroy both body and soul in hell.” In avoiding the annihilationist implications of this saying, the standard argument is to say that since the apoleia word group includes meanings other that destroy in the literal sense, we are justified in believing that apollumi may have one of those other meanings here. It is true, the word can indicate “ruin,” or when used in the passive voice it can mean “perish” (where a person is the object of the action rather than the subject). A similar thing is true of the middle voice, where a person is the object of a verb, and yet there is no other person who is clearly the subject. A good example is in Matthew 8:25 where the disciples are on a boat in a storm and they cry out to Jesus, “we perish” (apollumetha). This sort of counterexample would not necessarily work against annihilationism, since in many cases the perishing does indeed refer to literal death, which is precisely what the disciples feared.
However, one of our goals as interpreters of Matthew 10:28 should be to find out whether or not there is a clear pattern of how the term apollumi in relevantly similar contexts – the more similar the better. If we limit ourselves to the use of this term in the active voice (as it appears in Matthew 10:28), in the Synoptic Gospels (so that we do not stray into writings where significantly different rends of usage might appear), where the word in context clearly refers to the action of one person or agent against another, we will be able to see if there is a clear pattern that should inform our understanding of the term as it appears in Matthew 10:28, and we will hopefully be filtering out cases that have important differences from this instance of the word, which will help to safeguard us against the temptation to engage in illegitimate totality transfer. It may not be a perfect safeguard, and it may be that we will find examples that show a range of meaning even within this strictly limited sample that will require further work, but it is a start.
However, once we apply such limiting criteria, what we immediately find is that the range of meaning that was present in the entire apoleia word group is now filtered out entirely, and one clear emphasis of meaning remains. This is because in every single instance of the word apollumi where these criteria are met – The example is in the Synoptic Gospels, the active voice is used and the word clearly refers to the actions of one person or agent against another, the term apollumi – setting aside Matthew 10:28 – always refers to the literal killing of a person, with not a single exception. I will list just seven representative examples, but the reader is encouraged to check this for themselves:
In each and every other instance where all these criteria are met, the meaning is the same. There literally is no semantic range in these cases. Some claims in biblical interpretation are matters of opinion and open to question, but this is not one of them. This is a feature of the raw data itself – what we think it implies however may be questioned. But at minimum, it is clear that to take a meaning that arises from a significantly different usage of apollumi – a different voice, or a different body of literature, or a different context (e.g. where we are no longer looking at the actions of one person or agent against another), and to insist that we should attribute that meaning to a use of the word that conforms to the pattern described here, at very least requires a very robust defence. The mere fact that the wider apoleia word group is capable of expressing such meanings under different conditions (e.g. ruin, lose etc) cannot be the reason that we should find that meaning in Matthew 10:28, for this would be a perfect example of the illegitimate totality transfer. However theologically inconvenient it may be for defenders of the traditional doctrine of the eternal torments of hell, this is an instance where the exegetical evidence is very heavily against them, and there is no apparent escape route via an appeal to semantics.
[+]
Glenn Peoples