The Etymological Fallacy

Alias: Abuse of Etymology

Taxonomy: Logical Fallacy > Informal Fallacy > Red Herring > Genetic Fallacy > Etymological Fallacy


Formal logic fails us because of its assumptions. The postulates from which the mechanism springs are normally abstractions of a high order, words rather than things. The finest of automobiles will not run on a road of air; it must have solid ground under the wheels. The Greeks, with their assumption that words were real things, naturally enough soared into rarefied regions. Human thinking has been short of oxygen ever since. … "Logos" is Greek for "word"; "logic" is the manipulation of words.

Source: Stuart Chase, The Tyranny of Words (1938), Chapter 13, p. 226.



The etymology of a word is an account of its historical derivation from older words often from a different language. An older, usually archaic, word from which a current word is historically derived is called its "etymon". The term "etymological fallacy" is applied to two types of error:

  1. Semantic: The etymological fallacy as a semantic error is the mistake of confusing the current meaning of a word with the meaning of one of its etymons, or of considering the meaning of the etymon to be the "real" or "true" meaning of the current word. If one's goal is to communicate, then the "real" or "true" meaning of a word is its current meaning. Since the meanings of words change over time, often considerably, the meaning of an etymon may be very different from the current meaning of the word derived from it. The fact that a word historically derives from an etymon may be interesting, but it cannot tell us the current meaning of the word.

    For instance, the English word "decimate" means "to destroy a large portion of", but is derived from a Latin word having to do with a tenth of something―see the Source, below, for more on its etymology. Some people who know this insist that "decimate", therefore, means to destroy exactly one-tenth, rather than a large part, and may claim that as its "real" or "true" meaning. However, current English is not Latin.

    Source: Ammon Shea, "Does ‘decimate’ really mean ‘destroy one tenth’?", Oxford Dictionaries, 9/10/2012

  2. Logical: The etymological fallacy as a logical mistake results when one reasons about the etymon as if the conclusion applied to the current word. This is a logical error similar to equivocation, which involves confusing two meanings of the same word; but it differs from equivocation in that the etymological fallacy involves the meanings of two different words, though those words are historically connected.


Robert J. Gula, Nonsense: A Handbook of Logical Fallacies (2002), pp. 48 & 161.


The Etymological Fallacy, Fallacy Files Weblog, 6/12/2006

Analysis of the Example:

In the chapter from which this passage comes, Chase is arguing that formal logic is of little or no value because it involves manipulating words unconnected with reality. The passage commits a fallacy and sets up a logical boobytrap:

  1. The Greek word "logos" is actually highly ambiguous, having the following meanings in addition to "word": speech, argument, explanation, principle, reason, among others. But Chase mentions only one of these meanings, namely, the one which supports his argument, rather than others which undermine it―such as "argument" and "reason", which have an obvious bearing on the meaning of the English word "logic". However, assuming that "logos" really only meant "word", would that support his claim that formal logic is simply the manipulation of words? The suffix "-logy" which occurs in such English words as "biology", "psychology", and even "etymology", has the same source as "logic". Does this mean that every science ending in that suffix is really just words about its subject matter, unconnected to reality? Of course not. Just as "biology" came to mean the science of life, so "logic" has come to mean the scientific study of reasoning. Thus, Chase has fallaciously appealed to etymology to support his conclusion.
  2. Chase's analogy comparing formal logic to an automobile is very weak; hadn't he ever heard of airplanes? An alternative analogy would compare formal logic to a plane which can travel much faster than a car by lifting off of the ground and soaring into the atmosphere of abstraction. Of course, there may still be a danger of flying so high that the air becomes too thin, but what is Chase's evidence that this has happened to formal logic? Perhaps the analogy as Chase gives it should be taken only as a colorful illustration of his point, rather than as a reason in its favor, so that it doesn't constitute a fallacious analogy. However, it is at least a boobytrap that might trip up a naive or unimaginative reasoner.

Source: G. B. Kerferd, "Logos" in The Encyclopedia of Philosophy, Paul Edwards, Editor in Chief (1972), Volume 5, pp. 83-84.

Reader Response:

A reader known only as "Huns" has a problem with this entry. Huns' email is on the long side, and raises distinct issues that require different answers, so I will insert my replies on a white background in between Huns' comments on a yellow background. Take it away, Huns:

As I see it, stating that the currently accepted meaning of a word overrides its etymology incurs two problems. First, there is the abuse of language. For example, the word "homophobia" was originally coined to describe a person who was afraid of homosexuality, or of becoming a homosexual. It was internally consistent; we can pull the meaning of "homo" from the context (sexuality rather than human- or self-relation), and we also know what a "phobia" is because we've heard of dozens of phobias―and they are all about fear. Unfortunately, a journalist deliberately misapplied the term to people whose aversion is moral or aesthetic rather than fear-based. After that, more people followed suit, and the word became a generalized slur. In other words, as presented, the fallacy gives anyone a free pass to change the meaning of a word―deliberately or accidentally―as long as they were the first to successfully hijack it.

You're making two main points here: firstly, that I have an overly-permissive attitude to semantic change which, secondly, can lead to the kind of abuse illustrated by the word "homophobe". I'll address the second of these points later.

If you think about it, my view could be considered too restrictive rather than overly permissive. According to your account of the word "homophobe", there was a time when "homophobe" meant "a person with a phobia towards homosexuals or homosexuality". Then, the journalist that you referred to "hijacked" the word by giving it a new meaning. Specifically, what happened was a low redefinition of the word "homophobe", which was redefined to apply to anyone with a negative attitude towards homosexuality, including many who are not phobic.

The operative word in your description is "successful" as applied to "hijacking", that is, anyone who sets out to redefine a word must get enough people to use the word in its new sense that it becomes the meaning for the consensus of speakers of the language. This is a very high hurdle to jump, and it's likely that most attempts to "hijack" words fall short.

You may be right that the word "homophobe" has been hijacked, but I'm not so sure. There's no doubt that some people do use the word as you describe, but I'm not convinced that this is now the accepted meaning of the word. If not, the word has not yet been successfully hijacked. Moreover, you are free to resist hijacking, since you have as much input to the consensus as any other user of the language.

I'm sympathetic to your complaint that the meaning of "homophobe" has become divorced from the meanings of its prefix―"homo-"―and root―"phobe"―both of which come from ancient Greek. Compositionality―that is, the characteristic in which the meaning of something is a function of the meanings of its parts―is an important characteristic of language, making it possible to figure out the meaning of a new word from the meanings of its parts. Unfortunately, sometimes words lose their compositionality―for instance, the word "decimate", mentioned in the original entry above, comes from the Latin word for "ten". Other English words derived from the same source, such as "decimal", have retained a meaning close to the Latin meaning. In contrast, the month "December" also comes from the same root and was so called because it was the tenth month in the early Roman calendar. Should we change the name of October to "December" for that reason?

The lack of compositionality of the broad meaning of "homophobe" is a good reason to resist its hijacking, but it would be silly to insist that August should really be called "October"―the eighth month, from the Latin for "eight", also found in "octopus", an eight-legged animal.

If your goal is to communicate with people there's not much alternative to recognizing that the meaning of words in a public language is established by a consensus of its speakers. You can use a word in an idiosyncratic way on occasion, so long as you explain what you mean by it, but you couldn't do so in a wholesale way and still be understood. Language is, in a sense, essentially democratic. We can't, like Humpty Dumpty in Through the Looking-Glass, mean just whatever we want by the words we use.

Consequently, anyone wishing to shout down someone with an aversion to homosexuality now has an abusive term to use: "homophobe." The term, the meaning of which was utterly transparent before, has now become a loaded word. The implication that the target is afraid is used as a barb to upset them, as human psychology associates fear with weakness. If the person is actually afraid, it's at least accurate, but to impugn someone's personal taste as "fearful" is both inaccurate and unfair. At the same time, the accuser has a foothold for invalidating the psychological response they deliberately provoked, smarmily stating that "words change over time", they get to have their cake and eat it, too.

As a person who often takes up homophobes (and people who are homo-averse for reasons other than fear) on their arguments, I don't see this as a good thing. While I'm on the side that might seem to benefit from the use of this word as a bludgeon, I don't see it as honest. I see it as unnecessarily provocative, something to punish and badger the opposition with and keep the debate emotional when rationality would be far better. It makes debates take longer than they need to and it invites criticism from the homo-averse, who can hold up this very reasoning as evidence that the opposition is badgering them. So, not only is it unfair to my opponent―which makes it illegitimate for me to use―but it also potentially shoots me in the foot the minute I use it.

I certainly agree with you that "homophobe" is a loaded term―see the entry for Loaded Words―specifically, it has a negative emotive charge for the reasons that you suggest. However, you seem to be conflating two distinct aspects of words here: the literal meaning of a word and its emotive "charge". Even if the literal meaning of "homophobe" has broadened, it's still a highly-negative word. Recognizing the literal meaning of a word is not a license to ignore or exploit its emotive charge in the way you describe. For instance, consider any negative epithet for a racial or ethnic group: the fact that the literal meaning of the word may simply be any member of that group would not justify ignoring its negative force. Speaking of loaded language, I hope I wasn't "smarmy" in the entry, above.

The second problem is that "as a term is currently used" itself implies criteria without defining them, and implicitly appeals to popularity. If 75% of the people are using a definition that's been deliberately loaded, and 25% have rejected the loaded definition and prefer the original, who prevails and why? Would it be different if it was 60%/40% or 50%/50%?

As with all other informal fallacies, it isn't always fallacious to appeal to popularity. Language is a special case because we use it primarily to communicate with other people. However, this very fact means that we must also use words with sensitivity to their emotive overtones, the baggage that comes from their history, and the meanings of their parts.

How do we know when there is a consensus? Semantic change is usually a gradual process, and there's no precise point at which a word changes meaning, any more than there's a precise point at which a person ceases to be young and suddenly becomes old. Instead, there's a grey zone―no pun intended!―between youth and old age, and similarly between what "homophobe" used to mean and whatever it ends up meaning. I suspect that "homophobe" is still in the grey zone, but I'll leave it to the linguists and lexicographers to decide what the consensus is.

I would therefore propose that the etymological fallacy should include some clause to acknowledge the phenomenon of deliberately hijacking words, deliberately loading them up on meanings designed to incite emotional reactions, etc., as these tactics are used fallaciously.

In lieu of such a clause, I hope that this exchange will serve.

Acknowledgments: Thanks to Topher Cooper for suggesting the fallacy, and to Emil William Kirkegaard for pointing out an error in the analysis of the example.