*Totally revised 6/25/2018
Aristotle discussed logical fallacies in three places–On Sophistical Refutations (SE), Prior Analytics, Rhetoric. His approach to them is quite from the typical approach nowadays (to the extent that one can say there is an approach at all nowadays). The major difference is that, in the SE, he’s not concerned with the employment of fallacies in faulty reasoning, but rather with their intentional deployment in a formalized argument contest. The Sophistical Refutations is providing instruction in their use.
Nonetheless, it is true that Aristotle discovered (or invented) a lot of what now end up in lists of fallacies. Here again, however, they are sometimes different from what you encounter in a standard textbook treatment. In some cases, such as “non causa pro causa,” they merely share a name. Others, such as “accent,” are completely obscure to us and the examples Aristotle gives are untranslatable.
Aristotle defined a fallacy (which he called a “paralogism” in the SE) as a refutation or deduction that seems good but isn’t. Again, the idea that it’s a refutation tells us that he’s talking about a particular kind of logical game where one attempted to refute an opponent. There’s something of an analogue to this in the medieval obligation game (read about them here or listen to a podcast here). Another analogy might be Twitter (which too often seems like a contest of zingers).
Here below is a brief list based on the SE. This is (obviously) hardly a complete treatment. My point is rather primarily to signal where Aristotle’s meaning and ours diverges.
Ever fond of classification, Aristotle broke logical fallacies into two broad classes: fallacies within language (those that depend on verbal or grammatical ambiguity) and fallacies outside of language.
I. Fallacies dependent on language (De Soph Elen 4, 165b24-166b28)
1. Equivocation: This is when the same word has multiple distinct senses.
2. Amphiboly (or ambiguity). This is like amphiboly but it involves syntax having multiple senses—for instance: “I wish you the enemy may capture.” This has got a yoda-like syntax to it, but I think it gets the point across.
3. Combination: This is not the parts/wholes composition of contemporary use. Rather, Aristotle means it when words are understood in combination in a sentence. Here’s Aristotle’s example: “a man can walk while sitting and can write while not writing.” If you read it one way, it’s a contradiction, in another, it’s not.
4. Division: This is not division in the contemporary sense, but rather when words in a sentence are understood as divided. Aristotle’s example is very obscure: “upon division depend the proposition that 5 is 2 and 3.” What he means is that five is not both 2 and 3, but rather the combination.
5. Accent: the sense of “accent” here actually refers actually to the accent of words (like ancient Greek words which are accented for pitch and change meaning accordingly) so it’s hard to reproduce this in English.
6. Form of expression: He gives one example: flourishing is like “cutting” but one is a quality the other is an action.
II. Fallacies outside of language (De Soph Elen 5, 166b28-168a18):
7. Accident: mixing up the essence of something with an accident (non-essential property). Aristotle’s example: “Corsicus is different from Socrates, Socrates is a man, so Corsicus is different from a man.”
8. The use of words absolutely or in a certain respect (often called secundum quid): This is something like “hasty generalization” but not quite.
9. Misconception of refutation (ignoratio elenchi): this is kind of catch-all category for general irrelevance. Aristotle even says it’s the only real category of fallacy at one point in the SE.
10. Begging the question: assuming the thing to be proved.
11. Consequent: this is something like affirming the consequent (if A then B, B, so A) but Aristotle didn’t frame it that way (using the “if-then”).
12. Non causa pro causa: This is completely different from the “false cause” (an expression that didn’t appear for 2000 or so years) you often meet in textbooks. It has to do rather with a particular kind of argument technique known as reductio ad falsum (or impossibile).
13. Complex question: Asking two question disguised as one (e.g., when did you stop beating your spouse?).
Those are the 13. What you don’t see: any argument ad. Those kinds of fallacies (ad hominem, ad ignorantiam, ad verecundiam, and juidicium) didn’t show up until Locke’s Essay on Human Understanding (1689) and even then he didn’t consider them to be fallacies (and he understands them in a completely different sense). Missing also are false dichotomy and the straw man, which would be a part of most contemporary lists.