Is it absolutely certain that "bachelors are unmarried"? Are analytic propositions like these more certain than the cogito? Is this one area where philosophers throughout time have agreed unanimously without dissent?
"Bachelors are unmarried" is roughly a definition. Technically "Bachelors are unmarried men" is the definition, and then you'd have a tautology of "Unmarried men are unmarried". Or one might say "Bachelors are unmarried" forms part of the definition (with the other part being "Bachelors are men").
Definitions are not statements of truth, so cannot be false. You're just assigning a label of "bachelor" to the state of being unmarried.
Is it possible that there's some flaw in our reasoning that makes us unable to see how definitions can be false (or how me thinking somehow doesn't mean I am)? Perhaps. But it's about as close to "absolutely certain" we can get.
* This takes these words to have their current commonly-understood meanings, based on the premise that we're interested in analysing the underlying meaning, rather than analysing the words themselves. If some of their meanings change, obviously it may no longer classify as a definition, and we can't say much of anything about that then. This is merely a function of language than would apply equally to "I think, therefore I am", or any other argument.
Yes, it is possible. There are some words that have a specified meaning, e.g. the various scientific units, whose meanings are specified by the General Conference on Weights and Measures. But most words in natural languages do not have a specified meaning in this way. Their meanings just arise naturally from the way the words are used. Appealing to definitions is redundant. The compilers of dictionaries are not specifying the meaning of words: they are documenting how the speakers of a language use the words of that language.
Word meanings experience semantic drift and frequently change in meaning over time, sometimes quite radically. So at most, a sentence such as "bachelors are unmarried" is a statement that reflects current usage. It could change.
In fact, I once read an article in a magazine about the 10 richest bachelors in the UK and one of them was married. The author of the article justified his inclusion on the grounds that he was going through a divorce and he would be unmarried soon. Maybe you could say this was a mistake, or stretching the meaning, but it might catch on. Also, it is often said that a bachelor is an unmarried man, but I have heard the expression 'bachelor girl' used a few times.
The fact is that with most words, their usage answers to all kinds of requirements and these can change over time, often for empirical reasons. Even scientific terms have sometimes been revised for empirical reasons. The SI units were revised in 2019.
This is why the analytic/synthetic distinction is highly overrated. It does not represent a dichotomy but two ends of a spectrum. Analytic statements are not a distinct source of infallible knowledge.
Being married is a legal state (either legalised by the state or by your religion or both). Being a bachelor is often seen as a lifestyle. If you see it that way, you can be married, but never see your spouse in years, and live as a bachelor.
On the other hand, a couple living together without being married, doing everything together, raising children together and so on, are not bachelors even though they are unmarried.
There are at least a few interpretations of Quine's argument about the analytic/synthetic distinction. The one that I go with is:
- Take an analytic/definition sentence like, "Undead wizards are liches."
- Reformulate as a report of a stipulative definition speech act: "We have stipulatively defined 'lich' as 'undead wizard.'"
- Is the act of stipulation analytically true of "We," there? Presumably not, especially if our actions are robustly contingent.
- So every analytic-truth-by-definition can be rephrased as a synthetic report of acts of stipulation.
The other way to look at it is in terms of the trivial/nontrivial distinction. Trivial evaluations of assertoric functions use mere-identity functions, like, "If f(x) = x2, then f(2) = 22," which is less substantial than a full evaluation of the function. (Even more ethereal would be f(f(x)) = (f(x))2.) Stipulations then count as trivially justified, in that there is a meager transcendental argument for the justification of stipulative definitions as a stage-setting device in discursive presentation (it would be borderline unintelligible of us to write out the mass of our analysis if we could not decide some basic sentences as customary introductions of terms).
The bachelor example is hard to prove because usually terms are defined best based on senses data, so we say a vixen is a female fox, where fox is defined as some thing with certain properties such as fur, pointy ears which we can then prove by taking a photograph, sampling its fur.etc. but being unmarried is not something that lends itself to "the real world" well unless you count it as being something like "not having a ring on the third finger of the left hand". That is why it's hard to see when the definition does too. If we see a person with no ring on the left hand does that always show to us that they are unmarried?