Equivocation is the technical name for a logic fallacy, where an argument is made with a term which changes semantics in the course of the argument.
Equivocation in the context of information theory measures the amount of information that is contained in a random variable or other unknown quantity, given the knowledge over another random variable.
My favorite example of equivocation comes from my graduate logic professor, Dr. Johnstone (Penn State):
Hot dogs are better than nothing
Nothing is better than steak.
Therefore, hot dogs are better than steak.
The oddity of the conclusion should tell us that something has gone seriously awry with the argument - even though both premises are, on first blush, true.
What has gone wrong - "nothing" is used in two different senses in the premises. In the first premise, "nothing" means something like "nothing to eat at all," while in the second premise "nothing" means something like "no possible food choice on the planet." It is the slippage from the one sense to the second that allows for the peculiar conclusion - but this slippage rests on equivocation in the meaning of "nothing."
A more serious example is suggested by Mary Anne Warren, in her analysis of a familiar anti-abortion argument:
1) It is wrong to kill innocent human beings
2) Fetuses are innocent human beings
Therefore, it is wrong to kill fetuses.
Warren argues that, in fact, this argument faces its own dilemma: either it is guilty of equivocation - or it is guilty of question-begging.
It is easy enough to argue that "innocent human beings" means two different things in the premises - and thus the argument equivocates. For example:
innocent human being in premise 1 = "conscious of moral choice, but not guilty of committing/choosing an immoral act"
innocent human being in premise 2 = "innocent because the fetus is not capable of moral intentions and choices in the first place"
IF two different senses of "innocent human being" are thus at work in the argument - then the argument equivocates: while we might agree that it is wrong to kill innocent human beings in the first sense - this is not immediately relevant to fetuses as innocent human beings in the second sense, and so the conclusion does not follow.
[Alternatively, if we are not equivocating - if we somehow mean "innocent human beings" in the exact same sense in both premises - then the argument begs the question. That is, the premises taken together would then mean:
it is wrong to kill innocent human beings, where "innocent human beings" includes fetuses.
BUT: this is precisely the conclusion we're seeking to establish (or, the question under discussion). Our argument in this version then amounts to
[P] it is wrong to kill innocent human beings, where "innocent human beings" includes fetuses
[C] therefore it is wrong to kill fetuses [as innocent human beings]
The argument thus goes in a circle, as the premises already include everything we're attempting to assert in the conclusion - alternatively, the premises offer no additional argument, evidence, or information in support of the conclusion. Rather, the premises merely restate the conclusion, and the conclusion merely restates the premises. Or: the question, "Is it wrong to kill fetuses?" is begged as the answer to the question is merely assumed in the premises.
A still more serious example of equivocation comes in the debate among religious communities over abortion. See Judaism vs. Christianity: Hebrew vs. Greek texts as the source of divergent teachings on abortion