[I should warn you that this section of the site is incomplete and will be so for a while. I apologize for when I trail off.]

There are a lot of arguments bandied about as rationales for any given grammar prescription. Most of them are spurious, but a few have some merit. I’m listing them here so that I can refer to them in various posts to support or undermine prescriptivists’ arguments. After all, the best way to refute a prescriptivist’s argument is to simultaneously show them that their argument is ill-founded, but even if they persist with that ill-founded argument, they’re still wrong.

#1: Usage From Years Past
Perhaps the most common claim of prescriptivists is that they’re battening the hatches of our language against the ravages of our ill-educated age. This claim comes through in arguments that a certain form should be preferred because it is what the great writers of years past used. Oh, it’s a seductive mistress, this argument. Wouldn’t we all like to write like those great names of yore? Who doesn’t fancy himself an Alexander Pope or herself an Emily Bronte? But of course this isn’t a solid argument for what our modern language ought to be. For instance, if we went back to
Early Modern English, there would be a much more flexible word order: “as the law should them direct” (The Early Records of the Town of Providence, 1896) and “ye” would be an acceptable second person singular pronoun (e.g., “Hear ye, hear ye”). Another example is Strunk’s admonition of 1918 that “a conditional statement in the first person requires should, not would“. His example, I should not have succeeded without his help does not mean the same thing as I would not have succeeded without his help in contemporary English. Times change, usage changes. Sorry.
Summary: simple, seductive, and wrong

#2: Appeals to Logic
A lot of prescriptivists are enamoured of the idea that language is at its core a delightful, logical system. Take for example Brad of h3h’s argument that different than is “illogical”, and only different from should be used. The idea of such an argument is simple: (illustrated with the h3h examples)

1. constructions X and Y have the same meaning.
2. there is a construction Y’ with the same meaning and usage patterns as Y but some syntactic difference Z.
3. there is no construction X’ with the same meaning and usage patterns as X and the syntactic difference Z.
4. Therefore: since X and Y are equivalent, if X’ is ungrammatical, then Y’ must be ungrammatical as well

To illustrate this, consider Brad’s argument for the supremacy of different from. Let X be differs from and Y be is different from. He claims that (1) is met because John differs from Bill is equivalent to John is different from Bill. Now let Y’ be is different than, which is the same as Y but with than replacing from (this is the syntactic difference Z). Y and Y’ are equivalent, so (2) is met. However, if we replace from in differs from by than, we get *differs than, which is not grammatical, so (3) is satisfied. With (1)-(3) satisfied, the claim is that (4) must logically follow.
Which, of course, it does not. There are a number of flaws with this argument, which I will address at length in the future. But quickly:

1. language is not a logical system
2. no two constructions are perfectly equivalent
3. there is no reason to suspect that two equivalent constructions would undergo the same transformations

Summary: Misguided. Language isn’t math.

#3: Ambiguity Avoidance
[I'll fill this one in later when I get around to arguing against a specific usage problem.]

#4: Avoiding Overuse
Now here’s a relatively solid argument. Some rules of grammar are created because if youngsters don’t have them, they’ll be stylistically awful. I remember as a kid having it drilled into my head that you should never never never start a sentence with a conjunction. And I remember seeing red pen on sentences just like this one for daring to break this immutable rule that I now break regularly. But that’s sort of the point. Kids will start every sentence with a conjunction if you don’t stop them - because that’s what we do when speaking. (Just try to note all of the wells, anyways, ands, sos, buts, and the like used between sentences in your next conversation!) Adults, on the other hand, develop a sense of times when a sentence justifies being started with a conjunction, and should be willing to break this rule in those situations.

Likewise, one could argue that injunctions against passive voice, using first-person pronouns, semicolons, or using very are intended (at least in part) to prevent overuse by the stylistically-challenged. The problem is that too many people have reinterpreted these as inviolable rules and claim that these should never be used, not even sparingly.

Summary: These have their place, in moderation.

#5: Omit Needless Words
One of the catchiest three-word phrases in editing, this was Strunk’s mantra. Its idea is sound; a longer sentence is generally harder to understand than a shorter one, especially when the words have little semantic contribution. One example where Strunk is on the right track is his suggestion to replace there is no doubt but that with doubtless. But there is a problem in that it’s unclear what exactly a needless word is. If needless words are those that are not strictly necessary for grammaticality, that’s problematic. Consider a relative clause starting with that was, as in (1). Officially, such clauses in English do not need the that was, as shown in (1):

(1) The car (that was) driven past the barn crashed.
(2a) The horse that was raced past the barn fell.
(2b) ??The horse raced past the barn fell.

(1) is fine whether that was is in it or not, but (2b) is horrendous without these technically “needless” words. As it turns out, there are lots of cases where words unnecessary for grammaticality are essentially necessary for the sentence to be readable. More on this later.

Summary: Omit only words that are both truly superfluous and get in the way of comprehension.