On October 4th, I wrote about language, covering the concept that languages are essentially arbitrary. I wanted to reiterate this fact, and elaborate on it in a slightly different way.
All methods of communication are, at their base, arbitrary. There is no law of physics that requires that the number that comes after "one" be called "two". However, given that the purpose of language is to communicate effectively, some forms of language are more efficient than others. For example, there is no particular reason why every word in the English language could not be preceded by xxxthree xxxcopies xxxof xxxthe xxxletter xxx"x". But it's not a useful thing to do, since it adds no information to a given communication, and makes it harder to read.
A less absurd example, which came up during recent /. article, might be the following question: Should you use an apostrophe when pluralizing acronyms or certain abbreviations? For example, should the plural of CD be "CDs" or "CD's"? Well, there's several issues to consider, so let's explore them.
The normal pluralization of words in English is to simply add an "s" to the end, or an "es" if the word already ends in "s". For example, "car" becomes "cars" and "pass" becomes "passes". There are, of course, numerous exceptions -- for example, "bus" becomes "busses" (doubling the final "s" before adding the "es"), "ox" becomes "oxen" (adding "en" instead of "s" or "es"), "sheep" stays "sheep" (singular and plural are the same), and so forth.
The apostrophe is normally used in English to denote possession and abbreviation (most commonly, contractions). Possession is usually handled by adding "'s" (an apostrophe and an "s") to the word -- so "Jack's car" means "the car possessed (or owned) by Jack". Contractions are usually handled by removing part of a word, and substituting the apostrophe -- for example, "cannot" replaces the "no" with an apostrophe, becoming "can't", and "does not" removes the space and the "o" to become "doesn't". There are, as always, exceptions to this -- the commonly reviled contraction "ain't", for example, substitutes for "are not", "am not", and "is not", unlike most contractions, which only substitute for one word or phrase.
(Note the common misuse of the words "it's" and "its". "It's" (with an apostrophe) is always a contraction of "it is", and is never the possessive form of "it". "Its" (no apostrophe) is the possessive form of "it" -- it goes right next to "his", "hers", "theirs", and so on.)
If you're talking about an attribute of a CD -- for example, the contents of the CD -- it's entirely appropriate to use an apostrophe, since you're referring to the possessive -- the contents of the CD are the CD's contents. If you're talking about multiple compact discs, then it's reasonable to simply append an "s" to "CD", and you get CDs. But is this always the best answer? Imagine if you were talking about the grades you got this semester. Let's say you did really well, and in each class you got an A. If you were talking about your grades, you might write that you got all As! But that looks funny... kind of like the word "as". So maybe writing that you got all A's would be a little more readable. This defies the usual convention -- using an apostrophe to pluralize. However, "A" isn't an acronym the way that "CD" is -- but it is an abbreviation. Saying that you got "an A" in a class is really shorthand for saying that you got "an A grade" or "a grade of A". So if you got all A grades, perhaps it's reasonable to abbreviate "A grades" as "A's", since "A's" is functioning as a kind of contraction, rather than a pluralization. The pluralization here is from "A grade" to "A grades"; and it's the latter form that's contracted into "A's" (where the removed letters are the space between "A" and "grades" and most of the word "grades"). So "A's" wins, slightly, on readability, and (in a bizarre, roundabout way) makes sense with regard to contraction. I consider the apostrophe acceptable for pluralizing "A's", as long as the context is clear so that you don't mistake it for the possessive (after all, you can say that the influence Bob's single A grade had on him -- AKA, the A's influence -- is what made him want to go into grad school, for example).
Here's another twisty one: What about decades? Sometimes people refer to the 1990s, and sometimes they refer to the 1990's (also the 90s and 90's). Now, 1990 was the name of a particular year -- the one that came after 1989 and before 1991. You can normally pluralize proper names, if you're referring to multiple items with the name name (for example, if you have two people named John, you might refer to them as "the Johns"). But you can't have more than one of the year 1990! Or at least, we don't have more than one -- I suppose you could have two consecutive years both called 1990, although that would be rather confusing. So in this case, when the year's name is pluralized, we use that to refer to the entire decade -- all ten years from 1990 to 1999. That aside, what's better, "1990s" or "1990's" for pluralization? I would say that "1990s" is a better choice, since the apostrophe doesn't help readability, and can easily be mistaken for the possessive (e.g. "1990's biggest news story" is the biggest news story of 1990, not the biggest news story of the entire decade from 1990-1999). Also, what about the possessive case of the plural itself? Well, if we pluralize it as 1990s, then the standard possessive form for a plural (appending an apostrophe) seems to work fine. "The 1990s' biggest news story was..."
The upshot is, we use language to communicate, and it is in our best interests to communicate as effectively as possible. I consider it foolish to slavishly adhere to a language mechanism only because of tradition -- other important factors are commonality of usage (how many people will understand you?), simplicity of form (is the mechanism likely to be confused for another, or its meaning misunderstood?), and logic of construction (does the mechanism fit in, systematically, with the rest of the language?). That last factor is a tricky one, since any mechanism can be looked at in that way -- how well does one particular mechanism fit in with all the rest? But of course, any particular mechanism is as arbitrary as the rest; there is no "solid ground" on which to rest the language. As a result, there are a huge number of ways for any particular mechanism to exist, since there are a huge number of states of "rest of the language" that can exist -- for example, look how differently the possessive is handled in English, French, German, Chinese, and so on.