I'm curious: by this rule, would "the old gray stone Presbyterian church" be wrong? And why? Somehow it makes more sense to me (gray modifying stone, and Presbyterian modifying church), but I'm willing to be wrong....

It is said that laws, codes and rules are established because someone learned something the hard way. If we leave the Maryland General Assembly out of that discussion it remains generally true.

Similar rationality can be applied to the "rules" of grammar.

Like a template, the conventions and structures described in the OP are examples of what are known to work. And like the phrasing in the tenth amendment the detail is left to the devil to (and copy editors) to determine.

Article [X.] The powers not delegated to the United States by the Constitution, nor prohibited by it to the States, are reserved to the States respectively,or to the people.

Good post. I've tried to make a similar point on my own blog before. My conclusion is that it's not so important whether you call them descriptive rules and prescriptive rules or patterns and rules or whatever, but that you make the distinction in the first place. Different people can mean very different things when talking about the rules of language.

Eric P.'s version of the phrase is the one I prefer.

@ Eric P. – I believe it depends on whether you are thinking of a stone church that is gray or a church that is made of gray stone.

@Eric P. : In my graduate linguistics courses, which to be fair were many years and a few theories ago, we never quite learned the exact sequence John mentions above as a hard and fast rule, but we did discover empirically that the tighter the semantic relationship is between the adjective and the noun it modifies, the closer the adjective is to the noun. Based on that observation, I don't think your phrase is wrong, but John's phrase sounds correct also.

@everyone: As for rules, linguists are still determining the inherent rules of a universal grammar that applies to any human language. I've been out of the loop for quite a while, but Chomsky himself has evolved well past the transformational-generative grammar of his seminal work Syntactic Structures. When I was studying it, the popular theory was government/binding. Chomsky was developing the barrier theory when I dropped out. These rules generally are much less prescriptive than the ones we're taught as children. The idea is to do case studies of bits of the grammar of more than one language, and then derive generalities from those observations. So the reason we're unaware of the "rules" is precisely because they're so internalized. Some linguists even think they're hard-wired.

Sorry for the long post. Anyone with more recent knowledge of theory in the field, feel free to correct or update me.

Not all the sources one can consult about the grammar give identical versions of the sequence of adjectives in English. So there is some variation in the possibilities. But it's equally clear that combinations outside the sets of sequences that the grammarians and linguists have described sound bizarre to the native speaker.

I'm having fun creating phrases that use all of the adjective types in the order provided. To wit (with a tip of the hat to Scott Adams for the nationality he invented for the Dilbert comics): "a disreputable large old square black Elbonian orthodox metal community oven."

When I learned high school French, we were taught the order of adjectives was to be remembered as BANGS -- beauty, age, number, goodness, size. Heaven help me if I had to remember all those other restrictors above. :-)

There's give in English, but basically it's:

number
subjective qualities
inherent qualities

Inside inherent qualities we find

size
age
condition
color
breed/brandname/type/purpose

However, note that there can be sub-grouping (gray stone house = (gray stone) house), and also that there is overlap between the "age" and "condition" categories, so "battered old shoes" is as likely as "old rusty truck".

Source: Oxford Dictionary of English Grammar