Today I feel like blathering about an academic subject in which I am completely untrained, and yet have Opinions.
+ + +
Friend of mine on Facebook yesterday shared this post from Letters of Note, the photograph grabbed apparently from a book called 'The Elements of Eloquence,' author Mark Forsyth, and posted originally on Twitter by Matt Anderson.
Said friend and I -- alright, it was Melanie -- had a blast discussing it. The most fun I've had on a Sunday afternoon in a long while. It must be the long weekend.
Before we start to criticize, let's discuss the overall pattern. Have you noticed this feature of spoken English before? That when you prepend -- without conjunctions -- more than one adjective to a noun, often there's a particular way to order the adjectives in series that sounds better than the other possible orders?
Native English speakers would say, unless there was some special reason of emphasis,
The little red wagon
The red little wagon
The old brass key
The brass old key
The enormous antique rocking chair
The antique enormous rocking chair
The antique rocking enormous chair
The rocking antique enormous chair
and so on.
(Even though the author used an amusingly long string of adjectives as an example, I'm not going to go with more than three or four of them here, since our test of validity is simply to see what sounds okay. Overlong strings of adjectives don't sound like natural English to begin with, so they don't make good test examples.)
How do we know which order is right? And is there a permanent pattern? Mark suggested to me that we simply know what is right because we are used to hearing it some ways and not others, but I don't think so because I can construct an entirely novel phrase and we still know that one is better than another:
The delicious square German vegan liverwurst
is obviously (to me) preferable to
The German square vegan delicious liverwurst
but it isn't obvious why.
Some kind of pattern production program exists in our language brain. This isn't mere mimicking. We have a pattern inside our heads, and we know when something doesn't fit the pattern, without ever having been explicitly taught it. What is this pattern?
+ + +
The author of the photographed excerpt considers pre-nominal adjective sets as a string. Like beads, they go onto the noun one after another, and since we apparently prefer some permutations to others, he looks for a pattern that tells us what order we automatically identifies an ordered list of attributes that, supposedly, describes the order in which English speakers string adjectives. Here's the order again:
Opinion - size - age - shape - colo(u)r - origin - material - purpose - Noun.
This doesn't of course mean that we are likely, in ordinary speech, to craft sentences in which we describe a noun with a string of eight adjectives. It just means that we supposedly put size-adjectives before material-adjectives, opinion-about-the-thing-adjectives before origin-adjectives, and so on.
Let's see the examples I wrote just now:
"little red" wagon is size-color .... Check!
"old brass" key is age-material.... Check!
"enormous antique rocking" chair is size-age-purpose I guess... Check!
Or is it size-age-shape? What if the chair is, er, shaped like a rocking chair but intended for an office? Then it would be an enormous antique rocking office chair. Still fits the pattern, I suppose. Check!
"delicious square German vegan" liverwurst is opinion-shape-origin-(material or maybe purpose) ... Check!
+ + +
So far so good. But! If one sits and thinks for a while, one may come up with contrary examples. So the "absolutely" in the author's paragraph must certainly be struck out.
The first one I thought of was
She has a new little baby.
This pattern puts shape before age, yet it is clearly better than
She has a little new baby.
Later I thought of
The white crescent moon shone above
which, though it puts the white color before the crescent shape, sounds more correct than
The crescent white moon shone above.
+ + +
Okay, so the "absolutely" isn't warranted for our pattern. Yet it seems almost right. Is there a way we can generalize? And what makes this pattern appear in the first place?
One of our interlocutors suggested that it had to do with "essence" -- that the more essential qualities are placed closer to the noun, and the more superficial ones are placed farther away. So, for example, a little red wagon is more essentially red and only superficially little. A German rocking chair's rocking-ness is more essential to its nature, and its German-ness less so.
This seems like it is on the right track towards the pattern, but not so much on the how-we-apply-the-pattern. Deciding about the essential nature of things is a philosophical exercise. But we don't stop to practice philosophy every time we order our tall skinny decaf latte. We just ask for it, and make the poor barista hunt for the keys in a different order. (From the perspective of the cashier, the size of the latte is more essential, since it specifies the cost of the item; options like decaf and skinny are not.)
But there is something to this notion, and I think if we carry it farther we will find something better than the "string-of-attributes" theory.
Let's start by asking the question: Why is purpose placed in the closest spot to the noun?
+ + +
I don't think it's because the purpose of the object is the most essential quality of the object, but because of the natural way in which English speakers form compound nouns made up of multiple words.
A tennis racket is a thing. We aren't speaking French: We do not have to say "a racket of tennis." For all practical purposes, in English tennis racket is a single noun. We've never joined the words into tennisracket -- we aren't speaking German -- and that has created a bit of a red herring, in that, yes, "tennis" is a modifier, behaving like an adjective, describing the purpose of the racket -- but face it, a tennis racket is a thing. It is not a racket that might or might not be for tennis, at least not in the same way that it might be green or lightweight or expensive or homemade. It functions as a noun. You don't drop the "tennis" part of the noun unless it's obvious that you're in a tennis context.
Because a tennis racket is a thing, we don't divide the "tennis" from the "racket" by interspersing any other kind of adjective, any more than we divide the "dish" from the "washer" when I describe to you my new household appliance. I don't have a dish shiny washer, and for the same reason, I don't have a tennis new racket.
Is this because tennis-ness is the most essential quality of the racket? Indirectly, maybe; but the proximate reason is linguistic. When we talk of a tennis racket, we join the two words into a compound.
So the handheld implement for striking the ball in tennis, having a certain color and size is not described with a string of three adjectives: large-blue-tennis RACKET. Rather, tennis is joined to RACKET: blue (tennis racket).
+ + +
But purpose isn't the only kind of modifier that we make functional compounds with. I talked of liverwurst up there. We stick that together because, I suppose, liverwurst is German in origin, but we're really discussing liver sausage here -- "liver" denoting the material of which the sausage is made. And doesn't it often function as a compound too? Liver sausage is a thing. If you say "liver sausage" I get a mental image of a package of braunschweiger at the supermarket, followed by a strong association with rye bread and mustard -- I don't imagine a sausage (different mental picture) and then, thinking of liver, modify my mental picture.
"Pork sausage," same thing, until you want to mention that it's pork breakfast sausage (material-purpose-noun). But "pork breakfast sausage" is also a thing, although not as securely as is breakfast sausage and pork sausage.
I propose that in English, when we increase the number of modifiers on a noun, what we are really doing is creating nested compounds. Not sausage that is made of pork, and intended for breakfast, and which I think is expensive, and which is of a brown color, all on a string; but
expensive ( brown ( pork ( breakfast sausage ) ) )
That is, it's breakfast sausage that happens to be pork.
And, it's pork breakfast sausage that is brown.
And, it's brown pork breakfast sausage that is expensive.
+ + +
Let's look at our counter examples: the white crescent moon and the new little baby, both of which violate the prescribed order.
The white crescent moon is easy to explain. "Crescent moon" is a common compound. Yes, "crescent" literally describes the shape of the moon. But in the English language, a "crescent moon" is a thing in and of itself, much more strongly than is a "white moon." So while you might write about a crescent white moon, you are much more likely to write about a white (crescent moon).
Too, look at this:
white full moon (color before, nominally, shape)
round white moon (shape before color)
And that's because a "full moon" is an established compound and a "round moon" is comparatively only an adjective plus a noun.
How about that new little baby?
"Little baby" definitely functions as a compound in many contexts. Sweet little baby, good little baby, third little baby. Little almost goes before baby as a tic.
"New baby" is also a common utterance.
Why, then, "new little baby" and not "little new baby?" I think it's because new babies are by definition little, whereas little babies may or may not be new; so when we want to say, "She has recently given birth to a child!" we say "She has a new (baby)!" or "She has a new (little baby)!" almost interchangeably. The newness is up front. It's what we're emphasizing.
But if we want to say "She has a BIG new baby!" we are probably talking about a nine- or ten-pounder. It's a big (new baby).
+ + +
And what of compound adjectives?
Check out this triplet of adjectives along with the noun "baby", two nominally denoting size and one nominally denoting age, and consider how it works:
tiny new baby (but not new tiny baby)
little tiny baby = tiny little baby (interchangeable)
new little baby (but not generally little new baby)
What's up with this? Tiny must go pairwise before new. New must go pairwise before little. But tiny and little are pairwise interchangeable.
I think this is because "tiny little" and "little tiny" function, and equally well, as compounds. "Little baby" can be a compound, but we can force it apart with the right choice of modifiers and then you can get something like
(tiny little)(American baby).
With new, tiny, and little, there's six possible orders:
little tiny new baby
tiny little new baby
new tiny little baby
new little tiny baby
little new tiny baby
little tiny new baby
Some of these sound better to me than others. How about to you?
+ + +
Anyway, that's my untrained blathering about English syntax for the day. I have a vague notion that the concept of a "parse tree" is the one that trained linguists would use to point out the flaws in my crackpot analysis is. As they say, any fool can criticize, and I am always happy to play the fool for that purpose. But, as I said to Melanie via FB, the fun thing about linguistics is that any nerd can do original amateur research on your own brain and your friends' brains, something that is hard to do in other sciences unless you have an MRI machine in your cheerful spacious well-equipped American kitchen laboratory.