Multiple-level coordination

A few days ago I started writing up a note about Chinese stereotypes of Westerners, especially Western men, from my experiences teaching at Beijing Language Institute (as it was then) in 1985. The story I told began with this Chinese characterization of Western men:

(1) They are hairy, smelly, and have big noses.

(I once had a wonderful cartoon illustrating the stereotype, but I can’t at the moment locate it) — at which point I realized that (1), which was entirely natural for me, exhibited what many people take to be a failure of parallelism that results in ungrammaticality: the three apparent conjuncts (hairy, smelly, and have big noses) are flagrantly not of the same category. (Not that this is always a disastrous thing; see my posting here.)

Neal Whitman is the great student of this type of coordination, which he calls multiple-level coordination (MLC), because the conjuncts are structurally at different levels: in (1) the first two conjuncts, hairy and smelly, are structurally lower (they are complements of are) than the third conjunct, have big noses, which is structurally at the same level as are.

Before I go on to talk about the syntax, a few words about the stereotypes.

[First, hairiness. I have significant body hair, which is visible on my forearms, as well as facial hair. So I look like a typical Western man — rather ape-like.

While exploring Beijing on foot one weekend, I came across a group of small children in a park, who were fascinated by my appearance, eventually breaking out in the trochaic chant mei-guo ye-ye (in pinyin, without tones), but not realizing that I understood enough Chinese to know that they were chanting ‘American monster’.

On smelliness. I had never been so conscious of the possibility that the smell of my perspiration might offend other people (in this case, my Chinese colleagues and students) since puberty set in. Lots of showers and deodorant.

And then the big nose. I have a prominent Alpine nose, so I fit that part of the mei-guo ye-ye image as well. On the other hand, I had an American colleague teaching EFL at BLI who had a button nose. This became relevant when he had a particularly difficult day with his students, unable to get them to talk at all, on any of the topics he offered. Eventually, desperate to get them to speak, he offered to answer any question at all, on any topic, even personal ones.

And one student ventured a question: “Why is your nose so small?”

I don’t recall how he fielded the query.]

Back to the syntax. The handbooks, going back at least to Fowler’s Dictionary of Modern English Usage (Bastard Enumeration, under and), attribute the problem with examples like (1) to the omission of a conjunction, an omission that is the result of slovenliness (Fowler’s formulation), laziness, or inattention. (Much of the advice literature is quick to attribute motives to those who produce examples they deprecate, and even to level moral judgments on these people.)

The idea here is that someone who produces something like (1) is aiming at the correct

(1′) They are hairy and smelly and have big noses.

but lazily or inattentively omits the first and, truncating (1′) to the shorter but incorrect (1). They should be enjoined to Include All Necessary Words (see here).

One comment. The handbook discussions assume, without evidence or defense, but as a matter of “theoretical reasoning” about syntax, that parallelism — vaguely characterized, but let’s put that aside — is a (universally) necessary condition on (all) coordinate structures.

In my experience, people who take this position are almost impossible to challenge on it. For them, strict parallelism in all circumstances is God’s Truth, and evidence from actual usage, even the long-standing usage of competent and careful writers and speakers, is beside the point; people make a lot of mistakes, they say.

On MLC, linguists have been willing, sometimes reluctantly, to entertain the possibility that educated usage accommodates certain classes of MLCs. Whitman’s first posting on the subject “Be pompous, obese, and eat cactus” (2006, here) quoted CGEL (p. 1335) on “coordination at unlike levels”:

“The status of such examples is uncertain. They are more likely to be found in casual speech than in more carefully monitored speech or writing–but they do occur in the latter….”

By Whitman’s second posting, “It’s sick, twisted, and smells like old socks” (here), he too was considering that some of the examples might simply be grammatical. It’s certainly hard to notice them unless you’re looking for them — and then you find them all over the place. My own conclusion is that things like (1) are fully grammatical, instances of a coordination construction that’s available to speakers of English, though some may choose not to use it.

Another comment. The handbooks are inclined to characterize phenomena that they deprecate in terms of how examples can be fixed: (1) can be “fixed” by supplying another and, so the problem with it is that an and has been omitted. What’s missing here is an attempt to describe what people actually do, rather than what they “should” do.

If you think about MLCs in descriptive terms, then you notice other things, which have been brought out in Whitman’s postings and in papers on coordination, in particular a paper by Beavers and Sag that Whitman discusses in some detail (notably, here and here).

One important observation is that the coordinator that is supplied in the handbook analyses of MLCs is predictable from context: it’s and in (1) but or in

(2) They are (either) hairy, smelly, or have big noses.

This is in fact just like ordinary, perfectly parallel, coordination, where in

(3) They are hairy, smelly, and big-nosed.

the second conjunct is understood as connected to the others by and, while in

(4) They are (either) hairy, smelly, or big-nosed.

it’s or that’s understood.

Then there’s the fact that there’s another way to “fix” (1), namely by supplying an are to the second conjunct:

(1″) They are hairy, are smelly, and have big noses.

That is, the higher verb are is understood as distributing over the first two conjuncts, but is only realized on the first. Or in other terms, the second conjunct has an ellipted verb are, with its antecedent in the first conjunct.

(And the contrast between (1) and (2) now reduces to the interpretive mechanism for ordinary coordination, as in (3) vs. (4).)

Whitman has developed this simple but powerful idea, in which MLCs involve an ellipted verb, in a long series of postings covering many different sorts of MLCs (there’s a lot of complexity in this domain); you can get the full inventory by going to his Literal-Minded blog and clicking on the “Multiple-level coordination” tag.

Now to learn to feel comfortable about the fact that I’m hairy, smelly, and have a big nose.


13 Responses to “Multiple-level coordination”

  1. alltheprettylittlethings Says:

    Very interesting post. I LOVE stereotypes cause I feel they give an incredible insight on a culture’s view of themselfes through characterisation of the other.
    However I had quite a lot of chances to be that “other” and people sometimes seem rather disappointed if I don’t live up to my stereotype ( I am German).
    I have to admit that I had the strangest reactions to me being a foreigner in China so far. It seems that people are less inhibited than anywhere else to show their curiosity in otherness.
    For example: I got asked multiple times by young adults to pose with them for a picture. While I would consider it somehow inapropriate to ask someone for a picture just because he looks “strange” it didn’t bother me… just felt weird.
    Did you have similar experiences? Or is that a gender specific reaction?

  2. The Ridger Says:

    There are a lot syntactically ambiguous constructions that can be fixed by supplying an omitted word, but in most cases exactly identical constructions need no “fixing”. Surely, the fact that the “missing” word can be so immediately supplied means this construction is no more “wrong” than “I saw the man using binoculars” (as compared to “I saw the man using a map”).

    • arnold zwicky Says:

      Hmm, I don’t see MLCs as ambiguous, and the handbooks don’t characterize them as ambiguous either (and the handbook writers are just *demons* for finding ambiguities).

  3. Chung-chieh Shan Says:

    I understand “mei-guo ye-ye” as ‘American grandpa’, not ‘American monster’, and that fits much better with my mental image of you.

  4. Neal Says:

    Thanks for all the linkage! I agree that “It’s certainly hard to notice them unless you’re looking for them — and then you find them all over the place.” Even after the post where I aimed to record 100 examples, I still see them almost every day, and usually don’t even bother to record them now.

    • arnold zwicky Says:

      Yes, I’ve pretty much given up on recording them, too, though I’ll catch my own MLCs on occasion (as I did in this case).

      I’d put MLCs on my list of deprecated constructions that have become standard, but it’s not clear to me that they’re innovations (people have been peeving about them on theoretical grounds for at least a hundred years, and I suspect that they’ve been common for a very long time). They might be like “prepositions at end”.

  5. The Ridger Says:

    I didn’t mean they were ambiguous themselves, I was trying to compare the analyses that say “you fix this by adding a word” and yet assume that the “missing” word is obvious. To me, that kind of means the “missing” word really doesn’t need to be added.

    • arnold zwicky Says:

      Ah, got it. First cousin to the famous complaints about logical (rather than temporal) since and while; as Geoff Pullum wrote to me some time ago, the complainants (including the influential Chicago Manual of Style) insist that they be replaced by strictly logical connectives, because they’re ambiguous — but these authorities fail to notice that there’s almost never any question about which occurrences of the connectives are to be replaced!

      Similarly, the complaints about MLCs don’t ever seem to say that they’re unclear, only that they are “illogical”.

  6. Origin myths « Arnold Zwicky's Blog Says:

    […] stories that are colorful, full of specific details, and favor people you admire or revere. [Note multiple-level coordination, which I didn't notice until I was proofreading.] But it can take grinding work to check out these […]

  7. Data points: non-parallelism 3/19/11 « Arnold Zwicky's Blog Says:

    […] and to clean itself automatically (Remington shaver commercial, heard 21/21/04) [this one also has Multiple-Level Coordination in […]

  8. MLC in the news « Arnold Zwicky's Blog Says:

    […] this posting of mine. The perceived failure of parallelism in (1) can be fixed by supplying the “missing […]

Leave a Reply

%d bloggers like this: