Order Your Adjectives: Failed and Nuclear

A few weeks ago, I listened to a news story about the most recent of North Korea's nuclear tests, which they keep managing to pull off despite ever-increasing sanctions. Their neighbor China, one would think, would be nervous about North Korea acquiring nuclear weapons, the reporter said, because "China doesn't want a failed nuclear state on their doorstep."

That took me by surprise. Did China seriously want North Korea to succeed in their nuclear ambitions? Then I realized: The failure in question is a failure to be a functioning state, not a failure to acquire nukes.

Meanwhile, a tweet about English grammar was going viral. On September 8, Matthew Anderson of the BBC tweeted a photo of a page out of a book on English style. The text on the page explained how when various kinds of adjectives precede a noun, they have to be go in a certain order; for example, it's a little blue man, not a blue little man. Anderson titled his tweet "things native English speakers know, but don't know we know," and English speakers who had never consciously thought about this topic retweeted it by the thousands.

But the tweet ignored the fact that some adjectives, like failed, can be ordered in different ways in English, depending on what you want to say. I know, there's only so much you can do in 140 characters. Maybe the book gets into this in detail, but how many people who saw Anderson's tweet tracked down the book and read it? I know I didn't. Anyway, linguists call these adjectives subjective adjectives. To illustrate, here's what I wrote in a blog post I wrote on adjective ordering in 2011, using the example of fake Italian marble:

Italian modifies the noun marble, straightforwardly enough. Fake, however, doesn't modify just marble. It … modifies the nominal Italian marble. Fake Italian marble could be fake marble from Italy, real marble from someplace other than Italy, or even fake marble from someplace other than Italy. If you want to zero in on the "fake marble" meaning, you'd say Italian fake marble. So in this example, the order of fake and Italian depends on the meaning you're trying to convey. In other words, you can rearrange them, but you'll have a different meaning.

In the same way, failed nuclear state could refer to three possible things. It could refer to a state that has a failed nuclear program (the meaning that first came to my mind). It could refer to a failed state that has a successful nuclear program (what the reporter meant). It could even refer to a failed state with a failed nuclear program (which could happen). But if you really want to zero in on the "failed state" meaning, what you need is nuclear failed state. Why didn't the reporter say that?

Probably because the phrase failed nuclear state simply turns out to be more common than nuclear failed state. It took several searches to get that information, because both phrases are so rare that neither of them showed up in the corpora I checked, except for BYU's Corpus of Global Web-Based English (GloWBE), which had exactly one hit for each. I finally resorted to an ordinary Google search and got 3,280 hits for failed nuclear state, compared to 403 for nuclear failed state. However, Google lies. If you page through to the very last page of results for your search, you'll eventually come to a page that admits there are actually many fewer results than originally reported. Sure enough, in this case we end up with 100 hits for failed nuclear state, and about 40 for nuclear failed state. So to the extent that you can trust these numbers, people say or write failed nuclear state more than twice as often as nuclear failed state. But this raises another question: Why is failed nuclear state more common than nuclear failed state, when nuclear failed state has the more precise meaning?

One possibility is that nuclear and state "belong together" more than failed and state do, or as linguists put it, they form a tighter collocation. There are various ways of measuring this, but the one they use over at the Corpus of Contemporary American English (COCA) is called mutual information. So in COCA, we would expect the mutual information between nuclear and state to be greater than the mutual information between failed and state.

Surprisingly, it isn't. In the 520 million words in this collection, failed state is definitely a collocation—not as much of one as failed attempt, failed marriage, and failed experiment, but a collocation nonetheless, coming in at #68 among the top adjectives filling the spot just to the left of state or states. On the other hand, according to COCA's numbers, nuclear state isn't even a collocation at all, let alone a stronger one than failed state. Nuclear doesn't even appear in the top 200 (although non-nuclear and nuclear-armed do). That's not to say nuclear state never appears in COCA: There are 143 hits for it. But that's not enough for it to be a collocation, at least in this corpus.

Even though failed state is a stronger collocation than nuclear state, the term nuclear state is the older term, by several decades. That surprised me. After all, nuclear states have only been around since 1945, whereas failed states have been around for as long as there have been states—haven't they? But I was wrong. The earliest valid attestation of nuclear state that I found is from the Cleveland Plain Dealer, in an article from August 1, 1963, which states that "Red China will probably have the ability to threaten some unattached nuclear state before a decade is up." I suspect there are probably earlier attestations waiting to be found, but still, the earliest one I found was after World War II, as expected. (I did find many examples of nuclear state from as early as the 1920s, but those are in physics papers.)

As for failed state? In a March 2004 opinion piece in the Toronto Globe and Mail, titled "Failed States All Over," columnist Rick Salutin wrote:

You'd hardly know the term emerged a mere decade ago, as "a disturbing new trend," in Foreign Policy magazine. It has such a solid ring, like "empire" or "delegation." But it is really more a sign to cheer or boo, smile or shudder, than a way to describe a real society.

Salutin was right: In the spring 1994 issue of Foreign Policy, author Barry Schweid described events that took place the previous year in Haiti and Somalia, and mentioned how the term failed state was applied to both nations. In that article, failed state was still a new enough term to be placed inside quotation marks. It seems to have first shown up in print in August 1992, when the failed state in question was …

…Arkansas! Bill Clinton had just won the Democratic nomination for president, and during the subsequent Republican National Convention, the phrase "the failed governor of a failed state" was brought out as one of the slogans to be used against Clinton, according to an August 21 article by Don Noel in the Hartford Courant. Though the slogan ultimately didn't succeed against Clinton, it evidently did succeed as a term in foreign policy. These days, there's even a Failed States Index website.

Once failed state emerged as a collocation, what happened when nuclear got thrown into the mix? It looks like the more popular failed nuclear state was first on the scene. The Google Books corpus has it in a June 2004 publication, the Report of the Defense Science Board Task Force On Preventing and Defending Against Clandestine Nuclear Attack . The opposite-ordered nuclear failed state shows up the following year in a conference in Berkeley, California, where the president of the World Affairs Council of Northern California said, "I'm not just worried about what North Korea as a unitary state might do. I'm much more worried about North Korea selling that enriched uranium and other materials on the open market. A nuclear failed state is not a world we want to be in."

Looking toward the future, the term failed state may be on its way out. Consider the sentiment expressed by Salutin:

Talk about patronizing. Why not just call [failed states] losers? … It puts someone in their place while absolving the user of any blame for the failure, and simultaneously justifying either intervention (poor things) or abstention (they're hopeless). Neat trick.

Sentiments like that one are probably responsible for the name change in the Failed States Index: It's now the Fragile States Index. And as for fragile states with nukes? At this point, we're into "things we don't know we know" territory: To my ear, fragile nuclear state just sounds better than nuclear fragile state, and it doesn't create the same ambiguity as failed nuclear state. A fragile nuclear state is a fragile state, regardless of the robustness of its nuclear program. It looks like other speakers agree: Although Google returns less than 20 hits for fragile nuclear state, the string nuclear fragile state has none.