A few mornings ago I was half-listening to a radio piece that I think may have been about women’s kick-boxing in Jordan. (Forgive me for the vagueness, but it was way before 6 a.m., and I was half dozing to the early morning sound of my bedside clock radio playing the BBC World Service magazine program Boston Calling.) As my mind slowly rebooted, I heard someone quote an inspirational saying:
| [1] | Be the kind of woman that when you get up in the morning the devil says, “Oh crap, she’s up.” |
There’s something very interesting about the grammar of that sentence. But my guess is that only those who study syntax fairly carefully will have noticed it.
The sentence is a variation on a well-known theme; a Google Images search on “be the kind of woman” brings up scores of cards, wall hangings, fridge magnets, posters, coffee mugs, T-shirts, etc.; the image at left is just one example. (I don’t know the original author; I have seen attributions to Joanne Clancy, Jaimie Jacobs, and “Unknown"; the truth about the original source may be lost in the mist of time.)
All versions of the sentence share the interesting syntactic property I’m referring to. They contain a relative clause that lacks the vacant noun-phrase position (which I will call a gap) that English relative clauses are syntactically required to have.
Consider a typical relative clause, such as the underlined part of this sentence:
| [2] | Be the kind of woman that the devil will be afraid of. |
The gap is the unfilled noun-phrase position after of. The preposition of is syntactically required to have a noun-phrase complement. That’s why *The devil will be afraid of is ungrammatical as an independent sentence: The of demands a noun phrase complement, but it hasn’t got one.
The reason an open interrogative like Who will the devil be afraid of? is grammatical is that the initial who serves to satisfy the requirement imposed by of, despite being positioned at the beginning of the clause.
Relative clauses are somewhat like open interrogatives, in that they are required to have gaps, and often begin with either a wh-word (like who or which) or the subordinator that. Here’s what it’s like when you construct a sentence with a relative clause that has no gap:
| [3] | *Be the kind of woman that Paris is the capital of France. |
It’s ungrammatical (hence the star annotation). Paris is the capital of France is fully grammatical, precisely because it does not have an unfilled obligatory noun-phrase position; hence it cannot serve as a relative clause.
You might think that the problem here is relevance: The national-capital status of Paris has nothing to do with classifying women into different kinds. But a clause like people will be afraid of her is fully relevant; nonetheless, [4] is ungrammatical.
| [4] | *Be the kind of woman that the devil will be afraid of her. |
English relative clauses require a gap, not a pronoun. (This is specific to certain languages; the literal translation of [4] into Hebrew would be grammatical.)
So it’s very surprising that [1] feels completely natural. Somehow it manages to permit semantic naturalness to override syntax. It allows the property of being an x such that when x gets up in the morning the devil says, “Oh crap, x is up” to be expressed by the clause when you get up in the morning the devil says, “Oh crap, she’s up” despite its gaplessness.
Most people presented with [1] (and this is an experiment you could undertake) would probably see nothing wrong with it at all. If told there was something wrong and asked for a suggestion as to how to correct it, they would fiddle around making irrelevant changes. Yet a grammar for English that doesn’t demand gaps in relative clauses will wrongly allow [3] and [4] as grammatical, and we don’t want that.
Do not be too surprised at this categorization of [1] as “acceptable despite being ungrammatical,” however. Mark Liberman noted on Language Log in 2014 that there are precedents going back as far as 1890 for a strangely tempting error (coincidentally involving the word gap): People writing about something valuable will mistakenly write that it “fills a much needed gap,” clearly meaning that it is much needed because it fills a gap. The grammar of the phrase (which plainly says it is the gap that is needed) cannot conceivably support that meaning. The mind can play strange tricks, it seems. Optical illusions are one proof of that. And what we have here, I think, is a syntactical illusion.
So, be the kind of woman that when you speak people are so entranced by your meaning that nobody notices your syntax.
