## Ask Language Log: Is this a sentence?

D.M. writes:

A discussion on copyediting-l turns on whether one or more of the following are grammatical English sentences.

"That smile scares me," she said and swallowed hard.
"That smile scares me," she said and backed away.
"Anything for you, man," the captain said and extended his hand.

I'm in a minority that says they are not or that at best they are unintended examples of zeugma. What says Language Log?

I'll take this from three vantage points.

(1) My own intuitions:  The cited examples seem fine to me. I wouldn't have noticed them in reading, and on careful re-reading my inner Norma Loquendi still doesn't complain.

(2) General practice: The COCA corpus finds 1,674 examples of the pattern

said and [vvd]

with 1532 of them in the fiction category, for a frequency of 14.6 per million words in that genre. Some random examples:

"Rules rule," Coyote said and winked.
"That's fine," Pronek said and wiped his sweaty palms against his pants.
"It was your sister's," Fran said and resumed eating.
"Yes it is," she said and knocked my hand away.
"I could use your help," she said and patted his leg.
"At least let's keep in touch," he said and withdrew the package.

This frequency is strong support for the view that the construction is well within the norms of standard written English.

(3) Grammatical analysis: Here the picture is less clear, and we can see where D.M. is coming from. There's no problem, of course, with conjoining verbs with different complement patterns, including when the first verb is "said":

They said nothing and didn't move.
They both said their piece and walked away.
After a while she said thanks and handed it back to him.
Filled with gratitude to be here, he said his prayers and dozed off as soon as he finished.

But when the complement of said is a WH-trace in the first conjunct, the result is questionable at best:

*That's what he said and walked away.
*Tell me what she said and dozed off.

So why is it OK to have a fronted direct-quote complement, in what seems like the same sort of structure?

"Goodbye," he said and walked away.
"Goodnight," she said and dozed off.

We could follow D.M., and take the ungrammaticality of things like the WH examples to imply that the direct-quote examples must also be wrong. But instead, in a case like this, we should follow our instincts and the examples of respected writers, which imply that the structures must be different in some crucial way that prevents or excuses any culpable zeugma.

Update — a commenter points us to Neal Whitman's 2007 post at Literal Minded, "Tolkien, Rowling, and Quotative Inversion", which discusses a similar construction at length. Neal notes that Tolkien often uses non-parallel constructions of this type — though with the added feature of quotative inversion:

“We are plain quiet folk and have no use for adventures. …I can’t think what anybody sees in them,” said our Mr. Baggins, and stuck one thumb behind his braces, and blew out another even bigger smoke-ring.

In contrast, Neal finds that J.K. Rowling (or her copy editor) apparently doesn't like this construction, preferring things like

“Well, get back to your common room where you belong!” snapped Filch, and he stood glaring until Harry had passed out of sight.

Neal's conclusion:

Hastily generalizing from these books and the Beverly Cleary ones I wrote about, I’d say the avoidance of non-parallel coordination where quotative inversion is involved seems to have begun sometime in the past 20 or 30 years.

I'm skeptical that there really has been a change over time. At least, it's easy to find recent examples of the non-parallel pattern without quotative inversion:

“I'd like to hear about it,” he said, and opened his diary. [J.S. Foer, Everything is Illuminated, 2002]
"Of course," she said, and started briskly toward the house; [Sandra Marton, The Bride Said Never!, 1997]
“I shall return as soon as possible”, he said and held her hand. [Kefira Bar-Golani, Lions and Legacies, 2015]
"Take it one hundred and forty or leave,” she said, and started to take things from display on the front of her table. [Lazlo Ferran, Ordo Lupus and the Temple Gate, 2010]
"More like everyone forgot today was my birthday" I said and went back to my room [G-Jamz, Kaylin's Diary, 2014]
“It would break Gran's heart if I did,” she said, and started downstairs. [Susan Beth Pfeffer, Kid Power, 2015]
“Red seven," he said, and motioned toward a line of cards headed by a red seven. [Maris Soule, The Bachelor, The Beauty and the Blizzard, 2011]
“Come on, Mullins,” Jerry said, and went to the sofa and began to pull at one end of it. [Frances Lockridge, The Judge Is Reversed, 2016]
“Yeah, thanks,” I said and went and sat on the couch. [Scott Gallagher, Dancing Upon the Shore, 2010]
"This is all of my fault. I should have never left her,” I said and started crying harder. [Keyanna Ford, Living in Fear, 2014]
“And that's Ray,” Brent said and sat on the bed, rubbing his hands together. [Seth Johnson, The Things We Do for Women, 2013]
“I’ve seen enough,” I said, and started for the exit. [Paul Levine, Mortal Sin, 1994]
"It's a very low probability outcome," she said, and pushed the crashbar. [Cory Doctorow, With a Little Help, 2010]

Searching for the pattern with quotative inversion is harder to do on Google Books, but COCA allows a pattern like

said [np1] and [vvd]

which turns up plenty of examples like these

"No thanks," said Cory and turned away. [1996]
"Forget it," said Anita and turned her back on him. [1997]
"Sure he has," said Eli and pulled a face [1998]
"Sure you are," said Tom and took it. [2001]
"Of course I can," said Hector and stood up himself. [2003]
"I'm not going to skate this time," said Hans and slipped his strap into Peter's skate. [2006]
"Oh, please," said Rose and glanced down at herself [2009]
"Fine," said Jayjay and tuned out [2009]
"Jesus," said Jake and looked around for an enemy. [2009]

It may be worth noting that in many cases, the coordinated verb phrase functions to tell us more about the act of speaking, and could be replaced with a participial adjunct. Thus

“I'd like to hear about it,” he said, and opened his diary.

is functionally equivalent to

“I'd like to hear about it,” he said, opening his diary.

Update #2: A very relevant note in the comments —

The same pattern obtains with subject-verb inversion after fronted adverbials, e.g.

Itsy bitsy spider went up the water spout.
Down came the rain and washed the spider out.
Out came the sun and dried up all the rain …

Update #3: Beatrice Santorini points out that all of this is closely related to a phenomenon known as "Subjektlücke" ("subject gap") coordination in German, e.g.

Das Gepäck  liess er fallen und rannte zum Ausgang.
the luggage let   he drop   and  ran  to-the exit.
'He dropped the luggage and ran to the exit.'

See e.g. Hohle, 'Assumptions about Asymmetric Coordination in German,' 1990; Heycock and Kroch, "Verb Movement and the Status of Subjects: Implications for the Theory of Licensing", 1993.

1. ### David Denison said,

February 26, 2016 @ 6:41 pm

Those three sentences are fine for me too. Does this sequence help to license the last one?

They said their piece and walked away.
They said goodbye and walked away.
They said, "Goodbye!" and walked away.
"Goodbye!", they said, and walked away.

My only hesitation with the latter pair is whether or not to put a comma before the and, as I normally do without one before a conjoined clause with no expressed subject – but this is a matter of written style, not grammaticality.

The comparison with

*That's what he said and walked away.

is a red herring, isn't it? What is and walked away being coordinated with?

That is what (he said and walked away)
That is (what he said) and walked away

In the first bracketing, what is fronted wh-object of both verbs – but walk is intransitive. In the second, the understood subject of walk is obscure, the best bet being the subject of the preceding main clause, namely that, which obviously won't do. So of course, either way it's ungrammatical.

[(myl) What I had in mind was a violation of the coordinate structure constraint:

That is [what he said <trace> and walked away]

And so we can conclude that whatever is going on in

"Goodbye," he said and walked away.

it's not

"Goodbye" [he said <trace> and walked away]

No doubt there's some extant syntactic theory that explains, or at least is consistent with, the fact that

[QUOTATION SUBJECT [said and VERBPHRASE]]

allows the VP coordination even though the QUOTATION is the complement of "said"… In fact maybe this is a standard Syntax I quiz question these days? ]

2. ### Christopher Henrich said,

February 26, 2016 @ 6:55 pm

IANAL, but here goes anyhow…
The oddity of D.M.'s sentences seems to me very slight. In any case it has nothing to do with zeugma (or syllepsis). The issue is whether it's OK to "front a complement" when the verb phrase that it came from is part of a coordinated pair. I feel that it is OK. If I were reading one of those sentences out loud, I would put a tiny pause after "said"; in writing I would suggest a comma there.

Note that you can replace the complement without changing it, and have a satisfactory sentence:

She said, "That smile scares me," and swallowed hard.

With the "WH-trace" this doesn't work, even without the coordinated verb phrase:

That's what he said.
*He said that's what.*

Actually, your first example with a WH-trace seems acceptable to me, perhaps because it can be replaced with minimal editing:

He said that and walked away.

Your second WH-trace seems unacceptable to me. I feel that "Tell me" insists on governing the whole sentence

3. ### GH said,

February 26, 2016 @ 7:27 pm

As an alternative to:

*That's what he said and walked away.

How about "So he said, and walked away"?

4. ### Weltanschauung said,

February 26, 2016 @ 8:25 pm

From Milton's Lycidas:

But not the praise,
Phoebus repli'd, and touch'd my trembling ears;

5. ### Guy said,

February 26, 2016 @ 8:29 pm

What about other preposed complements? How do people feel about "We have pork and beef in the fridge. The pork, we'll eat for dinner tomorrow and go to bed"? Another thing to keep in mind is that "and" in these examples seems to convey a temporal order, with the situation expressed by the second coordinate occurring immediately after the first, and that's a situation where parallelism constraints are sometimes relaxed.

[(myl) The cases with topicalization seem just about as bad to me as the cases with WH structures.]

6. ### Jason Merchant said,

February 26, 2016 @ 10:01 pm

These seem like good counterparts to what are called "Subjektlücke" sentences in German and Dutch, which have generated a ton of discussion (of the form "In den Wald ging der Jäger und schoss einen Hasen": see Büring and Hartmann 1997 for an excellent discussion of these). Even worse is this NYTimes headline: Is Boxing Dead But Doesn't Know It? (screen grab here). Fun!

7. ### phspaelti said,

February 26, 2016 @ 11:12 pm

Well, since we're on the topic, I'd thought I'd throw in a favorite lyric from Lisa Loeb's "I do":

You can't hear it, but I do.
You don't seem angry, but I do…

8. ### Jason said,

February 27, 2016 @ 12:12 am

Also a favorite construction of one Mr Tolkien, eg

“We are plain quiet folk and have no use for adventures. …I can’t think what anybody sees in them,” said our Mr. Baggins, and stuck one thumb behind his braces, and blew out another even bigger smoke-ring. (p. 7)

9. ### rosie said,

February 27, 2016 @ 1:54 am

They sound wrong to me. No bracketing is possible.

1 *"That smile scares me," she [ said and swallowed hard ].
2 *["That smile scares me," she said ] and swallowed hard.

1 fails because the direct speech is not a complement of the second verb. 2 fails because a complete clause is being coordinated with a subjectless verb phrase.

To adapt a phrase used by Fowler to label a similar syntactic mistake with an object, there's a walled-up subject.

Guy's examples with a fronted object are clearly syntactically wrong, and the examples with direct speech are wrong for the same reason.

3a I had never before heard such a specious argument, and told him so.
3b *Never before had I heard such a specious argument, and told him so.

10. ### Bloix said,

February 27, 2016 @ 2:05 am

Is this all about the punctuation?
Does anyone have any trouble with:

"That smile scares me," she said, and swallowed hard.
"That smile scares me," she said, and backed away.
"Anything for you, man," the captain said, and extended his hand.

11. ### Fraser H. said,

February 27, 2016 @ 2:45 am

Is there a case to be made that "he said" enjoys a special status? Its frequency in standard fiction has caused some writing instructors to suggest that it goes almost unnoticed by most readers, analogous to the masking of saccades in visual processing.

If this possibility is stipulated, then maybe the verb "said" is somehow discounted in processing, making what seems logically like zeugma not be processed as zeugma.

12. ### CC said,

February 27, 2016 @ 2:48 am

As a grammar novice, this discussion is slightly over my head. Can I assume that the perceived problem with the sentences is that '"That smile scare me"' is the object of said, and when a second verb is added in the manner of the original examples, then the same object is implied for the second verb? By switching it around to "She said, 'That smile scares me,' and walked away' solves that problem by making it clear where the object belongs.

13. ### Dick Margulis said,

February 27, 2016 @ 6:29 am

@CC: Yes, that's the issue. "She said A and did B" is fine (compound predicate). "A, she said, and she did B" is fine (two independent clauses making a compound sentence). The problem is with "A, she said and did B," which seems defective.

Bloix's suggestion that a comma fixes it works pretty well from the norma loquendi point of view, but it is not a satisfying answer analytically. I've seen several people assert lately that one of the uses of the comma is "to replace missing words." I honestly don't recall hearing that and can't find it listed in any of the grammar or usage books I just checked, so I suspect it's an Internet rumor of recent vintage. In any case, it doesn't solve the problem here, because the "missing word" is the subject of the second clause, which would occur after the conjunction, not before.

Another interpretation of the suggestion to insert that comma is that the speech tag following the speech needs to be set off by commas because of the inverted order. And so "A, she said, and did B" (as in the Tolkien quote above) does seem far better than "A, she said and did B." Maybe the comma does remedy the defect. I'm half persuaded that's the case.

Part of the problem, I think, is that somewhere along the line we've adopted the convention of separating the speech tag from the speech with a comma. My hunch, although I don't have the inclination to do the historical research to back this up, is that before punctuation became quite so formulaic and conventional as it is now, it was probably typical to write "He said A" without a comma and that the inverted case, "A, he said," took the comma because of the inversion; and only later was that rule reinterpreted to mean that a comma always separates the speech tag, yielding "He said, A."

14. ### Joseph F Foster said,

February 27, 2016 @ 9:08 am

The three sentences are fine. Any analysis or theory that treats them as ungrammatical is simply wrong.

15. ### languagehat said,

February 27, 2016 @ 9:11 am

As a professional copyeditor, I find the sentences perfectly acceptable and it would never occur to me to change them. As a quondam linguist, I feel the same; I also feel obliged to point out that it is apparently the grammatical-analysis horse that is driving the acceptability cart for those who put their faith in such entities as WH-traces.

[(myl) I agree, as I hope the post made clear. But a question remains: What is the structure, and why is the apparent lack of coordinative parallelism not a problem in this case?]

16. ### GH said,

February 27, 2016 @ 9:30 am

Does it make a difference if you put the subject after the verb?

"Fine," said Tom and signed the contract.

Personally I'd still be inclined to add a comma:

"Fine," said Tom, and signed the contract.

17. ### David L said,

February 27, 2016 @ 9:51 am

I think Fraser H. is onto something.

"That smile scares me," she said, and backed away; or
"She said "that smile scares me" and backed away.

But then:
We had beef stew for dinner and went to bed; but
Beef stew for dinner we had and went to bed.

The second is OK only for Yoda. It seems fine to put 'she said' either in front of or behind the thing said, but that's far less acceptable with other verbs.

18. ### zythophile said,

February 27, 2016 @ 10:20 am

Rosie: is there any universe in which you would not understand immediately what that sentence meant?

The answer is surely "no", which means that whatever technical rules it might be held to break, as an English sentence it is entirely acceptable.

19. ### Ray said,

February 27, 2016 @ 11:33 am

this reminds me of free indirect speech getting tangled up with direct speech (or how writing a narrative gets entangled with writing a screenplay)… in the process of getting a character to say and do something, the narrator’s distance can shift from directing to reporting.

and along the way, it’s as if “s/he said” becomes something that we see/read as punctuation (end quotes), but that we don't hear so much as an action verb, which allows the 2nd verb to wander onstage, sometimes with bad timing…

“That smile scares me,” she said and wondered if she was jumpy.
“That smile scares me,” she said, and wondered if she was jumpy.

"That smile scares me," she screamed and wondered if she was jumpy.
"That smile scares me," she screamed, and wondered if she was jumpy.

likewise:

“Anything for you, man,” he said and imagined what would happen.
“Anything for you, man,” he said, and imagined what would happen.

"Anything for you, man," he shouted and imagined what would happen.
"Anything for you, man," he shouted, and imagined what would happen.

20. ### Nathan said,

February 27, 2016 @ 11:42 am

zythophile's position is a tad extreme. Not everything immediately understandable is grammatical/acceptable/idiomatic.

21. ### Weltanschauung said,

February 27, 2016 @ 2:05 pm

Itsy bitsy spider went up the water spout.
Down came the rain and washed the spider out.
Out came the sun and dried up all the rain …

A verb phrase can have an understood subject from an earlier term in the coordination, and if there's no way to isolate the subject with a pair of brackets, so much the worse for brackets.

[(myl) Good examples. But we should want to be able to say something more than "sometimes it works and sometimes (as in the case of WH-constructions and topicalization) it doesn't".]

22. ### Roscoe said,

February 27, 2016 @ 2:52 pm

"Courage!" he said, and pointed toward the land,
"This mounting wave will roll us shoreward soon."

– Tennyson, "The Lotos-eaters"

23. ### Bloix said,

February 27, 2016 @ 3:35 pm

dick Margulis – you remind me that I was taught that a comma can be used to indicate a missing word:
The Scots drink whiskey, the Dutch, gin, the French, brandy, and the Russians, vodka.

24. ### Mark said,

February 27, 2016 @ 5:35 pm

More random observations:

Nobody *converses* by fronting quotations. It's a convention of written language. And it seems to be a pretty limited one. It's always clause-initial, never preceded by "Maybe", "Sometimes," "Some people are so rude that…", etc. Nor can it follow "A believes/said/knows that", "A wondered whether", etc. Rarely does it follow a conjunction. It also seems to be specifically a device for specifying what *is* (was/will be) said, and so "or" cannot fill the role of "and" in the original examples, and "not" doesn't work in the verb phrase.

She will say "Yes" or remain quiet.
*"Yes," she will say or remain quiet.
?"Yes," she said, or she remained quiet.

She did not say "Do come in."
*"Do come in," she did not say.

(This is why the popular tag "said no one ever" is a joke.)

Also:
*"I've had enough," Josh shouted and Anita said.

These observations lead me to suspect that we're not going to get an elegant explanation of this as a special case of very general syntactic rules. It may just be that we have a special convention that, in certain contexts, "'Yes,' she said," can be used as a stylistic variation of "She said, 'Yes.'"

25. ### Ethan said,

February 27, 2016 @ 6:11 pm

@Mark

She will say "Yes" or remain quiet.
*"Yes," she will say or remain quiet.
?"Yes," she said, or she remained quiet.

All three of those strike me as unexceptional, although it might depend on a preceding sentence:
"We questioned her for hours. 'Yes', she said, or she remained quiet. I am sure if we call her in again tomorrow it will be the same. 'Yes', she will say or remain quiet."

26. ### Guy said,

February 27, 2016 @ 9:47 pm

@zythophile

By that reasoning, "I would like to can swim" would be grammatical, which isn't right.

27. ### Y said,

February 27, 2016 @ 10:02 pm

'"Tehee" quod she and clapte the wyndow to.'

28. ### rosie said,

February 28, 2016 @ 2:05 am

Languagehat: I also feel obliged to point out that it is apparently the grammatical-analysis horse that is driving the acceptability cart for those who put their faith in such entities as WH-traces.

Not in my case. The analysis doesn't cause the unacceptability, it explains the unacceptability that already existed.

Zythophile, you're too eager to say things about other people, where only they personally would know the truth. In this case, you're wrong about me. The meaning of the original sentence is not immediately apparent on my first reading it, because the second verb leads me up the garden path. Not only that, there is an easy fix.

29. ### GH said,

February 28, 2016 @ 6:03 am

@rosie

Could you explain a bit more how you think the first sentence is confusing, and how it should be fixed? The only thing I can imagine is that from a strictly analytical point of view, you might initially interpret "she swallowed hard" as referring to the direct quote; that she swallowed her words, hard.

I have to say I find it hard to really believe that sentences of this form could be genuinely confusing, since, whether or not one approves of them, they occur regularly enough that I would think any reader should be used to them.

@Mark:

*"Do come in," she did not say.

Although the statement is odd, that's because it's expressing an odd point; it looks grammatical enough to me.

@Guy:

"The pork, we'll eat for dinner tomorrow and go to bed."

I think part of the problem here is that the logical connection between the two parts seems very tenuous. Even when reordered, the change in topic is jarring and the timeline confusing: "We'll eat the pork for dinner tomorrow and go to bed." I think it becomes far more acceptable if you change it to something like:

"The pork we'll eat for dinner tomorrow, and have the cake for dessert."

@David L, Fraser H:

Regarding the "special status" of "X said", keep in mind that authors (even Elmore Leonard) mix it up with a lot of other words as well: "exclaimed", "replied", "wondered", "called", "asked", "answered", "whispered", "explained", etc. So I don't think there is anything special about "said" as such, but rather something special about reporting direct speech.

30. ### languagehat said,

February 28, 2016 @ 8:50 am

The analysis doesn't cause the unacceptability, it explains the unacceptability that already existed.

Nobody but you thinks it's unacceptable.

31. ### January First-of-May said,

February 28, 2016 @ 2:50 pm

@Y: I'm not as much surprised by the construction (though, aside from the quotative inversion, it's a nice example) as by the surprisingly modern sounding interjection (though the modern equivalent is more likely to be spelled "teehee"). Are there any other interjections that survived the passage of time so well?

@everyone else: I basically agree that the construction is even more acceptable with a comma between the two verbal structures, or with quotative inversion (which makes a zeugmatic interpretation less obvious), but it's still just fine even as given (especially in the examples found by the search, as opposed to those given in the original comment, but even those are nice enough).

32. ### Bloix said,

February 29, 2016 @ 12:18 pm

@David L"
"Beef stew for dinner we had and went to bed.
… OK only for Yoda."
Actually, it sounds vaguely Tolkienish.

33. ### Stan Carey said,

March 1, 2016 @ 7:51 am

I prefer them with a comma after said.

34. ### Usually Dainichi said,

March 1, 2016 @ 8:17 am

@Languagehat: "Nobody but you thinks it's unacceptable."

In the context of commenters on this post, you might be right. If you include people mentioned in this post, not so, unless D.M. and @rosie are the same person. In the context of all native English speakers, I highly doubt you are right.

And even if you're right, so what? Aren't people allowed to be different?

I sorely lament the kind of militant descriptivism you're showing here. If your point is that @rosie probably deep down accepts the sentences, but thinks they don't because of prescriptive rules clouding their vision, then state so, and be respectful. If you have scientific evidence that what @rosie says is unlikely to be true, present it. I'm sure people would appreciate it. I definitely would.

Too many times have I seen descriptivism as an excuse to suggest that minorities with slighly different grammaticality criteria are either lying or ignorant of their prescriptive super-ego. Frankly, I think it's a stain on an otherwise honorable cause.

35. ### Y said,

March 1, 2016 @ 5:50 pm

@J. 1st of May, I am no expert, and I don't know about other interjections, but I do believe that that quotation of Chaucer's is the first written record of "tee-hee", in whatever orthography. It's pleasantly startling when you first see it.

36. ### Ted said,

March 7, 2016 @ 7:10 pm

Guy:

How do people feel about "We have pork and beef in the fridge. The pork, we'll eat for dinner tomorrow and go to bed"?

The problem here is the pork. Consider instead:

The kugel we'll eat for dinner tomorrow and go to bed. The pork, I don't eat and wouldn't have in the house.

By me this is normal syntax. English, I'm not so sure.