“Best by May 2008,” I read on the bottom of the can of pears. Did that mean best by May 1, I wondered, or best by May 31? Probably May 31, I decided. In any case, even if it meant by May 1, that didn’t mean the pears were actually bad, did it? Just not at their peak of flavor, right? After all, best by wasn’t the same as use by, or even sell by. All the same, I knew if my wife saw that label, she’d throw the pears out. So I did what needed to be done: I opened the can and served those pears to Doug and Adam for breakfast.

As I fished out the pear halves, my thoughts drifted from pears in cans to pears in jars. Baby food jars. Three-ounce jars of pureed pears that we’d fed to Doug and Adam when they were babies. With the canned pears, I could count the halves and know that about 5 pears had gone in. With the pureed pears, it was impossible to say how many the jar contained, but it couldn’t have been more than one. Less than that, actually, since these pears were mixed with pureed pineapple. And yet the label had said, “Pears and pineapple”. Not “pear and pineapple”, or “pear and pineapples”, or “pears and pineapples”.

A joke about English linguistics is the so-called universal grinder (attributed to David Lewis in endnote 7 of this paper). Any count noun can be turned into a mass noun by means of this imaginary tool. The canonical demonstration is with the noun chair — it’s clearly a count noun, since you can talk about one, two, or more chairs. Like other count nouns, and unlike mass nouns, its singular form has to partner with a determiner in order to form a noun phrase — that is, you can’t say *I built chair or *Chair smells like mildew; it has to be I built {a, this, her, …} chair and {This, my, every, that, …} chair smells like mildew. However, if you imagine pushing a chair through the universal grinder, all of a sudden chair can be a mass noun: Chair was all over the floor; the box was filled with coarsely ground chair.

The universal grinder (or a suitable substitute) can also convert proper nouns to mass nouns. An episode from season one of Lost did this when an insufferable minor character named Arzt, in front of several regular characters, accidentally blew himself up with a stick of unstable, centuries-old dynamite. Minutes later, one of the remaining characters told another, “Dude, you’ve got some … Arzt … on you.”

Mass nouns often serve as the “food” versions of count nouns that refer to animals. Thus, you can eat chicken, turkey, duck, crab, lobster, fish, turtle, squirrel, and rattlesnake. Even if you reject the nouns beef and pork, you’d probably talk about eating cow and pig, not cows and pigs. And as Homer Simpson reassured the soon-to-be-vegetarian Lisa regarding the lamb chop on her plate, “Lisa, it’s lamb, not a lamb!” (Search for “vegetarian” here.)

But when it comes to fruits and vegetables, even after a pass through the universal grinder (or a real-world food processor), not all of the count nouns turn into mass nouns. Some, like pears, remain count nouns. Let’s say you see your friend sitting in front of a bowl, gobbling down spoonfuls of some unidentified goo, or drinking some kind of unfamiliar smoothie. You ask, “What are you eating?” or “What is that?” Your friend might say…

• pears
• peaches
• apricots
• plums
• bananas
• grapes
• {straw-, blue-, black-, snozz-}berries
• oranges
• (sweet) potatoes
• yams
• turnips
• beans
• radishes
• carrots
• beets
• mushrooms
• olives
• onions

Or they might say…

• pineapple
• coconut
• papaya
• guava
• kiwi
• durian
• watermelon
• cantaloupe
• grapefruit
• tangerine
• pumpkin

In my grammar, at least, I don’t think you can answer with the singular forms of the items I have in the plural, or vice versa, though there are some that I think could go either way. For example, I don’t know whether I’d say mango or mangoes, cucumber or cucumbers, or tomato or tomatoes.

As for why some words end up as mass nouns and others as plural count nouns, for a while I was thinking a rule like this would do it:

If someone could reasonably eat at least one of a given fruit or vegetable in a sitting, judging by size alone and not by taste, then its cut up/mashed/strained/pureed form will be referred to by a plural count noun. Otherwise, this form will be referred to by the singular form of the noun, used as a mass noun.

However, there are too many exceptions to it in my lists, not to mention the uncertainty with mango, cucumber, and tomato. Now I think that a rule like this may be in effect, but that it’s a weak rule, subject to override when one noun shares semantic properties with some others and gets attracted into their class regarding its mass/plural behavior. Thus, even though someone could probably eat an entire kiwi, mango, or papaya in one sitting, which would favor kiwis/mangoes/papayas for the “mashed” form, they get attracted into the semantic class of “tropical fruit” including pineapple and coconut, and end up as kiwi/mango/papaya.

The forum is open for alternative rule suggestions, including rules that describe the distribution of mass nouns and plural nouns for fruits and vegetables in your grammar, if your judgments don’t match mine.

:: :: :: :: :: :: :: :: :: :: :: :: ::

Two posts ago, I was talking about sentences like They must have loosened the pins and {he didn’t notice / him not have noticed}. Based on just examples with epistemic modals, the interim conclusion I reached was:

It looks like the pattern here is actually that the second clause must have tense, but person/number marking is optional.

Commentator Ellen K. added that she preferred the phrasing They must have loosened the pins and he not have noticed, so this is another possibility to consider. However, it is still consistent with the hypothesis that person/number marking is optional; the only detail is whether the no-person/no-number verb requires a nominative subject or not. For now, I’m going to avoid this third phrasing option, and just see what patterns there are with the phrasings I’ve been working with. The grammaticality judgments I’ll be giving are mine alone; however, my own intuitions have probably been compromised by thinking about these sentences and saying them to myself so much. I welcome your grammaticality judgments.

So, now I’ll look at some sample sentences with deontic modals, i.e. those that express obligation or permission. I’ll start with those expressing obligation, and go ahead and include the quasi-modal have to with them:

Deontic modals: requirement or obligation

• PRESENT TIME
  
  
  
  1. You must steal the medallion and {*they don’t see you / them not see you}.
  2. You have to steal the medallion and {?they don’t see you / them not see you}.
  3. You should steal the medallion and {*they don’t see you / them not see you}.
  4. You ought to steal the medallion and {*they don’t see you / ?them not see you}.
• PAST TIME
  
  
  
  1. You had to steal the medallion and {*they didn’t see you / them not see you}.
  2. You should have stolen the medallion and {*they didn’t see you / *them not see you / them not have seen you}.
  3. You ought to have stolen the medallion and {*they didn’t see you / ?them not see you / them not have seen you}.

With obligation deontic modals, then, it looks like the second clause again must have tense: You can see this in the past-time examples where them not see you is ungrammatical. Now, however, person/number marking is not optional; it’s forbidden. As for why the ought example sounds bad either way, I don’t know.

I’m not done with these wide-scoping modals yet. Soon I’ll look at dynamic modals (those that talk about ability or willingness), and I want to take a closer look at negations that scope over an entire coordination, too.

:: :: :: :: :: :: :: :: :: :: :: :: ::

From today’s Columbus Dispatch:

Most say the gas prices are outrageous, ridiculous and just plain suck.
(Tim Doulin, “Going numb, gallon by gallon,” p. A4)

I am shocked and disgusted to read this kind of language in the newspaper! Here, I’ll fix it:

Most say the gas prices are outrageous, ridiculous and just plain sucky.

That’s better. Instead of the non-parallel coordination of the tree on the left, we have the nice, parallel coordination of the tree on the right.

:: :: :: :: :: :: :: :: :: :: :: :: ::

I read in Dear Abby earlier this week about a nephew who was given some money to treat his grandparents to dinner, but for unknown reasons, did not do so. The current Abby responded in his defense:

He might have offered, and the offer was declined.

It’s another case of a modal that is syntactically part of just one clause (He might have offered), but semantically spreads its hypotheticality over two coordinated clauses (the second one being the offer was declined). The last example of something like this that I wrote about was

They must have loosened the hooks and Mr. Cleaver didn’t notice it.

Recently I’ve been reading a couple of papers on this topic, published in the 1980s by Muffy Siegel. (Hey, is that the same Muffy Siegel who got so much publicity a few years ago on her study of the word like? Yes, it is.) However, her sentences are a little bit different. Here are a couple of them (from “Compositionality, case, and the scope of auxiliaries”, 1987, Linguistics and Philosophy 10.53-75):

Ward can’t eat caviar and his guest eat dried beans.
Ward can’t go out drinking and his wife stay home with the baby.
For once, the hero didn’t end up with the beautiful woman and the villain end up with only his horse.

Her first two examples involve a modal (can’t), so they’re similar to my examples in that regard. Actually, it’s a modal plus a negation. The last one involves just negation, and I’ve collected a few of those too:

I nodded so hard I’m surprised my neck didn’t snap and my head fall to the floor.
I hope she didn’t die and nobody told me.

The main difference between Siegel’s examples and most of mine are that the second clauses in her examples all lack tense. It’s his guest eat, his wife stay and the villain end up, and not his guest eats, his wife stays and the villain ends up. Most of my examples, OTOH, have tensed second clauses: the offer was declined, Mr. Cleaver didn’t notice, nobody told me. The exception is the tenseless my head fall off.

Siegel’s analysis is good, but it depends on these clauses not having tense. She proposes that tenseless sentences can be freely generated in English, but when there’s no context providing the tense information, they have limited uses, such as expressing disbelief or scorn. (What, me worry? Her go out with him?!) For an example of a context that does provide tense information, there are perception verbs, which can take tenseless clauses as complements: I saw him do it. The next step in Siegel’s analysis is the observation that you can even coordinate tenseless sentences: Ward eat caviar and his guest eat dried beans?! All that remains is to assign the modal a semantics that applies to entire propositions (not just predicates), and a syntactic category that allows it to be shoehorned in just past the first subject, regardless of whether the clause is coordinated with another one or not. And then there are the details of making sure this first subject is in the nominative case, so you don’t end up with sentences like *Me can’t go and you not come with me!

So Siegel’s analysis won’t cover examples like mine, which have tensed second clauses. The question now is: Should it? Do you believe that her examples and mine are examples of the same kind of phenomenon, and should therefore be accounted for in a single analysis? Or does their superficial similarity conceal a deeper difference? Right now I’m playing around with different choices of modals and tenses, seeing which ones require the second clause to have tense, which ones allow it, and which ones forbid it. Here’s what I have so far:

First, let’s try out some epistemic modals — i.e. modals that let you express a conclusion you’ve made based on whatever facts or evidence you have.

Epistemic modals

• PRESENT TIME
  
  
  
  1. She must be in love with him and {he doesn’t know it / him not know it}.
  2. She might be in love with him and {he doesn’t know it / him not know it}.
  3. She could be in love with him and {he doesn’t know it / him not know it}.
  4. She may be in love with him and {he doesn’t know it / him not know it}.
• PAST TIME
  
  
  
  1. She must have been in love with him and {he didn’t know it / *him not know it / him not have known it}.
  2. She might have been in love with him and {he didn’t know it / *him not know it / him not have known it}.
  3. She could have been in love with him and {he didn’t know it / *him not know it / him not have known it}.
  4. She may have been in love with him and {he didn’t know it / *him not know it / him not have known it}.

So far, it looks like the epistemic modal can scope over both clauses, but it’s not a matter of whether the second clause is tensed or tenseless. The tensed version for the past time examples, he didn’t know it, is OK, but what of the other two? Him not know it doesn’t have tense, and it sounds bad. And what about him not have known it? That does have tense: the auxiliary have gives us the present perfect tense. Clearly, though, him not have known it cannot stand as a sentence on its own; in that regard it’s like the tenseless him not know in the present time examples. What is it missing, if not tense? It’s missing person/number marking; that is, the third-person singular he/she/it -s ending. It looks like the pattern here is actually that the second clause must have tense, but person/number marking is optional. The present time examples with him not know (as well as the earlier examples) are consistent with this rule because you can know just as easily as a present-tense form without person/number marking as a tenseless base form.

Next I’ll see how this pattern holds up when I try some deontic modals (modals that express obligation or permission) and some dynamic modals (ones that say something about their subject’s ability or desire). But I think I’ll leave those for the next post, since this one is getting kind of long, and I need to think about how I construct my examples.

:: :: :: :: :: :: :: :: :: :: :: :: ::

Matthew Watson asks:

For some time, I have been wondering about constructions like “He better tell me,” which use “better” as a modal verb. I have always used a separate auxiliary like “had” (e.g. “He had better tell me”) and parsed the sentence as a truncated sort of comparative statement (e.g. short for “He had better tell me than not”). However, I have read so many good writers now that use “better” by itself that I am beginning to think the construction has become an idiom.

Do you know what’s correct - should I use “had” with “better,” and how do you parse a solitary “better”?

CGEL is off its game on this one. It doesn’t mention solitary better at all in the section on modals, and offers only one paragraph on had better. And that paragraph says that had better always refers to present time, starring the sentence He had better have done it himself as ungrammatical. It sounds fine to me, as do these:

She better not have eaten all the cookies
You’d better have finished your homework by the time we get back.

According to both the Merriam-Webster Dictionary of English Usage and Garner’s Modern American Usage, the more formal expression is had better, with better by itself not actually wrong, but less formal. But why would plain old better come about in the first place? My guess is that it started out as a phonetic simplification. Using he had better as an example, the first step was probably the contraction of he had to he’d. Over time, speakers probably simplified the string he’d better, pronounced [hidbεɾṛ], to [hibːεɾr], with the [b] assimilating to the following [d]. The colon-like symbol [ː] indicates that the [b] is held for a longer duration than usual (in other words, it’s geminated). Finally, newer speakers simplified by degeminating the [bː], and pronouncing the string as [hibεɾr], or as it would be written out, he better.

The change would have been abetted by the fact that neither he had better nor he better really makes more sense than the other. Why is the had there, a speaker might wonder. With or without it, the main thrust of the expression comes from the better, which in a vague kind of way carries the idea that it’s better for someone to take one course of action than another (the idea that Matthew described). It’s vague in that the meaning of better is clearly involved, but what’s not clear is how the syntax of Subject+(had) better actually gets us there. You just have to write it off as one of those things, and if you leave out the had, then better acts more like other single-word modals, such as should.

However, a thousand years ago the syntax and semantics of better were more transparent. According to the OED, the earliest known citation of this use of better is from 971:

him wær better þæt he næfre ʒeboren

Literally, “for-him [it] were better that he never [was] born”; this is were in its archaic subjunctive use, which would be expressed today by would have been. So originally, be better was an impersonal verb — that is, one that doesn’t have a subject at all, or in Modern English, one whose subject is always it. But eventually the dative-marked “to him” (or whoever something was better for) came to be seen as the subject, and consequently got changed to he. A similar development occurred with best, although the earliest attestation for it is from 1330:

This still doesn’t explain how had got into the picture (or where the accompanying form of be went). The earliest attestation for had with better is from 1465, and the earliest for had with best is from 1559:

had be better … that it had never be done
you had best omit the worke

The better example suggests to me that the had may have come in as part of a have+be auxiliary, and stayed after the be disappeared, but I don’t know enough about Middle English to have confidence in this scenario. Maybe Karl Hagen would know. Also, there might be something on the topic in Visser’s An Historical Syntax of the English Language, but I don’t have that reference handy.

One thing I wondered as I researched this topic is why good didn’t get modalized in the way that better and best did. Maybe it’s related to the fact that modal better and best don’t have any difference in gradation — some people use one, some the other, and maybe some even use both, but the two words seem to have the same meaning. So in Modern English, you can’t say something like:

You had good get to the airport an hour early, but you had better get there two hours early, and you had best get there two hours early and already have printed your boarding pass before you left the house.

I’ve accumulated a number of links to linguistics posts that I’ve been meaning to recommend; now it’s time to get them all out of the docket at once.

First, here’s a guy giving a demonstration of the difference between [ʌ] and [ɘ], in a language where the schwa can be in a stressed syllable.

Folk etymology meets the acronym (OK, initialism or initialization, if you insist) in this discussion of courtesy copy from Josh Millard.

When you see a page of Old English text with stuff like Forþon him gelyfeð lyt, se þa ah lifes ƿyn on it, the first thing you notice is how different the words and letters are. And if you don’t know how to read Old English, that’s all you notice, so you never appreciate, for example, how different the syntax is. Karl Hagen of Polysyllabic meets this need by taking a piece of Modern English prose (by Dan Brown!), and putting it into Old English syntax while leaving the words and morphology unchanged. Interesting fact: Hagen was a consultant on the recent computer-animated movie version of Beowulf.

Next, Greg Larson goes on one of his celebrated rants, this time on an abuse of the adjective extreme by Pringles EXTREME Screamin’ Dill Pickle potato-chip-like salted snacks.

Finally, a couple of recent items of interest from Language Log, for any readers out there who don’t already read it. Here is Geoff Pullum on a simple argument that I’ve never heard made before that puts one more nail in the coffin of the case against singular they. And here is Arnold Zwicky on a construction that you would think just has to be — has to be — a mistake, but which seems to be produced intentionally by a number of speakers.

Here’s a post that’s been sitting in my pile of drafts for more than two years. I know it’s been that long; just look at the “current events” item from January 2006 that it starts out with:

Last week, David Lee Roth’s morning radio show in New York (a replacement for Howard Stern’s program) was canceled, after less than four months, following bad reviews, low ratings, and conflict with the station management. I learned about the situation in an AP story the week before the cancelation. In it, the editor of a radio trade publication was quoted as saying:

I think the radio industry expects this will end sooner than later.

I’ve heard plenty of people say that something should happen “sooner than later”, and I’ve always mentally corrected it to “sooner rather than later”, since sooner than later seemed uninformative. Something will happen sooner than some time in the indefinite future? If it happens at all, at any time before the end of the universe, there will be some “later” that this something happened sooner than, right?

Of course, sooner than later is perfectly sensible when it’s part of a verb phrase headed by rather (as in I’d rather have it sooner than later), or part of a prepositional phrase headed by the complex preposition rather than (as in It happened sooner rather than later), or some other comparison-inducing phrase. Some might quibble over the unspecified comparison — sooner than what? later than what? — but it’s no worse than the vagueness of sooner or later, and now the than makes sense.

At some point along the way, though, for some speakers the than seems to have lost its connection to the rather, and can stick around even when there’s not a rather in sight. Why might this have happened? Maybe such speakers hastily associate the comparative -er ending on sooner with the comparative-friendly than, so that I’d rather have it sooner than later has the same structure as I’d rather have it sooner than 4:00. It sounds reasonable, and who knows, maybe it’s even true. I’d go with that guess if it weren’t for something I once overheard at an airport giftshop. The customer didn’t know whether he should get a gift receipt or not for the object he was buying, and the sales clerk decided to go ahead and give him one, saying:

Well, let’s be safe than sorry.

She’s not alone in using safe than sorry without a better or a rather. Check these out:

School is starting and just to be safe than sorry, I was wondering..? (link)
RV Guidelines: Be Safe Than Sorry (link)
Be safe than sorry: Hard disk disaster can strike anyone anytime. (link)

In these examples, there’s no rather to license the than for real, and no comparative adjective form like sooner to confuse the speaker into putting in a than, but the than is there all the same. The obvious reason for its presence is that safe than sorry is a fixed expression showing up in idioms that do involve comparisons:

(It’s) better (to be) safe than sorry.
I’d rather be safe than sorry.

But how do we get from these expressions to standalone safe than sorry? One way is just clipping it in the same way as people who say “Long story short” instead of “To make a long story short” do. I don’t think that’s what’s happening here, though. If someone said, “Safe than sorry, you know,” and expected me to supply the rest of the meaning, I think that would be a clipping. But the safe than sorry’s I’m seeing are part of complete verb phrases and sentences. (Complete except for the missing comparative words, that is.)

Here’s my hypothesis. Someone gets used to saying a “safe than sorry” idiom without parsing it. Instead of taking the meaning as a comparison to the effect that being safe is better than being sorry, or that someone prefers being safe to being sorry, they just take safe than sorry to be a coordinated pair of adjectives, with than as a conjunction that just means “and not”.

So what allowed this post to finally graduate from the blogpile (as I’ve heard it called) to the published posts? The topic has come up on the American Dialect Society mailing list. You can read the first message here, and click on the arrows with light bulb icons to follow the thread.

A friend of mine recently wrote, asking:

I have a question regarding grammar. We have a book called “One Hungry Monster” and throughout the story, you get to count monsters from 1 to 10 as they beg to be fed. Finally, the little boy decides to feed them, and then you get to count from 1 to 10 the different types of food he brings (2 loaves of bread etc.) The 10th thing is “10 jars of peanut butter”, but the boy adds “and not a speck of jam, because I want every monster mouth shut tighter than a clam. Should it be “monster mouth” or “monster’s mouth”? I think they both sound correct, so I have no idea.

I was a little surprised by this question, since this book didn’t seem at all like my friend’s typical taste in leisure reading, and I’m almost certain she can count much higher than 10. Anyway, I’ll share what I wrote back:

Every monster mouth and every monster’s mouth are both correct. The first is just the compound noun monster mouth (it doesn’t matter that it’s written as two words) just like peanut butter or oven cleaner, put together with the determiner every to make a noun phrase. (A noun phrase is a noun plus any adjectives you care to add [in this case, none] and a
determiner if needed. Determiners include a, the, some, every, no, etc., as well as possessive forms: my, your, Neal’s, every monster’s.) Every monster’s mouth is a noun phrase, too, consisting of the plain old non-compound noun mouth and a determiner: every monster’s. (Inside that determiner lurks another noun phrase: every monster. But that’s another story.) In short:

Other grammar questions? Send them here!

Sometimes I find myself singing,

Kamuga junku, kamuga junku!

What does kamuga junku mean, you wonder? Well, me too. Unlike the helpful lyrics of “Hakuna Matata”, which explain very clearly that the phrase means “No worries,” the lyrics to “Kamuga Junku” state that there are no easy solutions for understanding this phrase:

Kamuga junku, kamuga junku!
There is no English equivalent.
Kamuga junku, kamuga junku!
Kamuga junku, kamuga junku!

The song is from a limited-edition 1987 cassette from a Houston-based band known as The Blanks. It was a limited edition because they made the album for their high school economics class, to sell for play money in a pretend marketplace as part of a toy economy that the teachers set up that week. I hadn’t played the tape in probably 20 years, but like that other song I wrote about, “Kamuga Junku” surfaces now and again as an earworm for a day or so. One day it happened while I was online, so I Googled the phrase. And what do you know — at long last, I found the English translation!

Well, no. But I did find that this hard-to-find album was available, digitally remastered, on CD! So now I have “Kamuga Junku” on CD, in my computer, and on my iPod. I played it in the car today, in fact. I got to thinking, though: When I first listened in 1987, I took the songwriters’ word for it that there was no English equivalent. But having read this 2003 Language Log post by John McWhorter, I’m more skeptical of these kinds of claims of linguistic exoticism. In this post, McWhorter reviews a book by someone named Mark Abley, and is impatient with Abley’s penchant for saying things like, “I had the impression that a three-hour philosophy seminar had just been compressed into a couple of minutes” regarding a revelation from a Mohawk speaker that in Mohawk, the same word means both “righteousness” and “law”. Here’s my favorite bit of McWhorter’s post:

Elsewhere in the book Abley marvels that the Boro language has words that mean specific things like “to love for the last time” and “to feel unknown and uneasy in a new place.” Okay — but English has a word for when two acquaintances, through sharing an experience or reminiscence, experience a sense of deeper connection for the first time: BONDING. How spiritual we English speakers must be … then — get this — we have a word for the first time a couple has sexual intercourse: CONSUMMATE.

So this stuff about there being no English translation for kamuga junku just won’t fly. There’s definitely an English translation. It’s probably not a single word, and maybe not a single phrase, or even a single clause. Maybe it would take a whole page of English to capture the full meaning of kamuga junku, or a whole book, but there is, there must be, an English equivalent!

Not too long ago, somebody lost track of how low we were getting on cat food, and consequently somebody found themself having to buy whatever cat food they could find in an unfamiliar pet-supply store. After walking past the bins of live crickets in the “weird pets” section, I got to the dog section and saw this sign posted in the aisle:

We’ll help you train your dog to behave more consistently.

That could lead to trouble, I thought. Just imagine…

Customer: I want a refund for the money I wasted on this stupid training course!
Employee: I’m sorry you’re dissatisfied. Did the course not work for your dog?
Customer: Hell, no! Riley used to sometimes pee on the floor instead of barking to go outside, but now he always pees on the floor! He used to chew up the newspaper every now and then, but now he does it every single morning!
Employee: The training worked! Your dog is behaving in a much more consistent manner than he did before!

Behave is an interesting verb. It’s a good example of how there are more classes of verbs than just transitives and intransitives. Transitive verbs are the ones that have to have a direct object. To illustrate with syntacticians’ favorite transitive verb, devour, this sentence is ungrammatical:

*They quickly devoured.

For it to sound right, devour has to have a direct object:

They quickly devoured {sixteen hot dogs / the crops in the field / the cannibals}.

An intransitive verb is one that doesn’t take a direct object, for example die. Put a direct object after this verb and you get nonsense:

*Everyone died the animals.

And of course, there are many verbs that can work both ways. In fact, there are so many that it can be hard to find a verb that is obligatorily transitive, and that’s why devour is such a popular example of one. However, when a verb can be transitive or intransitive, the intransitive version sometimes has a more specific meaning than just supplying a generic understood direct object. For example, a sentence like Have you eaten yet?, with the optionally transitive eat, has a more specific meaning than just “Have you eaten something?” The question is whether you’ve had a meal, not whether you munched some Cheerios out of the box, or consumed the piece of paper with the secret password on it.

Instead of direct objects, some verbs are looking for a phrase indicating a location or direction. For example, put with just a direct object is no good:

*He put the suitcase.

Better:

He put the suitcase {on the scale / down / in the unmarked van}.

Other verbs, such as sound, need an adjective. Ungrammatical:

*That sounds.

Grammatical:

That sounds {great / terrible / like something only an idiot would do}.

These phrases, whether direct objects, locations or directions, or adjectives, provide essential parts of the meaning of the verb, whether it’s a participant in the action, or a location that is crucially involved in the action, or a property that you’re saying the verb’s subject has. The general term for these is complements.

A few verbs, as it turns out, take manner-adverb complements, and behave is one of them. Any event described by behave involves two things: something animate, and a manner in which that something conducts itself. The adverb that goes with it doesn’t modify the verb (as an undiscerning sentence-diagrammer would tell you); it provides a crucial part of this verb’s meaning. Of course, just like with optionally transitive verbs like eat, it’s possible for behave to go without its complement, like this:

You’d better behave while I’m gone.

…but when it does, it doesn’t just mean “conduct oneself in some manner,” any more than intransitive eat and drink mean just “eat something” and “drink something.” Like them, it has a more specific meaning; in this case, “conduct oneself properly”. (Behave also has a reflexive version for this meaning: behave oneself.)

If you still don’t believe an adverb following behave is a complement instead of a modifier, try this. Across languages, complements by and large have to appear next to their verbs. You can’t slip an adverb in between them unless there are special circumstances. With eat, for example:

Today, they ate the hot dogs.
*They today ate the hot dogs. [ungrammatical, but for other reasons.]
*They ate today the hot dogs.
They ate the hot dogs today.

And now look at the same pattern with behave:

Today, the children behaved atrociously.
?The children today behaved atrociously.
*The children behaved today atrociously.
The children behaved atrociously today.

There are other tests, such as the “do so” test, but I won’t get into those now. Getting back to the sign at the pet-supplies store, the meaning they must have intended was intransitive “behave yourself” behave, with a structure like the one of the left. In this tree, behave is a VP all by itself (labeled VP₂); it doesn’t need any complements, and it means “behave well, conduct oneself properly”. (I put in the subscript “well” as a reminder of this meaning.) The adverb consistently is diagrammed as a modifier by having it and VP₂ as (to use quasi-family-tree terms) sisters, both daughters of the VP₁ node.

The uncooperative reading I got was with the behave that takes a manner-adverb complement, with a structure like the one on the right. This tree shows the adverb’s status as a complement by having it and the verb as sisters to each other, under the sole VP mother node.

So what’s the practical use of knowing these inaudible differences between intransitive behave and behave taking an adverb complement? When your child says, “Yes, I behaved while you were gone,” and after an angry report from the babysitter, protests, “But you didn’t ask how I behaved!”, and you ground them for lying, you can be secure in the knowledge that they didn’t merely mislead you; it really was a lie.