Category Archives: language and linguistics

Whoever is the subject?

Who will inherit the investigation?

Oh, whoever will inherit the investigation?

Whoever will inherit the investigation, he will be someone Mr. Trump nominates.

Whoever will inherit the investigation, Mr. Trump nominates him.

Whoever Mr. Trump nominates will inherit the investigation.

Wait, says the writer. Mr. Trump nominates him. So it must be whom. Whomever. And so, in The New York Times, appears this:

Whomever Mr. Trump nominates will inherit the investigation.

Because formally correct. So whom. Yeah?

Nah. Hyperformalism.

Of course cases like this bedevil writers. The construction is complex and whom is not part of standard daily English; in effect, it is a foreign word for most of us. Wherever we think it might be appropriate for formally correct speech, we are tempted to slip it in, sort of like how some people stick –eth on every conjugation when they want to sound old-fashioned. But sometimes we go overboard and use it where it doesn’t belong.

When people write sentences like the one in question, the rule they’re turning to is that the object must be whom, not who.

The rule that they’re forgetting is that every verb must have a subject.

What’s the subject of will inherit?

It has to be whoever, because whoever else would it be?

One loophole that writers miss that would resolve some grammatical dilemmas is that a whole clause can be an object, as in “Mr. Trump will nominate {whoever gives him the most money}.” Another loophole they miss is that the subject or object of an embedded clause can be made to disappear by what linguists call moving and merging, leaving just an embedded trace (that we know exists thanks to psycholinguistic experiments). That’s what goes on here. The him in Mr. Trump nominates him gets tossed like a baseball in a double play back to the Who, and the catcher’s mitt on the Who is ever. (It can also be an emphatic as in “Oh, whoever will help us?” but it’s not one here.)

Look at “Who(m)ever Mr. Trump nominates, he will inherit the investigation.” (I put the m in parentheses because if you use whom as the object you would use whomever here, but in normal non-prickly English we use whoever as the object too.) Notice that you (almost certainly) wouldn’t write “Who(m) Mr. Trump nominates, he will inherit the investigation.” The ever sets up a second reference, the he. It can also set up an object (him): “Whoever gives the most money, Mr. Trump will nominate him.” (All of this works with she and her too, but we can see that Mr. Trump does not work with very many shes and hers.) So the ever can refer to an object while attached to a who that’s a subject, or the converse.

Our sentence du jour, however, is not derived from “Who(m)ever Mr. Trump nominates, he will inherit the investigation.” Not quite. In “Whoever Mr. Trump nominates will inherit the investigation,” the main verb of the sentence is clearly will inherit (will is the auxiliary that takes the actual inflection, and inherit is the infinitive that conveys the sense); the subject of will inherit is Whoever, as already pointed out. Mr. Trump nominates is an insertion – a subordinate clause modifying Whoever. By itself it would be Mr. Trump nominates him, but, as I said, the him is tossed back and caught by the ever.

Let’s diagram that like a good linguist, shall we? This is the fun part! Syntax trees have details that non-linguists will be unfamiliar with, so let me set down a couple of basic facts:

  1. A sentence is a TP, which means tense phrase – because it conveys tense (when the thing happens), not because it’s too wound up. The heart of it is thus the part that conveys when it happens: the conjugation on the verb. The verb phrase (VP) is subordinate to that, but it merges with it unless there’s an auxiliary verb taking the tense.
  1. A subordinate clause is also a TP, because it has a conjugated verb, but it’s inside a CP, which means complement phrase, because it’s a complement to something else in the sentence. Often there’s a complementizer, such as that or which, but not always.

So.

The subject is Whoever. Because in English conjugated verbs (except for imperatives) have to have explicit subjects and they have to be in the subject (nominative) case, this can’t be Whom or Whomever. The tense goes on will. The verb is inherit. The object of that (its complement) is the noun phrase (NP) the freakin’ mess – sorry, the investigation. (I haven’t broken that down further, but actually it’s a determiner – the – and a noun.) The complement of Whoever, by which I mean the subordinate clause that describes who the Whoever is, is Mr. Trump nominates [him]. The him is tossed back to the ever.

Whoever Mr. Trump nominates will inherit the investigation.

Whoever will inherit the investigation?

Who will inherit the investigation?

He will inherit the investigation.

(Mr. Trump nominates him.)

So why doesn’t the NYT version instantly sound bad, as “Whom will inherit it?” would? It’s a more complex and unfamiliar construction, and what we tend to do in such cases is go with the salient rules we can remember and basically make up rules to make the rest work. For people who don’t balk at the “Whomever Mr. Trump nominates will inherit the investigation,” I believe what’s probably going on is that it’s an underlying “Whomever Mr. Trump nominates, he will inherit the investigation,” and the he is getting tossed back to the ever. So you have a trace of the subject rather than the object. Now, you can have a trace of a subject when you have more than one verb conjugated to the same subject – “Whoever gets the nomination inherits the investigation” – but it’s not normal formal standard English for a subject to be deleted and merge with an object that is not deleted. We need the subject!

But then, really, whoever speaks formal standard English all the time? Well, not whoever wrote that sentence, anyway, or it wouldn’t have been written, because it would have sounded wrong.

What’s logical about English?

A common complaint about English – by those who are inclined to complain about English – is that it’s not logical enough. Whatever that means. Words aren’t premises and sentences aren’t syllogisms, after all.

If you inspect the targets of their opprobrium, you find soon enough that what they mean is that English isn’t tidy enough for them. It’s inconsistent. Lacking in symmetry. Their experience has led them to believe that for every up there should be a down, for every in an out; when they see an over, they think “therefore under,” and if there is no under, they are… underwhelmed.

They’ve condemned themselves to a lifetime of disappointment. English does not satisfy their need for an overarching tidiness. It is not a Zen garden; it is a forested mountain, every tree grown unplanned in its place and conditions, every rock where the ineluctable complexities of physics left it. It is not an edifice of modernist design with proportions based on the golden mean; it is a Winchester House of a language, a veritable Heathrow Airport of accretions (for those who have not been to Heathrow, let me just say I suspect that J.K. Rowling based Hogwarts on it). Like any natural language, English has been built up by habit, need, association, and analogy. It does have structure – in fact, it has some inflexible syntactic requirements. We have slots to fill, and fill them we do. We just sometimes grab whatever’s ready to hand to fill them.

Let’s consider a few examples. One case where a desire for logic has actually prevailed is “double negatives.” Anyone who has studied logic will tell you that in “not not” the second not undoes the first one. “There will not be cake” is disappointing; “There will not not be cake” is affirming. Thus, the reasoning goes, “I do have nothing” and “I don’t have nothing” are opposite. But anyone who has learned a Romance language ought to know ça ne vaut rien, no vale nada – that ain’t worth nothing.

Nothing, you see, is not not. It’s a noun, not an operator. And one thing languages like is agreement. Concord. Adjectives tend to take the same gender as the nouns they modify, for instance. In English, we use concord with tenses in some contexts: “Should we expect them tomorrow?” “They said they weren’t coming.” Notice how we use weren’t even though we’re talking about the future? We even let negative concord pass unremarked in some contexts: “They won’t be coming, I don’t think.” This doesn’t mean I don’t think they won’t be coming; it just retains the negative aspect.

But since it’s possible, with shifting emphasis, to make “There ain’t no one here” and “There ain’t no one here” mean opposite things, an argument can be made for disallowing negative concord for the sake of unambiguity. So the proscription stuck, defended by pleas for logic – although “if negative noun, then negative verb” is perfectly reasonable if that’s the rule in the language.

Syntax has its requirements – as linguists would say, there are principles and parameters that specify how it functions in a given language. Negative concord is one parameter we have managed to turn off. Others are not so easily disabled. It’s necessary to have an explicit subject (except in imperatives), for instance; I can’t write “Is necessary to have an explicit subject,” so I stuff in an it that has no meaning. It may not seem logical to have a pronoun with no referent, but consider that, from the view of our syntax, “if it has a sentence then it has a subject” is solid. Sometimes we grab and stuff on the fly – we may jam a word in the place where a word like it normally goes, even if in this case it’s a whole nother thing and what even were we thinking? This, too, comes from a simple if-then – just a little simpler than it might have been.

Another plea for logic comes when a word is pressed into service in a way that seems untidy. One I saw recently was an objection to using disconnect as a noun, as in “There is a real disconnect between the labourers and the management.” We don’t say “There is a connect between them,” we say connection, so it’s illogical not to say disconnection. Indeed, this is untidy, in the same way as it’s untidy that when my wife is at home I heat two servings of food and pour two glasses of wine, but when she’s not at home I heat one serving and open a beer (or go out for sushi). But our little untidinesses have reasons: my wife doesn’t drink much beer and doesn’t like sushi. And disconnect is an allusive use borrowed from electronics and telephony.

A line of communication is expected to remain connected, so there is no instance where we would say that it has experienced a connect. We grabbed a bit and stuck it where it fit, and in so doing made a metaphorical connection. There’s no need to construct a symmetrical positive use any more than there is a need for a 33-storey building to have 33 levels of basement. And there’s no need to disallow allusions just for the sake of tidiness – we don’t forbid lights on Christmas trees just because there are none on the house plants. If you want to make a connection, you make it; if you don’t, you don’t. That’s logical, no?

Some people also like to laugh at how “illogical” English words are. “Why do our noses run and our feet smell? Why do we park in a driveway and drive in a parkway? Why do we say a bandage was wound around a wound? How come you can object to an object?” OK, now tell me why these are illogical.

Every one of them comes from a well-motivated historical development founded on consistent principles: metaphor, ergativity, historical sense developments and standard compounding rules, phonological shifts, stress-based differentiation of nouns from verbs. In every case there was an if-then judgement based on analogy. It just happened not to be exactly analogous to some other if-then judgements, and it produced results that seem inconsistent when juxtaposed. I think that’s fine – why not have funny things? But more than that, it’s not even illogical. In every case, we got to it from “if A → A´, then B → B´.” They just happened to be local judgements made in the context of a big, multifarious, inconsistent world.

But it would be illogical to treat a multifarious, inconsistent world as though it were elegant and pervasively consistent, wouldn’t it? It certainly wouldn’t be well adapted. It would be like laying down a strict grid street plan for a very hilly city (and San Francisco knows how well that worked out). It wouldn’t be as much fun, either. And it might do real harm.

croon

The lexis of our language is like a coral reef, full of wonders rich and strange. And, as with coral reefs, one of the threats to its diversity is bleaching.

Coral bleaching is a result of coral shedding algae due to rising sea temperatures. Semantic bleaching is a result of words shedding distinctions of meaning due to overuse, over-broad use, what one might call thesaurusitis: treating all words in a section of Roget’s as fungible. The words are still in use, but they lose much of their distinction of sense, thereby reducing the expressive power of the language. And modern electronic media can amplify this effect.

Expressive power and electronic amplification have much to do with the word of which I sing today, croon. It’s not a new word; it dates back half a millennium in Scotland and northern England, where it has for that long had the sense of a low sound, either (particularly in Scotland) a low, deep, loud, steady sound, or (more broadly) a low murmuring sound or soft quiet singing. It sounds like it should mean what it means. In singing, it’s the opposite of belting. Belting is the kind of singing you do when you have a noisy room and poor (or no) amplification. Once you have a good microphone and good speakers, you can draw the audience in with quiet, smooth singing. You can croon.

Which is what happened in jazz in the late 1920s onward. The first truly famous crooner was Rudy Vallée. His soft voice crept like a lover into millions of living rooms through the radio. Many others followed; the one probably most often thought of as a crooner was Bing Crosby. They all had a gentle, quiet singing style that worked closely with the microphone.

But as one technology giveth, another helpeth take away. Newspapers are written by people who are trained to be allergic to repetition. They seek different words for the same thing to make their prose seem more varied and expressive. We can never forget that a pumpkin is a gourd thanks to them; we are given the idea that every promise, however solemn and formal or not, is a vow; food writers talk of people munching even the softest, smoothest, quietest foods (ice cream? oatmeal?). And every act of singing may be called crooning.

It’s not that the word is always used over-broadly; it has not been utterly bleached. But a quick look through recent New York Times articles finds gospel choirs “crooning” more than once and a jazz singer who “crooned over the trio, belting the 1941 Duke Ellington classic” – yes, crooning and belting as the same act. Even non-singers, we are told, have sometimes “crooned”: soccer fans, en masse in a stadium, “crooned” “We’re not going home”; Donald Trump, at a rally, “crooned” “I love you! I love you!” to his supporters. Every one of these uses is amplified to an unlimited number of eyeballs through the wonder of the world wide web.

As a linguist, I can look at this and just write it down as instances of semantic broadening due to an evident desire for more expressive-sounding vocabulary (with the likely long-term effect of reducing the expressive value it draws on). As an editor and user of the language, however, I would rather resist it, because it ultimately reduces the expressive power. And there can be quite a lot of expressive power in the soft, quiet, focused, and amplified sound of crooning.

populism, populist

Language change is, generally, organic. It usually doesn’t happen by fiat (especially in English); it also doesn’t happen by vote. There may be some influence from “above” by people such as English teachers, but that mostly affects what rules people think they’re breaking when they’re speaking the way they want to speak anyway. You could say that language change – grammar, the meanings and pronunciations of words, and so on – happens by mass popular movement.

Which is not to say that it’s populist. Populism is a political stance that advocates the people – the general populace, hoi polloi – in opposition to a ruling élite. Language change is not the product of a program leading a movement of the populace in opposition to English teachers, editors, and others. There could be such a movement, of course, but de facto language change happens by popular will anyway. We’re all part of it, not just the people at the top. (The situation can at times be different for languages with official deciding bodies, such as in France.)

So, you could reply, if we all decided that the meaning of populism should shift to ‘following the will of the people as with a general tide, without a specific political program’, then that would be the meaning. Well, yes, if that sense shift happened and ultimately overcame opposition. The change would probably take quite a while and not be without some controversy. But it could happen.

But there are shifts that we have every right and reason to resist, no matter how many people use the new sense. We are all streamkeepers of this flowing language. When a word is being used as a euphemism to let something slide that should not, or when its use carries implications that have negative consequences, we should not let these pass unnoticed. If we speak up and point out the problems, we may help these shifts to become unpopular.

So, for instance, for a long time there was a default assumption that people in certain roles were masculine, and so he and him were used. Around the time that such assumptions started to be a bit less tenable, a common line from prescriptivists was that he and him were the natural universal gender-neutral pronouns. (Poor men, having to sacrifice the uniqueness of their pronoun! Ah, such sacrifices must be made.) At long last enough people pointed out that this did, in fact, convey the default assumption of masculinity – words carry resonances and implications whether you say they should or not – and so use of masculine pronouns as a universal has lost general acceptance. (Read more about this in my article on they.)

Which, I suppose, you could say was a populist movement – the neglected masses against the prescriptive authorities – though it was in particular a movement by and on behalf of the more neglected moiety of the populace, to influence the more dominant segment and thereby produce a fairer outcome for all.

But now let’s say that people start using populism to refer to a movement focusing on the desires not of the whole population but on a minority of it who consider the remainder to be of lesser status. Say, for instance, that there is a group we’ll call X in the population, and they feel that the government has been giving too many rights to that larger part of the population that is not X. This group has traditionally been the group that, for all its internal differences, has been ceteris paribus the more-advantaged group, and they’re seeing non-Xes get similar rights. This doesn’t involve the loss of any rights from X – unless you consider it a right to have things that other people you consider inferior can’t have. If some political leader or party rallies members of X against the government just so they can protect their perceived right to have more rights, would you call that populism? Would you accept seeing it called populism? When the movement is for the rights of not all the populace but just a subset of it, and strongly against the rights of others?

This is not a hypothetical question. I’m seeing populist used quite a lot by news media and the commentariat for racist, nativist, frankly sexist and reactionary movements. In countries across Europe and at least one in North America, leaders who advocate or enfranchise not just xenophobia but racism and sexism are being called populists, and the reactionary groups that support them are being described as having populist sentiment.

Which implies that women and non-white people are not part of the populace, or anyway are not relevant parts of it. In spite of being, in sum, the majority. And, for that matter, it also implies that white men are, en masse, in favour of such movements. Which is also not true.

I think we owe ourselves and everyone else a duty to make this use of populist and populism unpopular.

Calling them what they want

This article was originally published on The Editors’ Weekly, the official blog of Editors Canada.

We’re all professionally attentive to detail, so I’m sure we all appreciate that, having earned a PhD, I am technically Dr. Harbeck, and it could be rude to call me Mr. Harbeck. My wife, having a master’s, is Ms. Arro — not Miss Arro, because she’s married, and not Mrs. Arro, let alone Mrs. Harbeck. Letters addressed to us as “Mr. and Mrs. Harbeck” will be received as uninformed or rude, depending on who they come from.

Now, if I were a judge, you would call me “Your Honour”; if I were a lord, I would be “sir” or “my lord”; if I were a king, I might be “Your Majesty.” When we refer to politicians, nobility and high-ranking ecclesiastics, we have to make sure we include, as appropriate, “the Right Honourable” or “His Eminence” or whatever. We’re in the business of calling people the right thing: the title to which they are entitled.

Or calling them what they want to be called. Even non-editors know it’s rude to call someone something they don’t want to be called. We don’t call Sir Edward “Eddy baby” unless he asks us to. We also don’t call people who have changed their names by their old names, especially if their identity has changed. We don’t call Chelsea Manning “Bradley Manning” or Caitlyn Jenner “Bruce Jenner” (although we may use that name historically, for instance in stories on the Olympic Games).

We don’t always call people by names and titles, though. Sometimes we just use pronouns. There are languages (such as Turkish and Finnish) in which the sex of a person makes no difference in the pronoun, but English is not yet one such. Since the binary distinction is an unnecessarily restrictive imposition, the singular they is gaining currency (since number sometimes is relevant, however, expect to see they-all becoming popular in its wake). But some people do want to use pronouns for gender presentation. There are a few different pronouns in use, not just heshe and they, but also others such as zey. But not nearly as many as there are honorifics, let alone names.

And yet, some people — even ones apparently capable of attaining and requiring “Doctor” before their names — find it beyond endurance to have to keep track of these pronouns. They deride it as silly faddism or political correctness — terms of abuse for people who refuse to stay in the boxes you have made for them. They can manage to remember who is Mr., who Dr., who Your Excellency; they can get a grip on who is Alex, who Sandy and who Alexandra; but keeping track of pronouns is just too much for them.

Of course it’s not really. They just don’t want the dominance of their paradigm challenged.

As editors, we like to ensure adherence to chosen sets of arbitrary standards. But we also like to check our facts and get the myriad nice details right — such as what pronoun a person has asked to be called by. It’s not all that difficult, and it’s good manners, too.

In the original article, I didn’t include some further remarks on “freedom of thought,” which was a line taken by a professor who is a prominent opponent of following people’s choice of pronoun. But I would like to add them briefly here:

In the case of the professor in question, it’s obvious to onlookers that he’s incensed at having to defer rather than always be deferred to; it threatens his freedom of thought only inasmuch as it makes it difficult for him to maintain his hyperinflated self-estimation. (He has been heard to lecture women on the purity of his feminist bona fides. Not really the cuttiest butter knife in the drawer, this guy.)

But just to address the broader question: If you are of the opinion that strict nativist two-valued gender normativity is the only truth, I assure you that using requested pronouns will not force you to think otherwise. You are still able to think such things. If you are concerned about your reputation, lest you be mistaken for someone who respects others’ choices of gender identities, you are still free to make it clear that you are actually quite rigid in that regard, and are conforming to university policy out of respect for civility. You are even free to think that civility is stupid; your freedom to be a jerk in your mind is not impaired by a requirement to act nice. Most of us are jerks in our minds more often than we are in our words and deeds.

For a parallel: We can’t force people not to think racist thoughts (though we can do what we can to encourage them to revise their views), but we sure as hell can require them not to say racist things. Especially within the ambit of an educational institution, for instance. Part of existing in a civil society is agreeing that, however little you may like or agree with some people, you must at least recognize that they have certain rights, which must necessarily be extended to all for the functioning of society. One of which is to be treated like a human being, and not as something less due to some intrinsic part of their person.

See? You can think whatever you want. But you act in a way that shows the required acknowledgement of others’ humanity. This may threaten your freedom of thought if if interferes with your holding the view that you are already being more than accommodating enough for these people, or forces you to confront the possibility that, in spite of what you tell yourself, you do not view everyone equally. But I do not think freedom from having your thinking challenged is a freedom worth fighting for.

That old bad rule-seeking behaviour

Linguistics is great for making you aware of things you were already doing consistently but weren’t consciously aware of. In fact, that’s the basic point of several subfields of linguistics. There are a few particularly memorable examples that one learns in the course of an education in linguistics. One of these is the order of adjectives: we have a standard order for adjectives when there are several before a noun. We may not be analytically aware of it, but if someone says “a red big balloon” it will sound wrong.

If you’re a linguistics student, you take that as more data, and the point of such data is to use it to help you figure out why we tend to do that, and to do that you have to see what exceptions there are and sort out what the various inputs and influences are. It’s explanation-seeking behaviour.

If, on the other hand, you’re not a linguist but an ordinary English speaker, you may approach English with an eye to finding out what is right and what is wrong. We learn that good grammar is the great sorter, and we treat errors as evidence of flawed character and intellect. And so when you encounter a new fact about the language, you may well be inclined to turn that fact into a rule. It’s rule-seeking behaviour.

So when, recently, a book came out pointing out (in passing) the standard order of adjectives to the lay masses, many people were all “mind=blown” about it. The author, Mark Forsyth, stated confidently that the order absolutely has to be opinion-size-age-shape-colour-origin-material-purpose-noun, and “if you mess with that word order in the slightest you’ll sound like a maniac.”

Linguists, of course, quickly critiqued this overstatement; read Language Log for some responses. But non-linguists didn’t just say “Wow, I didn’t know I was doing that”; many of them were writing down that order and determining that any deviation from it must be wrong. I didn’t see any instances of people telling other people “You didn’t put it in this order, so you’re wrong,” but I did see instances of “What’s the correct order? I don’t want to get this wrong!” English speakers, you see, are in general convinced that they’re always making mistakes and doing things wrong, and they want to know the right way to do them for those instances where they’re going to be judged. (Ironically, many of them think one of those instances is when they’re talking to a linguist. Listen, honey, whatever you say or don’t say is great with us; it’s all data.)

So I feel I need to say this: There is no official correct order of adjectives you must adhere to. If something sounds wrong, then adjust it, but if it doesn’t sound wrong, it’s not wrong.

In fact, we vary the order depending on context, priority, and sets of words that we’re used to having together in a certain order. Consider a nice big old round red Spanish silk riding hood. Try moving any of the words and it may sound wrong (but not like a maniac!), although big old nice Spanish red round silk riding hood doesn’t sound awful to me. You might also get away with Spanish red silk. Some orders are more weighted than others. But change nice in the original phrase to good. Suddenly it doesn’t work as well: not good big old but good old big will sound better because of collocation – good old travels together. Also, any compound noun will defeat the word order. You have a tawny giant tortoise, for instance, because giant tortoise is a thing. And it will be different in what it can convey than giant tawny tortoise. We can even vary the order to adjust the sense: a little dumb ass is different from a dumb little ass because dumb ass is a common collocation.

So if you come up with a string of adjectives and it doesn’t sound right, just change the order so it sounds right. Make things that go together go together. Do not start overthinking it: don’t say “this one must go before that one because that is more common or more essential,” as is a common explanation for the ordering. Go with what sounds right. Afterwards, you can use that as interesting data to tell you what you see as more essential, but do not arrange the order by asking yourself which is more essential, because you could be wrong. For instance, we would all say big red building and not red big building, even though the building can be repainted more easily than it can have its size changed. And yet even that order can be changed in some circumstances: for instance, we use little closer to the noun to express an attitude towards it: I’m not gonna drive that purple little car of yours all over town communicates something not altogether the same as I’m not gonna drive that little purple car of yours all over town.

The number one thing we should learn from this, then, is that in general people aren’t fully aware of how their language works and yet they make it work, and when people start trying to analyze it and come up with prescriptions they very often miss things and get things wrong. The most poplar grammatical cudgels are based on thick-headed simple-minded misunderstandings of how and why we do things, and on grotesque overapplication of rules. That’s why they’re used as cudgels: not everyone follows them because they aren’t real rules, they’re made up.

Mistrust imposed rules, especially inflexible ones. You’ve been using English your whole life. If someone tells you that something that sounds right to you is wrong, and that something that sounds odd to you is right, they’re probably just wrong. And if your analysis tells you that a sentence that sounds awkward to you is right while one that sounds good is wrong, question your analysis. If you love the language and want to understand how it works and use it effectively, engage in explanation-seeking behaviour, not rule-seeking behaviour.

About this sentence that you’re reading

Originally published in Active Voice, the magazine of the Editors’ Association of Canada

About this sentence that you’re reading…

Should that be “About this sentence, which you’re reading”? After all, you’re not reading any other sentence, are you? So it’s not restrictive, so it must be non-restrictive, meaning it should have a comma and use which. Right?

I was talking to my wife Aina and a friend about this the other day…

What?

What do you mean, how many wives do I have? Look, if I set it off in commas, it would be “I was talking to my wife, Aina, and a friend about this the other day,” and you would be saying “So that’s three people? Who’s your wife if not Aina?”

Since you’re an editor, you’ve heard of restrictive and nonrestrictive clauses. A nonrestrictive clause is like a little Santa Claus in the sentence: it gives the reader a gift, with a comma or two as tiny ribbons.

The cakes are served warm.

The cakes, which are kept in the refrigerator but carried under the waiters’ arms, are served warm.

The sentence would be coherent without it, but it’s there as a special bonus of information. It’s also called a nonrestrictive relative, which works because it’s not really Santa Claus giving you gifts, it’s your relatives. (The rules apply similarly to non-clause modifiers as well, such as the name of my wife, Aina.)

A restrictive clause or relative (or modifier) is like your aunt who says “I’m giving you anything you want as long as it’s this sweater!” It no sooner gives than it starts to take away.

The tea is served cold.

The tea that is carried by Pat is served cold; the tea that is carried by Alex is served warm.

But where there is a Claus, there may be a Grinch. What is a restrictive Grinch? It’s one of those people who carp at things that are perfectly clear to reasonable people. Consider:

He went to a famous school called Eton.

The restrictive Grinch might say “Oh, so there are multiple famous schools called Eton?” In fact, the sentence does not necessarily imply that there are multiple famous schools called Eton (although it does imply that there are multiple famous schools), and ambiguity is not automatically a grammatical error – although it can be worth avoiding… when a reasonable person might reasonably misread it, or when too many people might deliberately (and unreasonably) misread it for their own entertainment.

Reasonableness is important. Communication normally (outside of contracts and courts of law) depends on people being reasonable. We make inferences on the basis of what’s reasonable, given our knowledge of the world. If we have some new data, such as a sentence,* and we choose among different interpretations on the basis of what we already know to be the case, we’re using a loose form of Bayesian inference.

Let’s take a Bayesian look at “my wife Aina and a friend” versus “my wife, Aina, and a friend.” In the version with commas, if we don’t know my wife’s name, we need to conjecture or determine whether I use the serial comma; if I don’t, it’s clearly two people, but if I do, it may be three. In the version without commas, the possibilities are (1) that I have more than one wife or (2) that I am disregarding the usual rule about setting off nonrestrictives with commas. Why would I disregard it? To avoid interrupting the flow or causing ambiguity with the extra commas. Given that bigamy is illegal in our society, option 2 is far more likely. Indeed, only a restrictive Grinch would raise the objection that the comma-free version must mean option 1.

There are, of course, many cases where proper use of commas is necessary to set off nonrestrictive clauses for clarity or legal defensibility. But there are also cases where the commas make no difference to the meaning but may make a difference to the flow or tone. The moon, which orbits the earth, is also the moon that orbits the earth (since there are many moons but there is one we call the moon); the sun, which we orbit, is also the sun that we orbit (for the same reason). This sentence that you’re reading is this sentence, which you’re reading, and the different possible uses of this make both defensible, but flow and tone may make one a better choice than the other.

So, when we are faced with a modifier such as a relative clause and we’re not confident about whether it’s restrictive or unrestrictive, we should ask the following questions:

  1. Will the commas help or hurt the flow?
  2. How likely is it to be misread accidentally?
  3. How likely is it to be misread deliberately?
    1. By which readers?
      1. And do they really matter?

 

*Or some new data such as a sentence – equally true.