Rotate, but don't turn

I've been looking at mattresses lately. It's bewildering. Not least is the instruction that some of them have:

No-turn mattress. Rotate regularly. 

Rotate and turn are one of many pairs of words that English has that are more or less synonymous but come from different sources. Typically, the one that we get from Romance (French or Latin) is used in more formal contexts or has a technical sense: here, that word is rotate. The Germanic counterpart (here, turn) is usually used in a more everyday sense.

That works for this pair. The OED has examples like The whole stage rotates concentrically and the kid turns on the spit, where each could be substituted for the other. The definitions are also more or less identical:

Turn (intrans): to move round on an axis or about a centre
Rotate (intrans): to turn about a centre or axis

We actually have the transitive senses here, as there is an implied object the mattress, but the transitive definitions are based on the intransitive ones.

The above discussion implies that it ought to be contradictory to say no-turn; rotate regularly, as how can you rotate something that can't or needn't be turned? Of course they mean it doesn't need to be flipped over, but you should turn it 180 degrees about its vertical axis (is that what I mean?? the thing stays flat, anyway) now and then to make sure it wears evenly. Here's an example of a pair of synonyms getting put to use in a situation where two different words meaning turn are needed. If one were so inclined, one could check whether the specific meanings each has (turn = flip over and rotate = stays flat) were generally consistent or if it's random which is used for which. That'll have to wait for another day, though, unless one of my readers wants to do it.

Gwynne's at it again

The odious Neville Gwynne is at it again, publishing books. This time, he's written a Latin book. He's so pompous, my immediate instinct is to disagree with anything he says, so when I saw an advert for it in the paper that said learning Latin would improve your English, I refuted it loudly and firmly to anyone who would listen.

I love Latin, and I think everyone should learn it. A friend who was subjected to the refutation pointed out to me several ways in which learning Latin can improve a person, and he actually mentioned things that most people never think of, such as scientific analysis (I think he said this, anyway - he said biology, so I suppose he may have meant that you'd understand binomial classifications better, which is true, but it would also help you with doing anything that requires careful, logical, rigorous analysis). This friend also agreed with Gwynne that Latin would improve your English, however, and until today I thought that I heartily disagreed with this point of view.

Latin grammar is sometimes held up to be 'better' than English, or alternatively the basis of English grammar. Although Latin grammar is a beautiful thing, it is not better and nor is it the root of our language. Gwynne says that Latin is the source of 'well over half' of English. This is sheer nonsense. I think he must be counting words, because I can accept that half of English vocabulary comes from Latin - we did borrow a lot when we were conquered by the Romans and then the French (I say 'we' - I've no idea whether I'm one of the conquerors or the conquered, or even if there's any way to tell at this distance). But even then, if you count tokens rather than types, it's nowhere near half. What that means is, rather than count the number of Latin-based words in the dictionary, you count a word each time it appears in use rather than only once. More common words tend to be Germanic rather than Latin, so the number of Latin-based words in any text is not likely to be half. Take my first paragraph: based on my instincts regarding the words' origins, only 10 of the 61 words are Latin-based. Hardly half. And then, words are so far from being the whole of language it's simply preposterous to say such a thing when our grammar is basically not Latin in any way whatsoever. So this is why I disagreed with the notion that knowing Latin improves one's English: a) There aren't really any similarities and b) it assumes that some people speak 'bad' English, and as a linguist I have to be a little bit forceful about making sure people know that there is no such thing (only inappropriate styles).

So, anyway, today I went on good old Amazon's 'look inside' thingy. And whaddaya know, I agreed with a lot of it (except that part about the influence of Latin on English). Let's look at what he says are the benefits of learning Latin:

1. 'To know the source of any word is to understand it better'. Well, in some ways this is not true. Take a word like 'foliage'. Does it help you to know that it comes from the French for 'leaf'? I'm not sure it adds much. What about a word that's changed a lot over time, like 'nice'? It doesn't really help you use it any more accurately if you know that it comes from the Latin word for 'ignorant, unaware', although is is interesting and telling people this kind of this in detail has the nice side effect of getting rid of unwanted company at parties. And what about 'enormity', which is so misused, if you believe the pedants? Well, it has the meaning of a terrible crime, but is supposedly used wrongly by people who don't know any better to mean 'hugeness'. Do you know where it comes from? The Latin word for 'hugeness'. Clearly, we cannot base modern usage on etymology. But on the other hand, it might make your vocabulary more nuanced if you know the origin of some words. Gwynne gives the example 'radically', which comes from the Latin radix, 'root'. 'Radically' therefore means 'from the roots', which doesn't help me use it any better but might help me to think of it when I need a work that means that. 
2. Translation from English to Latin requires you to reorganise the sentence, for which you need to understand how the sentence was put together. This is absolutely true and I have no quibble with it - it is good to understand how the mechanics of grammar works, and it does help you to write better sentences. 
3. You must revise your translations thoroughly so they read well. Also true. This might make your writing more elegant, as you will be practising writing elegant translations and editing prose. However, the Latin is merely a means of doing this, and is not essential to the process. You might just as well translate any language, or even simply rewrite English passages. 
4. The meaning of the Latin words and phrases we use in English will be 'very much clearer than if you were to rely solely on what a dictionary says as to their meaning'. Possibly. We don't use them all literally, so it won't always help, but I imagine it helps to remember what they mean if you recognise them, rather than simply having to memorise them. For instance, there's a mnemonic to get 'eg' and 'ie' the right way round, but I can never remember the mnemonic. I do know, however, that 'eg' stands for exempli gratia 'for example' and 'ie' stands for id est 'that is', so I get them right. (Side note: never follow 'eg' examples with 'etc' if you're writing an essay for me.) 

One benefit I'd add which Gwynne doesn't mention is spelling. I think if you know the root, you are more likely to spell certain words right. For instance, 'separate' is commonly misspelt 'seperate', but it comes from the Latin parare. Since I learnt that, I've never spelt it wrong.

So, unexpectedly, I'm broadly in agreement with Gwynne for once. Everyone should learn Latin right away.

We can't finish on such a positive note though. People will think I'm going soft. Here, have some criticism: Gwynne makes a number of mistakes regarding linguistic facts. Even without looking for them, I spotted two glaring inaccuracies, both, I think, a result of paraphrasing Wikipedia without understanding it properly and therefore introducing error. Here's the first:

Latin is the direct ancestor of, between them, the five so-called Romance languages (Italian, French, Spanish, Portuguese and Romanian) of the largest European language group, and of both of the official South American languages (Spanish and Portuguese).

Even ignoring the fact that it's odd to list Spanish and Portuguese twice (we could give him the benefit of the doubt as there are many differences between the European and South American varieties), this is a strange sentence. Those five are the Romance languages with the most speakers, but he implies that they comprise an exhaustive list of the Romance languages, which is not true. There are lots more. He also implies a contrast between 'the largest European language group' and the South American languages, but the language group he refers to is presumably Indo-European, and this is a classification based on relatedness, not geography, so Spanish and Portuguese are part of it regardless of where they are spoken. And perhaps the biggest inaccuracy of all here is his use of the phrase 'both of the official South American languages', which can only mean that there are two official South American languages, and those two are Spanish and Portuguese. This is wildly wrong: English, Dutch and French are all official languages in South American countries, and oh yes, there are lots and lots of indigenous languages, which Gwynne, in the grand tradition of the superior white man, overlooks. A brief glance at Wikipedia lists Quechua, Aymara and Tupi Guarani in Bolivia, Guarani in Paraguay, more than 60 indigenous languages with official status in Colombia, and so on. None of this is relevant to Gwynne's point, but getting the facts so wrong doesn't fill the reader with confidence about his linguistic expertise.

Here's the second, in a footnote:

The language that Latin replaced [in Britain] was the Celtic language.

No. It wasn't. There is no such thing, and there wasn't when Latin was around either. There are several Celtic languages, and although they all derive from a common ancestor, that was a heck of a long time before Roman Britain. I mean, really, the fact that I'm only giving Wikipedia references here tells you how easy it would have been to check a few of these claims.

[Update: I noticed that later in the book, he says that 'no modern language comes close to approaching Latin in difficulty'. I can't even be bothered to explain why this is idiotic, so I'll just invite readers to share their contenders for 'a language more difficult than Latin' (for an English speaker, presumably, as difficulty of learning varies depending on your native language).]

What came upon a midnight clear?

On Thursday, Jeremy Paxman asked this question in the Christmas university challenge: 'according to the carol, what came upon a midnight clear?'. Stephen Bayley, who as my parents noted made a bit of a tit of himself otherwise, correctly said 'it'. This was not accepted as the right answer, however. What Paxman wanted was 'that glorious song of old'. Here's the lyrics:

It came upon a midnight clear,
That glorious song of old

'That glorious song of old' is, in fact, the understood subject of the preceding clause. But it's extraposed (put outside) and the grammatical subject is 'it'. 'It' stands in here for a subject that is too long or (in this case) doesn't scan in the usual subject position.

Harsh, Paxo, very harsh.

UPDATE [23/12/12]: One of the twitterlinguists, @drswissmiss, pointed out the following:

@linguist_laura right dislocations are kind of special though, so the song is the subject in some ways and it is the co-referential pronoun
— Mercedes D (@drswissmiss) December 21, 2012

And she is, of course, correct. As I was typing the post I was thinking about the status of it - and one thing it isn't is an expletive. An expletive subject is something that English (among other languages) does when it just needs to fill in a gap. If you take a suitably seasonal verb like snow, then in Italian you can say nevica, 'snows', whereas in English you need to say 'it snows'. What snows? Nothing snows, and the Italian reflects that by not requiring a subject. English has this thing that it has to have a subject, so it bungs in this meaningless filler it. There does the same job in There's no easy way to say this, and you can check: it's not felicitous to ask 'Where's no easy way to say this?', so you're clearly not referring to a place when you say there.

But in the carol, something did come upon a midnight clear: the glorious song of old. So it is not an expletive it, it's a referential pronoun (i.e. a pronoun that refers to an actual thing rather than being a filler). That's why I called it extraposition in the original post. But that's not the full story. Because it does refer to the noun phrase 'that glorious song of old', we can say that it is co-referential with it (it refers to the same thing in the world). This is right-dislocation (putting some noun to the right of where it would normally be). When you do right-dislocation, you leave behind a pronoun to fulfil the grammatical requirements of the language (English's afore-mentioned need for a subject of some kind) and to point you in the right direction to get to the real subject.

You can also get left dislocation, which is pretty common. In some of the Romance languages it's called clitic left dislocation because the noun is moved to the front and a 'clitic' (a kind of pronoun which has certain special properties) is in the place where the noun 'should' be. Here's Spanish (example from Karlos Arregi's paper here):

Estos libros, Juan los leyó ayer.
These books, Juan read them yesterday (literally 'Juan them read'). 

The object of the verb is estos libros 'these books', but it's fronted. This often has the effect of making the noun into a topic or the focus of the sentence. In English we usually do this to contrast with something else ('these books he read, but those other ones he hasn't started'), but in Romance languages it's used for more purposes. Some languages, known as topic-prominent languages and including many (or all?) sign languages, put the topic of every sentence at the beginning (usually without a clitic though) simply to indicate that it is the topic of the sentence.

Back to the carol. I don't actually know enough about the syntax of dislocation to say whether the dislocated noun phrase is 'really' the subject of the clause. There is some evidence that the pronoun and the noun phrase are connected in this type of construction: for instance, if the language has case inflections on nouns, they may have the same case ending. (Non-linguists, move on to the next sentence; for linguists, here's a thesis chapter arguing that the pronoun in Czech left dislocation is a spelled out copy of the dislocated element.) OK, non-linguists, back in the room.

To settle this question once and for all, I'm going to refer to playground logic:

'Antidisestablishment' is a very long word. How do you spell it?

Answer: I T.