I have a number of pet peeves about how people use language. I am genuinely annoyed by the use of apostrophes before plural of numerals or acronyms like 50’s or ABC’s. But because I understand how language works, I keep my mouth shut. The usage has obviously moved on. I don’t think, ABC’s is wrong or confusing, I just don’t like the way it looks. But I don’t like a lot of things that there’s nothing wrong with. I get over it.
Recently I came across a couple of blog posts pontificating on the misuse or overuse of the word literally. And as usual they confuse personal dislike with incorrect or confusing usage. So let’s set the record straight! No matter what some dictionaries or people who should know better say, the primary function of the word “literally” in the English language is to intensify the meaning of figurative, potentially figurative or even non-figurative expressions. This is not some colloquial appendage to the meaning of the word. That’s how it is used in standard English today. Written, edited and published English! Frequently, it is used to intensify expressions that are for all intents and purposes non-figurative or where the figurative nature of the expression can be hypostesized:
1. “Bytches is literally a record of life in a nineties urban American community.” [BNC]
2. “it’s a a horn then bassoon solo, and it it’s a most worrying opening for er a because it is. it is literally a solo, er unaccompanied” [BNC]
3. “The evidence that the continents have drifted, that South America did indeed break away from Africa for instance, is now literally overwhelming” [BNC, Richard Dawkins]
The TIME magazine corpus can put pay to the non-sense about “literally” as an intensifier being new or colloquial. The use of the word in all functions does show an increase from the 40s, peak in the 1980s and 2000s returning to the level of 1950s. I didn’t do the counting (plus it’s often hard to decide) but at a glance the proportion of intensifier uses is if anything slightly higher in the 1920s than in 2000s:
4. This is almost literally a scheme for robbing Peter to pay Paul. [TIME, 1925]
5. He literally dropped the book which he was reading and seized his sabre. [TIME, 1926]
6. The Tuchuns-military governors are literally heads of warring factions. [TIME, 1926]
But there are other things that indicate that the intensifier use of literally is what is represented in people’s linguistic knowledge. Namely collocations. Some of the most common adverbs preceding literally (first 2 words in COCA) are graded: 1. quite (558), 2. almost (119), 5. so (67), 7. too (54), 9. sometimes (42), 12. more, 15. very, 16. often.
7. Squeezed almost literally between a rock and a hard place, the artery burst. [COCA, SportsIll, 2007]
Another common adverbial collocate is “just” (number 4) often used to support the intensification:
8. they eventually went missing almost just literally a couple of minutes apart from one another [COCA, CNN, 2004]
Other frequent collocates are non-gradual: “up”, “down”, “out”, “now” but their usage seems coincidental – simply to be attributed to their generally high frequency in English.
The extremely unsurprising finding is that if we don’t limit the collocates by just 2 preceding words, by far the most common collocate of literally is “figuratively” (304). Used exclusively as part of “literally and figuratively”. This should count as its own use:
9. A romantic tale of love between two scarred individuals, one literally and one figuratively. [COCA, ACAD, 1991]
But even here, sometimes both possible senses of the use are figurative but one is perceived as being less so:
10. After years of literally and figuratively being the golden-haired boy… [COCA, NEWS, 1990]
11. Mercy’s parents had pulled the plug, literally and figuratively, on her burgeoning romance. [COCA, Fic, 2003]
This brings us to the secondary function (and notice I don’t use the word meaning, here) of “literally”, which is to disambiguate statements that in the appropriate context could have either figurative or literal meaning. Sometimes, we can apply a relatively easy test to differentiate between the two. The first sense cannot be rephrased using the adjective “literal”. However, as we saw above, a statement cannot always be strictly categorized as literal or figurative. For instance, example (2) above contains a disambiguating function although it is not between figurative or non-figurative but rather between two non-figurative interpretations of two situations that it may be possible to describe as a ‘solo’ (one where the soloists is prominent against background music and one where the soloist is completely unaccompanied.) Clear examples are not nearly as easy to find in a corpus, as the prescriptivist lore would have us believe and neither is the figurative part clear cut:
11. And they were about literally to be lynched and they had to save their lives. [COCA, SPOK, 1990]
12. another guy is literally a brain surgeon [COCA, MAG, 2010)
Often the trope does not include a clear domain mapping, as in the case of hyperbole.
13. I was terrified of women. Literally. [COCA, LIFE, 2006]
This type of disambiguation is often used with numerals and other quantifiers where a hyperbolic interpretation might be expected:
14. this is an economy that is generating literally 500,000 jobs because of our foreign trade [COCA, SPOK, PBS, 1996]
15. While there are literally millions of mobile phones that consumers and business people use [COCA, MAG, 2008]
16. “Then literally after two weeks I’m ready to come back,” he says. [COCA, MAG, 2010]
Or sometimes it is not clear whether some vague figure is being intensified or a potential trope is being disambiguated as in:
17. He was the man who lost his wife when his house literally broke apart in the storm. [COCA, CNN, 2005]
These types of examples also sometimes occur when the speaker realizes that what they had previously only intended as an intensified use is an actual disambiguating use:
18. will allow us to actually literally X-ray the universe using these distant objects
Another common usage is to indicate a word for word translation from a foreign language or a component analysis of an etymology of a word. E.g.
19. theory of revolution (literally, an overturning) [BNC].
Sometimes this explanation includes side elaboration as in
20. “Ethnography — literally, textual description of particular cultures” [BNC].
“Literally” also has a technical sense meaning roughly “not figuratively” but that has nothing do with its popular usage. I could not find any examples of this in the corpus.
The above is far from an exhaustive analysis. If I had the time or inclination, we could fine tune the categories but it’s not all that necessary. Everyone should get the gist. “Literally” is primarily an intensifier and secondarily a disambiguator. And categorizing individual uses between these two functions is a matter of degree rather than strict complementarity.
None of the above is hugely surprising, either. “Literally” is a pretty good indicator that figurative language is nearby and a less good indicator that strict fact is in the vicinity. Andrew Goatly has described the language of metaphor including “literally” in his 1997 book. And the people behind the ATT-META Project tell me that they’ve been using “literally” as one of the indicators of metaphoric language.
Should we expect bloggers on language to have read widely on metaphor research? Probably not. But by now I would expect any language blogger to know that to look up something in a dictionary doesn’t tell them much about the use of the word (but a lot about the lexicographer) and the only acceptable basis for argumentation on the usage of words is a corpus (with some well recognized exceptions).
The “Literally Blog” that ran out of steam in 2009 was purportedly started by linguistics graduates who surely cannot have gotten very far past Prescriptivism 101. But their examples are often amusing. As are the ones on the picture site Litera.ly that has great and funny pictures even if they are often more figurative than the phrases they attempt to literalize. Another recent venture “The literally project” was started by a comedian with a Twitter account on @literallytsar who is also very funny. Yes, indeed, as with so many expressions, if we apply an alternative interpretation to them, we get a humorous effect. But what did two language bloggers think they were doing when they put out this and this on “literally”, I don’t know. It got started by Sentence First, who listed all the evidence to the contrary gathered by the Language Log and then went on to ignore it in the conclusion:
Literally centuries of non-literal ‘literally’ « Sentence first. Few would dispute that literally, used non-literally, is often superfluous. It generally adds little or nothing to what it purports to stress. Bryan Garner has described the word in some of its contemporary usages as “distorted beyond recognition”.
Well this is pretty much nonsense. You see, “pretty much” in the previous sentence was a hedge. Hedges, like intensifiers, might be considered superfluous. But I chose to use that instead of a metaphor such as “pile of garbage”. The problem with this statement is twofold. First, no intensifiers add anything to what they intensify. Except for intensification! What if we used “really” or “actually” – what do they add in that “literally” doesn’t? And what about euphemisms and so many other constructions that never add anything to any meaning. Steven Pinker in his recent RSA talk listed 18 different words for “feces”. Why have that many when “shit” would suffice?
Non-literal literally amuses, too, usually unintentionally. The more absurd the literal image is, the funnier I tend to find it. And it is certainly awkward to use literally and immediately have to backtrack and qualify it (“I mean, not literally, but…”). Literally is not, for the most part, an effective intensifier, and it annoys a lot of people. Even the dinosaurs are sick of it.
What is the measure of the effectiveness of an intensifier? The examples above seem to show that it does a decent job. And annoying a lot of prescriptivists should not be an argument for not using it. These people are annoyed by pretty much anything that strikes their fancy. We should annoy them. Non-sexist language also annoys a lot of people. All the more reason for using it.
“Every day with me is literally another yesterday” (Alexander Pope, in a letter to Henry Cromwell)
For sure, words change their meanings and acquire additional ones over time, but we can resist these if we think that doing so will help preserve a useful distinction. So it is with literally. If you want your words to be taken seriously – at least in contexts where it matters – you might see the value in using literally with care.
But this is obviously not a particularly useful distinction and never has been. The crazier the non-intensifier interpretation of an intensifier use of “literally” is, the less of a potential for confusion there is. But I could not find a single example where it really mattered in the more subtle cases. But if we think this sort of thing is important why not pick on other intensifiers such as “really”, “virtually” or “actually” (well, some people do). My hypothesis is that it’s a lot of prescriptivists like the feeling of power and “literally” is a particularly useful tool for subjugating those who are unsure of their usage (often because of a relentless campaign by the prescriptivist). It’s very easy to show someone the “error” of their ways when you can present two starkly different images. And it feels like this could lead to a lot of confusion. But it doesn’t. This is a common argument of the prescriptivist but they can rarely support the assertion with more than a couple of examples if any. So unless a prescriptivist can show at least 10 examples where this sort of ambiguity led to a real consequential misunderstanding in the last year, they deserve to be told to just shut up.
Which is why I was surprised to see Motivated Grammar (a blog dedicated to the fighting of prescriptivism) jump into the fray:
Non-literal “literally” isn’t wrong. That said… « Motivated Grammar Non-literal literally isn’t “wrong” — it’s not even non-standard. But it’s overused and overdone. I would advise (but not require) people to avoid non-literal usages of literally, because it’s just not an especially good usage. Too often literally is sound and fury that signifies nothing.
Again, I ask for the evidence of what constitutes good usage? It has been good enough for TIME Magazine for close to a century! Should we judge correct usage by the New York Review of Books? And what’s wrong with “sound and fury that signifies nothing”? How many categories of expressions would we have to purge from language, if this was the criterion? I already mentioned hedges. What about half the adverbs? What about adjectives like “good” or “bad”. Often they describe nothing. Just something to say. “How are you?”, “You look nice.”, “Love you” – off with their heads!
And then, what is the measure of “overused”? TIME Magazine uses the word in total about 200-300 times a decade. That’s not even once per issue. Eric Schmidt used it in some speeches over his 10-year tenure as Google’s CEO and if you watch them all together, it stands out. Otherwise nobody’s noticed! If you’re a nitpicker who thinks it matters, every use of “literally” is going to sound too much. So, you don’t count. Unless you have an objective measure across the speech community, you can’t make this claim. Sure, lots of people have their favorite turns of phrases that are typical of their speech. I rather suspect I use “in fact” and “however” far too much. But that’s not the fault of the expression. Nor is it really a problem, until it forces listeners to focus on that rather than the speech itself. But even then, they get by. Sometimes expressions become “buzz words” and “symbols of their time” but as the TIME corpus evidence suggests, this is not the case with literally. So WTF?
I just spent some time going after prescriptivists. But I don’t actually think there’s anything wrong with prescriptivism (even though their claims are typically factually wrong). Puristic and radical tendencies are a part of any speech community. And as my former linguistics teacher and now friend Zdeněk Starý once said, they are both just as much a part of language competence as the words and grammatical constructions. So I don’t expect they will ever go away nor can I really be too critical of them. They are part of the ecosystem. So as a linguist, I think of them as a part of the study of language. However, making fun of them is just too hard to resist. Also, it’s annoying when you have to beat this nonsense out of the heads of your students. But that’s just the way things are. I’ll get over it.
Well, I may have been a bit harsh at the blogs and bloggers I was so disparaging about. Both Sentence first and Motivated grammar have a fine pedigree in language blogging. I went and read the comments under their posts and they both profess anti-prescriptivism. But I stand behind my criticism and its savagery of the paragraphs I quoted above. There is simply no plausible deniability about them. You can never talk about good usage and avoid prescriptivism. You can only talk about patterns of usage. And if you want to quantify these, you must use some sort of a representative samples. Not what you heard. Not what you or people like you. Evidence. Such as a corpus (or better still corpora provide.) So saying you shouldn’t use literally because a lot of people don’t like it needs evidence. But what evidence there is suggests that literally isn’t that big a deal. I did three Google searches on common peeve and “literally” came third: +literally +misuse (910,000), preposition at the end of a sentence (1,620,000), and +passive +misuse writing (6,630,000). Obviously, these numbers mean relatively little and can include all sorts of irrelevant examples, but they are at least suggestive. Then I did a search for top 10 grammar mistakes and looked at the top 10 results. Literally did not feature in either one of these. Again, this is not a reliable measure, but it’s at least suggestive. I’m waiting for some evidence to show where the confusion over the intensifier and disambiguator use has caused a real problem.
I also found an elluminating article in the Slate by Jesse Sheidlower on other examples of ‘contranyms’ in English showing that picking on “literally” is quite an arbitrary enterprise.
A bit of corpus fun revealed some other interesting collocate properties of literally. There are some interesting patterns within individual parts of speech. The top 10 adjectives immediately following are:
- TRUE 91
- IMPOSSIBLE 24
- STARVING 14
- RIGHT 9
- SICK 8
- UNTHINKABLE 8
- ALIVE 6
- ACCURATE 6
- HOMELESS 6
- SPEECHLESS 6
- HUNDREDS 152
- THOUSANDS 118
- MILLIONS 55
- DOZENS 35
- BILLIONS 17
- HOURS 14
- SCORES 14
- MEANS 11
- TONS 11
The top 10 numerals (although here we may run up to the limitations of the tagger) are:
- ONE 25
- TWO 12
- TENS 9
- THREE 8
- SIX 8
- 10 7
- 100 7
- 24 6
- NEXT 5
- FIVE 4
There are the top adverbs:
- JUST 91
- OVERNIGHT 24
- ALMOST 19
- ALL 17
- EVERYWHERE 14
- NEVER 13
- DOWN 12
- RIGHT 12
- SO 11
- ABOUT 11
And the top 10 preceding adverbs:
- QUITE 552
- ALMOST 117
- BOTH 91
- JUST 67
- TOO 50
- SO 38
- MORE 37
- VERY 31
- SOMETIMES 30
- NOW 26
One of the patterns in the collocates suggests that “literally” often (although this is only a significant minority of uses) has to do with scale or measure. So I was thinking is it possible that one can use the intensifier literally incorrectly (in the sense that most speakers would find the intensity inappropriate). For example, is it OK to intensify height of a person in any proportion? Is there a difference between “He was literally 6 feet tall” (disambiguator) and “He was literally seven feet tall.” (intensifier requiring further disambiguation) and “He was literally 12 feet all” (intensifier). The corpus had nothing to say on this, but Google provided some information. Among the results of the search “literally * feet tall” referring to people, the most prominent height related to literally is 6 or 7 feet tall. There are some literally 8 feet tall people and people literally taller because of some extension to their height (stilts, helmets spikes, etc.) But (as I thought would be the case) there seem to be no sentences like “He was literally 12 feet tall.” So it seems “literally” isn’t used entirely arbitrarily with numbers and scales. Although it is rarely used to mean “actually, verifiably, precisely”, it is used in proportion to the scale of the thing measured. However, it is used both when a suspicion of hyperbole may arise and where a plausible number needs to be intensified. And most often a mixture of both. But it is not entirely random. “*Literally thousands of people showed up for dinner last night” or “*We had literally a crowd of people” is “ungrammatical” while “literally two dozen” is OK even if the actual number was only 18. But this is all part of the speakers’ knowledge of usage. Speakers know that with quantifiers, the use of literally is ambiguous. So if you wanted to say “The guy sitting next to me was literally 7 feet tall”, you’d have to disambiguate and say “And I mean literally, he is the third tallest man in the NBA.”