In reading Richard Dawkins' The Greatest Show on Earth, I noticed that  Prof. Dawkins is rather fond of the word literally, using it 38 times in the roughly 130,000 words of the cited work, for a rate of about 292 per million words. This is more than eight times greater than the overall rate of about 35 per million words in the COCA corpus, and 15 times greater than the rate of 15 per million words in the British National Corpus.

Most of Prof. Dawkins' uses of literally do not mean "word for word" or "in a literal as opposed to figurative way", but instead are a sort of intensifier. This is not at all surprising, since the emphasizing sense has been the commonest meaning of literally for a century or more, and Richard Dawkins is a very emphatic person. But all the same, I doubt that the legions of peevers who believe that literal should only be used to mean "not figurative" will even notice Prof. Dawkins' usage, much less work themselves into a froth over it. That's because his usage occupies a sort of middle ground, whose inconsistency with the "word for word" and "not figurative" meanings is subtle rather than blatant.

Consider how the entry for literally in Merriam-Webster's Dictionary of English Usage analyzes the semantic drift of literally. This narrative, which is not as well known as it deserves to be, follows the Oxford English Dictionary's entry through four stages.

The first … means "in a literal manner; word for word": the passage was translated literally. The second … means "in a literal way": some people interpret the Bible literally. The third … could be defined "actually" or "really" and is used to add emphasis. It seems to be of literary origin. […] The purpose of the adverb in [these] instances is to add emphasis to the following word or phrase, which is intended in a literal sense. The [fourth,] hyperbolic use comes from placing the same intensifier in front of some figurative word or phrase which cannot be taken literally.

In one instance, Prof. Dawkins uses literally in the first of this succession of meanings, "word for word" (or here "morpheme for morpheme"):

Theologians worry about the problems of suffering and evil, to the extent that they have even envented a name, 'theodicy' (literally, 'justice of God'), for the enterprise of trying to reconcile it with the presumed beneficence of God.

In a few cases, Prof. Dawkins uses literally with the second of these meanings, "in a literal (as opposed to a metaphorical, figurative or symbolical) way". In these cases, we could replace "literally" with "non-figuratively" or "non-symbolically" and retain the sense, if not the style:

[The enlightened bishops and theologians] may add witheringly that, obviously, nobody would be so foolish as to take their words literally. But do their congregations know that? How is the person in the pew, or on the prayer-mat, supposed to know which bits of scripture to take literally, which symbolically?

But in the great majority of the 38 examples in this book, Dawkins uses MWDEU's third sense of literally, meaning something more like "truly" or "precisely" — or simply "pay attention, now!":

A tree-ring clock can be used to date a piece of wood, say a beam in a Tudor house, with astonishing accuracy, literally to the nearest year. […]

And the amazing thing about dendrochronology is that, theoretically at least, you can be accurate to the nearest year, even in a petrified forest 100 million years old. You could literally say that this ring in a Jurassic fossil tree was laid down exactly 257 years later than this other ring in another Jurassic tree!

It's true that he means these chronological assertions to be literal rather than figurative, but it would be weird to substitute "non-figuratively" or "non-symbolically" or "non-metaphorically":

?A tree-ring clock can be used to date a piece of wood, say a beam in a Tudor house, with astonishing accuracy, non-figuratively to the nearest year. […]

?And the amazing thing about dendrochronology is that, theoretically at least, you can be accurate to the nearest year, even in a petrified forest 100 million years old. You could non-symbolically say that this ring in a Jurassic fossil tree was laid down exactly 257 years later than this other ring in another Jurassic tree!

In the many cases where Prof. Dawkins uses literally to emphasize a negative, substituting a literal reference to non-figurative meaning would be even stranger:

?It is a fact that non-metaphorically nothing you could remotely call a mammal has ever been found in Devonian rock or in any older stratum. They are not just statistically rarer in Devonian than in later rocks. They non-figuratively never occur in rocks older than a certain date. […]

?There are non-symbolically no trilobites above Permian strata, non-metaphorically no dinosaurs (except birds) above Cretaceous strata.

In such examples, there is no opposition is between a literal meaning and a figurative or metaphorical of symbolic one. Instead, the role of literally is to emphasize the absolute truth of the negative.

When literally is weakened to "almost literally", it would be stranger still to substitute a denial of figurative, metaphorical, or symbolic intent:

The number of individual birds in these flocks can run into thousands, yet they almost literally never collide.

?The number of individual birds in these flocks can run into thousands, yet they almost non-metaphorically never collide.

Here it seems that literally adds nothing to the truth-conditional meaning: "they almost literally never collide" seems to be true in just the same set of circumstances that "they almost never collide" is. The only effect of literally in this sentence is to emphasize that the statement is true.

It's easy to see how this purely-emphatic sense of literally turns into the hyperbolic sense, as it did by 1839 when Dickens wrote in Nicholas Nickleby:

His looks were very haggard, and his limbs and body literally worn to the bone, but there was something of the old fire in teh large sunken eye notwithstanding, …

And also:

"Lift him out," said Squeers, after he had literally feasted his eyes in silence upon the culprit. "Bring him in; bring him in."

The second of these quotes is in the MWDEU entry, as is the fact that the 1903 OED included a note by the editor, Henry Bradley, to the effect that literally is

Now often improperly used to indicate that some conventional metaphorical or hyperbolical phrase is to be taken in the strongest admissible sense.

It seems to me that the last quotation from Richard Dawkins, the one about how flocking birds "almost literally never collide", helps us to understand the transition from emphasis to hyperbole.  Consider this passage from a recent restaurant review (Sam Sifton, "A Modern Italian Master", NYT 9/28/2010):

21. ### Ran Ari-Gur said,

March 6, 2011 @ 4:26 pm

The "literally = without exaggeration" explanation, aside from being so obviously correct that I don't understand how Dr. Liberman can quibble with it, also has the virtue of accounting quite nicely for "almost literally". Notice that "almost without exaggeration" is also well attested, with the same sense.

22. ### Sid Smith said,

March 6, 2011 @ 4:28 pm

@JL

;-))

True! Tho it perhaps underlines our point that 'almost literally' is a matter of fact (in this case of kitchen economics) rather than of grammar. So 'almost literally' can be acceptable, provided that the point of fact is *almost* true.

Perhaps we could agree on: 'The truffle was almost literally worth its weight in gold.'

23. ### Andrew (not the same one) said,

March 6, 2011 @ 5:33 pm

I think one upshot of this discussion is that 'literally', even when understood literally, is not equivalent to 'non-metaphorically', since it also implies 'non-hyperbolically'. If I say 'I've been waitng for you for years', when in fact I have been waiting only half an hour, what I say is clearly not meant literally, but it would be odd to call it metaphorical.

24. ### Jahi Chappell said,

March 6, 2011 @ 5:37 pm

Anyone else find Diane's statement "but there's really no other word that means "literally", in the sense of "actual"" amusingly self-contradictory? And it's followed with "So if you use "literally" to mean actually…"! As John Cowan points out, it seems like "actually" is a substitute that clarifies the ambiguity Diane is literally being driven to distraction about :)

Also, it seems to me in her example that there is a mutual lack of successful communication. She is intent on expressing that there are actually more than 100 cousins; her conversation partner seems to be unconcerned with actual exact numbers but is rather trying to communicate mutuality. Re-phrased, it does make Diane sound, perhaps, a bit neurotic, though for reasons possibly other than she realizes. She seems intent on expressing quite specifically the fact that her husband has more than 100 cousins–her conversation partner seems to miss the importance Diane places on large but specific number, trying to imply mutuality because they have some number of cousins they interpret as being similarly large.

"It's over 100!"
"I know, I have a large family too."
"I'm not just saying large, I'm saying over 100!"
"Yeah, we have like over 20."
"I'm talking 1/2 an order of magnitude more cousins than that!"
"Yes, we both have many cousins."
"But my husband has more than 99, less than 150 of them!"
"That is much larger than normal, just like the larger than normal number in my family."
"$#%%$*!"

25. ### Dan K said,

March 6, 2011 @ 5:45 pm

@myl: A reader would be tempted to interpret this kind of categorical statement as hyperbole because people often use categorical words like "never" and "nothing" when they don't mean them literally. I'm not savvy enough to document this using google, but I'm going to assert that there's a rich history of "our team never wins," "mom never lets us do anything fun," "I got nothing out of that book," "nothing good ever comes out of Hollywood," etc. (The last example, come to think of it, may be literally true.) I think it's more plausible than not that Dawkins wanted to rule out that kind of misunderstanding here, and used the word "literally" for that reason. But we're all entitled to our own judgments of plausibility when it comes to an author's intentions, I guess. The paragraph indeed emphasizes the categorical nature of the evidence, and his use of the word "literally" is a big part of that. After using the dreaded word once, he writes, "They are not just statistically rarer… They literally never occur…" To me, it seems almost impossible to avoid the conclusion that he is desperately pleading with readers not to take the words "nothing" and "never" as mere hyperbole (i.e., here meaning "statistically rarer"). I certainly agree that the use is emphatic, but I can't see how to avoid the conclusion that it's emphatic specifically because it rules out a common figurative use?

I should add that my assertion that "literally" is used here to rule out a hyperbolic reading does not in any way imply that everywhere it's absent is therefore hyperbolic. In many cases, I imagine he omitted it because a hyperbolic reading was unlikely in context, or because he didn't feel it was important to make the distinction. In this paragraph, it's clear enough from the entire paragraph that he was concerned that "nothing" wouldn't be taken literally.

26. ### army1987 said,

March 6, 2011 @ 6:56 pm

What the hell would be wrong with "yet they hardly ever collide"?

27. ### Andrew (not the same one) said,

March 6, 2011 @ 7:02 pm

Jahi Chappell: While in some contexts 'actually' will convey the point – and certainly if you say 'When I say literally, I mean actually' it is fairly clear – this won't work in every context, because 'actually' can also be used in other ways, for instance to contrast with what might have happened or what would have happened in a counterfactual case. In such contexts it can be used metaphorically. (E.g 'If he had had no cousins he might have reacted differently, but actually he has hundreds of cousins'.)

March 6, 2011 @ 7:05 pm

It's interesting how adding yet another modifier might reverse the metaphoric drift. If I say "We literally have hundreds of these things", that means that we have a lot of "these things" and the number might count in the hundreds, although that hardly is certain. If this is changed to "We quite literally have hundreds of these things," suddenly there seems to be a considerably greater certainty concerning the number of "these things" actually counting in the hundreds.

29. ### Jonathan D said,

March 6, 2011 @ 7:39 pm

I, too, have always understood usage like Dawkins as meaning "without exaggeration", and the "almost literallys" as acknowledging exaggeration while insisting that fact of exaggeration can't be relevant. Sure, he's trying to hammer something home, and he uses "literally" where he might not need to, to do that, but how does that take away from the idea that he is using it to meaning not-hyperbolic in particular?

30. ### Dominik Lukes said,

March 7, 2011 @ 4:06 am

If I could offer an addition to the analysis. In my post on this subject -http://metaphorhacker.net/2011/02/literally-triumph-of-pet-peeve-over-matter – I did some collocate hunting and found that "almost" is the third most common word immediately preceding "literally" with "quite" being first. This (and other things) suggest that "literally" is often associated with scalar concepts and therefore as both a potential trigger and disambiguator of hyperbole.

[(myl) Thanks — I missed this when it came out, and I'm glad to read it now.]

31. ### pj said,

March 7, 2011 @ 6:24 am

@army1987
For me, birds that 'hardly ever' collide would still collide significantly more frequently than birds that 'almost literally never' collide. Not you?

32. ### Arnold Zwicky said,

March 7, 2011 @ 7:16 pm

Yet another cartoon, on my blog, here.

33. ### Diane said,

March 8, 2011 @ 5:51 pm

@Jahi Chappell

Actually, when I was writing that post I thought about whether actually could replace literally in that sentence and I literally decided that it could not. :)

Seriously, though, neither "actually" nor "really" would have been natural for me to say in the first sentence in that dialogue. (John Cowan's use of actually also sounds awkward to me.) To me they would both be used later, to add emphasis or clarify, but not the first time you bring up a topic. But perhaps that is just how I talk and others would use it differently.

It's not that there aren't ways to rephrase to get the point across. It's the fact that I said what I meant and I was misunderstood that frustrated me. But Jahi, you're more right than you know (*laughing*) That bit of dialogue came from a conversation about the craziness of marrying into a big family. I was trying to one-up her and she wasn't letting me!

34. ### Ellen K. said,

March 8, 2011 @ 8:13 pm

35. ### Ellen K. said,

March 8, 2011 @ 8:29 pm

@Sid Smith: But the price of gold is way higher than \$100 per ounce.

36. ### Dan H said,

March 10, 2011 @ 4:47 am

In Diane's example above, I tend to agree with Jahi Chappell, it sounds like a fundamental disconnect rather than an actual language issue.

I can only speak personally, but I generally *would* take "literally over a hundred" as meaning you cared more about the specific value (if you were just trying to give the impression of large numbers I'd expect you to say "literally a hundred" or "literally hundreds"). For that matter I'm pretty sure I'd take "over a hundred" without qualification as being more specific about number.

37. ### Jahi Chappell said,

March 20, 2011 @ 11:39 pm

@Dan H: Interesting point–"literally" is possibly an unnecessary embellishment. If I said "I have over one hundred cousins", I would quite expect people to take me at my word. I was going to say that my intonation (inflection? -not-a-linguist-) could possibly clue in whether or not I was being literal, but I *cannot* imagine making the (what I perceive as) inflection pattern hinting that I'm engaging in hyperbole or rough approximations without adding "…,like,…" as in, "I have, like, over a hundred cousins." The raw statement conveys itself–and of course, an exact number is all the better: "My husband has 103 cousins" has absolutely no need of "literally".

Funny that I was getting a similar gist of the actual conversation, which was a bit of one-upmanship. Thanks for the background, Diane!! So I still maintain that this is an example of unavoidable human misunderstandings, not ameliorable by more precise language. A more precise language takes two to tango–not only do you have to mean it precisely, the other person must a) understand the content you're meaning precisely, b) believe that you mean it precisely, and c) care that you mean it precisely. It seems like (c) was missing in Diane's example, and no amount of optimal language design would get by it :)

Ultimately, what irks me about peevologist complaints based on "it [should] convey information more logically" is that what they want is not just a more logical language, but a more logical (and conforming) people who will correspondingly agree to *never mean anything than the given meanings*, where [given] = [what the peevologist assumes it "originally" meant, often shown on Language Log to be a new, not original, meaning]. I just have no desire to live in such a conformist world; I *like* playing with language and layering meaning and implication; using conflicting "literal" meaning and implication to construct irony, hyperbole, or sarcasm, etc. etc. And I've never met a situation where fluid word definitions couldn't be solved by context. Of course, I've also never met a situation where I can't shove my excessive verbiage "down someone's earhole" :p (Anyone wondering what that latter is from, I couldn't remember, but a little Google-fu refreshed my memory of the Ron White routine it came from)

38. ### Jahi Chappell said,

March 20, 2011 @ 11:43 pm

Or rather, there are few situations where I'm able to *contain myself* and stop from shoving unwanted information down others' earholes… :) And excuse the many "slips of the tongue" and accidental word omissions in the above post…

39. ### Rodger said,

June 28, 2011 @ 1:13 pm

An odd sighting of "literally" in today's Guardian: "an American Muslim …, who quite literally has foregone his right to an eye for an eye", referring to Rais Bhuiyan who is actually campaigning for his attacker's death sentence to be commuted. http://www.guardian.co.uk/commentisfree/cifamerica/2011/jun/28/us-death-penalty-victim

40. ### Sadly, Jamie Redknapp is literally correct – Telegraph Blogs said,

March 12, 2012 @ 5:02 am

[…] Women (published 1868), Louisa May Alcott wrote "the land literally flowed with milk and honey." Dickens wrote in Nicholas Nickleby (1839): "His looks were very haggard, and his limbs and body literally worn to the bone", and "'Lift him […]

41. ### H said,

April 23, 2013 @ 9:13 am

Two years late comes:
'In one respect, almost literally so' in the Economist.

42. ### No we haven't 'literally killed' the English language. Or metaphorically killed it. Stand down, semantics nerds – Telegraph Blogs said,

August 13, 2013 @ 5:56 am

[…] Women (published 1868), Louisa May Alcott wrote "the land literally flowed with milk and honey." Dickens wrote in Nicholas Nickleby (1839): "His looks were very haggard, and his limbs and body literally worn to the bone", and "'Lift him […]

43. ### Delightful oddities | MasterWord said,

August 20, 2013 @ 5:56 am

[…] that "literally" has been literally used non-literally for literally centuries: even Dickens said in Nicholas Nickleby (1839): "His looks were very haggard, and his limbs and body literally worn to the bone", and […]