Over the years, we've discussed many cross-cultural comparisons based on the "No Word for X" meme. In the most recent LL post on the subject ("No word for integrity?", 12/31/2008), I asserted that
[W]hen someone makes a sociological point by saying that language L has no word for concept C, you'll rarely lose by betting that they're wrong.
But a recent assertion by Bart Wilson seems more promising — the linguistic part is supported by reference to a chapter in a recent book by an actual (and eminent) linguist, and the socio-cultural part is supported by reference to a large body of empirical research, some of which was done by Wilson himself.
Here's Wilson ("Fair's fair", The Atlantic, 1/25/2009):
For the past 25 years experimental economists–of which I am one–have been infatuated with a pie-splitting problem known as the Ultimatum Game. […] The Ultimatum Game is so popular because it is simple to explain and simple to run, yet its results involve one of the most complex problems of society: what are we saying when we say something is "fair"?
Because the word [fair] is so central to how markets work, I've been engaged in a somewhat dizzying struggle to understand it better, or at least clarify how economists apply it to economic behavior. […] Early in my project I turned to one of my favorite linguists, Anna Wierzbicka of Australia National University, for her insight into the meaning of the word. Luckily for me, when I emailed her she pointed out that she had recently published a whole chapter on this very topic in her book English: Meaning and Culture .
Did you know that fair is one-to-one untranslatable into any other language–that it is distinctly Anglo in origin?
The very next day, James Surowiecki came back with the obvious counter-argument ("Is the idea of fairness universal", The New Yorker, 1/26/2009):
The economist Bart Wilson has an interesting post on one of the more important experiments that behavioral economists have run, the ultimatum game. The core insight that economists have gleaned from ultimatum-game experiments is that people care quite a bit about fairness, and are actually willing to turn down free money if they feel that the amount of money they’re being offered is unfair. Wilson’s post, in turn, is a meditation on what, exactly, we mean “when we say something is ‘fair’.” Starting from the assertion that the word “fair” is apparently untranslatable into any other language, Wilson goes on to argue that if this doesn’t mean that fairness is peculiar to Anglophone cultures, it at least suggests that “the concept of fairness is not necessarily ubiquitous to human cultures.”
Let’s set aside the fact that the definition of “untranslatable” is itself hard to pin down, as are its implications for thinking about human feeling. (English doesn’t have a single word that means the same thing as the German schadenfreude, but no one thinks this means that Americans don’t take pleasure when others fail.) What’s more perplexing about Wilson’s argument is that the ultimatum game has been played in many different countries with similar results.
Yesterday, Bart Wilson tried to argue back ("Is fairness cross-cultural, or not?", 1/27/2009):
It's less interesting to me that people nearly universally offer (and accept) more than $1; the benchmark of a game-theoretic automaton is a low standard. […]
I think there's a lot more value to be gained in observing how mean offers vary by context. After translating the instructions and altering the protocols to fit different cultural settings around the world, we observe that mean offers vary from 48% to 26% of the pie. We may all have a sense of fairness, but our sense of what is fair varies pretty dramatically.
As I mention we can also change the experimental protocols for participants right here in the US and observe nearly the same variance, 44% to 28%. The interesting question is, how do people come agree on the tacit social rules for a particular context?
Well, another interesting question would be, is there really any evidence to support the view that differences between languages (as opposed to other cultural differences, or more simply, differences in the way the game is set up) actually affect how people play this game? And then there's the more difficult question of whether the existence of a word "one-for-one translatable" as "fair" makes a critical difference. So far, I haven't seen any evidence for a believable effect of language at all — but perhaps it's out there. We know that linguistic priming can make a moderately large difference in the behavior of bilinguals on certain sorts of psychological tests — roughly half a standard deviation, in some well-designed and controlled experiments. Is this true of the Ultimatum Game? If it is, how much of the effect can be predicted by the distribution of senses among words whose meanings are in the general area of equitable, just, fair, etc.?
Pending an examination of the literature on this point, I'll turn to a couple of more mundane issues that came up in Wilson's original essay. One is peripheral, but easy to test:
If you're anything like most Americans, you probably hear or speak that word many times a day. "It's not fair!" screams the petulant child. "That's a fair price," claims the smooth salesman. "I'm trying to be fair," sighs your boss.
I wondered whether Wilson's estimate of "many times a day" is accurate, so I took a quick look at some corpus frequencies:
|Source||fair count||Total corpus size||1 "fair" in __ words|
|LDC English CTS||1,804||26,151,602||14,496|
|LDC English News||190,063||2,559,992,056||13,469|
|Time Magazine 1923-2006||9,632||100,000,000||10,382|
These estimates — from conversational transcripts and from various kinds of written texts — all agree that a word spelled fair comes up about once in 10,000-15,000 words. Some of these are other uses (like "World's Fair" and "fair complexion" and "a fair amount of trouble" and "fair to poor"), and perusing a sample of the LDC conversational transcripts suggests that these amount to about a quarter of the total (24 out of a random sample of a 100, to be precise). Applying this correction, we'd get roughly one Wilson-relevant fair per 15,000-20,000 words
At a typical average apeaking rate of 150 words per minute, one fair in 15,000 words would be one fair per 100 minutes of talk. Given three to four hours of conversation time per day (180 to 240 minutes), which seems to be about average (though the distribution has a lot of variance), this would be around 2 uses of fair per day. Putting it all together, it seems more accurate to say that "if you're like most Americans, you probably hear or speak that word a couple of times a day". (Depending on how much and how fast you read, you'd get a few more by eye as well as by ear).
The other concrete issue is Wilson's claim about the lexicographical (and socio-cultural) history:
Did you know that fair is […] distinctly Anglo in origin? And a relatively new word at that? (Late 18th century, actually–the industrial revolution apparently also vastly enhanced our capacity to complain.)
The evidence for this historical claim seems weak to me. (And as stated, I think it's at least partly false.) Wilson got it (I think) from chapter 5, "Being FAIR: Another Key Anglo Value and Its Cultural Underpinnings", in Anna Wierzbicka's book English: Meaning and Culture. Wierzbicka's argument is more subtle than Wilson's (in part simply because it's longer), but I'm still not sure that it's really true. And it's clear that Wilson's summary gets some of the facts wrong.
Wierzbicka analyzes the OED's citations for forms of fair and unfair going back to the 9th century, and argues that "the familiar twentieth-century sense of unfair (as, for example, in unfair competition) emerged and spread widely only in the nineteenth century", with the earlier uses that seem similar being in fact a vaguer and more general word of condemnation. This seems to be true, but her argument about fair struck me as much less convincing.
In the OED's entry for fair, the closest thing seems to be sense 4:
Equitably, honestly, impartially, justly; according to rule. Also in phr. FAIR AND SQUARE.
This seems rather close to the sense that Wilson wants, and citations for it go back to 1300:
c1300 Havelok 224 Al was youen, faire and wel.
1603 SHAKES. Meas. for M. III. i. 141 Heauen shield my Mother plaid my Father faire.
1680 OTWAY Orphan II. vii, I can never think you meant me fair.
The entry also cites "fair-dealing" back to 1711:
1711 SHAFTESBURY Charac. (1737) I. 63 There is as much difference between one sort and another, as between fair-dealing and hypocrisy.
1718 Freethinker No. 14. 96 A fair-dealing, honourable Merchant.
Wierzbicka writes that
The earliest examples that sound "modern" in their use of fair are those with the phrase fair play, attested from the end of the sixteenth century onward. In fact, however, even the phrase fair play is misleading as putative evidence for a pre-eighteenth-century emergence of the modern concept of 'fairness': fair play had its opposite in foul play, not in unfair play, whereas fair competition has its opposite in unfair competition. […]
In fact, one can speculate that the notion of 'fair play' as a whole played a significant role in the emergence of the new concept of 'fair': in fair play, the notion of, roughly speaking, rules was part of the meaning of the word play, whereas in fair competition (and other comparable modern collocations), it became part of the meaning of fair itself. To put it differently, the notion of 'fair play' appears to have become absorbed, as it were, in the meaning of 'fair' itself; the meaning of fair, as applied to all types of human interaction, appears to have been built on the model of 'fair play'.
This strikes me as a plausible line of reasoning, but I wonder whether it's entirely true.
In the first place, the relevant sense of "fair play" seems to start about a hundred years earlier than the OED citations than Wierzbicka relies on. A bit of search on LION turns up e.g. Robert Henryson (1430-1506). "The Taill of the Wolf and the Wedder":
2560 'Is this your bourding in ernist than?' (quod he),
2561 'For I am verray effeirit, and on flocht;
2562 Cum bak agane and I sall let yow se.'
2563 Than quhar the gait wes grimmit he him brocht.
2564 'Quhether call ye this fair play, or nocht?
2565 To set your Maister in sa fell effray,
2566 Quhill he ffor feiritnes hes fylit up the way.
Or John Skelton (1460-1529) "The auncient acquaintance, madam, betwen vs twayn":
41 Play fayre play, madame, and loke ye play clene,
42 Or ells with gret shame your game wylbe sene.
Or Anon, 1100-1500, "Robin Hood and the Shepherd" (apparently in modernized spelling):
55 'Arise, arise, thou proud fellow,
56 And thou shalt have fair play,
57 If thou wilt yield, before thou go,
58 That I have won the day.'
Not only are these examples earlier than Wierzbicka's "end of the sixteenth century" — and certainly well before the "late eighteenth century" and "industrial revolution" times periods that Wilson cites — but they're all figurative, applying the notion of "fair play" outside of a literal game-playing context. And whether or not the word fair was used before 1500 in this sense, outside of its collocation with play, Wilson's concept of fairness-with-respect-to-social-norms is clearly present.
I'd be surprised, furthermore, not to find things from the same period in French or Italian or Dutch expressing similar sentiments.
It's also not clear that Tudor and Elizabethan authors would have paired "fair play" with "foul play" as opposed to "unfair play" — even in other uses, Shakespeare was happy to pair "fair" with "unfair" (as in Sonnet 5 "And that vnfaire which fairely doth excell") as well as with "foul" (as in Sonnet 137 "To put faire truth vpon so foule a face").
Wiezbicka's general argument about changes in political culture may be right:
"Fair play" as a model of human interaction highlights the "procedural" character of the ethics of fairness. Arguably, the emergence of the concept of "fairness" reflects a shift away from absolute morality to "procedural (and contractual) morality," and from the gradual shift from "just" to "fair" can be seen as parallel to the shifts from good to right and also from wise (and also true) to reasonable: in all cases, there is a shift from an absolute, substantive approach to a procedural one.
But it would be nice to find a way to make the linguistic arguments — whether about changes across time or across space — in a more testable form.
As George Wither (1588-1667) put it:
VVith these Cards , I, an After-game have play'd,
But, there's one Card , by shuffling, so mislaid,
That, now my fore-game's lost; yet, if I may
Just Dealing find, Fair Gamsters , and Fair-play ,
What ever happens, I no question make
But, I at last, may thereby save my stake ;
And, when the Sett is ended, win much more
Then ere I got by Play , or Work before.
To play the winning linguistic card in a fair game of social science, it seems to me, we need a better way to track usage across time and space in a quantitative way.