I took a couple of weeks off recently for my summer holidays during which I started reading an “airport book” (picked up at W.H. Smith’s in the new Heathrow Terminal 5 under one of those ubiquitous “buy one get one half price” deals also offered by Waterstones, Blackwells and Borders throughout the UK — even my local Tesco supermarket offers 50% discount on trade paperbacks). It is called The Black Swan by Nassim Nicholas Taleb (Penguin Books, 2007), and what attracted me to shell out my 6 pounds (sorry, readers in Australia) was the subtitle The Impact of the Highly Improbable and the blurb:
“This book is all about Black Swans: the random events that underlie our lives from bestsellers to world disasters. Their impact is huge: they’re nearly impossible to predict; yet after they happen we always try to rationalise them.”
Taleb is currently Dean’s Professor in the Sciences of Uncertainty at the University of Massachusetts, Amherst, and has a background in probability theory, the study of empiricism and randomness, and Wall Street trading. A nice break from linguistics and endangered languages I thought.
Taleb’s main thesis is that there are certain discoveries (“Black Swans”) which are entirely unexpected (“outliers”) but which have a major impact on beliefs and theories of the world that require post-hoc revisions to accumulated wisdom, attempting to make the discovery explainable and predictable. His writing style is rather egotistical, repetitive and dressed up in pop jargon for my taste (and, as I found out when I had finished the book, for other reviewers), however he does make a number of interesting points. One of these is a contrast between two contexts, what he calls “Mediocristan” and “Extremistan”, as set out in his Table 1 (page 36) which I partially reproduce here:
|Mild randomness||Wild randomness|
|Most typical member is mediocre||Most “typical” is either giant or dwarf, ie. there is no typical member|
|Corresponds (generally) to physical quantities||Corresponds to abstract elements, eg. numbers|
|Total is not determined by a single instance or observation||Total can be determined by a small number of extreme events|
|Short-term observation identifies trends||Long-term observation required|
|Routine, obvious, predictable||Accidental, unseen, unpredictable|
|History crawls||History makes jumps|
|Events are distributed according to the “bell curve” or its variants||Events are either Mandelbrotian (tractable scientifically) or totally intractable|
The context of what Taleb calls “Mediocristan” includes such things as height, weight, calorie consumption, income for a dentist, or mortality rates. The context of “Extremistan” includes (ultimately socially constructed) values like wealth, number of book sales, number of references on Google, commodity prices, and populations of cities (page 35). To help understand the contrast, Taleb gives an example: if we take 1000 random people and calculate their weights, we can identify a range of values and a total — addition of one further individual (even the heaviest person on the planet) will make little difference to the total or range. If we look at their wealth, on the other hand, addition of a single individual, eg. Bill Gates, can result in an unpredictable jump (as Taleb surmises, Bill Gates’ wealth will represent 99.9% of the new total, with all the others representing “no more than a rounding error for his net worth, the variation of his personal portfolio over the past second” (page 33)). The same would be true for book sales, and the subsequent addition of J.K. Rowling to the group.
What about language? Taleb mentions number of speakers and token frequencies of vocabulary items as being in the domain of “Extremistan” — speaker numbers vary wildly with extreme outliers (Chinese with 1,200 million, other languages with 1 or 2), and as well known from corpus linguistics, a small number of word forms in any language are highly frequent while others can be vanishingly rare.
It seems to me that if Taleb’s thesis is correct (and I have not been able to do justice to all the complexities of his arguments here), it has a further application in the realm of endangered and under-documented languages. It could form the basis for an (attractive) epistemological argument to respond to the question (which I have been frequently asked by members of the general public, at least) “Why study under-documented and endangered languages?” This argument can stand beside, or instead of, the intangible cultural heritage arguments promoted by Unesco, among others (that have been criticised for their fundamental neo-Whorfianism). Under-documented languages are potentially the domain of Black Swans, discoveries that are outliers in terms of currently constructed typologies (formalised or not) of human language (one thinks of extremes of phoneme inventories seen in small languages like !Xóõ or Rotokas, for example). This argument would provide a potential philosophical underpinning for the famous quotation by Martin Joos that “[L]anguages can differ from each other without limit and in unpredictable ways.” (Martin Joos (ed.)1957 Readings in linguistics: the development of descriptive linguistics in America since 1925. Washington: American Council of Learned Societies)