Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

Zwicky's Trifecta of Illusions

18 Post author: thomblake 17 July 2009 04:59PM

Linguist Arnold Zwicky has named three linguistic 'illusions' which seem relevant to cognitive bias. They are:

  1. Frequency Illusion - Once you've noticed a phenomenon, it seems to happen a lot.
  2. Recency Illusion - The belief that something is a recent phenomenon, when it has actually existed a long time.
  3. Adolescent Illusion - The belief that adolescents are the cause of undesirable language trends.

Zwicky talks about them here, and in not so many words links them to the standard bias of selective perception.

As an example, here is an exerpt via Jerz's Literacy Weblog (originally via David Crystal), regarding text messages:

  • Text messages aren't full of abbreviations - typically less than ten percent of the words use them. [Frequency Illusion]
  • These abbreviations aren't a new language - they've been around for decades. [Recency Illusion]
  • They aren't just used by kids - adults of all ages and institutions are the leading texters these days. [Adolescent Illusion]

It is my conjecture that these illusions are notable in areas other than linguistics. For example, history is rife with allusions that the younger generation is corrupt, and such speakers are not merely referring to their use of language. Could this be the adolescent illusion in action?

So, are these notable biases to watch out for, or are they merely obvious instances of standard biases?

Comments (26)

Comment author: dclayh 18 July 2009 04:00:27AM *  5 points [-]

I think the Adolescent Illusion could be generalized simply by removing the word "language".

Comment author: thomblake 18 July 2009 04:33:58AM 0 points [-]

Agreed. My conjecture was that there were similar moves to be made for each of them.

Comment author: eirenicon 17 July 2009 08:16:26PM 4 points [-]

The frequency illusion seems to be the recency effect in the form of the Baader-Meinhof phenomenon. Amusingly, if not ironically, this is the third time this week I have had cause to reference the Baader-Meinhof phenomenon, having not thought of it for months prior.

Comment author: thomblake 17 July 2009 09:07:30PM 2 points [-]

Baader-Meinhof phenomenon

It appears several times on this very page - uncanny!

Comment author: Psychohistorian 19 July 2009 02:31:07AM *  0 points [-]

Text messages aren't full of abbreviations - typically less than ten percent of the words use them. [Frequency Illusion]

It would be interesting to see where this data comes from. Teens are probably a minority of texters. The older people get, I would bet the less likely they are to use abbreviations, and the less likely they are to use mostly abbreviations. If you also count business-generated text messages, it would be pretty easy to dilute the extremely heavy use of text messages by a subgroup. Also, "typically less than 10% of words use them," is highly ambiguous; if it's per-distinct-word rather than per-word-used, you could have 10% of words (I, you, can, later, soon, etc.) being used, say, 60% of the time.

These abbreviations aren't a new language - they've been around for decades. [Recency Illusion]

This doesn't really convince me. Yeah, the Atkins diet was invented in 1972. But you probably never heard of it until about 2002, when it actually became popular. Calling it a new diet in 2002 seems legitimate to me. The fact that some guy in the 40's used "l8r" because he wanted a unique license plate doesn't seem relevant to teenagers using it now texting; when people think of it as a recent phenomenon, they are thinking of its prevalence, not whether someone, somewhere used it once. I've seen no evidence that many texting abbreviations have been prevalent before now, even if they have been used occasionally. The other examples on the Wikipedia page make more sense; I'd be more inclined to believe they've been in common use for a while, but people deploring them now claim they're a recent development.

Comment author: thomblake 19 July 2009 02:58:14AM 0 points [-]

It's funny how many of the comments focused on these examples. I probably should have written up more about the 'text messages' angle; I was just using that as one example of someone referring to Zwicky's 'illusions'.

I'm not sure if you looked at Crystal's blog entry I linked to above. He spells out some of the interesting bits, albeit anecdotally. Presumably, the book goes into much more detail. For example:

For every one instance of u, there are nine of you, they found.

Also, I can't seem to find the reference at the moment, but I was recently reading a list of common abbreviations used in letter writing long before texting, like SWAK for "Sealed with a kiss".

Comment author: astray 17 July 2009 05:36:14PM 0 points [-]

I will note a shortcoming in Jerz's analysis - whether or not kids are the leading texters, they may be the leading originators of undesirable language trends.

The adolescent illusion seems tied to representativeness, with perhaps a tinge of in/out groupness.

The frequency and recency illusions show up in cases like the Baader-Meinhof phenomenon.

Comment deleted 17 July 2009 07:31:55PM [-]
Comment author: Emily 17 July 2009 09:58:23PM 4 points [-]

Many people seem to think virtually any kind of language trend is undesirable. They perceive practically all change as degeneration. Needless to say, this attitude is considered completely baseless by actual linguists.

Comment author: dclayh 18 July 2009 03:57:02AM 3 points [-]

Many people seem to think virtually any kind of language trend is undesirable.

If true, this is sad. Personally I applaud language changes that increase complexity (e.g. neologisms) and disparage ones that reduce it (words or grammatical constructions falling out of use; two words that mean different things being used to mean the same thing).

Comment author: MichaelBishop 18 July 2009 05:08:43PM 2 points [-]

Not all language changes that increase complexity are useful. In many cases, the use of particular words, or ways of speaking, can best be understood as wasteful signaling.

Comment author: dclayh 18 July 2009 05:25:54PM 1 point [-]

True. But the purpose of language is not merely to be useful, but also beautiful and fun.

Comment author: thomblake 18 July 2009 03:58:48AM 1 point [-]

Needless to say, this attitude is considered completely baseless by actual linguists.

There are different kinds of linguists, and in my experience your generalization is incorrect. France, for instance, has an entire government agency of linguists devoted to resisting change in the French language.

Comment author: SoullessAutomaton 18 July 2009 10:55:56AM *  5 points [-]

The distinction drawn here is linguistic descriptivism vs. prescriptivism. If we take "actual linguist" in this context to mean scientific-minded academic researchers, the field largely requires itself by nature to be descriptive--they're attempting to document and describe actual human behavior "in the wild". It is, typically, not the business of scientists to be dictating terms to reality (they leave that to the engineers). As Emily points out, L'Académie française doesn't seem to contain a single academic linguist.

I suspect this is mostly a disagreement on the definition of "linguist".

Some academic linguists further take an active stand against prescriptivism, even outside the context of their field. Language Log has of course discussed the issue, such as some recent posts here and here, which you've probably already read, given that you linked to Language Log in the main post.

Comment author: thomblake 18 July 2009 04:52:07PM 0 points [-]

Indeed. If someone were to say, "There are no prescriptivist linguists" I'd take that to be a false statement. I've known at least one academic who argued in favor of prescriptivism, and more are out there (thus, the need for academic linguists to argue against prescriptivism).

Comment author: SoullessAutomaton 18 July 2009 05:53:32PM 3 points [-]

It is a rule of thumb that at least one academic somewhere will sincerely hold any possible position you can think of. Nevertheless, a prescriptivist approach is a fringe position in the field of linguistics, if only because it is usually incompatible with a scientific approach to the subject.

Comment author: Emily 18 July 2009 09:15:49AM *  4 points [-]

The French Academy is looked down on as being conservative almost to the point of absurdity by just about every linguist I've read on the topic.

ETA: I just skimmed through the Wikipedia article on the topic, which gives this:

although most academicians are writers, one need not be a member of the literary profession to become a member. The Académie has included numerous politicians, lawyers, scientists, historians, philosophers, and senior Roman Catholic clergymen.

Not a linguist mentioned.

Comment author: thomblake 18 July 2009 04:56:55PM 0 points [-]

Currently there are a number of philosophers and one philologist, but it's primarily composed of writers. 'Linguist' isn't necessarily a designation every student of language would take.

Comment author: djcb 18 July 2009 11:59:49AM 0 points [-]

But to what extent is the value judgment of linguists about language change more important than that of other language users?

Comment author: SoullessAutomaton 18 July 2009 12:10:42PM 2 points [-]

It's not. Academic linguistics attempts to avoid value judgments entirely and rely on observations of how typical users of a language actually communicate.

Value judgements on language use are typically the province of high-status individuals who use language professionally (e.g., writers) or who teach language to others.

Comment author: komponisto 18 July 2009 01:09:35PM 1 point [-]

Academic linguistics attempts to avoid value judgments entirely and rely on observations of how typical users of a language actually communicate.

Which may of course include studying the value judgments made by language users (sociolinguistics) .

Comment author: thomblake 17 July 2009 07:39:02PM 1 point [-]

In this case specifically, David Crystal is arguing against folks who bemoan the effect of text messages on English. An example of what they might complain about: "I H8 U jk LOL. B @ my plc @ 9". Haters think that this will destroy the English language and civilization as we know it, while Crystal points out that these sorts of conventions are not new, children do not try to use this sort of language on school papers for the most part, and texting seems to aid in literacy.

Comment author: SilasBarta 17 July 2009 09:47:18PM 1 point [-]

Well, he might not want to point to such a text message in his defense, because it would work against his claim that less than 10% of the words are abbreviated!

Seriously, only 10%?

Comment author: thomblake 18 July 2009 03:48:02AM 1 point [-]

It sounds plausible to me, depending on what's being communicated. There aren't really a lot of words with convenient shorthand. Though you can usually get away with leaving out vowels.

Comment author: astray 17 July 2009 07:46:42PM 0 points [-]

I, admittedly, haven't read enough of the posts to know the specific cases, but I presume uptalk and quotative like rank highly.

The trends themselves are secondary, I was mostly just commenting on the supposed one to one mapping between adults being the main texters and adolescents being the supposed main originators of the trends. As thomblake notes, this may merely be an artifact introduced in the noise of reproduction and reporting.

Comment author: thomblake 17 July 2009 06:55:41PM 1 point [-]

I will note a shortcoming in Jerz's analysis

To clarify, that's actually from an article written by a reporter after having interviewed Crystal. I don't have a reference to Crystal's work handy to determine whether his claim would have the same form. Given that his aim was the study of language, I would guess that he had originally been referring to origination of 'undesirable' trends in texting by adults.

Comment author: thomblake 17 July 2009 05:27:53PM 0 points [-]

The Wikipedia article for Adolescent Illusion seems to list a few others that might be relevant, but I didn't include them in the article because I was unsure of the sources and they don't seem to be originated in the same context.