[From Bruce Nevin (2003.09.19 11:28 EDT)]
You may well have seen this little exercise, which has proliferated
explosively in email and in blogs, beginning about September 12.
PRETTY WEIRD
Aoccdrnig to a rscheearch at Cmabrigde Uinervtisy, it
deosn’t mttaer in waht oredr the ltteers in a wrod are,
the olny iprmoetnt tihng is taht the frist and lsat ltteer
be at the rghit pclae. The rset can be a total mses and
you can sitll raed it wouthit porbelm. Tihs is bcuseae the
huamn mnid deos not raed ervey lteter by istlef, but the
wrod as a wlohe.
As the cousin who passed it along to me said, “Dyxlesics of the
wrlod untie!”
Well, it’s more like: All the letters are present, but only the first and
last letters must be in the correct order. There can be wrong letters
(researcher == rsereearch instead of rscheearch above) or possibly extra
letters (omit the indefinite article “a”, then research ==
rseearch, racseerh, etc.). Even though it probably overstates the case,
nevertheless, this seems to be evidence against the simple model of a
word recognizer as checking off one after another of the phonemes (or
letters) in a fixed sequence. It seems unlikely that the speed and ease
of recognition could be entirely due to semantic expectations. After all,
the entire text is written this way. Comprehension is not delayed, but
begins immediately. (My oldest daughter, Ruby, wrote back that she didn’t
even realize anything was odd until the text told her so.)
It lends credence to a kind of linguistics known as optimality theory
(OT).
http://roa.rutgers.edu/
OT is a selectionist theory, as in Gary Cziko’s book Without
Miracles. On the speech-production side of things, an ‘underlying
form’ (reference perception) for a word is retrieved from memory. A
diverse population of candidate pronunciations is tested against a set of
‘constraints’ (reference perceptions in systems controlling aspects of
speech). The constraints are ranked, so that some have higher priority
(gain) than others. The optimal candidate (least error) is the one that
is produced. This simple framework produces some very sophisticated
results. The literature now is quite large.
A stock intro example: In an OT description of English phonology, the
Onset constraint (a syllable must begin with a consonant) ranks higher
than Faithfulness constraints (produce the underlying form faithfully).
Consequently, before a word such as apple whose underlying form
begins with a vowel, we insert a ‘default’ consonant when it follows a
word that ends in a vowel (and usually when it is utterance-initial), but
not when it follows a word that ends in a consonant. In English, the
default consonant is a glottal stop:
-
Apples are good.
Note the initial glottal stop. (A glottal stop is the initial and medial
sound in the English negation that is usually written
“uh-uh”.)
The pronunciation of a preceding word may in some cases be changed to end
with a consonant when a vowel-initial word like apple
follows: -
An apple a day …
By contrast, final n of an (which has the same historical
origin as one) is not retained when a consonant follows: -
*A bird in hand …*Moving along to the definite article, before a vowel the
rhymes with thee: -
The appleThe effect is that a y sound intervenes before apple, a
consonantal sound even though no consonant letter represents it.
By contrast, the vowel of the rhymes with the vowel of cup
when a consonant-initial word follows: -
The cupIn those dialects – often perceived as somewhat uncouth – in which
one says
“Thuh apple”, a glottal stop intervenes.
In many of the languages that rank Onset over Faithfulness, the ‘default
consonant’ is not a glottal stop. For example, it is t in Tagalog
(an important language in the Philippines). Yet other languages coalesce
the vowels into one, or elide one of the two vowels. Sanskrit and Greek
are examples, and to some degree this was also formerly heard in English,
e.g. Shakespeare, “Oh, let me see’t!”
Languages which, on the other hand, rank Faithfulness constraints above
the Onset constraint do not insert a default consonant between a
word-final vowel and a word-initial vowel, but instead allow a syllable
to begin without a consonant in its onset. An example would only be
convincing if you knew the language.
There has also been some work in this theory on the question of how the
underlying forms (reference perceptions) of words are learned (or
‘acquired’). It explains how it can be that young children can have what
appear to us to be oddly scrambled pronunciations, and apparently not
notice the discrepancy between what they produce and what they hear from
others. Examples from one 2.5 year old:
Child Adult
----- -----
gichys chicken
nowsman snowman
dans stand
This sort of phenomenon, and the variability of it from one child to
another, is hard to explain otherwise.
This seems consistent with many control systems controlling in parallel.
It obviously is not scrambling of the sort proposed above for written
English (that may in part be an artifact of the arbitrariness of English
spelling, which puts some distance between the written word shape and the
corresponding spoken word shape), but it may be related. A few salient
points from web discussions (refs below):
-
The faster you read the less trouble you have with it.
-
Breaking up a letter cluster that represents a single phoneme (ch, sh,
ow, etc.) is more problematic. (Indeed, the initial example above is
pretty respectful of syllable and morpheme boundaries.) -
It doesn’t work with inflected languages (e.g. German, Russian) if the
inflectional ending is scrambled into the middle of the word.
“Uncle Jazzbeau” may have tracked down the source
(http://www.bisso.com/ujg_archives/000224.html).
Here’s a quote:
···
ø¤º°°º¤ø,¸¸,ø¤º°°º¤ø,¸ begin quote
¸,ø¤º°°º¤ø,¸¸,ø¤º°°º¤ø
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
The original reference is [in] a letter to New Scientist magazine by
Graham Rawlinson of Aldershot, Hampshire (vol 162 issue 2188 - 29 May
1999, page 55) titled “Reibadailty”.
Rawlinson writes: ‘You report that reversing 50-millisecond segments of
recorded sound does not greatly affect listeners’ ability to understand
speech (In Brief, 1 May, p 27).
'This reminds me of my PhD at Nottingham University (1976), which showed
that randomising letters in the middle of words had little or no effect
on the ability of skilled readers to understand the text. Indeed one
rapid reader noticed only four or five errors in an A4 page of muddled
text.
'This is easy to denmtrasote. In a puiltacibon of New Scnieitst you could
ramdinose all the letetrs, keipeng the first two and last two the same,
and reibadailty would hadrly be aftcfeed. My ansaylis did not come to
much beucase the thoery at the time was for shape and senqeuce
retigcionon. Saberi’s work sugsegts we may have some pofrweul palrlael
prsooscers at work.
'The resaon for this is suerly that idnetiyfing coentnt by paarllel
prseocsing speeds up regnicoiton. We only need the first and last two
letetrs to spot chganes in meniang.
‘This was not easy to type!’
The letter is in New Scientist’s searchable online archive
(archive.newscientist.com)
ø¤º°°º¤ø,¸¸,ø¤º°°º¤ø,¸ end quote
¸,ø¤º°°º¤ø,¸¸,ø¤º°°º¤ø
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
At
http://www.bisso.com/ujg/
he says where the dissertation can be found.
LanguageHat has an extended discussion with some good stuff in it:
http://www.languagehat.com/archives/000840.php
Also Slashdot
http://science.slashdot.org/article.pl?sid=03/09/15/2227256&mode=thread&tid=133&tid=134&tid=186
These three sites are now mentioned on the Snopes folklore site:
http://www.snopes.com/language/apocryph/cambridge.asp