|
Last Update: Wednesday, 7 January 2015
Note: |
|
Last Update: Wednesday, 7 January 2015
Note: |
See also: "Buffalo buffalo Buffalo buffalo buffalo buffalo Buffalo buffalo", Wikipedia (accessed May 4, 2008).
The syntax is the same as that of
Several of us students found the plural "-s" endings to lack a certain aesthetic simplicity, and we searched for a better word. I came up with
I.e., buffalo who are buffaloed by other buffalo themselves buffalo still other buffalo.
However, my fellow graduate students and I were not satisfied. So I concocted:
The syntax of the following sentence is close to the previous one:
I.e., mice who are in Boston, and who are chased (in a way unique to Boston) by cats who are in Boston, eat (in a way unique to Boston) cheese that comes from Boston.
So, buffalo who live in Buffalo (e.g., at the Buffalo Zoo, which does, indeed, have buffalo), and who are buffaloed (in a way unique to Buffalo) by other buffalo from Buffalo, themselves buffalo (in the way unique to Buffalo) still other buffalo from Buffalo.
I used the following question on AI and computational linguistics exams:
(though the capitalization is a bit off for that). Another of my students notes that the 10-word version is also ambiguous, along the lines of this:
parsed as meaning that "(The) Buffalo buffalo [i.e., the buffalo who live in Buffalo] (that) [other] Buffalo buffalo (often) buffalo (in turn) buffalo other Buffalo buffalo", and attributed it to his student Annie Senghas.
I had an email exchange with Pinker about the above history. Pinker responded as follows:
Dear William,
That's terrific! Thanks for letting me know that history, and if I use the example again, I will surely give you priority. (I believe it was independently invented by Annie -- I checked specifically with her when someone else had told me that she may have borrowed it.) I might balk a bit at the verb "to Buffalo-buffalo" as verb-compounding is fairly rare and stilted in spoken English, but the set of examples is certainly enlightening.
Thanks for letting me know about the review, and about your priority in the example.
Sincerely,
Steve Pinker"
I replied as follows:
Steve-
Yes, I also balk at the compound verb "to Buffalo buffalo", but I couldn't resist. My model was the Tennesee waltz; presumably, when one dances it, one could be said to be Tennessee-waltzing, no?"
Steve-
Remember the email exchange we had last summer about Buffalo sentences?
Well, Stuart Russell and Peter Norvig's new AI text, Artificial Intelligence: A Modern Approach, claims that Barton, Berwick, and Ristad came up with it. The exact reference is:
Barton, G. Edward, Jr.; Berwick, Robert C.; & Ristad, Eric Sven (1987), Computational Complexity and Natural Language (Cambridge, MA: MIT Press), Ch. 3, Sect. 3.4, esp. p. 100.
Nonetheless (I checked my records), I independently devised it in 1972 and have been using it in lectures since 1976. So, possibly there were 3 independent discoveries: mine, your student Annie, and Barton/Berwick/Ristad. In any case, I'm going to check with Berwick (but you're closer; you might check, too, if you're interested).
-Bill"
Pinker replied:
Thanks! If the book ever goes into a second edition, I will surely bring this all up.
Best,
Steve"
OK guys; it was bad enough that Steve Pinker, in his The Language Ins tinct, claimed that one of his grad students discovered/invented:
but now *you* claim, without citation, that Barton, Berwick, and Ristad came up with it (p. 690). Can you document that?
It's possible that Pinker's student came up with it independently, but I can document that I devised it in 1972. Here's the story:
In 1972, I took a grad course in philosophy of language from John Tienson at Indiana University. In that course, he presented the sentence:
which is grammatical and meaningful, if not acceptable, with no punctuation changes, having, of course, the same syntactic structure as:
Finding the "-s" morpheme unaesthetic, several of us grad students sought something better.
doesn't quite hack it, since "fish" requires an indirect object: one fishes *for* something. At that point, I came up with the Buffalo sentence.
I began using it in courses at SUNY Fredonia in 1976. One of the students in my first course there is now an ESL teacher in ... Buffalo, of course, and uses it in his classes.
I publicized it first to the SUNY Buffalo linguistics department that year, and then gave it more celebrity at ACL-88, when I put a parse tree for it in the registration packet (I was the local arrangements coordinator) and used an overhead transparency of it during my welcoming remarks. And a version of your problem 22.8 appeared as a question on our department's Graduate Qualifying Exam in 1988.
Since then, I've heard others claim it, but with the less interesting reading of the form Adj N V Adj N (like your "Dallas cattle..." sentence).
My favorite version requires the introduction not only of the modifier "Buffalo" for the animal (Buffalo buffalo are the ones in the Buffalo zoo), but also for the verb "to buffalo": You see, the Buffalo buffalo's style of buffaloing other buffalo is *so* unique that, like Tennessee waltzing, it's called Buffalo buffaloing, so:
Bottom line: When/where did Barton et al. devise it? "
To which Norvig replied:
I did no research at all to determine the origins of the buffalo sentence; I originially [sic] wrote the exercise without any attribution; then I saw a second-hand citation (either in [Michael A.] Covington's NL and Prolog book or James Allen's 2nd edition, I think) and put it in. Next week, when I get back from a trip, I'll put your story up in the "Clarifications" section of the web pages, and schedule a correction for the next printing of the book.
Being authoritative and finding original sources is important to us, so I'm glad you sent this note. We just didn't have time to do it completely, especially in exercises and the like.
-Peter"
They did, indeed, update the information; it now appears as Exercise 22.12 in Russell & Norvig's Artificial Intelligence: A Modern Approach, 2nd Edition (Upper Saddle River, NJ: Pearson Prentice-Hall, 2003): 833.
Here's *a* source:
Barton, G. Edward, Jr.; Berwick, Robert C.; & Ristad, Eric Sven (1987), Computational Complexity and Natural Language (Cambridge, MA: MIT Press), Ch. 3, Sect. 3.4, esp. p. 100.
They give an analysis of Buffalo sentences, plus others, like:
Now I need to contact them to find out where *they* got it from :-)
-Bill"
Covington replied:
I got it from Barton, Berwick, and Ristad, but have a dim recollection of having heard it somewhere before that. In citing them I did not mean to indicate that they were the definite originators.
- Michael A. Covington, Assc. Director mcovingt@ai.uga.edu -
- Artificial Intelligence Center University of Georgia -
- Athens, Georgia 30602-7415 U.S.A. phone 706 542-0358 -
and followed up:
One more place to look: Dover Publications publishes a reprint of a late 19th century book called "Oddities and Curiosities of Language and Literature" or something like that. If the Buffalo-sentence (or the police sentence, etc.) was current in Victorian times, it will be there.
[note added 9/17/06 by Rapaport: I looked in this book and did not find any sentence like it]
Here's a sentence from H. P. V. Nunn's _Elements of New Testament Greek_ (1951, originally published 1916):
Dear Professor Berwick:
I'm trying to track down the various origins of the Buffalo sentences that you discuss in your book with Barton and Ristad.
Here's my version of the history:
1. In 1972, I took a graduate course in Philosophy of Language with John Tienson at Indiana University. He gave the sentence:
as an example of a syntactically and sematically correct sentence that was difficult for humans to parse without already understanding it (along the lines of "mice cats chase eat cheese"). My fellow grads and I tried to come up with a more aesthetically pleasing sentence without the -s plural marker. We rejected "fish fish fish fish fish" since one normally fishes *for* something. I then devised "buffalo buffalo buffalo buffalo buffalo". Not being satisfied, I considered the buffalo in the Buffalo zoo (the Buffalo buffalo) and their unique way of buffaloing the other Buffalo buffalo, so unique that, like Tennessee waltzing, it's called Buffalo buffaloing, whence:
2. Since 1976, I have been using those sentences in my courses at SUNY Buffalo.
3. Around 1976 or shortly thereafter, I told the SUNY Buffalo Linguistics Dept. about them
4. In 1988, when I was the local arrangements coordinator for ACL-88, I put a parse tree for the simpler sentence (what you call "Buffalo^5") in the registration packet, and had an overhead transparency of it in my welcoming remarks. That same year, I also put it on our department's AI Ph.D. qualifying exam.
5. A few years ago, someone told me they heard it at a conference, possibly attributed to Dan Dennett (or one of his students), but giving the syntactic analysis as NP + V + NP.
6. Then I read Steven Pinker's new book, The Language Instinct, in which he attributes it to a student of his.
7. I then came across a reference, attributing it to you (and Barton and Ristad) in Russell and Norvig's new AI text (AI: A Modern Approach), though they don't give the reference to your book.
Now, given that I *know* that I came up with it myself, and giving the benefit of the doubt to Pinker's student, my question is: where did you folks get it from (and when)?
Thanks for any enlightenment you can give me on this.
-Bill Rapaport"
Berwick replied:
Hi Bill,
Well, hard to tell about these urban legends, you know.
I just recall reading about it when I was 10 or something years old---before 1972, to be sure. Then Ed Barton and I sat around
discussing it in 1982, and we just thought it was part of
common parlance (or urban legend) by then also. Even in the
police police police form.
For a very hilarious take on all this, you might want to
email
carl demarcken
(cgdemarc@ai.mit.edu) for one of his
famous "friday afternoon gsb" abstracts--in his abstract,
he works out the exact algebraic formula for any number of buffaloes,
as a joke, etc. These are stored on a web page somewhere in
http://www.ai.mit.edu.html
take a look...
[rest of message, on another topic, deleted]
best,
I replied:
Bob-
Thanks much for the info on buffalo! So if you heard it before 1972,
and I know I came up with it on my own, that means that at least 2
native speakers of English discovered this odd sentence--odd not only
because of its opaque surface syntax vis-a-vis its deep syntax, but odd
also because of its lexicon--independently, which makes one wonder about
how odd it is after all! :-)
[rest of message, on the other topic, deleted]
-Bill
Bill,
I first heard of the Buffalo sentences from Bob, who also mentions in his
book "French" and "police". I might add "char" to the list. (It would be
easy enough to do a search for more appropriate words).
I actually use the sentences occasionally to verify performance of parsing
systems- to make sure they conform to expected polynomial parsing times.
And I have found over the years that I can produce and analyze the
sentences pretty easily.
In the GSB message below (I send out a humorous message to a large mailing
list at the AI lab here every Friday inviting everyone to an afternoon beer
bash), I mention the Buffalo sentences and give a small grammar. I did
once calculate the largest eigenvalue of the matrix for the grammar, which
turned out to be about 1.33 (I did it by hand so it might not be correct),
but later noticed that I should have added a rule to handle the use of
Buffalo, the city, as the object of a verb: "I like Buffalo", "Gnus fool
Buffalo", "Buffalo buffalo Buffalo". Anyway, the message is below and I
even append a short lisp program the counts buffalo sentences of length n
exactly (but not using the rule just mentioned).
Thank you (and John Tienson and ...) for the sentences. They've provided
amusement to me for a number of years, and I even considered submitting a
silly ACL or LI squib about them. I never did, largely because I had no
idea who invented them. But if you're interested...
Carl
To: all-ai
Consider, for the sake of posterity, the word "Buffalo". The word in
isolation can take on at least one of three possible senses:
Now, gedanken about a string of N consecutive occurrences of the word
Buffalo. A grammar that seeks to describe the variety of meaningful
interpretations for this string must generate at least the interpretations
that the following context-free rules do:
So, "BUFFALO BUFFALO BUFFALO BUFFALO BUFFALO" can mean something analogous to
Indeed, as N -> infinity, the number of interpretations for Buffalo^N is
approximately proportional to 1.3312^N. For 200 Buffalos there are
121,030,872,213,055,159,681,184,485 easily understood interpretations.
To find out more about buffalo, police, char, and the French join us at
this week's
G I R L S C O U T B E N E F I T
at 5:30 p.m. Friday in the seventh floor playroom. Beer, root beer, and
tales from the dark side of linguistics will be in plentiful supply.
-----------
Dear Professor Rapaport,
There is an ongoing discussion of "punning" in comp.lang.lisp which refers
to your 1992 article on parsing challenges (
http://www.linguistlist.org/issues/3/3-175.html ).
I don't follow how the example sentence, which contains ten uses of the
word "buffalo", proceeds from the explanatory text. Using the rules of
English grammar and without changing word form, the longest sentence I can
construct contains only six uses:
I would greatly appreciate elaboration of the additional use cases in the
example sentence.
Additionally, I don't see how the dance reference is applicable. With
respect to the "buffalo" example, the equivalent, unmodified form,
"Tennessee Waltz Tennessee Waltz", is not a proper English sentence. I am
not a dance devotee, but accepting your premise that the term denotes both
form and method, the minimally correct English sentence requires addition
of the definitive article "the", as in:
Incidentally, I spent a good part of my childhood in Buffalo and much of my
family is still in the area. I have seen the bison in the zoo many times,
but have never before heard that their behavior is so peculiarly different
as to have its own descriptive term.
Thank you for any time you deign to spare to this note.
Very truly yours,
MS.CS. Northeastern University - Programming Languages & Operating Systems
...to which I replied:
George:
Thanks for your interest in {B|b}uffalo :-)
The sentence "Buffalo buffalo buffalo buffalo buffalo."
is actually ambiguous with both your reading, where "Buffalo
buffalo" is an Adj+N phrase as subject, "buffalo" is the
verb, and "[B]uffalo buffalo" is a miscapitalized Adj+N
phrase as direct object. However, the reading I prefer,
and which is historically more accurate, has essentially
the same syntactic structure as "Mice cats chase eat cheese":
the first "buffalo" is the subject; the next "buffalo buffalo"
is a relative clause (with omitted "that") whose structure
is N+V; and the final "buffalo buffalo" is a VP (=V+N)
containing the predicate of the sentence. So, paraphrased,
it becomes: Buffalo that (other) buffalo buffalo, themselves
buffalo yet other buffalo. Imagine 3 buffalo in a row,
each "buffaloing" the one in front of it.
The modification you question--which is admittedly a bit of
a stretch, but then English admits of such stretches (as
the computer scientist Alan Perlis said:
in English, any noun can be verbed)--involves prefixing
each word in the original sentence with the adjective "Buffalo",
referring to the name of your former and my fair city, with
the same grammatical structure as above, but with the (humorous)
interpretation that the buffalo who live in the Buffalo Zoo, viz.,
the Buffalo buffalo, not only buffalo the other buffalo in the zoo,
but do so in a way unique to Buffalo, called "Buffalo buffaloing",
and, moreover, they themselves are buffaloed in that way, viz.,
Buffalo buffaloing, by other Buffalo buffalo in the zoo:
Buffalo buffalo Buffalo buffalo Buffalo buffalo Buffalo buffalo Buffalo buffalo.
Other variations are possible.
Incidentally, to update and correct the history of the original, 5-word
sentence, although I did create the sentence as I said in my LINGUIST
posting, I have since learned that an earlier incarnation (invocation?)
of it occurs in a book by Robert Berwick (possibly Computational Complexity and Natural
Language--I'm at home and don't have the exact reference), which also
cites "police police police police police". Berwick tells me that he
first heard the sentence growing up in NYC in the 1950s, as I recall.
There may be a brief discussion of this in the Russell & Norvig AI text.
As for Tennesee waltzing, all I meant was to give a parallel construction
to "Buffalo buffaloing"; I did not intend that there was a similar
sentence, since, as you rightly note, waltzes don't waltz, nor is
waltzing a transitive verb.
-Bill Rapaport
... to which Neuner replied:
Dear Professor Rapaport,
I appreciate very much your thorough explanation of the "buffalo"
example. It occurred to me that "buffalo" is not a transitive verb, but I
failed to recognize the appositive form and so the sentence failed to parse
beyond six buffalos. It's a splendid example of both the fluidity of
English and the general difficulty in parsing natural language - even for
native readers.
I suspect it would also help not to read news in the middle of the night 8-)
Thanks again,
I further inquired:
George:
You wrote:
>There is an ongoing discussion of "punning" in comp.lang.lisp which refers
Can you give me a pointer to the comp.lang.lisp article that cites
this?
Thanks.
-Bill
... and received this reply:
Hi Professor,
Buffalos became a very popular topic. There are discrepancies between the
lists on Google and on my news server - the postings don't quite match up.
AFAICT, this article by Ray Dillinger contains the first reference to you
personally. Buffalos and other examples appeared earlier.
http://groups.google.com/groups?hl=en&lr=&ie=UTF-8&oe=UTF-8&selm=m3ad2ex2qh.fsf%40javamonkey.co
The thread starts at:
http://groups.google.com/groups?dq=&hl=en&lr=&ie=UTF-8&oe=UTF-8&threadm=873c85kdee.fsf%40g.mccaughan.ntlworld.com&prev=/groups%3Fdq%3D%26num%3D25%26hl%3Den%26lr%3D%26ie%3DUTF-8%26oe%3DUTF-8%26group%3Dcomp.lang.lisp%26start%3D50
and continues at:
http://groups.google.com/groups?dq=&hl=en&lr=&ie=UTF-8&oe=UTF-8&threadm=pcoptb88ody.fsf%40thoth.math.ntnu.no&prev=/groups%3Fdq%3D%26num%3D25%26hl%3Den%26lr%3D%26ie%3DUTF-8%26oe%3DUTF-8%26group%3Dcomp.lang.lisp%26start%3D25
and becomes ridiculous here:
http://groups.google.com/groups?dq=&hl=en&lr=&ie=UTF-8&oe=UTF-8&threadm=m3vfl05psp.fsf%40banff.eder.de&prev=/groups%3Fdq%3D%26num%3D25%26hl%3Den%26lr%3D%26ie%3DUTF-8%26oe%3DUTF-8%26group%3Dcomp.lang.lisp%26start%3D25
YMMV,
A
copy of the relevant pages of Borgmann's book is available from me;
send email to the address below.
"Date: Wed, 19 Apr 95 19:04:34 EDT
From: berwick@ai.mit.edu (Professor Robert C. Berwick)
To: rapaport@cs.buffalo.edu
Subject: buffaloes, etc.
bob berwick
Date: Thu, 20 Apr 1995 09:01:27 -0400
From: "William J. Rapaport"
To: berwick@ai.mit.edu, rapaport@cs.buffalo.edu
Subject: Re: buffaloes, etc.
From: cgdemarc@ai.mit.edu (Carl de Marcken)
Date: Thu, 20 Apr 95 13:40:07 EDT
To: rapaport@cs.buffalo.edu
Cc: berwick@ai.mit.edu, cgdemarc@ai.mit.edu
Subject: Re: buffalo sentences
Subject: GSB 5:30 p.m. Friday, Sept. 16, seventh floor playroom
--text follows this line--
Buffalo [Noun] ;; As in, Buffalo the animals that are not gnu.
Buffalo [City] ;; As in, Buffalo the city that is not Boston.
Buffalo [Verb] ;; As in, to bewilder or to baffle.
Root -> NounPhrase | Sentence
NounPhrase -> NBar [ Sent/NP ] ;; "Badgers" or "Beavers [that] Bovine bang"
Sentence -> NounPhrase VerbPhrase ;; "Brontosaurus beat Boas"
NBar -> [ City ] Noun ;; "Bison" or "Boston Butterclams"
VerbPhrase -> Verb [ NounPhrase ] ;; "{Bowfin} break [ Braconid ]"
Sent/NP -> NounPhrase VP/NP ;; "{Bumblebees that} Bushtit bedeck"
VP/NP -> Verb ;; "{Beagles that Beetles} belay"
"Bangor behemoths blame Brighton bats" or
"Bigfoots bedazzle birds [that] bears besmirtch" or
"Bobolinks [that] bharal betray beshrew biflagelates" or
"Bombay billfish [that] bighorn bind bivy" or
"Billygoats [that] Bangkok bivalves bifurcate bite" or
"Bedford broncos [that] Bridgewater bedbugs block" or
"Beasts [that] beluga [that] bluefish blight bewitch"
(defun print-buffalo (n &optional upper)
(if upper
(progn
(format t "Buffalos Sentences~%")
(loop for i from n to upper
do (format t "~D~10T~D~%" i (buffalo i))))
(print (buffalo n)))
(values))
(defun buffalo (n)
(+ (np n) (s n)))
(defun np (n)
(if (<= n 3)
1
(+ (np (- n 2))
(np (- n 3)))))
(defun np (n)
(flet ((it (n i r1 r2 r3)
(if (= n i)
r1
(it n (+ 1 i) (+ r2 r3) r1 r2))))
(if (<= n 3)
1
(it n 4 2 1 1))))
(defun s (n)
(cond ((< n 2)
0)
((= n 2)
1)
(t
(+ (np (- n 1))
(loop for i from 1 to (- n 2)
summing (* (np i)
(np (- n i 1))))))))
Date: Thu, 18 Mar 2004 04:26:40 -0500
To: rapaport@cse.Buffalo.EDU
From: George Neuner
Subject: "buffalo" ad nauseam
Buffalo buffalo Buffalo buffalo Buffalo buffalo.
adj n adv v adj n
|-- subject --| |--- verb ----| |-- object ---|
Tennessee Waltz the Tennessee Waltz.
George Neuner
Undergraduate SWE test score: 754
Vocal advocate of remedial grammar for exposed media employees.
Currently unemployed.
Date: Thu, 18 Mar 2004 07:50:40 -0500 (EST)
From: "William J. Rapaport"
To: gneuner2@comcast.net, rapaport@cse.Buffalo.EDU
Subject: Re: "buffalo" ad nauseam
[Note added 10/5/2012: Thanks to reader Abdullah A. Khan, 10/5/12,
who noticed that I was missing a couple of "buffalo"s, which I've now
corrected]
Date: Thu, 18 Mar 2004 09:05:52 -0500
To: "William J. Rapaport"
From: George Neuner
Subject: Re: "buffalo" ad nauseam
George Neuner
Date: Tue, 23 Mar 2004 14:39:09 -0500 (EST)
From: "William J. Rapaport"
To: gneuner2@comcast.net
Subject: reference to me on comp.lang.lisp
>to your 1992 article on parsing challenges (
>http://www.linguistlist.org/issues/3/3-175.html ).
Date: Tue, 23 Mar 2004 16:21:58 -0500
To: "William J. Rapaport"
From: George Neuner
Subject: Re: reference to me on comp.lang.lisp
George
From: thelewolfs@comcast.net
To: rapaport@cse.Buffalo.EDU
Subject: Buffalo Buffalo Buffalo Buffalo
Date: Tue, 12 Dec 2006 17:47:46 +0000
Hello,
This is in reference to your website on the creation of the "Buffalo
Buffalo" sentence.
Actually, this appears earlier than your 1972 claim. Dmitri A.
Borgmann's classic book, "Beyond Language--Adventures in Word and
Thought," has a copyright date of 1967; it was published by Charles
Scribner's Sons, New York. (I am sure that you are aware that the late
Dmitri Borgmann was the founder of Word Ways, the Journal of
Recreational Linguistics.)
I refer to the "Problems" section on pp.70-71 ("Repetitive Homonymy"),
in which Borgmann challenges the reader to devise an English homonymic
sentence. On page 190 ("Resolutions"), one of the examples given is
"BUFFALO BUFFALO BUFFALO BUFFALO." Borgmann goes on to say: "For
anyone puzzled by this statement, we elaborate: 'Wild oxen (roaming
the streets) of Buffalo, New York bewilder (visiting) North Carolina
coast dwellers.'"
Sincerely,
Shirley Wolf
17016 Freedom Way
Rockville, MD 20853
link:
Tristan Miller
reminds me (see letter below) that the correct page number is
290, not 190
And here is some history about an even earlier occurrence of a "Buffalo"
sentence:
Subject: Another antedated "buffalo"
From: Tristan Miller
Date: 12/7/14 11:38 AM
To: William J. Rapaport
material is highlighted