Language Log

Orthocorrection

Tidy-whiteys

Although the Eggcorn Hoard has been moving from the gleaming towers of Language Log Plaza to Chris Waigl's elegant new warehouse,  those of us here at the home office haven't lost interest in the phenomenon.  Indeed, eggcorniacs (Chris included) are forever faced with trying to judge what went on in people's minds during the first productions of some form-meaning pairing that diverges from the pairing previously current in the speech community.  Making these judgments can be a tough task.

Case in point: tidy-whitey, as in this cite from Mark Morford's column "Attention, Liberal Shoppers!" in the March-April 2005 Gay and Lesbian Review, p. 4:

France defies Google

Perhaps inspired by Jean-Noël Jeanneney, Agence France Presse (AFP) is suing Google for copyright infringement. According to Reuters:

The French news service is seeking damages of at least $17.5 million and an order barring Google News from displaying AFP photographs, news headlines or story leads, according to the suit filed on Thursday in the U.S. District Court for the District of Columbia.

According to the same story, Google must be prepared to argue this one out in court:

AFP said that it had informed Google it was not authorized to use AFP's copyrighted material as it did and that it had asked Google to stop.

Google has ignored those requests, AFP's suit alleges.

I suppose that it's for this reason that Google News continues to be called "beta", and also that it doesn't offer any advertising. If Google can win this fight, it would be a rare legal triumph for the concept of "fair use", which I assume would be the basis of their defense. As the U.S. Copyright Office explains, section 107 of the U.S. copyright law "sets out four factors to be considered in determining whether or not a particular use is fair:"

1. the purpose and character of the use, including whether such use is of commercial nature or is for nonprofit educational purposes;
2. the nature of the copyrighted work;
3. amount and substantiality of the portion used in relation to the copyrighted work as a whole; and
4. the effect of the use upon the potential market for or value of the copyrighted work.

In this case, the argument for Google would be (1) that Google News has an educational (and experimental) function and does not generate revenue for Google; (2) that the material cited is available to the public on the web in any case; (3) that only minimal amounts are cited, just enough to allow readers to determine whether or not to click through to the source; and (4) that the effect on AFP is positive, since it brings more readers to sites that run their stories (and may run advertising as well).

This doesn't take into account the "Anglo-Saxon perspective" that pained M. Jeanneney so keenly, though congeniality of national perspectives is not one of the factors recommended by section 107 for determining fair use. But I'm not raising Jeanneney's arguments as a joke, since AFP is chartered by the French government, giving it roughly the same status as Xinhua or Voice of America. Well, it may be technically somewhat more independent, but it would be surprising if AFP took a step like this without government consultation.

I'm not a lawyer, and have very little understanding of the current dynamics of copyright law. However, this strikes me as a fortunate case for defenders of fair use, since it pits one of the most popular and innovative U.S. companies against the news agency of the French government, who presumably have less influence in the Federal Court system (and on Capitol Hill) than Disney does. Of course, Google might avoid the fight, for instance by agreeing not to index AFP.

[Update: some news reports quote "Google spokesman Steve Langdon" as saying "We allow publishers to opt out of Google News but most publishers want to be included because they believe it is a benefit to them and to their readers." This is in direct contradiction to AFP's charge that they've asked Google to stop crawling their stuff but that "Google has not stopped despite repeated requests."

This odd discrepancy will presumably be cleared up soon. My own guess is that it's a question of where the material is aggregated from: not from AFP's site, but from the sites of newspapers that subscribe to AFP and reprint its stuff under license.]

[Update #2: Edward Hasbrouck thinks that writers ought to be rooting for AFP. And Dana Blankenhorn says that AFP's site has no robots.txt file.]

Fo did it

Alexander, King of Jesters, seems to be wrong when he says that "Many jesters and fools spoke a gibberish language called Grammelot that was first described over 500 years ago." Similarly, Gianni Ferrario seems to be wrong when he says that "Grammelot is a form of theater invented by the comic actors of the Commedia dell'Arte of 1400, and is organized in an onomatopoeic mode, that is, it manages to evoke concepts by means of sounds that are not established or conventional words." At least they're wrong to imply that the term Grammelot dates from the 15th or 16th century. (See this earlier post for references and links). In fact, the term was apparently invented by Dario Fo, perhaps in connection with his 1969 play Mistero Buffo.

The confusion seems to have arisen because of Fo's references to the 16th-century playwright Angelo Beolco. In Fo's Nobel acceptance speech, he gave credit to "Ruzzante Beolco, my greatest master along with Molière", called him "until Shakespeare, doubtless the greatest playwright of renaissance Europe", and referred to the inspiration of Ruzzante's linguistic inventiveness:

Ruzzante, the true father of the Commedia dell'Arte, also constructed a language of his own, a language of and for the theatre, based on a variety of tongues: the dialects of the Po Valley, expressions in Latin, Spanish, even German, all mixed with onomatopoeic sounds of his own invention. It is from him, from Beolco Ruzzante, that I've learned to free myself from conventional literary writing and to express myself with words that you can chew, with unusual sounds, with various techniques of rhythm and breathing, even with the rambling nonsense-speech of the grammelot.

Stefano Taschini told me by email that his dictionary "suggests that Grammelot might result from the composition of the French words grammaire,  mêler, and argot, but its etymology stays uncertain." Stefano checked the "Encyclopedie Larousse, without finding any matches (trying all possible accent combinations). Despite being a french-looking and french-sounding word (I always heard it pronounced as [gram'lo]), apparently it isn't."

My own searches for grammelot in the BNF's Gallica archive and the Dictionnaire de l'Académie Française (both 8th edition and 9th edition) likewise produced nothing. In addition, the word grammelot isn't in large English dictionaries such as the OED and Webster's 3rd International, nor is it in the LION ("Literature Online") database. The ARTFL project's Opera del Vocabolario Italiano search form and the Ricerca page of the Tesoro della Lingua Italiana delle Origini also return "Nothing found matching specified search term(s)" and "Nessun elemento trovato" for various types of searches for grammelot. If such a word had really been in use since the 15th and 16th centuries to describe the language of the Commedia Dell'Arte, or even Ruzzante's language in particular, this absence would be very surprising. It seems much more likely that the term was invented in the 1960s by Dario Fo and Franca Rame to describe their own linguistic experiments.

This leaves the question of whether the language of Ruzzante's plays was simply a representation of the local vernacular, or rather (as Fo says) a theatrical invention. Stefano isn't sure:

As for Ruzzante's language, I really cannot say. My unsubstantiated feeling is that it looks too much like a parody of Po plane dialects to be really one of them, and I'm rather inclined to consider it more an invention à la Fo than a faithful representation of the vernacular spoken in the countryside around Padua. At some point in La Moscheta, Ruzzante dresses up as a foreigner, and speaks in a mocked-up Italian (rather funny, actually). There, the language is obviously invented, but probably reflecting the efforts of his farmers to speak "properly".

Stefano (one of those special people whose email has footnotes) points out that a book and video tape of Fo's Mistero Buffo is available here, and that the texts of two of Ruzzante's plays are on line here.

Continuing on the trail of grammelot, the only return for this term from the MLA bibliography is Erith Jaffe-Berg, "Forays into Grammelot: The Language of Nonsense", Journal of Dramatic Theory and Criticism (JDTC) 2001 Spring; 15 (2): 3-15, which is not available on line. The fact that there are no other references seems to be further evidence that this is a new term, not an old one. However, Ray Girvan sent a reference to a paper by Adrienne Ward ("'Imaginary Imperialism': Goldoni Stages China in 18th Century Italy". Theatre Journal 54.2 (2002) 203-221) which is available through Project Muse, and which uses the term grammelot as if it were a traditional name for (certain kinds of) theatrical double-talk.

Ward's paper is about a scene of "pretend Chinese" ("finti cinesi") in Carlo Goldoni's comic opera L'isola disabitata (1757). In footnote 36, Ward writes:

The deployment of a variety of languages and dialects (existing or imaginary) was relatively common practice on the Italian stage in Goldoni's time, hailing back to a classic bit in the commedia dell'arte tradition. A key feature of the commedia's poetics of improvisation involved virtuoso linguistic performances, in which characters would spew forth exaggerated and comical but authentic-sounding streams of foreign language, sometimes switching deftly between diverse tongues. Grammelot is the term used to describe this early practice of replicating the Other's language, which then became particularly suited to eighteenth-century exotic theatre. In Dario Fo and Popular Performance (Ottawa: Legas, 1998), Antonio Scuderi defines grammelot as ". . . a fake language which consists of nonsensical sounds that imitate the inflection and cadence of real speech," 8. He also notes that "The insertion of a limited number of key words that are identifiable to the audience conveys a sense of semantic value and thus a sense of real speech to the otherwise nonsensical sounds," Dario Fo: Stage, Text and Tradition (Carbondale: Southern Illinois University Press, 2000), 58. The scene of the "finti cinesi" clearly exhibits these traits of grammelot, and shows Goldoni's continued reliance on certain aspects of commedia dell'arte performance. For more on the history of grammelot and its manifestations in twentieth-century theatre, see the aforementioned works by Scuderi, as well as Dario Fo, Tricks of the Trade, trans. Joseph Farrell (New York: Routledge, 1991). [emphasis added]

So either Ward and Scuderi have misread Fo, or this word has someone survived for half a millenium in the theatrical demimonde, without leaving any detectable traces in the literary and linguistic history of Italy, France and England.

[Update 12/10/2006 -- I was wrong to suggest that Antonio Scuderi might have misread Fo. In Scuderi's discussion of the subject, which I had read only in Ward's quotations, he makes it clear that he does not know the origin of the term, and he does not claim either that it is an old word or that it is a new one. Ward's discussion, which I had read in full, certainly implies that the term dates back to the early days of the commedia dell'arte tradition. However, Scuderi's only role here was to provide the definition that she used.]

Just for fun, here's Goldoni's passage of fake Chinese:

and the "translation" given in Ward's paper:

about which she writes

My English translation of this portion of the scene of the "finti cinesi" can only hope to approximate both the sense and the nonsense of the Italian wordplay. The Italian version makes great use of the occlusive phoneme /k/ and truncated syllables at the end of lines, most likely to imitate the clipped sound of Chinese speech. I have tried to imitate this in an English parody as closely (and creatively) as possible. The repeated use of the syllable "ka" in the Italian contributes to the scatological tone of the exchange, sharply reminiscent of standard commedia dell'arte sketches. Other verbal constructions from which meaning can be extracted include threats ("Ti menaccà—paraticà"), and more overt scurrilous phrasing, such as "Chichirichi caccaraccà, / Quaiotta squaquarà," where "squaquarà" evokes the colloquial expression for diarrhea, "la squaquerella."

[Update 12/7/2006 -- Antonio Scuderi writes:

The other day, in a moment of egoistic self-indulgence, investigating books and articles where my work has been cited, I came across my name on your blog, Language Log, in an installment entitled "Fo Did It" (2005). Apparently without opening a book, the author concludes that the Italian Nobel Playwright, Dario Fo, invented the term grammelot, which refers to an aural performance technique. He quotes a long passage by someone else who quotes me, and then concludes, "So either Ward and Scuderi have misread Fo, or this word has someone (sic) survived for half a millennium in the theatrical demimonde, without leaving detectable traces in the literary and linguistic history of Italy, France and England" (2005:3).

Okay, it's just a blog, no responsible peer reviewing required, etc. But still... If the author had cracked open the two books of mine that are mentioned, he would have found that I cite where the term is listed in a major Italian dictionary (Zingarelli 1995:797), and a survey of attempts to trace the origins of the word. In my essay, "Updating Antiquity," in Dario Fo: Stage, Text and Tradition, I explain that according to John Rudlin (Commedia dell'Arte: An Actor's Handbook, 1994:59-60), Fo most likely learned the technique from Jacques Lecoq, who "definitely" learned it from Jean Dasté, who had used it with the Copias troupe, which had called it grummelot. I explain that the etymology is uncertain and provide several hypotheses. None suggest that Fo invented the term, and in fact, Fo himself makes it clear that it did not originate with him: "termine di origine francese, coniato dai comici dell'Arte e maccheronizzato dai veneti che dicevano 'gramelotto'."

Wrestling with this enigmatic term in a scholarly endeavor was not easy. In any instance, of course, it is disappointing to see the results of such research bandied about in an off-hand irresponsible manner. In the present case, the invention of Fo inventing the term has reached Wikipedia by way of Language Log, so that myth is not just lost in a blog, but, alas, presented in a forum that some will trust.

I'm grateful for the additional information. However, the word grammelot (under whatever spelling) still seems to be a modern invention, contrary to the implication of some prominent contemporary users. Although Fo says that the term is "di origine francese", it is apparently not attested in any French dictionaries or other historical sources that I have been able to find. Specifically, the 8th edition of the Dictionnaire de l'Académie Française has neither grammelot nor grummelot nor grammelotto. Likewise, Gallica tells us that "aucun document ne répond à la requête".

And neither grammelotto nor relevant similar word appears to occur in the Opera del Vocabolario Italiano textual database, which includes more than 20 million words of pre-1375 Italian text.

Pending any evidence of use before 1960 or so, it still seems to me that the word grammelot is a modern invention -- whether the inventor was Fo or Jacques Lecoq or someone else -- which has been introduced into general use by Fo. The fact that the word occurs in an Italian dictionary published in 1995 doesn't address this question one way or another, unless it gives citations from an earlier time.]

Franklin Medal to Aravind Joshi

Aravind Joshi has been awarded the 2005 Benjamin Franklin medal in Computer and Cognitive Science. The citation, quoted below, focuses on his many research contributions and his other honors. It doesn't mention some of the the most important things about Aravind, though. There's his 40 years of tireless work to build a culture of interdisciplinary research on human language, involving linguists, psychologists, philosophers and mathematicians as well as computer scientists and engineers. There's his unwavering intellectual tolerance and respect for alternative approaches, combined with the utmost rigor and the highest standards of evaluation. And at an age when some people might think about retiring, Aravind has been forging into a completely new area of research, modeling the language of biological macromolecules.

Here's what the Franklin Institute has to say about him:

Citation: The 2005 Benjamin Franklin Medal in Computer and Cognitive Science is awarded to Aravind Joshi for his fundamental contributions to our understanding of how language is represented in the mind, and for developing techniques that enable computers to process efficiently the wide range of human languages. These advances have led to new methods for computer translation.

Aravind Joshi is a world leader in the interdisciplinary research that covers linguistics, cognitive science, and computer science. His theoretical insights regarding the structure of human language, and the resulting tools and techniques he developed, help us understand how we communicate and allow us to interface with technology more naturally.

After receiving his B.E. in electrical and mechanical engineering from Pune University and his D.I.I.Sc. in communication engineering from the Indian Institute of Science, both in his home country of India, Dr. Joshi began a fruitful academic career in the United States. He received his M.S. (1958) and his Ph.D. (1960), both in electrical engineering, from the University of Pennsylvania, where he has been a professor for the last 40 years. Currently Henry Salvatori Professor of Computer and Cognitive Science, Dr. Joshi is also co-founder and co-director of the University's Institute for Research in Cognitive Science.

Dr. Joshi's numerous awards include the David Rumelhart Prize for 2003 from the Cognitive Science Society, a Lifetime Achievement Award from the Association for Computational Linguistics, the Research Excellence Award of the International Joint Conference on Artificial Intelligence, and an honorary doctorate from the University of Paris 7. He was elected to the National Academy of Engineering, and is a fellow of the Institute of Electrical and Electronics Engineers (IEEE), a founding fellow of the American Association of Artificial Intelligence (AAAI), and a member of the Association for Computational Linguistics, the Linguistic Society of America, and the Association for Computing Machinery.

The other 2005 citations can be found here.

 

Surnames

Linking to a post on Japanese surnames at Butterfly Blue, Steve at Language Hat asks "Did you know (to take one startling fact) that Japan has more different surnames than any other country in the world (about 120,000)?" Backed by Steve's well-deserved authority on the net, I'm afraid that this meme may start to propagate, although it's far from being true.

20-odd years ago, I worked on software to determine the pronunciation of names for text-to-speech applications. We worked from lists of American surnames, derived from phone books and other sources, that comprised several million distinct (orthographic strings representing) surnames, and were far from complete. This is also the kind of number that emerges from the description of NameX, a software product that can be seen as a "thesaurus containing 132 million variants for 2.6 million distinct Surnames," though this is for names "from all over the world with comprehensive coverage of names with European origins".

You can download a list of "Frequently Occurring Names" from the U.S. Census Bureau that includes 88,799 distinct surnames from 6,290,251 records. According to the documentation, this is based on the records of people living in the 5,300 blocks where the "post enumeration sample" was done, along with "additional surrounding ring blocks", amounting to about 1/40 of the overall 1990 census. This was then pruned further: "For purposes of both confidentiality and elimination of data noise we restricted the number of unique names available at this internet site to the minimum number of entries that contain 90 percent of the population in that data file."

The file was sorted in inverse order of frequency, with the names for each frequency count sorted in inverse alphabetical order. The last batch before the file is cut off includes 13,124 names from

ZYSETT 0.000 89.231 75677

to

AALDERINK 0.000 90.483 88799

The count of individuals with each name is not given, but the numbers mean that these 13,123 names comprise 90.483-89.231= 1.252% of the total set of people in the sample, so that we can determine that the sample count for the names in this set must have been 6. This allows us to provide a lower bound on the number of distinct names in rest of the sample as about 99,774 (i.e. assuming that all additional names occur six times each), which would yield 88,799+99,774=188,573. In fact this is surely much too small, since the tail of names will normally include increasing numbers at lower frequency counts (of 5, 4, etc. people per name). In addition, the other 39/40 of the census will add considerably to the tail of infrequent names, not only because of the expected effects of a larger sample, but because of the bias introduced because the existing list sampled whole households from a particular set of compact geographical areas ("blocks").

I don't present this as any sort of estimate of the number of distinct surnames among American households (which I believe to be well over a million), but rather as a demonstration that whatever the true number is, it's much larger than 120,000.

The winter eggcornucopia

The eggcorns mount up alarmingly here at Language Log Plaza. I've started to move items to Chris Waigl's eggcorn database and don't expect to provide further catalogues of examples here. (If you have contributions to make, drop them off at Chris's warehouse.) But here's an inventory of the winter crop (including genuine eggcorns, dubious ones, and examples of related types).

First, an appreciation of eggcorns from my soc.motss friend Michael Siemon, in e-mail on 2/19/05:

... The whole eggcorn thing, since you nudged us a while back, has fascinated me. I used to pass over these, with a small wince or blink of amazement/amusement, but passed on without dwelling on the possible mental processing involved. But now, they seem to be jewels (kitschy jewels, to be sure) cast before the usenet swine who do their best not to notice them. And so I now pause and allow them their voice.

Meanwhile, Rich Baldwin asked me in e-mail on 1/21/05, "Is the birth of an eggcorn an eggcoronation? Or is that taking the joke too far?" Well, I have no shame, as you can tell by the title of this posting. Let the eggcoronations begin!

1. Stuff already in the eggcorn data base. These are items from my winter list that others have already entered into the data base:

deep-seeded [ < deep-seated ]

(run the) gambit [ < gamut ]

(take for) granite [ < granted ]

hare's breath [ < hair's breadth ]

off-times [ < oft-times ]

sorted [ < sordid ]

2. Stuff I just entered. Quite an assortment of things today.

catchnap [ < catnap ]

centrifical (force) [ < centrifugal ]

centripedal (force) [ < centripetal ]

content (with) [ < contend ]

far be it for [ < far be it from ]

girdle (one's loins) [ < gird ]

granola (oil) [ < canola ]

Grumman [ < Grauman ]

languistics [ < linguistics ]

marball [ < marble ]

metal [ < medal ]

(in the) mist (of) [ < midst ]

passed [ < past ]

powerhorse [ < powerhouse ]

pre(-)fix, pre(-)fixe, pre-fixed [ < prix fixe ]

skeletal (staff) [ < skeleton ]

soak (one's wild oats) [ < sow ]

3. Still in the to-do files. Some of these might already be in the database; I haven't checked them all. A number of them present special problems of analysis that need comment. Some are old standards.

black and (red fish) [ < blackened ]

carrot on a (stick) [ < carrot and ]

chomping (at the bit) [ < champing ]

(French) crawlers [ < crullers ]

(make) due [ < do ]

flaunt [ < flout ]

flush (out) [ < flesh ]

font (of knowledge) [ < fount(ain) ]

lazy-fare [ < laissez-faire ]

mano y mano [ < mano a mano ]

(comedy of) manors [ < manners ]

mitigate (against) [ < militate ]

pack [ < pact ]

pain-staking [ < pains-taking ]

(set) perimeters [ < parameters ]

prolongered [ < prolonged ]

(on) queue [ < cue ]

same-oh same-oh [ < same-old same-old ]

verses [ < versus ]

wonderlust [ < wanderlust ]

4. Cases that are even more problematic. Here's a collection of things that might be simple misspellings, blends, classical malapropisms, or whatever.

adapt [ < adept ]

bare (responsibility) [ < bear ]

beg to pardon [ < beg (your) pardon ]

canape [ < canopy ]

coward [ < cowered ]

exult [ < exhort ]

hold (up) [ < holed ]

kick in the bucket [ < kick the bucket ]

perceptively [ < perceptibly ]

picture (of beer) [ < pitcher ]

waiver [ < waver ]

zwicky at-sign csli period stanford period edu

The experience of capricious censorship

I'm frustrated. Three separate attempts to produce a blog entry this morning have been stopped by the weirdest netnannity that I've ever seen.

I'm at the Airlie Center in Warrenton, VA, attending a workshop, on which more later. Airlie is a lovely place, whose web page says that it was "founded in 1960 as an 'island of thought', and has provided a unique environment for the the creative exchange of ideas ever since." This environment features hundreds of acres of lovely countryside, swans in mating duets, and an internet connection, available in the rooms for the usual $9.95 per day, that capriciously blocks about half the web.

I first noticed this when I tried to access the comics archive at http://www.doonesbury.com. What I got instead of Doonesbury was a big dark-blue page, blank except for a light-blue box reading:

The page title is "SonicWALL -- Web Site Blocked." Gee, I thought, that's funny. Did one of Trudeau's strips insult the swans?

Then I started to compose a post on the origins of the term Grammelot. Stefano Taschini wrote that his Italian dictionary says it "might result from the composition of the French words grammaire, mêler, and argot", but he couldn't find it in any French dictionaries. So I thought I'd check the Gallica website at the Bibliothèque Nationale de France. Oops -- blocked by the IT department again. Now, I've been critical of the BNF, but blocking their site doesn't make it better.

So to soothe myself, I took a quick tour through our blogroll. But http://ablauttime.blogspot.com/ and
http://mixingmemory.blogspot.com/ and apparently every other site at blogspot were blocked. Also blocked were:

http://www.raygirvan.co.uk/apoth/thought.htm
http://piginawig.diaryland.com/
http://www.audhumla.org/
http://blinger.org/
http://www.languagehat.com/
http://keywords.oxus.net/
http://www.painintheenglish.com/
http://www.wordlab.com/

Allowed, on the other hand, were

http://semanticcompositions.typepad.com/ (and apparently all other typepad sites)
http://mattweiner.net/blog/
http://www.livejournal.com/users/q_pheevr/ (and apparently all other livejournal sites)
http://www.margaret-marks.com/Transblawg/
http://www.bisso.com/ujg/
http://www.multilingualblog.com/

This doesn't make any sense at all as an attempt to protect America's conference-goers from nefarious influences.

I can see three alternative explanations. One is that it's pure pharmaceutical-grade incompetence -- someone's pet cat sleeping fitfully on a keyboard connected to SonicWALL's Content Security Management control screen. A second possibility the Airlie IT department is staffed by a manager who is worried about ritual pollution by words and images, and a lackey who has no idea how the web works ("sexually explicit lyrics over at blogspot? check, boss, it's taken care of; and those French postcards won't be a problem again, not at Airlie, no sir."). The third and most probably theory is that the Bastard Operator from Hell has retired to the Virginia countryside and is now running the Airlie IT department. This page might provide some insight -- I don't know, I can't read it, myself, because

[Seriously, you can look up the rating of URLs according to SonicWALL's Content Filtering Service database, and you'll find that the BNF, the bartleby.com site, all blogspot weblogs, Ray Girvan's site, etc. etc. are all classified as "Category 14 - Arts/Entertainment". Apparently the Airlie IT department has voted arts and entertainment off their "island of thought". Or could that just be the default configuration for SonicWALL's software? ]

The moose who chose to lose whose nose?

Arnold Zwicky remarked in the first version of this post that lose is the only English word with /uz/ spelled <ose>. [The remark is not there now, because he's thoroughly on the ball, and corrected himself just a few hours later, while I was writing this one. We tussle like this all the time at Language Log Plaza. They say that in the corporate world it's dog eat dog. Well, at Language Log it's exactly the reverse.] ’Tisn't true [as he now says in the corrected version]: there is also whose (which he might perhaps have initially missed because it's an inflected form, not a citation form of a lexeme). But the temporary slip only underlines his point about the orthographic minefield of English words ending in /uz/ and /us/ (likewise /eiz/ as in phrase and /eis/ as in base, and so on). If I had talent as a poet I would attempt to write poems about it for the bemusement (one can scarcely say amusement) of children:

If a moose should choose to lose his nose,
His does, I s'pose, would get morose.
A nose does have use for a moose on the loose:
It's his nose he would use when close to a rose,
For roses aren't food a loose moose would choose
(Whose moose would choose a dose of rose?)...
[etc. etc., approx. 165 stanzas]

But first, I have no talent as a poet (my stuff is right down there with mortgage company poetry); and second, confusing poems aren't going to help the little mites with their spelling problems, are they? Best to use old-fashioned methods: if you spell lose as loose or choose as chose you will be punished. We are Language Log, and we will brook no ilittera illitter illiteracy.

Ke'o and Ngadha

Last summer I learned about a couple of truly odd languages spoken by small groups in Indonesia, Ke'o and Ngadha. They have no prefixes or suffixes at all. All grammar is taken care of by separate words -- no WALK-ED, no FRIEND-SHIP, no RE-SEND. These days I am thinking that this, of all things, may be evidence of an interspecies encounter between two different species of HOMO.

Now, there are plenty of languages that are awfully low on prefixes and suffixes, although typically languages have at least something along these lines according to some linguists' analysis. But Ke'o and Ngadha are particularly odd in that they belong to the Austronesian family, which is not exactly known for being affix-shy. Ke'o and Ngadha's relatives include languages like Tagalog, bristling with prefixes and suffixes, and Malay, which shaves down the chaos somewhat, but still challenges the learner with affixes galore. Even Ke'o and Ngadha's VERY closest relatives, in what is called the Central Malayo-Polynesian group, are usually chock full of, at least, prefixes.

So what worried me about Ke'o and Ngadha was why languages would just dump their affixes. When this happens elsewhere, typically the affixes leave behind some kind of footprint -- in Chinese, for example, many prefixes faded away eons ago -- but often morphed into a change in pitch on the word that remains today. But in Ke'o and Ngadha, nothing of the sort. No tone, no fossilized warts. They're just nude. Their grammars are, in this aspect, simpler than what is typical of their relatives. They are haikus amidst limericks.

In fact, whenever grammars just let it all go in this way, it is because at some point, they were learned more by adults than children. Adults aren't as good at learning languages as kids -- as we all know from language classes -- and so in cases like this, the grammar becomes more streamlined. The demonstration case here is the extreme one of creoles, where a European language full of AMO, AMAS, AMAT-type baubles becomes one where the verb stays the same in all persons, and so on.

But Ke'o and Ngadha are not creoles. They have had nothing to do with plantation slavery or anything similar. I scratched my head over these languages for months.

But then last fall, we all learned about the discovery of skeletons of the "little people" on the Indonesian island of Flores. The skeletons date from 13,000 to 18,000 years ago, and are so unprecedently small in size that they have been classified as a new species of HOMO.

And get this -- Ke'o and Ngadha are spoken on none other than Flores.

This got me thinking: could it possibly be that what made these languages go "creole" was that the "little people" learned them so extensively that the way they spoke them became what children heard most? This, after all, is how creoles come to be. It is also why Swahili, used as a second language more than as a first one for centuries, is significantly more user-friendly than other Bantu languages. It is even why, in some linguists' opinion (including mine) English is the only Indo-European language in Europe without pesky gender of the LE BATEAU/LA LUNE sort.

Of course, I can only venture this as a nervy guess at this point. But the evidence is interesting indeed.

The skeletons were unearthed in western Flores, while Ke'o and Ngadha are spoken somewhat eastward, in the South Central part. But there are two things. First, the language of western Flores is Manggarai, and while it has a little scattering of affixes, even it is "naked" enough that when I read a description of it last summer, long before the "little people" had been reported, I wondered just what was up even with it.

But second, not only the Manggarai but also the Ngadha people have legends about little people who once lived among them until as late as the 1500s. It may well be that "little people" skeletons have yet to be unearthed in the Ngadha area, since Flores is very small.

And in any case, the legends say that the little people spoke in an incomprehensible babble but couldn't repeat back in the Homo sapiens' language. This can't be taken as evidence that the little people did not have language, since the "babbling" impression is a typical judgment of foreigners' language by people unschooled in the esoterica of modern linguistics. And wouldn't you know that two weeks ago we learned that the "little people"'s brain case suggests advanced cognition -- just as does their hunting and tool-making, which, in the analysis of Homo sapiens, is regularly treated as evidence that natural language had arisen.

Then there is one more thing. Many of the languages of Flores are barely documented, and this includes various ones spoken between Manggarai and Ngadha. But documentation is beginning for one of them, Rongga, which sits right between the other two. And it turns out to be one more language with, mysteriously, no prefixes or suffixes.

So -- a preliminary hypothesis is that when a descendant of the ancestral Proto-Austronesian language came to Flores, which archaeology and comparative linguistic reconstruction suggest was between 4000 and 4500 years ago, its speakers encountered "little people" already there, speaking their own language. These would not have been the only people around, as Homo sapiens is thought to have gotten to the Flores region about 50,000 years ago.

But we might suppose that the "little people," perhaps because of their smallness and less "advanced" society, were incorporated into the newcomers' society at some point, possibly by force. Their non-native rendition of the newcomers' language gradually became the model that all children learned. The result -- "schoolboy Proto-Austronesian," kind of like our version of French or Spanish in middle school. Except in this case, this was refashioned into a full, nuanced language. Full, nuanced language does not require affixes.

There are models for this. For example, Afrikaans in South Africa is a variety of Dutch as filtered through Khoi ("Bushman") servants and nursemaids, the result being the second most streamlined Germanic language after English.

We could also take this hypothesis as support that the "little people" indeed had the language capacities that we do, since it would appear that they acquired the newcomers' language just as successfully -- albeit incompletely -- as Homo sapiens does worldwide in similar situations. They managed a version of Proto-Austronesian (for linguists, technically, Proto-Malayo-Polynesian) robust enough that children could acquire it and use it everyday without surrounding adults conclusively dismissing it as utter gibberish.

Obviously, we need a lot more data before I could go further -- genetic analysis of the skeletal material, documentation of more of the languages of Central Flores, and more thorough searching for fossils (i.e. we have no fossils of "little people" after 13,000 years ago as of yet).

But the sheer weirdness of Ke'o, Ngadha and Rongga strikes me as begging for an explanation beyond ordinary tendencies in how languages change over time when nothing interrupts them -- such as, just maybe, being learned by hobbits.

Two types of "errors" (again) and it's all "grammar" (again)

Over on the eggcorn database site, I've added a comment on the lose-to-loose entry, which Chris Waigl has annotated with:

This entry has been assigned to the "questionable" category pending further discussion: [it] looks like a simple misspelling to me.

I agreed with Chris (and MWDEU), but the other comments mostly maintain that this can't be a spelling error -- several point out, in support of this claim, that it is very frequent -- and one (from Spindoc) even declares it to be "one of the more common and genuine grammatical errors in the language". I see now that I have once again failed to appreciate that ordinary people just don't distinguish two types of "errors" (Spindoc puts them together in the judgment: "Not a misspelling/typo") and also lump everything that's regulated in language under the heading of "grammar". Let me try to sort this stuff out one more time.

People who study errors in language make a systematic distinction between inadvertent errors -- in the case at hand, slips of the pen or typos -- and another type of mistake, which arises from imperfect command of the conventions at work in the larger community of language users -- in the case at hand, "spelling errors" in the sense of errors involving the conventions of spelling. Writing or typing "teh" for "the" is an inadvertent error, and a very common one. Writing or typing "loose" for the present tense or base form /luz/ of the verb whose past tense is spelled "lost" is, I maintain, almost always something else; people who spell this way, and there are a great many of them, almost always intend that spelling (while those who spell "teh" surely do not intend that spelling).

Now, spelling /luz/ as "loose" is a very likely and natural error to make: "lose" is one of only two English words, and the only verb, with /uz/ spelled "ose" (the spelling of possessive "whose" is equally surprising); the very common verb "choose" has /uz/ spelled "oose"; and anyway there's another very common word "loose" already hanging around. So plenty of people, having learned that /luz/ has a "-se" spelling with an "o" in it (rather than a "u"; "luse" would have been the most likely candidate spelling), opted for the reasonable (though incorrect) "oo" spelling. Unfortunately, the conventional spelling, "lose", is even odder than they realized.

Once you have a lot of people spelling /luz/ as "loose", they can serve as models for other people, so the spelling error will tend to maintain itself and spread. After all, this spelling makes sense; if you can't have "luse", "loose" is the next best thing.

Unless it's done very carefully, instruction might actually make things worse. A kid who learns that there's something odd about the way /luz/ is spelled, and that it has something to do with /lus/, and that lots of people spell it wrong, might simply conclude that the oddity is that these words are homographs, like the present and past tense verb forms spelled "read", and that it's the spelling "lose" that's wrong. That would reinforce "loose" as the spelling for /luz/. Another possible baleful effect of instruction is that the spelling of /lus/ could be cast into doubt; see below.

In passing, I note (once again) that very different kinds of conventions and knowledge are involved in spelling vs. pronunciation, syntax, morphology, vocabulary, and so on (not to mention discourse organization, linguistic politeness, and a variety of other skills of language use). Using "grammar" as a cover term for all this stuff merely sows confusion and misunderstanding.

Some of the comments on lose-to-loose on the eggcorn site suggest that there is confusion in both directions (though I have not seen examples of this), and even that pronunciations are affected (though I have not heard this). Both possibilities are alluded to in a comment from Patricia:

I agree this is not simply a spelling error or typo. "Lose" should not be pronounced in the same way as "Loose" or vice versa, but often is. As an English teacher in the public schools for 25 years, I have marked and remarked, discussed and rediscussed this error in both group settings and one-on-one. My limited successes have been the result of one-on-one, dogged persistence in correcting pronunciation and in insisting on visualizing the difference in the one "o" and the "double o."

I'm inclined to think that Patricia is showing a deep confusion of pronunciation and spelling here, and that her report is really just about more instances of the spelling "loose" for /luz/. But it would be possible for spelling reversal -- "lose" for /lus/ -- to take place, if the spelling of /lus/ (which is not very odd, although you do have to learn to select between "oose" and "uce" for /us/) has been cast into doubt. It's hard to see how you could get the pronunciation of a verb spelled "lose" as /lus/, but I can see how you could end up with a verb with the meaning of "lose" but pronounced /lus/; it would be spelled "loose", of course. This is one thing that could happen from students reading things like "You'll loose your wallet" in the writing of others who are imperfect spellers; after all, "loose" is pronounced /lus/, right? Such a student would have two verbs denoting accidental dispossession -- /luz/, learned from speech, and /lus/, learned from writing. You know, there are people who think that there are two different cities in southern California -- one with a name pronounced like La Hoya and learned from speech, and one with the name La Jolla, pronounced just like it's spelled (in English), learned from writing.

It's worth noting, once again, that very common spelling errors are always the consequence of imperfections in the conventions of spelling. People seek out system in these matters, and resist isolated peculiarities in favor of spellings that play by the general rules.

Somewhere back of the teeth in Glasgow

Another story about language, another case of attributional abduction.

Actually, there are two stories, one by Fiona MacGregor in The Scotsman, and another by Gerard Seenan in the Guardian (based on MacGregor's piece, apparently). Both are about Jim Scobbie's research on sound change in progress in Glasgow. The stories both put forward an interesting research question, and discuss an on-going research project that has turned up new evidence. However, both stories also quote the featured scientist as saying some things that don't quite make sense. In this case, as we've often had occasion to suggest before, it's a good bet to attribute the mistake to the journalists.

The two stories discuss changes in Glaswegian pronunciation that apparently bring it closer to Home Counties standards, and especially the loss of trilled /r/ in some contexts.

Residents of Maryhill, an area of Glasgow perceived to have a "strong" accent, are now dropping the consonant altogether at the end of words with words such as ‘car’ and ‘bar’ are changing in pronunciation to ‘cah’ and ‘bah’.

However, contrary to claims that English television programmes are influencing the way Scots speak, Scobbie believes the dropping of the final R is a natural development.

"To the casual listener it may sound like the R has been dropped completely as it is in England, but when you look at the ultrasound images you can see the tongue starts to shape an R before the sound trails off."

He said if it was a case of merely copying English pronunciation there would be no evidence of the R at all. And he claimed that within Glasgow there were also differences with the dropped R more common in Maryhill than Bearsden.

"There’s a natural tendency in all languages for consonants to become weaker at the end of the word than at the beginning. So you might still get the burred R at the start of word even if people are losing the final R. [from the article in The Scotsman]

So far, so good.

In fact, it's terrific to see questions like these discussed in the popular press. The basic facts are clear and easy for everyone to understand at an elementary level, people are interested in what's happening, and there are some fundamental issues in play. For example, this is the latest in a lengthening series of examples where "loss" of a phoneme has been shown to be the result of weakening a speech gesture to the point where it no longer has its normal acoustic effect, though it still exists in a vestigial form. Phoneticians have traditionally assumed that this sort of thing happens, but as far as I know, the first clear instrumental documentation of this phenomenon was by Louis Goldstein about 20 years ago. The availability of inexpensive, safe ultrasound imaging technology has made it possible to find many additional cases.

You could object a bit to the way the disjunctive choice is set up. Couldn't the influence of r-less mass media be interacting with -- not in opposition to -- the effects of articulatory lenition? And in any case, it's a little odd to frame things as a contrast between the influence of television and "a natural development". If television is part of the world people live in, then their response to television is as natural as their response to anything else, isn't it?

However, the bit that puzzled me came later on:

Scobbie added other changes that had been noticed in pronunciation including a loss of the aspirating H sound. As a consequence, words such as ‘which’ were beginning to be pronounced ‘wich’ north of the border and some young Scots were even being recorded as saying ‘lock’ instead of ‘loch’. [from The Scotsman again]

Now one of the alternative pronunciations of which really does involve "an aspirating H sound", that is, noise generated by turbulent flow of air through the glottis. But the difference between the two pronunciations of loch has nothing to do with "aspirating H" or the larynx, or for that matter anything else connected to the pronunciations of which in any way. The one that MacGregor writes "lock" ends in [k], a voiceless velar stop, whereas the one she writes "loch" ends in [x], a voiceless velar fricative.

I'm sure that Jim Scobbie knows all this at least as well as I do, and probably better, so I'm going to assume without argument that this mistake was introduced by Fiona MacGregor, and then picked up in an even more muddled form by Gerard Seenan in his Guardian story:

Dr Scobbie also noticed a decline in the aspirating "h" sound, with some young Scots committing the cardinal sin of pronouncing loch as lock. [from The Guardian]

Sidestepping the theology here, let's get the articulatory phonetics straight.

From the lips back to the larynx, the IPA names 11 places of articulation:

On this drawing of a midsagittal section of a human head, I've indicated with red and blue arrows the approximate location of a velar stop or fricative, and of the glottis where the noise in [h] (and the aspirated [w^h] in some pronunciations of which) is created.

As you can see, the velar and glottal positions are both "in the back of the mouth" but otherwise are not very close. Uneducated people often get confused about all those places in back of the teeth, taking a perspective sort of like Saul Steinberg's famous map of the view to the west from 9th Ave.

Looking in from the lips, I guess you can get just as confused about velar vs. glottal articulations as Steinberg's New Yorkers might be about the relative locations of Albuquerque and Los Angeles. Add the fact that loch is spelled with a digraph "ch" to represent the velar fricative written in IPA as [x], and that if you remove the "h" you get "loc", which looks like it should be pronounced the same as "lock", and there you are.

The mistake is not essential to the story -- but think of the fuss if a story in the Guardian referred to "Eastern European countries such as Bulgaria and Belgium".

And whose fault is it that today's intellectuals (including journalists) don't know the difference between spelling and sound, and are ignorant of the elementary geography of the vocal tract? I blame the linguists.

Unwinese

In response to Friday's post about Grammelot, Donal Lynch sent in a pointer to the British comedian Stanley Unwin. This BBC Unwin sampler offers two short passages in Unwinese about language. Here's my quick attempt at an orthographic transcription of the first one.

England joys all concentrate. Corruption of the English, well, English language. So manifest was so many careless. ((They mission it)) a sillibold, like "partic'ly". And ((intrudes ye)) an extra one, like "renovenate" instead of "renovate", as I heard by a zealous cockney who knowed it all. No bother, 'e think it, ((before)) out of the voice box he ho. Oh, no.

Corrections are welcome, since my linguistic and cultural background is inadequate to deal with some of this. For example, what I've transcribed as "intrudes ye" seems to be something like [ɪnˈtru.ʤi] in broad IPA (i.e. it rhymes with "tin fuji").

The practicioners of doubletal, comedic and otherwise, are legion. Their gobbledygook comes in many different forms -- perhaps there is a (folk-) taxonomy of doubletalk somewhere. In any case, doubletalk is often combined with bits of dialect or special spoken registers, perhaps because it's easier to persuade someone they're not quite getting it if the accent and other features index a form of speech that might have unexpected features.

Some more Unwin links: The World of Stanley Unwin; a BBC obit; some Stanley Unwin Transcripts; memorial at Bikwil.

Rats off to ya!

In today's NYT, Dave Itzkoff has an article on Cartoon Network's Adult Swim. If you haven't heard of this late-night collection of idiosyncratic animations, there are two reasons why you might want to check it out. The first reason is demographic:

When Nielsen begins tracking the block's ratings separately from Cartoon Network's daytime numbers at the end of this month, Turner Broadcasting expects "Adult Swim" to be the top-rated basic cable network among 12-to 34-year-old males.

The second reason, of course, is linguistic.

Itzkoff mentions in passing that the creators of one Adult Swim cartoon are consciously setting out to create catchphrases that will bond the audience to the show and to one another. Fair enough -- this mechanism, whether conscious or not, has worked for entertainers from Shakespeare to Matt Groening. But there's something new this time: it seems that these catchphrases are meaningless to start with, gaining significance only from their role in the show and its subculture. At least that's how Tim Heidecker and Eric Wareheim talk about a phrase promoted in their new series Tom Goes to the Mayor:

The playful visuals of "Tom Goes to the Mayor" are matched by its idiosyncratic sense of humor, one in which a meaningless catchphrase like "rats off to ya!" (a caption from a T-shirt that Tom fruitlessly tried to market) can form the basis for a secret language shared by fans of the show.

"There's no reason for anyone to be saying that," Mr. Wareheim said. "There's no basis for it at all. But people find little sayings like that, and it just takes over their lives. It becomes special to them because it's theirs."

Well, it worked for Monty Python and Star Trek, not to speak of various advertisers and academics too numerous to mention, even if it's traditional to start with a phrase that means something in its original context. But I'm not sure that this one is going to make the grade. A Google search for {"rats off to ya"} produces only 195 hits. Compared to 27,800 for {"I for one welcome"}, or 4,570 for {" No one expects the Spanish Inquisition"}, this suggests that relatively few lives have been taken over by TGTTM so far (though in fairness the episode only aired 12/19/2004). There seems to be some other evidence of lack of TGTTM life-taking-overness as well.

Perhaps you need something more than catchphrases? Or maybe it works better to start with the traditional sort of catchphrase, which functions as an ordinary piece of language in its original context, before being generalized by cultural resonance? I'm not sure about all that, but I can confidently tell you that we haven't heard the last from the catchphrase industry.

[Update: Tim May points out that { "nobody expects the spanish inquisition"} gets "an even more impressive ~14,600" Google hits.]

[Update #2: and of course the disjunction of both patterns gets more hits yet, not to speak of various generalizations of the pattern, as various other readers have written to explain. Here's a table of some additional results:

]

No word for sex

Novelist Frank Delaney was on NPR's Weekend Edition today, babbling to Scott Simon about the Irish and how different their world view is from everyone else. We Irish, he explained, have no word for yes in our language, and no word for no. And, he added, "no word for sex! But we're not going to get into that." Not him, maybe, but Language Log is not afraid to get into sex. I immediately suspected that Delaney was gilding the lily, and that this notion — a language with no word for the absolute best way to have fun that does not involve laughing — would turn out to be utter nonsense. I therefore turned, as I have before, to one of the foremost modern Irish specialists in the world, my colleague at UC Santa Cruz, Jim McCloskey. Could it be true, really, I asked him, that Irish has no word for sex?

He sighed the deep sigh of a man who often has to listen to people talking ignorant nonsense about a dying language that he loves and has studied for forty years. And he remarked that FrankDelaney couldn't have spent much time around people who actually speak the language. And he then listed the first few Irish terms for various kinds of sex and sexual activities that came to his mind:

ag bualadh leathair
ag déanamh pite boid
ag déanamh pís ghlaice
ag ealáin
an cleas a dhéanamh
collaíocht
craiceann a bhualadh
craiceann a dhéanamh
cuid a dhéanamh de
focáil
gnéas

(I dare not tell you what these mean, because of FCC regulations. I've put them in alphabetical order. Some dictionaries fail to list some of them, but all are in use in the native-speaker community. Two of the terms refer to masturbation, but hey, as Woody Allen said, that's sex with someone you love. Notice that three of the terms are single words.)

So what on earth can Frank Delaney be thinking of? Why do people say these things about languages they purport to care about, and do absolutely nothing to check up on whether the things they are saying are even remotely close to the truth? Why is everyone so given to bullshitting about language and thought, even about the language of their own countrymen?

The story about Irish lacking particles meaning "yes" and "no" is true, by the way. But it has nothing to do with the Irish mind or spirit or way of looking at the world or the notion of neither agreeing nor disagreeing. In Irish you repeat the verb of someone's clause to agree with it (as if someone said "Got milk?" and the way you gave an affirmative response was to say "Got"), and you repeat their verb with the negation particle in front to deny it ("Not got"). But the same is true of Chinese. Anyone want to suggest that the Chinese have exactly the same cultural propensities and outlook on life as the Irish? More bullshit about language and thought.

Here's some advice. Whenever you hear someone starting to say something that begins with "The X have no word for Y", or "The X have N different words for Y", never listen to them, and always check your wallet to make sure it's still there.

It's not our fault

Eric Gibson doesn't like "Art Since 1900". It's not the art he dislikes, but the book, authored by Hal Foster, Rosalind Krauss, Yves-Alain Bois and Benjamin Buchloh, and just published (3/15/2005) by Thames & Hudson. And he really, really, really doesn't like this book, to the point where he offers this "suggestion for parents of high-school students":

Find out whether the college that your child hopes to attend plans to assign "Art Since 1900" in its art-history courses. If so, apply elsewhere.

Since Gibson is The Wall Street Journal's Leisure & Arts features editor, this may be bad news for admissions statistics at Princeton, Columbia and Harvard, where the book's authors teach. Or maybe not, I don't know. My own interest in the matter is not institutional but disciplinary. Gibson blames everything on psychoanalysis, Marxism -- and linguistics:

Since the early 1980s, the authors have been at the forefront of "the new art history," an interpretive school whose fullest expression can be found in October magazine, a quarterly they founded in 1976. (The name is meant to evoke revolutionary associations.) It was in October that Hal Foster, Rosalind Krauss, Yve-Alain Bois and Benjamin H. D. Buchloh led the charge against the traditional, aesthetic appreciation of art, supplanting it with psychoanalysis, political ideologies such as Marxism and various forms of French theory, like those derived from linguistics. According to this school of thought, a painting isn't merely an abstract or representational image on canvas but a social "text" to be interpreted and deconstructed. [emphasis added]

I'm not sure which of the "various forms of French theory" particularly offend him, but his review mentions Jacques Derrida and Michel Foucault. To the extent that it's possible to determine what these authors mean, I don't find anything important in their "theories" that can be plausibly said to be "derived from linguistics". I'll continue to put off an evaluation of Derrida's language-related ideas -- including the "everything is a text" meme that so clearly offends Gibson -- but let me say for now that his ideas were "derived from linguistics" roughly in the sense that the views of the Marquis de Sade were derived from Catholicism.

As for Foucault, the article about him in the Stanford Encyclopedia of Philosophy does mention (the famous linguist) Saussure, but describes nothing that I would call a consequential influence. And along with works on prisons, clinics and sex, Foucault did once write a book called "Les mots et les choses" ("words and things"), in which he (very obscurely) discusses (the early linguist) Rasmus Rask along with David Ricardo and the Comte de Buffon. But on that evidence, you might as well say that Foucault's theories are "derived from economics" or "derived from biology".

So why might Gibson blame Foucault & Co. on linguistics? Perhaps it has something to do with the modern continental-philosophy sense of discourse:

1976 T. EAGLETON Crit. & Ideology ii. 54 A dominant ideological formation is constituted by a relatively coherent set of ‘discourses’ of values, representations and beliefs.

I'm sorry to say that I found that quotation in the OED's 1993 additions to the entry for discourse the noun, as a citation for an additional sense attributed to linguistics:

[3.] e. Linguistics. A connected series of utterances by which meaning is communicated, esp. forming a unit for analysis; spoken or written communication regarded as consisting of such utterances. Also transf. in Semiotics.

But the Eagleton quote is not about that sense of discourse at all. The other citations are mostly appropriate ones for the gloss given:

1951 Z. S. HARRIS Methods in Structural Linguistics iii. 28 For the incidence of formal features of this type only long discourses or conversations can serve as samples of the language.
1957 G. L. TRAGER in Encycl. Brit. XIV. 162H/2 The syntax of any language can be arrived at in analogous ways. The phonologically determined parts of a discourse are found, and their constituent phrases separated out.
1983 BROWN & YULE Discourse Analysis ii. 29 We can see little practical use, in the analysis of discourse, for the notion of logical presupposition.

This sense of discourse is relatively straightforward -- it just means "chunks of speech or text bigger than a sentence", such as stories, conversations, arguments and so on. "Discourse analysis" in this sense is just "the analysis of the form, meaning and use of language in pieces bigger than a sentence".

But what Foucault and his followers mean by a discourse is something completely different. I haven't been able to find a good definition -- the Foucaultian culture is not one in which clarity is prized -- but it seems to mean something like "a set of social norms that define who can talk how about what, to whom and when, including provision of words, phrases and larger linguistic forms, as well as restrictions on accessible ideas and assumptions".

Here is a set of somewhat random selections from the first page or so of what Google returns for {Foucault discourse}:

To Foucault, discourse is just one of the rules of society he examines critically. Language too, he argues, transcends and even obviates individual perception rather than allows our independent existence to flourish.
What interested him were the rules and practices that produced meaningful statements and regulated discourse in different historical periods.
Discourse is "a group of statements which provide a language for talking about ...a particular topic at a particular historical moment".
Discourse, Foucault argues, constructs the topic. It defines and produces the objects of our knowledge. It governs the way that a topic can be meaningfully talked about and reasoned about.
In more general terms for F. it is discourse as a medium for power that produces subjects or, as he puts it, "speaking subjects," which, for him, are the only kind there are.
"[I]n every society the production of discourse is at once controlled, selected, organised and redistributed according to a certain number of procedures, whose role is to avert its power and its dangers, to cope with chance events, to evade its ponderous, awesome materiality"
Discourse operates by "rules of exclusion" concerning what is prohibited. Specifically, discourse is controlled in terms of objects (what can be spoken of), ritual (where and how one may speak), and the privileged or exclusive right to speak of certain subjects (who may speak).
This kind of discourse is produced by a "will to knowledge" or "will to power," wherein discourses "discipline" us: "a will to knowledge emerged which . . . sketched out a schema of possible, observable, measurable and classifiable objects; a will to knowledge which imposed upon the knowing subject-in some ways taking precedence over all experience-a certain position, a certain viewpoint, and a certain function . . . " (218). So the will to knowledge or truth or power has a history-"of a range of subjects to be learned, the history of the functions of the knowing subject," etc.
Some of the most important contemporary discourses are "disciplines," by which he particularly means academic disciplines that discipline our thinking. Disciplines are "opposed to . . . the author, because disciplines are defined by groups of objects, methods, . . . the interplay of rules and definitions, of techniques and tools: all these constitute a sort of anonymous system . . . without there being any question of their meaning or their validity being derived from whoever happened to invent them"

You can argue about whether particular pieces of this stuff are meaningful or meaningless, profound or trivial, useful or useless, true or false. You could even argue that this is the kind of thing that linguists ought to have thought about and worked on, in principle, since it purports to establish a general framework for how language can be used in a given place and time. But in sober historical fact, none of this is "derived from linguistics", and linguistics has been less affected by this discourse discourse than any of the other fields in the humanities and social sciences. [Well, except maybe for economics..]

It's ironic, even poignant, to blame the discipline of linguistics for postmodern "theory", since its proponents in the American academy generally appear to despise linguistics. They haven't in general studied linguistics themselves, and they've removed not only requirements but even recommendations for linguistics courses from the curriculum for their undergraduate majors and graduate students alike. Some of the blame for this estrangement belongs to the linguists, no doubt, but regardless of fault, the rise of postmodernism has been a disaster for our field.

So, Mr. Gibson, give us linguists a break. We didn't do it!

More WTF coordinations

As I noted in an update to my post on my reaction to an odd coordination that I read, Neal Whitman wrote to tell me about his recent article in Language 80.3 (pp. 403-434) on this topic, entitled "Semantics and Pragmatics of English Verbal Dependent Coordination" (sorry, access to Project Muse required for the link to work). Neal also provided the following additional examples:

It makes it hard for him to get [his stuff done] and [to bed on time].
She wants [an engagement ring] and [her boyfriend to stop dragging his feet].
Don't eat [fast food], or [at restaurants, food-service companies, or caterers].

The last of these examples is perfectly fine for me, underscoring the apprehensiveness I had about saying that only phrases with the same syntactic category can be conjoined: [fast food] is a noun phrase, and [at restaurants ...] is a prepositional phrase. The other two examples are different, though; my knee-jerk, WTF reaction is to give the first a question mark (by which I mean that it's somehow borderline between grammatical and ungrammatical) and the second a star (by which I mean that it's ungrammatical -- except that it improves somewhat if a for is added before the second conjunct).

Mike Pope also wrote to comment:

Would you say that this is a form of zeugma? The small child of a friend of mine once said "The sun makes you hot and sneeze," which seems at least similar in spirit to what you've got here.

As explained here, zeugma is "A construction in which a single word, especially a verb or an adjective, is applied to two or more nouns when its sense is appropriate to only one of them or to both in different ways, as in He took my advice and my wallet."

This [my advice] and [my wallet] (noun phrase and noun phrase) example is fine for me; Mike's [hot] and [sneeze] (adjective and verb) example is not (but it must have been fun to hear a kid say it). If they're both just examples of zeugma, why is that? WTF?

Now consider the following example (from Life of Pi, pg. 37):

I nodded so hard I'm surprised my neck didn't snap and my head fall to the floor.

When I first read this a few months ago, I had an even bigger WTF reaction than for any of the others. But I immediately reasoned through it and now find it almost perfectly grammatical. All that it took was the recognition that the negation expressed by "didn't" in the first conjunct takes scope over both conjuncts ...

NOT [ [my neck snap] and [my head fall to the floor] ]

... and that this means something subtley different from having two negations, each taking scope over one of the conjuncts ...

[ NOT [my neck snap] and [ NOT [my head fall to the floor] ]

as in:

I nodded so hard I'm surprised my neck didn't snap and my head didn't fall to the floor.

Which is ambiguous (as my colleague Andy Kehler pointed out to me) between a reading in which the neck-snapping causes the head-falling and one in which there is no causation (as pragmatically odd as that might be); in other words, causation between the first and second conjuncts is not necessary in this second sentence while it is in the original.

(Andy also reminds me of Arnold Zwicky's post from last August about grammatical and ungrammatical coordinations, sparked in part by a suggestion by Neal Whitman. Some of us here at UCSD are planning to read and discuss Neal's Language paper sometime next quarter; if anything particularly interesting comes out of that -- or if we have any more WTF reactions worth commenting on -- you'll hear about it here on Language Log.)

[ Comments? ]

Simlish as 21st-century Grammelot?

The things I've read about Grammelot, and the bits of it that I've heard, remind me of Simlish, the fake language used in The Sims and its follow-on games. In case you're not a Sims person, here's a bit of Simlish motherese and some Simlish food enjoyment, just to give you the flavor of this "language".

According to the article on Simlish where I found those samples:

... when The Sims was originally designed Will Wright wanted the language the Sims spoke to be unrecognizable but full of emotion. That way, every player could construct their own story without being confined to a Maxis-written script (to say nothing of the mind-numbing repetition). We experimented with fractured Ukrainian (one of the original The Sims designers was a native speaker), and the Tagalog language of The Philippines. Will even suggested that perhaps we base the sound on Navajo, inspired by the code talkers of WWII. None of those languages allowed us the sound we were looking for – so we opted for complete improvisation, originated and performed by some SF Bay Area professional actors whose specialty was improv; Stephen Kearin and Gerri Lawlor.

So now it is five years since Simlish was born, and the tradition is carried on by an additional nine actors performing all the age range voices: baby, toddler, child, teen, adult and elder. The auditions were held January through May 2003 with over 100 actors from SF and LA trying out. We selected a highly talented cast and quickly got them to work recording many full 8 hour days, recording voice to over a hundred animations a day, resulting in thousands of takes a day.
With such a huge amount of voice data - 40,620 samples at the moment our pre- and post-production processes have been streamlined to get the voice into the game so it can be listened to, assessed and either re-recorded (in rare cases) or hacked (constructed out of similar-sounding files) . Currently there are at least six Maxoids dedicated to getting the sound effects and voice in the game.

And even though I said Simlish is not a language per se, there are some common words that we directed all the actors to perform. If you listen closely, you’ll hear a word that means baby (nooboo), another for pizza (chumcha), and another phrase said during the Dirty Joke interaction that isn’t exactly defined, but just seemed to fit. That one was invented by Liz Mamorsky, our elder female voice.

Near the bottom of this page, there's a link to a video interview with one of the people who does Simlish "voiceovers" (described as "a behind the scenes look at how Simlish is created"). There are now pop songs in Simlish, and Ravi Purushotma thinks that Simlish is a good way to teach foreign languages. Well, really, it's the textual instructions that give the foreign language practice -- but the characters give you feedback in Simlish if you misunderstand and (say) have your character take a shower when what he really needs is lunch.

Here's a recent interview with Will Wright, the creator of The Sims and its sequels and add-ons.

[Update: A sort of Simlish dictionary is available.]

Bovine excrement on NPR

In connection with Harry Frankfurt's book Bullshit (discussed briefly on Language Log here, here, and here), I forgot to mention a week ago that I heard a National Public Radio talk show devoted to the book, with Professor Frankfurt on the line to talk about his book and various other professors and comentators brought in from time to time; but because of the goddamn Federal Communications Commission, no one was at any time during the hour permitted to mention the name or the topic of the book. Makes you proud to be an American, don't it? You can use it, you can speak it all the time, but you won't ever have to face it being named on the radio...

Gibberish by any other name

David Donnell sent in a link to an item in the Dover-Sherborn Press concerning the "magical sounding gibberish language called Trammeled that originated in Europe centuries ago":

Alexander, King of Jesters, will perform his unusual comedy at Kraft Hall in Dover Church Sunday, March 20, 3 p.m. [...] This offbeat performance includes Renaissance water spitting, the nose flute serenade, jingle bell juggling, twisted sticks of the forest, balancing stunts high above the audience and other unusual routines inspired by the tradition of the fool.

Alexander enjoys surprising his audience with a rapid fire of visual gags. In one routine, he observes them through a darting eyeball in his mouth shortly before he crams three flutes into his mouth. Without missing a beat, he plays a sweet tune in staggered rhythms on all three at once.

During the act, Alexander speaks a magical sounding gibberish language called Trammeled that originated in Europe centuries ago. In those times, nearly every village had its own dialect, sometimes even a completely different language. Trammeled helped traveling jesters and minstrels cut through language barriers that presented themselves every few miles. It serves Alexander today, since he often performs abroad.

David was intrigued:

I don't know what kind of lingo "Trammeled" is, but it sure sounds magical... On one hand it's "gibberish", on the other it's useful "abroad" and helps "cut through language barriers"!

I Googled it quickly, but didn't see any refs to a language by that name.

Me either. But checking for "Alexander, King of Jesters" turned up this study guide from the Alaska Junior Theater, in Fairbanks, which mentions "the jester gibberish language 'Grammelot'". Alexander's Alaska hand-out says that

Many jesters and fools spoke a gibberish language called Grammelot that was first described over 500 years ago. It consisted of funny sounds along with a few real words from different languages. Although Grammelot could not say everything quite as clearly as a real language, it could express general ideas and it engaged people's imaginations. It also turned out to be very practical because:

1. Villages were remote centuries ago. They were separated by dark woods. The terrible roads made it was hard to leave town, and without TV or radio, the peasants of one village may never hear the accent of the people in the next town. As a result, even neighboring villages might not understand each other. Every town spoke a little differently, and so each town had their own dialect. Sometimes they spoke very differently, and had their very own language. Not surprisingly, there were far more languages then than there are today.

2. Free speech was not a right centuries ago In the days before mass media, it was the traveling perform-ers who gave peasants much of their news of the outside world. If anyone said something that angered the king or queen, he or she could easily be thrown in jail. The censors watched performers very closely. The censors were the people hired by the king or queen to make sure that nothing was said that could upset them or the royal court. If the jesters spoke Grammelot, the censors were less likely to give them a hard time, since nobody knew exactly what they were saying.

The "study guide" also provides a glossary for "Alexander's version of Grammelot":

co yo yo - curly or twisted. It is also used in slang to mean 'wow' since jesters prefer things that are twisted and bent.
hodio tonada - an exclamation of surprise like “woah, check it out!”
Waga dee bwa - I am
Waga da bala - He is
Kafuggo! - darn it!
bo-whoo - (low voice) the big one
wuh hoo - the medium one
eee oo - (pronounced in a high voice) the little one
bweesta - fish
galeggwi - up
jiffa - middle
peet - down
Basnop ka dipple yadda yadda - rather than try to fix the problem, just validate my feelings
Kafuggo! Snippa blop! - But I'm just trying to help!

Right.

Well, Grammelot is apparently for real:

Il Grammelot è una forma di teatro inventata dai comici della Commedia dell’Arte del 1400 ed è organizzata in chiave onomatopeica, ossia il riuscire a far arrivare concetti attraverso suoni che non sono parole stabilite, convenzionali.
Questi giullari usano intruglio di dialetti e parole inventate che rendono immediata e molto colorita la recitazione, nella quale predomina anche una gestualità ed una mimica molto accentuate.

Dall’insieme di queste componenti viene fuori un tipo di teatro estremamente espressivo, iperbolico, esilarante, viscerale, diretto e quindi comprensibile un po’ dappertutto e ad ogni tipo di pubblico.

Un modo di recitare in cui il linguaggio usato perde di significato letterale, per diventare suono, vibrazione, musicalità che comunica emozioni e suggestioni.

Grammelot is a form of theater invented by the comic actors of the Commedia dell'Arte of 1400, and is organized in an onomatopoeic mode, that is, it manages to evoke concepts by means of sounds that are not established or conventional words. These jokers use a mixture (?) of dialect and invented words that make for a vivid and very colorful performance, in which exaggerated gesture and mimicry are also very prominent.

From the combination of these components there emerges a type of theater that is extremely expressive, hyperbolic, hilarious, visceral, direct and therefore somewhat comprehensible everywhere and to every type of audience.

A way of speaking in which the language that is used loses its literal meaning, in order to become sound, pulsation, musicality that communicates emotions and suggestions. [apologies for my translation -- I read Italian by triangulation from Latin and French...]

There are some quicktime videos with transcripts on the same site.

The word grammelot is not in the OED -- which is a surprise -- nor in the other dictionaries that I checked. Apparently this word has never caught on in the history of the English language. For example, it doesn't occur in the LION database. Perhaps there is some other word that's used instead.

Anyhow, the one remaining mystery is how grammelot got morphed into trammeled over at the Dover-Sherborn Press. A clue is provided by the fact that the very same wording recurs in other announcements for Alexander, King of Jesters:

During the act, Alexander speaks a magical sounding gibberish language called Trammeled that originated in Europe centuries ago. In those times, nearly every village had its own dialect, sometimes even a completely different language. Trammeled helped traveling jesters and minstrels cut through language barriers that presented themselves every few miles. It serves Alexander today, since he often performs abroad. (from the Dover-Sherborn Press)

During the act, Alexander speaks a magical-sounding gibberish language called Grammelot that originated in Europe centuries ago. In those times nearly every village had its own dialect, sometimes even a completely different language. Grammelot helped traveling jesters and minstrels cut through language barriers that presented themselves every few miles. It serves Alexander today since he often performs abroad. (from the Wilbraham Public Library News)

As you can see, these passages are identical except for some differences in punctuation and the Grammelot -> Trammeled substitution. I think we can assume that this was a scribal error in a chain mediated by speech. Perhaps the reporter called in the story, reading Alexander's press materials, to someone who typed it up from the tape. Or perhaps the reporter used an automatic speech-to-text dictation system, which of course did not have grammelot in its lexicon, and then didn't proofread carefully enough. My money is on ASR.

Please notice that I resisted the temptation to make the audio track of a Grammelot monologue into a language identification quiz. (I do have another quiz queued up, contributed some time ago by Stefano Taschini, which I have neglectfully failed to post...)

[Update: several people have written in to point out that Microsoft Word's spelling checker corrects "Grammelot" to "Trammeled", and is therefore the most probable culprit in this case.]

[Update 3/12/2005: Ray Girvan has turned up some more accurate information on the origins of (recent versions of) Grammelot:

The modern popularisation seems to have sprung from Dario Fo's one-man touring show, "Mistero Buffo", Various references imply that while Grammelot existed generically, he has reinvented it in a personal form tailored for modern relevance. The Times review of "Mistero Buffo" (Saturday, Apr 30, 1983) mentions that his three grammelot sketches, though based on mediaeval texts, were performed in separate Italian, French and American grammelots.

Dario Fo biography:
"'grammelot', a language derived from the mixture of modern phonemes and dead dialects from Italy's Po valley area".

Fo's own Nobel speech:
"Ruzzante, the true father of the Commedia dell'Arte, also constructed a language of his own, a language of and for the theatre, based on a variety of tongues: the dialects of the Po Valley, expressions in Latin, Spanish, even German, all mixed with onomatopoeic sounds of his own invention. It is from him, from Beolco Ruzzante, that I've learned to free myself from conventional literary writing and to express myself with words that you can chew, with unusual sounds, with various techniques of rhythm and breathing, even with the rambling nonsense-speech of the grammelot".

http://mercury.web.ca/archives/caea-l/2004-October/001672.html:
The Italian Cultural Institute, in association with BellaLuna Productions Presents a one-day workshop with Mario Pirovano, Italian protégé of Nobel Prize-winning Playwright Dario Fo.: Part 2 ... Grammelot re-invented by Dario Fo and its present-day relevance.

http://lists.econ.utah.edu/pipermail/marxism-thaxis/1997-October/003972.html:
His one-man tour de force, "Mistero Buffo" ("Comic Mystery"), written in 1969, finally had its United States premiere at the Joyce Theater in Chelsea in 1986. It has its stylistic roots in the strolling players and minstrels of the Middle Ages. But it was also a timely satirical blast at religion and politics, delivered in Grammelot, a kind of double-speak masquerading as a language, wholly invented by Fo himself.

Apparently "Ruzzante" (real name Angelo Beolco) wrote in Paduan dialect. Whether the bits of German and so on that Fo remarks on were part of the rural speech of the period, or Beolco's introductions, is not clear to me.

Although similar languages have precedents - eg Lingua Franca, Polari - it wouldn't surprise me if Fo had coined the term "Grammelot".

Reinventing macaronic language seems to be a particular fascination for modern Italian authors (this recalls Umberto Eco - Salvatore in "The Name of the Rose" and the peasant hero of "Baudolino" use similar languages).

Ray also points out this article entitled "The Modern Macaronic", by Albert Sbragia, dealing especially with Carlo Emilio Gadda. An interesting quote:

The Renaissance macaronic in its purest form is a northern Italian creation with its precedents in medieval burlesque, goliardic verse and sacred parodies, and with extra-Italian continuators and resonances in various European countries and in Rabelais. Its origins lie in the late fifteenth-century Benedictine athenaeum of Padua and specifically in the linguistic experimentalism of Tifi Odasi, whose poem Macaronea defines the genre. Its fame was assured in the first half of the following century by Odasi's Mantuan pupil and emulator Teofilo Folengo (pseudonym Merlin Cocai). Folengo's Baldus (four editions: 1517, 1521, 1534-35, and posthumously in 1552) is a mock-epic poem of giants and farfetched chivalric adventures including the discovery of the mouth of the Nile and a final descent into Hell. Baldus is the genre's acknowledged masterpiece, and it enjoyed a notable popularity in the 1500s with over a dozen editions and reprintings. It was not without influence on Rabelais's Gargantua and Pantagruel, in which it is cited more than once. Such was the perceived connection that the first French translation of Folengo's works in 1606 bore the title Histoire maccaronique de Merlin Coccaie, prototype de Rablais.

The original macaronic is characterised linguistically by its vocabulary of Italian, dialect, and Latin words within a substantially Latin morphological, syntactic, and prosodic form. The hybridisation is typically trilingual in the northern Italian macaronic poets involving Latin, Italian (Tuscan), and Po Valley dialects. Not the natural or ingenuous product of a native plurilingualism, the Italian macaronic is a sophisticated caricatural artifice, a linguistic parody which exploits the situation of polyglossia experienced by the cultural elite. The demise of the original macaronic is due precisely to the success of the Italian humanists in their philological recuperation of classical Latin which had made possible the complexity of macaronic verse in the first place. Latin as a literary language was frozen in a normative a chronological straitjacket as an eminently noble but irremeably ancient language. As a result, Tuscan Italian was finally able to assert itself fully as the contemporary national language of letters, and it underwent much of the same normative and chronological classicising restrictions as had Latin. From a state of triglossia the linguistic and literary evolution the Italian peninsula would evolve more clearly as a case of fragmented diglossia, with numerous epicentres of dialect in tension with written and literary Italian.

]

[Update -- more on grammelot here.]

Stunningly inept modifier manners

A small circle of grammarians that I belong to (Rodney Huddleston, Chris Potts, Arnold Zwicky, and me) collects "dangling modifier" examples (I said something about some related examples in this recent post). That is, we are collecting published sentences in which a fronted modifier constituent that is intended to have a predicative interpretation (roughly, it is meant to have an understood subject) gets interpreted with something other than the matrix clause subject as its intended subject, or fails to be interpreted with an appropriate subject at all. Arnold Zwicky recently collected a speciment in a Palo Alto Daily News story (March 5, page 69) about the Palo Alto Toyota dealership, which is probably going to have to move to a larger site outside of the city to gain higher sales volume, according to the manager, Mr Kopacz. The story focuses mainly on the plans for the new site. But then it continues:

Generating $66 million in sales revenue last year, Kopacz estimates that a larger dealership with a freeway billboard could generate $130 to $140 million in sales.

Who's doing all that generating? Kopacz?

As Arnold remarked when he sent this example around, there's no way you can get that initial clause generating $66 million in sales revenue last year to feel like it's about the present dealership. Struggling to find a suitable subject for generating, one comes upon the main clause subject Kopacz, and one inexorably comes to think for a few seconds that he is the one generating that sales revenue. But's that's not what they meant. They meant that the old dealership soon to be replaced generated $66 million last year. But they didn't make that clear.

Arnold calls the sentence "stunningly inept". I agree with him. The line we take on examples of this kind, you see, is not that they violate the syntactic correctness conditions for English — they are simply too common for that to be the case. Roughly, what we think is that the syntax of English leaves things open for you to design your paragraphs in such a way that preposed non-finite adjunct clauses will, in context, be easily and naturally linked up with suitable understood subjects. And as always when you are left some freedom to do things whichever way you judge to be appropriate, you can screw it up. You can write something stunningly inept that baffles the heck out of an intelligent reader for several seconds. As I've said before, that should at the very least count as bad grammatical manners, the syntactic analog of dining goofs like rinsing your fingers in the consommé, or eating the butter from the butter dish.

Fragments from my to-blog list

Blogdigger is a useful way to search blogs, somewhat complementary to technorati (and lots faster).

Jean Véronis documents a strangely exact doubling of Yahoo's search count. And I've been meaning for a while to link to his exploration of yoghurt spelling on the web (in French).

Chris Waigl links to Avibase, "a multilingual bird database with 1.9 million entries on 10,000 species and 22,000 sub-species. In case you need to know that the Black-crowned Sparrow-Lark is called Saharanvarpuskiuru in Finnish."

There's a story in Wired on the Phraselator speech-to-speech translation device, which has been used in recent tsunami relief operations as well as in Afghanistan and Iraq.

Cultural Constraints on Grammar

Dan Everett wrote to say that his paper "Cultural Constraints on Grammar" will be published in Current Anthropology as a main article, for which the journal will solicit 15 commentaries.

Here's the abstract:

The Pirahã language challenges simplistic application of Hockett's (1960) nearly universally-accepted 'design features of human language', by showing that some of these design features (interchangeability, displacement, and productivity) may be culturally constrained. In particular Pirahã culture constrains communication to non-abstract subjects which fall within the immediate experience of interlocutors. This constraint explains several very surprising features of Pirahã grammar and culture: (i) the absence of creation myths and fiction; (ii) the simplest kinship system yet documented; (iii) the absence of numbers of any kind or a concept of counting; (iv) the absence of color terms; (v) the absence of embedding in the grammar; (vi) the absence of 'relative tenses'; (vii) the borrowing of its entire pronoun inventory from Tupi; (vi) the fact that the Pirahã are monolingual after more than 200 years of regular contact with Brazilians and the Tupi-Guarani-speaking Kawahiv; (vii) the absence of any individual or collective memory of more than two generations past; (viii) the absence of drawing or other art and one of the simplest material cultures yet documented; (ix) the absence of any terms for quantification, e.g. 'all', 'each', 'every', 'most', 'some', etc.

It's worth underlining that Dan sees the situation in terms of "cultural constraints on grammar". Some people see the question as whether language constrains (individual) thought, with culture emerging as a kind of amalgam of the psychology of individuals. Dan takes a different perspective:

Before beginning in earnest, I should say something about my distinction between 'culture' and 'language'. To linguists this is a natural distinction. To anthropologists it is not. My own view of the relationship is that the anthropological perspective is the more useful. But that is exactly what this paper purports to show. Therefore, although I begin with what will strike most anthropologists as a strange division between the form of communication (language) and the ways of meaning (culture) from which it emerges, the conclusion of the paper is that the division is not in fact a very useful one and that Sapir, Boas, and the anthropological tradition generally has this right. In this sense, this paper may be taken as an argument that anthropology and linguistics are perhaps more closely aligned than, say, psychology and linguistics, as most modern linguists (whether 'functional' or 'formal') suppose.

Thus Dan is separating Whorf (at least as he is generally understood) from Sapir, and aligning himself with Sapir:

I also argue against a simple Whorfian view, i.e. against the idea that linguistic relativity or determinism alone can account for the facts under consideration. In fact, I also argue that the unidirectionality inherent in linguistic relativity may offer an insufficient tool for language-cognition connections more generally, for failing to offer a more fundamental role for culture in shaping language.

Dan's article is an important one, and if you're interested enough in language to be perusing this blog, you should read it.

Exercise for the reader: if Dan's analysis of Pirahã is correct, what are the implications for studying languages like English? Turning it around, what are the implications for these ideas of what we know about languages like English and the cultures they're embedded in?

Up for the downshift

My radio alarm is set to my local public radio station. This morning, I woke up to hear the following:

... and then rotate it clockwise, as I'm looking at it ...

It turns out this was the second part of a two-part story on two wounded Marines, and Marine 1st Sgt. Brad Kasal was talking about a contraption on his wounded leg; as NPR's Joseph Shapiro explains: "Twist one of the screws, and Kasal's bone is literally pulled apart, just a tiny bit. That stimulates the bone to grow; the muscles, nerves, blood vessels, and skin too. Eventually, Kasal will turn these screws himself, a little every day." Kasal goes on: "Right now the doctors are still doing it, and it's relatively simple. It's a little bit uncomfortable ... well, quite a bit uncomfortable ..."

Now I have no idea what this contraption (called a "ring external fixater") looks like or how the screws are oriented, but I think it's safe to assume that the distinction between turning them clockwise and turning them counterclockwise is pretty important. This reminded me of the (comparatively far more trivial) discussion a few months ago about shuttle-loop directions and related matters (see here, here, here, here , here, here, here, and even over there).

Actually, I've just been searching for an excuse to come back to this issue, because I spent quite a bit of time thinking about it in relation to bicycles -- specifically, the gear shifters on mountain bikes. (If you're really interested in the technicalities of this topic, I suggest looking here and/or here. I need to get into a few technicalities here, but I'm putting aside the more complicated and interesting ones in the interests of my own time.)

In case you didn't know, gear shifters are the things on your handlebars that allow you to change gears. The shifter on the left-hand side of the handlebar (as you're mounted on the bike -- the shifter you would control with your left hand while riding normally) controls the front derailleur and the one on the right-hand side controls the rear derailleur. As explained here:

The derailleur is the device that changes gears by moving the chain from one sprocket to another. There are two derailleurs: one on the rear and one on the front. The highest ratio (when the bike can go fastest) is produced when the chain is on the biggest sprocket in the front and the smallest in back. The lowest ratio (the bike is easiest to pedal up hills, but very slow) is produced when the chain is on the smallest sprocket in front and the biggest in back.

In the front, the biggest sprocket is on the outside and the smallest is on the inside; in the back, the smallest is on the outside and the biggest is on the inside. Thus, when you're in the lowest gear, the chain is more or less parallel and flush with the bike frame; when you're in the highest gear, the chain is parallel but furthest away from the frame.

A lot of this information is second nature to many biking enthusiasts, but the average bike-riding person doesn't care about anything other than the following distinction: shifting one way makes it easier to go uphill, shifting the opposite way makes you go faster otherwise. Reflecting this need of the average Joe, many (standard) gear shifters have numbers to indicate the difficulty of the gear. For example, a 21-gear bike (3 sprockets in the front, 7 in the back) has numbers 1-3 (easiest-hardest) indicated on the left-hand shifter and 1-7 (also easiest-hardest) on the right-hand shifter.

Over the past 15 years or so I've ridden mountain bikes with three types of shifters, which I believe entered the market more or less in the following order. (There are probably other, newer, cooler, better, more expensive ones now; I limit my attention to these three.)

1. "lever shifters", with which you shift gears by moving a lever back and forth with your thumb and/or forefinger -- or maybe left and right or up and down, depending on the lever's orientation on the handlebars. Moving the lever one way shifts to a higher gear, moving it the other way shifts to a lower gear.
2. "click shifters", which which you shift gears by clicking one of two buttons, usually oriented one on top of the other within easy access of the thumb and/or forefinger. Clicking one button shifts to a higher gear, clicking the other shifts to a lower gear.
3. "grip shifters", with which you shift gears by turning a knob embedded in the handlebar grips themselves, such that you simply have to rotate your wrist one way or the other (toward you or away from you). Rotating your wrist one way ... well, you get the idea.

Here in San Diego I own a bike with click shifters, but while I was in Vancouver last term I bought a used bike with grip shifters. While riding the Vancouver bike I noticed something interesting (and relevant to the shuttle-loop business): with the right-hand shifter, I had to rotate my wrist away from me to shift to a higher gear (which is sort of intuitive -- it feels more or less like "up", and "up" = "higher"), but with the left-hand shifter I had to rotate my wrist toward me to shift to a higher gear (which is sort of counterintuitive, mutatis mutandis).

If you're having trouble picturing this, here's another way to look at it. Suppose I'm on my bike and you're standing beside it, facing it from the right-hand side (like this). To shift to a higher gear, I would rotate my right wrist in such a way that it would turn clockwise from your point of view. If you changed your point of view by standing on the other side of my bike, my left wrist would also turn clockwise when I shift to a higher gear.

But as the person riding the bike, I only have one possible point of view -- and from that point of view, my wrists rotate in opposite directions to achieve what is essentially the same goal. Why did the bike (part) manufacturers do this? My brother, an avid cyclist, suggests that maybe it's because the real goal is not higher vs. lower gear, but smaller vs. larger chain sprocket. (Recall that the effect of shifting to the larger chain sprocket in the front is the same as shifting to the smaller one in the back and vice-versa.) He may be right, but this seems about as un-user-friendly as you can get in terms of design.

Another possibility I'm entertaining (at least until someone else suggests something else, or better yet settles the matter) is that someone at least as convincing as Geoff Pullum was somehow involved in the early design phase of the original grip shifters. This hypothetical person persuaded the engineers that the way to look at the shifters was from each side of the bike, and that the distinction between "clockwise" and "counterclockwise" should be perfectly understandable to anyone. The engineers, not wanting to seem like idiots, agreed.

[ Comments? ]

More bullshit grammar

In addition to referring to a bunch of nouns as adjectives, Timothy Noah's review of Harry Frankfurt's On Bullshit makes a second error in grammatical analysis.

Noah writes:

Although Frankfurt doesn't point this out, it immediately occurred to me upon closing his book that the word "bullshit" is both noun and verb, and that this duality distinguishes bullshit not only from the aforementioned Menckenesque antecedents, but also from its contemporary near-relative, horseshit. It is possible to bullshit somebody, but it is not possible to poppycock, or to twaddle, or to horseshit anyone. When we speak of bullshit, then, we speak, implicitly, of the action that brought the bullshit into being: Somebody bullshitted. In this respect the word "bullshit" is identical to the word "lie," for when we speak of a lie we speak, implicitly, of the action that brought the lie into being: Somebody lied.

But actually, of the 14 "Menckenesque antecedents" that Noah cites (humbug, poppycock, tommyrot, hooey, twaddle, balderdash, claptrap, palaver, hogwash, buncombe (or "bunk"), hokum, drivel, flapdoodle, bullpucky), four are given a verbal sense by the American Heritage Dictionary: humbug, twaddle, palaver, and drivel.

A quick check in the OED adds some others from the list that are cited as verbs:

1821 W. IRVING in Warner Life (1882) 136 A fostered growth of poetry and romance, and balderdashed with false sentiment.
1893 Westm. Gaz. 11 July 2/1 He flapdoodled round the subject in the usual Archiepiscopal way.

The OED also reveals that some of these were verbs before they were nouns (at least in the bullshit-related sense) -- thus drivel as a verb meaning "To talk childishly or idiotically; to let silly nonsense drop from the lips; to rave" from 1362, vs. drivel as a noun meaning "Idiotic utterance; silly nonsense; twaddle" from 1852.

Several of the others on Noah's list are sometimes painlessly verbed, as a simple net search verifies:

Balderdashing into doom [6/22/2004 WaPo headline]
Who's this hokey honky tryin to hooey? (link)
When Gianera let the casual acquaintance into his home, Mullin cried “You’re claptrapping me!” and shot Jim as he tried to escape. (link)
And I am not horseshitting you, I really did do it. (link)
It appears Marilyn Stowe was hogwashing us. (link)

It's certainly true that bullshit is used more commonly as a verb than these other words are -- but it's also used more commonly, period (e.g. 3,040,000 ghits for bullshit vs. 171,000 for twaddle or 84,800 for horseshit).

Unlike the "nouns are adjectives" mistake, this one seems to arise from carelessness about what is being claimed and lack of concern for what the relevant facts really are. So ironically, the insight about the word bullshit that Noah starts his column with -- is itself bullshit. I mean this in the specific technical sense of the term that derives from Frankfurt: "bullshitters seek to convey a certain impression of themselves without being concerned about whether anything at all is true".

Noah's review seems to be calculated to place the rhetorical ball so that he can spike it into the teeth of the Bush administration. He ends this way:

The Bush administration is clearly more bullshit-heavy than its predecessors. Slate's founding editor, Michael Kinsley, put his finger on the Bush administration's particular style of lying three years ago:

If the truth was too precious to waste on politics for Bush I and a challenge to overcome for Clinton, for our current George Bush it is simply boring and uncool. Bush II administration lies are often so laughably obvious that you wonder why they bother. Until you realize: They haven't bothered.

But by Frankfurt's lights, what Bush does isn't lying at all. It's bullshitting. Whatever you choose to call it, Bush's indifference to the truth is indeed more troubling, in many ways, than what Frankfurt calls "lying" would be. Richard Nixon knew he was bombing Cambodia. Does George W. Bush have a clue that his Social Security arithmetic fails to add up? How can he know if he doesn't care?

Social Security surely matters more than grammatical terminology does. But Noah's disdain for GWB's carelessness about the analysis of budgets would be more convincing if Noah himself were not so careless about the analysis of words.

The grammar of bullshit

On March 2, Timothy Noah in Slate reviewed Harry Frankfurt's 1986 monograph on bullshit. Here's Noah's fourth sentence:

How does bullshit differ from such precursors as humbug, poppycock, tommyrot, hooey, twaddle, balderdash, claptrap, palaver, hogwash, buncombe (or "bunk"), hokum, drivel, flapdoodle, bullpucky, and all the other pejoratives* favored by H.L. Mencken and his many imitators? [emphasis added]

The asterisk on "pejoratives" leads to this footnote:

Correction, March 4, 2005: An earlier version of this article mistakenly described these words as adjectives. In fact, they are nouns.

Kudos to Slate (and Noah) for footnoting the correction rather than just fixing it silently. The mistake is not critical to Frankfurt's ideas or Noah's review of them. (Well, maybe it suggests a certain lack of concern for what words actually mean, which is not entirely unconnected to what Frankfurt thinks bullshit is.) However, it does underline a point that we've made again and again in this blog. Most Americans learn almost nothing about how to describe and analyze the sound, structure and meaning of the English language. This includes most American intellectuals, whose degree of ignorance in this area is historically unprecedented. It extends to many of those who are being trained, at the best universities, in the discipline known as "English", and even more strongly to those trained in other fields.

As I've explained before, I blame the linguists.

[Note, by the way, that Noah's confusion about humbug etc. can probably be explained by the same reasoning that Geoff Pullum used in the case of Jon Stewart's confusion about terror. ]

OED Science Fiction update

A note from Jesse Sheidlower about improvements in the (already very cool) OED Science Fiction site:

For those interested: the Oxford English Dictionary Science Fiction project at http://www.jessesword.com/sf has been redesigned and relaunched.

The biggest change is that the OED's database of citations of SF words is now made (mostly) available via the website. The OED does not usually make its work available in this way, but OED has agreed to publicly open up this part of its database to acknowledge the great contribution volunteers have made to this project.

That means that if you contribute a cite, it's viewable by everyone. Here's a link with more information about the citations

http://www.jessesword.com/sf/about_citations

We are also adding quite a few new words: there is an internal list of pending words we have been maintaining and over the next few weeks many of those words will be moved to the main pages. This link:

http://www.jessesword.com/sf/newest_adds

takes you to a list of the most recent additions.

We hope these changes make the site into more of a general resource for the vocabulary of SF, instead of just a catalogue of OED research needs.

Jive etymologies

In reference to the October post "Trevor's Law of Hip Etymology," I must agree with the verdict that the black English terms HIP, DIG, and JIVE are not borrowings from Wolof. However, the proof that most compels me is from a wider view of where slaves in America came from and how Black English arose.

Clarence Major proposed that, for example, HIP comes from Wolof's HEPI "to see" in the early seventies, when the study of just where slaves in a given colony were drawn from in Africa was in its infancy. At the time, the general sense was that slaves had come from a wide variety of locations, and that for that reason, pretty much any African language was fair game for an etymology. In these days, it was typical to see even obscure languages like Vai and Susu and Kpelle treated as sources for this word or that feature.

But nowadays, creolists and specialists on Black English know more about how these varieties arose and which slaves were most important in creating them. And that light, 35 years later Major's etymologies, piquant though they were, must be gently consigned to the realm of history.

For one thing, although the Wolof are relatively prominent to many Americans because of the large number of Senegalese immigrants in this country, and to black Americans because the Goree Island slaving settlement is a popular tourist attraction, the fact is that there is no evidence that Wolof speakers were predominant among slaves in the United States, numerically or culturally. For example, Lorenzo Dow Turner's seminal work on the provenience of African names among Sea Islanders (rural blacks who speak Gullah Creole) points to countless languages up and down the African coast: Wolof is just one of the many.

Of course we might say that lots of other groups were making their contributions, and that HIP, DIG and JIVE were just the two cents that the small number of Wolof speakers tossed in. But this means that we should expect that dozens of other Black English words had been traced to, say, Bambara, Mende, Twi, Yoruba, Efik, Umbundu, and so on. But they haven't. Most Black English slang clearly traces to words that started out in Merrie England. No one proposes that a word like PHAT -- roughly equivalent to the once-celebrated use of BAD to mean GOOD -- just sounds like FAT but actually traces to Igbo. There is no handful of cool black slang words that we are told trace back to Kikongo. Rather, the grand old HIP/DIG/JIVE trio stands out by itself, endlessly quoted over the decades.

In fact, if there any one African language that we could even begin to treat as "black Americans' native tongue," it is Mende of Sierra Leone. This is the language that some old Sea Islanders still sing folk songs in, although no longer knowing their meaning. If Wolof had ever had enough juice that its speakers would contribute vibrant words that would last centuries, then we would expect Wolof folk songs.

Finally, the Wolof etymology doesn't even help if we look at earlier stages of African slaves' varieties of English that fed into what happened here in America. A good portion of the slaves who helped to found the Charleston colony, for instance, were brought in after having served in Barbados, rather than directly from Africa. They spoke the Barbadian variety of Caribbean Creole English, such that Gullah is one more variation on that pattern. But Caribbean Creole English, again, exhibits no especial Wolof contribution. Its grammar, for example, reflects languages of Ghana, Togo, Benin and Nigeria -- but there is not a hint of Wolof's very different structure.

The HIPI myth has even made it across the ocean. Years ago I gave some lectures in Dakar where a local English teacher was quite taken with the idea that the HIP and JIVE he had learned about in black Americans' English were Wolof words. I didn't see any point in arguing with him, but for Language Log I thought it might be better to give the facts -- you dig?

Box spaghetti straight

Harry Shearer's radio program "Le Show" has a segment nearly every week in which Harry does all the voices for a sketch involving the 60 Minutes team and various other TV newspeople. Last Sunday morning it was Dan Rather chatting in folksy Texas phrases as he packs his boxes to move out of the NBC building, and at one point he describes another TV newsman approvingly as "box spaghetti straight". (The phrase appears to get no Google hits, by the way.) The phrase provides not just one but two beautiful arguments against the policy of almost all traditional English grammar as regards defining notions like "adjective" and "adverb".

Traditional grammars always tell you that adjectives are defined as words that modify nouns, and adverbs can be defined as words that modify other parts of speech -- they modify verbs, adjectives, other adverbs, and prepositions. As Shearer's fictionalized Dan Rather would say, that dog won't hunt. I have no idea why it has proved so robust, but it sure isn't correct.

Under the traditional view, since things that modify nouns are ipso facto adjectives, it follows that in box spaghetti (meaning "spaghetti that comes in a box"), the word box has to be an adjective. But if it were an adjective, although you would expect a comparative form (*boxer than the other one) and a superlative form (*the boxest I ever saw), instead you get a plural form boxes and a genitive form box's. Clear signs of nounhood. If box isn't a solid, true-blue noun, there aren't any. Calling it an adjective is completely nuts.

And considering the whole phrase box spaghetti straight, what the fictional Rather meant by it is "straight in the way that box spaghetti is straight". It's parallel to rock solid, which means solid in the way that rocks are solid: rock modifies the adjective solid. But if the things that modify adjectives are ipso facto adverbs, then rock has to be an adverb in rock solid — and in box spaghetti straight, the phrase box spaghetti has to be an adverb.

All these consequences come right from the planet Zorbo. The right way to describe things is the way it is done in The Cambridge Grammar of the English Language. Instead of confusing the idea of being an adjective with the idea of being a modifier of a noun, and confusing the idea of being an adverb with the idea of being a modifier of a non-noun, we separate these notions.

The constituent that comes before a head in a phrase to qualify its meaning has the function modifier. Although the modifier in a noun phrase will often be an adjective, it doesn't have to be. In a phrase like London fog, we have a proper noun serving as a noun modifier. In a phrase like box spaghetti, we have a common noun serving as a noun modifier.

And although the modifier in an adjective phrase will often be an adjective, it doesn't have to be. It can be a noun, as in rock solid, or a phrase formed by a noun and its modifier (The Cambridge Grammar calls this a nominal), as in box spaghetti straight.

Box is a noun, spaghetti is a noun, straight is an adjective. Together they form an adjective phrase. This is not what your traditional grammar books will tell you. But that's because those books get it wrong. You can trust me; on matters like this, I'm box spaghetti straight.

The progress and prospects of the digital BNF

A terrific post by Andrew Joscelyne at Blogos discusses a Le Monde interview with Jean-Noël Jeanneney, the head of the French National Library ( Bibliothèque Nationale de France, or BNF) who wrote a few weeks ago about Google's challenge to Europe. The most interesting part, to me anyhow, is Andrew's memory of an earlier BNF disaster. More precisely, this was an earlier case in which an obsession with industrial competitiveness, combined with top-down decision making by technically clueless bureaucrats, wasted many tens of millions of new francs.

Andrew wrote:

I remember attending a demo in the early pre-web 1990s given by Cap Gemini (or whatever the IT services company was called at the time) which had been charged with designing a ‘scholar’s workstation’ for the brand new BNF, looming with its four monstrous bookend towers and damp wooden platform over the Seine opposite the old wine depot of Bercy. The idea was to offer serious readers digitized and bitmapped versions of books from every age, allowing both access to the text as a corpus, and as a set of specially designed original pages. All wonderful stuff, yet predicated on a massive digitization campaign. However, as Jeanneney admits today, the BNF has managed to digitize only 80 thousand works in a decade, compared with Google’s project of 15 million in about half that time. Why so few?

Alas, the BNF has created nowhere near 80,000 e-texts, as we'll see shortly. But let's focus first on this BNF workstation. I never saw a demo, but I did attend more than one presentation by its funders and developers. "Wonderful stuff" is a phrase that never occurred to me at the time, and it seems even less appropriate in retrospect.

If my memory is correct, the following things were true:

1. The workstation (called PLAO, "poste de lecture assistee par ordinateur" or "computer-assisted reading environment") was based on new, proprietary hardware and software. Why? To promote the French IT industry.
2. According to the original plan, the BNF digital library would be accessible only via special PLAO workstations at the BNF site in Paris. There would be no provision for any sort of remote access, not even dial-up, much less via the (American!) internet. I recall asking questions about this at one of these presentations, around 1994 or 1995: the speaker looked pained, as if I'd suggested putting ketchup on my croque-monsieur.

The PLAO thus set its face directly against the two biggest technology trends of the decade, namely commodity computing and the internet.

I emphasize that this is my memory from a couple of presentations that I heard a decade ago or more, and may be inaccurate or incomplete -- I welcome additions and corrections from more knowledgeable readers. I presume that PLAO is in the dustbins of technological history; in any case, BNF's digital library is now accessible on the web via the Gallica site.

I've used Gallica with gratitude in the past, and expect to use it again in the future, but based on my interactions with the site, I'm confident that many fewer than 80,000 works are really available -- in text form, anyhow. The number of works returned by my searches seems to suggest a significantly smaller number, and indeed Gallica has a list of documents in text mode, which includes merely 1,118 works.

A quick check on the Interrogation du catalogue page suggests that this list might be a bit out of date, but not by much. There are six types of document: "Ouvrages en mode texte", "Monographies en mode image", "Périodiques en mode image", "Lots d'images", "Documents sonores", and "Documents manuscrits" ("works in text mode", "monographs in image mode", "periodicals in image mode", "sets of images", "audio documents", "manuscript documents"). If you search the "Ouvrages en mode texte", you'll get hit counts like 935 for France, 1,135 for terre, 1,131 for chose, 1,043 for parmi, and so on. (It won't work to search for common function words like "mais" or "non" -- these are apprently stop words, and -- silently -- return nothing.)

So apparently the BNF's efforts over the past decade have given Gallica e-text holdings of about 1,200 works. The all-volunteer Project Gutenberg has produced more than 13,000 e-texts over roughly the same period of time.

A couple of other comparisons: if I ask Google for pages in the French language that contain the word France, I get 24,000,000 hits. If I ask amazon.com for books on the subject of France, I get 26,100 hits. If I ask the Literature Online ("LION") database for texts containing the word France, I get 26,880 instances in 7,959 distinct works. Asking LION for earth gets 201,326 instances in 69,425 distinct works -- out of the "more than 350,000 works of poetry, drama and prose" that LION offers.

If Brussels gives M. Jeanneney the "plan pluriannuel" with a "budget généreux" that he's asking for, let's say that it'll be a triumph of hope over experience.

[Let me be clear that I'm all for pluralistic efforts, and especially European efforts, in the digital libraries arena. And I also believe that government-supported efforts could have a crucial role to play, especially if the result were to be e-texts in the public domain or otherwise openly available (and not just through one entity's web site). But I also believe that the best way to predict what (well-established) institutions will do in the future is usually to look at what they've done in the past.]

WTF grammar

We need a new term. Prescriptive grammar says "thou shalt not say (things that meet conditions) XYZ". Descriptive grammar says "love the vernacular, and say what you like". But what do we call it when you're taken grammatically aback by something you hear or read, and then try to figure out what the problem was?

This process is somewhat prescriptive, in that it starts with a perceived violation of internalized norms (what Geoff Pullum and Barbara Scholz call correctness conditions). But it's also rather descriptive, in that you try to understand the problem by means of a systematic investigation of relevant patterns of usage. So how about reactive grammar? Or more informally, WTF grammar?

Case in point: Eric Bakovic's reaction to a sentence in an old Mac OS X manual.

If you have an older Mac and upgraded the processor, don't expect it to work or support from Apple.

Eric's reaction: "It's just bad." He has a story to tell about why it's bad, but his badness reaction comes first, and his explanation comes second.

And sometimes the explanation phase turns out to be remarkably difficult.

Back last summer, Eric had another one of those WTF moments when he read this sentence about resistentialism:

Here, at last, was a word for the rug that quietly curls up so it can snag your toe, the sock gone AWOL from the dryer, the slippery piece of toast that always hits the floor jelly side down.

This hit Eric with an image of the writer experimentally dropping the same bit of breakfast over and over again, though it's clear that we're supposed to generalize over encounters with many different instances of jelly-clad toast. Eric analyzed the problem as "the use of the word always, universally quantifying over the predicate of the relative clause (hits the floor jelly side down) that in turn modifies a singular definite noun phrase (the slippery piece of toast)".

David Beaver suggested a different analysis: "The problem is simply that relative clauses, as has often been observed, are what we term scope islands".

I in turn objected to David's story, pointing out examples like

(link) Their Barbera is a fun and fruity wine that always pleases us...
(link) My mother used to make a fantastic beef stew that always tasted better the next day.

David responded with an account in terms of Carlson's distinction between different types of generic sentences, "only some of which involve reference to kinds", suggesting that "'piece of' resists being kind denoting". (If this isn't completely clear to you, that's the point -- it's high semantic wonkery, and if David's account is correct, you'll need quite a bit of background to understand what's really going on here.)

This morning, Gabriel Nivasch wrote to me with a different proposed explanation:

Regarding your post on the toast falling with the jelly side down:

There is a difference between "food", "wine", and "seafood" on one hand, and a "toast" on the other. The former are non-countable nouns, and the latter is a countable noun. Therefore, you can say

"food that always satisfies"

because each time it is a different piece of food that is being eaten, yet they're all placed together under the word "food". The same applies to "wine" and "seafood".

On the other hand, if you say

"a piece of toast that always falls"

it seems to say that the *same* piece of toast is falling one time after the other, so it sounds a little weird.

One problem up front is that toast isn't all that countable -- we usually say "Do you want some toast?" or "Do you want a piece of toast?", not "*Do you want a toast?"

Anyhow, I don't think that Eric's original WTF reaction was caused by "the piece of toast" being a count-noun kind of expression. Consider this quote from the brochure for the 1968 Renault 1100:

This is the car that's always been DIFFERENT and for 1968 it's differenter!

This quantifies over generic experiences with cars in different Renault model-years, not over experiences with any specific vehicle -- but car is certainly a count noun.

The best story still seems to be David Beaver's observation that piece of "resists being kind denoting". This is not entirely unconnected with the mass/count distinction, since piece of was motivated in the original sentence in order to countify toast, so to speak. However, I'm still not sure that this story is the right one.

So eight months and five analysts later, we've still got Eric's bizarreness reaction -- which most but not all people seem to agree with -- without any clear prescription about how to write so as to avoid it. This situation is not at all typical, because most grammatical WTFs have a simple explanation, easily accessible to someone with linguistic training. Nevertheless, this case emphasizes the fact that explicit grammatical principles are post hoc explanations of the phenomenology of linguistic experience. Norma Loquendi rules.

Don't expect this post to make sense or any help from me

It's amazing the things you find when you're not really looking. While working on some paper revisions today, I had this extraordinary urge to clean up the pile of books and papers that had gathered around my feet over the past week (of working on paper revisions, of course). Somehow, a three-year-old note to myself found its way into this pile. It's a sentence that I wrote down, from an old book on Mac OS X. On p. 206, Gene Steinberg wrote:

If you have an older Mac and upgraded the processor, don't expect it to work or support from Apple.

I can see what this sentence means, and I can understand how Steinberg may have reasoned that it was better than the alternative:

... don't expect it to work, and don't expect support from Apple.

Problem is, the sentence Steinberg wrote does not satisfy (what I think are) my syntactic correctness conditions. It's just bad. The problem, as I see it, is that Steinberg has attempted to conjoin an infinitival sentence (it to work) with a noun phrase (support from Apple).

Now I'm sure that CGEL has some more accurate analysis of the relevant facts (perhaps even a better term than infinitival sentence?) but I'm too tired and lazy to walk all the way across Language Log Plaza to Geoff's office to see what it is. Anyway, I know there are some patterned exceptions to the generalization I'm about to make, but typically, only syntactic phrases of the same type can be conjoined. So it should be perfectly fine to conjoin two noun phrases as complements of expect, and indeed it is:

Don't expect my everlasting love or a bouquet of roses.

It should also be perfectly fine to conjoin two infinitival sentences as complements of expect, and it is -- but for me, if the subject of the second infinitival sentence is a pronoun, a for is (mildly) required before that pronoun. On the other hand, for is absolutely forbidden before the first infinitival sentence (whether or not its subject is a pronoun), and it's completely optional otherwise:

Don't expect (*for) the car to drive itself or (for) the sky not to fall on your head.

Don't expect (*for) it to work or *?(for) them to help you.

Well, that kept me occupied for a full half hour or so. Then it was back to work ... cleaning up the mess on the floor. Then the revisions. Now bed.

Update, Mar. 9: Neal Whitman writes:

Your latest post on Language Log was of special interest to me, since I have a collection going of utterances such as, "It makes it hard for him to get [his stuff done] and [to bed on time]," and "She wants [an engagement ring] and [her boyfriend to stop dragging his feet]," and "Don't eat [fast food], or [at restaurants, food-service companies, or caterers]." In case you're interested, I discuss coordinations like these in a paper that appeared in the Sept. 2004 issue of Language. It covers adjunct-with-adjunct, complement-with-complement, and adjunct-with-complement coordinations.

The "She wants ..." example is OK for me, but only with "for" before the second conjunct; the "Don't eat ..." example is perfect. I look forward to reading Neal's paper.

[ Comments? ]

Raising standards -- by lowering them

Chris at Mixing Memory reports that blogging about (his end of) science is discouraging:

As an academic, I have spent a lot of time hiding away in the ivory tower, oblivious to the larger world around me. As a graduate student, especially, I had almost no time to pay any attention to what non-scientists were saying about cognitive science. However, on a fateful day in early 2004, I chose to crawl out of my hole and actually look at what other people were saying. I started reading blogs. And now I want to crawl back in!

I'm happy to say that he's decided to resist the urge, and to keep fighting the good fight. His stuff is smart, well informed, and well written -- take a look at his posts on Lakoff and framing, for example, or on recovered memories, or his most recent post on corpus-based approximations to meaning. However, I'm going to disagree with his advice to people writing about science.

Actually, his advice is specifically directed at "anyone who wants to talk about cognitive science, but has not spent a lot of time studying it", but the things that bother him are characteristic of science writing in general, and so his prescriptions apply more generally as well. You should read his Five Points for yourself, but here's my summary:

1. Do a lot of background reading before writing anything.
2. Ignore what's in the popular press (aka MSM).
3. Ignore popular books.
4. Read the peer-reviewed literature.
5. Consult experts.

Chris says that "if everyone followed these guidelines when they wrote about fields in which they are not experts, maybe the public wouldn't have such a god-awful understanding of the sciences."

Though I sympathize deeply with Chris' frustration, I disagree completely with his prescriptions, for two main reasons.

First, if everyone followed his guidelines, there'd be an order of magnitude less science writing than there is, and there's already too little. Instead of putting up higher barriers to entry, we should be encouraging more people to do more thinking and writing about mathematics, science and technology (and history and literature and art, too, but that's another story). A rising tide of interest and involvement will lift all intellectual boats, even if some unsavory stuff floats up off the mud flats.

Second, the peer-reviewed literature may be the best thing we've got, but it's not very good. There's an enormous quantity of irrelevant junk in it, and a certain amount of out-and-out crap. Much of it is unreadable or misleading. Worse, a lot is missing -- questions that don't get asked, negative results that don't get published, whole problem areas that don't get addressed for decades at a time. And you don't have to agree with Steve McIntyre's views on global warming to sympathize with his complaints about "disclosure and due diligence" in the refereed literature.

So here's my alternative prescription for improving scientific communication:

1. Encourage everyone to think about science, and to write about it on the web, whether they know anything about it or not. And encourage them to criticize what others write, and to read others' criticisms, and to tell their friends about the best stuff that they find, whether in the popular media, or in the technical literature, or in weblogs. I claim that open intellectual communities intrinsically tend to generate a virtuous cycle: if there were an order of magnitude more science writing in blogs, there'd be less than an order of magnitude more crap, and more than an order of magnitude more good stuff. (The same is probably true for science writing in newspapers, though the network effects are smaller there.) This follows from a scientific version of Moglen's Metaphorical Corollary to Faraday's Law: add more wires, lower the resistance, and more intellectual current is induced.

2. Improve the (professional) scientific literature. Here's a three-point plan:

1. Open access on the web for all scientific publications, with durable doi-style references.
2. Open access on the web for all data and programs involved in scientific publications.
3. Standard APIs for references in all scientific publications, and methods for inducing trackbacks across all achives of such publications.

Point 1: Open access lowers encourages people to read (and evaluate!) primary sources, not just someone's summary. More people reading more papers is good.

Point 2: All the data and programs behind published claims should be published in electronic form, so that readers can check methods and results, try alternative models, and (most important) build on others' work. This shortens the half-life of mistakes, and accelerates the spread of good ideas.

Point 3: Now that nearly all journals, proceedings etc. are on the web, there's no excuse not to make it trivial to extract the citation graph (i.e. who cites whom for what). Then users can wander around in the graph, use it to calculate value via the analog of page-rank, and do all sorts of other neat things. The way things are currently done, finding the citation graph is a non-trivial exercise in text analysis and reference normalization, even for the documents that are not hidden behind a publisher's barrier. This is one place where "semantic web" ideas really ought to be imposed.

There will of course be cases where this much openness is not possible -- e.g. where data can't be published for privacy or intellectual property reasons. But such cases should be treated like the use of anonymous sources in journalism -- permitted only where the results are valuable enough, and there's no alternative.

All this stuff is happening anyhow. Let's do it faster.

W.O.D.

Andrew Joscelyne at Blogos explains what's going on with "Webster's Online Dictionary (The Rosetta Edition)", a vast project initiated and managed by Philip Parker. The goal: "the biggest multilingual dictionary site on the web", constituting an “N-dimensional cube of words in every language to every language,” currently said to weigh in a terabyte or so. Margaret Marks at Transblawg complains about the treatment of translator, and calls the WOD ""huge and bizarre".

Take a look at the entry for language. There's some useful stuff in there, but I think I'd think twice before believing what this source told me about words in a language I didn't know. You might get something like the "synonyms within context", which seems to include every Roget's category in which any term contains the word language, and therefore covers many things that are not synonyms at all, within context or without it -- personality, vigor, raciness, authorship, gravity and so on.

[Update: Abnu from Wordlab emailed to point out that he "was not impressed with the naming of Webster's Online Dictionary. Shouldn't we expect something original from a Chaired Professor of Innovation?". Well, perhaps sometimes it's innovative to be derivative.

butt to buck, start to stark, or vice versa?

Back in August, I looked into the butt naked vs. buck naked issue, and concluded that it's a draw. Scholars disagree about which was the original expression, and current usage is roughly split as well. I referenced that discussion in my post today on Hobbesian choice, and Prof. Paul Brians wrote to disagree.

On your very interesting Web site you challenge my claim that "buck naked" is older than "butt naked." I'd be interested to see evidence of any use earlier than my first encounter with the latter, less than ten years ago. In 37 years of reading student writing and popular journalism, it's only recently that I've encountered the "butt" variation.

The Cassell Dictionary of Slang lists "buck naked" as early 19th Century and speculates, as did one of your sources, on "buck" as a variation on "butt"; but until someone comes up with an actual early citation, I'll stand by my etymology as more likely. Lightly clad blacks and Indians were commonly called "bucks" in the 19th century.

Though it has a different origin, I associate this also with the common "nip it in the butt."

I certainly don't have any 19th-century citations for "butt naked", and I agree that Prof. Brians might well be right that "buck naked" is the original phrase. I'm not challenging his claim, just observing that others disagree, and determinative evidence seems to be lacking. However, I'm fairly confident that "butt naked" is more than ten years old. I recall hearing it as a child, as does someone posting on the phrases.org.uk bulletin board:

Half century ago in the rural southeast US the expression was "butt naked" (only we said something closer to "butt necked") and it just ment naked.

And several fairly authoritative dictionaries (in addition to the Cassell's reference that Prof. Brians mentions) suggest that "buck" was originally a euphemism or other alteration of "butt". The Dictionary of American Regional English (DARE) has

BUCK NAKED - adjective. Also buck-ass naked, buck-born ~, stark buck ~. Origin uncertain, but perhaps alteration of butt/buttocks. Entirely unclothed.

and the American Heritage Dictionary (perhaps echoing DARE) has

Etymology: buck- (perhaps alteration of butt) + naked.

So as far as I can tell, the jury is still out.

And though I don't have a citation, I do have a curiously exact precedent to point to. Starting around 1530, another expression of the form "X naked", where X was a word for "buttocks" ending in a /t/, was changed into a new expression "Y naked", where Y was X with the final /t/ changed to /k/. "Y naked" made just as much sense as "X naked", but it was politer. And this time the OED gives us the citations.

The OED glosses start naked as

Entirely naked; = STARK-NAKED a.

and explains the etymology as

[App. f. START n.1 + NAKED a.
The literal sense would seem to be ‘naked even to the tail’. Start has not been found in Eng. with the sense ‘buttocks’ (= TAIL n.1 5), but the MDu. and Ger. equivalents are so used.]

Citations start in 1225 -- and continue to the 19th-century American south:

a1225 Juliana 16 (Roy. MS.), & he het hatterliche strupen hire steortnaket [Bodl. MS. steort naket].
a1225 Ancr. R. 148 Heo haueð bipiled mine figer..despoiled hire stert [printed sterc] naked, & iworpen awei [etc.].
Ibid. 316 Bicleope þine sunne steornaked; þet is, ne hele þu nowiht of al þet liþ þer abuten.
13.. Pol. Songs (Camden) 336 Sholde he for everi fals uth lese kirtel or kote,..He sholde stonde start [printed starc] naked twye o day or eve.
c1320 Cast. Love 431 in Minor Poems fr. Vernon MS. xxxviii, And I-strupt him al start-naked.
a1325 in Horstm. Altengl. Leg. (1878) 140 Þai lay þerin all star naked.
1892 Dialect Notes (Amer. Dial. Soc. 1896) I. v. 234 Start-naked: stark naked. ‘He is a start-naked villain.’.. Mr. A. W. Long, of North Carolina, reports that he never heard any other form than start-naked used in conversation in that state; and that two of his friends --- one from Virginia, and the other from South Carolina -- make the same statement for those two states.

The OED entry for stark naked states unequivocally that it is "altered from the earlier START-NAKED", with the earliest citation from 1530:

1530 PALSGR. 842/1 Starke bely naked, tout fin mere nud... Starke naked, tout fin nud.
1560 J. DAUS tr. Sleidane's Comm. 356 They left them starcke naked.

So "butt naked" would be a straightforward calque of a common expression whose word for "butt" had dropped out of the language. Of course, this would be much more persuasive if someone could find a citation...

By the way, the "bely" in the first citation is belly, and the expression "stark belly naked" also has a citation in the entry for belly (and several others in the LION database):

1611 COTGR., Tout fin mere nu, all discouered..starke bellie naked.

The OED entry for stark naked references sense 2 of stark as adverb, glossed as "To the fullest extent or degree; absolutely, utterly, quite". A phrase meaning "absolutely naked" is at least as reasonable as one meaning "naked to the tail" -- though for the first hundred years or so, people often stuck in another piece of anatomy, just for good luck:

165 And then forth shalt thou, sterk belly naked,
166 With dogs arrand quen, thou shalt be bayted.
(Walter Smith, 1525, "The fyfth mery Iest, or etc.")

601 Glad was this man, and with his gladnesse waked,
602 But scarcely had he opened both his eyes,
603 Before he felt his wife starke belly naked:
604 And found his finger hid betweene her thighes.
(Robert Tofte, 1611, "THE FOVRTH SATYRE OF Ariosto")

[A curious side note: according to an article in Slate by Mark Scheffler, in Liberia around 1989, Joshua Milton Blahyi went under the name of "General Butt Naked", commanding the "Butt Naked Battalion". That takes butt naked back 15 years, but more important, it suggests that "butt naked" is the normal expression in vernacular forms of Liberian English, where it may have traveled an independent path for the past 150 years or so.]

An eggcorn hatches

OK, it's official. Hobbesian choice is now the blue-state equivalent of the French choix cornélien: a forced choice between two unattractive alternatives. Well, it's not exactly official -- there's no Blue State Academy to decide these things -- but after reading James Wolcott's 3/2/2005 attack on Kurt Anderson, it looks to me like we've passed the sociolinguistic tipping point on this one.

You can read all about the Great Hobbesian Choice Brouhaha in this 2/19/2005 post, but here's a quick summary. Back in April of 2003, Peter Wood at the National Review trashed John Payton, who argued for the University of Michigan in Gratz v. Bollinger, for using the phrase "Hobbesian choice". Wood took this to be a malapropism for the long-established idiom "Hobson's choice", indicating that "diversity's defenders came across as stridently self-righteous and pretty sloppy about the details". Then Kurt Anderson wrote in the 2/21/2005 New York Magazine about our "Hobbesian choice" in Iraq -- "either we hope for the vindication of Bush’s risky, very possibly reckless policy, or we are in a de facto alliance with the killers of American soldiers and Iraqi civilians", according to him. Some bloggers rapped his knuckles for the apparent mistake, and I agreed: "Even if there is a valid and coherent reason for Anderson to see his choice as a 'Hobbesian choice', he can't use that phrase without taking literate readers aback, and leading some of them to make fun of him. Unless, of course, he can convince them that the whole thing was a clever pun all along." Then Anderson wrote to Jim Hanas that indeed, he meant the phrase as a piece of word-play.

Now James Wolcott has posted a stinging attack on Kurt Anderson under the heading "Take this binary mode and shove it": "bright and intelligent in a completely uninteresting way", "glib in a Manhattan Mandarin manner that conceals the glibness behind a knowingness that itself conceals a lack of deeper, driving conviction", "one of those media personalities who's always 'positioning' himself without ever taking a real position".

The character evaluation aside, Wolcott quotes approvingly from Matt Taibbi:

"'Each of us has a Hobbesian choice concerning Iraq' [Andersen writes]. This is horseshit on its face. Even the original Hobbesian choice was horseshit, especially in the eyes of the stereotypical New York liberal Andersen is addressing. We no more have to choose between chaos and authoritarianism than we do between rooting for Bush and rooting for the insurgents. There is a vast array of other outcomes and developments to root for."

Taibbi assumes without comment that "the original Hobbesian choice" was between chaos and authoritarianism, and Wolcott accepts that assumption without comment. I think it's going to be impossible to put this horse back in Hobson's stable, even if we wanted to -- there's a second idiom now.

There are plenty of precedents for the stable [sorry] co-existence of eggcornic idiom pairs: "butt naked" vs. "buck naked" (where scholarship seems unable to establish which was the original), and "home in" vs. "hone in", "deep seated" vs. "deep seeded", and many more. What's interesting about this case is that there's also a subtle but clear difference in meaning: a Hobson's choice between something bad and nothing, and a Hobbesian choice between equally unwanted opposites.

*

Google usage

I'm fond of McSweeney's. For example, the article currently featured on the web site is Chris Gavaler's "Who's on First?", which is pretty funny in a quiet sort of way. And many of the things on the new lists page are also funny. But when I read Adam Koford's list of "Adjectives rarely used by wine tasters", I was skeptical. After a series of posts last summer on the language of wine tasting and similar enthusiasms, I retain a sort of general sense of what kind of descriptions wine aficionados are likely to use, and I was pretty sure that Adam had not pushed the envelope nearly far enough.

So I checked with Google. Adam's first "rarely used" adjective is chunky -- but {chunky wine} has 208,000 web hits on Google, most of which seem to just the sort of thing that Adam thinks shouldn't come up. Nine out of the first ten, for example:

From the hot 2003 vintage in Europe, this is a chunky wine that is perfect for your mid-week meals with our Own Pasta and Meat Sauce with Roasted Eggplant and Peppers as a side dish.
A well structured, chunky wine with hints of figs and raisins.
A big, chunky wine that is typical of the Okahu style.
A solid, chunky wine with a Vogel's toast nose, lashings of creamy, rich fruit through the middle and a savoury finish.
Chewy, big chunky wine with a lot of tannin.
Chunky, tarry, spicy wine. ... Firm tannins, however not slaughtered by oak and despite its chunky and rustic nature, it was a tasty wine with this dish.
As for a "Chunky wine" [...] That would mean young,huge tannins, toasty oak and a big finish...almost "Chewy".
The Amarone, a proverbially big and chunky wine, had died in the bottle.
This is a nice savoury, chunky wine that's drinking beautifully now.
Made with extra thick glass, this chunky wine goblet is a durable and versatile choice for casual entertaining.

Adam's second "rarely used" adjective is super-charged -- but {super-charged wine} has 37,800 whG. Again, nine of the first ten hits are just what Adam doesn't expect:

This was the third time I had encountered Chateau Angelus, a massive, super-charged wine that just seems more monolithic each time I taste it.
Leonetti winemaker Gary Figgins has it down, blending his best selections from the Seven Hills, Windrow and Spring Valley vineyards into this super-charged wine.
Huge and thick with fruit on the palate, this super-charged wine is impeccably well balanced, with fine focus, ripe tannins, and a long, complex, tidal wave of a finish.
Wow! This is like a supercharged Rhone.
From a newish estate in South Africa, this has a striking, paint-box colour of vivid purple and intense aromas of super-charged blackcurrant and mint, brambles and an earthy depth.
A brilliant, supercharged white wine.
It is richly oaked, with minty fruit and a slightly syrupy texture. Extraordinary stuff. Excellent, provided that you like this supercharged style.
A super charged Sauvignon Blanc.
I’ve heard of using strong black tea as a tannin additive and of using coffee in brewing recipes for a super-charged stout but I’ve never heard of making coffee wine.
Full-bodied, with great intensity, tremendous purity, sweet, well-integrated tannin, and a long, blockbuster finish that lasts for 40+ seconds, it offers both power and finesse, a rarity in the super-charged world of big California reds.

(And the tenth is a "super-charged stout", discussed in the context of wine-making. Note, by the way, that Google now seems to match super-charged, super charged and supercharged to "super-charged" -- the last especially is a pleasant surprise.)

Well, I won't go on beating the dead horse. And perhaps Adam knew about this all along, in which case I've fallen for the joke hard enough already. Anyhow, folks, remember -- if you're about to make a generalization about usage, even as a joke, check with Google first.

Computer does something-or-other to Moby Dick -- in 9.5 seconds

According to a March 3 NYT article by Noah Shachtman, a company named Attensity has developed software that can "parse" Moby Dick in 9 and a half seconds. Why? "By labeling subjects and verbs and other parts of speech, Attensity's software gives the documents a definable structure, a way to fit into a database. And that helps turn day-to-day chatter into information that is relevant and usable." And the CIA, which helped fund the company, doesn't care about Moby Dick, but wants to use Attensity's software to "comb through e-mail messages and chat room talks".

This general sort of technology comes under the heading of "Information Extraction from Text", sometimes abbreviated "IE", or "text data mining", or "Automatic Content Extraction". (I'm a member of a group at Penn that's been working on information extraction from biomedical text.) The NYT article also describes applications that are really "information retrieval" (IR) rather than IE: "Looking through a company's customer file for a person named Bonds, for example, is fairly simple. But if the data is unstructured - if the word 'bonds' hasn't been classified as the name of a ballplayer or as an investment option - searching becomes much more difficult."

The NYT article also mentions Inxight, Intelliseek, and some other companies. To put this work in context, you'd want to sketch the history of DARPA's TIPSTER program, the series of "Text Retrieval", "Multilingual Entity" and "Message Understanding" conferences (TREC, MET and MUC), DARPA's ACE program, DARPA's CALO project, the failure of Whizbang! (some of whose technology went to Inxight, who in turn licensed it to Intelliseek, who also hired some of the people from Whizbang!'s Pittsburgh lab), and a few other things as well. Against this background, you'd want to know what kinds of analysis Attensity's software is really doing, and how accurate the various types of analysis are. That would enable you to (begin to) evaluate how well the software works in one sort of application or another. The software's speed is also worth knowing, but it's a second-level question. (And it also matters what the hardware is: "analyzing 200,000 words in 10 seconds" doesn't mean much, unless we know whether this is being done on a single machine, or a cluster of 1,000.)

Alas, the article gives us no real idea at all what Attensity's software is doing. Shachtman doesn't give us any coherent technical description of its analyses, or any examples of the analysis that it performs on any specific sentences. As a result, we can't tell whether it's trying to provide a full parse, or is just doing part-of-speech tagging and perhaps some noun-phrase or clause chunking. It's apparently doing some tagging of some types of entity references, but we have no idea which ones, or how well. It may be trying to infer some relations among text strings, or relate text strings to stable cross-document references (e.g. as identifiable people, places, organizations and so on). It's possible that it's even trying to do some sort of predicate-argument analysis, or at least analysis of the relations implicit in certain specified types of events and actions. That's certainly the implication of the description given by an Attensity customer:

"Attensity shows how the words all relate to one another - all the actors, objects and actions in a document, and how they connect."

But Shachtman seems confused about what the differences might be among these various sorts of analysis:

MAYBE sixth-grade English was more helpful than you thought. One of the dullest grammar exercises is being used to help find potential terrorists, and save companies a bundle.

Diagramming sentences - picking out subject, verb, object, adjective and other parts of speech - has been a staple of middle and high school grammar lessons for decades. Now, with financing from the Central Intelligence Agency, a California firm is using the technique to comb through e-mail messages and chat room talks, which can be a rich lode of corporate and government information, and a tough one to mine.

Shachtman seems to think that "diagramming sentences" is a matter of assigning part-of-speech labels to words. But actually, it's a kind of parsing, which assigns structural labels and relationships, recursively, to groups of words. On the other hand, "subject" and "object" are not "parts of speech", but rather (simplifying a bit) relationships between a noun phrase and verb. So Shachtman is recursively confused -- "diagramming sentences" is more than "picking out parts of speech", but two of the four examples he gives of "parts of speech" are actually examples of the type of relationships among groups of words that "diagramming sentences" is supposed to describe. And showing "all the actors, objects and actions in a document" would be another level of analysis entirely.

Shachtman's confusion, I'm afraid, reflects a historical mistake in his lede. Grammatical analysis of whatever kind -- whether diagramming sentences, assigning parts of speech, determining (co-)reference, or analyzing semantic relationships among words or their referents -- is far from being a "staple" of middle and high school education. Rather, it's become more and more rare in the American educational system at all levels. When it's done at all, it's less and less likely that the teachers themselves actually know how to do what they're trying to teach, since they themselves have never learned. And neither, it seems, have the reporters.

The botched description of "grammar exercises" in Shachtman's lede is not important at all. But it does help us understand why he apparently had no conception at all of what sorts of analysis Attensity's software might be doing, and therefore didn't ask any of the relevant questions while reporting his article, or present any of the relevant answers when he wrote it.

[Update: Martha Palmer emailed:

Check out David L. Bean and Ellen Riloff, "Corpus-Based Identification of Non-Anaphoric Noun Phrases", ACL-99,pp. 373-380.

It looks like good ol' muc technology, souped up regular expression pattern matching....w/ some pos tagging and some semantic grammar rules...

(and lots of hype, or course!)

(David L. Bean is co-founder and CTO of Attensity)

Well, there might have been some changes in their algorithms since 1999. And there's nothing wrong with a little good old American hype when you get a chance to be featured in the NYT. But in the best of all possible worlds, the tech writer assigned to a story like this at the NYT would understand the linguistic issues well enough to identify the underlying technology briefly but accurately -- instead of incoherently and misleadingly. ]

[Update 3/7/2005: Shachtman's article appeared on 3/5 in the IHT under the headline "Grammar become tool for CIA and businesses." ]

[Update 2/7/2005: Cassandre Creswell points out a more recent article: David Bean and Ellen Riloff, " Unsupervised Learning of Contextual Role Knowledge for Coreference Resolution", HLT-NAACL-2004.

You Ain't Haydn Nothin'

Amandalucia at We are Free Morphemes followed the links from Geoff Nunberg's recent Annals of Ambiguity post, and she has some editorial advice for presenters at the Eastern Psychological Association meeting:

Have these people not heard of the "Short Catchy Phrase: (colon) Longer Explanatory Phrase Detailing the Study" school of paper titling?

Her illustrative examples include (her addition in italics): "You Ain't Haydn Nothin': Gender Differences in Evaluations of Classical Musical Performances".

There's an opportunity for some historical sociostylistics here. Conference programs all the way back to the founding of scholarly and scientific organizations are available in libraries, and many of them are on line now. When did the CatchyPhrase: ExplanatoryPhrase model get started? In what fields? How did it spread? And what about variants such as DescriptivePhrase: CuteQuote, or ExplanatoryPhrase: AdditionalExplanatoryPhrase?

For particular examples, see for instance the program of the AHA's 2005 meeting, whose first few sessions include these colonated titles:

Existential Thought and Culture in Transnational Perspective: Authenticity, Morality, and Murder (session title)
"Dionysian Enlightenment": Walter Kaufmann and the American Nietzsche
Dasein, Death, and Dope: Martin Heidegger’s Lost Place in the History of American Existentialism
"Many Great Deliverances": Boston King’s Atlantic Revolution
Revisiting the Victorian "Crisis of Faith": The Transformation of the Religious Impulse in Physics, Feminist Literature, and Art (session title)
The Modern Art Museum as Sacred Space: Reformers, Aesthetes, and the Crisis of Faith
New Women, New Religion: Feminism and the Victorian Crisis of Faith
In Search of Epistemological Certainty in the Wake of the Victorian “Crisis of Faith”: Mind, God, and Meaning in the Physics of Sir Oliver Lodge
Racial Anthropology: The Galton Society and its Effects on American Physical and Cultural Anthropology
The Strange Death of Sociobiology and Evolutionary Psychology: The Case of Roswell Hill Johnson
Eugenics, Biomeritocracy, and the Reconstruction of Democracy: The Case of the High School Textbook, 1914–48
Defective or Disabled? Southern Physicians, Eugenics, and the Therapeutic Relationship, 1890–1930
War and the Modern City: Community Building and Urban Reconstruction during World War II (session title)
Global War, Local Battles: World War II and the Growth of Compton, California
Working-Class Utopia: Community Centers, Neighborhood Units, and American Architecture during World War II
Disease, Health, and the State in the Late Nineteenth- and Early Twentieth-Century United States: Charting the Colonial Connections (session title)
"The First Duty of a Citizen Is to Be Healthy": Health Education and U.S. Imperialism in the Early Twentieth Century
"Loathsome Necromancy": Health and Power around Puget Sound, 1967–90
"Suitable Care of the African When Afflicted With Insanity": Race and the Insane Asylum in the Nineteenth- and Early Twentieth-Century United States

There are some nice examples of the pattern Amandalucia referenced: "Dasein, Death, and Dope: Martin Heidegger’s Lost Place in the History of American Existentialism"; "'Loathsome Necromancy': Health and Power around Puget Sound, 1967–90". However, others suggest a more elaborate taxonomy of colon-divided titles, still to be catalogued by some enterprising meta-scholar. There's plenty of data out there.

[Update: I should mention that I've known several academics -- at least one linguist and one philosopher -- who were willing to do several months of research in order have a good reason to use a clever pun in a paper title. The Red Queen said "First the sentence, then the evidence". These scholars followed the maxim "First the title, then the evidence". In both cases, the goal is a useful source of motivation, I guess. ]

Stop him before he invents again

A brief piece in this month's Technology Review chilled me to the bone:

Imagine picking up a novel at a bookstore, and instantly your cell phone receives a text message containing your friends opinions of the book, as well as suggestions for films you might enjoy.

Media Lab doctoral candidate Hugo Liu is creating just such a system, called Ambient Semantics: a sensor embedded in a ring or wristwatch will read a radio frequency identification tag affixed to an object; the system will then search a database for information about the object and, on the basis of the wearers interests, send pertinent data as well as recommendations to his or her cell phone or PDA.

Ambient spam is more like it. I don't want to seem like an antitechnology sort of person, but the day that picking up a novel in a bookstore causes my cell phone to ring with a text message will be the day that I give up either bookstores or cell phones.

Ambient Semantics is powered by a database Liu built by gleaning information from Web pages, online communities, and social networks. The database, comprising some two million relationships between 100,000 items, can be used to predict personal preferences. For example, if a person likes Led Zeppelin, the database might indicate that she would also like the film School of Rock.

I see, it's like turning your real-world shopping over to amazon.com's feeble-minded model of personality. Future plans include putting your social life into the cyber-hands of a psychotic cruise line social director:

Liu says the next step for the system is feedback during personal interactions. Eventually, he says, when two strangers shake hands, their sensor devices will display their common interests and mutual friends, preventing missed opportunities.

This leads towards one of the stock characters encountered in dystopian science fiction novels, the animated billboard with the personality of a streetwalker or a used car salesman. These are generally backgrounded as routine annoyances of future life. But Mr. Liu's invention could spark a whole new back-to-nature movement.

I'm afraid that steering one Media Lab grad student in more reasonable directions would not help -- NTT's RedTacton technology is designed for exactly this sort of thing. (NTT site here, with sample applications starting here). Seriously, there are a lot of neat possibilities for this as well as now-conventional wireless technologies, but somehow I have a feeling that for every application I like, there are going to be several that I don't.

[Update 3/5/2005: Benoit Essiambre emailed:

The tech review article dumbed down his research so much that I can understand your concerns. Having your cell phone ring in the common circumstances described would be very irritating. However Liu's specialty is NLP. Looking at his website I am convinced that the choice of cell phone as an example medium was done because it is an example that concretizes the theory to a gadget a layperson can relate to. All the appeal of the system is in the powerful common sense semantic analysis going on in the background. Liu's specialty is natural language processing. His inclination towards art ( http://web.media.mit.edu/~hugo/research/index.html#aesthetiscope) and his work based on common sense knowledge (http://commonsense.media.mit.edu/cgi-bin/search.cgi) makes me think he could turn his system into something appealing.

Let me make it clear that I'm not against merchandising. I like advertising. But I'm not a big fan of spam, and a world full of networked objects engaged in common-sense reasoning about my goals, preferences, social networks and personal history is going to be a world full of spambots.

The idea of networked "do what I mean" technology embedded in everyday objects also raises its own reasons for concern, marketing misuses aside. But that's a topic for another post. ]

Speaking truth to TOEFL

There is much to shock the director of any program in the teaching of English as a foreign language (or indeed, the teachers) on the englishdroid site. For example, in its glossary of English language teaching it says this about the TOEFL test:

TOEFL  Bizarre American exam, in which candidates listen to robots intoning things such as, 'Wow, I sure hope my meticulously assembled entomology collection has not gotten misplaced by the faculty janitors.' A deep-voiced robot then asks, 'What does the woman mean?'

And there's more under "Types of classes":

TOEFL classes  Peculiar American test of a language deceptively similar to English. Combines archaic, formal grammar with folksy idioms. Mostly multiple-choice.

The listening test features unemployable actors or robots reading out unnatural sentences in plodding monotones, or with inflections in the wrong places. Childishly crude 'distracters' (red herrings) enable you to answer most of the test without even listening to the tape. There are some tricky ones, eg 'It is not impossible that the team will be less than successful in the final.' - 'What does the man mean?' (Answer: Fuck knows.)

TOEFL tests language structures that are obsolete outside the snootier American universities. For example, inversion after a comparative: 'Roger likes classical music more than does Rita.'

Ghastly stuff. But I fear it's not much of an exaggeration. This the sort of thing that does go on in the English language testing industry, as we have observed several times before on Language Log. The ETS is responsible for most of the sins committed in this area. Why should their Test Of English as a Foreign Language be any different?

I shudder to think that when I was a Dean of Graduate Studies and Research on my campus I had to oversee the process of making sure all foreign admittees had taken the TOEFL, and of denying admission to those whose scores were lower than the Graduate Council deemed acceptable. I really have been responsible for great evil. There must be dozens of talented Chinese physicists whose ambition to study at my university was forever frustrated because they could not tell from one hearing of a robot voice whether it was or was not impossible that the team would be less than successful in the final, or whether Roger did like classical music more than did Rita.

I feel honor bound to warn the casual surfer of the gross cynicism and unforgivable tastelessness of the englishdroid site. It can really only be excused on the grounds that teaching English as a foreign language is such hard work that after school hours its practitioners need a cheap laugh and a cold beer. And it will provide the former. You will giggle, as long as you can just prevent any goodness, decency, sympathy, respect or optimism from welling up inside you unbidden. I mean really, check this out: one-stop shopping for bad taste and slack professional ethics. Considering what a serious-minded, ethical, pro-feminist, responsible scholar I am, I can't imagine why I browsed this site for as long as I did...[Should that be "for as long as did I"? Just asking. —Ed.]

Especially since the glossary has the temerity to say this about grammar:

grammar  The G word. Once taught only by unimaginative fascists, but now possibly coming back into vogue.

(Possibly? Grammar is back, buddy, so watch it! And don't you call us fascists. Tomorrow belongs to us.)

On the other hand, you simply have to love a site that says this:

This site has no fucking pop-ups, no fucking adverts, no fucking Flash, no fucking frames, almost no fucking JavaScript (just the two quizzes, which wouldn't work otherwise, and the link on this page, an attempt to avoid fucking spam), no fucking fixed font sizes, no image rollovers that take fucking ages to load, no links that open in new fucking windows, no pages that 'work best' in Internet fucking Explorer, nor, I hope, anything else that drives cantankerous web users like me fucking bonkers.

The englishdroid site was recommended to me by Stephen Jones, who should be fucking ashamed of himself.

[All material in this post has been sent for review by the Federal Communications Commission for advance approval, though admittedly they have not yet reported back.]

Disagreements continue about "Hobbit" brains

Science Express has just published a new study of (a virtual endocast from) the skull of (specimen LB1 of) Homo floresiensis. The abstract:

The brain of Homo floresiensis is assessed by comparing a virtual endocast from the type specimen (LB1) with endocasts from great apes, Homo erectus, Homo sapiens, a human pygmy, a human microcephalic, Sts 5 (Australopithecus africanus) and WT 17000 (Paranthropus aeithiopicus). Morphometric , allometric and shape data indicate that LB1 is not a microcephalic or pygmy. LB1's brain size versus body size scales like an australopithecine, but its endocast shape resembles that of Homo erectus. LB1 has derived frontal and temporal lobes and a lunate sulcus in a derived position, which are consistent with capabilities for higher cognitive processing.

However, an AP wire service story on the paper quotes "some other researchers" as "sticking to their opinion that the Hobbit probably suffered from a form of microcephaly, a condition in which the brain fails to grow at a normal rate, resulting in a small head with a large face, and even dwarfism."

What it is not, says primatologist Robert Martin, provost of the Field Museum in Chicago, is a scaled-down version of a Homo erectus or a new transitional species that held on for millennia in tropical isolation.

As I noted in an earlier post, "this is all a bit of a black eye for physical anthropology. If paleontologists (and Science magazine!) can't agree about such a basic question, something is wrong."

In particular, the amount of quantitative data on statistical norms for various modern populations seems to be surprisingly small. At least, not much of it is used in this Science paper, whose plots and tables generally represent each species or group as a single point (e.g. in the plot of endocast Height/Breadth against Breadth/Length reproduced below, where there are individual points for "Homo sapiens", "Pan", "Pygmy" and "Microceph", as well as for various individual fossil skulls).

[Update: There's a NYT article today (2/4/2005) by John Noble Wilford, with a somewhat clearer quote from Martin:

Dr. Robert Martin, a primatologist at the Field Museum in Chicago, said that he and colleagues in Britain were preparing a paper contending that the examined braincase was too small to be explained easily by ordinary dwarfism or the tendency of isolated people on islands to become smaller over generations. In such cases, Dr. Martin said, brain size would usually not diminish by the same amount as body size. Dr. Martin said that he was not ready to rule out microcephaly on the basis of a test of a single microcephalic braincase.

One other thing that bothers me a little about this discussion is that the original find was not actually a fossil -- in the sense of being mineralized -- but was preserved in some other way that left it with a consistency described as similar to "mashed potatoes" or "wet blotting paper" (see the quote from the original issue of Science in my post here). Given this, how secure can anyone's estimate of the skull's aspect ratio really be? ]

Hindi and math skills

Let me add a brief remark to Sally Thomason's comments in the previous post, but couched in somewhat stronger terms. The conclusions she is dubious about concern the hypothesis that mathematical ability is retarded for English speakers because the root "ten" is not visible in a word like "fifteen" (indeed, the root "five" isn't very clearly visible there either). But these conclusions seem to me not just questionable but completely fatuous. It is having learned to count to 100 in Hindi that convinces me of this. To a rough approximation, there are no real signs of any transparency at all in the numeral words from 1 to 100 in Hindi. The morphophonemic alterations that have taken place over the millennia (a consequence, I would think, of a long tradition of rote learning of the numerals in purely oral form) are so radical that you could learn to count all the way up to n and still not be able to guess what the word for n + 1 will sound like, for any n < 101 (it gets easier after 100). Yet — and this is the point — anyone who thinks Hindi-speaking people are among the mathematically incapable knows nothing about India, or about California higher education, or the sociology of present-day Silicon Valley.

[Added later: All right, all right, I know I'm being utterly and unscientifically careless here, because India has thousands of languages and dialects, and many Silicon Valley programmers come from the Dravidian area in the south where the languages are less closely related to Hindi than English is. I know, I know. So all I'm saying is that I simply don't believe that a math deficit will some day be identified in Hindi speakers as a group that isn't there in Tamil speakers as a group and the explanation will turn out to be the transparency of the morphology in the numeral system. I'm not going to investigate it because I think the idea is completely loony. It wasn't me that raised this crackpot idea, was it?]

I won't type from 1 to 100 — spare me — but just to give a sense of the horrors that await, of how opaque the Hindi numerals are, I'll give the list from 1 to 20. (Phonetic nerd note: Single a is schwa, double is long [a], e and o are long, au is Cardinal Vowel #6 and ai Cardinal Vowel #3 for many speakers; ch is prepalatal stop, capital T is retroflex and t and d are dental, all aspirated if and only if h follows.)

1	one	ek		11	eleven	gyaaraa
2	two	do		12	twelve	baaraa
3	three	tiin		13	thirteen	teraa
4	four	chaar		14	fourteen	chaudaa
5	five	panch		15	fifteen	pandraa
6	six	chhai		16	sixteen	solaa
7	seven	saath		17	seventeen	satraa
8	eight	aaTh		18	eighteen	aTharah
9	nine	nau		19	nineteen	unniis
10	ten	das		20	twenty	biis

The really extraordinary thing is that it doesn't get much easier for the next eighty numeral words.

Does English Hinder Math Skills?

In the March 4 edition of The Chronicle of Higher Education an article on `Why Chinese Students Score High in Math' (p. A16) reports a claim by a University of Michigan psychologist that the greater transparency of numeral words for 11-19 in East Asian languages accounts in part for young Chinese, Japanese, and Korean students' superior learning of math by comparison to American students. Professor Kevin F. Miller comments:

"American children are confused by the fact that you name seventeen but you write it as ten-seven...Most first graders, at the end of the school year, if you ask them how many 10s there are in 17, they'll say 7."

The article (and presumably the scholar as well) acknowledges that other factors also contribute to Asian students' greater success in learning math, but the article emphasizes the argument that both the ordering of the numerals in the compound English 'teen words (3+10, 4+10, ...) and the semantic opacity of the words eleven and twelve make math learning harder for American first-graders.

The article is too short to raise, much less answer, some obvious questions: what about Asian-American students raised in traditional households but with English as their only first language? Is the same effect detectable in young students from other language backgrounds with numeral systems similar to the East Asian one reported here? How about students whose first languages have numeral systems that are more transparent or significantly less transparent than the English and East Asian patterns?

I was curious, so I conducted a quick and dirty survey of some of the grammars on my shelves. The Chinese-Japanese-Korean pattern is a bit more common in my nonscientific sample than the English pattern, with varying degrees of opacity in both types. Some examples: Hungarian, Swahili, Tagalog, Cherokee, Mundari, Turkish, Vietnamese, Dakota, and several Salishan languages all have the East Asian pattern, often with a particle or other connective separating the parts of the complex numerals. Most of these are fairly transparent systems, but the usual Dakota words for the 'teens seem to be composed of a particle meaning `again' + the unit (`again' + 1, `again' + 2, etc.). One grammar explains this as a shortening of the full form, which is (10 +) `again' + 1, etc.

On the other side, Russian, Finnish, Cherokee, Comanche, Colloquial Egyptian Arabic, Kurdish, and possibly Hindi and Urdu have systems similar to the English one, at least in that the words for 11-19 start with `1', `2', and so forth. But in this set of languages there are many very opaque 'teen forms. The Hindi and Urdu words don't appear to have `10' in them at all, and some of them don't even clearly have the numerals 1-9. Kurdish (which is not too distantly related to Hindi and Urdu) has more recognizably compound forms for the 'teens, but the phonetic distortion is sufficient to obscure the connection between the `10' element in these compounds and the word for ten. The Russian system also has some phonetic distortion in the `10' half of the compound, but overall it's more transparent than the English 'teen words. The second element in the Finnish 'teen words means `second', not `ten', and the second element in the Comanche words probably means `to go in/out'.

It would be nice if scholars would investigate a greater diversity of languages and systems before drawing conclusions about links between numeral structure and math learning. As reported in the Chronicle, the claim about the linkage strikes me as simplistic. In particular, the relative opacity of the English system could well mean that young children have learned the 'teen words as wholes, or as units plus an otherwise meaningless suffix, rather than as combinations of unit + ten; in that case, the compositional opacity might be the relevant factor, not the ordering unit + ten vs. the East Asian ten + unit pattern: as far as I can tell from the grammars (which, admittedly, usually give only the orthography for these numerals, not the pronunciation), the three East Asian languages mentioned in the article all have quite transparent compound numerals for the 'teen words.

Pechelingues

What with getting ready for Thanksgiving and all, I somehow missed Trevor's 11/15/2004 post entitled "Pirates and Kleinecke's etymology of 'pidgin'".

He starts from the multinational pirates known as pechelingues, pichilingues, or pechelingas in Spanish-American slang of the 16th and 17th centuries. Another point of reference is "a trade pidgin known as Pichingli" in the Canary Islands in the 19th century. There are several other intriguing historical notes, including a 1641 term for small change of diverse origin.

Trevor makes a suggestion about the origin of the word pechelingue (he finds the story about Spanish mispronunciation of the Dutch port of Vlissingen (Flushing) unconvincing), and a second suggestion about the origin of the word pidgin, namely that it was originally a term for the language of the pechelingues, which might originally have been spoken among the ethnically and linguistically diverse pirates of the Barbary coast (scroll down to "No hay nación de cristianos en el mundo de la cual no haya renegado y renegados en Argel"). He puts this up against the OED's story about the origin of pidgin, namely that it's

A Chinese corruption of Eng. business, used widely for any action, occupation, or affair. Hence pidgin-English, the jargon, consisting chiefly of English words, often corrupted in pronunciation, and arranged according to Chinese idiom, orig. used for intercommunication between the Chinese and Europeans at seaports, etc. in China, the Straits Settlements, etc.; also transf. (quot. 1891).

and a more recent suggestion by Kleinecke " that ‘pidgin’ may derive from a Yayo (South American) form ‘-pidian’, meaning ‘people’ and occurring in such tribal names as ‘Mapidian’, ‘Tarapidian’."

A useful introduction to the "Barbary pirates" can be found here. Last fall, on the occasion of "type like a pirate day", I pointed out their role as opponents in an earlier American-led global war on terror, complete with reluctant Europeans and other historical analogies.

annals of ambiguity

Scott Parker of American University sends me the intriguing title of a paper to be presented at the poster session of the Eastern Psychological Association in Boston next week:

Gender Differences in Rates of Breast Self-Exam and Testicular Self-Exam Amongst A Sample of Residents In A College Community

Not a very newsworthy result, was my first thought.

Gari

Today's New York Times (p. F11) has a review of a Japanese restaurant called Gari. It describes the food and atmosphere in some detail, but doesn't explain the name, so I will.

/gari/, usually written がり, is the special term used for red pickled ginger when it is served with sushi. The ordinary term is 紅生姜 or in kana べにしょうが /beni sjo:ga/ "red ginger". When you buy pickled ginger in a store the container is labelled 紅生姜, but when you eat it in a sushi shop you call it /gari/.

Ginger isn't the only thing that has a special name in a sushi shop. Green tea is normally called 煎茶 せんちゃ /sentja/, but in a sushi shop the special term 上がり /agari/ is used. Strictly speaking, /agari/ means "freshly brewed green tea", but you hear it mostly in sushi shops. /agari/ is a word with quite a few meanings. The most generic is "rise, slope, ascent", but among other things it is the term used to describe the death of fish and insects.

Structures of words vs. structures of numbers

A paper has just been published in the Proceedings of the National Academy of Sciences (PNAS) that offers a new perspective on several recent themes. The authors are Rosemary A. Varley , Nicolai J. C. Klessinger , Charles A. J. Romanowski and Michael Siegal, and the paper is called "Agrammatic but numerate" (PNAS, March 1, 2005, vol. 102 no. 9, 3519-3524).

Here's the abstract:

A central question in cognitive neuroscience concerns the extent to which language enables other higher cognitive functions. In the case of mathematics, the resources of the language faculty, both lexical and syntactic, have been claimed to be important for exact calculation, and some functional brain imaging studies have shown that calculation is associated with activation of a network of left-hemisphere language regions, such as the angular gyrus and the banks of the intraparietal sulcus. We investigate the integrity of mathematical calculations in three men with large left-hemisphere perisylvian lesions. Despite severe grammatical impairment and some difficulty in processing phonological and orthographic number words, all basic computational procedures were intact across patients. All three patients solved mathematical problems involving recursiveness and structure-dependent operations (for example, in generating solutions to bracket equations). To our knowledge, these results demonstrate for the first time the remarkable independence of mathematical calculations from language grammar in the mature cognitive system. (The full paper is available on the PNAS site only to subscribers, but a copy of the .pdf version appears to be here.)

By "bracket equations" the authors mean arithmetic expressions in which the scope of operators is indicated by matched parentheses or brackets, e.g. 4 + 11 x (3 x 2) vs. (4 + 11) x 3 x 2.

The results are certainly striking. These patients are profoundly aphasic, as would be predicted given the extent of their dominant-hemisphere perisylvian lesions -- each row of images below belongs to one of the three patients surveyed:

Their "severe disruptions in grammatical performance across language modalities" included ability to "[perform] no greater than at a chance level on understanding reversible sentences in both spoken and written modalities". ("Reversible sentences" are things like "The lady is calling the man", where it's equally plausible for the man to be calling the lady, so that you need to know more than the identity of the verb and the two noun phrases in order to guess correctly who did what to whom.) The patients also had other linguistic disabilities.

Despite their negligible syntactic ability in English, these patients could still do quite a bit of mathematics, of a kind that seems to require analogous syntactic processing. For example, in the "bracket expression" subtest,

Each patient calculated the sum of 90 expressions containing brackets. These included 64 expressions where the brackets were syntactic; i.e., if the participant adopted a serial order strategy, the result would be incorrect; e.g., 36 ÷ (3 x 2). The remaining interspersed 26 items were nonsyntactic: e.g., (3 x 3) - 6. The syntactic bracket expressions consisted of 38 items with a single level of embedded brackets and 26 items with apparent doubly embedded bracket structure. To avoid training performance, only 13 of these 26 items required serial computation of numbers contained within both sets of brackets, i.e., 50 - [(4 + 7) x 4] versus 3 x [(9 + 21) x 2]. Responses to the syntactic bracket expressions were scored for accuracy and presence of serial order calculation errors, e.g., 2 x [(5 x 2) + 5] = 25.

The results ( the three patients are called S.A., S.O. and P.R.):

This was just one of 14 mathematical subtests, and the patients' performance on all of them was strikingly better than their performance on analogous linguistic tasks.

This connects, at least loosely, to different sides of two recently-discussed stories. One is the role of language in mathematics, and especially in the most basic form of mathematical thought, namely counting. We discussed this in posts on the Pirahã (here, here, here) and the Mundurukú (here and here). The other is the role of recursion in language, discussed here, here, here and here.

In their conclusions, Varley et al. say that

In terms of the relationship between language and mathematics, our findings indicate considerable independence between the structure-dependent operations of language and number in an established cognitive architecture. Although agrammatic, all patients displayed sensitivity to, and use of, parallel syntactic principles in mathematics. Their responses are incompatible with a claim that mathematical expressions are translated into a language format to gain access to syntactic mechanisms specialized for language.

These results allow consideration of two alternative interpretations regarding the syntactic mechanisms of language and mathematics. One is that a common and domain-general syntactic mechanism underpins both language and mathematics but that mathematical expressions can gain direct access to this system without translation into a language format. In the case of patients with agrammatic aphasia, language representations are disconnected from the syntactic mechanism, but mathematical expressions can still gain access. The second alternative is that in the mature cognitive system, there are autonomous, domain-specific syntactic mechanisms for language and mathematics. Autonomy in the adult state does not entail independence throughout the developmental course of a system, and one mechanism might bootstrap the second. However, the presence of dissociations between mathematics and language in people with developmental language impairments indicates the potential for autonomous mechanisms throughout the lifespan and suggests that a language-specific mechanism does not bootstrap a nonlinguistic syntactic system.

[...]

With regard to the number lexicon, number words were unlikely to be the code in which calculations were performed; both S.A. and S.O. showed inefficiencies in using phonological and orthographic number words. Despite this, both were able to perform exact calculations involving two- and three-digit numbers. If, indeed, linguistic number words were the code in which calculations were performed, the inefficiencies inherent within these codes would have resulted in high error levels in mathematical tasks. All patients were efficient in processing Arabic numerals, suggesting that this code and its underlying conceptual base are sufficient for calculation.

The article left me with several questions.

One is whether this pattern (relative sparing of mathematical abilities) is typical of similarly agrammatic patients. If so, how has this fact remained unknown for so long? What other surprising dissociations are out there waiting to be discovered?

Another is how much use the patients made of pencil and paper techniques in solving the mathematical problems, since to some extent the learned procedures for arranging partial results on paper might substitute for (or at least assist) impaired ability to retain and manipulate the structure of mathematical expressions.

[Varley et al. link via email from Cosma Shalizi.]

Baby got what?

In my humble opinion, this is sacrilegious, while this is merely profane.

It was a clever pun all along :-)

With respect to discussions like this one about his use of the phrase Hobbesian choice in a New York magazine column, Kurt Anderson has explained himself by email to Jim Hanas, who posted the message at Encyclopedia Hanasiana.

Revenge of the Codex People

I've made a couple of references in passing to Michael Gorman's anti-Google and anti-blog screeds. Now the Online Computer Library Center blog reveals that these documents have deep historical roots.

The OCLC blog has several other relevant posts as well. And also see this OCLC report "2004 Information Format Trends: Content, Not Containers".

[Update 3/2/2005: for an antidote to Gormanitis, see Karen Schneider's post at Free Range Librarian. ]

[Link via email from Rich Alderson]

Now the FCC tells us, three months too late

It is astonishing that FCC chairman Michael Powell should tell us now, weeks after 66 TV stations made the decision not to show the movie Saving Private Ryan on Veterans' Day in case they were fined for the swearwords in the soundtrack (which for contractual reasons could not be edited or bleeped out of Spielberg's film), that really (now it can be told) the movie was OK for network television after all! My own opinion is that in a free society it is unacceptable for any words in the lexicon to be tabooed to such an extent that radio and TV stations can be retrospectively fined if they turn up in a broadcast (even one that the station did not write or produce). But if we are to adhere to the quaint old idea that there are taboo words of such potency that the government should be in charge of their use, the FCC should at least be prepared to say in advance what is going to be allowed and what is not. How can a responsible and decent organization like Language Log make its decisions about what words are fit to put before the public if the answer to its questions about what is currently taboo aren't even known by the fucking FCC? Oops...

"The horror of war and the enormous personal sacrifice it draws on cannot be painted in airy pastels. The true colors are muddy brown and fire red, and any accurate depiction of this significant, historical tale could not be told properly without bringing that sense to the screen," said Mr Powell, in some unusually colorful prose. (Who writes this stuff? And why do they take so long? The decision was actually reached on February 3rd.) Well, I should damn well think so: war is hell. But lots of things are hell. Are we allowed to swear about them occasionally in a free country, or not? The one prime condition to be met by a society whose laws make sense is that you should be able to know up front whether you're about to do something illegal. "We don't know whether we'll fine you or not" isn't good enough. I can hardly believe there has been so little critical coverage of this ridiculous episode of tacit censorship.

Horace diagnoses Ivan Watson's mistake?

This morning, Geoff Pullum noticed an "astonishing dangling modifier" on NPR:

Without Washington's support, however, Saddam Hussein quickly crushed the revolt.

Geoff was "quite surprised to catch such a great example of the sort of dangler you should avoid at all costs ... in scripted speech on National Public Radio". I agree, but I have a suspicion about how it happened.

Perhaps Watson (or whoever wrote the piece) originally had something like

Without Washington's support, however, the revolt was quickly crushed [by Saddam Hussein] ...

However, a later editorial eye saw an opportunity to replace a passive verb ("was quickly crushed") with its active counterpart, in order to become "more direct and vigorous". Result: the modifier was left dangling.

As Horace explained more than 2,000 years ago, "in vitium ducit culpae fuga, si caret arte". Avoidance of an error leads to a fault, if it lacks skill.

Without Washington's support... who??

An astonishing dangling modifier from Ivan Watson on National Public Radio's "Morning Edition" show this morning (listen to the story here; the example is just after three and a half minutes in). Talking about the Kurds and the brief period during which they overcame old feuds and rose in a united rebellion against Iraqi Arab rule, Watson goes on:

Without Washington's support, however, Saddam Hussein quickly crushed the revolt.

How's that again? Who was without the support of Washington?

Here's the technical grammatical description. The preposition phrase without Washington's support functions here as a clause adjunct at the beginning of the clause. It is understood as modifying the clause, but in a predicative way. It means what "without receiving Washington's support" would mean. We have an implicit argument slot to fill: who is it that didn't have Washington's support?

What we need is a target for the predication — roughly, a logical subject we could put with "receive American support" to make a clause with a meaning that makes things explicit in the right way. In such cases, it is extremely common for the subject of the matrix clause to be the key to making things clear. To take a couple of random examples pulled from the text of Bram Stoker's Dracula, when we read "Without saying any more he took his seat", we understand that the person who did not say any more was the person referred to as "he" (Quincy Morris, in this case). When we read "Without taking his eyes from Mina's face, Dr. Van Helsing motioned me to pull up the blind", we understand that it was Dr. Van Helsing who did not take his eyes from Mina's face.

"Dangling modifier" is the name prescriptive grammarians have given to the kind of construction where the main clause subject does not make clear the identity of the unexpressed target of the predication that is expressed in the adjunct. In many cases this causes little trouble, as the better usage books agree. But in the worst cases for intelligibility, the matrix clause subject is a disastrously wrong choice for the target of predication, with sometimes misleading and sometimes ludicrous effect. The first example cited by The Penguin Dictionary of American English Usage and Style (by Paul W. Lovinger; New York: Penguin Reference, 2000) is as clear a case as one could want, taken from a book about cannabis (if you could steel yourself, please, I want no politically incorrect giggling at this one):

Although widely used by the men, Bashilange women were rarely allowed to smoke cannabis.

Adopting the matrix clause subject (Bashilange women) as the target of predication for used by the man yields a truly unfortunate misunderstanding.

I know that a linguist like me is always assumed by the prescriptivist community to instantiate what E. B. White calls "the modern liberal of the English Department, the anything-goes fellow", as he tells his editor firmly that he will make no compromise with "the Happiness Boys, or, as you call them, the descriptivists" ( click here for White's remarks in context). But in fact, despite my scientific interest in describing languages as they actually are, I am as free as anyone else to have negative reactions to unintentional bathos or unhelpful confusion caused by bad writing. I think cases as plangent as the Bashilange example fully deserve the ridicule and censure that prescriptivists are so eager to heap upon them. However, it should not be overlooked that they're actually rather rare. Most dangling modifier cases slip by smoothly in context without anyone noticing them, which probably does mean there is no rigid syntactic prohibition against them built into the correctness conditions for the language; the principles they violate are more subtle pragmatic ones about normal understanding of implicit arguments in context.

I was quite surprised to catch such a great example of the sort of dangler you should avoid at all costs, and to find it in scripted speech on National Public Radio. Leaving it unclear whether it was the betrayed rebels or the nightmare dictator who lacked American support is a pretty gross error, especially given the history of America's vacillating alliances in Iraq during the 1980s and the 1990s.

On the characteristics of natural languages and certain European nationalities

Ever since I was an undergraduate, Barbara Partee has been one of my heros in the field of linguistics. So I was tickled to discover that the reading list for her (fall 2004) course "Linguistics 726 – Mathematical Linguistics (really: mathematics for and in linguistics)" includes a number of Language Log posts. The context is Barbara's Lecture 13: "Are Natural Languages finite-state languages? (and other questions)".

In her lecture notes as well as in the reading list, Barbara intersperses discussion of published articles (from Science, Psychonomic Bulletin and Review, and Cognition) with some less formal publications, including the Language Log pieces. If you're interested, read the whole thing: as you'd expect if you know Barbara's work, it gives a lucid and crisp exposition of a complex set of interesting issues.

You'll find a very different sort of exposition in Michael Gorman's much-discussed commentary in Library Journal, "Revenge of the Blog People!". Stung by negative reaction to his snarky op-ed on Google Print, Gorman strikes back:

A blog is a species of interactive electronic diary by means of which the unpublishable, untrammeled by editors or the rules of grammar, can communicate their thoughts via the web. (Though it sounds like something you would find stuck in a drain, the ugly neologism blog is a contraction of "web log.")

[...]

Given the quality of the writing in the blogs I have seen, I doubt that many of the Blog People are in the habit of sustained reading of complex texts. It is entirely possible that their intellectual needs are met by an accumulation of random facts and paragraphs. In that case, their rejection of my view is quite understandable.

Given all of this, I was surprised to find that one of the papers on Gorman's web site, entitled "How the English see the French", starts with this quote:

"A Frenchman must be always talking, whether he knows anything of the matter or not; an Englishman is content to say nothing when he has nothing to say -- Samuel Johnson (1790)"

Gorman goes on to assert that he is a "typical Englishman". Apparently things have changed since 1790.

The dogs of speech technology

The association of dogs and speech research in Geoff Pullum's recent post reminds me of a story. The time was 1977, about 3:00 a.m. one cold winter night. The place was Murray Hill, N.J., the home of AT&T Bell Labs research. Building 2, wing D, 4th floor, in the console room of the DDP-224 interactive computer. Joe Olive and I had been programming since dinner, and our new speech synthesis system was pronouncing its first phrases: "The birch canoe slid on the smooth planks." "Mesh wire keeps chicks inside." "The spot on the blotter was made by green ink." It sounded really good to us, maybe not totally natural, but clear as crystal.

One of the night-shift cleaning crew came in to empty the trash can. He looked around, puzzled, for the source of the sound. "Y'all got dogs in here?"

This was a good lesson on the role of expectations in speech perception, and the importance of formal intelligibility testing.

A similar point is one of the themes of Solzhenitsyn's novel "The First Circle", which deals with speech research in a Stalinist labor camp. But that's a story for another time.