Sedaris endorses compositionality
« previous post | next post »
Thanks to Graeme Forbes for alerting me to this! He has given me permission to post his note to his pro-compositionality friends. [For readers for whom compositionality is a new concept: it's a central tenet of formal semantics, usually credited to Gottlob Frege (but not without some controversy): The meaning of the whole is a function of the meaning of the parts and of the way they are syntactically combined. See, for instance: this introductory handout or the entry on Compositionality in the Stanford Encyclopedia of Philosophy.]
From Graeme Forbes:
You may have already seen this, but in case not, here's an excerpt from an article in the current New Yorker, "Easy, Tiger", by David Sedaris (July 11/18 2011, p.40). It's an entertaining piece about how he "mastered" Mandarin, Japanese and German with the aid of tourist-courses on his iPod, including one from a company called Pimsleur. The "Easy, Tiger" alludes to a phrase in the section on romance in the Mandarin course. Or was it the German course? Surely not!
The key paragraph from Sedaris's piece:
Pimsleur's a big help when it comes to pronunciation. The actors are native speakers, and they don't slow down for your benefit. The drawbacks are that they never explain anything or teach you to think for yourself. Instead of being provided with building-blocks which would allow you to construct a sentence of your own, you're left using the hundreds or thousands of sentences you have memorized. That means waiting for a particular situation to arise in order to comment on it; either that or becoming one of those weird non-sequitur people, the kind who, when asked a question about paint color, answer, "There is a bank in front of the train station", or, "Mrs. Yamada Ito has been playing tennis for fifteen years."
Steve Kass said,
July 18, 2011 @ 2:45 pm
Or “My hovercraft is full of eels.”
Paul McCann said,
July 18, 2011 @ 3:26 pm
Your introductory handout link is dead. Have a fresh one?
Dan Lufkin said,
July 18, 2011 @ 3:31 pm
Didn't N. Chomsky deal with this colorless green idea about 50 years ago?
Nathan Myers said,
July 18, 2011 @ 4:12 pm
Douglas Hofstadter proposed "Politicians lie", and "Cast iron sinks", uncontroversial statements of fact, and composed them into "Politicians lie in cast iron sinks". Is this an example, a counterexample, or something else?
Adding moderatory context doesn't help much: "Most politicans lie, much of the time, in cast iron sinks in melted butter" is no better.
Henning Makholm said,
July 18, 2011 @ 5:46 pm
There is such a thing as pro/counter-compositionality people? What exactly are the opposing positions? Surely nobody could take the extreme position that an analysis of natural language can only be valid if it is strictly compositional. Or, on the other hand, that compositional analyses can never be valid.
In computer science, we recognize compositionality as a desirable design goal in the construction of programming languages (or other formal notation). It is also recognized that there is often tension between this goal and the goal of making it reasonably easy to say things one wants to say often. Many of the trade-offs in language design contrast systematic simplicity with ease of expression. There are plenty of examples where fairly smart people have allowed short-term convenience to completely wreck the semantic tidiness of their programming language, so certainly it is not the case that the human mind automatically favors expressive systems that have a natural compositional structure.
(Everything can be given some compositional analysis, if only you allow sufficiently tortured and indirect senses of what is meant by "meaning". Doing so is not always useful).
Barbara Partee said,
July 18, 2011 @ 6:51 pm
@Paul – Oh, sorry about the bad link! It should have been:
https://udrive.oit.umass.edu/partee/MGU_Web_11/materials/MGU111.pdf
@Henning — I was careless, and Graeme himself certainly didn't say anything like 'pro-compositionality' — I should probably have just said 'friends interested in compositionality'. I agree with you. I think people differ by how strongly they favor trying as hard as possible to give a compositional analysis of a given construction. There are many different apparent desiderata for a good analysis of something, and such high-level principles as compositionality or modularity or staying close to surface syntax aren't really empirical, they're more like methodological principles. I've tended to be strongly on the compositional side, but not dogmatically. And I also see it as a kind of ideal, and as something with a pretty strong explanatory bite, but not absolute.
And sure, if compositionality is your only constraint, and you don't care what it takes to achieve it, it becomes pretty empty. As Emmon Bach used to say, it has to be accompanied by a "No Funny Business" constraint to be serious.
Dan T. said,
July 18, 2011 @ 8:45 pm
The "WikiMarkup" used by Wikipedia (and other Mediawiki-based sites) is an example of language-syntax designers painting themselves into a corner for short-term convenience; the syntax arose by developers adding features over the years as they felt a need for them, and trying for simplicity on the part of editors using the markup language rather than consistency of syntax from the machine's point of view; the result was a language that's extremely difficult to formally define or to build rigorously valid parsers for, which is causing great headaches in the current efforts to build "WYSIWYG" tools to make editing convenient to non-programmers.
Mark F said,
July 18, 2011 @ 11:03 pm
Following Henning's comment — what did people think meaning was a function of before Frege? Doesn't the fact that we have the term "idiom" tell us that compositionality is the general rule but that it has well-recognized exceptions?
This reminds me of Saussure's discovery that language is an arbitrary system of signs. I never understood what people thought language was before he came along.
Daniel Barkalow said,
July 18, 2011 @ 11:17 pm
Nathan: that whole is not combined out of those parts. (a) there are two meaning of "lie", and two meanings of "sinks" with different syntax; (b) there's more to the way parts are syntactically combined than linear order, even considering nothing further than grouping. For that matter, "(true statement) in (true statement)" is nonsense, and certainly not a true statement. It therefore demonstrates that, even if you've decided on compositionality, you still need to work out syntax as well as what meanings are and how they can compose when put together.
Philip Spaelti said,
July 19, 2011 @ 1:08 am
@Mark F: Saussure's arbitrariness was not a "new" idea. People had been going back and forth for centuries about whether meaning was arbitrary or somehow inherent to the words themselves. Each side would bring up its examples, the defenders of non-abitrariness would usually refer to things like onomotopoiea.
Saussure merely puts an end to this discussion. And then he goes on to show that arbitrariness of meaning explains sound change.
linda seebach said,
July 20, 2011 @ 9:44 am
As the eminent French mathematics consortium N. Bourbaki used to say, whenever they allowed themselves to lapse into what mathematicians actually say when they're working, "Par abus de langage . . ."
Tracy Canfield said,
July 23, 2011 @ 9:30 am
Actually Sedaris's criticism is less true of Pimsleur than it is of many other phrasebook-style recordings. A Pimsleur lesson routinely follows the format of giving you "I'm hungry. I'm not hungry. You're hungry." in the target language, then asking you in English how to say "you're not hungry" – asking you to, well, assemble the building blocks of what you've learned in order to produce a sentence you've never heard.
I'd say the biggest limit of the Pimsleur sets is that the written materials are ridiculously minimal – the Mandarin set has a little booklet that tells you 木 is "tree", and assures you you'll pick up the rest of the characters by seeing them around. That means your vocabulary will be limited to what you've actually been drilled on, and you might not be able to consistently extend what you've learned to new vocabulary items – you'll learn a number of first- and second-person verb forms, for example, but not that one ending is regular and can be used all over the place, and another is irregular and can only be used with the verb you learned it on.