Brandeis University
A M AG A Z I N E D E V OT E D TO R E S E A R C H AT B R A N D E I SVO L U M E 4 / N U M B E R 2Brandeis UniversityWhy Your ComputerDoesn’t Understand YouJames Pustejovsky has his eyes on the prize—a computer that talks in any language andknows what’s in your fridge.ENCORE!In creating music, these award-winningcomposers know the score—anything goes aslong as it’s good.Catalyst \\BRANDEIS UNIVERSIT Y1
FrontlinesSPRING 2010VOL. 4, NUMBER 2Dear Reader,Catalyst, the Brandeis University research magazine for the sciences,social sciences, and humanities, chronicles activities that make thiscampus an exceptional center of learning, teaching, and scholarship.From new programs focusing on entrepreneurship and computational linguistics, to forefront research on race and free speech inthe nineteenth century and the biology of memory, Brandeis ishelping define liberal arts for the twenty-first century. Catalyst isone compelling way we are telling that story.In this issue, James Pustejovsky, professor of computational linguistics, describes the complex challenges that artificial intelligencefaces in communicating through language—any language—theright nuances of meaning and word senses. You can try your hand atcomputational linguistics by taking the quiz on page twelve.The language of music—composition—is also thriving at Brandeis.In these pages you will discover how three award-winning composers are creating dynamic new music that embraces diverse soundsand styles. From the early days when Aaron Copland and LeonardBernstein were faculty members, music composition has been aparticular forte here.You’ll also learn how psychologist Derek Isaacowitz has pioneeredthe use of eye-tracking technology to show that, contrary to conventional wisdom, older adults do not view the world negatively, butrather are able to maintain an even, sunny outlook. But it’s a different story for young people.CATALYST IS PUBLISHED TWICE A YEARBY THE OFFICE OF COMMUNICATIONS.PROVOST AND SENIOR VICE PRESIDENTFOR ACADEMIC AFFAIRSMARTY WYNGAARDEN KRAUSS, PHD’81SENIOR VICE PRESIDENT FORCOMMUNICATIONSANDREW GULLYEDITORLAURA GARDNERDESIGN DIRECTORDEBORAH WIEDERPHOTOGRAPHERMIKE LOVETTSTAFF WRITERSMARJORIE LYONMAX PEARLSTEINTHERESA PEASECHARLES RADINI am confident you will enjoy these stories and the other profiles,commentary, and news briefs in Catalyst.CONTRIBUTING WRITERSIRENE ABRAMSDAVID C. ENGERMANCHANDLER FULTONDAVID E. NATHANSincerely,COVER ARTROBERT BRINKERHOFFAdam B. JaffeFred C. Hecht Professor in Economicsand Dean of Arts and SciencesCORRESPONDENCECATALYSTMS 064 BRANDEIS UNIVERSITYPO BOX 549110WALTHAM, MA02454-9110BRANDEIS.EDU/CATALYST2BRANDEIS UNIVERSIT Y// Catalyst
contents2 THE PIPELINE3 NOTABLE RESULTS4 ENCORE!Brandeis is nurturing a culture of entrepreneurshipthrough expanded industry relationships and newacademic programs that incorporate businessbuilding know-how.A towering figure in labor law, Jules Bernstein ’57made his name fighting for workers’ rights.In creating music, Brandeis’s award-winningcomposers know the score—anything goes aslong as it’s good.8 WHY YOUR COMPUTERJames Pustejovsky has his eyes on the prize—aDOESN’T UNDERSTAND YOUcomputer that talks in any language and knowswhat’s in your fridge.13 HANDS ONTaisha Sturdivant ’11 has never let her tragicbackground get in the way of life or her plans toimprove it for others.14 DEREK ISAACOWITZThe psychologist’s eye-tracking research reveals rosyIS IN THE MOODresults in older people.16 OBSERVATIONS18 RESEARCH NOTESSoviet area studies helped us understand our coldwar enemy. So why not follow the same model tounderstand the terrorist threat or Afghanistan?From love and death in lyric poetry, to research onhigh-energy physics and active matter, Brandeisscholars generate new knowledge to make the worlda better place.20 THE SHORT, SIGNIFICANTLIFE OF THE FRIEDLAND How one building came to epitomize the pace ofBIOCHEMISTRY BUILDING scientific discovery at Brandeis.Catalyst \\BRANDEIS UNIVERSIT Y1
( )(VP(V)(NP))( P)Y(S(NP)(VP(V)(NP)))8BRANDEIS UNIVERSIT Y// Catalyst
(NP)(v)UTER(S(NP)( (np)(vp(v)(V(VP)(V)( WHY YOUR COMPUTERDOESN’T UNDERSTAND YOU)JAMES PUSTEJOVSKY HASHIS EYES ON THE PRIZE—ACOMPUTER THAT TALKS INANY LANGUAGE AND KNOWSWHAT’S IN YOUR FRIDGE.By Theresa PeaseCatalyst \\BRANDEIS UNIVERSIT Y9
Therapist: Hello, I am Eliza. Howcan I help you?Patient: I need my head shrunkreally fast.Therapist: Why did you say yourhead shrunk really fast?Patient: I didn’t say it shrunk. I saidI need you to shrink it.Therapist: You are being a bitnegative.Patient: I’m going to jump off abridge. I guess you could call thatnegative.Therapist: Do you wish to guessme?Patient: No, I wish to swing from atree screaming nasty words.Therapist: You are being a bitnegative.Patient: Just as I thought; you can’thelp me. Life stinks. Don’t youthink life stinks?Therapist: Oh, I think life stinks.JAMES PUSTEJOVSKYBelieve it or not, dialogues like this onefeaturing MIT computer scientist JosephWeizenbaum’s program Eliza riveted thenascent artificial intelligence communityforty-five short years ago. Designed tosimulate a Rogerian psychotherapist,Eliza parroted back key phrases soconvincingly that some users scarcelybelieved they were speaking to amachine. Despite the awkward phrasingand frequent errors, philosophers,linguists, and computer scientists theworld over thought that the day whenyou could comfortably converse with arobotic assistant was just aroundthe corner.Fast-forward almost half a centuryand you’ll find another generation ofscholars far more cautious in their estimateof exactly when you’ll be able to tradegossip with R2-D2, C-3PO, or, for thatmatter, your automobile. Just think aboutthe chats that take place when you are,10BRANDEIS UNIVERSIT Y// Catalystsay, on hold with American Airlines, andyou’ll realize that Eliza, history’s mostfamous therapist wannabe, was but achimera.easily confused cybersourcesIt’s not that progress has not been made.If you want to know yesterday’s sportsscores, learn whether your neighbor’shouse is for sale, or get the names of SnowWhite’s seven dwarves, you’ll surely beable to find your answer with a few clicksof the mouse. You can get reasonablyaccurate directions from your TomTomand dispute consumer problems with arobotic customer-service agent on eBay.But try to ferret out more open-endeddata—say, what books have beenwritten about post–Civil War reconstruction since 1942—and you’ll confuse yourcyber sources woefully.That, says James Pustejovsky, headof the linguistics section in Brandeis’sDepartment of Computer Science, isbecause Google and its cousins have noway to discern references to the U.S. CivilWar from any other sentences thatcontain the words “civil” and “war,” andno knowledge of the word “since.” Infact, the search engine has no sense oftime at all. To make matters worse,computers have no appreciation for irony,and they don’t understand the relationship of words within a sentence: whetherJoe killed Mary or Mary killed Joe is allthe same to them. They are not good atsynonyms; if you ask, “Who purchasedthe Sears Tower?” your search enginemay completely skip a headline aboutsomeone who “bought” or “acquired”the building. Further, such programs donot grasp common idioms or have acommand of what Pustejovsky calls “thesocial clues.” Greet Eliza with “Give mefive, Baby,” and she’ll respond with,“Don’t you ever say hello?”
teaching smart computers to thinkIn short, those “smart” computers can bepretty dense compared to the moresophisticated and nuanced communications that engage the human brain—asituation that graduate students in a newMA program at Brandeis are tryingto correct.“Human beings, even by the age ofthree or four, do an amazing job atunderstanding words in context, with alltheir different nuances and meanings,”explains Pustejovsky, who created a newgraduate program in computationallinguistics. “You can give a computer atemporal narrative about Johnny. Youunderstand the implied context: Thedesk was in his house, there was a chair atthe desk, and to get to the door Johnnychanged his posture and walked. Heprobably turned a doorknob. Understanding the narrative’s hidden meaning is alarge part of artificial intelligence, and allour efforts are aimed at building analgorithm that will make it possible formachines to do that.”Understanding this complex subtext(Pustejovsky calls it the “dark matter” ofsemantics) involves multiple parametersin the brain that he and his team are reet Eliza with “GiveGme five, Baby,” and she’llrespond with, “Don’t you ever say hello?”trillion numbers and it will make senseout of them. But if you tell a computer,‘When the alarm rang, Johnny got upfrom his desk, went out the door, and leftthe house,’ it won’t be able to follow thestory. It can’t figure out what happened,why, or in what order.”Indeed, it was an underestimate ofhow much knowledge the human brainreally contains that caused mid-twentieth-century artificial intelligence pioneersto misjudge the enormity of the taskbefore them. Discovering how the brainwas “hardwired” and then mimickingthose processes to produce machine“intelligence” sounded infinitely moredoable to Weizenbaum’s colleagues thanit does to contemporary researchers.But the difficulty doesn’t dilute thethrill of the quest, says Pustejovsky, whofor the past twenty years has led aBrandeis research group trying to teach acomputer to visualize and re-create asimple scenario like Johnny’s.the dark matter of semantics“If you understand English,” Pustejovskynotes, “you can create a spatial andstriving to replicate in the computer.Noting that such efforts involve studyingthe way children learn to tell stories,Pustejovsky says, “I’m not a cognitivepsychologist, but I am really interested incognitive development, because thesecomputational models will succeed onlyinsofar as they reflect the way weourselves learn.”Mathematicallyinclinedasayoungster, Pustejovsky first becamecurious about language when as ahigh-school student he went on a scubadiving trip in Mexico. During a dayashore, he visited Mayan ruins and wasriveted by the hieroglyphs he saw.Although the youth entered MITwith the intention of majoring in electrical engineering or math, he heard inhis sophomore year about a fascinatingsyntactical analysis class his roommatewas taking with world-renowned linguistKenneth Hale. It was the first year of anew undergraduate degree programin linguistics and philosophy, and Haleshared faculty billing with other superstars like the linguist Noam Chomsky andJohn Robert Ross, a syntactician.“That day,” Pustejovsky says, “Idropped physics and signed on for Ken’sclass. Within two weeks, I had droppedtwo more courses and was enrolled inthree linguistics classes.”Pustejovsky was no polyglot. In fact,he grew up in a bilingual English- andCzech-speaking home in Texas withoutamassing a Czech vocabulary of morethan four dozen words. Instead, it was thetechnical aspects of language thatfascinated him.“Linguistics was analytic, and it wasa lot like math,” he says. “It was patternanalysis, really, looking for the structureof language. I wasn’t interested in theemotive content, and I certainly was notdrawn in by the literary aspect; I hadnever been excited by English classes. Itwas the scaffolding, the underlyingarchitecture of language, that got myattention.”applying mathematical principlesto languageBy making logical and algebraic connections, Pustejovsky realized, he could applymathematical principles to language.This enabled him to mine various tonguesfor similarities and differences in wordorder, sentence construction, subjectverb agreement, and other properties ofspeech that shed scientific light on thebrain’s innate ability to receive and sendcommunications. It was like solving anarcane and very complex puzzle.After graduating from MIT in 1978,Pustejovsky received a DAAD grant—aGerman Fulbright Scholarship—to explorelinguistics and philosophy at PhilippsUniversity in Marburg, Germany. In 1985,he received a doctorate in linguisticsfrom the University of Massachusetts atAmherst, then stayed around for aneighteen-month postdoc in artificialintelligence. After joining the Brandeiscomputer science faculty in 1986, hecontinued to do linguistics research,often with funding from the NationalScience Foundation, National Institutesof Health, or DARPA (Defense AdvancedResearch Projects Agency), and topublish theoretical linguistics articles inscholarly journals.Linguistics classes, however, tookplace in Brandeis’s psychology department until 2005, when the under graduateCatalyst \\BRANDEIS UNIVERSIT Y11
linguistics program was reorganized as aninterdepartmental program. Pustejovskyjoined the program faculty and nowserves as chair. The program is designedto provide a broad, liberal-arts viewof linguistics, while the new master’sprogram concentrates more heavily onunderstanding how languages operate inorder to put that information to work inartificial intelligence applications.from english to mandarin“We chose to move the Brandeis programin this direction because computationallinguistics is a growth industry. Peopleneed to develop communication andlanguage software for the Web, forindustries, for government, and fortranslation purposes. This is not aboutcreating applications that make moneyfor Google. This is about offering a solid,cross-disciplinary, foundational programthat has ties to anthropology, classics,psychology, and neuroscience,” saysPustejovsky, recently named to the TJX/Feldberg Chair in Computer Science.The fledgling graduate program hastwo tracks: a five-year BA/MA degreeand a two-year master’s program. Thefirst class of five-year students graduatedin spring 2009, and the inaugural groupof two-year MA candidates arrivedlast fall.Among them are students who aspireto teach at the university level, developapplications for various industries, andput computational linguistics skills towork deciphering ancient scripts. Otherpotential applications of computationallinguistics include facilitating literaryanalysis and the study of literary history;performing complex archival tasks; andenabling the cataloging and retrieval ofphotographs and other images that donot include text, Pustejovsky notes.“The Holy Grail of computationallinguistics and artificial intelligence,” hesays, “would be a reliable hand translator. Someone could speak English into adevice in Beijing, and out would come atranslation in Mandarin, with just the rightword senses and nuances of meaning.“If we are successful in our efforts,”Pustejovsky says, “more mundaneaspects of our lives will also be impactedby this work. For example, the time willcome when you will be able to ask yourrefrigerator what you need to buy at themarket, and it will tell you, ‘Well, the milkhas been in here for two weeks, and thefruit section is empty.’ Or maybe it willjust print out your shopping list for you.”Theresa Pease is senior editor in theOffice of Communications and editor ofBrandeis University Magazine.BREAK THE CODEtenji karaokeBraille is a tactile writing system, based on a series of raiseddots, that is widely used by the blind. It was invented in 1821by Louis Braille to write French but since has been adapted tomany other languages. English, which uses the Roman alphabetjust as French does, required very little adaptation, butlanguages that do not use the Roman alphabet, such asJapanese, Korean, or Chinese, are often organized in a verydifferent manner.The following tenji words represent atari, haiku, katana, kimono,koi, and sake. Which is which? You don’t need to know eitherJapanese or Braille to figure it out; you’ll find that the system ishighly logical.a. b.c.d.Below is a Japanese word written in the tenji (“dot characters”)writing system. The large dots represent the raised bumps; thetiny dots represent empty positions.e.f.karaokeWhat are the following words?g.Answers: A. haiku B. sake C. katana D. kimono E. koi F. atari G. karate H. anime1212 B RBARNADNEDI SV IEVRESRI TS IYT Y// //CatalystE IUS NUI NCatalysth.
catalyst \\ brandeis university 1 spring 2010 vol. 4, number 2 2 the pipeline 3 notable results 4 encore! 8 why your computer doesn’t understand you 13 hands on 14 derek isaacowitz is in the mood 16 observations 18 research notes 20 the short, significant life of the friedland how one building came to epitomize the pace of biochemistry building
Brandeis University. This report also includes in stitutional policies concerning campus security, such as those regarding sexual assault, alcohol, and other drugs. Brandeis University distributes a notice of the av ailability of this Annual Security and Fire Safety Report by Oct. 1 of each year to each member of the Brandeis University community.
Brandeis University Class of 1977 Reunion. Brandeis. University: September. 1973--May. 1977. . University Plans Dropping of 16 Faculty for '74-75; Cross Country Wins Again; Outruns Wesleyan, Bowdoin . Brandeis Expands Pre-Med Advising. The Origin of Species (review of. 2001: A Space Odyssey) 4/23/74.
Welcome to the Brandeis University Defined Contribution Retirement Plan! Start investing in your future— and yourself—today. Brandeis offers Fidelity Investments and TIAA as retirement service providers. This User Guide will walk you through how to enroll in your Brandeis Retirement Plan online and how to make changes to your account.
Requirements for U2snRNPaddition to yeast pre-mRNA Xiaoling C.Liao, Hildur V.Colot, YueWang and Michael Rosbash* Howard Hughes Medical Institute, Department of Biology, Brandeis . Department of Biochemistry, Brandeis University, Waltham, MA02254, USA \11--D 1992 Oxford University Press. 4238 Nucleic Acids Research, Vol. 20, No. 16 MATERIAL .
Professor of International Economics, Brandeis International Business School, 2014-2017. Barbara and Richard M. Rosenberg Professor of Global Finance, Brandeis
Fall 2021 Staff Department Plan. Please describe you. r overall Fall 2021 operational plans and how you will provide support to the Brandeis community. Please consider and include: The core components of support needed for the Brandeis community.
BRANDEIS BLAST June 1, 2021 Have a great summer!! The Brandeis Blast will return on August 15th. STATE OF THE CHAPTER: A Report on Our Activity July 1, 2020 – June 30, 2021 I think that everyone can agree on this word to describe the past year: UNEXPECTED.
Resume Summary Education · Harvard Ed School ’89 ’98 · Boston University Corp. Education o Business Analysis · Brandeis University 1966 o French (concentrations: Russian, mathematics) · Bryn Mawr College in France 1965 · URF College Math 1961 Summer Employment ·
