Penn A&S masthead

Computers and Learning:
We Got Rhythms, Algorithms, and Wrinkles in Time

Computer technology has more applications in teaching and research with every passing day.
Here are only a few of the ways SAS faculty are exploiting the new possibilities.

Taking the Pain Out of Practice:
Cristle Collins Judd and the Department of Music

photo of Cristle Collins JuddWhen Maria, the spunky governess in The Sound of Music, discovered the musical deficiencies of the von Trapp children, she started them at the beginning. We all hummed along as the kids learned their scales and sang "do -- a deer, a female deer -- re -- a drop of golden sun," and were warmed as they achieved perfect harmony within five minutes, even as they romped on a mountain peak and hung from a carriage! The bad news is that Maria's time and circumstances are long past, and achieving perfect harmony takes years of practice. The good news is that there are now computers almost as talented at teaching as Maria was -- and Penn's Music Department has them.

The new Music Computer Lab was established in January with start-up grants from the University and the Pew Charitable Trusts, specifically to assist undergraduate students. Cristle Judd, assistant professor of music and the Lab's faculty administrator, explains that "we realized we were not teaching students certain skills as efficiently as we should. These include ear-training and sight-singing, the rudiments of piano, writing music notation, and eventually more sophisticated techniques like spectral analysis. We wanted a situation in which students could work individually or in groups, going at their own pace. If they could practice in a well-equipped lab, then they could do more creative things in their classroom time."

The new Lab makes use of the latest technology; it has eleven workstations, each with its own computer, interface, and synthesizer. (Word of these synthesizers got out early and proved quite a lure to students.) With the help of this equipment, students can produce camera-ready copy of music, either their own compositions or assignments for class. They can also hear scores that they have just written (these are sometimes more complicated than they can play for themselves). In addition to the skills of writing music, students are trained to "hear" notation without having the notes physically played for them. Computer programs can produce notation on the screen and evaluate how accurately a student plays it on the attached keyboard. Or the student may sing into a microphone on the computer, matching a given pitch or singing an unfamiliar melody. Errors -- those annoying wrong notes -- show up immediately on the screen.

As Judd points out, this technology is extremely helpful when "you have 15 students in a beginning class with widely varying skills and training. Labs used to wind up geared toward the middle of the class. So the best student was bored and didn't want to be there, and the weakest student never really got up to speed. But with a computer, a student can work in an interactive way, and lab time can be used much more productively."

Music theory courses require these essential skills, but there are computer programs available for music history courses as well. For example, students can listen to a piece on CD-ROM and have a score in front of them, plus a diagram that shows how the piece is put together. The Lab can also assist undergraduates to produce their own electronic compositions. With a keyboard and a computer editing program, a beginning composer can summon up and manipulate standard instrumental sounds, plus infinite other noninstrumental sounds. Once the composition is finished, it can be recorded.

The Lab is also helpful with more advanced techniques like the spectral analysis of music. What is a spectrograph? "It gives you a graphic image of the contents of a sound. It tells you about pitch, timbre, dynamics, overtones." Judd notes that this analytical technique is especially useful for non-western or other non-notated music. "A spectrograph produces a kind of transcription for music that resists conventional transcribing. You can then filter in or out certain kinds of sounds that you're particularly interested in studying. For example, if you're listening for a vocal inflection like a hard glottal stop, you can locate all of them and filter them out to see what else is happening around them."

Most of the software that enables students to do all these things can't be picked up at your local computer store. Some of it comes from specialty firms. Some was originally developed in Switzerland, in a sound lab dedicated to speech analysis. Some has been devised in academic settings. As a further electronic aid, Penn faculty have put some of their own work on the World Wide Web so that students can access it. For example, when students buy a music textbook, it always has musical notations interspersed through the text. The best students can look at the notation and recreate the sound in their heads. Others can sit at a piano and work out these illustrations. But some have trouble doing even that.

To help, Penn faculty take the illustrative notations and make sequenced files of them. Then a student reading a text can go to the correct Web site, point and click, and hear the eight measures of Mozart that appear on page 50 -- and all the other musical illustrations. They hear synthesized, not recorded sound. "The lab keyboards are very good and can bring the sounds alive for students who are not themselves practicing musicians."

Judd emphasizes that the benefits of the Music Lab apply to the full range of students in the department. In addition to its majors and minors, the department has a large service course enrollment. After getting the bugs out of the system, faculty hope to use the new technology in most music classes. The Computer Lab will then provide service for about 250 students each semester, though the time spent at workstations is expected to vary widely.

What does the future hold for the pre-professional and the interested beginner? Among other things, multi-media journals of music on the Web. Judd notes that there is already a Web journal for 17th-century music. "One article by a woman singer has recorded examples of early performance practice. Point and click: you see a modern transcription of 17th-century notation and hear her sing the example in more than one way -- with French and Italian pronunciation, for example." Music faculty, too, will increasingly unite graphic and sound capabilities in smooth, seamless presentations. "The computer can do all the timing and coordinating much better than a person trying to match transparencies with musical examples on CD." It can present a beautiful Vatican manuscript on screen while viewers hear the music exactly when they should. "In the past," Judd laughs, "you could do it, but what an effort! and it didn't always work." Even an experienced teacher could suffer from two left hands during such presentations. In the future, the only real worry will be a power failure -- and the sudden need to find a keyboard without a plug.

Collaboration at the Cutting Edge:
Donald Ringe and Linguistics

Donald Ringe photoIt sometimes happens that a tool invented for use in a particular field is less helpful than its inventor had hoped. An unforeseen hitch keeps most people from adopting it. "Yes, it's a great lettuce-corer, but we don't core our lettuces anymore. Thanks anyway." It would be satisfying if someone in a different line of work decided that the tool was useful after all. "I need to dig a lot of miniature post holes. This gismo is perfect! How soon can I have one?"

Something like this happened recently to Tandy Warnow, Assistant Professor of Computer and Information Science. Working in computational biology, she and her colleagues used cutting-edge mathematics to produce a variety of computer algorithms that would draw evolutionary trees from biological data. The algorithms would run in a reasonable amount of time and return true evolutionary trees -- each under particular circumstances and according to certain criteria. .

Warnow had beautiful research tools that would construct evolutionary trees in a variety of ways. She wondered if they might be applicable to other fields. What sort of tools would be needed in linguistics, for example? Languages are said to "speciate" much as biological species do, and their ramifying branches and development can be pictured as evolutionary trees. Warnow went to the head of the Institute for Research in Cognitive Science and asked if anyone at Penn working in historical linguistics might be interested in her work. The name that came up was that of Associate Professor Donald Ringe.

It took some time and discussion for Ringe to see just how Warnow's mathematical tools might be helpful to him. "It would be idle to suppose that you just crank linguistic data through a computer algorithm, and the Truth comes out," he says ruefully. "At first I hoped it might be that simple, but soon all sorts of questions arose." To get realistic results, Ringe, Warnow, and Ann Taylor -- a Penn Ph.D. in linguistics serving as a co-Principal Investigator on the project -- had to work out a whole new methodology based on linguistic data. It turned out that one of the algorithms was ideal for this kind of data -- an algorithm that had been much less feasible for use in biology.

Ringe thinks the new tool is more than promising. He can use it to test linguists' hypotheses about the development of particular language families. One example involves a long-standing dispute among scholars of Indo-European languages about the position of a subgroup called Anatolian. Languages of the subgroup, which include Hittite, are all extinct but were once spoken in what is now Turkey.

Likely tree for Indo-European Family"Is Anatolian just a subgroup or did the parental language split into two languages, one the ancestor of the Anatolian group and the other leading to all the other Indo-European languages? Do we have that binary split at the top of the tree?" Ringe began by disbelieving in this second view. But the algorithm kept returning trees with exactly that branching at the top. He is now at least tentatively convinced. "Since I began with a bias against what we got, I have to figure that what we got is probably correct."

But do languages really evolve, speciate, have descendants, and become extinct? Isn't that just a metaphor? "A biological population reproduces itself biologically," Ringe explains, "and a population speaking a given language reproduces itself through first language learning." Languages, however, don't change in response to natural selection as biological species do, but for other kinds of reasons. These may involve the structure of the human mouth and vocal apparatus. Or they may resemble the social reasons -- often involving prestige -- that other cultural artifacts change. "Right now, for instance, there is a vowel shift going on in Chicago that is changing the way 'Jan' and 'John' are pronounced. The changes are being quickly picked up in outlying suburbs and small towns, presumably because the way Chicago people talk is cool and to be imitated.

"So the metaphor is pretty good so long as you realize that although languages do die; they more often transmogrify into something else. Language death can certainly occur, however. The Gauls gave up their language to learn Latin, which is why French is a descendant of Latin. Hittite died, though it's unlikely that its speakers were exterminated. They probably switched to another language."

Ringe is clear that the new method of investigating these issues would be impossible without the computer. "The algorithm is so complicated that it can't be implemented by hand. If you can't do it computationally, you can't do it at all." The reaction to the method by other linguists has been mixed. Some definitely want to use it. "Really conservative linguists don't find this the kind of work they like to do." And some are interested but doubtful and waiting to see more. "This is still in progress."

As yet, the team has no user-friendly software, though they hope to rectify that in a matter of months. Better software should make the new method more accessible to colleagues and even to students. Meanwhile, the team has presented papers at a number of computer science conferences as well as linguistic conferences at Harvard and Manchester, England. Ringe is enthusiastic about tackling new problems with the algorithm, specifically the development of the Dravidian languages of India. There are also other fields where the algorithm could be useful, specifically in the study of certain kinds of manuscript transmission. A scholar might determine, for example, which medieval manuscripts have been copied from which other manuscripts. Conceivably there are other evolving "populations" that could be studied with this method or with some modification of it. In the long term, it's hard to keep a good algorithm down -- or to predict just how many applications it may have.

Going Where No One Has Gone in a Long, Long Time:
Holly Pittman and the History of Art

The glories of Assurnasirpal II, a king of ancient Assyria, survive but only in bits and pieces. His once splendid palace has long been a ruin buried in northern Mesopotamia (modern Iraq). When the ruin was excavated and dismantled in the nineteenth century, the breath-taking carvings that lined the throne room walls -- vivid scenes of hunting and war on eight-foot-high stone slabs -- wound up in the British Museum. With the help of books and a good spatial imagination, a student unable to visit the Museum might conjure up an impression of what the throne room once looked like. But the palace as a work of architecture is gone forever. No one will walk through it again.

Or so it seemed until very recently. The advent of computer technology and its graphical capabilities, however, has changed this situation. Now Dr. Holly Pittman, associate professor of history of art, can send her students on an electronically simulated walk-through of the palace, with its stone carvings in place on the throne room walls. A specialist in ancient Near Eastern art, Pittman finds much of the material that she studies distributed in museum collections all over the world and published in individual book plates. As she points out with Assurnasirpal's palace, "Students who don't normally work with architecture and who aren't used to reconstructing things in their minds really have a hard time going from individual pictures in books to reassembling one dark and overwhelming architectural space from 3,000 years ago." But now computer technology offers a solution to this problem. It has allowed talented art history majors to work with Pittman to reconstruct monuments and visual presentations that no longer exist in their original form.

Bull-headed lyre from the tombs of UrExamples include not only the Assyrian palace but reconstructed royal tombs from the city of Ur. Accoutrements of the kings and queens buried in these tombs -- objects fashioned 2,500 years ago from gold, lapis, carnelian, and other materials--now reside in the University Museum and were featured in a recent seminar of Pittman's. An electronic fly-through and reconstructions of the tombs and death pits from which the treasures came give a sense of the structures' interrelationships. There are also computerized plans of three of the graves. A student can go from point to point of a grave plan, calling up pictures and descriptions of the finds made there.

Computer technology, then, has allowed Pittman to give students an experience that was otherwise unavailable to them. And this can be done for any building or artifact that must be reconstructed. For example, she teaches an introductory class about Chartres Cathedral. "While I could do an elaborate 'virtual' presentation of its interior, I could also achieve much the same effect with a video." But if Chartres should disappear one day, then computer technology alone could allow students to experience the vanished interior of the cathedral "virtually" in three dimensions.

Are students responsive to these reconstructed presentations?

"Absolutely. After viewing a reconstruction, students have said, 'Now I understand what you're getting at.' These techniques allow them to see why it matters where things are in relation to each other. Showing one slide after another can't help them comprehend those relations in the same way. Also, students are cognitively ready for this technology from their exposure to it in the wider culture. They know how to process it."

Ram in the ThicketArt History professors use these techniques pretty regularly now to augment their slide lectures. Pittman stresses the word "augment." With their focus on the work of art and its aesthetic components, she and her colleagues do not yet routinely use computers to display objects during their lecture presentations. "The quality of reproduction just isn't as high as it is with transparency slides." So slides will survive for a while in the classroom. Students have review material available to them via computers and the World Wide Web. Even when the library is closed, works of art discussed in class can be reviewed by going to the relevant Web page. (Holly Pittman's site will even teach students the basics of constructing their own Web page.)

Preparing for the future, History of Art is also creating a library of digitized images, working from the 500,000 images in the slide library. Pittman is sure that delivering images digitally in the classroom will come before long. Quality must improve and costs go down, but "it's just a matter of time." There are also copyright issues to be resolved in the larger world. "Once the Louvre and the British Museum and other institutions make their images available, that, too, will remove a deterrent to use. The next three years will see an ever-increasing range of pedagogical applications. We're all teaching ourselves about these things, and our graduate students are becoming ever more educated in the technology."

Headdress of Lady Pu-abiWhat, specifically, does Pittman see coming along? "Sections of large survey courses will be augmented through interactive Web programs. Long-distance learning is going to explode." Penn scholars are talking now to colleagues in Oxford about having a satellite relay seminar on a museum topic. But however interesting the possibilities, "I myself do not think that anything will replace the value of face-to-face contact between student and teacher. I'm not afraid that we are all going to be replaced by machines. Human contact is too important to the learning process."

Pittman is eager to use the new technology as a way of exploring her mission, but she hasn't put it at the center of things. "I'm not a pioneer, I'm just comfortable with a new set of tools." Eventually, these tools may change the way we think and perceive, "but that will take a couple of generations." In the meantime, "I want presence. I don't want everybody to be virtual but to be flesh and blood."

Holly Pittman home page:http://ccat.sas.upenn.edu/arth/hp/hpittman.html


Return to Table of Contents