Interaction

February 15, 2006

The logic of language and learning

Photo: L.A. Cicero

Professor Byron Reeves is the director of the Center for the Study of Language and Information and is also
director and co-founder of Media X.

It should come as no surprise that the interface between human beings and technology is an area teeming with researchers at Stanford.

What exactly goes on in that interface? People use technology to study how people learn. They observe how people react—psychologically, cognitively, emotionally—to technology. They invent software that speaks to users in a language that makes sense to them; figure out if babies actually learn something from computer screens; study the similarities between human and artificial brains. In short, they explore the limitless facets of communication among and between humans and machines.

An enterprise like this is an obvious candidate for independent labs, which provide not only a space, physical and/or virtual, for cross-disciplinary collaboration, but also the flexibility and autonomy to go wherever the research needs to go.

L.A. Cicero

Linguistics professor Tom Wasow is a former director of the Center for the Study of Language and Information and co-founder of the Symbolic Systems Program.

At present much of that activity is taking place at two such labs: the Center for the Study of Language and Information (CSLI) and the Stanford Center for Innovations in Learning (SCIL).

Their missions and research often overlap. The director of CSLI, Byron Reeves, for example, is also the director and co-founder of Media X, a partnership network with private companies devoted to studying interactive technology. Media X is housed in the tech-rich Wallenberg Hall, which is managed by SCIL. Reeves also co-directs a project with Roy Pea, co-director of SCIL, called LIFE, dedicated to figuring out how people learn and also located in Wallenberg (http://life-slc.org/).

Altogether, Reeves said, there are some 10 centers, labs or groups on campus studying language-related technologies and the intersection between the human sciences and technology.

Back in the 1980s, when Apple was hiring philosophy graduate students to apply their powers of abstract thinking to the challenges of computers, a group of scholars at Stanford founded CSLI to study the emerging science of information, computing and cognition. They hailed from linguistics, philosophy, education, psychology and the relatively new field of computer science. Their mission was to develop software and other technology that responded to the structures and needs of human intelligence and language.

"The aim was to get computers to be more intelligent," recalled John Perry, the Henry Waldgrave Stuart Professor of Philosophy and a former CSLI director. "Understanding how people did complex projects, we thought, would help us improve computers. So philosophy was an important part of the mix. You had computer scientists reading Heidegger, for goodness' sake." But as the computers got much, much smarter, the philosophers became less valuable.

When CSLI was born, it was well funded thanks to a start-up grant from the Systems Development Foundation, an indirect offshoot of the RAND Corporation. (The foundation had issued a request for multidisciplinary proposals, and when those from Stanford, SRI International and Xerox PARC looked similar, SDF suggested they combine their efforts, which they did.) The money went into projects, some outstanding hires (including logician John Etchemendy, Perry's graduate advisee and current provost) and a beautiful new building, Cordura Hall. Since the money ran out, researchers have raised their own, which they've found mostly at the National Science Foundation and at various military agencies.

Linguist Stanley Peters, for example, another former CSLI director and the current director of the Computational Semantics Laboratory, which studies the intersection of linguistics and computer science (also called computational linguistics), is working on a project funded by the Office of Naval Research to see how (or if) robots can teach humans in a language that is mutually intelligible and accessible.

One of CSLI's earliest offspring was not a project at all but a program: Symbolic Systems, which currently is authorized by the Faculty Senate to grant bachelor's and master's degrees through 2008 (http://symsys.stanford.edu/).

"John Perry and I were running CSLI," linguist Tom Wasow said, "and John said, 'You know, we really ought to start up an undergraduate program.' I said, 'Great idea,' and we contacted people in computer science, philosophy, psychology and linguistics. We hammered out a curriculum, I wrote the boilerplate text on the reasons for the program, and we got it approved."

The program, administratively housed in the Linguistics Department, today graduates between 40 and 60 students a year. Beyond occupying some of its space, SymSys affects Linguistics in another way only fitting for an interdisciplinary program. Over the years, Wasow said, because SymSys and Linguistics students take many of the same classes, the far more numerous techies end up driving the debates, and Linguistics has changed as a result.

"SymSys is one of the most successful IDPs [interdisciplinary programs]," Wasow said proudly. "It fills a huge gap, because no other program combines fuzzy and techie work like it does. Students can study both operating systems and how the mind works. It sucks up the best students like a magnet."

Reeves, the current CSLI director, is the Paul C. Edwards Professor in the Department of Communication. His office is in Wallenberg Hall, with Media X and SCIL, but across campus from Cordura Hall. So what does being or not being an independent lab mean?

Above all, he said, the value is practical. The labs administer multidisciplinary research, manage grant money and aren't bogged down with instruction and hiring decisions like departments are.

However, they are not necessarily physical entities.

"New buildings are overestimated," he said, though he admits he likes Cordura Hall. "The question is, what level of administration and formalism is necessary to create interdisciplinarity?" In his view, you need a light rather than a heavy model, you need "stuff you can do, with boots on the ground." If you've got that, then a common building, while nice, is less important.

"But faculty can't do that on their own," he added. "So we stake out a middle ground, try to create a hub for a virtual network, which is incredibly valuable. The lighter you stay, the easier it is for institutions to come and go. If they have their own building, they never go."

In celebrating CSLI's existence, Peters and Wasow particularly praised its internship program, funded by the university, which pairs undergraduates with professors or postdocs over the summer. Bragging on Wasow's behalf, Peters said he has between three and five SymSys interns a year at the Computational Semantics Laboratory and they're all wonderful.

For Perry, the independent labs "have to provide people with something they don't have from their departments, and that varies from department to department." Some people need money; that was the case with the philosophers when CSLI started. Others need more space for themselves; still others need a venue where they can meet with researchers from other fields, which was the case with the people from SRI International.

The present-day Wallenberg Hall occupies the last portion of the Quad repaired after the 1989 Loma Prieta earthquake. With a grant from two of the private Swedish Wallenberg foundations, supplemented by the university, the building was gutted and then equipped with offices and classrooms. Five of the classrooms are managed by SCIL, though they can be used by anyone who applies; they are showpieces of what technology (both high and low) can do for learning.

In a marriage of genius and common sense, the five "flexible classrooms" allow students with laptops to use their own cursors to get onto big common screens, save their work on a common website, and work collectively on collapsible whiteboards and then photograph their results onto the website. This ability to employ technology while simultaneously recapturing and nurturing the collaborative mood of the classroom is emblematic of the mission of SCIL, which was launched as an independent lab in 2002.

The top floor of Wallenberg Hall provides space for some of the more remarkable endeavors of SCIL, CSLI or both.

Chief among them is the Learning in Informal and Formal Environments (LIFE) project, which in 2004 received a five-year, $25 million grant from the National Science Foundation (NSF). In conjunction with colleagues from the University of Washington and SRI International, researchers here are using technology to, in the words of Roy Pea, co-director of SCIL and a professor at the School of Education, "develop a science of learning." Reflecting the NSF's willingness to embrace multi-institutional and multidisciplinary projects, there are five PIs; Pea is one, and one of the research "strand leaders" is Reeves.

"We could never have done this without the independent labs," said Reeves, referring to the fact that the LIFE award was made to SCIL. "It combines neuroscience, anthropology, curriculum development and the humanities, and it's multi-campus." LIFE itself has spawned some 30 research projects.

Also on Wallenberg's fourth floor—what Pea called the "mix-it-up space"—is another child of CSLI, Media X, where Reeves wears yet another of his hats. Media X brings together researchers, faculty members, students and industry designers to study and develop interactive technology. Among the dozens of projects it has funded since 2002 is one led by Pea called "Science Education Network of Sensors," one led by Reeves investigating how computer-supported stories can change people and one led by Wasow on "Learning English via Robust Conversation." Many of the Media X researchers are affiliated with Symbolic Systems; they come from communication, engineering, law, linguistics, medicine, psychology and education, among other areas (http://mediax.stanford.edu/).

Media X projects are funded by the industry partners, which at present number around 30 companies worldwide. They establish the fields to be researched, and Media X then issues requests for proposals.

In a way, the organizational mesh of research and development about the interaction of humans and technology is as complex as the object of the study itself. Twenty years ago, no one could have imagined the directions technology would take or how the intimate relations between humans and their inventions would change the way we think and learn. Visionary and audacious researchers at Stanford figured out ahead of time that they needed to be there. They're still there, changing with the times, both catching up to innovations and leading the way.