Researchers Win Creative IT Grant

By Anna Lynn Spitzer

Irvine, Calif., April 14, 2008 -- Teaching students how to use metaphors is an everyday occurrence in English composition or creative writing instruction. Now scientists believe the descriptive phrases can help students understand and conceptualize scientific material as well, and two Calit2 researchers are part of a team using information technology to help teachers put these figures of speech to better use.

Ben Raphael

Bill Tomlinson, UCI assistant professor of informatics, along with education assistant professor Lindsey Richland and graduate student Eric Baumer, recently won a two-year, $200,000 CreativeIT grant from the National Science Foundation to develop and test a text-mining tool that can automatically identify metaphors in manuscripts.

“Computational Metaphor Identification for Supporting Creativity in Science Education” seeks to refine a computational method that identifies metaphors in large bodies of text, allowing researchers to evaluate their effectiveness in fostering creative learning in high school science students.

“The research is built upon the idea that the metaphors people use can help reveal the way they conceptualize a topic,” says Tomlinson. “One of the ways students learn new things about science is by using metaphors to relate new concepts to some other phenomenon that
they’re already familiar with.”

Genome Diagram

Patterns in Word Usage
The research builds on a technique called “selectional preference learning,” which finds patterns in the usage of similar words across different domains. Once these patterns are identified, they will be applied to classroom education, serving as a way to categorize those metaphors students use most successfully when they learn specific scientific concepts.

In turn, the selected metaphors will help students critically examine their own comparisons. They will alsopromote creative thinking by encouraging learners to develop additional metaphors.

The work is a continuation of research begun by informatics graduate student Eric Baumer, who was also a 2006-07 Calit2-Emulex graduate fellow, for his Ph.D. thesis. “Previous

researchers have explored how many scientific and technical innovations, from the cotton gin to Einstein's theory of general relativity, were all highly influenced by metaphorical thinking,” Baumer says. “One of the strengths of these techniques is that they analyze trends over large bodies of text, bringing to students' attention aspects of their own writing they might not otherwise notice.”

Tomlinson and Baumer seek to create new technologies that advance computational metaphor identification (CMI) techniques. Specifically, Baumer is developing new algorithms that will more efficiently analyze large chunks of text to identify metaphors and catalog them to determine which are most commonly used.

Richland and her students will develop innovative ways of teaching science that employ CMI to help students and teachers better understand their metaphor usage.

Encouraging Critical Thinking
While education researchers have successfully documented a link between learning and the use of metaphors, thus far they have not been successful in gaining wide access to students’ writings.

The use of CMI as a learning technology will provide a platform for identifying which metaphors underlie students’ scientific thinking, says Tomlinson. Integrating this technology into existing science software modules will allow researchers to test its effectiveness as a teaching tool.

“Critical thinking is a huge part of science, especially at higher levels,” Tomlinson says. “Having these kids develop an association between science learning and metaphor may put them in a better position to become innovative scientists later in their careers.”