Skip to main content
Expand AU Menu



CAS Dean's Office
Fax: 202-885-2429
Battelle-Tompkins, Room 200

CAS Dean's Office
4400 Massachusetts Avenue NW
Washington, DC 20016-8012


Talking to Avatars: Prof Creates a Learning Device

By Ariana Stone

Photo credit: Ariana Stone

Photo credit: Ariana Stone

When it comes to learning new foreign languages, real-life context is everything, according to Luis Cerezo, new professor of Spanish linguistics and translation and director of the Spanish language program.

Cerezo holds degrees in machine translation, computational linguistics, and Spanish applied linguistics. His varied training enabled him to create Talking to Avatars, a computerized language tutor that places users in realistic speaking situations.

The software has two purposes: to serve as a learning tool for students and as a “peephole” for instructors. That is, Cerezo has used the program to research which teaching styles work best for different areas of grammar.

“What I really love about my research is that I can later incorporate it in the classroom,” says Cerezo. “You’re walking on solid ground because you know which teaching strategy is going to work best. You have the data to support that.”

Talking To Avatars offers the student the opportunity to interact with filmed native Spanish speakers in two different scenarios. In one, a real estate agent assists the student in finding an apartment. In another, the user approaches a police officer about a robbery. For both scenarios, the video “avatars” ask the user questions, such as ‘how many rooms would you like in your apartment?’. The user is then given the opportunity to respond, and if they makes grammatical mistakes, the avatars give feedback in different ways.

Cerezo wrote most of the software and shot the hundreds of video recordings needed to simulate the conversations as part of his Spanish linguistics dissertation. His study involved 230 intermediate Spanish students from Georgetown University.

First, Cerezo tested the students before they used the program to access their language skills, and then he incorporated several variables to see which methods worked best for different areas of grammar. One control group received neither positive nor negative feedback when they used the program. The other groups received one of four types of corrective feedback. Some received instruction on grammar, while others were simply informed that they had made a mistake in conjugating a verb or placing a preposition. Some were asked to fix their errors, while others were not.

Additionally, Cerezo investigated a second variable: active or passive participation. Some students used the program, while others observed their interactions.

Cerezo found that no hard and fast rule exists for language acquisition; different styles of feedback, such as explicit or implicit, work for different grammatical structures, as do different styles of student interaction, such as passive or active. This has implications for both the language classroom and the development of interactive pedagogical materials: in helping determine when computers have an edge over non-interactive audiovisual media and in lighting the way towards hybrid learning models.

“I’d like to incorporate my software as a teaching tool here at American,” says Cerezo. “We are trying to make our curriculum more and more hybrid so that students can practice with computer tools and come to class better prepared to interact with each other. This is what it’s all about: interaction in real-life situations.”