As a child, Umashanthi Pavalanathan had a morning routine.
She and the other four members of her immediate family would wake up and get ready. As a group, they would walk out of the house in the direction of her school, not knowing whether all five would meet again at the end of the day.
An uncertainty foreign to most growing up inside the United States, it was a way of life for Pavalanathan, who grew up during the height of a violent civil war in Sri Lanka that claimed the lives of more than 100,000 and displaced nearly one million.
“It just happened,” says Pavalanathan of the sporadic bombing that affected the northern area of Sri Lanka, where she lived. “It was normal. It was part of our life. We couldn’t ... [just] stay home and stop living our lives. We had to go.”
And, so, every day, Pavalanathan and her family went through that same routine, committed to pursuing one thing they felt could help them achieve a better future: an education.
That commitment put Pavalanathan on a path that has led her to Georgia Tech’s School of Interactive Computing, where she has contributed to important research in computational sociolinguistics and was recently named a Foley Scholars finalist.
Hope in the Dark
The violence began in 1983 and lasted until 2009. Pavalanathan was born in 1986 and didn’t move to the United States until 2011, meaning the vast majority of her life was spent living under these tenuous circumstances.
“It was scary,” she admits. “It was just the uncertainty. To see friends and family members, and you don’t know what’s going to happen.”
One thing she had to look forward to, though, was her education. It was a common point of emphasis among parents of Tamil families living in northern Sri Lanka: Vigorously pursue an education in order to advance to a university with the hope of preparing for a better future when the country regains peace and stability.
“Even with the hopelessness, there was some hope that someday things can change through education,” says Pavalanathan. “So, we had a goal. Even though there were bombings, and friends and neighbors were being killed, we had a goal.”
There were challenges. Beyond the obvious – the bombings and displacement – Pavalanathan also grew up without electricity until she was 12 years old. They had kerosene lamps they would use at night. The scarcity led to innovation, she says, as people would come up with novel ways to limit the amount of kerosene being used.
She was displaced for an extended period of time once in 1995, when she was 9 years old. A heavy attack forced her family and many others out of their homes for about six months. Living as a refugee within her own country, she attended school in the evenings to keep up with everyone else her age.
Along the way, she was introduced to computing.
Before high school, she saw her first computer at an exhibition at a university. There, she learned about the internet.
“I was excited about it,” she says. “That was something I enjoyed.”
When she was about 12 years old, her family had electricity for the first time. It wasn’t available for 24 hours a day, so when her family got its first computer two years later she was only able to use it at certain times of the day.
“The dial-up was faster at night, so I would stay up late and try to do as much as I could,” she said. “I enjoyed solving problems in that way. That’s when I knew I wanted to do something in computing.”
A Helpful Challenge
The story may have ended there, if not for the high school she attended in Sri Lanka. It was a missionary school that focused on more than just standard education.
There were extracurricular activities like sports and fine arts, things that pushed Pavalanathan to be more outgoing.
“I think I realized that [as an undergraduate] when I met students from other top schools in my hometown known for good grades that there was a difference in the way I was brought up,” says Pavalanathan. “Many could do well in exams, but they couldn’t present themselves or go up and speak to people. But I think my school was very influential in giving us the challenge in those areas.”
They were vital skills she claims helped her when she made the big decision to come to the United States following her undergraduate studies. Taking that course was considered very much outside the norm in her family.
“I asked my dad not to tell that to the relatives, because they were brought up in a strict, tight-knit environment,” she says. “It was something that wasn’t really accepted by everyone.”
However, when she made it to the United States, first as a visiting scholar at Indiana University and then as a Ph.D. student at Georgia Tech, she found that things came naturally.
She had a number of Ph.D. offers from other schools but says she chose Georgia Tech because of the welcoming and diverse environment in the School of Interactive Computing. Also key to her decision was the relationship she established with her advisor, Jacob Eisenstein, and the research she was able to pursue.
Pursuing Impactful Research
Pavalanathan’s research is focused on the field of computational sociolinguistics, a fusion between computer science, social computing, and natural language processing that studies the relationship between language and society in a computational way.
Sociolinguists have long studied the impact context has on the development of language, but only recently have they had large online social systems like Facebook, Twitter, and others to observe natural communication on a large scale.
Pavalananthan is interested in studying why and how people say the things they do in a given context.
“When we speak, we have non-verbal cues,” says Pavalanathan. “Those don’t exist in writing, so we are trying to invent new ways.
“Say that they’re happy or sad or want to argue. Sometimes they’ll use all capitals or punctuation. On Twitter, you’ll see repeating characters. That’s not just random. There is a reason people do this.”
Additionally, she examines how language changes with the audience. For example, when someone speaks directly to a peer on Twitter, they will speak in a different manner, likely more informal, than they would to a broader audience.
“We see that in face-to-face communication, as well,” she says.
One paper published last year looked at how the introductions of emojis on Twitter caused any changes in writing style from the more dated emoticons.
The ultimate goal of the research is to improve language tools to make them more aware of linguistic patterns in different social contexts.
“But we’re still a long way from that,” says Pavalanathan. “We are trying to understand the patterns of variation in online language, and this could potentially help us to improve language processing tools in the future.”
The work has gotten her recognized as a finalist in the 2017 Foley Scholars program. Winners will be announced at the GVU 25th Anniversary celebration on Oct. 18 at the Tech Square Research Building.
“It was a nice surprise to be selected as a finalist,” she says. “It really validates the work that we’ve done.”