Jodi Tommerdahl, an associate professor of curriculum and instruction in the College of Education and Health Professions, and Cynthia Kilpatrick, an assistant professor of linguistics in the College of Liberal Arts, found that language concerns among children could be reliably diagnosed by analyzing samples about 100 spoken words at a time rather than evaluating longer samples that take more time to analyze. The team argues that the shorter samples could speed up the diagnosis process and enable clinicians help children earlier in life.
“This is key because at the moment, language impairment is often undiagnosed. And when it is diagnosed, it’s diagnosed later than you’d want it to be,” Tommerdahl said.
She added: “If we can find meaningful differences in typical and atypical language in children at age 2 instead of at age 6, we’re just that much closer to developing more appropriate treatments for those kids and getting them back on track both in the linguistic sense, but also educationally.” The new findings are published in this month’s journal Language Testing.
For their study, the pair used a test-retest procedure to look systematically at the morphology and syntax ability in 23 typically developing children between ages 2 and 3. They set up toys, hidden cameras and microphones in a flexible, large room at the University of Birmingham, U.K., which provided some funding for the study. As each child and parent entered the room, their interactions were recorded for 40-minute intervals over several days within a one-week period.
“We had very good quality in a natural play setting because most of what we know about child language comes from spontaneous samples or observing children communicate naturally,” Tommerdahl said. “Most tests involve elicited language or children being asked to do something or to respond to specific stimuli.”
Some researchers record in a clinic with a professional interacting with the child. The team says that using spontaneous samples, however, the children are not in a test situation, not nervous, and researchers are able to see real-life language that children use on a daily basis.
Kilpatrick said their research shows that 200 utterances is not better than 100 utterances in terms of reliability within a spontaneous language sample, so 100 utterances might be the ultimate sample size.
“Clinicians, linguists and people who work with language samples spend a lot of time recording language samples, and you want the shortest sample possible to give you the best results or most reliable result that you can get,” Kilpatrick said.
The team’s next step is to acquire funding to either develop or contribute to an enormous database of child language that would be freely available to researchers who want to use it.
Tommerdahl and Kilpatrick’s new paper is called “The Reliability of Morphological Analyses in Language Samples.” It is also available via podcast at the Language Testing website.
About UT Arlington
The University of Texas at Arlington is a comprehensive research institution and the second largest institution in The University of Texas System. Total research expenditures reached almost $78 million last year. UT Arlington ranks fifth in the nation for undergraduate diversity and was ranked as the seventh fastest-growing public research university by The Chronicle of Higher Education in 2013. Visit www.uta.edu to learn more.
The University of Texas at Arlington is an Equal Opportunity and Affirmative Action employer.