An Indiana State researcher and her colleagues are changing the way educators think about poverty’s impact on children’s exposure to language.
Recently published research by an Indiana State professor is dispelling the often-cited stereotype of a “30-million-word gap” between children from middle-class and low-income homes.
Linda Sperry, department chair of communication disorders and counseling, school, and educational psychology at State, is a co-author of findings published in the academic journal Child Development. Fellow authors include her husband, Douglas Sperry, associate professor of social and behavioral sciences at Saint Mary-of-the-Woods College, and Peggy Miller, professor emerita of University of Illinois.
Their research finds no statistical difference between children raised in low-income families and those children raised in middle- or upper-income families.
The “30-million-word gap” dates back to 2003, when researchers Betty Hart and Todd Risley reported a “Welfare” (low-income) group of children heard 616 words per hour, while a “Professional” (upper-middle class) group of children heard 2,153 per hour. As the name suggests, the gap widens exponentially as the children age.
Sperry’s research, however, found its two lowest-income groups, named “South Baltimore” and “Black Belt,” spoke an average of 1,061 and 1,838 words per hour to their children, respectively. And when adding in extended families who also talk to Black Belt children, they hear an “astounding” 3,203 words per hour, Sperry said.
“The ‘word gap’ disappears,” she said. “Our results suggest several important conclusions. First, and most significantly, not all poor families are the same. Whereas the majority of the Kansas Welfare families were single mothers living in a housing project, the Black Belt families were large and extended, with most of the children having extensive opportunities to talk with not only their mothers, but also with their fathers, siblings, other relatives and friends. Second, this result suggests that the extreme numbers of the Hart and Risley study were statistical outliers.”
The Sperrys, Miller and others conducted their research from 1975 until 1991.
“Data collection was ethnographic in method (participant observation) and longitudinal by design (sample of population studied at intervals),” Sperry said. “Data were specifically collected on underrepresented groups in the language development literature.”
The South Baltimore data were collected in an impoverished, urban, European-American community beginning when the child was about 18 months old and continuing until the child was approximately 3 years old. The Black Belt data were collected in an impoverished, rural, African-American community beginning when the child participant was 24 months to 42 months old. The Indiana data were collected in a working-class, rural, European-American community along the same timeline as the Black Belt data. Two data sets were collected in Chicago between when participant children were 30 to 48 months old; one of these communities was working class, and the other was middle class and was used as a comparison group.
“During the consent process, families were asked to help keep their child talking (as they normally would),” Sperry said. “The camera was set up in the room where the child was, and the researcher followed the child wherever he or she went over the course of the visit, except to the bathroom. If she went outside, the videographer followed her outside.”
Sperry said no restrictions were placed on the child or family’s behavior. Once the video was made, the second half hour was transcribed verbatim. The method of data collection for all five data sets was participant observation. Scientifically speaking, the child was “observed,” and the researcher acted as a “participant” in the process of the observation. Since the children were so young, the families typically participated as well.
“In this way, we made an effort to collect data that were ecologically and culturally valid,” Sperry said.
The vocabulary coding project became Douglas Sperry’s dissertation project when he was a doctoral student at the University of Illinois. Douglas Sperry identified every new word and every word spoken to and around the target child by primary caregivers, siblings, other family members and the researcher in the five data sets. All talk by the researcher was set aside and did not count in this study.
Douglas Sperry framed three situations: the number of words spoken to the target child by the primary caregiver; the number of words spoken to the target child by all other family members; and the number of words spoken in the presence of the target child (but not to him or her) by others. He counted words by the tens of thousands and then compared the five groups in his study with the four groups originally discussed by Hart and Risley in 1995.
When the Hart and Risley data began to be released, Sperry’s team “became quite convinced that the 30-million-word gap was a mischaracterization of the facts.” The biggest reason was the Hart-Risley data differed from the data already collected by Sperry’s team.
“Our study has generated an uncommonly intense reaction, both within the academy and within the policy world,” Sperry said in mid-June. “The journal Child Development solicited a commentary from five developmental psychologists who have interest in the home language environments of young children — Roberta Michnick Golinkoff, Erika Hoff, Meredith Rowe, Catherine Tamis-LeMonda and Kathy Hirsh-Pasek. The Brookings Institute is soliciting blog posts from various authors. NPR and Atlantic Monthly are both writing short articles, and a number of parenting blogs and education think tanks have also solicited interviews. Interest continues as the Chicago Tribune published an article just this week!”
In addition to Linda Sperry, Douglas Sperry and Miller, others who contributed to this study included: Judith Mintz, who collected data for the middle-class sample in Longwood, in Chicago; and Lisa Hoogstra, who collected data for the working-class sample in Daly Park, in Chicago.
Linda Sperry, Douglas Sperry, Mintz and Hoogstra were all dissertation students of Peggy Miller at different times. In addition, college students — too many to mention — at Indiana State, Catholic University of America, Columbia University, University of Illinois and University of Chicago contributed to transcription of these five data sets.
“Douglas Sperry did final cleaning of the majority of transcripts — a task that took years of effort — and all of the data reduction and coding of the vocabulary words (both unique types and tokens) for this particular study, as well as all of the analyses associated with the SSM study,” Sperry said.