EVER SINCE a small but groundbreaking study in 1995, it’s been accepted wisdom that a child’s academic success is directly related to the amount of talk the child hears from adults in the first few years of life. Children in higher-income families hear more language than those in lower-income families; this disparity, the theory runs, leads to a “word gap” that puts poorer children at a disadvantage when they enter school.
Now, the city of Providence is set to put this theory to the test through new high-tech means, in the much larger setting of a city population—and then try to narrow the word gap for children in real time. Earlier this month, the city won the $5 million grand prize in Bloomberg Philanthropies’ Mayors Challenge, based on its proposal for a project called Providence Talks. The plan is to equip low-income children with recording devices that calculate how many words they hear, and then coach parents on how to boost their children’s language exposure.
Through Providence Talks, researchers and policy makers are likely to learn much more about whether pulling this language lever can really help level the academic playing field. At the same time, however, by asking scores of regular parents to opt into massive, data-driven recording and analysis of all the language their children hear in their first few years, and then encouraging them to change the personal matter of how they talk to their kids as a result, they are launching a project of unprecedented scope and audacity—one that opens up fascinating questions about language, social engineering, privacy, and parenting.
For Providence, the needs are pressing: as Mayor Angel Taveras and schools Superintendent Susan Lusi explained in a video pitch for the program, only one in three children in the city enter school at the appropriate literacy benchmarks, and closing the word gap in disadvantaged families is seen as an efficient, early way to change that. Taveras told me that early intervention appeals to him as a graduate of the Head Start pre-kindergarten program. (He’s also closely watching the linguistic progress of his own 14-month-old daughter.) “I really wanted to do something on early childhood education because of what I went through,” he said. “I felt this was a big issue we could attack on a citywide level.”
Certainly, a whole metropolis has yet to take on the word gap as an engine for major change. The notion itself only hit the public consciousness in 1995, when University of Kansas child psychologists Betty Hart and Todd Risley released a study based on a decade of painstaking research. Over a period covering the first three years of children’s lives, the team recorded 42 families for an hour every month, capturing every word spoken between parents and children. Then they created and analyzed 30,000 pages of transcripts and followed up on the children’s academic progress.
Hart and Risley found a significant disparity when comparing the vocabulary exposure of six families on welfare to 13 “professional class” families: children in the former group heard 616 words per hour, while children in the latter heard 2,153 words per hour. Extrapolating those results to 14-hour days, they estimated that underprivileged children were hearing about 30 million fewer words through the age of 3 than their upper-income counterparts.
The “30 million word gap” became a central talking point in early childhood education, despite the small sample size in Hart and Risley’s study (a limit that was inevitable, given the painstaking manual analysis required). But in 2006, technology arrived that made possible a new kind of research: LENA (short for Language ENvironmental Analysis), developed by researchers based in Boulder, Colo. With LENA, researchers record children for 12-hour periods by attaching small devices to their clothes. Once all the data is collected, algorithmic methods make it possible to come up with reliable word counts while screening out background noise, including television, and non-word sounds.
The LENA Natural Language Study, published in 2008, crunched through 32,000 hours of data from 329 participants and found that the Hart and Risley findings held up, though they found that the earlier research had overestimated the number of words used in a rich language environment (LENA found 20,000 words a day, as opposed to the previous estimate of 30,000). LENA researchers also found that the more parents talked directly to their children, the better the children performed in language assessment tests. The next step was intervention. A pilot study found that monthly meetings with parents could boost children’s levels of word exposure by 55 percent.
Providence, which is using the
LENA technology to power their program, is banking on similar improvements. The voluntary program will start with a small number of participating families, solicited as early as when the kids are born. The city anticipates gradually expanding to thousands of families by 2018, with the goal of eventually offering it to all qualified families in the city. Brown University researchers will serve as independent evaluators; though the details have yet to be worked out, Taveras expects that the evaluation will compare enrolled and unenrolled families as a gauge of effectiveness. Participating parents will be coached on how to improve language exposure and vocabulary development for their kids by home visitors, who already help guide low-income parents in other areas. Providence’s grant proposal also includes privacy safeguards for the families: once recordings have been run through the software analysis, they will be deleted.
Most language researchers I talked to seem optimistic about this policy initiative—for one, Deb Roy of MIT’s Media Lab, who observed his own child’s language development in a home-recording study called the Human Speechome Project. “For a topic as important as early language development (and later literacy), it is surprising how little basic data we have of what actually happens in home environments,” Roy said in an e-mail.
Roy thinks that providing feedback to parents about their children’s language development is “marvelous.” “As many parents know, life with a young child is demanding and the experience can be a blur,” he said. “Having a simple way to look back and understand the language experience of your child, and understand what that might mean for his/her future seems to me a potential big win.” LENA’s own research has found that parents have difficulty judging how much they talk to their children, so hard data ends up being valuable.
While the scholarly community has embraced LENA’s approach, meanwhile, some linguists do wonder how much of a difference Providence Talks can make. Kyle Gorman, a researcher at Oregon Health and Science University’s Center for Spoken Language Understanding, thinks the program could turn out to be “a boondoggle.” “There is reason to doubt that the intervention will be effective either in its proximate goal (increasing language attainment) or the tacit social goal of leveling the socioeconomic playing field for children,” he wrote in an e-mail.
Taveras, for his part, is confident that the project will prove to be a boon. “We believe there’s going to be an impact, and we believe the data will show that,” he said. He also stressed that early intervention would be just a first step: “It will help level the first day of kindergarten, but then we need to make sure that the kids are in school, that they’re learning, and that they realize learning is year-round.”
But if narrowing the vocabulary gap with new technology really does pay off for the children of Providence, then this might be the start of wider efforts around the word gap: as New York’s Michael Bloomberg said when awarding the Mayors Challenge prize, the hope is that the program will “take root locally and spread across the nation.” If so, we might all get used to a peculiar sight: toddlers rigged with
tiny recorders that are silently counting the words around them.
Ben Zimmer is the executive producer of VisualThesaurus.com and
Vocabulary.com. He can be reached at benzimmer.com/contact.