Summary: A baby’s first words are, most likely, based on their visual experiences, a new study reports.
Source: Indiana University.
Study reveals that visual memory’s role in early language learning may advance treatments for delayed speech, autism.
Indiana University psychologists have shown that a baby’s most likely first words are based upon their visual experience, laying the foundation for a new theory of infant language learning.
The findings also suggest new possibilities for the treatment of children with language deficits and autism.
Drawing on theories of statistical learning, IU researcher Linda Smith and colleagues found that the number of times an object enters an infant’s field of vision “tips the scales” in favor of associating certain words with certain objects.
The research appears in the journal of the Royal Society Philosophical Transactions B in a special issue on statistical learning.
“We think that children’s first words are predictable based on their visual experience with objects and the prevalence of those objects in their visual world,” said Smith, a professor in the IU Bloomington College of Arts and Sciences’ Department of Psychological and Brain Sciences and senior author on the study.
“Visual memory may be the initial key to getting words stuck on objects — familiar visual objects like table, shirt, bottle or spoon,” she added. “It’s an aggregated experience; those very first words may be learned — slowly and incrementally — for a few visually pervasive objects. This may be how infants begin to break into language before their first birthday.”
The study’s results could also help inform interventions for children with delayed speech and other language disorders.
“Difficulty learning words could stem from visual processing problems,” Smith added. “Children who are late talkers have slow or age-delayed visual processing skills for objects, for example. Children with autism have object-processing problems as well.”
Although many researchers have studied infants’ first words to understand learning, Smith said none have approached the question from the visual side.
“While studying language acquisition from the ‘word side’ may benefit those studying later stages of language learning — at the ages of 18 months to 3 years — it cannot account for how children break into language,” she said.
Under the new theory, which Smith and colleagues call the Pervasiveness Hypothesis, a few highly prevalent objects stand out to infants among the “clutter” of other less frequent objects to become their first words.
To conduct their study, IU researchers looked at videos that showed the visual field of eight children — five girls and three boys — between 8 and 10 months old, the period before children engage in verbal interactions with parents and caregivers.
The videos came from head-mounted cameras worn by the children an average of 4.4 hours. Caregivers were told the cameras would observe children’s daily activities, not words or objects specifically. Caregivers could choose when to activate the camera.
For the purpose of the study, researchers observed mealtime scenes, defined as any eating by anyone at any time or location — in cars, at playtime or in a high chair, for example. The recordings yielded 917,207 mealtime frames, with one image sampled every five seconds. Five objects were recorded for each frame: a total of 745 objects.
Using an accepted method to index child vocabulary, the researchers then divided the named objects into “first nouns,” which are acquired by half of all 16-month-olds; “early nouns,” which are known by half of all 30-month-olds; and “late nouns,” which are acquired at later stages of learning.
First nouns include words such as table, shirt, chair, bowl, cup, bottle, food, spoon and plate.
The study’s results revealed a strong correlation between the most frequently appearing objects and “first nouns,” with the top 15 of these words appearing in the images collected by the study.
“The comparison of first and early nouns was particularly striking, since both sets of object names are acquired quite early in childhood and refer to objects common in households with infants,” said Elizabeth Clerkin, a Ph.D. student in the IU Bloomington Department of Psychological and Brain Sciences and first author on the study.
“That infants’ visual environment during mealtime consistently involves a very small number of objects — and the names of these high-frequency objects are among those normally learned first by infants — suggests visual experience is doing the heavy lifting in very early word learning,” she added.
Whether children who experience speech disorders are not picking up visual regularities in the environment or simply live in households with fewer regularities, Smith said it’s vital to explore the role of both words and vision in language learning.
“Taking account of the visual brings a whole new dimension of word-learning into view,” she added. “If all you ever worry about is the word side of word-learning, you may be missing half the problem: visual cues that aid language learning.”
In addition to Smith and Clerkin, co-authors on the study are professor Chen Yu and laboratory technician Elizabeth Hart of the IU Bloomington Department of Psychological and Brain Sciences and Georgia Institute of Technology professor James M. Rehg.
Funding: This research is supported in part by the National Science Foundation. The study grew out of a larger NSF grant to IU to create a collection of over 500 million images to track the visual regularities in the lives of children from birth through age 24 months.
Source: Kevin D. Fryling – Indiana University
Image Source: NeuroscienceNews.com image is credited to Indiana University.
Original Research: Abstract for “Real-world visual statistics and infants’ first-learned object names” by Elizabeth M. Clerkin, Elizabeth Hart, James M. Rehg, Chen Yu, and Linda B. Smith in Philosophical Transactions of the Royal Society B. Published online November 21 2016 doi:10.1098/rstb.2016.0055
Real-world visual statistics and infants’ first-learned object names
We offer a new solution to the unsolved problem of how infants break into word learning based on the visual statistics of everyday infant-perspective scenes. Images from head camera video captured by 8 1/2 to 10 1/2 month-old infants at 147 at-home mealtime events were analysed for the objects in view. The images were found to be highly cluttered with many different objects in view. However, the frequency distribution of object categories was extremely right skewed such that a very small set of objects was pervasively present—a fact that may substantially reduce the problem of referential ambiguity. The statistical structure of objects in these infant egocentric scenes differs markedly from that in the training sets used in computational models and in experiments on statistical word-referent learning. Therefore, the results also indicate a need to re-examine current explanations of how infants break into word learning.
“Real-world visual statistics and infants’ first-learned object names” by Elizabeth M. Clerkin, Elizabeth Hart, James M. Rehg, Chen Yu, and Linda B. Smith in Philosophical Transactions of the Royal Society B. Published online November 21 2016 doi:10.1098/rstb.2016.0055