Learning to see how white-matter pathways develop before speech emerges
Before the first word
For parents, a baby’s first spoken word seems to arrive suddenly—almost like a gift. Yet speech development begins long before that moment.
A newborn’s first cry is not speech, but it already brings breathing, movement, and sound together—and draws a caregiver’s response. In the months that follow, infants vocalize and gesture, listen to voices, watch faces and mouth movements, and gradually gain control over the movements needed to make sounds—from the visible movements of the lips to the less visible movements of the palate. At the same time, they learn that their actions can change what happens around them.
Early action, sensory experience, and social interaction together help create the context in which communication develops. But what is changing inside the brain during this preparation remains difficult to observe. Our work thus unfolded through three questions, each building on the last. Could we make the relevant pathways visible and measurable? Once we could, what did their development mean for emerging speech? And would the same framework remain useful when early auditory experience took a different path? Each answer made the next question possible.
First, we had to learn how to see
White matter is often described as the brain’s communication infrastructure. It contains bundles of nerve fibers that connect distant regions, allowing areas involved in movement, hearing, vision, and language to exchange information. Diffusion MRI offers a non-invasive way to study how these pathways develop.
Yet infant imaging is unusually difficult. Infants cannot simply be asked to keep still in a scanner, and their brains change rapidly across the first two years of life. Differences across scans or brain regions can reflect both technical artifacts and real developmental change. Some of the small auditory pathways we cared about were especially difficult to identify reliably. Behavioral assessments posed a parallel challenge: each offered only a brief window into abilities that were changing from month to month.
Our first challenge was therefore methodological. To reliably identify and measure these pathways, we developed methods tailored to infant brains, initially focusing on auditory pathways and later extending to motor, visual, ventral-language, and dorsal-language pathways. Using longitudinal data from the Baby Connectome Project, we charted the development of 33 tracts from 2 weeks to 24 months across 5 systems, and related these measurements to 11 assessments spanning 6 developmental domains.
The first word is a team effort
Classical language pathways, a longstanding focus of the field, were a natural place to begin. But once tracts across five systems became measurable, the question became broader: How did tracts across multiple systems relate to the early skills that precede speech?
At two weeks of age, several motor and auditory tracts showed higher values on our diffusion MRI measure (NDI), whereas several long-range language-related tracts showed lower values. These lower-starting tracts subsequently showed the steepest age-related increases. By 24 months, differences in the cross-system profile had narrowed, although individual tracts remained distinct. We described this pattern as progressive cross-system rebalancing: tracts began from different states and changed at different rates, while the overall profile became more balanced with age.
Their links with behavior were also distributed. Motor abilities showed the clearest positive association pattern. Gesture-related communication looked partly motor-like, whereas vocabulary showed a different, predominantly negative pattern. This was not evidence that higher NDI harmed language. Rather, it suggested that brain microstructure, early skills, and vocabulary growth were not all changing on the same clock.
Exploratory follow-up analyses suggested two possible time-ordered patterns. Early motor abilities were associated with later social development and communication, while early social experience was associated with later motor development and communication. These analyses cannot establish cause and effect, but they are consistent with a potentially reciprocal developmental process: action may create opportunities for social engagement, while social engagement creates new opportunities for action.
The surprise was not simply that many pathways were involved. It was that no
pathway—or behavior—told the whole developmental story on its own.
Reaching this interpretation took time. Reviewers pushed us to separate the strongest evidence from the more exploratory patterns and to ask what the distributed findings meant, rather than simply how many associations were statistically significant. Professor Janet F. Werker helped us describe infant perception as an active process of reorganization, rather than simply a loss of sensitivity. Professor Angela D. Friederici helped us distinguish an early-developing motor-related branch of the dorsal-language pathway from a later-maturing language-related branch. These perspectives helped us move from ranking isolated findings to understanding a developing system. In that sense, the manuscript itself also became a team effort across disciplines.
Beyond the scan
What does this mean outside an MRI scanner? Our study cannot tell us whether everyday experiences cause particular changes in white-matter pathways, but its distributed associations are consistent with a broader multisystem view of development. Infants learn while acting, listening, watching, and engaging with other people, gradually discovering how actions, sounds, sights, and social responses belong together.
Speech acquisition makes this especially clear. Although a recording can provide useful exposure to sound, a responsive exchange adds what happens next: an infant attempts a movement or vocalization, watches a face, receives a response, and tries again. Through this back-and-forth, action, perception, and social feedback continually shape one another.
Our findings offer a complementary view. Tracts associated with movement, hearing, and language developed on different timelines, yet became increasingly coordinated over the same period. Rather than pointing to a single language pathway, our findings suggest that the first words emerge as a developing infant gradually learns to coordinate multiple systems.
Where the framework lead next
Once speech acquisition came into system-view, a third question followed: what happens when one of the system’s major inputs—hearing—is altered early in life?
We are now extending the framework to infants and children with hearing loss, including those assessed before and after cochlear or auditory brainstem implantation. These cohorts offer a rare opportunity to study how altered auditory experience, developing neural pathways, intervention timing, and later communication outcomes interact over time.
Implants can provide new access to sound, but the system-view can change the clinical question. The issue is not only whether auditory input can be provided, but also how a developing infant learns to integrate and use that sound-input with action, vision, and social interaction. This process unfolds over time and cannot be captured by a single scan. Our future longitudinal work will therefore follow infants and children before and after intervention to ask whether differences in brain and behavioral development help explain their later developmental trajectories. This is a hypothesis generated by the present study, not a conclusion that has already been established. Ultimately, understanding these trajectories may support more individualized assessment and follow-up.
A baby’s first spoken word may sound like the beginning of communication. For us, it is the recognizable milestone of a process already unfolding through changes in the developing brain, the infant’s own actions, and the earliest exchanges with others.
The first spoken word is not where communication begins. It is where a much longer developmental process takes the form we recognize as speech.