Before Words - The Other Already Participates in the Architecture of Language

Before a child pronounces their first word, a great deal has already happened.

They have already heard thousands of voices. They have encountered rhythms, intensities, pauses, and repetitions. They have heard words in particular rooms, in front of certain objects, during meals, play, farewells, requests, movements, and emotional exchanges.

But perhaps there is an even more interesting question:

does the baby learn only which sounds repeat — or also where, when, with whom, and under what conditions those differences appear?

This question allows us to bring together three lines of research that, although they pursue different goals, seem to enter into an especially fertile dialogue: Marina Kalashnikova’s recent studies on infant-directed speech, Patricia Kuhl’s classic work on phonetic specialization, and the Human Speechome Project, initiated by Deb Roy at MIT.

Together, they suggest something important:

language does not emerge only from the repetition of sounds. It emerges from differences that repeat within encounters.

Before meaning, the signal is already being organized

In 2026, Marina Kalashnikova, Laura Fernandez-Merino, and Nicola Molinaro published the article Neural encoding of infant-directed speech during infants' first year of life in Developmental Cognitive Neuroscience.

The study investigated how the infant brain tracks speech addressed directly to the baby — infant-directed speech, or IDS — compared with speech directed toward adults.

The question matters because when we speak to babies, we do not simply produce the same words in a more affectionate tone. We change physical properties of the signal: rhythm, intonation, pitch, duration, and spectral distribution.

The results indicated more efficient neural encoding of infant-directed speech, suggesting that its acoustic properties facilitate the tracking of the signal during the first year of life.

But this still does not mean meaning.

The baby does not need to know what a word means for temporal, spectral, and phonetic differences to already be transduced.

This distinction is central:

a difference may begin to modify the Body before a consolidated lexical meaning exists.

But not every difference will continue to carry the same weight

This is where Patricia Kuhl and a long tradition of research on perceptual reorganization during the first year of life become essential.

During the first months, human infants can discriminate many phonetic contrasts that do not belong to the language they hear every day. As the first year progresses, this perceptual openness changes: discrimination of many non-native contrasts decreases, while specialization for regularities in the ambient language increases.

Classic studies place much of this reorganization between approximately six and twelve months.

This is sometimes summarized too loosely as:

“the baby stops hearing certain sounds.”

That is not quite accurate.

The physical difference may still be present.

The auditory system still receives acoustic energy.

But that difference may no longer occupy the same perceptual or categorical boundary.

Therefore:

what does not repeat in a functionally meaningful way within the linguistic territory may progressively lose relative weight in what is perceived as an important difference.

And the reverse also occurs.

What repeats in a statistically organized way may gain relevance.

Kuhl described this specialization within the idea of neural commitment to the patterns of the native language: experience progressively facilitates the processing of frequently encountered regularities, while making some non-native distinctions less readily accessible.

It is not simply loss.

It is specialization.

The Body-Territory learns which differences will continue to count as differences

This point is especially important for our Body-Territory proposal.

A child is not born knowing which acoustic boundaries will matter within the linguistic community in which they live.

They encounter distributions.

Some differences repeatedly appear associated with meaningful linguistic consequences.

Others are rarely or never used to distinguish words in that territory.

So the system does not merely learn:

“this sound exists.”

It begins to learn:

“this difference matters here.”

Perhaps learning a language means, in part, learning which differences will continue to count as differences.

But one dimension is still missing.

Because these sounds do not repeat in a neutral space.

Where does a word happen?

This is precisely the problem that made the Human Speechome Project so extraordinary.

Deb Roy and his team instrumented a residence with 11 cameras and 14 microphones, seeking to record, at an unprecedented scale, the everyday audiovisual experience of a child during the early years of language development. The goal was to create an extremely dense longitudinal corpus of what the child saw and heard at home.

The question was no longer only:

how many times was a certain word spoken?

It began to include:

where?

when?

during which activity?

in which sequence?

by whom?

In later studies derived from this corpus, Brandon Roy, Michael Frank, and Deb Roy analyzed the relationship between caregiver word use, activity contexts, and the age at which specific words appeared in the child’s own production.

They found something important:

frequency mattered, but context mattered too.

Words used more frequently tended to be learned earlier, but words concentrated within more consistent activity contexts also tended to appear earlier in development.

This substantially changes the simple idea of repetition.

We do not have only:

word → many repetitions → learning.

We have:

word + activity + space + objects + people + moment → situated recurrence.

Perhaps it is not the word that repeats

Imagine the word “water.”

It may appear in front of a glass.

When someone is thirsty.

During a meal.

While a faucet is being opened.

After play.

During a bath.

In every encounter, something is similar.

But the experience is never exactly the same.

This allows us to recover a formulation that has become central in BrainLatam:

No sign enters the same Body-Territory twice, because the previous encounter has already participated in its history.

The signal may repeat.

The object may appear to be the same.

The place may be similar.

But the one who encounters the word again has already been modified by previous encounters.

Therefore, repetition does not mean accumulating identical copies of the same data.

Repetition produces history.

And history changes interpretation.

Context does not come after meaning

There is a common image in which we first receive a word and then use context to discover what it means.

The Speechome Project allows us to imagine something deeper.

Perhaps context is not merely a clue added later.

Perhaps it participates from the beginning in the formation of what that word may come to mean.

Sound.

Place.

Object.

Movement.

Person.

Bodily need.

Consequence.

Affect.

The word emerges within this network.

The results from Roy and colleagues do not demonstrate that all these dimensions are directly incorporated into a single neural representation. That would go beyond the evidence.

But they do show that the contextual structure of everyday experience adds explanatory power beyond simple word frequency.

And this connects directly with our formulation:

data are not meaning.

The acoustic sequence alone does not carry everything it will come to mean for that Body-Territory.

When the other also changes

There is still another fascinating dimension.

A child’s linguistic environment is not static.

Adults change the way they speak to the child.

The infant-directed speech studied by Kalashnikova already demonstrates this: the other reorganizes their own signal in the presence of that Body.

The experiment by Kuhl, Tsao, and Liu adds another piece.

Nine-month-old American infants were exposed to Mandarin during 12 sessions with native speakers. After this relatively brief social exposure, they showed improved discrimination of Mandarin phonetic contrasts. When equivalent material was presented through audiovisual recordings or audio only, the same effect did not appear in that experimental paradigm.

This result became famous for suggesting that social interaction may function as a kind of gateway for phonetic learning.

But from a Body-Territory perspective, we can ask an even larger question.

When the object of the encounter is another subject, both sides can modify the next encounter.

The baby responds.

The adult changes voice, rhythm, expression, or repetition.

The new signal meets a baby who is no longer exactly the same.

We then have:

baby ⇄ caregiver ⇄ signal ⇄ context ⇄ new encounter.

Language begins to look less like a transfer and more like a coordination.

5D Consciousness: Attention does not create the difference

This discussion also requires us to preserve an important principle of 5D Consciousness.

A difference does not need to be under focal attention in order to begin participating in the configuration of the Body-Territory.

Attention may amplify it.

It may increase its priority.

It may favor learning, response, or explicit availability.

But it does not need to create its existence.

Likewise, when a non-native contrast loses perceptual weight over development, we do not need to conclude that its physical difference has stopped reaching the organism.

It is more precise to say:

the history of that Body has reorganized the relative weight of that difference.

This connects Kuhl, Kalashnikova, and Roy.

Kalashnikova asks how certain signals are tracked neurally.

Kuhl shows that experience reorganizes which differences receive perceptual relevance.

Roy shows that this experience is distributed across places, activities, and relationships.

Perhaps learning a word is learning a trajectory

So perhaps the child is not simply storing words.

They are encountering regularities among differences.

Certain sounds appear with certain objects.

Certain words appear during particular actions.

Certain places concentrate particular experiences.

Certain people change how they speak.

Some phonetic differences continue to produce useful distinctions.

Others lose priority.

And every encounter modifies the one who will encounter the next.

For this reason, perhaps the sequence is not simply linear:

sound → word → meaning.

Perhaps it is closer to:

encounter → difference → recurrence → expectation → category → transformation → new encounter.

And it is within this movement that a word may be born.

Not necessarily at the moment when the child finally manages to pronounce it.

That may simply be the moment when we are finally able to hear it being born.

Before that, hundreds of encounters have already taken place.

Some sounds have repeated.

Some differences have gained weight.

Others have lost priority.

Certain places have made certain relationships more probable.

The other has modified their own signal.

The child has been modified by that signal.

And the next word has encountered a Body-Territory already carrying the previous encounters.

Perhaps learning a language, then, is much more than learning the meanings of words.

Perhaps it means learning:

which differences matter, in which territories they matter, and how each new encounter can change what they will mean the next time.

References

Kalashnikova, M., Fernandez-Merino, L., & Molinaro, N. (2026). Neural encoding of infant-directed speech during infants' first year of life. Developmental Cognitive Neuroscience, 81, 101816. https://doi.org/10.1016/j.dcn.2026.101816
Investigates how the infant brain tracks infant-directed speech, showing encoding advantages that help explain how the form of the social signal participates in early linguistic experience.

Kuhl, P. K., Tsao, F.-M., & Liu, H.-M. (2003). Foreign-language experience in infancy: Effects of short-term exposure and social interaction on phonetic learning. Proceedings of the National Academy of Sciences, 100(15), 9096–9101. https://doi.org/10.1073/pnas.1532872100
Shows that brief social exposure to Mandarin can alter phonetic discrimination in American infants and that, in that experiment, equivalent exposure through recordings did not produce the same effect.

Werker, J. F., & Tees, R. C. (1984). Cross-language speech perception: Evidence for perceptual reorganization during the first year of life. Infant Behavior and Development, 7(1), 49–63. https://doi.org/10.1016/S0163-6383(84)80022-3
A classic reference on perceptual reorganization during the first year, showing declining discrimination of certain non-native contrasts as a function of linguistic experience.

Roy, D., Patel, R., DeCamp, P., et al. (2006). The Human Speechome Project. Proceedings of the Cognitive Science Society.
Introduces the unprecedented longitudinal observation project that instrumented a home in order to densely record the audiovisual environment in which a child’s language developed.

Roy, B. C., Frank, M. C., & Roy, D. (2012). Relating Activity Contexts to Early Word Learning in Dense Longitudinal Data. Proceedings of the 34th Annual Conference of the Cognitive Science Society.
Shows that both frequency of use and contextual consistency predict the age of word acquisition, indicating that language learning depends not only on how many times a word appears, but also on how it is distributed across everyday activities.

Roy, B. C., Frank, M. C., DeCamp, P., Miller, M., & Roy, D. (2015). Predicting the birth of a spoken word. Proceedings of the National Academy of Sciences, 112(41), 12663–12668. https://doi.org/10.1073/pnas.1419773112
Uses the ultradense Speechome Project corpus to show that words encountered in more distinctive spatial, temporal, and linguistic contexts tend to appear earlier in the child’s own speech, reinforcing the importance of multimodal context in lexical learning.