← all systems · see it in The System
Language acquisition
Language acquisition is the process by which humans come to perceive, comprehend, produce and use a language, spoken or signed, as linguistics and developmental psychology describe it: usually first-language acquisition by infants and young children (including bilingual first-language acquisition), set apart from second-language acquisition; it runs from fetal and newborn listening through phonetic tuning, babbling, word segmentation and first words to grammar, which children in a speech community converge on by about five, and it is shaped by the language a child hears or sees, live social interaction, sensitive periods and the maturing brain. How much of it rests on an innate, language-specific endowment (universal grammar) and how much on general learning from use is a standing dispute.
Properties published scienceour proposal
Sourced fields, and our own readings.
| born at | before birth, in each child |
| born precision | weeks (third trimester) |
| boundary | Inside: how a child (or a later learner) comes to perceive, understand and use a language: hearing or seeing the language around it from before birth, tuning to its sound (or sign) categories, babbling, segmenting words, building a vocabulary and a grammar, the brain's commitment to the language, the sensitive periods in which this goes best, bilingual and signed acquisition, late first-language acquisition after deprivation, and the theories that explain it (nativist universal grammar, usage-based, statistical learning, social interactionist, emergentist). Outside: the language itself (language, phonology, lexicon, grammar, sign language), the child as a whole (child development, human), the organ (human brain), the people who talk to the child (family), learning a later language as an adult (second language, walked), and reading and writing (literacy). |
| nested level | A part of child development (the language line among its capacities) running on the growing human; its output, a speaker of the community's language, is what the language and the speech community are carried by. |
| scale | organism |
Measures (26) published science
| phonetic narrowing window | 6-12 months of age |
| decline of nonnative r-l discrimination (Japanese infants) | 8-10 months of age |
| phonemes per language | about 40 distinct phonetic categories |
| native-language discrimination in newborns | 4 days of age (Mehler et al. 1988) |
| neonates tested for prenatal vowel learning | 80 (40 Sweden, 40 USA), mean 33 h after birth infants |
| onset of babbling, hearing vs deaf infants | about 6 vs about 11 months of age |
| canonical babbling onset window | 5-10 months of age |
| vocal imitation first elicited | 20 weeks of age |
| statistical word segmentation exposure | 2 minutes of a nonsense-syllable stream (8-month-olds) |
| vocabulary burst | 18 months of age |
| Mandarin syllables heard in live-tutor study | about 33,000 from 4 talkers over 4-5 weeks syllables |
| syntactic learning flourishes | 18-36 months of age |
| new word meanings learned per day | 10-15 meanings per day (average) |
| age by which children converge on the community grammar | about 5 years |
| age at which grammar-learning ability begins to decline | 17.4 years (best-fitting model, R2 = 0.89) |
| age at which native speakers reach grammar asymptote | around 30 years |
| child vocalizations added per 100 adult vocalizations per hour | 27 (growing by 16 per year of age) child vocalizations per hour |
| child vocalization rate growth with age | 66 more vocalizations per hour per year of life |
| child-directed speech to under-4s (Tsimane) | < 1 minute per daylight hour |
| variance in outcomes 10 years later explained by conversational turns at 18-24 months | 14-27 % of variance (IQ, verbal comprehension, vocabulary), SES controlled |
| gain in comprehension growth per year younger at cochlear implantation | 1.1 (95% CI 0.5-1.7) points per year of language growth |
| deaf or hard-of-hearing births (United States) | 2-3 per 1,000 children |
| words heard before fluent use (children) | 10s to 100s of millions words |
| language-model pretraining data | hundreds of billions to trillions words |
| word-learning trajectories estimated | about 400 words in each of 10 languages, from 32,000+ children |
| bilingual sample age range | 1;10 to 2;6 years;months |
Open contentions (13) open contention
Where sources or readings disagree. Both stay on the map.
- C-261 Is grammar learned from other speakers, or shaped by an inborn universal grammar?
booked: The map follows the view that most grammar is learned from others, and leaves open how much is inborn.
the other reading: innate language-specific constraints (Chomsky, Fodor, Lenneberg; the language faculty) / language learned from use by general learning mechanisms (Bates, Tomasello, MacWhinney; statistical learning; chunking) / behaviourist conditioning (Skinner, relational frame theory) / social interaction with knowledgeable adults (Vygotsky, Bruner) / Bayesian learnability: an ideal learner with domain-general biases (Perfors, Tenenbaum and Regier 2011) / children's early combinations are productive, unlike Nim's (Yang 2013) / negative evidence exists in adult reformulations (Chouinard and Clark 2003) - C-413 Where do thoughts come from?
booked: No theory is proven, so the map lists the main ones side by side and doesn't pick one. It shows where thinking is handled in the brain, not where thoughts start.
the other reading: as in C-413 - C-1120 Does being bilingual sharpen the mind?
booked: The map treats it as contested: benefits reported at various ages, but no advantage in adults after correcting for publication bias.
the other reading: as in C-1120 (no executive-function advantage in adults after correction) - C-1123 Is there a critical age for learning a second language's grammar?
booked: The map follows the 2018 finding that learning ability holds to about 17.4 years and then declines, and keeps the other proposed age patterns alongside.
the other reading: as in C-1123 (an optimal period of 0-7 years; a continuous decline from infancy; never completely closing); the critical period hypothesis's own estimates for SLA range from 2 to 13 years - C-1126 Should deaf children with implants also learn sign language early?
booked: The map makes no ruling and keeps the different positions side by side, including those within the Deaf community.
the other reading: as in C-1126 - C-3482 What opens and closes the sensitive period for learning a first language?
booked: The map keeps the competing readings; none is ruled.
the other reading: superior neural plasticity / an earlier start (more years) / limits in cognitive processing / no interference from a first language / willingness to err / desire to conform / immersion in a community of native speakers (Hartshorne's list) / excitation-inhibition balance opening and molecular brakes closing (Werker and Hensch) / neural commitment to the first language (Kuhl) / delayed prefrontal maturation (critical period hypothesis article) / lateralisation (largely discredited, per the same article) / social gating (Kuhl) - C-3502 Is a child's grammar complete by about age five?
booked: The map keeps the competing readings; none is ruled.
the other reading: converged by about 5 / still developing into adolescence and early adulthood / adults differ in grammatical knowledge (Dabrowska 2012, the specialist's lead, not opened) - C-3514 Does how much adults talk to a child shape the child's language?
booked: The map keeps the competing readings; none is ruled.
the other reading: a large SES-linked gap in words heard drives language outcomes (Hart and Risley 1995, as Romeo 2018 and Sperry 2019 report it) / quantity of directed adult talk matters / conversational turns rather than sheer words matter (Romeo) / overheard and child-to-child speech count as input (Sperry; the Mayan and Tsimane studies cited by Cristia) / reply to Sperry et al.: the gap should not be abandoned; directed quality speech, not overheard speech, is the vital ingredient (Golinkoff, Hoff, Rowe, Tamis-LeMonda, Hirsh-Pasek 2019) - C-3519 Is language in the infant brain spread out at first, or on the left from the start?
booked: The map keeps the competing readings; none is ruled.
the other reading: spread then concentrate (src:wp:language-acquisition) / a left-biased network present from the first months, tuned and strengthened by experience (src:dehaene-2002) / dorsal and ventral pathways, with the dorsal path to Broca's area maturing late (specialist, unsourced here) - C-3540 Do children learning a word hold several possible meanings or one at a time?
booked: The map keeps the competing readings; none is ruled.
the other reading: cross-situational accumulation (many candidate meanings held) / propose-but-verify (one at a time) / both, by situation - C-3553 Does infants' statistical word-finding work on real speech?
booked: The map keeps the competing readings; none is ruled.
the other reading: transitional-probability segmentation as a core mechanism of word finding / a mechanism that needs other cues (prosody, known words) to work on natural, variable speech - C-3555 Do signing children start out ahead in vocabulary?
booked: The map keeps the competing readings; none is ruled.
the other reading: an early sign advantage (src:wp:language-acquisition) / broadly parallel milestones, with part of any early advantage coming from counting communicative gestures as signs (the specialist's reading; the counting point is unsourced here) - C-3567 Is there a vocabulary spurt at 18 months?
booked: The map keeps the competing readings; none is ruled.
the other reading: a vocabulary spurt or naming insight around 18 months / no discrete spurt in most children / acceleration as a by-product of parallel learning
Cross-scale patterns seen here (19) our proposal
Shapes that recur when the same kind of problem is solved at different scales, and how each shows in Language acquisition. Each links to every other system that shows it.
- same role, different builds (structural)
One job, learning a first language from the language around, met through different channels: by ear (speech) and by eye (sign), on a parallel timetable (manual and vocal babbling, early sign vocabulary). And by different learners: children reach fluency on 10s-100s of millions of words; language mod - symmetry breaking (name the cut) (structural)
Before 6 months infants distinguish the phonetic units of all languages (no preferred language); by 12 months perception is language specific. The cut is named: the distribution of sounds in the language heard. The broken direction sets what follows (words that fit the native scheme are learned fast - trade one property for another (structural)
Native for nonnative: neural commitment raises learning of native-compatible patterns and lowers nonnative perception; at 7.5 months better nonnative discrimination predicts slower later language, better native discrimination faster. Children spread along the trade. - Umwelt: each system carves its own slice of one shared medium (structural)
Each infant carves its own slice of the one shared acoustic medium: the same English r-l sounds are two categories to American infants and become one to Japanese infants between 8 and 10 months; native-vowel prototypes exist only for the language heard. - who opens the door (structural)
Social gating: the infant's phonetic learning door opens only for a live, interacting person; television and audio deliver the same sounds and nothing is learned. Fetal hearing, built before birth, is the port the first speech arrives on. Narrowed (Roseberry 2014): what opens the door is contingent - borrowed-then-held (partial)
The zone of proximal development: linguistic tasks a child cannot do alone at a given time but can with an able adult, later done alone; conversational scaffolding first borrowed from the caregiver. - nested ladder (partial)
Built up one step at a time, from phoneme distinctions to words to grammar; in the structure-building model lexical categories come before functional categories. - outside-to-inside (partial)
Speech perception always precedes production: the language's categories are outside first (in the community's speech) and are taken inside the child before it produces them. Caveat: perception precedes production in onset; production then feeds perception back (perception-action links by about 3 mon - same toolkit, independent builds (partial)
Vocal learning with a live tutor in a sensitive period, built separately in songbirds and humans: zebra finches need visual interaction with a tutor; white-crowned sparrows learn alien song only from a live tutor; a richer social environment extends the sensitive period in barn owls and sparrows. Me - seed (partial)
A grammar arrives and unpacks an unbounded space: a finite capacity, with recursion, yields an effectively infinite number of sentences; in Nicaragua a few hundred children with home signs unpacked a full grammar within a few cohorts. - templating (partial)
One conserved learning module reused across domains and species: statistical learning works on speech, music and visual patterns in the same way, and in nonhuman animals; and the community's grammar is copied into every child by about five. - The carry (partial)
The wait: the infant's task in the first year is to work out the composition of the 40-odd phonemic categories before acquiring the words that depend on them; words wait for categories, syntax (18-36 months) for words. - The Payment to Entropy (partial)
A signal handed over without the social gate arrives unusable: the same Mandarin material on television or audio taught nothing, while live sessions did. Form: signal (the loss is in the hand-off's medium, not in the sounds). No per-hand-off efficiency is sourced beyond all-or-none. - the whole sets the local reference (partial)
The speech community fixes the reference each child's perception is measured against: 6-month-olds respond specially to vowels at the distributional mean of adult speakers' productions; every child converges on very much the same grammar by about five. - conservation law plus closure (surface)
Poverty of the stimulus: the input is compatible with infinitely many grammars and needs a second source to close it (innate constraints, by the nativist reading; social and statistical biases by others). No balance law is involved, so surface. - lightning (surface)
Vocabulary 'explodes' at 18 months, but whether a literal word spurt exists is debated; read as a stored difference released, it fails the firing (no discharge shown). - namespace completeness (the map predicts its gaps) (surface)
A child's productive rule fills slots it never heard ('gived', 'runned'), and the poverty-of-stimulus argument says children know grammar not in their input: a structure predicting members, though here the predicted members are often wrong. - same shape, opposite sign (surface)
One process (neural commitment) with opposite effects on two classes of input: native patterns up, nonnative down. Booked surface: it is one mechanism's two outputs at the same time, not the mechanism found again with its sign reversed (sharpened dark twin, 2026-11-09). - smooth to sharp in finite time (surface)
Hartshorne et al. report a 'sharply-defined critical period': learning rate flat to 17.4 years, then a sigmoid decline. Steep but never singular, so surface (the dark twin).
Valence: what it trades, and with whom (12) published science
Each exchange with a neighbour, read from both sides: who gains, who pays, and what carries it.
- Cognition via sys:mental-representation · Language acquisition gain
- Human (Homo sapiens) via sys:sensory-environment · Language acquisition gain
- Language (human language) via sys:acoustic-wave · Language acquisition gain
- Lexicon: the stock of words and word parts of a language, and as each person holds it (the mental lexicon), as linguistics and psychology describe it via sys:acoustic-wave · Language acquisition gain
- Grammar (the structure of a language, as linguistics describes it) via sys:acoustic-wave · Language acquisition gain
- Sign language (the natural signed languages) via sys:electromagnetic-field · Language acquisition gain
- Human brain via sys:electrochemical-gradient · Language acquisition gain
- Family (the human family as a system) via sys:sensory-environment · Language acquisition gain
- Phonology: how languages organise the meaningless units their words are built from (phonemes in spoken languages; handshape, location and movement in signed ones), together with the field of linguistics that studies those systems via sys:acoustic-wave · Language acquisition gain
- Learning a second language (second-language acquisition) via sys:acoustic-wave · Language acquisition mixed
- Human voice via sys:acoustic-wave · Language acquisition gain
- Child development via sys:sensory-environment · Language acquisition gain
Partners and parts published science
Built from / builds (6)
- Statistical learning
- Phonetic learning (speech perception in infancy)
- Babbling
- Vocabulary development (word learning)
- Acquisition of grammar
- Child development
Sits inside / holds (2)
expressed by (13)
- Generation/Synthesis
- Recombination/Composition
- Scaffolding/Enablement (Meta-Tooling)
- Templating / Standardization
- Niche Construction (alter environment)
- Interoperability/Transduction (Bridging)
- Evolvability/Plasticity Engineering (Design for Future Variation)
- Sensing/Scanning
- Uncertainty Reduction/Learning
- Multi-Scale Data Fusion / Triangulation
- Social Learning / Public Information Use
- Quality Control & Error Correction
- Scheduled Build-and-Dismantle (On-Demand Structure)
floor filled by (20)
- Human (Homo sapiens)
- Language (human language)
- Lexicon: the stock of words and word parts of a language, and as each person holds it (the mental lexicon), as linguistics and psychology describe it
- Grammar (the structure of a language, as linguistics describes it)
- Human brain
- Family (the human family as a system)
- Joint attention
- Phonology: how languages organise the meaningless units their words are built from (phonemes in spoken languages; handshape, location and movement in signed ones), together with the field of linguistics that studies those systems
- Eye
- Ear
- Statistical learning
- Attention
- Infant-directed speech (motherese)
- Critical period
- Child development
- Phonetic learning (speech perception in infancy)
- Babbling
- Vocabulary development (word learning)
- Acquisition of grammar
- Overheard speech
Kinds (10)
- Universal grammar
- Learning a second language (second-language acquisition)
- Simultaneous bilingualism
- First-language acquisition
- Sign-language acquisition (language acquisition by deaf children)
- Late first-language acquisition (after language deprivation)
- Usage-based and emergentist accounts
- Social interactionist theory
- Behaviourist account (operant conditioning)
- Bayesian learnability accounts
wired (1)
Notes published science
- Babies need a responsive person to learn a language's sounds: 9-month-old Americans who played with Mandarin-speaking tutors for 4-5 weeks (about 33,000 syllables) told Mandarin sounds apart as well as Taiwanese babies with 10 months of listening, but babies shown the same material on television or audio learned nothing at all (src:kuhl-2010); for toddlers learning verbs, a partner over contingent video chat works too, a recording does not (src:roseberry-2014).
- Children can make a language: deaf children brought together in Managua from 1977 built Nicaraguan Sign Language out of home signs, and each younger cohort made its grammar tighter; a brow furrow that hearing Nicaraguans rarely use with questions became the grammatical question marker by the second and third cohorts (src:wp:nicaraguan-sign-language, src:kocab-2022).
- Children learn a language from tens to hundreds of millions of words; language models are trained on hundreds of billions to trillions. In one test, training a model on child-directed text did not close the gap, and the authors read the difference as a more efficient learner (src:feng-2024). Meanwhile native speakers keep getting better at grammar until about 30 (src:hartshorne-2018).
Layers
The ways it takes in its world, where they meet, and how it acts back, each from L1 (where it meets the world) up.
language acquisition: from the language around a child to a speaker of it (reader's view)
- L1: The language around the child: speech (or sign) from adults and other children, addressed or overheard; amounts differ widely between and within communities
- L2: Taking it in: hearing from the third trimester and sight of faces and hands; newborns already tell their native language from others
- L3: Sound (or sign) categories: from all languages' contrasts to the native ones between 6 and 12 months; babbling tuned to the language around
- L4: Learning: statistical learning from birth, gated by a responsive (contingent) partner and shared attention to objects, within sensitive periods
- L5: Words and grammar: segmentation, first words (gestures often first), the 18-month vocabulary burst (contested), word combinations, morphology and syntax
- L6: The language held in a committed brain, concentrated over time in left-hemisphere language areas
- L7: A speaker of the community's language: the same grammar as peers by about five, refined into adulthood, used in conversation, and sometimes reshaping the language
the boardroom (shared floors)
- L4: Learning: implicit statistical (distributional) learning from birth, gated by live social interaction, inside sensitive periods; whether the learner is language-specific (universal grammar) or general (usage-based) is disputed
- L5: Words and grammar built: segmentation by 8 months, first words, the vocabulary burst at 18 months, 10-15 new meanings a day, two-word combinations, morphology (overregularisation) and syntax flourishing at 18-36 months
- L6: The language held: the brain's circuitry committed to the native language(s), concentrating over time in left-hemisphere areas (Broca's, Wernicke's); late first-language learners never fully reach it
- L7: A speaker (or signer) of the community's language: converging on the community grammar by about five, still refining it into adulthood, able to say and understand an unbounded number of sentences, and taking turns in conversation
way in: A second language heard: dual-language input
- L1: Two languages in the input, each in its own share; growth in each tracks its share
- L3: Two streams kept apart: a bilingual brain forms; monolingual 8-month-olds struggle to track two streams in sequence
way in: Own voice: the child's babbling and words heard back
- L1: The child's own speech-related vocalizations (protophones, babbles, syllables, words), rising 66 per hour with each year
- L3: Babbling tuned to the language around: canonical syllables in the first year; by 10 months babbling differs by country; deaf infants babble later and less
way in: Sign seen: a sign language used around and to the child
- L1: A natural sign language in the child's world (deaf parents, a signing community); deaf children so exposed parallel or lead hearing children early on
- L3: Signs taken in as language: early sign vocabulary larger at 12-17 months, evening out at the two-word stage; spatial reference and complex handshapes hard until 5-9
way in: Speech heard: the language spoken around and to the child
- L1: The language around the child: adult and child talk, directed or overheard; more adult talk goes with more child speech across 1,001 children on six continents
- L2: Heard, from before birth: fetuses learn word variants and native vowels; newborns tell their native language from others
- L3: Phonetic statistics: the distribution of sounds heard tunes perception between 6 and 12 months (native up, nonnative down)
way in: The endowment: what the child brings (genes, a growing brain)
- L1: The inherited endowment; how much is language-specific is the standing dispute (universal grammar against general learning); genes act only with environment; FOXP2 mutations cause verbal dyspraxia
- L2: A brain growing fast, with more connections than in adulthood
- L3: Maturation: synapses validated by use over a long postnatal period; windows of plasticity opened by excitation-inhibition balance and closed by molecular brakes
way in: The moving body: sitting, walking, handling objects
- L1: The infant's body in the world: sitting up and walking change what can be reached and seen
- L3: Motor milestones bearing on vocabulary: acquisition as an embodied process
way in: The responsive partner: a person answering in time, turn by turn (live or by video chat)
- L1: A responsive partner, contingent in time: a live person, or a partner over video chat; the same material by television, audio or yoked video teaches nothing (Kuhl 2003 study as reviewed; Roseberry 2014 for verbs at 24-30 months)
- L2: Attention raised: infants attend more to a live person than to an inanimate source
- L3: Social gating and conversational turns: interaction gates the learning mechanisms; more turns, more Broca's-area activation
way in: The talking face and gaze: seeing speech and where the speaker looks
- L1: The speaker's face articulating and gazing at objects while talking
- L2: Seen speech joined with heard speech: place of articulation abstracted across sight and sound by 3-5 months; the face shapes statistical learning
- L3: Gaze following links words to objects; gaze shifts between tutor and toy go with stronger neural discrimination; part of triadic joint attention (child, adult, object), which develops from sharing to following to directing attention between 9 and 15 months (Carpenter et al. 1998)
way in: Thought: the child's own concepts and categories that words are mapped onto
- L1: Contested: where the concepts that words map onto come from; positions listed, none ruled: innate in some form (Plato), god-given or learned from trusted speakers (twelve centuries of Sanskrit grammarians' debate), abstracted from sense impressions (Hobbes, Locke)
- L3: Concepts and categories used to guess word meanings: lexical categories by 21-23 months, the whole-object assumption, grouping by shared properties
way in: Touch and the other near senses: objects handled, tasted and smelled while named
- L1: Objects touched, tasted, smelled and seen while their words are heard
- L3: Richer multisensory experience of a word's referent supports learning and recognising the word
way out: Gesture: pointing and gesture-plus-word combinations (hearing and deaf children)
- L1: The pointing hand in the shared light, seen by the adult
- L3: Imperative and declarative gestures, directing others' attention, from about 9-15 months
- L5: Gesture-plus-word combinations (point at bird, say 'nap') predict two-word combinations
way out: Hands: manual babbling and signs
- L1: Signs in the shared light, seen by others
- L3: Manual babbling
- L5: Signs more regular and more frequent than hearing babies' gestures
way out: Reshaping the language: children's learning changes the language the next cohort hears
- L1: The language as the next cohort meets it: changed by those who learned it
- L7: Younger children take the older children's language to greater complexity
way out: Voice: cooing and babbling, words, sentences, conversation
- L1: The child's voice in the air, 66 more vocalizations an hour each year
- L3: Canonical babbling in the first year (usual window 5-10 months)
- L5: Words and sentences: vocabulary bursts at 18 months
- L7: A thought put into words