Child Language Decoded · 4 of 6

The Five Theories

Edexcel A-level English Language · Paper 2 · C1/C2 · Skinner, Chomsky, Bruner, Halliday and Tomasello — and how to make them earn marks in an essay

What comes back to you?

SPEAK No notes: a child says "I drawed two sheeps." Name the feature (twice over), explain why it CANNOT be imitation, and recall the invented-creature experiment that proved the point.
Overgeneralisation (virtuous errors) on verb and noun · no adult models "drawed"/"sheeps", so the forms must come from an internal rule · the wug test (Berko, 1958).
Today: the five theories every top essay draws on, so you can attach an explanation — and a famous name — to any feature you find in the data.

The question behind the questions

Every child on Earth — every language, every culture — learns to speak in roughly the same sequence, mostly without teaching. The deepest question in this subject is simply: how?

SPEAK Before meeting the official answers, commit to your own. How do YOU think children learn language — copying? An inborn instinct? Conversation? Argue your position for thirty seconds; today will test it against the evidence.

By the end you'll hold five answers at once — because the exam's highest levels go to candidates who show the data supports different theories at different moments, not to candidates with a favourite.

Answer 1 · Behaviourism — Skinner

B.F. Skinner (1957) — "children imitate and are rewarded"
Language is learned behaviour: copy what you hear; what gets smiles, food and attention gets repeated.

A child says "milk", receives milk and delight, and says it again. Adults model; children imitate; reinforcement carves the habit deeper.

The data supports it when… a child echoes an adult's form moments after hearing it, or fixed polite phrases ("ta", "peas") appear exactly as trained.
The data challenges it when… "goed" and "sheeps" appear — forms no adult ever modelled — or when a child resists correction and keeps their own version.

Answer 2 · Nativism — Chomsky

Noam Chomsky (1965) — "children are born wired for grammar"
The brain ships with a Language Acquisition Device: an innate capacity to extract grammar from messy input.

Chomsky's argument: the speech children hear is limited and full of fragments, yet every child builds a full grammar from it — so the essential machinery must be inborn. Virtuous errors are the LAD's fingerprints: rules applied by a system nobody taught.

The data supports it when… overgeneralisations appear ("runned", "mouses"), or the child produces novel sentences they cannot have heard.
The data challenges it when… the adult's contribution is doing visible heavy lifting — Chomsky's account says little about conversation, and transcripts are full of it.
SPEAK Thirty-second courtroom: you're Skinner's lawyer — make his best case using "peas" (please). Now switch sides: you're Chomsky's — make the case using "goed". Which brief felt stronger, and why?

Answer 3 · Social interactionism — Bruner

Jerome Bruner (1983) — "language is learned inside conversation"
Whatever is inborn, it only ignites through interaction. Adults run a Language Acquisition Support System (LASS) — Bruner's deliberate reply to Chomsky's LAD.

Scaffolding is the key term: adults support the child exactly where needed, then remove support as skill grows — like scaffolding coming off a finished building. In transcripts, scaffolding is visible and quotable:

Adult moveWhat it looks like
Modelling / recastChild: "the /wɪŋ/" → Adult: "the swing!" — correct form returned, no criticism
ExpansionChild: "it big" → Adult: "it IS a very big one" — the grammar installed
Questions & reformulation"What did you have? What did you eat?" — repeated invitations to talk
The data supports it when… the adult's questions, recasts and expansions visibly drive the child's next utterances (watch for a child reusing a structure the adult just modelled).
The data challenges it when… the child produces forms no interaction supplied ("goed") — scaffolding shapes, but cannot fully explain, the building.

Answer 4 · Halliday — "children talk to DO things"

Michael Halliday (1975) — the functions of early language
Ask not what the child says, but what the child is trying to DO with it.

Halliday catalogued the jobs early language performs. The four most useful in exam data:

FunctionThe jobSounds like
Instrumentalgetting needs met"more juice"
Interactionalbonding with people"night-night mummy"
Imaginativecreating pretend worlds"I going to be a crocodile"
Informativegiving information"bees sad now"

(The full set also includes regulatory — controlling others; personal — expressing self; heuristic — exploring the world with "why?".) Naming the function a child's utterance performs links language directly to the question's focus — what the language allows the child to do.

Answer 5 · Usage-based theory — Tomasello

Michael Tomasello & Elena Lieven (2000s) — "children learn the chunks they hear and love"
No special grammar organ needed: powerful pattern-finding + rich experience does the job.

Children first learn whole chunks around things they care about ("eat him up" from stories), then gradually abstract the patterns inside them. Frequency and passion decide what's mastered first.

The data supports it when… a child nails a beloved complex word (a dinosaur name) while simplifying easy ones, or recycles storybook chunks.
The data challenges it when… productive rules outrun any plausible frequency — "wugs" was produced for a word heard once.

Theory courtroom

Each exhibit is a real-style transcript moment. Decide which theory it most strongly supports — argue aloud first, then click your verdict.

Exhibit A:

Child (2;9): it broken
Parent: oh no (.) it IS broken isn't it
Child: it is broken (.) mend it

Exhibit B:

Child (3;1): the mans digged a big hole

Exhibit C:

Child (3;0), to a toy: you be the dragon (.) I be the knight (.) fee fi fo fum

Exhibit D:

Parent: say thank you
Child (2;4): /tæŋ ku:/
Parent: good girl!
SPEAK Verdict time: having seen four exhibits, deliver a one-minute closing speech — "No single theory explains all the data, because…" — using at least two exhibits as evidence.

Exam voice: weaving theory into analysis

Theory bolted on ("this supports Chomsky") earns little. Theory woven in — support AND challenge — earns the top levels. Compare:

Bolted onWoven in
"The child says 'eated'. This supports Chomsky.""'Eated' cannot be imitated — no adult models it — so it evidences the productive rule-building nativists describe; yet the same child's 'it is broken', produced seconds after her father's expansion, shows interaction shaping her output exactly as Bruner's scaffolding predicts. The data rewards both accounts at different moments."

Your turn

Data: Noah, aged 3;2, helping his grandfather water the garden:

G: what do the flowers need?
N: they need a drink (.) I giving them a drink
G: you're giving them a lovely drink
N: I'm giving them a drink (.) grow flowers grow! (.) abracadabra
G: is the magic working?
N: yeah (.) they growed a bit

Write one theory-woven paragraph. Challenge: use at least TWO theories, with the data deciding when each applies.

"Noah's exchange shows two acquisition mechanisms operating side by side. His auxiliary-less 'I giving them a drink' is expanded by his grandfather to 'you're giving them a lovely drink', and Noah's very next turn — 'I'm giving them a drink' — adopts the contracted auxiliary, a textbook instance of Bruner's scaffolding steering output through interaction. Yet minutes later 'they growed a bit' applies the -ed rule to an irregular verb no adult models, evidence for the internal rule-building nativists describe. Meanwhile 'grow flowers grow! abracadabra' performs Halliday's imaginative function, turning watering into a spell. The transcript rewards no single theory; it shows a child learning through conversation and beyond it at once — which is precisely what the highest-level essays are expected to argue."
SPEAK Audit the model: where exactly does it switch theories, and what words manage the switch? ("Yet…", "Meanwhile…") Steal those hinges for your own writing.

Recall zone

From Lesson 3: the four developmental stages, in order.

Holophrastic → two-word → telegraphic → post-telegraphic.

From Lesson 2: "spaghetti" → /ˈgeti/. Name the pattern.

Weak syllable deletion — the unstressed opening syllables drop, the stressed core survives.

From Lesson 1: what one framing must never appear in your essay?

The deficit model — treating the child's language as faulty rather than developing.

From memory: the five theories

Close everything. Say all five — name, one-line idea, one piece of supporting data, one piece of challenging data. Then check the full table.

Skinner — behaviourism: imitation + reward · supported by trained politeness ("/tæŋ ku:/" → "good girl!") · challenged by "goed".
Chomsky — nativism: inborn grammar device (LAD) · supported by virtuous errors and the wug test · challenged by how much visible work conversation does.
Bruner — social interactionism: scaffolding, LASS · supported by expansions the child then reuses · challenged by forms interaction never supplied.
Halliday — functions: language does jobs (instrumental, interactional, imaginative, informative…) · applies wherever you ask WHAT the utterance is for.
Tomasello — usage-based: chunks + frequency + passion · supported by perfect dinosaur names and story chunks · challenged by rules outrunning frequency.

One-minute closing argument

1:00
SPEAK "Which theory best explains how children learn language?" Answer the way a top essay would — no single winner, evidence deciding moment by moment — for one full minute.
REFLECT Which theory was hardest to argue FOR? That's the one to reread before any timed essay — weak spots, not favourites, decide grades.
I can name all five theories, attach each to the data that supports AND challenges it, and weave them into analysis rather than bolting them on.