Child Language Decoded · 3 of 6

Words, Meanings and Grammar Under Construction

Edexcel A-level English Language · Paper 2 · C1/C2 · Stages of development, stretched word meanings, and the brilliant logic behind "I goed"

What comes back to you?

SPEAK Warm the engine, no notes: a child says /ti:/ for "tree" and /ˈjæmə/ for "pyjamas". Name the two different sound patterns — and then the Golden Rule of how to frame them.
/ti:/ = consonant cluster reduction · /ˈjæmə/ = weak syllable deletion · framed as typical development, never deficit.
Today: the developmental stages plus the word- and grammar-level features, so you can place a child's age band from evidence and explain what their "errors" prove.

The puzzle of the invented word

A child of 3;4, watching snow for the first time: "look! (.) the sky is snowing down the garden"
SPEAK Nobody taught this sentence. Nobody has ever said it to the child. Yet it is understandable, grammatical in shape, and rather beautiful. What must be true about how children learn language, if they can produce sentences they have never heard?
Children are not recording machines — they are pattern factories. They extract rules and generate new language. Today's evidence builds the case; Lesson 4 attaches the theories and the famous names.

The four stages

Examiners expect you to place the child's language against the developmental sequence. Ages are averages — real children wobble around them, and saying so is itself a good analytical point.

~12–18 m
Holophrastic (one-word) stage. One word does the work of a whole sentence: "juice!" can mean I want juice, look, juice, or more juice. Context and intonation carry the rest.
~18–24 m
Two-word stage. Mini-sentences with consistent order: "mummy sock", "more juice", "daddy gone". Grammar has begun — the order is rarely random.
~24–36 m
Telegraphic stage. Three or more words, keeping content words and dropping the small grammar: "me go park now", "where teddy gone". Like an old telegram — every word costs, so only the essential survive.
~36 m +
Post-telegraphic stage. The little grammar words move in: determiners, prepositions, auxiliary verbs. Utterances lengthen; questions and negatives become adult-like. Still plenty of variability.

Place the child

Diagnose the stage from each single utterance — say your answer and your evidence aloud first.

1. "where daddy put the big spade?"

2. "gone" (said while pointing at an empty plate)

3. "I can't find my other shoe"

Stretchy words

A two-year-old owns a few hundred words but needs to talk about everything. The solution: stretch what you have. Two directions, two terms:

TermWhat happensExample
OverextensionOne word covers too much territory"dog" for every four-legged animal; "daddy" for every man
UnderextensionOne word covers too little"duck" only for the yellow bath duck — real ducks aren't "duck"

Overextensions are rarely random — the stretched word usually follows shared features: shape, size, movement, texture. A round moon called "ball" is stretching by shape; a cat called "dog" by four-legged-ness.

Diagnose the stretch

1. A child calls the orange harvest moon "ball" — what's the logic of the stretch?

Overextension by shape — round things share a label. The child's categories are built on perceptual features.

2. A child uses "hat" for hats, hoods, and a saucepan on someone's head — and refuses "hat" for a bike helmet in a shop. What two things are happening?

Overextension by function/position (things worn on heads) — and simultaneously an underextension-style limit (the unfamiliar helmet in an unfamiliar place doesn't qualify). Word meanings are being tested and tuned against experience.
SPEAK A child's vocabulary is also a map of their world. What kinds of words would you expect to dominate a three-year-old's word-hoard, and what would that let you say about context in an essay? (Think: home, food, toys, family, favourite media.)
Early lexis clusters in concrete, here-and-now semantic fields — food, family, animals, toys, screen characters — because words are learned where life happens. Naming the child's semantic fields, and what they reveal about home and interests, is direct context evidence (the 15-mark strand).

Brilliant errors

Now the single most powerful evidence type in child language. Read these, all real-style, all from children aged 2–4:

"I goed to the shops" · "look at the sheeps" · "she runned away" · "two mouses"
SPEAK Answer this carefully, because everything hangs on it: has the child ever heard an adult say "goed" or "sheeps"? So where do these forms come from — and why might a linguist call them brilliant?
No adult models these forms — so they cannot be imitation. The child has extracted the rules (past = add -ed, plural = add -s) and applies them productively, even where English makes exceptions. The term is overgeneralisation, and such forms are often called virtuous errors: the mistake that proves rule-building intelligence.

The famous experiment

In 1958, researcher Jean Berko showed children a picture of an invented creature: "This is a wug. Now there are two of them. There are two ___." Children who had never heard the word confidently said "wugs". Why is that the perfect proof?

"Wug" doesn't exist, so no child could have memorised its plural. Producing "wugs" proves the child owns the rule, not a list of remembered words. Quote "the wug test (Berko, 1958)" when you argue children generate grammar — it's exam-grade support.

Spot the evidence type

Which of these is a virtuous error — and why aren't the others?

The little words arrive last

Full words for things and actions come first. The small machinery of grammar — is, am, the, a, to, my — arrives last, because it carries no picture a child can point at. Four patterns to name:

PatternExampleWhat an adult says
Copula omission — missing is/are as the main verb"it yummy" · "teddy sad"it is yummy
Auxiliary omission — missing helper verbs"I going park" · "where daddy gone?"I am going · where has daddy gone
Determiner variability — the/a appearing, vanishing, or doubling"want the a ball" · "on tablet"a ball · on the tablet
Preposition omission — missing position words"sit next mummy"next to mummy

And the counterweight — always give the examiner what the child CAN build. The showpiece is the expanded noun phrase: "a big one with a long neck" — a noun dressed in layers of description. One quote like that balances a whole paragraph away from the deficit model.

Interleaved round — everything so far

Name the feature. Careful: today's and Lesson 2's features are mixed together.

1. "me want the /nænə/" — TWO features

Weak syllable deletion on 'banana' (Lesson 2) + non-standard pronoun "me" for "I" (grammar still sorting its pronoun case). Bonus if you noticed the determiner "the" IS present — say what's there, not just what's absent.

2. "the sheeps runned away"

Double overgeneralisation — plural -s on an irregular noun and past -ed on an irregular verb. Two virtuous errors in four words: a child applying its rulebook everywhere.

3. "daddy /tʌmɪŋ/ (.) daddy tummin a party"

Substitution /k/→/t/ ("coming" fronted to "tumming"), auxiliary omission (no "is"), preposition omission (no "to"). Three levels of analysis in one tiny utterance — exactly how top answers mine a quote.

Exam voice: the grammar paragraph

Data: Ava, aged 2;7, "reading" a picture book with her mother:

M: what's happening on this page?
A: the bear eated all the honey
M: he ate ALL the honey? (.) oh no
A: yeah (.) he a naughty bear (.) bees sad now
M: the bees are sad (.) what will they do?
A: they chase him (.) look (.) a big big cloud of bees

Write a full grammar paragraph: at least two quotes, two named features, the stage placement, and one thing Ava CAN do. Then compare.

"Ava's grammar at 2;7 sits late in the telegraphic stage. Her 'eated' overgeneralises the -ed past rule to an irregular verb — a virtuous error showing productive rule use rather than imitation, since no adult models this form. Copula omission appears twice in 'he a naughty bear' and 'bees sad now', where is/are are absent, yet in the same turn she builds the expanded noun phrase 'a big big cloud of bees', showing her syntax is already layering description. Her mother's expansion 'the bees ARE sad' quietly models the missing verb, and the shared book keeps every meaning recoverable — development on full display, communication never in doubt."
SPEAK Score the model against the ingredient list: quotes, terms, stage, a CAN, context link. Now score your own. What's the one upgrade for your next paragraph?

Recall zone

From Lesson 2: a child says /bɒbɒ/ for "bottle" — name the pattern and what it hints about age.

Reduplication — a repeated syllable replacing a harder sequence; typical of very early speech, usually fading before three.

From Lesson 1: what are the two mark blocks of Paper 2 worth, and what does each reward?

30 marks: terminology + theory applied to quoted data in organised writing. 15 marks: linking language to context and how meaning gets made.

From memory

1. The four stages, in order, with one identifying feature each.

Holophrastic (one word = one sentence) → two-word (mini-sentences, consistent order) → telegraphic (content words only) → post-telegraphic (grammar words move in).

2. Overextension vs underextension — one example each.

Overextension: "dog" for all four-legged animals. Underextension: "duck" only for the bath duck.

3. Why is "I goed" better evidence about how children learn than "I went"?

"Went" could be memorised imitation. "Goed" can only come from an internal rule — it proves generation, not copying (the wug test's point exactly).
REFLECT Which stuck faster today — stages, stretchy words, or virtuous errors? What made that one stick?
I can place a child's stage from transcript evidence and analyse word-meaning and grammar features as development, with the terminology examiners reward.