Paper 1 Section B — children's language development.

HookThe imaginary bird that settled a real argument

ModelPhonological development — from babble to system

The broad sequence: crying and vegetative sounds first; cooing from around six to eight weeks; babbling from around six months, first reduplicated ('ba-ba-ba') then variegated ('ba-gi-da'); and by roughly ten months the babble narrows towards the sound inventory of the language the child actually hears — phonemic contraction. First recognisable words arrive around twelve months, and every age in this paragraph is a typical range, not a deadline: individual variation is enormous, and saying so in the exam is a mark of quality, not hedging.

Early words are systematically simplified, because a child's perception runs ahead of their articulation. The named processes to spot in data: deletion of final consonants ('do' for dog) and of unstressed syllables ('nana' for banana); cluster reduction ('pane' for plane); substitution of harder sounds by easier ones ('wabbit'); reduplication ('choo-choo'); and assimilation, where one sound colours another ('gog' for dog). Plosives and nasals tend to arrive before fricatives — /θ/ in 'three' is famously late.

The elegant proof that these are output limits rather than knowledge gaps is the 'fis phenomenon' from a 1960 case study: a child who said 'fis' for fish flatly rejected an adult who said 'fis' back — demanding 'fis', but accepting only 'fish'. The child could hear the difference perfectly; the mouth just could not deliver it yet.

ModelLexis and semantics — the labelling problem

Vocabulary grows slowly, then detonates. A typical child holds around 50 words at eighteen months, then enters the naming explosion, adding words at a rate that carries them to several thousand by school age. Early lexicons are dominated by concrete nouns from the child's immediate world, with social words ('bye-bye') and a scatter of verbs.

The interesting data is in the mistakes. Overextension stretches a word beyond its adult range — 'dog' for every four-legged animal. Eve Clark and later Leslie Rescorla showed these are not random: categorical overextensions stay within a category (all animals become 'dog'), analogical ones follow a shared property (the moon becomes 'ball' because it is round), and mismatch statements use a word to comment rather than label ('duck' said while pointing at an empty pond). Underextension is the quieter opposite — 'duck' meaning only the yellow one in the bath, not the things on the pond.

Jean Aitchison packaged the whole progression as three overlapping tasks: labelling (attaching sounds to things), packaging (working out each label's boundaries — where over- and underextension live), and network-building (wiring words together into hypernyms, opposites and fields, so 'dog' finds its place under 'animal' and next to 'cat'). In writing data from older children, network-building shows up as synonym choice and semantic precision.

ModelGrammar — the rule factory

Grammatical development runs through four broadly sequenced stages. Holophrastic (roughly 12–18 months): single words doing whole utterances' work — 'juice' as request, observation or complaint, disambiguated by prosody and context. Two-word (around 18–24 months): consistent mini-grammar, 'mummy sock', 'more juice', with word order already largely correct. Telegraphic (around two to three years): three or more words keeping the content and dropping the function words and auxiliaries — 'daddy go work now'. Post-telegraphic (from around three): function words, auxiliaries and clause-joining arrive, and utterances start to look like the adult target.

Inflectional morphemes appear in a strikingly consistent order — Roger Brown's longitudinal work found present progressive '-ing' and plural '-s' early, past '-ed' and third-person '-s' later. And the crown jewels of any data response: virtuous errors. 'Runned', 'sheeps', 'more better' are forms no adult models, so they cannot be imitation; they are overgeneralisations of internalised rules, often following a U-shaped curve where a correct irregular ('went') is displaced by the rule ('goed') before both settle. Question formation climbs its own ladder — rising intonation, then fronted wh-words without inversion ('where daddy go?'), then auxiliary inversion — and Ursula Bellugi traced the matching path for negation: external 'no' ('no bed'), internal 'no/not' without auxiliaries ('I no want it'), then integrated auxiliary negation ('I don't want it').

ModelPragmatics — learning what language is for

Before children control much grammar, they already use language to get things done. Michael Halliday, from a case study of one child's development, catalogued the early functions: instrumental ('want juice' — obtaining), regulatory ('push me' — directing others), interactional ('night-night' — maintaining relationships), personal (expressing self), heuristic ('wassat?' — probing the world), imaginative (play voices and stories), and representational (conveying information — the one adults assume is primary, and one of the last to dominate). Labelling functions in transcript data is quick AO2 credit.

Conversational competence has its own timetable. Turn exchange is rehearsed before words exist — in peekaboo and feeding 'proto-conversations', which is one of Jerome Bruner's central observations. Then come adjacency pairs (question–answer, greeting–greeting), explicitly coached politeness routines ('say thank you'), and repair strategies when communication fails. The late arrivals are the socially expensive skills: judging what a listener already knows, framing indirect requests, and handling non-literal language — irony and sarcasm are often shaky until eight or later. In data, watch how a child manages the conversation, not just the forms inside their turns.

ModelWriting — a second acquisition, on different terms

Speech is acquired from immersion; writing has to be taught — which makes written data developmentally younger than spoken data from the same child, and the comparison is a gift in the exam. Emergent literacy starts long before school: mark-making that distinguishes drawing from 'writing', scribble that runs left-to-right, letter-like shapes, the totemic first word — the child's own name.

Richard Gentry's spelling stages give you labels for what comes next: precommunicative (letters without sound-links), semiphonetic (letters for salient sounds — 'KT' for cat), phonetic (all sounds represented, conventions absent — 'wonse' for once), transitional (visual patterns of English arriving, not always correctly placed), and conventional. The analytical reflex: an invented spelling like 'monsta' is not failure, it is a phonological analysis performed by a six-year-old — data about what they can hear.

For composition, Barry Kroll's stages chart the writing–speech relationship: preparation, then consolidation (around seven to eight, writing sounds like transcribed speech — 'and then... and then'), differentiation (around nine to ten, written genres and registers pull away from speech), and integration (a controlled personal voice, for many writers well beyond eleven). Joan Rothery's classification of early school writing adds genre labels — observation/comment, recount, report, narrative — and UK data often reflects synthetic phonics teaching: grapheme–phoneme correspondences first, digraphs, then the 'tricky' common exception words.

MechanismThe theory wars — and how to referee them

AO2 is 20 of the 30 marks, and it is not earned by listing theorists; it is earned by matching each theory to the data it explains and confronting it with the data it cannot. Skinner's behaviourism — imitation plus reinforcement — handles accent, vocabulary choice and coached politeness routines well, and collapses on virtuous errors and wug plurals: children produce forms no adult has ever modelled. Chomsky's nativism — an innate language faculty, the 'poverty of the stimulus' argument that input is too thin and error-filled to explain the grammar children build — predicts exactly the uniform stage sequences and rule overgeneralisation the data shows, but says little about interaction, and the tragic case of Genie (discovered in 1970, aged thirteen, after years of extreme isolation; she learned words readily but never full grammar) suggests innateness has a sensitive period and needs interaction to fire. So does Bard and Sachs' study of Jim, a hearing child of Deaf parents whose exposure to spoken English came mainly from television: his speech lagged badly until sessions with an adult — input alone, without interaction, was not enough.

Piaget's cognitivism ties language to general cognitive development — children talk about absent objects only after object permanence; comparatives follow seriation — a good fit for semantics, though grammar sometimes outruns general cognition. Bruner's interactionism supplies what Chomsky leaves out: a Language Acquisition Support System of scaffolding routines and child-directed speech — raised pitch, exaggerated intonation, simplified lexis, expansions and recasts. Yet Ochs and Schieffelin's fieldwork found communities, such as the Kaluli of Papua New Guinea, where adults do little of this and children acquire language on schedule — so CDS helps but is not necessary. No account survives all the evidence. That sentence, demonstrated with data, is the top band.

VocabularyKey terms the mark scheme pays for

TrapsMisconceptions that cost marks

ExamWhat examiners want