You are viewing a free preview of this lesson.
Subscribe to unlock all 10 lessons in this course and every other course on LearningBro.
Lexical and semantic development describes two intertwined achievements: building a vocabulary (the lexicon, the store of words a child knows — the lexical side) and working out what those words mean and how they relate to one another (the semantic side). The two are distinct. A child can have a word in their lexicon while still having only a partial, fuzzy or wrong sense of its meaning — which is exactly why a toddler can confidently say "dog" and yet apply it to a cow. From the first tentative word around twelve months to a vocabulary of many thousands of words by the start of school, this is the most visible and dramatic strand of language acquisition, and a rich one for data analysis because lexical errors (over- and under-extension, novel coinages) are so revealing of the child's underlying conceptual organisation.
A useful preliminary distinction is between receptive vocabulary (the words a child understands) and productive (expressive) vocabulary (the words a child uses). Comprehension consistently runs ahead of production: children understand many more words than they can say, at every stage. When you analyse data, remember that absence of a word from a child's speech is not proof they do not know it.
First recognisable words typically emerge around 12 months, with wide individual variation (some children produce words by 9 months, others not until 18). What counts as a "word" is a stable sound-sequence used consistently for a meaning — so a child's invented baba reliably used for bottle qualifies, even though it is not the adult form (such forms are sometimes called proto-words).
The period from roughly 12–18 months is the holophrastic (one-word) stage, in which a single word is made to do the work of a whole sentence.
Key Definition: Holophrastic stage — the period (approximately 12–18 months) in which a child uses single words (holophrases) to convey meanings that an adult would express in a full clause. Milk might mean "I want milk", "There's the milk", or "I've spilled the milk", depending on context.
| Holophrase | Possible meanings (depending on context, intonation, gesture) |
|---|---|
Milk | "I want milk" / "That is milk" / "I've spilled the milk" |
Daddy | "Daddy is here" / "Where is Daddy?" / "That is Daddy's" |
Up | "Pick me up" / "I want to go up" / "Look up there" |
No | "I don't want that" / "That is wrong" / "Don't do that" |
Because a single word is so underspecified, the meaning of a holophrase is co-constructed: it depends on context, intonation and gesture, and the caregiver supplies a great deal of the interpretive work. This collaborative meaning-making is itself evidence cited by interactionists for the role of the caregiver (Bruner's scaffolding) in early language.
Katherine Nelson (1973) carried out a foundational study, recording the first 50 words of 18 children and sorting them into categories. Her percentages have become a staple of the specification:
| Category | Approx. proportion | Examples |
|---|---|---|
| General nominals (naming a class of things) | ~51% | dog, ball, car, milk, shoe |
| Specific nominals (naming a particular thing/person) | ~14% | Mummy, Daddy, the family pet's name |
| Action words | ~14% | go, up, sit, give |
| Modifiers | ~9% | big, hot, mine, nice |
| Personal-social words | ~8% | no, yes, please, bye-bye |
| Function words (grammatical glue) | ~4% | for, is, what |
The headline finding is the dominance of nouns, especially names for concrete, manipulable, dynamic things (objects that move, make a noise, or can be acted upon) — the so-called noun bias. A plausible explanation is cognitive salience: a ball or a dog is a stable, bounded, pointable referent, whereas an action like give is fleeting and an abstraction like fairness has nothing to point at. Joint attention (the caregiver and child looking at the same object and the caregiver naming it) makes nouns especially easy to map.
Nelson also identified individual variation in style:
This style difference is important evidence against any single, rigid universal route: children can take measurably different paths to the same destination.
Key Definition: Noun bias — the finding (Nelson, 1973) that early vocabularies are dominated by nouns, particularly names for concrete objects, people and animals. It is often attributed to the cognitive salience and picturability of objects relative to actions or abstractions.
A caution on cross-linguistic claims: the noun bias is robust for English, but it is not strictly universal. Studies of verb-prominent languages (such as Korean and Mandarin) report relatively earlier and stronger verb learning, which suggests input and language structure shape the bias. If you mention universality, qualify it.
As children fit words to the world, they make two characteristic, opposite errors that lay bare their developing semantic categories.
Overextension is using a word too broadly, stretching it beyond its adult range. Rescorla (1980) distinguished three types:
| Type | Basis of the extension | Example |
|---|---|---|
| Categorical | Other members of the same broad category | Calling all four-legged animals dog |
| Analogical | Perceptual resemblance | Calling a round ball, then calling the moon ball because it is round |
| Predicate / relational | A relationship or association rather than the thing itself | Saying hat while pointing at a head; saying door for a key (the thing it relates to) |
Overextension is most visible between roughly 1 and 2½ years and tends to involve a sizeable minority of a child's early words. It typically resolves as vocabulary grows and finer distinctions become available — the child acquires cow, horse and sheep and stops calling them all dog.
Key Definition: Overextension — applying a word to a wider range of referents than its conventional adult meaning (e.g. all men
Daddy, all round thingsball). Rescorla (1980) classified it as categorical, analogical or predicate/relational.
Underextension is the mirror image: using a word too narrowly, for only a subset of its proper referents.
| Example | What it shows |
|---|---|
cat used only for the family cat, not other cats | The word is bound to a single exemplar |
shoe used only for the child's own shoes | The category has not been generalised |
duck used only for ducks on the pond, not a toy duck | Over-reliance on the original context of learning |
Underextension is thought to be especially common in comprehension, whereas overextension is more conspicuous in production. It is under-researched partly because it is harder to notice: you immediately spot a child calling a horse a dog, but you may never observe that they fail to call an unfamiliar cat a cat. An important interpretive point: a child can overextend in speech (saying dog for a horse) while not being fooled in comprehension — asked to "point to the dog" among pictures, the same child may choose correctly. This again warns against reading production errors as straightforward evidence about underlying knowledge.
Eve Clark (1973) proposed the Semantic Feature Hypothesis to explain overextension. The core idea:
| Phase | Features the child has linked to dog | Range of application |
|---|---|---|
| Early | [four legs], [fur], [moves] | dogs, cats, horses, cows (broad overextension) |
| Intermediate | + [barks] | dogs and dog-like animals (narrower overextension) |
| Mature | + size, shape, finer features | dogs only (correct extension) |
hat), which is about association, not shared physical features.Jean Aitchison proposed that learning a word involves three tasks, which map onto three developmental stages:
| Stage | Task | What the child is doing |
|---|---|---|
| Labelling | Linking sounds to referents | Grasping that sequences of sounds (words) are names for things — e.g. that dog picks out the family pet. This is the insight that things have names. |
| Packaging | Working out the extent of a word's meaning | Discovering the boundaries of a category — what does and does not count as a dog. Over- and under-extension belong to this stage, as the child mis-judges the package's size. |
| Network-building | Connecting words to one another | Building a semantic network of relations between words: that a poodle is a kind of dog (hyponymy), that dog belongs under animal, that big contrasts with small (antonymy), that words collocate. |
Key Definition: Network-building (Aitchison) — the stage in which a child stops treating words as isolated labels and begins to relate them: hyponymy (type-of links), synonymy, antonymy and collocation, forming an organised mental lexicon.
Aitchison's framework is examiner-friendly because it lets you place a child precisely: a child still mis-judging which animals are dogs is packaging; a child who can tell you a spaniel is a kind of dog is network-building.
A deep puzzle underlies all of this. When a caregiver points and says "rabbit", how does the child know the word means the whole animal rather than its ears, its colour, its hopping, or the field it sits in? The philosopher W. V. O. Quine framed this as the problem of the indeterminacy of reference: in principle the input radically under-determines a word's meaning, yet children map words quickly and usually correctly. Researchers have proposed that children come to the task armed with lexical constraints or biases that drastically narrow the possibilities:
cup will tend to map a new word heard in the presence of a cup and an unfamiliar object onto the unfamiliar object. This bias powers fast mapping and helps explain how second labels (synonyms, superordinates) are resisted at first.These constraints are themselves theoretically loaded: if children bring innate biases to word learning, that is a (lexical) cousin of Chomsky's argument that they bring innate structure to grammar. An interactionist would counter that joint attention and the caregiver's pointing already do much of the disambiguating work, so the biases may be supported by social cues rather than purely innate. You can present the constraints as a strong illustrative example of the active, structure-seeking child, while noting the live debate over how much is innate versus socially scaffolded.
Words are not stored as an unordered list but as an interconnected mental lexicon, and the kind of links a child has formed is diagnostic of their stage (Aitchison's network-building). Useful relations to name in analysis:
| Relation | Definition | Example |
|---|---|---|
| Hyponymy | a "kind-of" hierarchy (subordinate under superordinate) | spaniel is a hyponym of dog; dog of animal |
| Synonymy | words with closely similar meaning | big / large |
| Antonymy | opposite meaning | hot / cold |
| Collocation | words that habitually co-occur | fish and chips; salt and pepper |
| Polysemy | one word, several related senses | mouth (of a person / of a river) |
Children's developing control of these relations is observable: superordinate terms (animal, furniture) are typically acquired after basic-level terms (dog, chair), which are in turn earlier than subordinate terms (spaniel, rocking-chair) — the basic-level bias. A child who spontaneously offers a superordinate ("a dog is an animal") or an antonym is demonstrating network-building and a more mature lexicon, which is precisely the sort of observation that earns AO1 credit when justified from data.
Subscribe to continue reading
Get full access to this lesson and all 10 lessons in this course.