Language development

Generative structure of language

👉 Language is generative, meaning that we can generate new sentences using a finite set of grammatical rules, allowing us to understand sentences we have never heard before, even when they contain novel combinations of words.

👉 Because language is generative, it allows for displacement, meaning we can talk about things that are not physically present, including the past, the future, and abstract ideas.

👉 This contrasts with animal communication, which relies on fixed signals tied to specific situations and does not allow flexible recombination.

👉 Additionally, language has no theoretical limit on sentence length or complexity, since we can embed clauses within clauses (recursion).

👉 Finally, grammar and meaning are semi-independent, meaning sentences can be ungrammatical but still understandable, or grammatical but meaningless.

Generative structure of language: Syntax and Grammar

👉 Syntax = rules for how words are arranged to form sentences

👉 Grammar provides structure that:

• clarifies meaning

• reduces ambiguity

👉 Without grammar:

• meaning is unclear

• listeners must guess what is being referred to

Biological bases of language

👉 During the first and second years, the brain shows increasing activation and specialization for language

---

🧠 Key regions (left hemisphere)

👉 Broca’s area

• language production

• grammar and articulation

• damage → difficulty producing speech

---

👉 Wernicke’s area

• language comprehension

• damage → difficulty understanding speech

---

⚠ Important clarification

👉 These areas are present early, but

• not fully specialized at birth

• become more responsive to speech gradually over the first few years

---

Impact of early versus late brain damage on language function

🔍 Findings

👉 Early damage (prenatal / infancy):

• often little to no lasting language deficit early on

• other brain regions can compensate

---

👉 Later damage (toddler → adult):

• clear language delays/impairments

• effects become more pronounced with age

---

🎯 Interpretation

👉 Language functions are not fixed at birth
👉 They become specialized over development

👉 Early brain = highly plastic
👉 Later brain = more specialized, less flexible

Developmental trajectory of word learning

👉 Birth:

• ~0 words

---

👉 End of 1st year:

• ~10 words

• slow growth (~1 word every few weeks)

---

👉 2nd year (~18–24 months):

• ~300 words

• vocabulary explosion

---

👉 After ~19 months → ~5 years:

• ~9 words/day on average

• rapid, continuous growth

The problem of reference (Quine)

👉 The problem of reference = figuring out what a word refers to in a complex environment

---

🔍 Why it’s hard

👉 When a child hears a word, there are many possible meanings

Example:
👶 hears “rabbit”
Could mean:

• the animal

• its ears

• its color

• the action (hopping)

👉 There is no obvious correct mapping

---

🎯 Key idea

👉 Word learning is difficult because of multiple possible interpretations

---

💡 Importance

👉 Despite this problem, children learn words quickly and efficiently

👉 → shows they use strategies to solve the problem

Novel noun learning task

👉 In a novel noun task, infants are shown a new object and told:
“This is a riff. Can you give me another riff?”

👉 They must choose between:

• a part of the object

• an object with the same overall shape

---

👉 Finding:
➡ infants choose the whole object (same shape)

---

🎯 Interpretation

👉 Infants assume words refer to entire objects, not parts

---

💡 Why it matters

👉 Reduces ambiguity → helps solve problem of reference

---

Fast mapping

👉 Fast mapping = ability to learn a new word–meaning association after minimal exposure

---

🔍 How it works

👉 Children use constraints/biases to quickly narrow meaning:

• Whole object bias → word refers to whole object

• Shape bias → extend by shape

• Mutual exclusivity → new word → new object

---

👉 Also relies on social cues:

• joint attention

• caregiver labeling

• contingent responses

---

🧪 Evidence (Linda Smith)

👉 After shape-bias training:

• children could map new word → new object immediately

• even with no direct training on that object

👉 → shows rapid generalization

---

🎯 Function

👉 Allows children to:

• learn words very quickly

• build vocabulary efficiently

---

💡 Important point

👉 Initial mapping is quick but incomplete
👉 refined over time (slow mapping)

Cognitive constraints on word learning: (1) Shape bias (Smith)

Core idea

👉 Infants can learn that words refer to shape

➡ this is called the shape bias

---

🧪 Study setup

👉 7 weeks of training during play:

• experimenter labels different objects with same shape using same word
  ➡ draws attention to shape as category

---

🔍 Test 1 (Week 8 – First-order generalization)

👉 Same object as training
👉 Choices:

• shape match

• color match

• texture match

---

👉 Result:
➡ trained infants chose shape more often than control

---

🔍 Test 2 (Week 9 – Higher-order generalization)

👉 Brand new object + new word

👉 Question:
➡ Will infants extend the word by shape without training?

---

👉 Result:
➡ YES — trained infants used shape bias again

---

⚡ Key concept: Fast Mapping

👉 Infants can learn a new word quickly
➡ using prior knowledge (shape bias)

👉 No direct training needed

---

📈 Long-term effect

👉 1 month later:

• trained group → 256% vocabulary increase

• control → 78% increase

---

🎯 What this shows

👉 Experience → builds cognitive biases

👉 Shape bias + whole object constraint →
➡ faster word learning

👉 Learning becomes easier over time

Cognitive constraints on word learning: (2) Mutual exclusivity

🧠 Core idea

👉 Children assume that each object has only one label

➡ new word = new object

---

🔍 Example

👉 Child sees:

• one known object (already has a name)

• one unknown object

👉 Hears a new word

---

👉 Child will:
➡ match the new word → unknown object

---

🎯 What this shows

👉 Children prefer to map:
➡ new words to new things

---

👉 This is a cognitive bias that:

• simplifies word learning

• reduces confusion

---

⚡ Connection to learning

👉 Helps with:
➡ fast mapping (learning words quickly)

---

⚡ Quick recall

👉 New word → new object

Cognitive constraints on word learning: (3) Whole object constraint

🧠 Core idea

👉 When children hear a new word,
➡ they assume it refers to the whole object, not its parts

---

🔍 Example

👉 You show a child a dog and say “blicket”

👉 Child assumes:
➡ “blicket” = the whole dog

NOT:

• its tail

• its color

• its fur

---

🎯 What this shows

👉 Children naturally:
➡ map words to entire objects first

---

👉 This simplifies learning by:

• reducing possibilities

• making word learning faster

---

⚡ Connection

👉 Supports fast mapping

---

⚡ Quick recall

👉 New word = whole object

Social guidance of attention during word learning (Yu and Smith study)

🧠 Core idea

👉 Children use social biases (not just hearing words)
➡ to support fast mapping and word learning

---

🔍 Types of social biases 👉 1. Intersensory redundancy

👉 Caregivers label objects that are:

• in their hand

• or in the child’s hand

➡ combines multiple senses (seeing, hearing, touching)
➡ strengthens word–object connection

---

👉 2. Social cue following (hands & eyes)

👉 Infants track:

• what caregivers are looking at

• what caregivers are holding

➡ helps them identify what is being labeled

---

👉 3. Joint attention

👉 Child and caregiver focus on the same object at the same time

👉 Often occurs when:

• parent looks at object child is holding

• child looks at object parent is holding

➡ creates optimal conditions for word learning

---

🎯 What this shows

👉 Word learning is socially guided, not just passive

👉 These biases:

• reduce ambiguity

• make learning faster

---

⚡ Connection

👉 Supports fast mapping