Chapter 9: Language Development in Infancy and Early Childhood

Key Focus: Prenatal Learning, Phoneme Perception, Segmentation, Word Meaning, & Syntax

• Traxler (2nd ed.), Chapter 9 Link to Chapter 9

Overview

This chapter explores language development from prenatal stages through early childhood, examining how infants learn to perceive speech sounds, segment words from continuous speech, acquire word meanings, and develop grammatical structures. We investigate the interplay between innate capacities and environmental learning, addressing fundamental questions about how children accomplish the remarkable feat of language acquisition.

Learning Goals

After studying this chapter, you should be able to:

• Evaluate evidence for prenatal language learning and its implications for nativist vs. behaviorist debates
• Explain how infants develop and refine phoneme perception, including categorical perception and perceptual narrowing
• Analyze the segmentation problem and how infants use multiple cues (prosody, phonotactics, statistics) to solve it
• Describe the role of infant-directed speech and statistical learning in early language acquisition
• Apply word learning biases and social-cognitive skills to explain how children solve the word meaning problem
• Compare nativist and usage-based theories of syntax and morphology acquisition
• Understand the U-shaped development pattern in morphological acquisition and competing explanations
• Analyze cross-linguistic differences in language acquisition and their theoretical implications

1. Prenatal Learning (pp. 331-335)

1.1 Debate: Behaviorism vs. Nativism

• Modern Consensus: Prenatal learning shapes “innate” skills—much early knowledge is learned before birth, not genetic.

1.2 What Fetuses Hear

Fetal auditory system works in the third trimester (28–40 weeks)—mother’s body filters sound:

• Filtered: High-frequency sounds (>2,000 Hz, critical for phonemes like /p/ vs. /b/) are muted—no word recognition.
• Preserved: Low-frequency sounds (500–1,500 Hz, carry prosody: rhythm/pitch) pass through.
• Mother’s Voice: Loudest (10–20 dB louder than others) via bone conduction—explains newborn preference (DeCasper & Fifer, 1980).

1.3 Definitive Prenatal Learning Evidence

A. “Cat in the Hat” Study (DeCasper & Spence, 1986)

• Training: 16 mothers read The Cat in the Hat twice daily for final 6 weeks of pregnancy (control = no story).
• Testing: Newborns used High-Amplitude Sucking (HAS) to “choose” sounds—sucked harder for Cat in the Hat (even by strangers).
• Conclusion: Infants retained the story’s prosodic patterns (not just mother’s voice).

B. Fetal Heart Rate Studies (Lecanuet et al., 1988)

• Procedure: Mothers recited a nursery rhyme daily; fetal heart rate slowed (attention) more for familiar rhymes.
• Older Fetuses (37–38 weeks): Stronger responses—auditory system matures with age.

1.4 Why Prenatal Learning Matters

1. Native Language Preference: Newborns of French parents prefer French over Russian (Mehler et al., 1988).
2. Bootstrapping Segmentation: Prosody (e.g., stress) helps infants find word boundaries post-birth.
3. Speech Attention: Prenatal exposure makes infants focus on speech—accelerating later learning.

2. Infant Phoneme Perception (pp. 335-339)

2.1 The Phoneme Challenge

Infants must:

1. Form Categories: Learn which acoustic differences matter (/b/ vs. /p/) and which to ignore (different speakers’ /b/).
2. Tune to Native Language: Retain only native contrasts (English /l/ vs. /r/; Japanese does not use this).

2.2 Categorical Perception

Infants treat continuous acoustic sounds as discrete categories (present at 1 month).

Classic Study: Eimas et al. (1971)

• Stimuli: Syllables varying in Voice Onset Time (VOT): /ba/ (0–20 ms VOT) vs. /pa/ (40–60 ms VOT).
• Procedure: Infants habituated to /ba/ (sucking slows) → tested on same-category (/ba/ 20 ms) or different-category (/pa/ 40 ms) syllables.
• Result: Dishabituated (sucked faster) only for different-category syllables—categorize like adults.

2.3 Is Speech Perception “Special”?

No—relies on general auditory mechanisms:

• Animals: Chinchillas/quail learn /ba/ vs. /pa/ (Kluender & Kiefte, 2006).
• Nonspeech: Infants categorize violin sounds (bowed vs. plucked) (Cutting & Rosner, 1974).

2.4 Universal Sensitivity

Infants detect phoneme contrasts from all languages until ~12 months:

• Werker & Tees (1983): 2–6-month-olds detect Hindi/Thompson contrasts (not in English); 12-month-olds lose this sensitivity.
• Kikuyu Infants: 2-month-olds detect /b/ vs. /p/ (not in Kikuyu)—innate, not learned.

2.5 Perceptual Prototypes (6+ Months)

Infants develop “average” phoneme representations (e.g., English vs. Swedish “ee” vowels).

• Kuhl et al. (1992): Swedish infants treat Swedish “ee” variations as same; English infants do the reverse—prototypes sharpen category boundaries.

3. Solving the Segmentation Problem (pp. 339-345)

3.1 The Problem

Fluent speech has no “spaces”—infants must segment into words (emerges 6–7 months).

3.2 Tool 1: Prosodic Bootstrapping

Use rhythm/stress—metrical segmentation strategy: stressed syllables = word starts (English).

• English Trochaic Stress: 90% of bisyllabic words (e.g., COOkie, BAby)—7-month-olds use this to segment.
• Cross-Linguistic: French infants use iambic stress (weak-strong); Dutch infants use trochaic (like English).

3.3 Tool 2: Phonotactic Knowledge (7–9 Months)

Language-specific rules for sound combinations (e.g., English allows /spl/ at word start, not /gd/).

• Learning: Infants deduce rules from utterance edges (start = word start, end = word end).
• Example: bigdog → boundary between /g/ and /d/ (English does not allow /gd/ at word start).

3.4 Infant-Directed Speech (IDS/“Motherese”)

Adults’ specialized speech style— aids segmentation:

• Features: Higher pitch, slower rate, exaggerated stress, short utterances.
• How It Helps:
  1. Clearer prosody (e.g., COOkie! highlights word boundaries).
  2. Key words at utterance ends (e.g., Look at the DOG!).
• Cross-Cultural: Universal in industrialized cultures; infants exposed to IDS segment 2–3 months earlier.

4. Statistical Learning & Segmentation (pp. 345-349)

4.1 What Is Statistical Learning?

Detect transitional probabilities (TP): likelihood one syllable follows another.

• High TP = syllables belong to the same word (e.g., /pre/-/tty/ in pretty, TP ~80%).
• Low TP = word boundary (e.g., /tty/-/ba/ in pretty baby, TP ~5%).

4.2 Classic Study: Saffran et al. (1996)

• Stimulus: Artificial mini-language (jikpel, rudneb, mibvot)—continuous stream (no pauses/stress).
• Training: 8-month-olds listened for 2 minutes.
• Testing: Looked longer at “words” (high TP) vs. “non-words” (low TP)—used TP to segment.

4.3 Development

• 6–7 Months: Rely only on statistics.
• 8–9 Months: Combine statistics + prosody (prefer high TP + trochaic stress).
• Beyond Speech: Use stats for visual/musical patterns—general cognitive skill.

4.4 Limitations

• Natural Language TPs: Variable (not 1.0 for words, 0.33 for boundaries)—insufficient alone.
• Multi-Cue Strategy: Infants combine stats, prosody, and phonotactics for accuracy.

5. Learning Word Meanings (pp. 350-357)

5.1 Quine’s Gavagai Problem

When a speaker says gavagai while pointing to a rabbit, meaning is ambiguous (whole rabbit? ears? “furry?”). Infants use biases and social skills to solve this.

5.2 Innate Biases

5.3 Social-Cognitive Skills

A. Joint Attention

Follow speaker’s gaze to infer reference. Example: Adult looks at distant toy + says dax → infant looks at toy (not adult’s hand).

B. Speaker Reliability

Trust accurate speakers. Example: 3-year-olds ignore labels from speakers who misnamed familiar objects (Birch & Bloom, 2002).

5.4 Syntactic Bootstrapping (Verb Learning)

Use sentence structure to infer verb meanings.

• Yuan & Fisher (2009): 2.5-year-olds heard She blicked! (intransitive) → looked at 1-person action; She blicked her! (transitive) → looked at 2-person action.

5.5 Noun Bias

Infants learn nouns faster than verbs (cross-linguistic):

• Conceptual: Nouns = concrete objects (easy to track); verbs = abstract actions.
• Input: Nouns are more frequent/prominent in IDS (end of utterances).

6. Acquisition of Morphology & Syntax (pp. 357-365)

6.1 Key Terms

• Morphology: Word form rules (past tense: walk→walked; plurals: cat→cats).
• Syntax: Word combination rules (English SVO: She ate cake).
• MLU: Mean Length of Utterance (2 years = ~2 morphemes; 4 years = ~4).

6.2 Morphology: Past Tense (U-Shaped Curve)

Theories

6.3 Syntax Acquisition: Theories

A. Nativism (Universal Grammar, Chomsky)

• Innate Principles/Parameters: All languages have nouns/verbs; parameters (e.g., word order) set via input.
• Example: English infants set SVO parameter; Japanese set SOV.
• Evidence: 2-year-olds master word order quickly (rarely say dog big the).

B. Usage-Based (Tomasello)

• No Innate Grammar: Learn via:
  1. Item-Based Learning: Memorize frequent phrases (want milk).
  2. Generalization: Extend to new phrases (want water).
  3. Abstract Rules: Develop later (e.g., [Want] + [Noun]).
• Evidence: 2-year-olds only use verbs in structures they’ve heard (no Mommy drank milk until hearing the transitive form).

6.4 Cross-Linguistic Differences

• Word Order: English (SVO) vs. Japanese (SOV)—infants master by 2–2.5 years.
• Case Marking: Russian/Finnish use endings (not word order)—mastered 3–4 years.
• Pro-Drop: Italian/Spanish allow omitted subjects (Ate cake); English does not—learned by 3 years.

Quick Review Questions

1. What do fetuses learn in utero, and how is this measured?
2. Why do infants lose sensitivity to non-native phoneme contrasts by 12 months?
3. What is prosodic bootstrapping, and how does it help segmentation?
4. What is the whole-object bias, and how does it solve Quine’s Gavagai Problem?
5. Why do toddlers over-regularize past tense (e.g., goed)?