Speech Production and Early Lexicon Notes

Speech Production

Overview:
- Development of vocalization
- Early words' phonological processes
- Speech production explanation
Perception vs. Production
- Measurement differences
- Spontaneous speech and imitation analysis
  - Caregiver diaries
  - Caregiver reports: checklists
  - Recordings
- Infants recognize phonemes before producing them
- Understanding more about phonology surpasses their pronunciation skills
  - Example: "wistener" vs. "listener"

Early Speech Production

Tasks:
- Vocalize native phonemes
- Match adult sounds to own repertoire
- Associate sound patterns with meaning
Production Development:
- Bio maturation
- Social experience
Maturation:
- Brain maturation
  - Growth and myelination
- Vocal tract changes
  - At birth, the vocal tract is three times smaller than an adult's.
  - Infant: one tube for breathing, sucking, and swallowing

Development of Vocalization

0-2 months: Reflective stage
- Basic cry patterns (discomfort sounds)
- Vegetative sounds (burp, gurgle, suck, swallow, snore)
2-4 months: Cooing stage
- Vowel-like sounds
- "Comfort" sounds (pleasure in interaction)
- Laughter: ~4 months
4-6 months: Expansion stage
- Vocal play – manipulation:
  - Pitch (e.g., squeals and growls)
  - Loudness (e.g., yells)
  - Oral/nasal closures (e.g., snorts and raspberries)
6-10 months: Canonical Babbling stage
- Syllable-like sequences
- Via opening and closing jaws, lips, and tongue
- Reduplicated: repeat CV sequence (e.g., ma ma ma)
  - C: most often stops
  - V: tend to be low and open
  - Important auditory feedback loop
  - Most common phonemes are $|m|$ and $|d|$ .
10-12 months: Variegated Babbling stage
- Varied CV patterns
  - E.g., di bo ga
  - Vowels begin to sound similar to native language
  - Start producing intonational contours
12-18 months: Single Word stage
- First words: open syllables
  - CV or CVCV; rarely CVC
  - Use of gestures, eye gaze, and pointing
    - Demonstrates interest prior to formal language
  - Sometimes invent own symbols
    - Idiomorphs: idiosyncratic words used in consistent ways
    - E.g., "ca ca" for milk, "whew" for hello
  - Often contextually-bound
18-24 months: Two Word (telegraphic) stage
- Word combos
- Convey semantic roles and relationships
  - E.g., baby sleep, kick ball
  - Function words omitted

Phonological Processes in Early Words

Phonological Acquisition: “Holly Jolly Christmas”
- Norah: 24 months
- Kids’ pronunciation differs from adults’
  - E.g., “dat” vs. “that”
- All kids make mistakes before mastering phonology
- Differences persist, despite:
  - Correction
  - Child’s own realization
- Errors reveal that kids have learned about phonology
  - Systematic
    - E.g., wed, wight, awound
Order of sound acquisition
- 1. Sound pairings that differ maximally
  - E.g., C and V
- 2. Consonant clusters
  - E.g., “sp” in spill, “tr” in tree
- 3. Final consonants
  - E.g. “t” in cat
Reduction: delete or eliminate sounds
- Weak-syllable deletion
  - E.g., “nana” for banana
- Final-syllable deletion
  - E.g., “ba” for bottle
- Cluster reduction
  - E.g., “tor” for store, “pider” for spider
Assimilation: replace sound(s) w/similar (easier) sound
- E.g., “tee” for see, “lellow” for yellow, “guk” for truck
Reduplication: 1 syllable repeated
- E.g., “baba” for bottle
“Ana” for banana, “mama” for mommy, “bup” for truck, “ca” for car, “kee” for kitty, “cooky” for cookie

Segmentation

Example sentence used for segmentation analysis: "The feet were all different sizes. This girl has very big feet. Even the toes on her feet are large."
Test: feet… feet… feet

Infant-Directed Speech (IDS)

Exaggerates pauses, pitch, and syllable length
- Highlights boundaries between syllables and words
- Interaction with stress
  - Example: ↑ stress on strong syllable
Infant-directed speech is the most exaggerated amongst other speech.

Probabilistic Cues

Syllables within words co-occur more frequently than those across word boundaries
Transitional Probabilities of Syllables (TP)
- E.g., likelihood of syllable B following syllable A
- Example:
  - PRE → TTY (High likelihood)
  - BA → BY (High likelihood)
  - Low likelihood of TTYBA
Saffran et al. (1996)
- Do infants use TP to segment?
- 8 months old
  - Habituation: 2-min. exposure to syllable stream
    - E.g., tudaropigoladabikutibudopa
    - Synthesized speech: no prosodic info, stress, or additional cues (Just TP)
    - Some 3-syllable combos more common than others
  - Test: isolated words (high TP; TUDARO and PIGOLA)
  - Non-words (across word boundaries, low TP (ROPIGO)

Summary

Infants use a variety of cues in segmentation
- Before attaching meanings to words
- Facilitates lexical development
Which cues are most influential?: Depends on:
- Language being learned
- Sophistication of learner (e.g. age, experience)

Early Lexicon: Content and Acquisition

First words
- ~12-18 months
- Primarily signal intentions
  - Negation (“no”)
  - Recurrence (“more”)
  - Notice: call attention to (“hi”)
- Most contextually-bound (e.g., ‘no’ = only for food) and not “true words”
  - Lack abstractness
  - Don’t refer to absent objects
  - Don’t represent meaning beyond context
First 50: Course
- Slowly added
  - ~10 per month (some don’t stick)
  - Comprehension > production
- First 50: Content
  - Noun bias in English learners
    - Specific (e.g., Mommy, Blanky) and general (e.g., kitty, bottle)
- After 50 words, rate ↑ to ~30 words per month: Vocab Spurt
  - Controversy re: universality and time
  - Doesn’t happen w/all kids
  - Individual differences: later for expressive (vs. referential) kids

Why Noun Bias?

1. Refer to obvious things in the environment
- Easy to learn with ostensive definition (‘that is an X’)
- Natural partitions
- Don’t require relational meanings, like verbs
2. More common in English
- But, NOT in languages where nouns often dropped (e.g., Turkish)
3. Eng nouns more salient (end of utterances, extra emphasis in IDS)
4. Eng moms use more nouns
- But, NOT Japanese moms, who use more verbs (and child vocabs reflect input)

Vocab Spurt

Causes?
- 1. Naming insight (all things have names)
- 2. Ability to fast map - learn from 1 exposure
  - E.g., familiar words (shoe, ball, car) + unfamiliar
    - “Shoe” vs. “zib”
- 3. Development of categorization
  - Strong + correlation w/spontaneous sorting
    - E.g., give pile of dolls and blocks
- Fast-mappers have larger vocabs (already spurted)

Early Lexicon: Individual Differences

Language Styles
- Individual differences in styles and strategies
  - Referential style (naming): nouns, single-word utterances
  - Expressive style (Social): diverse vocabs, multiword utterances
- Sources of differences:
  - Input: caregivers of ref kids label ↑ objects
  - Child’s strategy: language as labeling vs. social device
- Significance: one not better (ref kids don’t learn words faster)
- Single word utterances and usually verbs*

Expressive Child

The child is using language in a different way that may not always be correct (go “drown” when they mean swim)*

Rate of Vocab Acquisition

Environmental influence
- Frequency: ↑ words in environment = ↑ child vocab
  - #1 predictor of vocab
- SES (Social Economic Status): #1 predictor of frequency
- Birth order: 1st borns reach 50-words earlier (no diff at 100 words)
- # of siblings: as sibling # ↑, variation in vocab ↓
- Sex: caregivers talk more to females
- We start to see differences in merge in children once they reach the point where they speak 100-1000 of words*
- There is a 30 million word difference from lower SES to higher SES, this is because lower SES parents have to work a lot of hours that limit their ability to talk to their children, and the quality of resources of child in their life (schools)*
Child influence
- Phonological memory:
  - Kids with w/↑ phon memories learn ↑ words, even when controlled for IQ and vocab size
- Sex diff: F > M, but diff is very small (2% of variance)
  - Females mature faster
    - Caregivers talk more to F

Concept Formation

Follows systematic pattern:
- Order reflects complexity (easier concepts learned earlier)
- Definitions develop towards adults’ in systematic ways
- E.g., characteristics for set of objects
Clarke’s Component-by-Component Theory
- Learn semantic features one at a time
- Early on, few features for many words → ↑ overextension
- As concept features ↑, overextensions ↓
- Pro: predicts how kids learn referential concepts
- Cons: can't explain other concepts (e.g., verbs)
Developmental Evidence
- Overextension: use word for overly large class of items (missing necessary feature)
  - E.g., ‘dog’ = 4-legged animal (call a horse a dog)
- Underextension: use word for fewer items than appropriate (attach unnecessary feature)
  - E.g., ‘dog’ = only their stuffed animal

Word Learning

Induction Problem
- When I point to and how do I know what is referent?

Word Learning Strategies

Whole object constraint
- Word refers to entire object and not a part
  - E.g., ‘tree’ → whole tree and not branch or leaf
Taxonomic constraint
- Word refers to thing of same kind
  - ‘This is a sud’
  - ‘Find another one that is the same as the sud’
Mutual Exclusivity
- New words refer to new things
  - ‘Give me the chromium one’
    - Already have name for ‘blue’. So must be… ^this one^
  - ‘Give me the blicket’
Principle of Conventionality
- Use conventional labels for things
- Prefer specific rather than general items
  - ‘Zebra’ instead of ‘animal’
  - ‘Coat’ instead of ‘clothing’

Word learning: Methods

Looking Behaviors
- Preferential Looking Paradigm (PLP)
- Eye-Tracking
Behavioral Tasks
- E.g., forced choice, productive task

Word learning Cues

Frequency: ↑ words = learning
- Cross-situational learning: hearing word in different contexts helps learning
Sociopragmatic Cues (Eye Gaze, Joint Attention, Pointing)
- Use eye gaze to infer meaning by 13 months
- Sample by Task:
  - Ex. look at novel object w/ child & label it Vs.
  - Look away & label it
  - Only map word w/joint attention
- “The baby needs to follow both the pointing and the gaze of the person speaking” “He won’t know the meaning of the object because the lady wasn’t looking” *
Morphological Cues
- Word endings modify meaning
  - -ing & -ize: verbs
  - -ish: adjectives
- Even modify endings to coin new words
  - E.g., noun → verb: ‘brooming”
- “These are blickets” vs. “These are blickish"
Syntactic Cues
- In English, word order cues subject & object (SVO)
- ‘Big Bird is graphing Cookie Monster’

Word Learning Challenges

Relative concepts (meaning depends on context)
Ambiguous words (non-literal): joke, sarcasm, metaphor
- May laugh, but don’t get why funny
- ‘E.g., kids take things seriously like the song “I didn’t know that I wanted until I tasted you”’.

Syntax and Morphology

Early Syntax

Considerable differences in rate of development
- Age is a poor index

Measuring Early Syntax

MLU (mean length of utterance)
- Take 100 spontaneous utterances
- Count # morphemes per utterance
  - More cookie = 2: baby sleeps = 3
  - Birthday = 1 (unless understand birth & day)
- Calculated mean
- More predictive index of language growth than age
Brown’s Grammatical Stages - 5 MLU-defined stages
- Stage I
  - MLU: 1 - 2
  - Word combos
- Stage II
  - MLU: 2 - 2.5
  - Grammatical morphemes
- Stage III
  - MLU: 2.5 - 3
  - Sentence modalities

Stage 1

Word combos (1 -2 MLU)
- Semantic categories appear
  - Agent + action/object___ “ Dog barks”
  - Action + object/location_ “ walk street/eat grape”
  - Possessor + object “ dog’s bone”
  - Demonstrative + object__ “ this phone”
  - Recurrence_ “ more juice/’nother cookie”
  - Nonexistence (very common)____ “ allgone ball”
- First multi-word utterances
  - Not imitations or random combos
  - Indicate grammatical knowledge
    - 1. Follow native word order
    - 2. Use content, but not function words (more informative)

Stage II

Grammatical morphemes (2-2.5 MLU)
- E.g. “Daddy driving”, “my cars”
- Add complexity
- First acquired (in order):
  - Progressive: -ing (e.g., singing)
  - Prepositions: in, on (e.g., in cup)
  - Regular plural: -s (e.g, books)

Stage III

Sentence modalities (2.5 - 3 MLU)
- Negatives
  - 1. “No” in front of utterance to negate it
    - E.g., “no Hannah sleep”
  - 2. Use negative (no, not, can’t, don’t) between S & V
    - E.g., “Hannah no sleep”
  - 3. Learn rule, but slip w/double negatives
- Interrogatives (questions)
  - Use rising intonation
    - E.g., “Mommy cup?”
  - Then, “can, will, are” in order
    - E.g., “Are you sad?”

Syntax Comprehension

Test understanding of sentences
- Watch 2 reversible scenes
- Hear sentence matching scene
  - E.g., “The Bird is carrying the Horse”
  - If understand S and O: ↑ looking to matching scene
    - 17 months!

Comprehension

Eng kids use 2 strategies to infer meanings of sentences:
- 1. SVO strategy
  - Interpret 1st noun as subject & 2nd as object
  - BUT → wrong interpretation of passive
2. Event probability
- Use knowledge of real-life events
  - “The mouse was chased by the cat” (Mice don’t chase cats)
  - “Mary baked John a cake” (Mary can’t bake John)

Brown’s 14 morphemes

Longitudinal study of grammatical morpheme production
Findings:
- Order of acquisition highly similar
Order of acquisition
- 1. Progressive (-ing): singing, playing
- 2 & 3 prepositions (in, on)
- 4 regular plural