Lecture 2 - Aquiring words and their meanings

• Babbling


  Repetitive use of syllables with no recognizable meaning
• Babbling(6-15 months)


  
  

  • Motoric constraints: stops (t a p), nasals (n o se), and glides (y ou, w e) more than fricatives (f eet) and liquids (l ow, r ow)
  • Early phase is "universal" and later phase may reflect attributes of the ambient language
  • Goes from non-referential to referential (intentional?)
  • Deaf babies babble too, orally and manually
• What did Petito and Marentette (1991) find in relation to babbling in deaf and hearing children?

  
  

  • Babbling is an automatic and amodal stage of language development
• What are the two hypotheses arguing whether babbling is a precursor of language?

  
  

  • Discontinuity hypothesis (Jacobson, 1968)
  • Continuity Hypothesis (Mowrer, 1960)
• The DISCONTINUITY hypothesis claims that:

  • babbling is a preprogrammed, muscular exercise (reflexes)
  • all sounds can be babbled
  • silent period -> no correlation between babbling and language onset
• CONTINUITY hypothesis claims that:

  • acoustic analyses - differences in prosody & phonemes between french and german babbling
  • has linguistic manifestations (Holowka & Petito, 2002) LH associated with language, right side of mouth opens more when babbling
• Phonemes


  
  

  • Classes of sounds in a language that affect meaning (ten, den)
  • Toddlers tend to simplify - avoid 'difficult' phonemes (dropping the final C of CVC, substituting stops for fricatives)
  • At 1.5 years, 50-60% of phonemes are articulated
  • At 3 years, 90% of consonants and 100% of vowels are correctly articulated
• The classic method to study early word learning is indirect count and reports (around 9 months)
• Tincoff & Jusczyk (1999) used the preference-looking paradigm and found that 6-month-olds look longer at a picture of 'mummy' when hearing the word 'mummy'
• What does Tincoff's study suggest infants show?

  
  

  • shows specificity - infants know the word 'mummy' refers to a specific female, not just any female
  • infants can attach meaning to a sound string
• Bortfeld et al. (2005) found that 6-month-olds who heard 'mommy' passages showed a preference for 'feet' over 'cup', while those who heard 'Tommy' passages did not show a preference
• This shows that lexical knowledge can help segmentation and learning of adjacent words, and demonstrates phonological specificity of existing lexical knowledge
• The reference problem


  It is difficult to infer a child's intention through their behaviour (e.g. a child holding a duck and saying 'bear')
• Less than 10% of infant-directed speech consists of isolated words
• Segmenting speech into words is a challenge for children
• Statistical learning


  1. Progressive detection of sequential regularities (transitional probabilities between syllables)

  2. Infants associate high transitional probability between syllables as likely words (points of cohesion) and low transitional probability between syllables as likely word boundaries (points of transition)
• Types of words acquired first

  • Nouns (object words)
  • Verbs (action words)
  • Modifiers ('big')
  • Social words ('thank you', 'no')
  • Grammatical function words ('to', 'for')
• Vocabulary growth


  • Comprehension: At 6 months, recognise a few familiar words; at 12 months, avg = 100 words, top 10% ~ 200 words, bottom 10% ~ 50 words
  • Production: At 12 months, mean = 10 words, wide variation - some start only at 20 months; at 24 months, mean = 300+ words, top 10% > 600 words, bottom 10% ~ 150 words
• Factors correlated with vocabulary size

  • Age (but less than 50% variance)
  • Biological factors (synaptogenesis)
  • Other cognitive abilities (e.g., pretend play, theory of mind, symbolism)
  • Environmental factors (talkative parents, ambient speech)
• Comprehension vocabulary correlates more with these factors than expressive vocabulary, because comprehension comes first and other factors can intervene and interfere (articulatory, social) for production
• The lexical spurt might not be a systematic/necessary stage in vocabulary acquisition
• Computational simulations can account for a lexical spurt (exponential growth) without postulating qualitative changes in learning mechanisms
• Exponential learning is observed if: (1) words can be acquired in parallel and (2) few words can be acquired quickly and a greater number of words take longer
• Vocabulary growth


  More with comprehension vocabulary than with expressive vocabulary
• Why is there more vocabulary growth with comprehension than expressive vocabulary?
• Lexical explosion


  Sudden word learning spurt in the second year
• Lexical explosion


  • Initially explained by qualitative changes in learning mechanisms: Naming insight (onset of symbolism), Syntactic development, onset of lexical spurt
• Simulation when learning a word offers a cost or a benefit to future learning

  Exponential learning is observed if: (1) words can be acquired in parallel and (2) few words can be acquired quickly and a greater number of words take longer. Both are reasonable and realistic assumptions. An exponential function is seen even if learning of word leads to a cost in learning other words.
• Multiword utterances


  Combinations of words occur with an expressive vocabulary of 50-100 words
• Mean Length of Utterance (MLU)

  The number of words in a sentence even if the sentence is syntactically incorrect (e.g., "no eat" = 2)
• Children differ in the age at which their sentences start to increase in length, independent of their vocabulary size
• At 24 months onward, longer MLU comes with an increase in syntactic complexity
• Short MLU at 24 months


  • Hot. This hot. This hot this time.
  • Mommy, help me. This hot.
  • Cinnamon on them.
  • Ow. Hot! This burn my hand.
  • Mommy, put butter on mine.
• Longer MLU at 36 months


  • Mom fixed this for me and I don't like it.
  • Abe's gonna eat rest of it.
  • I like a mayonnaise sandwich.
  • Mom will come home and not like it.
  • Put it right here so I can eat it.
• Multiword utterances at 24 vs 36 months


  • Few function words, Single-clause sentences, Present tense
  • More function words, Multiple-clause sentences, Various tenses
• Semantic development


  How do children attach meaning to words?
• Nouns are often 'referential' (physical representation), verbs are not 'do it' 'eat up' meaning of verbs can be inferred from syntactic context
• Syntactic bootstrapping


  The meaning of words (and verbs in particular) can be learned from the syntactic structure of the carrier sentence
• Whole-object bias


  • Words refer to objects, not object parts. Teaching features is hard!
• Mutual Exclusivity


  • If you already know the name of an object, a novel word must refer to something else. Learning two languages is cognitively counter-intuitive!