week 2- STATISTICAL LEARNING AND OTHER LEARNING MECHANISMS

Question 1

Variegated babble

Answer 1

moving towards phonological production, combining consonants.

Question 2

what knowledge do children need to acquire language?

Answer 2

Word meanings: specific utterances relate to conventionalised entities
World knowledge: one word relates to a range of entities that share a set of properties
Language-specific knowledge: e.g. /dɔɡ/ and /dʰɔɡ/ are not contrastive in English but /dɔɡ/ and /dɔk/ are

Question 3

without knowledge, what is input?

Answer 3

without knowledge, the input is just an incomprehensible stream of noise
→ A set of automatic learning mechanisms may support the establishment of linguistic knowledge

Question 4

what is speech segmentation?

Answer 4

breaking down the speech stream into segments (words, short phrases, phonemic cues)

Question 5

what factors help to segment words?

Answer 5

• Prosodic cues: knowledge from birth supports segmentation based on stress, e.g. KINGdom, HAMlet
• Phonotactic regularities: infants are sensitive to phonological patterns, e.g. /br/ at onset, /nt/ at offset
• Allophonic variation: infants use allophonic cues to detect word boundaries, e.g. night rates vs. nitrates
• Isolated words: infants’ earliest words tend to be those produced in isolation • Transitional probabilities: infants are sensitive to co-occurring syllables in the input (/pɹɪ/ /tɪ/)
• A combination of the above: different approaches are relevant to different stages of development

Question 6

discuss a study based around prosodic cues.

Answer 6

(Jusczyk, Houston & Newsome, 1999)

• Black room
• Babies sat on parent’s lap
• Two speakers on either side
• red light flashes at the front, centralizes the gaze
• one of the stimuli is played from one of the speakers (kingdom v hamlet)
• baby’s head turns towards the noise
• when the baby is bored their head will turn back to the center
• the other word is then played
• how long will the infant attend to one word compared to the other?
• If they stay attentive to kingdom for longer, then we know they’ve successfully isolated the word.

Question 7

How does an infant even begin to make sense of language?

Answer 7

1. Isolated words in the input support segmentation
   → word learning
2. Reliable prosodic cues in the ambient language support segmentation
   → word learning

Question 8

Input is messy – how do infants know what is important in the ambient language, and what is not?

Answer 8

→ Infants acquiring English: /pæl/ and /pʰæl/ are the same word
→ Infants acquiring Korean: /pæl/ (foot) and /pʰæl/ (arm) are different words

Question 9

discuss a study on discriminating phonemes.

Answer 9

Discriminating Phonemes (Werker & Tees, 1984)
• 12 English-speaking 7-month-olds and 10 English-speaking adults • 5 Thompson-speaking adults
• Participants heard contrasting English (/ba/ vs. /da/) and Thompson (/ˀk/ vs. /ˀq/) consonants
• Infants tested using head-turn paradigm

Young infants can differentiate phonetic contrasts that adults cannot
Over time, they become less able to differentiate phonetic contrasts that do not exist in their ambient language
Experience of the ambient language leads language perception to become language-specific
Phonological knowledge of the ambient language can then go on to guide word learning

Question 10

discuss Phonological Retention in Adult Adoptees

Answer 10

Choi, Broersma & Cutler, 2017
• 29 monolingual Dutch-speaking adults, adopted from Korea; half before 0;6, half after 0;6
• 29 monolingual Dutch adult controls
• All participants trained on [t*] ~ [t] ~ [tʰ] contrast (not present in Dutch)
• Participants tested on ability to produce contrast after 10-12 days of training
• Both groups learned the contrast, but adoptees were faster
• Early-adoptees outperformed both groups
• Phonological learning must take place very early on, before age 0;6 • Phonological knowledge is retained from very early language experience, even if this is never activated through production
• (remember that non-linguistic babble begins at around 0;6)
• These results contradict all expectations from the established literature

Question 11

prosody

Answer 11

the patterns of stress or intonation in a langauge

Question 12

phonotactics

Answer 12

the study of phoneme sequences in a language

Question 13

Allophones-

Answer 13

phonemes that don’t contribute to distinctions of meaning. Eg; spin and pin are allophones of p (they do contrast in Koeran).

Question 14

light L and dark L

Answer 14

• L at the beginning of a word- light L. e.g. light

- L at the end of a word- dark L. e.g. bottle

Question 15

Transactional probabilities (statistics

Answer 15

the probability of two things cooccurring. E.g. ‘the’ is usually followed by a noun.