Week 9-Language Development 2 Flashcards
What is segmentation and how do infants try to resolve this in word learning?
S=learning words from a continuous stream of speech with no gaps
*They use transitional probabilities to know when one word ends and another begins (Saffran, Aslin & Newport, 1996)
– prettybaby, prettydress, prettyflower
– Pri-tee is probably a word
– Tee-bay, tee-dress and tee-flow
probably aren’t
How did Saffran, Aslin &
Newport (1996) investigate segmentation in infants?
-Used the Head-Turn Preference Procedure in 8-month-olds
-Infants exposed to continuous stream of syllables for 2 minutes
-Then given choice of whether to
listen to words or non-words
(Experiment 1) or words or part-
words (Experiment 2)
What occured in Saffran, Aslin &
Newport’s (1996) first experiment + results?
-Aim:Can infants distinguish
words (e.g., golatu) and non-words
(e.g., lagotu)
*Words=Sequences where 1st syllable (e.g., go) always followed by 2nd syllable (e.g., la) always followed by 3rd syllable (e.g., tu)
-Transitional probabilities=1.00
*Non-words=Sequences where
syllables never followed each other (distorted/messed up version)
–Transitional probabilities=0.00
Results:Infants listened longer too (were more interested in) non-words
What occured in Saffran, Aslin &
Newport’s (1996) second experiment + results?
*Aim:Can infants distinguish words (e.g., golatu) and part-words (e.g., latupa)
*Part-words = Sequences where 3rd syllable only followed 2nd syllable 1/3rd of the time (latupa/latuda/latuti)
–Transitional probability=0.33
-Results:Infants listened longer too (were more interested in) part-words
What did Saffran, Aslin &
Newport (1996) find overall in their experiment on transitional probabilities/segmentation?
*8-month-olds can discriminate 3-
syllable sequences with high
transitional probabilities (Words=1.00 ) from:
*Sequences they have never heard
before (Non-words = 0)
*Sequences with lower transitional
probabilities (Part-words=0.33)
*Infants can use transitional
probabilities to find words in the
speech stream
What did Fló, Brusini et al.
(2019) find how early infants are sensitive to transitional probabilities?
*Used functional Near-Infrared Spectroscopy (fNIRS) to look at this in newborns
*Played newborns 3 mins of a stream of syllables
*Found that newborns showed different patterns of brain response to words and part-words
How do infants move beyond transitional probabilities?
*Transitional probabilities are a good way to break into the speech signal
*Dominant stress pattern of English words is TROCHAIC (STRONG-weak, e.g., CAN-dle, DOC-tor)
*Infants soon learn to ignore
transitional probabilities in favour
of language-specific cues (e.g.,
stress)
How did Jusczyk, Houston &
Newsome (1999) test trochaic stress pattern in infants?
*Played ‘English-speaking’ 7.5-
month-old infants passages
containing 2-syllable words
*Some words had the trochaic stress pattern typical of English (e.g., CAN-dle) but others had an iambic stress pattern (e.g., Gui-TAR)
What did Jusczyk, Houston &
Newsome (1999) find in relation to infants and the trochaic stress pattern?
*Words then presented separately
* Infants listened longer to the words they had heard embedded in the longer passages than to isolated words (e.g., CAN-dle v DOC-tor)
*But only if the target words had the trochaic stress pattern
What did Gerken (1994) suggest the trochaic stress pattern could explain?
*It can explain the kind of segmentation errors we see in children’s early speech:
– NA-na for ba-NA-na
– PU-ter for com-PU-ter
– RAFFE for gi-RAFFE
When are infants sensitive to language-general/language-specific cues?
L-G(transitional probabilities):from birth
L-S(stress):from 7.5 months
How is learning word order rules a challenge for infants?
*SVO in English=Big Bird daxes Cookie Monster
*VSO in Welsh=Daxes Big Bird Cookie Monster
*How do children know who is doing what to whom?
* When do they develop verb-general understanding?
SVO=Subject-Verb-Object
Explain Tomasello’s (1992) Verb-Island Hypothesis:learning word order rules
*Use of new verbs started simple regardless of how complicated sentences with old verbs were e.g.,
Mummy cut bread vs Draw
*Travis did not have ‘joined-up’
knowledge of word order but instead had islands of knowledge about how to use and understand particular verbs such as cut
*2-year-olds do not have verb-general knowledge of word order
How did Akhtar & Tomasello
(1997) test the verb-island hypothesis?:learning word order rules
*Taught children novel verbs without any info about word order
–This is wugging
*Asked children to act out sentences with novel verb in SVO word order
–Make Big Bird wug Cookie Monster
What did Akhtar & Tomasello
(1997) find testing the verb-island hypothesis?:learning word order rules
*Children performed at chance
–Sometimes made Big Bird do the novel action to Cookie Monster
–Sometimes made Cookie Monster do the novel action to Big Bird
*Concluded that children did not have verb-general knowledge of word order
How did Gertner, Fisher &
Eisengart (2006) simplify testing verb-general knowledge of word order in 2 year olds?
*Maybe Act-Out test is too
difficult for young children
*They used preferential-looking
technique to measure
looking time to the correct
v the incorrect alternative as a simpler technique
What did Gertner, Fisher &
Eisengart (2006) find about knowledge of word order in infants?
*Both 21- and 25-month-olds showed verb-general understanding of word order
*21-month-olds with Boys and Girls
*25-month-olds with Bunnies and Ducks
*Even 21-month-olds know that the first-mentioned Noun/subject (e.g., Boy is the Agent or Do-er of the Action)
Toddlers have to learn how to mark for what verbs?
–Agreement (e.g., she play-s)
–Tense (e.g., she play-ed yesterday)
What kind of mistakes do children make in verb marking and why? What did Brown (1973) find?
*Looked at early multi-word
speech of 3 children in Boston
*Early speech was ‘Telegraphic’
(only including the most important
words like in a telegram)
*Verb-marking was often absent too e.g., We go to the park yesterday
*Correct use of verb marking develops only gradually
*Adam started using s on verbs at around age 2.5but did not use it
in 90% of obligatory contexts until age 3.5
What is Developmental
Language Disorder (DLD)?(previously called Specific language impairment (SLI))
*Defined as a language deficit (1.5+
SDs below the mean) in the absence of any known sensory or cognitive deficits
*Norbury et al. (2016) report
prevalence of around 7.5% (8% boys and 6% girls) in the UK (equivalent to 2 children in
every classroom)
What particular language deficit
problem do English-speaking children with DLD tend to have?
verb marking:
*Rice, Wexler & Hershberger
(1998) report significant deficits
in the provision of 3sg –s (3rd singular) and past tense –ed relative to both age-matched and language-matched controls
Why do chilren drop endings according to Bloom (1990)?
-They lack the processing capacity to include all they know in their speech
*But in other languages children
make errors where they add the wrong (infinitive) ending:
–German Kaffe trink-en
(Coffee to drink)
–French Poupée dorm-ir
(Dolly to sleep)
–Spanish Niños cant-ar
(Children to sing)
What did Freudenthal, Pine,
& Gobet (2010) find on why infants make errors using a computer model called MOSAIC to simulate errors in English, Dutch, German, French and Spanish?
*MOSAIC learns slowly from transcripts of CDS building sentences from the right edge of the utterance (recency effect)
– Mummy likes to drink coffee
– Coffee
– Drink coffee etc.
*Errors because children learn from the right edge of the utterance (recency effect)
*More errors in Dutch and German because infinitives always come at the end (right edge)
*Could not simulate very high rate of errors in English (> Dutch)
What did Rasanen, Ambridge
& Pine (2014) find when testing if English-speaking children replace verb forms they don’t know (well) with forms they do?
*See is much more common than sees in English child-directed speech (CDS)
*BUT fits is relatively frequent in English CDS NOT fit
*The more children heard the
–s (e.g. fits) vs bare (fit) form in the input the more often they correctly produced the –s vs bare form themselves
