SLHS 3131 MOD 2 papers - methods and conclusions

Lew-Williams et al., 2011 (Isolation)

METHOD: English-speaking babies (8-10m) listen to Italian. Looking times measured

FINDINGS: stat. significance in looking time difference for high TP words when the words are isolated for them

CONCLUSION: Isolated words help with segmentation when words have high TP and enhance statistical language learning in infants

Graf Estes & Lew-Williams, 2015 (Transitional probability)

METHODS: (8-10m) looking times male vs. female voice; frequency (# of times presented) must be controlled when looking at high (8 speakers) vs low (2) TP

CONCLUSION: Babies track TPs better when there is more acoustic variation (more speakers). More voices = variation = better segmentation

Aravind et al., 2018 (Fast mapping)

METHODS: novel word tests on preschool aged children; 1) propose (tests fast mapping) 2) ME application (verifies FM)

FINDINGS: The ability to make a good initial hypothesis was a function of various factors, including language ability and experience, but across-the-board, children were no better than chance after a wrong initial hypothesis

CONCLUSION: Fast mapping depends on age, and it's inconclusive whether or not they do this robustly. Children fast map and then use multiple exposures to "propose but verify". Signs of cross-situational learning

Singh et al., 2012 (segmentation)

METHOD: longitudinal study of infant (7m) word segmentation

CONCLUSION: Word segmentation abilities at 7 months can predict vocab size at 24 months (correlation vs causation)

*Caveat: bilingual children start with smaller vocabs, and children can catch up later on

Schneidman et al., 2013 (cultural, x-speaker households)

METHOD: infants from single-speaker and multi-speaker households recorded every 4 months (tests predicted vocab for multipeaker household 3.5yr children)

CONCLUSION: Directed language, especially from the primary caregiver, is most important for early word learning, even if child is exposed to overheard speech.

Haebig et al., 2017 (cultural, learning mechanisms, children with ASD and DLD)

METHODS: Statistical learning was assessed through a word segmentation task and fast-mapping (eye-gaze) was examined in an object-label association task (school aged children)

CONCLUSION: some, but not all, children with DLDs struggle with statistical learning and word segmentation; Children with Specific Language Impairments struggled more than those with Autism

Children with DLDs did fine when fast mapping and word segmentation were combined

Antonvich & Graf Estes, 2018 (cultural, TP, mono- & bilingual infants)

METHOD: assessing monolingual and bilingual 14-month-olds' abilities to segment two artificial languages using transitional probability cues

CONCLUSION: Early bilingual exposure prepares infants to navigate challenging aspects of dual-language learning, particularly with segmenting multiple speech streams

Wasserman et al., 2019 (cultural, associate learning in other animals)

CONCLUSION: Associative learning mechanisms in other animals

Cross-situational learning = multiple exposures, so brain can test out different hypotheses and choose the best association with the most evidence (probabilities again!)

Pigeon vs. Human

pigeons' associative learning shows many hallmarks of learning including categorical coherence, generalization as well as elimination of error factors.

Pigeons sort photographs into human language categories

learning may be determined both by strengthening correct stimulus-response associations and by weakening incorrect stimulus-response associations

Kandhadai et al., 2017 (data literacy, ME)

Bilingual and monolingual infants differ in how they interpret a novel word that is applied to a familiar object.

Bilingual infants interpret a novel label that is applied to a familiar object as a second category label.

Monolingual infants interpret the novel word that is applied to a familiar object as a salient property of the object.