The more they hear the more they learn? Using data from bilinguals to test models of early lexical development

被引:3
|
作者
Sander-Montant, Andrea [1 ]
Perez, Melanie Lopez [1 ]
Byers-Heinlein, Krista [1 ]
机构
[1] Concordia Univ Canada, Dept Psychol, Concordia Infant Res Lab, 7141 Sherbrooke St West,PY-033, Montreal, PQ H4B 1R6, Canada
基金
瑞典研究理事会;
关键词
Word learning; Maturation; Language experience; Bilingualism; Looking; -while; -listening; Infants; Children; PREFERENTIAL LOOKING PARADIGM; EARLY WORD COMPREHENSION; LANGUAGE EXPERIENCE; ENGLISH BILINGUALS; VOCABULARY SIZE; CHILDREN; INFANTS; EXPOSURE; PERIODS; BRAIN;
D O I
10.1016/j.cognition.2023.105525
中图分类号
B84 [心理学];
学科分类号
04 ; 0402 ;
摘要
Children have an early ability to learn and comprehend words, a skill that develops as they age. A critical question remains regarding what drives this development. Maturation-based theories emphasise cognitive maturity as a driver of comprehension, while accumulator theories emphasise children's accumulation of language experience over time. In this study we used archival looking-while-listening data from 155 children aged 14-48 months with a range of exposure to the target languages (from 10% to 100%) to evaluate the relative contributions of maturation and experience. We compared four statistical models of noun learning: maturationonly, experience-only, additive (maturation plus experience), and accumulator (maturation times experience). The best-fitting model was the additive model in which both maturation (age) and experience were independent contributors to noun comprehension: older children as well as children who had more experience with the target language were more accurate and looked faster to the target in the looking-while-listening task. A 25% change in relative language exposure was equivalent to a 4 month change in age, and age effects were stronger at younger than at older ages. Whereas accumulator models predict that the lexical development of children with less exposure to a language (as is typical in bilinguals) should fall further and further behind children with more exposure to a language (such as monolinguals), our results indicate that bilinguals are buffered against effects of reduced exposure in each language. This study shows that continuous-level measures from individual children's looking-while-listening data, gathered from children with a range of language experience, provide a powerful window into lexical development.
引用
收藏
页数:15
相关论文
共 50 条
  • [1] Learn More From Your Data With Asymptotic Regression
    Clarke, Alasdair D. F.
    Hunt, Amelia R.
    JOURNAL OF EXPERIMENTAL PSYCHOLOGY-GENERAL, 2025,
  • [2] Using Bayesian parameter estimation to learn more from data without black boxes
    Rachel C. Kurchin
    Nature Reviews Physics, 2024, 6 : 152 - 154
  • [3] Using Bayesian parameter estimation to learn more from data without black boxes
    Kurchin, Rachel C.
    NATURE REVIEWS PHYSICS, 2024, 6 (03) : 160 - 161
  • [4] Using Google Trends Data to Learn More About Survey Participation
    Gummer, Tobias
    Oehrlein, Anne-Sophie
    SOCIAL SCIENCE COMPUTER REVIEW, 2023, 41 (06) : 1968 - 1985
  • [5] Using PIAAC data to learn more about the literacy practices of adults
    Nienkemper, Barbara
    Grotlueschen, Anke
    INTERNATIONAL JOURNAL OF LIFELONG EDUCATION, 2019, 38 (04) : 393 - 405
  • [6] MORE FROM ROZANSKA ON EARLY SPEECH DEVELOPMENT
    ROZANSKA, E
    VOLTA REVIEW, 1967, 69 (05) : 294 - 295
  • [7] Enhancing Synthetic Test Data Generation with Language Models Using a More Expressive Domain-Specific Language
    Tan, Chao
    Behjati, Razieh
    Arisholm, Erik
    TESTING SOFTWARE AND SYSTEMS, ICTSS 2023, 2023, 14131 : 21 - 39
  • [8] Using more realistic data models to evaluate sensor network data processing algorithms
    Yu, Y
    Estrin, D
    Rahimi, M
    Govindan, R
    LCN 2004: 29TH ANNUAL IEEE INTERNATIONAL CONFERENCE ON LOCAL COMPUTER NETWORKS, PROCEEDINGS, 2004, : 569 - 570
  • [9] What do Models Learn From Training on More Than Text? Measuring Visual Commonsense Knowledge
    Hagstrom, Lovisa
    Johansson, Richard
    PROCEEDINGS OF THE 60TH ANNUAL MEETING OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS (ACL 2022): STUDENT RESEARCH WORKSHOP, 2022, : 252 - 261
  • [10] Learn more from less: Generalized zero-shot learning with severely limited labeled data
    Lu, Ziqian
    Lu, Zheming
    Yu, Yunlong
    Wang, Zonghui
    NEUROCOMPUTING, 2022, 477 : 25 - 35