2022-2023 Doctoral Awardee: Alex Silver

Social And Linguistic Influences on Number Word Acquisition

Alex Silver, University of Pittsburgh

Project Abstract

Early numeracy skills are important for mental and physical health later in life, yet over the last few years children’s mathematical knowledge has dropped at alarming rates. Parental talk about numbers affects math performance in preschool- and school-aged children. However, how parental input relates to math skills in infants is unclear, particularly as infants learn that each number word represents a set of a particular size, a concept foundational for later math skills. Prior to understanding how all number words refer to quantities, infants have some recognition of numbers and can use linguistic cues to infer quantity. Whether this reflects genuine numerical understanding and whether infants can map number words to quantities without other cues remains unknown. The goal of this research is to investigate the process by which infants learn number words and test the role of parental input in this process. These results will have implications for understanding the development of mathematical thinking and variability in children’s earliest math skills that will allow us to make recommendations for home environments to support math learning.


Mathematical knowledge is dropping at an alarming rate, with math scores on national tests declining for the first time since the 1970s [1], and low-performing students showing worse declines than their higher-achieving peers in math [2]. Critically, these early numeracy skills are important for health and financial outcomes later in life [3-5]. Parental input, particularly via talk about numbers, affects math performance in preschool- and school-aged children [6]. However, little work has examined how parental input is associated with developing math skills in infants and toddlers, particularly as they learn that each number word represents a set of a particular size, a concept that is foundational for the development of later symbolic math skills [7]. Prior to understanding how all number words refer to quantities, infants have some recognition of the count list, and toddlers have some recognition of number word meaning [8, 9]. Infants can also use other linguistic cues to infer quantity [e.g., number marked by singular or plural verbs, quantifiers, and noun morphology; 10]. However, whether this reflects genuine numerical understanding and whether infants can map number words to quantities without these other cues remains unknown.

The overall goal of this project is to test the process by which infants learn to map number words to their referred quantities and the role of parental input in this process. Specifically, I will test infants’ number word comprehension in a preferential looking paradigm where infants view two images containing sets of identical objects differing in quantity and are prompted to look at one of the images. I will then test the association between infants’ number word comprehension and parents’ number talk (i.e., use of number words) during a semi-structured observation with their infants.

Aim 1: Test whether linguistic cues are necessary to map number words to referred quantities. I will investigate whether infants aged 18-24 months can map number words to their referred quantities in a preferential looking task where linguistic cues to quantity are present for half of trials (e.g., “Where are two birds?” and “Where is one bird?”) and no linguistic cues beyond the number word are present for the other half of trials (e.g., “Which has two fish?” and “Which has one fish?”). If infants understand number words, their looking time to the target should be above chance for all trials, whereas if they require redundant linguistic cues to aid in number word comprehension, looking time to the target should be above chance only on trials containing redundant cues to differentiate the quantities, and at chance on trials with no linguistic cues.

Aim 2: Test whether parental input explains individual differences in number word comprehension. I will explore whether parents’ number talk to their infants during a semi-structured observation where they view a picture from a storybook and discuss it is predictive of individual differences in infants’ number word comprehension, above and beyond parents’ overall talk and family demographic factors. I hypothesize that parents’ number talk will be positively related to infants’ number word comprehension.


Sample size was determined based on power analysis with effect sizes from past literature for the associations between linguistic cues and infants’ looking time, as well as for associations between parents’ number talk and children’s math performance. Participants will be 100 monolingual English-speaking infants aged 18-24 months born full-term (>37 weeks gestation). Each infant will participate with one parent and dyads will be compensated for their time. Participants will be recruited via Lookit, an unmoderated online research platform that families can access anywhere with an internet connection, allowing us to reach families from across the United States, broadening the diversity of our sample and gaining a more naturalistic picture of the process in question than traditionally possible in lab-based settings [11]. Infants will complete a measure of number word comprehension and a semi-structured observation with their parent, both of which will be videorecorded for later coding. The order of the number task and the parent-child observation will be counterbalanced across participants.

The number word comprehension preferential looking paradigm timing and presentation of stimuli is based on the design used in the most relevant prior study [10]. Importantly, since that study did not test number word comprehension, I have begun piloting the paradigm and will finetune it if necessary. Infants will view two sets of identical objects on the computer screen (either sets of birds or sets of fish) which are commonly known by infants this age [according to WordBank; 12]. Infants will be prompted to look at one set, and the prompts for each trial will vary in the presence or absence of redundant linguistic cues to quantity. For half of the trials, redundant morphological cues (singular or plural marking of the verb and noun) will be present in addition to the number word (e.g., “Where are two birds?” and “Where is one bird?”), whereas for the other half of the trials no redundant linguistic cues will be present (e.g., “Which has four fish?” and “Which has one fish?”). Infants’ looking time to each set will be coded for each trial.

Parental number talk will be coded from observations of infants with their parents. Dyads will view an image on the computer screen and will be asked to talk about the scene as if it were a picture in a book. As in work using similar parent-child observations [e.g., 14, 15], parents will not be prompted to discuss number, and although the image contains sets of objects that can be counted and Arabic numerals, there is no requirement to talk about numbers and parents may choose to discuss other aspects of the scene. Interactions will be transcribed verbatim and the percentage of number words out of all words used will be calculated [e.g., 16].

Finally, child vocabulary and family demographics will be assessed. Parents will complete the MacArthur-Bates Communicative Development Inventories Short Form Level II [normed for children aged 16-30 months; 17] and report demographic variables including child age and gender.


Preliminary data collection using similar stimuli suggests that my approach to measuring number word comprehension and parental number talk is feasible. Remote preferential looking paradigms to collect data from infants and toddlers have been remarkably successful, replicating results from data collected in controlled laboratory settings [e.g., 18, 19-22] and other remote, unmoderated studies have successfully collected videos of naturalistic parent-child conversations in the home [23].

To address Aim 1, I will first test whether number word comprehension is significantly different from chance for each trial type (i.e., trials containing redundant linguistic cues and trials not containing redundant linguistic cues). I will employ one-sample t-tests (two-tailed) comparing the percentage of looking time to the correct set to chance (50%). I will then conduct a Bayes factor analysis [24] to indicate the relative likelihood of the null hypothesis versus the alternative hypothesis. To address Aim 2, I will test for effects of parental number talk on infants’ number word comprehension using a series of hierarchical linear regression models. I will first regress infants’ number word comprehension (i.e., percentage of looking to the correct image on the preferential looking task) on covariates including infants’ age, infants’ gender, infants’ vocabulary, parent education, and order of tasks (i.e., number task prior to or after parent-child observation). In a second step, I will regress infants’ number word comprehension on parental number talk with those same covariates to test whether parental number talk is a significant predictor of infants’ number word comprehension and explains significantly more variance than the baseline covariates. As for Aim 1, I will conduct Bayesian analyses in addition to frequentist analyses [25].


Results from this project will have significant implications for the conceptualization and understanding of the development of mathematical thinking in typical development and individual differences in children’s earliest math abilities. Findings will advance our understanding of the process by which young children learn number words and the contexts required to assess numerical understanding, and future studies of number skills will be able to incorporate this nuanced understanding of linguistic cues into their study design. Furthermore, results from this project will have direct application for early childhood development both in the home and in childcare settings: Families, educators, and researchers will be able to design interventions and learning materials that incorporate linguistic cues to aid in children’s number word acquisition and subsequent math skills (e.g., materials that highlight the importance of number talk in infancy, emphasizing that children need lots of experience with sets of different objects and quantities to learn number words, and modeling the most effective ways to talk about and use number for infants’ learning). Given the urgent need to improve early math skills, these approaches will be particularly important for supporting math learning for children who are at risk for math learning disabilities (e.g., children born preterm).

Impact Statement

Math ability is a critical predictor of academic, financial and health outcomes, yet math knowledge is dropping at an alarming rate in the past few years, with the COVID-19 pandemic exacerbating math achievement gaps. Urgent intervention is needed, as variability in math is present by the beginning of formal education and remains remarkably stable. This project will test whether linguistic cues help infants learn number words, a foundational math skill required for later math development, and will examine how parents’ support for infants’ math learning is associated with infants’ number word knowledge. Given the urgent need to improve math skills and promote early math learning environments prior to school entry, understanding the strongest predictors of infants’ number word knowledge will help us identify which targets will be most successful to support young children beginning to learn math and particularly those at risk of poor math achievement.


  1. Mervosh, S., The Pandemic Erased Two Decades of Progress in Math and Reading, in The New York Times. 2022: New York.
  2. Meckler, L., Scores fall coast to coast, especially in math, under pandemic’s toll, in The Washington Post. 2022: Washington, D. C.
  3. Reyna, V.F. and C.J. Brainerd, The importance of mathematics in health and human judgment: Numeracy, risk communication, and medical decision making. Learning and Individual Differences, 2007. 17(2): p. 147-159.
  4. Duncan, G.J., et al., School readiness and later achievement. Dev Psychol, 2007. 43(6): p. 1428-1446.
  5. Agarwal, S. and B. Mazumder, Cognitive Abilities and Household Financial Decision Making. American Economic Journal: Applied Economics, 2013. 5(1): p. 193-207.
  6. Elliott, L. and H.J. Bachman, How Do Parents Foster Young Children’s Math Skills? Child Development Perspectives, 2018. 12(1): p. 16-21.
  7. Geary, D.C. and K. vanMarle, Growth of symbolic number knowledge accelerates after children understand cardinality. Cognition, 2018. 177: p. 69-78.
  8. Wang, J.J. and L. Feigenson, Infants recognize counting as numerically relevant. Dev Sci, 2019. 22(6): p. e12805.
  9. Wagner, K., J. Chu, and D. Barner, Do children’s number words begin noisy? Dev Sci, 2019. 22(1): p. e12752.
  10. Kouider, S., et al., Acquisition of English Number Marking: The Singular-Plural Distinction. Language Learning and Development, 2006. 2(1): p. 1-25.
  11. Scott, K. and L. Schulz, Lookit (Part 1): A New Online Platform for Developmental Research. Open Mind, 2017. 1(1): p. 4-14.
  12. Frank, M.C., et al., Wordbank: an open repository for developmental vocabulary data. J Child Lang, 2017. 44(3): p. 677-694.
  13. Libertus, M.E. and E.M. Brannon, Stable individual differences in number discrimination in infancy. Dev Sci, 2010. 13(6): p. 900-6.
  14. Bachman, H.J., et al., Triangulating multi-method assessments of parental support for early math skills. Front Educ (Lausanne), 2020. 5.
  15. Duong, S., et al., What’s in a question? Parents’ question use in dyadic interactions and the relation to preschool-aged children’s math abilities. J Exp Child Psychol, 2021. 211: p. 105213.
  16. Elliott, L., E.J. Braham, and M.E. Libertus, Understanding sources of individual variability in parents’ number talk with young children. J Exp Child Psychol, 2017. 159: p. 1-15.
  17. Fenson, L., MacArthur-Bates communicative development inventories. 2007, Baltimore, MD: Paul H. Brookes Publishing Company.
  18. Scott, K., J. Chu, and L. Schulz, Lookit (Part 2): Assessing the Viability of Online Developmental Research, Results From Three Case Studies. Open Mind, 2017. 1(1): p. 15-29.
  19. Smith-Flores, A.S., et al., Online measures of looking and learning in infancy. Infancy, 2022. 27(1): p. 4-24.
  20. Morini, G. and M. Blair, Webcams, Songs, and Vocabulary Learning: A Comparison of In-Person and Remote Data Collection as a Way of Moving Forward With Child-Language Research. Front Psychol, 2021. 12: p. 702819.
  21. Bacon, D., H. Weaver, and J. Saffran, A Framework for Online Experimenter-Moderated Looking-Time Studies Assessing Infants’ Linguistic Knowledge. Front Psychol, 2021. 12: p. 703839.
  22. Nelson, C.M. and L.M. Oakes, “May I Grab Your Attention?”: An Investigation Into Infants’ Visual Preferences for Handled Objects Using Lookit as an Online Platform for Data Collection. Front Psychol, 2021. 12: p. 733218.
  23. Rhodes, M., et al., Advancing Developmental Science via Unmoderated Remote Research with Children. J Cogn Dev, 2020. 21(4): p. 477-493.
  24. Rouder, J.N., et al., Bayesian t tests for accepting and rejecting the null hypothesis. Psychon Bull Rev, 2009. 16(2): p. 225-37.
  25. Greenland, S., Bayesian perspectives for epidemiological research. II. Regression analysis. Int J Epidemiol, 2007. 36(1): p. 195-202.