Page 50 - IJPS-11-3

International Journal of
Population Studies Early marriage and birth in Bengali women

proportion of women giving birth before the age of 18 states share similar socioeconomic characteristics, with
declined from 28% to 22% between 1993 and 2006 (Moore a large proportion of their population comprising native
et al., 2009). Nevertheless, in 2019 – 2021, 7% of women Bengali speakers.
aged 15 – 19 had already begun childbearing, marking
only a 1% decline compared with 2015 – 2016 (IIPS & ICF, 2.2. Outcome variables
2022). Among the states, higher levels of teenage pregnancy The outcome variables were (1) early marriage, defined
were observed in Tripura (22%), West Bengal (16%), and as the percentage of women married before the age of 18,
Assam (12%). West Bengal has the country’s highest and (2) early marriage and childbearing, defined as the
incidence of child marriage and teenage pregnancy, with percentage of women married before the age of 18 who had
Tripura ranking second. Meanwhile, Assam ranks fourth begun childbearing (they had either given birth or were
in teenage pregnancy (IIPS & ICF, 2022). pregnant at the time of the survey) before the age of 19.
The states of Assam, Tripura, and West Bengal share 2.3. Predicting variables
a commonality: A significant portion of their population
speaks Bengali. With 272.2 million native speakers, The main predictor variable was women’s culture, which
Bengali is the seventh most spoken language in the world was measured using their native language (Bengali or non-
(STATISTA, 2022), and it is primarily spoken in Bangladesh Bengali). The native language was considered a proxy for
and the aforementioned Indian states. According to the broader social and cultural norms. Studies from India,
2011 Indian Census, 87% of West Bengal’s population Pakistan, and Bangladesh have similarly used language
speaks Bengali, and the figures are 67% in Tripura and 28% as an indicator of cultural identity and practices (Basu &
in Assam (ORGI & MHA, 2011). Amin, 2000; Muhammad, 1996; Routh & Maji, 2021).

This raises a pertinent question: Does language, Several socioeconomic and demographic factors were
which conveys cultural norms and values, influence early also included to analyze the adjusted effect of culture
marriage and childbearing? To the best of our knowledge, on early marriage and childbearing. These included the
no study has explored the potential role of language, as following: Women’s education level (no schooling, up
a representation of a community’s cultural norms, as to 9 years, 10, or more), mass media exposure (no or
a determinant of early marriage and pregnancy in the yes), sex of the household head (male or female), social
Indian context. Thus, this study assesses the role of culture, group (scheduled caste [SC], scheduled tribe [ST], other
as measured through language, in early marriage and backward classes [OBC], non-SC/ST/OBC, and who did
childbearing among Bengali speakers in India. not know their social classification), religion (Hindu,
Muslim, or others), household economic status (poorest,
2. Data and methods poorer, middle, richer, or richest), and place of residence

2.1. Data sources (urban or rural).
This study used data from the fifth round of the NFHS, 2.4. Statistical analysis
conducted between 2019 and 2021. The NFHS-5 is a
nationally representative survey that covers 636,699 The study used bivariate analysis to examine socioeconomic
households and provides data on various health, nutrition, and demographic differences in early marriage and
and women empowerment indicators, including marriage childbearing. To assess the adjusted association between
and fertility. The survey followed a two-stage stratified culture and early marriage and childbearing, the study
sampling method, achieving a 98% response rate. Primary employed binary logistic regression. Multicollinearity was
sample units – survey villages in rural areas and census verified using the variance inflation factor; only predictor
enumeration blocks in urban areas – were selected using variables with acceptable multicollinearity levels were
probability proportion to size sampling. Trained research included in the final regression. Sample weights were
investigators collected data through computer-assisted applied to adjust for non-response. The analyses were
personal interviews, and participants provided either conducted using Stata (Version 15), and the results were
verbal or written informed consent. Only respondents who reported at a significance level of 5%.
voluntarily consented to participate were interviewed. The 3. Results
published report offers more details regarding the survey
design, questionnaire, quality control procedures, and 3.1. Sample characteristics
management information (IIPS & ICF, 2022). Table 1 presents the socioeconomic, demographic, and
The study sample included 51,069 married women health profiles of the women aged 15 – 49 years included in
aged 15 – 49 from Assam, Tripura, and West Bengal. These the analysis. Among the sample, 46% were married before

Volume 11 Issue 3 (2025) 44 https://doi.org/10.36922/ijps.2068