家庭環境影響幼兒語言發展因果關係之研究

Causal Relationships between Aspects of Family Environment and Language Development in Young Children

本研究依據文獻分析結果，提出家庭環境影響幼兒語言發展的因果模型，並採用臺灣幼兒發展調查資料庫（KIT）中的36與48月齡的二波的資料，來驗證之。其中男童966人（51.3%），女童916人（48.7%）。並利用偏最小平方結構方程模型（PLS-SEM）來分析。結果顯示:在控制幼兒性別、父母親教育程度等變量下，結果發現有五：一、36月齡學習刺激對48月齡的讀寫萌發則達顯水準；二、36月齡的學習材料對48月齡的語言發展的三個層面都達顯水準；三、36月齡的條件式管教對48月齡語言發展的三個層面上都沒有顯著；四、36月齡父母回應對48月齡語言發展的三個層面上都沒有顯著；五、36月齡的環境多樣對48月齡對語言理解與讀寫萌發等二層面都達顯水準。依據此等發現，提出建議：一、多給予幼兒學習上的刺激，以增加讀寫與識字能力；二、家庭應提供豐富的學習材料，以增進語言發展；三、布置環境多樣，以增進幼兒的語言理解與讀寫萌發；四、未來研究上，建議深入分析雙語環境在家庭環境與語言發展關係上的調節效果，或進行雙領域的潛在成長模型之分析。

Family environment refers to the parents or primary caregivers provide stimulation and support to children. After a child is born, their physical and mental development is affected by their family environment. One aspect of this development is the development of language, which is increasingly influenced by a child’s family environment as the child grows older. Family environment is positively association with the language ability of young children; for example, the amount of parent–child conversation and interaction a child has can positively predict the child’s later language development. In Taiwan, the levels of family–child interaction are positively correlated with language development in young children. However, a research gap exists; the studies addressing this association in Taiwan have generally used cross-sectional data to analyze the association rather than used longitudinal data to explore a possible causal relationship. In addition to longitudinal data analysis, structural equation modeling (SEM) can be used to analyze and demonstrate causal relationships between variables. Hair et al. (2021) argued that partial least squares SEM (PLS-SEM) is particularly valuable for validating causal models. Studies in Taiwan have generally used statistics relating to family environment and language development to verify their association, and research on causal relationships between the dimensions of the two variables is lacking. Therefore, the present study thoroughly analyzed the causal relationships between the dimensions of family environment and language development. Five dimensions of family environment were measured: Environmental diversity, the presence of learning materials, learning stimulation, parental responsiveness, and the use of conditional punishment. Three dimensions of language development were measured: Language comprehension, language expression, and emergent literacy.
This study used samples obtained from children at 36 months (n = 2,164) and 48 months (n = 2,031) old from the Kids in Taiwan: National Longitudinal Study of Child Development & Care (KIT) conducted by Chang (2019, 2021). In total, 100 samples were used to pretest the reliability and validity of the KIT family environment and language development scale, and the remaining data from the two sample waves (36 and 48 months) were merged in the final analysis. After the data merge, the samples with missing values were deleted. Finally, 1,882 valid samples were obtained for 966 boys (51.3%) and 916 girls (48.7%).
Scores for the family environment and language development scale were extracted from the KIT data obtained from the children at 36 and 48 months old. Items on both subscales (family environment and language development) were rated using 4-point Likert scales. The content of the scale was determined through rigorous processes, such as an expert review, structured interviews, and pretests. Although the family environment and language development scale exhibited content validity and reliability, it did not sufficiently explain the variance in the data and had to be retested. Because the scale tests specific factors that can be tested the validity using confirmatory factor analysis, only 100 samples were used to retest it. PLS-SEM was used to perform confirmatory factor analysis, with statistics software used to analyze small samples (30–100 individuals). The criteria for confirmatory factor analysis are as follow: Convergent validity is that the factor loading (λ) of the observable indicators of the factor should fall between .50 and .95; the average variance extracted should be greater than 50%, and the composite reliability greater than .80, and discriminant validity between factors was measured by comparing the average variance extracted for each factor with the square of the correlation between the factors, and two factors exhibited discriminant validity when the average variance extracted was greater than the square of the correlation between the factors. The results of the confirmatory factor analysis revealed that the family environment and language development scale exhibited high validity and reliability when used with 36-month-old and 48-month-old children.
On the basis of a literature analysis, hypothetical models of causal relationships between the dimensions of family environment and dimensions of language development were developed.
Five dimensions of family environment were measured at 36 months, and three dimensions of language development were measured at 48 months. In total, we tested 15 path models. A significant path coefficient indicated a causal relationship between the two variables in the path model. Three dimensions of language development measured at 36 months were considered the foundations for language development at 48 months and were incorporated into the models as a control variable.
Demographic variables, such as kindergarten attendance, gender, and paternal and maternal education levels were also included in the models as control variables. Because gender and kindergarten attendance were categorical variables, they were included as dummy variables in the 48-month-old data. Gender was coded as follows: 1 = female, and 0 = male. Kindergarten attendance was coded as follows: 1 = attends kindergarten, and 0 = does not attend kindergarten. The education levels of the parents were represented in terms of years of education: Master’s degree or higher qualification = 18 years, bachelor’s degree = 16 years, junior college completion = 14 years, high school completion = 12 years, junior high school completion = 9 years, primary school completion = 6 years.
The causal hypothetical models were tested using PLS-SEM in Smart PLS 3.0 software. The analysis revealed the following results: First, learning stimulation in the family environment at 36 months did not significantly influence language comprehension and language expression at 48 months but significantly influenced emergent literacy at 48 months. Second, the presence of learning stuffs at 36 months significantly influenced language comprehension, language expression, and emergent literacy at 48 months, indicating that the presence of learning materials affects later language development. Third, the use of conditional punishment at 36 months did not significantly influence language comprehension, language expression, or emergent literacy at 48 months, indicating that the use of conditional punishment did not significantly affect language development over 1 year. This indicates that conditional punishment has little effect on language development. Fourth, parental responsiveness at 36 months did not significantly influence language comprehension, language expression, or emergent literacy at 48 months, indicating that parental responsiveness has little effect on later language development. Fifth, environmental diversity at 36 months did not significantly influence language expression but did significantly affect language comprehension and emergent literacy at 48 months, indicating that environmental diversity affects later language comprehension and emergent literacy.
On the basis of these findings, this study proposes the following suggestions: (1) Children should receive stimulation to ensure they learn and have stronger literacy skills. For example, primary caregivers should teach young children about colors, sizes, and shapes or provide children with children’s books to encourage children to learn letters, numbers, and the names of items. (2) Families should provide rich learning materials to promote language development, such as balls, baby walkers, throw pillows, sets of cups, puzzles, soft building blocks, play dough, clay, beads, and drawing materials. (3) Caregivers should ensure environmental diversity to improve children’s language comprehension and emergent literacy. For example, parents should guide their children to engage in indoor and outdoor activities often. They can do so by accompanying them to visit relatives and friends, parks, markets, temples, bookstores, and libraries. (4) Future research should investigate the moderating effect of bilingual environments on the relationship between family environment and language development in depth, and the latent growth model of dual fields should also be analyzed. The findings of the present study can be used as a theoretical framework in future research.