Natural Sciences Meet Social Sciences: Census Data Analytics for Detecting Home Language Shifts

As we are living in a global environment, it is not unusual to have more than one languages or dialects used in a country. Examples include Canada in the Americas, Singapore in Asia, and Switzerland in Europe. With the initiatives of globalization, many people immigrate or live in a country other than their birthplace. As a result, different people in the same country may have different home language (i.e., first language). For instance, as a nation composed of a highly diverse language population, Canada provides a unique opportunity to study the factors causing certain languages (or families of language) to be lost over subsequent generations among allophones (i.e., people whose mother tongue is neither English or French). In this paper, we focus on census data analytics. Specifically, we analyze census microdata by exploring machine learning and data mining techniques-such as decision tree induction, random forest, and categorical naive Bayes-to study the influence of various social and economic factors on the probability that allophones adopt official languages as their language spoken at home. This study is a showcase where natural sciences and engineering (NSE) meet social sciences, in which NSE solutions (e.g., census data analytics) are applicable for the study of social science related phenomena (e.g., successful detection of shifts in home languages).