A century of U.S. censuses and the language characteristics of immigrants

Since 1890, every U.S. census but one has asked about the language characteristics of the U.S. population. This almost uninterrupted data series, however, has been shaped by contemporaneous presumptions about the ties between language and ethnicity, the likelihood of proficiency in English among various subgroups, and practical constraints. I describe shifts across censuses in the phrasing of questions about language, the coding of responses, and the subpopulations for which the questions were asked and the results were published. I then describe the data generated by these items and discuss their interpretation. I conclude with a summary of the major insights and limitations of a century's worth of data.