Acceso usuarios
[1]
;
Hammarström, Harald
[1]
Many national censuses collect data on language use, offering valuable insights for tracking language vitality and guiding policy decisions. However, little is known about the reliability of these data. This article proposes a framework for assessing the reliability of census-based speaker counts, focusing on three key aspects: coverage, accuracy, and consistency. We apply this framework to Canadian census data from 2001, 2006, 2011, 2016, and 2021, examining nearly 60 Indigenous languages. Our analysis shows significant improvements in coverage over time, particularly for languages with small speaker populations or those previously grouped under broader ‘macrolanguages.’ However, we find that the census tends to overestimate speaker numbers for less commonly spoken languages and underestimate them for more widely spoken ones, especially between 2006 and 2016. Overall, the consistency of census data aligns with that of traditional sources. In light of our results, we recommend prioritizing more recent census data and exercising caution when using figures for languages that were previously grouped under macrolanguages.
