2016 ACS Yiddish + Other West Germanic languages

That's really disappointing, but it's a technically accurate classification. Yiddish and PA Dutch are Germanic languages. The ACS tends to create overly broad and often less accurate language groups in order to get the numbers high enough to publish - grouping all the Native American languages together for example. But the grouping of Yiddish is part of a changing demographic landscape. This year is the first time "African" is broken up into more than one language.

You could still use PUMS though you won't get the fine level of detail.

Checked in with some colleagues and here is some further information. -Justin

From the Education and Social Stratification Branch at the Census Bureau: Yiddish, Hungarian, Scandinavian languages, Thai, and Lao were combined with other languages in B16001 this year because the number of speakers for those languages was less than many other individual languages in the US. This was a very difficult decision. Imposing the 200,000 speaker threshold on our language categories was done for disclosure avoidance reasons as well as to highlight other languages that are more in need of translation services in the US. Categorization decisions were based on linguistic groupings and we did not attempt to factor in other measures of cultural similarity or difference. Remember that if you are looking for languages that are not on B16001 you should use the PUMS files, where you can find data for Yiddish and 132 other languages at the state and PUMA levels.

Checked in with some colleagues and here is some further information. -Justin

From the Education and Social Stratification Branch at the Census Bureau: Yiddish, Hungarian, Scandinavian languages, Thai, and Lao were combined with other languages in B16001 this year because the number of speakers for those languages was less than many other individual languages in the US. This was a very difficult decision. Imposing the 200,000 speaker threshold on our language categories was done for disclosure avoidance reasons as well as to highlight other languages that are more in need of translation services in the US. Categorization decisions were based on linguistic groupings and we did not attempt to factor in other measures of cultural similarity or difference. Remember that if you are looking for languages that are not on B16001 you should use the PUMS files, where you can find data for Yiddish and 132 other languages at the state and PUMA levels.

Hi Justin,
Thanks for the explanation, it will be helpful for us in explaining the change to others.

Both of these groups are so geographically concentrated that using state and PUMA levels isn't very useful. In some census tracts in Brooklyn and elsewhere, Yiddish is the dominant spoken language, more than English. Though I don't believe Pennsylvania Dutch has this level of predominance anywhere, census tracts where the current category of "Other West Germanic languages" have relatively high values line up perfectly with the locations of Amish private schools.

Since these groups are so concentrated, and have so little overlap, I have to imagine it won't impact disclosure avoidance too much; the new category will still have similarly low numbers outside their enclaves.

As I wrote, it will probably be pretty easy for policymakers and researchers on the ground to know which group exists in a heavily represented area, but it'll just make it that much more difficult to explain to people not familiar with the communities.

All that said, I am excited about the new language categories you've added, which will help serve those growing communities (which I imagine are more comprised of recent immigrants than the Yiddish and Pennsylvania Dutch speaking communities; I'm curious to see how their English-speaking ability compares).

I'd also add that the Census documentation on the comparability of languages from 2015 to 2016 is very useful: www2.census.gov/.../2016_Language_User_Note.pdf