In the languages New Yorkers speak at home, sized by number of speakers
American Community Survey, 2022 one-year estimates
Methodology and sources
What this shows
Each sentence reads "I live in New York City" written in a language spoken by New Yorkers at home. The font size of each sentence is proportional to the number of speakers of that language in a given decade.
Sizing
Font size is scaled using a cube-root function of speaker count. A purely linear scale would make any language under ~100,000 speakers invisible next to English (4+ million) or Spanish (1.8+ million). The cube root preserves the rank order and relative magnitude while keeping smaller languages legible.
Data sources
1990: Decennial Census, tabulated by NYC Department of City Planning (Population Division, April 2004 report).
2000: Decennial Census, Summary File 3, Table PCT010. The last decennial census to include detailed language questions.
2010: American Community Survey (ACS) 2006-2010 five-year estimates, Table B16001, supplemented by ACS 2009-2013 detailed language tables for finer breakdowns.
2022: ACS 2022 one-year estimates, Table B16001.
Caveats
The U.S. Census first asked about "language spoken at home" in 1980, but city-level data from that year is not accessible through the Census API. The earliest reliable city-level data is from 1990.
Census language categories have changed significantly over time. Bengali was grouped under "Other Indic languages" until the ACS 2009-2013 detailed tables broke it out separately (85,205 speakers). African languages were reported as a single group until the 2022 ACS split them regionally.
The 2022 "Yiddish" category includes Pennsylvania Dutch and other West Germanic languages. "Amharic" represents "Amharic, Somali, or other Afro-Asiatic languages." "Thai" represents "Thai, Lao, or other Tai-Kadai languages."
From 2010 onward, language data comes from the ACS (a survey with margins of error), not the full decennial count.
Most figures come directly from Census Bureau Table B16001 or the 2000 Decennial Census Table PCT010. Some languages not separately tracked in the standard tables (Albanian, Turkish, Romanian, Ukrainian, Tibetan, Nepali) use estimates from the ACS detailed language tables or secondary analyses. These are marked with approximate figures.
"Chinese" is reported as a combined category. The 1990 data comes from an NYC Department of City Planning tabulation, which used broader groupings for some language families.
What's missing
New York City is home to an estimated 800+ languages. This visualization includes approximately 35 languages that appear in census tabulations. Many languages spoken by smaller communities, indigenous peoples and recent immigrants are not captured in the census at all.