Most Common Third Languages, Updated.

A couple of years ago I posted an update on The Most Common Language In Each US State—Besides English And Spanish; now WordFinderX has a new version, with a somewhat updated national map (German seems to be spreading) and more granular maps showing the West (Hindi in Fremont), the Southwest (Tagalog in Gilbert), the Midwest (Swahili in Fargo), the Southeast (Tamil in Winston-Salem), the Northeast (Greek in Wilmington), and New York City (as impressive a spread as you might imagine, with Staten Island featuring, from north to south, Russian, Chinese, and Italian). Thanks, Taylor!


  1. J.W. Brewer says

    For the benefit of those who don’t know the local geography, the white spaces on the map of NYC represent parks, airports, and other such areas that for whatever bureaucratic reason are technically excluded from the territory of any of the city’s officially-defined “community districts.” In most cases such areas have no “official” permanent residents. However one of the excluded areas is Rikers Island, which does not have “permanent” residents but does have at any given moment on the order of five thousand involuntary “until further order of the court” residents, whose linguistic diversity you would hope the Census Bureau and/or other researchers would be interested in tracking. On the other hand Hart Island is coded pink for “Italian is third-most-common,” even though it has no _living_ permanent residents. (It is used as the city’s potter’s field and by some accounts has over a million deceased permanent residents; the two islands are connected in the sense that much of the labor connected with burials on the latter is done by work details of inmates from the former.)

  2. cuchuflete says

    The language information is interesting. The geographic designations are curious. Assigning Maryland and Delaware to the Northeast is useful only if laziness compels one to divide the Atlantic seaboard into northeast and southeast, without a middle.

  3. I strongly suspect the spread of “German” is due to the demographic growth and geographical expansion of various Pennsylvania Dutch (AKA Pennsylvania German) speakers: I am certain that if the various Germanic varieties spoken by Anabaptist groups (Pennsylvania Dutch mostly, Amish Swiss German, Amish Alsatian, Mennonite Low German and Hutterite to a lesser degree) were treated as separate languages from German proper, the latter language would have exhibited a clear decline over recent years, and only the former varieties would have become more important.

    Indeed, it is my understanding that Anabaptist Germanic varieties, plus (Hassidic) Yiddish, are some of the fastest-growing languages anywhere on the planet.

  4. That makes sense.

  5. ə de vivre says

    Who are these French speakers in Indianapolis and Kansas City? I’m guessing they must be from some former French colonies rather than Hexagonais or wayward Québécois?

    Also, does Oregon not publish this kind of language data at the municipal level? It seems odd to pick out Spokane for further analysis and leave off Portland…

  6. What’s going on with the “other English creoles” in Lubbock?

  7. I’m somewhat surprised that Vietnamese is given for Portland, Maine, where I lived until yesterday (I moved to a small town not far away). The city, and Maine overall, have a significant immigrant population from Somalia and Ethiopia and I would have expected their languages to be more prominent. I guess the Vietnamese community is longer established but they seem less noticeable, in terms of businesses and the like.

  8. J.W. Brewer says

    Note that how significant versus de minimis the “third-most-common” language is in percentage terms may vary widely by location. In some places it’s gonna be what’s spoken by 0.23% of residents as opposed to the fourth-place finisher at 0.22%. (The Census Bureau stats, if you find the complete versions, do tend to have plus-or-minus error bars, acknowledging that they are incomplete estimates, but people who like to rank things are not going to be dissuaded by a situation where #3’s lead over #4 is dubious because the high end of the estimated range for #4 is higher than the low end of the estimated range for #3.)

  9. What’s going on with the “other English creoles” in Lubbock?

    which are the non-“other” ones*, and which creoles are they counting as languages, for that matter? i was wondering about that in relation to some of the brooklyn census districts, too. the one that covers prospect heights and northern crown heights, for instance – the narrow band headed east from the top of the blank area marking prospect park – certainly has some kreyòl speakers, but its caribbean population is largely from jamaica, trinidad/tobago, and the smaller islands rather than haiti. i think if those creole lects were treated as separate from english – by the census bureau, but also by their u.s. speakers – the Third Largest Language for the district would depend entirely on the lumping/splitting choices made about them.

    * tok pisin? nigerian pidgin? krio? singlish?

  10. Giacomo Ponzetto says

    Those maps are asking a lot of 1-Year ACS PUMS data. Sure, it’s designed to be a representative sample; but once you focus on the 8% of the population that speaks languages other than English or Spanish, building county-level estimates seems rather over-optimistic.

    I can see no reason to use 1-Year rather than 5-Year PUMS for this question. It seems advisable to sacrifice recency further and look at 2009-13 data, for which the Census has published detailed tabulations based on the confidential full sample rather than the public-use microdata subsample.

    Minus the cool maps, those full-sample estimates are online for our enlightenment and entertainment:

  11. Giacomo Ponzetto says

    @Etienne, Hat:

    WordFinderX doesn’t provide much detail about their methodology, but I’d be surprised if they had the desire and even the ability to change the Census classification.

    Normally (and I presume in ACS PUMS data, though I haven’t checked) the Census defines 42 language groups.

    One of those is German, including self-reported German and Luxembourgish. Another is Yiddish, Pennsylvania Dutch or other West Germanic languages, including self-reported Yiddish and Dutch.

    I cannot say if the Pennsylvania Dutch are fooling the Census by reporting they speak “German”.

    The detailed 2009-13 tabulations should give a clue, because they distinguish not the usual 42 but a richer 380 languages. I haven’t had time to look, and I know too little about the Pennsylvania Dutch to know where to look for suspicious pockets of “German”.

  12. @Giacomo Ponzetto:

    Because Pennsylvania Dutch Anabaptists use a version of the Bible in High German it is all too easy for them to simply claim themselves to be German speakers, and indeed, I suspect that “Pennsylvania Dutch” is as close to Standard High German, if not indeed closer, than many a German dialect spoken by non-Anabaptists in the U.S. whose native speakers would not dream of calling it anything except “German”.

    Considering their lifestyle, however, there is another complication I wonder about: could a (substantial?) number of Anabaptist Germanic speakers simply avoid being counted by the census altogether?

  13. David Marjanović says

    What Étienne said – Pennsylvania Dutch is a bit Lower than Standard German, but as a western Central dialect it’s closer to it than most.

    Yay, sociolinguistics.

  14. @ə de vivre

    What do you mean by Hexagonais?

    As for French speakers in Kansas City, I don’t know of any particular French speaking population. But it could be affected by simply a lack of any single-language group (besides Spanish speakers) recently coming in any any large numbers specifically into the city limits of Kansas City, Missouri, which is I assume what is measured (as distinct from the larger area).

  15. What do you mean by Hexagonais?

    Hexagone (France).

  16. Giacomo Ponzetto says

    To the best of my knowledge, language is self-reported in the ACS. All the Census does is combining self-reported languages into a smaller number of language groups.

    In the 2009-13 estimates, there are 1,060,000 speakers of German and 130,000 of Pennsylvania Dutch.

    Almost all speakers of Pennsylvania Dutch are in ten states: the same ten states and also in close to the the same rank ordering as the states with the most Amish population according to the (seemingly serious) data cited on Wikipedia. Measuring in thousands:
    – Pennsylvania: Pennsylvania Dutch 52 – German 51 – Amish 59
    – Ohio: Pennsylvania Dutch 29 – German 54 – Amish 59
    – Indiana: Pennsylvania Dutch 16 – German 35 – Amish 44
    – Wisconsin: Pennsylvania Dutch 7 – German 37 – Amish 15
    – New York: Pennsylvania Dutch 6 – German 65 – Amish 12
    – Missouri: Pennsylvania Dutch 4 – German 23 – Amish 9
    – Kentucky: Pennsylvania Dutch 3 – German 13 – Amish 8
    – Michigan: Pennsylvania Dutch 3 – German 40 – Amish 11
    – Illinois: Pennsylvania Dutch 2 – German 46 – Amish 7
    – Iowa: Pennsylvania Dutch 2 – German 12 – Amish 7

    Conversely, the ten states with the most German speakers are simply the nine most populous states, plus Indiana (instead of North Carolina). In this case too the rank ordering is very similar.
    – California: German 111 – Pennsylvania Dutch 0 – residents 37 million
    – Florida: German 69 – Pennsylvania Dutch 0 – residents 19 million
    – Texas: German 69 – Pennsylvania Dutch 0 – residents 25 million
    – New York: German 65 – Pennsylvania Dutch 6 – residents 19 million
    – Ohio: German 54 – Pennsylvania Dutch 29 – residents 12 mission
    – Pennsylvania: German 51 – Pennsylvania Dutch 52 – residents 13 million
    – Illinois: German 46 – Pennsylvania Dutch 2 – residents 13 million
    – Michigan: German 40 – Pennsylvania Dutch 3 – residents 10 million
    – Indiana: German 35 – Pennsylvania Dutch 16 – residents 6 million
    – Georgia: German 26 – Pennsylvania Dutch 0 – residents 10 million

    I’m sure some Anabaptists must have reported their language as German, but I don’t think they can account for the bulk of reported German speakers. There are virtually no Amish in California, Florida and Texas. Surely there cannot be hundreds of thousands of Mennonites and Hutterites there either.

    More likely, a bunch of those German speakers are simply first-generation immigrants from Europe. Germany is still a top-twenty country of birth for foreign-born US residents, accounting for more than half a million.

  17. Keith Ivey says

    It’s odd that DC is not included in the cities, but it appears the answer for it is French, ahead of Amharic (my guess) by a significant number.

  18. Giacomo Ponzetto says


    What’s going on with the “other English creoles” in Lubbock?

    Almost surely sampling error. Even with the confidential full sample, reliable estimates can be constructed only for places with 100,000 or more total population and 25,000 or more speakers of languages other than English and Spanish. Lubbock, TX has more than enough population (both as a county and as a metropolitan area), but it did not have enough language variety in 2009-13. Most likely it still doesn’t.


    which are the non-“other” ones*, and which creoles are they counting as languages, for that matter?

    The usual short list of 42 includes Haitian, which is surely the same language also known as Kreyòl.

    The longer list is not fully transparent to me, and to my surprise it is not exactly a refinement of the shorter list.

    Under Spanish and Spanish Creole it includes trace amounts of Ladino and Pachuco. Is either a creole? Are there about 100 speakers of Judeo-Spanish in the U.S., or does the Census mean something else by Ladino?

    It has a heading for French Creole instead of Haitian.

    Under Portuguese (incl. Portuguese Creole) it includes Papia Mentae, which I take to be an unusual spelling of Papiamento.

    Under Other Indo-European languages it includes Jamaican Creole, Krio, Hawaiian Pidgin, Pidgin, Gullah and Saramacca. Instead, the shorter lists gives “Jamaican Creole English” as an example of Other and unspecified languages.

    Under Other Native North American languages it includes French Cree and Chinook Jargon.

    It may include other creoles, pidgin or mixed languages that I have not spotted.

  19. Giacomo Ponzetto says

    @Keith Ivey:

    It’s odd that DC is not included in the cities, but it appears the answer for it is French, ahead of Amharic (my guess) by a significant number.

    I’m too used to this to tell if it’s nerdview, but it seems logical to me that DC is included along the states (as is Puerto Rico), whereas among the cities you find the Washington-Arlington-Alexandria metropolitan area.

    In DC, French is third at 8 thousand speakers and as you guessed Amharic is fourth at 5.

    In Washington-Arlington-Alexandria, it’s a three-way statistical tie between Korean (63), French (58) and literal write-ins of “Chinese” (58). More precise respondents report Mandarin (13), Cantonese (6), Formosan (2) and trace amounts of Fuchow, Wu and Hakka. I’d hazard the guess there are enough Cantonese speakers answering “Chinese” for Mandarin to fall behind Korean and French, but we cannot really tell.

  20. at least some, and possibly a large percentage, of the texas, illinois, indiana (and arkansas) german-speakers are neither anabaptists (except perhaps incidentally) nor recent arrivals. the 1848er wave of immigrants from the german-speaking lands settled heavily in those parts of the u.s. i can’t say much about the midwestern end of things, but i have a few friends from texas/arkansas-german backgrounds, and in their families, german (of some variety, i assume not standardized) was the cradle-tongue up to at least WWI and possibly WWII.

    separately, some of the florida and california numbers are going to be retirees from german jewish families of the mid-20thC exodus, some of whom have remained quite attached to the language. (it’s also possible that a chunk of the californians are from german-speaking arkie families.)

    @GP: ladino/judezmo is certainly a fusion language (for those who believe in the category; i’m agnostic but find it useful), but not a creole (at least for typologies that think creoles come from pidgins).

    and it’s bizarre that they bother to have these “X Creole” categories and then put all these english-related ones into an “Other Indo-European” catch-all category – especially since their only indo-european connections are by way of being english-related creoles!

  21. Keith Ivey says

    I’m too used to this to tell if it’s nerdview, but it seems logical to me that DC is included along the states (as is Puerto Rico), whereas among the cities you find the Washington-Arlington-Alexandria metropolitan area.

    That makes sense to me too, but I was talking about what the WordFinderX maps do, and they don’t include DC as either a state equivalent or city, in either Northeast or Southeast.

  22. Giacomo Ponzetto says

    The first WordFinderX map, in Every State has “District of Columbia,” but you need to magnify it to full screen size and then some to see that it’s the same French color as surrounding Maryland.

    Their city maps are all bizarre because they focus on incorporated cities and towns. This is hardly ever a good idea, unless you are studying municipal government. A perfect illustration of the problem is the disappearance of the whole Washington-Arlington-Alexandria metropolitan area.

    The largest incorporated cities nearby are, in order, Washington, DC; Baltimore, MD; Virginia Beach, Cheaspeake, Norfolk and Richmond, VA. Each of the six has far fewer residents than either Fairfax County, VA, Montgomery County, MD, or Prince George’s County, MD. Norfolk and Richmond also have fewer than Arlington County, VA, completing the ring around DC. But the counties encircling DC are not legally incorporated, so they get ignored.

    Then you remove DC because you already mapped it as a state, and there goes the sixth largest metro area in the country.

  23. Thanks, I did miss that. I guess I’m so used to seeing DC left out that I see it even when it hasn’t happened.

Speak Your Mind