How many languages?

« previous post | next post »

From a Globe and Mail story about the census in India (hat tip to Michael Kaan):

India concluded its national census this week, having tallied up some 1.2 billion souls, and the last night of counting focused on homeless people – of whom there are an estimated 150,000 in Delhi alone. Getting them into the count was just one in an array of staggering challenges: how to enumerate in the dozen areas under control of various armed rebel movements, and in the 572 tiny islands that make up Andaman and Nicobar; how to train 2.5 million enumerators and handle answers in 6,661 languages.

Whoa! 6,661 languages? The Ethnologue site says it has information about the 6,909 "known living languages" in the world, and lists only 438 living languages for India (for comparison: it lists 176 living languages for the United States, 86 for Canada, and 12 for the United Kingdom).

But if you look at the entries in the Ethnologue, you'll see that most languages have alternative names (sometimes a lot of them) and most languages have recognized dialects listed (sometimes a lot of them). That's probably enough to inflate the language count by more than one order of magnitude. (It's also true that "immigrant languages" — for India, the site mentions Armenian, Burushaski, Judeo-Iraqi Arabic, Northern Pashto, Uighur, Walungge, and Western Farsi — aren't included in the count, but they're probably a small contribution to the problems of the national census of India.)

So it all depends on how you count.



11 Comments

  1. Mike Aubrey said,

    March 3, 2011 @ 2:15 pm

    Indeed. A language is political entity as much as it is a linguistic one.

  2. Quendus said,

    March 3, 2011 @ 3:18 pm

    The 6661 figure was the raw number of mother tongues with differing names from the last Indian census. From http://censusindia.gov.in/Census_Data_2001/Census_Data_Online/Language/gen_note.htm :
    "… the respondent was made to feel free to return the name of his mother tongue and the same was recorded faithfully by the enumerator. This has led to the recording of a very large number of mother- tongue names from all over the country. At the 2001 census, the number of such raw returns of mother tongues has totaled 6,661. Since mother- tongues as returned in the census are basically the designations provided by the respondents of the linguistic mediums in which the respondents think they communicate, they need not be identical with the actual linguistic mediums. For assessing the correlation between the mother tongue and designations of the census and for presenting the numerous raw returns in terms of their linguistic affiliation to actual languages and dialects, 6,661 raw returns were subjected to thorough linguistic scrutiny, edit and rationalization. This resulted in 1635 rationalized mother tongues and 1957 names which were treated as ‘unclassified’ and relegated to ‘other’ mother tongue category. The 1635 rationalized mother tongues were further classified following the usual linguistic methods for rational grouping based on available linguistic information. Thus, an inventory of classified mother tongues returned by 10,000 or more speakers are grouped under appropriate languages at the all India level, wherever possible, has been prepared for final presentation of the 2001 mother tongue data. The total number of languages arrived at is 122."

    [(amz) Many thanks for the clarification.]

  3. Carl said,

    March 3, 2011 @ 6:49 pm

    Well, Hindi and Urdu, that's two languages right there, right? ;-)

  4. Rob P. said,

    March 3, 2011 @ 8:17 pm

    Oddly I just yesterday heard someone on my local public radio station discussing India and saying that it had 27 languages. Twenty seven seems as shockingly small as 6,661 seems shockingly large.

  5. Stephen R. Anderson said,

    March 3, 2011 @ 9:09 pm

    Given that there are at least 500 distinct languages spoken in Northern Italy, this doesn't seem too far out of line … :-) In shameless self-promotion, see https://lsadc.org/info/ling-faqs-howmany.cfm for some comments on the general problem of individuating languages, which is actually just about as hard as the speciation problem in biology.

  6. Craig said,

    March 4, 2011 @ 12:28 am

    @Rob P., India has 29 languages with more than a million speakers.

    While that might have been the source for what you heard, I also note that the External Services Division of All India Radio broadcasts in 27 languages (16 foreign, 11 Indian).

  7. Mo said,

    March 4, 2011 @ 1:48 am

    I wonder whether there's any way to estimate our confidence interval for the number of languages in the world… Is it justified to say that there are about 7k languages spoken in the world, or is the level of uncertainty too high?

  8. C Thornett said,

    March 4, 2011 @ 2:56 am

    At the adult education centre where I teach, the majority of ESOL students speak one or more of the sub-continental languages, some which are minority languages or dialects; the boundaries between dialects and languages seem to be quite indistinct, even considered as gradients.

    How difficult does it have to be for one speaker to understand another before 'different dialect' becomes 'different language'? Some students state that official classifications are often political, rather than linguistic, and religious and social considerations may play a part as well.

  9. Tadeusz said,

    March 4, 2011 @ 3:15 am

    @Stephen R. Anderson. Thank you, that was very interesting, and your publication contains good empirical evidence for the argument that there is no "language" but a number of distinct (though related) idiolects, though you end your text by proclaiming your belief in the one language thesis.

  10. Emilio said,

    March 4, 2011 @ 11:02 am

    I agree. Anderson's article is both clear and subtle. I'm going to reccommend it to a few non-linguist friends. To be honest, I guess many linguists would benefit quite a lot from it as well.

  11. Robert Moore said,

    March 26, 2011 @ 9:32 pm

    Follow-up to my last comment: Sorry, I see to my chagrin that the journal seems to want to charge for the article, even through the library database at my own institution (U Penn). I'll send a pdf file to anyone who contacts me via email, rem10us at yahoo.com. Profuse apologies for the bad link.

RSS feed for comments on this post