June 17, 2004

Toys in Babel-land


Lance Knobel points me to one of those productivity-destroying internet tools which can quickly eat up hours of your day: the Modern Language Association's interactive Language Map. You can see the density of speakers of thirty-seven languages and language groups from the national level right down to individual zip codes, or even by age groups; you can generate charts of the distribution of languages in each state; you can compare and contrast different states or towns... the list is almost endless. Above is the density of Spanish speakers in New York: pretty, no?

Posted by Felix at 12:40 PM GMT

Interestingly, according to this site, the average zip code has about 6,000 individuals. Which makes the NYC zips with more than 50,000 Spanish speakers even more impressive.

Posted by: Felix on June 17, 2004 07:24 PM

Those who spent a week at a postal economics conference recently feel that we have the right to chime in here. It isn't often that a fleeting knowledge of postal economics allows for that, so I'm going to grab it.

Anyhoo, loads of post codes have zero residents (e.g. the Pentagon's postcode), which will drag down the average somewhat. Nonetheless, because individual postcode area coverage is linked to delivery networks which can deliver more per postie in dense urban areas, postcodes in cities will cover larger populations (even while they are geographically much smaller, obv.).

This, as it happens, links to a theoretical problem of introducing competition in the postal sector which took up most of the week I spent at the postal conference. However, it turns out the problem is porbably purely theoretical and has almost no relevance to the real world, so I left thinking that I could have spent my week more wisely.

Posted by: Charles on June 17, 2004 10:09 PM

Very interesting site but two shortcomings limit its usefulness. First, the color scheme. Yes it's pretty, but it is also hard at first glance to make out patterns, because the order of color to population size isn't intuitive. It would be easier to have it follow a rainbow scheme, and also to go from flatter tones to bright ones.

Second, the data is all there but it's all in straight population numbers, which is information of a kind, but not a kind that really says a lot. What we need are percentages of population. If you look at any language's use across the US, yes you'll see, say, more people speaking French in Louisiana and New England. But the fact that 1-99 people speak French in certain Western counties could be even more interesting, if we knew that this represented, say, 15% of a thinly populated area. Since the data already exists, not putting it in percentage terms is a major oversight.

It's also not clear how the census handles double counting. If someone says their language is Polish, does that mean they also don't speak English, or is to be categorized as "English speaking" mean you cannot speak another language? I realize this is too complicated a question for the purpose of a heat map but it does make me wonder exactly what these definitions mean.

Posted by: Jame on June 18, 2004 10:33 AM

Dead men don't talk. That must be why there are no Spanish speakers in the East River.

Posted by: Stefan Geens on June 18, 2004 02:14 PM

Jame, the FAQ tells us that "Of the 47.0 million respondents (18% of the entire population over five years old) who reported that they spoke other languages, 55% said that they also spoke English "very well," 22% that they also spoke English "well," 16% that they also spoke English but "not well," and 7% that they did not speak English at all."

If you go to the data center, you can get a lot more detail, some of it fascinating. For instance, in my zip code (10002), there are 19,142 Spanish speakers, 20,815 English speakers, and a stunning 36,235 Chinese speakers. (If the Spanish speakers had been just 4.5% more numerous, we'd have moved up from magenta to lime on the map.) I have to say I'm quite proud to live in a zip code where English is a minority language! In 11211, there are two languages more widely spoken than English (20,131 speakers): Spanish, with 26,571 speakers, and Yiddish, with 20,446.

Posted by: Felix on June 18, 2004 02:16 PM