As of this morning, 59 members of the U.S. House of Representatives have signed on in support of H.R. 997, a bill to declare English the national language of the United States and (with some exceptions for national security, criminal justice and the like) to mandate that only English be used in federal and state government-related activities. Let’s set aside the question of whether this is a good or bad bill that is right or wrong. Instead, let’s consider why these 59 members of the House have cosponsored this legislation — and why 376 members of the House have not.
The question of why some members of Congress support a bill and why others don’t can’t be legitimately answered with the response, “because it is the right thing to do.” That’s a constant, and you can’t explain a variable with a constant; if the constant rightness of the bill were really the driving force behind support for a bill, then every member of Congress would be supporting the bill. Conversely, if the bill’s wrongheadedness were the driving force, then no member of Congress would be supporting the bill. Some members of Congress consider this bill to be the right kind of bill to support, and some members of Congress consider this bill to be the wrong kind of bill.
What are the factors that influence a politician’s consideration? The most typical answer to that question has to do with representation: since members of the House of Representatives owe their jobs to voters, representing the will of voters in their districts, it ought to be in their interest to do what voters want… right? Well, there are some other legitimate possibilities — perhaps politicians are responding to the other people they’re indebted to, campaign contributors, perhaps politicians respond to the people they rub shoulders with on a social basis, and perhaps no matter what the pressures upon them politicians have robust personality traits (even principles, maybe) that will consistently drive their behavior. But theories of representation have formed the backbone not just of contemporary political science, but of the pluralist depiction of politics we receive in our high school civics curricula:
Legislators are inclined to ask, explicitly or implicitly, with regard to just about every contested issue, “how would its enactment affect my constituency” and “how would my support of its enactment affect me with my constituency?”
So how well do theories of constituent representation predict variation in support for this English-only policy? The answer depends on what you what you think representation means and how you think it works.
Below let’s consider a number of different ways of thinking about political representation. To assess their rough fit to the facts of this case, I’ve gone to data regarding the demographics of congressional districts collected by the U.S. Census Bureau, which is the source of all data I’ll discuss here.
Perhaps a member of Congress is more likely to respond to an issue when it has a more conspicuous relevance in his or her district. If there are hardly any people who speak a language other than English in a district, then the issue of language shouldn’t be a matter of local contention. The more non-English speakers there are in a district, the more salient the issue of spoken language should be. A simple issue salience hypothesis would therefore predict that the more non-English speakers there are in a district, the more likely it would be for a district’s representative to cosponsor H.R. 997.
This hypothesis is not borne out by available data. The main non-English language spoken in the United States is Spanish, and in the average U.S. congressional district (including delegations) 11.1% of the population speaks Spanish. (I’m referring to a mean for a district; all trends I discuss here also appear when comparing median values.) But the average percent speaking Spanish in the districts whose representatives support English as a national language is lower, not higher, than the overall average: just 6.5% of the population speaks Spanish on average in these districts.
If simple issue salience in a district doesn’t explain the actual pattern of support for H.R. 997, what might? The contact hypothesis predicts that under certain circumstances, if people come into contact members of another group more often, they are less likely to hold feelings of animosity toward that group. If familiarity breeds amity, then that might explain why the districts of representatives supporting H.R. 997 have fewer, not more, Spanish speakers in them.
The circumstances under which contact occurs are vitally important according to the contact hypothesis: if contact occurs under conditions of marked inequality, then intergroup animosities can actually become more pronounced. Research indicates (see references especially to Sigelman and Welch 1993, Jackman and Crane 1986 and Pettigrew and Tropp 2006) that intergroup animosities lessen most when members of different groups meet under conditions of equality (or in some cases higher status for minority groups). This would suggest an additional effect of economic inequality between Spanish and English speakers in a member’s district on the likelihood of support for H.R. 997. Unfortunately, data directly connecting language spoken with economic condition aren’t easily available from the census.
When thinking about the negative correlation between percent Spanish speakers in a district and support for H.R. 997, it’s important to recognize that the salience of an issue to a district is not enough to drive members of a district toward support for a particular policy related to that issue. Indeed, while language may be a salient political issue in a district where there are many Spanish speakers, if there are enough speakers of Spanish then support for (not removal of) Spanish-language infrastructure may become a winning issue. This suggests a curvilinear relationship between percent Spanish in a district and support for English as a national language. In districts where the percent speaking Spanish is very low, the issue is not salient and so a representative might not have a reason to cosponsor a bill like H.R. 997. In a district where the percent speaking Spanish is very high, a representative might have good reason to oppose H.R. 997. Only in a district where there are enough speakers of Spanish to make the issue salient, but not enough speakers of Spanish to make supporting H.R. 997 risky, would we expect to see representatives adding their cosponsorship to a bill like H.R. 997.
That is precisely the pattern we see. The Spanish-speaking composition of the districts of cosponsors of H.R. 997 cluster an average of 3.7 percentage points away from the mean for all districts; the Spanish-speaking composition of districts of those who don’t cosponsor H.R. 997 lies on average 7.5 percentage points away from the mean, both below and above it. Cosponsorship of H.R. 997 more commonly occurs by representatives of districts who have a fair but not very large number of Spanish-speaking residents.
Descriptive Representation associated with Substantive Representation
A rather different version of the idea of representation is descriptive representation. A term coined by Hanna Pitkin some four decades ago, descriptive representation happens when members of Congress are also members of a racial, ethnic, religious, gender or some other categorical group. Women in Congress can be said to descriptively represent women, white people in Congress to descriptively represent white people, English speakers in Congress to descriptively represent English Speakers.
It is an unsettled empirical question as to whether descriptive representation leads to substantive representation: that is, action in accordance with the prevailing sentiment of members of a social category. In this case, it is interesting to note that not one of the 59 supporters of H.R. 997 is a member of the Congressional Hispanic Caucus, a group within the Congress organized around members’ personal Hispanic status.