As someone who is interested in both data analysis and the study of foreign language, I was intrigued by the Washington Post's recent Wonkblog article entitled “Americans are beginning to lose their love for foreign languages.”
Using data from the recent Modern language Association (MLA) report, the story highlights the “drastic fall-out” in university students' enrollment in foreign language studies in recent years. To illustrate this premise, the article included a graphic, which highlights the three most recent data points and points out the decline:
However, using this chart to support the article’s thesis raises several issues – which can generally be categorized into “visualization” problems and “data” problems.
First, the visual problems:
The y-axis is absent, and the data points are unlabeled. How are we supposed to know how many of every 100 students are enrolled in foreign language classes?
The x-axis is deceptive. The bars are evenly spaced, but the years they represent are not. The gaps between bars from left to right represent, respectively, five, three, four, four, three, and four years. This is a grievous error for charts that purport to show how data changes over time.
The color change arbitrarily marks the last three bars, which in series decline over time. But the caption reads that enrollment has been falling “since the recession.” The National Bureau of Economic Research (NBER) dates the beginning of the recession as December 2007 and the end as June 2009 in the United States. Why does the color change at 2006, which was prior to the recession?
1991 – 1994 was also a period of recession in the United States, but neither 1990 nor 1995 are colored in dark blue to reflect this.
On top of these issues we also have data problems: based on what’s shown in the ENTIRE graphic, there doesn’t seem to be a drastic “fall-out;” there is a decline in the early 90s, an increase in the late 90s through 2006, and then a symmetrical decline afterwards.
This didn’t seem to me to be “drastic fallout.” However, MLA has years of data—helpfully, referenced in the Wonkblog article--so I decided to analyze it further. After taking the raw data from the report, just for the years shown in the Washington Post blog, I created a similar graph and fit a trendline.
The first thing I noticed was that the trend was positive! The author had highlighted the last three data points (which I left highlighted here), but the entire graphic argued against the title of the blog post. In fact, a linear fit had a slope of just under 0.02; admittedly, not a massive increase, but an increase nonetheless.
But, I couldn’t leave well enough alone. I had access to all the MLA data, so I went ahead and graphed it:
Now THERE is a drastic fallout! Since 1960, there is a dramatic decrease in the number of foreign language enrollments (linear trend of -0.13) per 100 students. However, the trend since 1972 has been roughly level (linear trend of -0.006). Perhaps the title of this blog posting should have been “Americans Lost Their Love of Foreign Languages in the Seventies”--more honest, but less likely to go viral.
On second thought: would that title actually be more honest? There are several factors that could have led to the decline shown above. It is important to note that the data represented is a per-capita number. Overall college enrollment is more than three times higher now than it was in 1960, and as such, the actual count of foreign language enrollments have also increased, per the same report:
What we can't tell from these numbers is which languages, specifically, are being studied, and if there's a distinct subset of languages that is gaining or shrinking more dramatically. Fortunately, we can get some idea of these answers from a different source. The Modern Language Association of America collects that data in intermittent surveys, and what we learn from these responses is that some language study is actually on the rise.
In absolute terms, as shown in the chart below, Spanish enrollment is more than four times higher in 2013 than it was in 1960...while the combined enrollment in French, German, Russian, Latin, and Ancient Greek has actually declined, in actual terms, by 20%. The most dramatic growth is in traditionally less-popular Romance languages (Italian, Portuguese), Far Eastern languages (Chinese, Japanese, Korean) Middle Eastern languages (Hebrew, Arabic) and in American Sign Language. Taken as a group, enrollment in these languages is 16 times greater in 2013 than in 1960.
This shift away from French/German/Russian/classical languages is even more noticeable when we look at their share of all foreign language enrollments at American colleges. These languages accounted for two-thirds of all foreign language enrollments in 1960; by 2013, their share was barely more than one-fifth. Spanish, on the other hand, had grown to account for more than half of all enrollments alone.
It is often the case that one data set is not enough, on its own, to support any real analytic result. In this case, the original article used a single subset of data from one source, and posited a sweeping cultural conclusion. However, even a small amount of follow-up analysis showed that the conclusions were overstated at best, and incorrect at worst.
In order to get to the real truth behind these numbers, more and deeper analysis would be required; we still don't know, for instance, whether a change in secondary school education policies affected collegiate enrollments; or if the rise of outside sources of language education (i.e., self-learning and independent courses) made a difference; or if universities simply offer a broader range of course offerings now than they did 50 years ago.
What we can say, with the support of data, is that more people are studying Spanish in college now than in the past, and by a significant amount; more people are studying Asian languages than ever before; and in general, more students are studying at least one language in college today--in fact, almost three times as many students--as they were in 1960. None of these facts support the conclusion that Americans are "losing their love for foreign languages."