Ask a science student at a top American university which are the best cities in the United States for pursuing advanced study in science or engineering, and chances are that Cambridge, Massachusetts, and Palo Alto, California, will come to mind. But pose that same question to a comparable student in Ankara, Turkey, and the answer will be different, says Zeynep Esra Tanyildiz, a Turkish scholar who is examining the relationship of aspiring foreign scientists to American academic science. Close to the top of the list will be Columbus, Ohio, but pride of place will almost certainly go to Atlanta, Georgia.

It’s not that life on the Anatolian steppe necessarily breeds an interest in Scarlett O’Hara or CNN, Tanyildiz explains. Rather, there’s a “tunnel” running between the Turkish capital and the Peach Tree City that carries a steady stream of scientific talent direct from Middle Eastern Technical University (METU), Turkey’s premier science institution, to the labs of the Georgia Institute of Technology. The Ramblin’ Wrecks host the largest concentration of Turkish grad students in the country, she says, followed by the Buckeyes of Ohio State University, and Atlanta has a very active Turkish student community. Similar express connections appear to direct grad students and postdocs of other nationalities to labs on particular American campuses, she adds. These passageways, although largely invisible to Americans, appear to play a crucial role in determining the distribution of scientific talent in this country.

The network effect

Tanyildiz revealed the existence of the “tunnel” while talking with Science Careers about research she is conducting for her doctorate in public policy at the combined program of Georgia Tech and Georgia State University, both in Atlanta. Numerous scholars have looked at the growth in numbers and importance of foreign graduate students and postdocs in American science in recent decades. Tanyildiz, however, is one of the first to do so from the point of view not of American institutions and interests, but of the young scientists themselves. She is investigating the processes and motivations that cause people to end up where they do.

The ethnic distribution of those staffing this nation’s university labs is far from random, she declared in a paper she presented in March at the annual meeting of the Midwestern Political Science Association in Chicago, Illinois. She bases that conclusion on data she gathered from 164 randomly selected labs, representing 12 science and engineering disciplines and 30 U.S. universities, a third each from the top, middle, and bottom ranks of American graduate schools. She compared the placements of American-born grad students and professors in science and engineering with those of comparable students and faculty members born in China, India, Korea, and Turkey, four of the top ten suppliers of science talent to the United States.

“Within the same university and department,” Tanyildiz wrote in the paper, “the percentage of students from a specific country of origin [is] higher in labs with a faculty member from the same country of origin, compared with labs that are directed by native [i.e., American-born] directors.” Overall, two thirds of the foreign grad students in her sample work in labs with foreign-born directors, compared to only one-third of the American-born students.

One third of foreign grad students in the sample work in the labs of American-born directors, but the majority of the foreign grad students work in labs with directors of their own nationality. This ethnic affinity is particularly strong for students whose native tongues differ greatly from English, and less strong for the students from India, where English is an official national language and widely used in education at all levels. The tendency to work with a director of the student's own nationality was strongest in the lowest-ranked schools and weakest in the highest-ranked ones.

To get some insight into how this sorting by nationality actually happens, Tanyildiz is conducting structured interviews with Turkish graduate students about the decision-making process that led them to a particular professor’s lab. She has found that the “tunnel”--a network of personal relationships running between Ankara and Atlanta--accounts for the presence of the many graduates of METU (her own alma mater) studying at Georgia Tech. Professors at METU who have contacts at--and often degrees from--Georgia Tech smooth the students’ path to Atlanta. Georgia Tech professors with contacts at--and, often, degrees from--METU attract promising candidates from Ankara. She cites, for example, an engineering professor who “has a best friend in Middle East Technical University, [and] is always contacting him and just requesting those students,” she says. Students still in Ankara also contact students in Atlanta for information and advice.

Tanyildiz has discerned evidence of similar “tunnels” to America from the three other countries. “In this data set,” she writes, “80% of the Korean [lab] directors graduated from Seoul National University, 57% of the Indian directors from the Indian Institute of Technology, and 21% of the Chinese directors from University of Science and Technology of China.”

Talking Turkey

From the standpoint of foreign students--especially science students--this arrangement has enormous advantages, not the least of them linguistic. Working with a professor and lab mates from back home make for an easier, more congenial experience than plunging into a strictly American cultural environment. Many students of science and engineering--even graduates of METU, where the language of instruction is English--bring less-than-outstanding English skills, Tanyildiz says. She originally intended to do her interviews with graduate students in English, but she found that many subjects had difficulty. So she changed to conducting the interviews in Turkish. Mastery of English matters less in science than it does in some other fields, Tanyildiz notes. Many graduate schools, for example, require Graduate Record Examination verbal scores for humanities and social science programs but not for science and engineering.

Regardless of one’s field, however, imperfect English is a major impediment to making close friendships with Americans, which increases the importance of having a local network that speaks one’s native tongue. These contacts matter “economically and socially, [for] everything from studying together, sharing notes written in Turkish,” and collaborating on experiments, to finding familiar food and people to hang out with, she says. This greatly increases the attractiveness of a campus with an established community of compatriots.

Methodological limitations prevented Tanyildiz from extending her formal analysis to postdocs. Anecdotal evidence, however, suggests that the factors influencing grad students work even more strongly on postdocs, she says. “They need the networks even more than the students do” because the often-amorphous nature of many appointments provides much less structure and support than a formal graduate program. Even so, many of Turkey’s newly minted Ph.D.s eagerly seek postdoc positions in the United States because a year of research abroad is a prerequisite for tenure at some of Turkey's leading universities. So aspiring academics in Ankara generally take the well-trod path of “applying to the United States and finding Turkish professors. ... They’re not just going to go [and] randomly shoot CVs to all the places. The professors, who have been to the United States earlier, the senior professors … say to them, ‘Go and work with Dr. [so-and-so] in America and then come back.’ ”

And there’s more to the “tunnel” than student preference, she believes. Because labs are “semiautonomous groups within the university,” Tanyildiz writes in her paper, lab chiefs generally exercise considerable independence in staffing them. Her research does not examine directors’ actions or motivations, nor does it explain how they choose the particular students who work in their labs. She suspects that some professors give preference to students of their own nationality if other qualifications are equal, but she does not have evidence on this question.

The patterns Tanyildiz has identified mirror in many ways this nation’s long immigration history. A number of differences, of course, distinguish today’s science graduate students from the “huddled masses” of largely ill-educated immigrants of yesteryear. Unchanged, however, is every newcomer’s need for information, connections, and assistance in learning the ropes in a new land. Whether people intend to remain in the United States, return to their own countries, or go elsewhere, ethnic networks and active local communities of compatriots play an important role in helping them meet those needs while they’re here. Forces that have long encouraged new arrivals to cluster in certain locations and occupations appear to be, as Tanyildiz persuasively argues, still at work in the labs of America’s research universities.

This article is second in a three-part series on foreign scientists in American university labs.

Comments, suggestions? Please send your feedback to our editor.

Text corrected: 4 May 2007.

Beryl Lieff Benderly writes from Washington, D.C.

10.1126/science.caredit.a0700063