Grad Student Wins Ph.D. Challenge. Wait Till His Adviser Finds Out.

Gabriel Parent says he loves a challenge. Let’s hope his adviser is as passionate about surprises.

Mr. Parent was named on Wednesday night as the winner of the Ph.D. Challenge, a contest that invited grad students to insert the words “I smoke crack rocks” into the final version of a peer-reviewed academic paper. A master’s candidate in Carnegie Mellon University’s Language Technologies Institute, he put the phrase into a paper on the use of crowdsourcing to produce more-accurate transcriptions of dialogues collected with speech-recognition technology. The paper’s co-author is his adviser, Maxine Eskenazi, who Mr. Parent says knew nothing about the late edit.

“She kind of knows me,” he admits mischievously. But “she’s definitely going to be surprised.”

The people behind the Ph.D. Challenge are anonymous and have consented to be interviewed via e-mail only.

In its announcement, the group said that Mr. Parent was one of only three entrants and that his selection was unanimous: “Gabriel’s bold writing style seamlessly weaved the special phrase directly into the text in such a way that one cannot help but to think that the paper would be incomplete if it was removed.”

The paper, “Toward Better Crowdsourced Transcription: Transcription of a Year of the Let’s Go Bus Information System Data,” concerns the speech-recognition telephone system used by the municipal transit agency that serves greater Pittsburgh. Callers can give their location and destination, and the automated system will respond with schedule information and other instructions.

But anyone who has ever used such a system knows that such technology is imperfect. Mr. Parent says that callers’ pronunciation and syntax can be problematic, and that because such systems are programmed to deal with a narrow range of queries, their word-recognition vocabularies are typically very limited.

Cue the paper’s contest-winning sentence: ”For example, the caller could yell: ‘I smoke crack rocks,’ which isn’t likely to be correctly parsed by the grammar, and for which words are not in the dictionary.”

Those words did not appear in the version of the paper that passed through peer review, Mr. Parent says. The original example was “I lost my cell phone.” He substituted the contest sentence in the paper that he submitted to the Workshop on Spoken Language Technology, a conference held this week in Berkeley, Calif., by the Institute of the Electrical and Electronics Engineers.

That version was distributed to all the participants at the conference. But who actually reads beyond the abstract until perhaps months later, when a paper seems relevant to one’s own research?

For his efforts, Mr. Parent will receive a box of Maruchan ramen noodle soup (his choice of flavor), a pack of leather elbow patches, and a paper certificate. A tentative prize, an autographed photo of the Princeton professor Paul Krugman, is looking unlikely, as the economics Nobelist has not responded to repeated email entreaties from organizers of the contest.

A French-speaking native of Montreal, Mr. Parent says he anticipates plenty of email messages of his own when word circulates of his contest victory. He is not entirely sure of what else to expect.

“Some people are going to think it’s funny, some are going to think it’s stupid, and some people will probably wonder why someone would ever do something like that,” he says. Mr. Parent, who plans to finish his master’s degree this summer, says he places himself in the first group: “I think research doesn’t have to be always so boring.”

He said he might even be up for participating in next year’s Ph.D. Challenge, which the organizers say will be even more daunting than this year’s: getting a peer-reviewed paper published with a co-author named “Genital Warts.”  —Don Troop

(Photo illustration by Bob McGrath)