Peggy's response to my newsletter bit on Turnitin.com came back to me because I was in the original cc: line, and Ellen kindly passed on Grant's reply to Peggy and invited me to respond.

Naturally, I agree with both Grant and Peggy. Students are mostly honest, but cheating does happen. And I think sometimes you do have to check on papers by searching electronic databases.

My particular problem with Turnitin.com--well my two main complaints--are with its performance and its rhetoric. My article addressed the rhetoric a lot.

So let me say something more about how it performs.

Here's the extent of searching that Turnitin.com can do: it searches their archive of papers--and they copied free papers from free term paper mills to start, but now keep a copy as well of every paper submitted--as well as the open WWW. That leaves out any electronic writing behind a password--Lexus/Nexus, Infotrac, and others usually accessed via your campus library's subsription. Not to mention all those print sources. Google and other web-based search engines can search the same corpus *except,* of course, for the submitted papers Turnitin.com archives.

So here's what I did, and it's worth trying something like this on your own:
 

1. I went on the WWW and used 'the simpsons' and 'the hobbit' in Google as search terms. I took writing from the first few pages of hits and cobbled them into papers. Both short. Google ranks its search returns such that the pages that are linked to the most from other pages come first, meaning the returns on the first few pages are fairly well traveled pages.

2. I got a guest account for free at Turnitin.com--and anyone can get a free guest account and run up to five papers through them. I loaded two versions of the Simpson paper--one with sources cited parenthetically), one without. I did one version of the hobbit paper. On the Simpson paper, the only passage found--out of four sources cribbed, was one that was set off, correctly, as a long, indented citation. Since the paper was short, the Turnitin.com report rated the paper as having a significant match. But it missed the other uses of sources, which it should have caught because they were verbatim. It didn't catch them in either the cited or uncited version, missing significant plagiarizing.

3. In the Hobbit test, it caught two phrases. One was a quote from the book,and the Turnitin.com report showed a link to a site on the WWW where the book was excerpted. That's not much help. In a paper on a book the book was quoted. And there were quote marks around the passage. But more to the point, I've never read the book. I took the quote not from the book, but from a free paper mill paper on the Hobbitt, and in addition to the quote from the book, I took the two sentences leading up to the quote, changing only two words. So I created an intentional swipe, with very minor--and for plagiarism
purposes--insignificant alterations.

Click on the Image to See Full Size.
 

It did find one plagiarized passage in the Hobbit paper. Sort of.  Turnitin.com found and flagged a portion of sentence that had been plagiarized from a paper found on the WWW. However, the full plagiarized passage wasn't flagged. What was 'caught'
by Turnitin.com was: "The most important thing to notice here is. . ."

Well, that's a very common construction. And the source Turnitin.com gave for this phrase wasn't a link back to the Hobbit paper I lifted it from, where, if I followed the link, I would have seen the rest of plagiarized original at least. Rather, it was to apaper on the WWW talking about how to do some scripting on a Macintosh. That is, it matched a pretty common phrase to an irrelevant source.

So in these tests, the service failed at its fundamental promise, which is to find exact *or similar* matches to documents in its database or on the WWW.

If you use Turnitin.com, you have to be very careful. It's results--the reports it issues--can be very persuasive looking. The page that comes back, with passages underlined and a list of pages matches were found, and the sliding scale that goes from Green (little matching) to Red ("danger, Will Robinson" levels of matching) are visually and rhetorically convincing on first glance.

I think knowing those defects is important. If for example, Grant runs a paper or two through the system, that he suspects is plagiarized, and Turnitin.com doesn't turn up anything useful of definitive, that won't change the fact that the paper was likely plagiarized. He can then do phrase searches in Google or another good Internet search engine. He can try searching private databases available via the campus library. In fact, I would tell students that you know how to do that. In addition to using Turnitin.com, you know how to do phrase searches in Google and other net search engines, as well as in library databases and catalogues. If students know you know how to search, whether you use Turitin.com (as Grant will) or not (as Peggy won't), it's as more effective than only saying you have Turnitin.com.

The bottom line is that there is no perfect search, but as writing teachers you all have advantages over your colleagues in other courses. You read and are generally more attentive to your students' prose. The trick is having a plagiarism policy that's clearly explained, that plagiarism and honesty are talked about in class, that your campus has a clear and fair process for dealing with plagiarism, that you have lattitude (plagiarism caused by learning how to integrate sources, i.e. technical plagiarism differs mightily from cheating and fraud) to make distinctions, when proceeding with a plagiarism inquiry and charge that you have procedures for protecting the student's confidentiality and due process, and finally, and most important, given the likelihood that searches for original sources will come up nil quite often, that you have a way to proceed with a plagiarism charge even without a copy of the original source.

Sorry to go on so long.

Thanks for reading this far.

Best, Nick
nick_carbone@hotmail.com