August 25, 2005

Journal-mediated scholarly debate: slow and ceremonial conversation

Whether or not you're interested in the content of the on-going debate in Cognition about approaches to language evolution, you might find it interesting to contemplate its schedule. Here's a summary of the time line:

Hauser, Chomsky and Fitch (HCF): published in Science November 22, 2002.
Pinker and Jackendoff (PJ): Received 16 January 2004;  accepted 31 August 2004.  Available online 19 January 2005. Published March 2005.
Fitch, Hauser and Chomsky (FHC): Received 5 November 2004;  accepted 15 February 2005.  Available online 19 August 2005. Not published yet.
Jackendoff and Pinker (JP): posted on J's website March 23 2005. No information yet available from Cognition.

So Pinker and Jackendoff took a year or so to decide to respond to HCF and to send in their critique PJ, which arrived at Cognition roughly 14 months after HCF was published. It took 7.5 months for it to be accepted, and then another 4.5 months for it to be put on line, and another 2 months to appear in paper form. The response FHC to PJ was sent in 2 months after PJ's acceptance, 2.5 months before it appeared on line, and 4.5 months before it appeared in print; FHC was accepted 3 months after submission, published on line after an additional 6 months, and not published yet. Meanwhile, the response JP to FHC was completed and put on line (and thus I assume was sent to Cognition) about 1 month after FHC was accepted, and five months ago; apparently Cognition has accepted it, but it has not yet appeared on the Cognition web site.

(I'm not singling Cognition out for criticism — this is a typical sort of schedule for such sequences.)

This conversational tempo is reminiscent of 18th century correspondence between Europe and North America, or Europe and India, when a message could take as much as six months to reach its destination. If we start the conversational clock at the point where PJ was accepted by Cognition, and entirely ignore the timing and distribution of actual print media, we get the following sums for the conversational sequence PJ/FHC/JP:

Thinking and writing: 3 months
Review: 10.5 months + unknown time for JP, still not known to be accepted -- say 13.5 months?
Waiting for publication on line after acceptance: 10.5 months + unknown time for JP -- say 11.5 months?

The sums are uncertain because the time periods involved are not entirely disjoint (so that only 19 months in total have elapsed since PJ was received by Cognition), but it still seems likely that the mechanics of the system have slowed this conversation down at least as much as sending the manuscripts by square-rigger across the oceans would have done. Shouldn't there be a way to carry out scientific discussion that's a bit brisker? Certainly one good candidate for elimination is the 11.5 months or so clocked in this case by waiting for accepted articles to appear on line.

Somewhat more tentatively, I'd like to raise the question of whether the review process is always worth its cost in time. As is normal and probably inevitable in the refereed literature, this particular back-and-forth includes a number of factually doubtful statements and presuppositions. I'll cite just one example: in PJ we read that

HCF do discuss the ability to learn linearly ordered recursive phrase structure. In a clever experiment, Fitch and Hauser (2004) showed that unlike humans, tamarins cannot learn the simple recursive language AnBn (all sequences consisting of n instances of the symbol A followed by n instances of the symbol B; such a language can be generated by the recursive rule S→A(S)B).

and in FHC, we read the response that

The inability of cotton-top tamarins to master a phrase-structure grammar (Fitch & Hauser, 2004) is of interest in this discussion primarily as a demonstration of an empirical technique for asking linguistically relevant questions of a nonlinguistic animal.

The reader will naturally conclude from this that Fitch and Hauser (2004) actually did establish something about the abilities of tamarins and humans to learn the language AnBn, whereas the sad fact is that this conclusion was a serious over-interpretation of a rather limited experiment, and seems to be incompatible with later research.

In addition to such (inevitable) mistakes, the programmatic nature of this exchange results in an unusually large fraction of statements of opinion, where the role and value of the review process is especially unclear. I'll also point out that the review process, though regarded as a sacred ritual by our academic culture, is a relatively recent development. I recall reading that when Albert Einstein moved to the U.S. in the 1930s, and first submitted an article to an American journal, he was shocked and offended to learn that that it was being sent out for review. This was not because he thought himself in particular above such things, but rather because he had never encountered the practice before, so that his first reaction was that he was being singled out as an untrustworthy source. (Memory says, perhaps falsely, that I read this in Abraham Pais' wonderful biography of Einstein, Subtle is the Lord, which I don't have at hand.)

I'm a conservative sort of person, though not nearly as conservative as most academics are about their culture, so I'm not about to propose that we scrap the existing journal system. As Churchill is said to have said about democracy, it's the worst possible system, except for all the others.

But all the same, among the emerging technologies of networked text archives, links, indices and so on, there are a wide range of other possible solutions to the problems of scientific and scholarly communications that refereed journals have evolved to solve. And as a result, I'll predict that within 50 years, scientists and scholars will use a very different set of methods for communicating and discussing their research results, and the existing system of scientific and scholarly journals will survive only in a vestigial form, analogous to the caps and gowns that academics once wore all the time, and now put on only for ceremonial occasions.

[Update: Jay Cummings writes

In the upcoming Physics Today (September 2005), the story of Einstein's objection to being reviewed is told. The reviewer (with some trepidation, because after all, he knew this was Einstein) suggested a correction. Einstein refused the correction, but he turned out to be wrong.

I haven't seen the article -- I'll look forward to learning the details.

Posted by Mark Liberman at August 25, 2005 11:20 AM