Monday, May 24, 2010

The Politics of Publishing

Let's talk about class warfare in academics, shall we? I just read a nice little article on speech production from Cognition and while I enjoyed it, I couldn't help but wonder how it got published because it was rather light weight. To be fair, Cognition published it as a "Brief article" so it was meant to be short*; nonetheless, it had the feel of a grad student poster, not a publication. You might argue that this is the point of a "Brief article",  but I will argue that similar content would likely not have been published had it not been recognizably associated with a well known scholar. Despite the precautions of blind reviews, it is not uncommon for a linguistics reviewer to have a pretty good idea of who authored or co-authored a paper, simply because linguistics is a small field, and the sub-fields even smaller. Most scholars have easy-to-recognize methodologies, content areas, or style that acts almost as a scholarly fingerprint. I don't mean to be mean-spirited, I hope this doesn't come across that way, but minus the second author's fingerprint, I don't see this paper getting accepted.

But first, let's look at the paper itself: A purple giraffe is faster than a purple elephant: Inconsistent phonology affects determiner selection in English (full citation below).

From the abstract: "during the production of a determiner–noun phrase, nouns automatically activate the phonological forms of their determiners, which can compete with the phonological forms that are generated by an assimilation rule." As I understand it, this means that nouns activate default articles when we're about to say them. Show me a picture of an orange giraffe, and I think a orange giraffe before I say an orange giraffe. because I haven't yet applied the phonological process that says the indefinite article a becomes an when followed by a vowel. This leads to competition between a and an to see which one will actually be said out loud. What the researchers found was that the phrase an orange giraffe was produced more slowly than a purple giraffe and they argue that this slowness (aka, naming latency) is the result of the extra time it takes to apply the phonological process to the indefinite article (aka determiner competition).  By putting the adjective orange in between the article and the noun, they were able to show that it was the noun driving this effect, not the adjective (because the indefinite article agreed with the noun originally, thus the slowness).

Like I said, a nice little article. Cute little paradigm, good results, nice work. But here's the thing: more than a dozen previously published articles say the same thing. There's nothing new here. What this research does is drill down to test a detail of a well known phenomenon (determiner competition), namely that phonology alone can invoke this competition. In the authors' own words: "the lexical-syntactic level may not be necessarily involved."  Let me repeat: X may not be necessarily involved in Y. Wow, that's hardly a bold statement worthy of a journal publication. There was really only one experiment reported (a second experiment was included, but imho, it was so similar to the first, it hardly counts as a second). I have no problem with this as an experiment and it would be a good poster at a psycholinguistics conference or the LSA. but I can't image being able to publish something like this myself, nor anyone I went to grad school with because we simply didn't have the institutional ooomph to guide something as light as this through the review process.

Yes yes, again I know the process is "blind", but this article does have Kathryn Bock's fingerprint. She's well known for publishing on sentence production in general as well as number agreement in sentence production, a similar if not exact match to determiner agreement. And she's a well respected psycholinguist by any measure. I'd have to re-read a bunch of her papers to see how closely the methods and presentation match her work, but I'd guess that a "blind" reviewer, who would by necessity be familiar with sentence production literature, would not have a hard time guessing that this work was done in conjunction with Bock.

 The blind process actually allows such unintentional and intentional biases to fester because it's hidden and hard to prove. If the process were transparent, I suspect this sort of thing wouldn't happen as much. I'm all for open reviewing. I'd prefer all open, transparent reviewing. Why hide?

Spalek, K., Bock, K., & Schriefers, H. (2010). A purple giraffe is faster than a purple elephant: Inconsistent phonology affects determiner selection in English Cognition, 114 (1), 123-128 DOI: 10.1016/j.cognition.2009.09.011

*The only guideline regarding the content of a "Brief article" I could find on Cognition's site is this: "Brief articles must be no more than three thousand words long."

No comments:

Putting the Linguistics into Kaggle Competitions

In the spirit of Dr. Emily Bender’s NAACL blog post Putting the Linguistics in Computational Linguistics , I want to apply some of her thou...