2004-08-19

If sounding sexy is the name of game, choose your vowels carefully

An MIT linguist, Amy Perfors, found it out by posting photos with fake names on the Web site, "Hot or Not", which allows the face police to rate strangers' looks. She found that men's photos tagged with "front vowel" names (say, Matt) were rated as more attractive than the same photos labeled with "back vowel" names (Paul). The opposite was true for women. (Rose: not sexy. I think.)


I don't know what those zany MIT linguists will come up with next. My feelings are mixed on this study, as my first name vowel is front, but not tensed. I think this would make me sexy. On the other hand, I've never been accused of being such.

http://www.clarionledger.com/apps/pbcs.dll/article?AID=/20040819/COL0204/408190349/1023/FEAT05

1 Comments:

Blogger Experimental linguistics said...

Ok, I read the 1-page paper. First of all, the author is not a linguist. She is in the Cognitive Science program (although she does have an M.A. in linguistics, but not from MIT). The paper was presented in a Cog Sci meeting. Close, but not quite there. The article starts with two very controversial ideas in linguistics: Sapir-Whorf hypothesis and sound symbolism. More than, it intends to challenge the the notion of arbitrariness of the symbol.

The study itself. A rating experiment conducted on-line? Not exactly controlled. From what I saw in the website used, there is not much control in the pictures either. More details would be welcome, but the it is a too short article. Despite all that, there is a statistically significant difference in the attractivenes ratings according to vowel frontness in the monossylabic (I assume) names. But it is in opposite direction for women! Also, the mean ratings presented are very close numerically, what suggests that maybe the statistical test (t-test) could be improved. The author acknowledges that in her website:

"1 - Because I got my data off of hotornot.com, I was able to gather thousands of data points, which is generally a good thing (the more you have, the more reliable the finding). But it also means that smaller differences are more likely to be found "statistically significant" than if the dataset had fewer points"

Personally, I think the overall goal is misguided. Even if the notion of arbitrariness is not the ultimate truth for language in all cases, it seems to be close enough to it, i.e., a really good approximation.

Now, what needs to be really praised is that she got the article to be cited in Nature (http://www.nature.com/news/2004/040809/full/040809-4.html), New Scientist, Guardian, CNN, and among other places. THAT is really impressive for such a small scale study

LO

3:19 AM  

Post a Comment

<< Home