Are prejudices magnified depending on the language being spoken? An experiment based on a standard Implicit Association Test suggests yes.
In an Implicit Association Test pairs of words appear in sequence on a screen. Subjects are asked to classify the relationship between the words and then the time taken to determine the association is recorded. In this experiment the word pairs consisted of one name, either Jewish or Arab, and one adjective, either complimentary or negative. The task was to identify these categories, i.e. (Jewish, good); (Jewish, bad); (Arab, good); (Arab, bad).
The subjects were Israeli Arabs who were fluent in both Hebrew and Arabic.
For this study, the bilingual Arab Israelis took the implicit association test in both languages Hebrew and Arabic to see if the language they were using affected their biases about the names. The Arab Israeli volunteers found it easier to associate Arab names with “good” trait words and Jewish names with “bad” trait words than Arab names with “bad” trait words and Jewish names with “good” trait words. But this effect was much stronger when the test was given in Arabic; in the Hebrew session, they showed less of a positive bias toward Arab names over Jewish names. “The language we speak can change the way we think about other people,” says Ward. The results are published in Psychological Science, a journal of the Association for Psychological Science.
Nice. But this leaves open the possibility that, since Hebrew is the second language, all response times in the Hebrew treatment were increased simply making it harder to see the bias. I would still prefer a design like this one.
Balaclava bluster: Johnson.
Cheap Talk 
4 comments
Comments feed for this article
July 13, 2010 at 10:18 pm
twicker
So, from the short report (available here: http://pss.sagepub.com/content/21/6/799 ), here are the full results (I’ve done a wee bit of formatting):
“The predicted interaction between task and language was significant,
F(1, 40) = 5.4, p = .025.
For the Arab-Jew IAT, D scores were smaller in Hebrew (0.39) than in Arabic (0.70), t(40) = 3.04, p < .005,
whereas for the weapon-instrument IAT, they were of similar size in Hebrew (1.10) and in Arabic (1.08), t(40) < 1.
These effects were not qualified by higher-order interactions involving task order or language order.
Thus, we observed overall positive associations to both Arab names and musical instruments. The positive bias toward Arab over Jewish names was reduced in the Hebrew context relative to the Arabic context. However, the positive bias toward instruments over weapons was not affected by language context."
Looks pretty convincing to me. FWIW, they derived the D-scores using the "second method" from the 2003 Greenwald, Nosek, & Banaji improved IAT scoring algorithm, which theoretically takes care of the issues you discussed.
Greenwald, Nosek & Banaji (2003). Understanding and using the Implicit Association Test: An improved scoring algorithm. JPSP, 85(2), 197-216.
EBSCOHost Permalink here:
http://search.ebscohost.com/login.aspx?direct=true&db=pdh&AN=psp-85-2-197&site=ehost-live&scope=site
All this also reminded me of Lera Boroditsky's work out at Stanford. Sample from Edge:
http://www.edge.org/3rd_culture/boroditsky09/boroditsky09_index.html
All *that* said, your study design *is* still cool. 🙂
July 14, 2010 at 11:48 am
jeff
thank you for the details. i couldn’t access the gated paper from home where i was posting.
July 15, 2010 at 8:41 pm
michaeleriksson
I stumbled across the same article by Lera Boroditsky last year—and I found it sufficiently naive that I even wrote an email to her with some issues. (A nutshell abbreviation: Little support for the title is given, and the overall discussion is simplistic.) I fully grant the possibility that she was lying to laymen; however, it is not an article that I would recommend.
July 14, 2010 at 1:18 am
michaeleriksson
As you implicitly point out at the end, it is very important not to jump to conclusions, but to investigate all possible explanations. While there are some indications that language influences thought, this idea is occasionally abused for political purposes (most notably by feminists and the politically correct), which makes it vital to be vigilant against over-interpretation and over-estimation of the influence. (In particular, a common error among the ideologically driven laymen is to assume that this would be a one-way street, where the word used affects thought, but the fact that thoughts and culture affect the words used, or at all present, is ignored.)
Above e.g., an alternate explanation is that the subjects were used to hearing anti-Jewish sentiments in Arab, but not in Hebrew, which resulted in slightly different associations being triggered.) A few others can likely be found on closer thought.