Korean Sociological Image #67 — Comparing Image Searches of Different Nationalities and Races


As The Sociological Cinema put it, these search results “speak volumes about objectification, sexualization, and Orientalism.” So, out of curiosity, I decided to compare them with those for the same search terms written in Korean, and then on Naver also, Korea’s biggest internet portal.

Unfortunately, an exact comparison was impossible, as Korean has separate words for European (etc.) people and European things, so I chose the former. Also, note that Naver doesn’t allow any pornographic images, although rest assured that many image searches will readily provide anything just short of that — type in any young female singer or actress’s name for instance, and most likely nohchool (노출; lit. “exposure”) will be the first additional search term suggested. But nothing like that came up in the results for these searches:

유럽사람 / European Person — Google Images:

유럽사람 / European Person — Naver Images:

미국사람 / American Person — Google Images:

미국사람 / American Person — Naver Images:

아프리카인 / African Person — Google Images:

아프리카인 / African Person — Naver Images:

아시아인 / Asian Person — Google Images:

아시아인 / Asian Person — Naver Images:

Very interesting, but this post would be incomplete without including the results for “White Person”, “Black Person”, and “Foreigner” too. And indeed the results on Google Images for the first term, directly below, did include some NSFW ones, while the results on Naver for the second were probably the least flattering of all the searches (see here and here for more information about the Blackface and bus incidents depicted in them):

백인 / White Person — Google Images (NSFW):

백인 / White Person — Naver Images:

흑인 / Black Person — Google Images:

흑인 / Black Person — Naver Images:

외국인 / Foreigner — Google Images:

외국인 / Foreigner — Naver Images:

Do you think there’s any significance in these image results? Or do they only give at best a very superficial, maybe even quite misleading glimpse at Koreans’ associations with various races and nationalities, as evidenced by several images coming up in searches that they’re obviously not appropriate for? Please let me know!

(For more posts in the “Korean Sociological Image” series, see here)

20 thoughts on “Korean Sociological Image #67 — Comparing Image Searches of Different Nationalities and Races

  1. Very interesting experiment… It is weird the stuff that pops up in Google Images. I remember once trying to make flashcards at my hagwon in Daegu, and I kept coming up with porn, even though I was typing innocent phrases with the safety filter ON. I’d never really thought about it at home. But then my boss walked past and laughed, “Are those your friends?” he asked.

    Very interesting about the blackface pics, too. I wonder if Naver’s search algorithm takes into account recent events a little more than Google’s. Or has this incident been that significant?

  2. A naver search for ‘mbc 흑인 마이콜’ turns up almost 50 news articles related to this incident (mostly internet news services and sports versions of the big paper, but also papers like Chosun Ilbo, Maeil Gyeongje and Hanguk Ilbo). So it seems to have gotten a reasonable amount of attention. If I had to guess, I’d say that yes, Naver’s search algorithm does take into account recent events more than Google’s.

    • It’s interesting that it does that, considering — for anyone that doesn’t already know — Naver already has a “most recent images” option (which I wish Google had). By default, image searches are at the “most accurate” setting, and when you use that, like I did here, popular search terms often have a separate section for “주제별” image results at the top too. I’m not exactly sure what that means — literally it’s “theme/subject star [images]“, but it usually seems to bring up popular and/or very recent ones.

  3. Pingback: WTF du jour – Comparaisons des nationalités dans Google Image | Ufunk.net

  4. I’m not sure the experiment holds much water in either case. Basically, Google analyzes 1. everything you search on their site and 2. what you click on from their site and maybe even 3. other websites you visit and probably 4. what is in your Gmail. It does this and then uses its algorithms to predict what results you may want for your next search.

    Thus, the methodology in this study is incredibly, incredibly important. Someone who searches significantly for political themes on Google, for example, is more likely to turn up politicians when they search for people. If you search a lot of music, perhaps you’ll get musicians.

    You and I could search the same time on Google and our top 10 results are probably more likely to be different than the same.

    • If you use Google without signing in or with cookies turned off, it won’t be able to use your history, gmail, etc. when searching.

      When I searched for “Asian” all I got was porn, too.

  5. Pingback: Quand les comparaisons de nationalités dans Google dérapent… | Sotica

    • Me too actually — I should have stressed that I didn’t intend for this to be taken as a serious investigation at all sorry, and would add to the things that Nate mentions that simply different locations will result in different search results too.

  6. The original comparison shows all too well the standards of reasoning to which so many groups of sociologists hold themselves. The first response of any intelligent and reasonably educated person would be to consider confounding factors, such as the facepalm-obvious fact that “Asian” is a common term for a race (which stimulation-searchers would be expected to care about), while “African” by itself is much less so and the others hardly at all. Let’s say there’s a dude of whatever race who totally sexually objectifies white women, as the comparison implies about Asian women. Is it remotely plausible that he searches for “European” instead of, say, “blonde” or “redhead”? This is a trivial and obvious difference in the meanings of the English words chosen. It looks for all intents and purposes like these people (i) didn’t even want to take a second to think things through, (ii) because the grand unreasonable interpretation supports their preconceived agenda.

    Forgive my bluntness, but what kind of a dumbass does one have to be to jump to “The queries speak volumes about objectification, sexualization, and Orientalism”, or even to give someone else a pass on such crap?

    • Actually, your points about the confounding factors are well taken, and I challenge you to find so much as a single sociologist that wouldn’t readily agree with them. Like I said, both the original English search results and my own Korean ones are clearly not serious investigations, and that those sociologists linked to don’t provide such caveats — not unreasonable in a one sentence comment on a picture — does not automatically imply that they’re unaware of what you mention. To assume otherwise smacks of confirmation bias by someone who obviously already has a very low opinion of sociologists.

      I do accept though, that, given those issues, it may have been more accurate to say that the “The query about Asians speak[s] volumes about objectification, sexualization, and Orientalism”, which clearly it does. However, you could still argue that the original line is fine. It does not for a moment imply that only Asian women are objectified and so on, for instance, even if you do personally believe that that is the sociologists’ “preconceived agenda” (again, do you have actual evidence of that agenda?). Moreover, it is still hugely significant that image search results for “Asian” — 60% of the world’s population — give mostly Japanese porn stars, whereas searches for other races (etc.) need further refined into redheads and so on before you likewise get mostly pictures of naked women. Only the results for “Latina”, say, even come close, but of course that term is already much more specific than the sex/gender-neutral one “Asian”.

      Is that indeed due simply to the way the words are used differently in the English language? If so, why do we use the word “Asian” so differently to the others? Or are there some other reason for the results entirely? Whatever the answer(s), that a 2 minute investigation produces such similar results whatever computer you’re using and wherever you are in the world, definitely shows that something’s going on with the way English-speakers conceive of Asians.

      I can easily see myself in the future giving this as a quick and enlightening homework assignment to students, to whom it would be far more beneficial asking them to explain the differences rather than simply giving my own explanation. But if they begin by saying that “Most sociologists are [idiots]” though, add later that my choosing the search terms I did is evidence of my “preconceived agenda”, and finish by saying that I’m “a dumbass”, then they’re not exactly going to get a good grade. And I’m certainly not going to waste any time on your next comment if you do the same again either.

  7. Thanks, James, for your thoughtful response.

    Your point about confirmation bias is well taken, although it leads me to wonder how many blog posts — let alone comments — are capable of avoiding it while also making substantive points. It seems to me that that’s what footnotes and references are for.

    Cards on the table, though. I’m a linguist with education in psychology, and side interests in the origins of language and evolutionary psychology (which very much go hand-in-hand), so that most of my felt hostility to what I called “sociologists” has to do with obscenely distorted summaries of important evopsych research in top-tier sociology journals. Naturally, I don’t think that all or even most sociologists are unprofessional — I’m a regular reader and admirer of your own blog, after all. It occurs to me just now that my opinion might be fairly summarized as that peer-review at the highest levels of the discipline is broken, letting many people get through who think the plural of anecdote is data and that the rhetorical force of a conclusion is more important than the precision of an argument, this in a way that would my thoroughly embarrassing to a reputable cognitive psychology journal.

    You write:

    Moreover, it is still hugely significant that image search results for “Asian” — 60% of the world’s population — give mostly Japanese porn stars, whereas searches for other races (etc.) need further refined into redheads and so on before you likewise get mostly pictures of naked women.

    As I see it, this sentence has three problems. First is the claim that “Asian” refers to 60% of the world’s population. I feel something of an intrusion into my area of expertise by your use of a technical definition appropriate for rare communicative contexts, when the everyday use of the term is very different. I can put the point this way. If I witness a crime in, say, San Francisco, and I tell the police that the perpetrator was “Asian”, when he was Pakistani, do you think the cops are going to question the right person, or be satisfied with my answer? Probably not, because in ordinary usage the term calls up an image of East Asians (and to some lesser extent Central and Southeast Asians), and very many of your 60% don’t fit the bill. If the perpetrator was a white guy speaking with a clear Australian accent, hell, the police would probably want to charge me with obstruction. The primary semantics of the word “Asian” is a certain prototypical appearance (which is mostly what race is), and only secondarily the residents of a certain macrocontinent and nearby island chains.

    Second, neither are “American” nor “European” anything like common racial terms, and “African” by itself is pretty strained if it’s supposed to mean a race and not a place. “White/black/asian/hispanic/Indian/Middle-Eastern” are terms I’d expect witnesses to use in reporting a suspect to the police, for example. The problem with comparing searches for “white” and “black”, of course, is that these are highly frequent color adjectives that return images like polar bears, black cats, chess boards, and outer space. Apples to apples, it’s an accident that “Asian” is the common racial classification that isn’t an even more common word in the language. You’ve got to add a noun to make a reasonable comparison. (I just searched for [Black/White/Asian teens] — “Black teens” gives an even mix of guys and girls looking respectable; “Asian teens” gives sexy girls with only about 3% guys, and “White teens” gives drug using, mocking memes, mugshots, and — oddly — quite a few pics of black people. Which leads into my last point…

    Yes, “white” is not normally used as a modifier when white people are talking about or searching for white people in the U.S., for example. Instead, more precise classification in terms of hair color comes into play. But this seems to be basic human processing of marked vs. unmarked. I live in Korea, and whenever I’m talking about a non-Korean using their Korean name, you can bet that the first categorization I make is their ethnicity and/or nationality, not whether they’re tall or had long hair. But if I’m talking about a Korean, am I going to lead off with the fact that they’re Korean? Why would I do that?

    Yes, there are certainly important racial biases in perception and categorization. I know about the numerous psychological studies on implicit racism, and I also know from personal experience that Koreans used to kinda look the same and now they don’t. My brain has adjusted in the way that brains are wont to do. If English-language porn-searching classifications break down into Asian/Latina/Black/Blonde/Brunette/Redhead, this in and of itself doesn’t tell me much about relative sexual objectification, just a well-known fact about how the brain processes appearance based upon statistical exposure.

    Well, I guess that’s enough for now. Keep up the good work!

  8. I don’t know if this has been brought up, but “Asian” is an age old porn term for east (and south east) asians. It’s almost always on top of the list of the categories on pretty much all porn sites since it begins with an “a”. Maybe that has helped in the way we think of the word since pretty much everyone in the west looks at porn nowadays. The other terms aren’t really porn terms at all. So, to see if it was just limited to “asian” I did a search for some other non-sexual words such as “ebony” and “teen”, and … well… I got even more NSFW results than for “asian”. “granny” (oh god), “shaved”, “hairy”, “schoolgirls” (this was obvious maybe) also gave me an abundance of NSFW images.

  9. Now, this was quite interesting!
    It only signifies that search engines tend to bubble the results quite a bit. And as sad as it is, people spent most of the time searching porn.

    I too was a bit surprised, when I wanted to read up on certain aspects of Mongol customs and was bombarded with sites and pictures of Mongol Mail-order Brides. 8/

    Using bubble & tracking-free search engines like DuckDuckGo yields better results.

  10. That original post is a bit dodgy, even on its own terms: They have different search settings for different searches.
    I did my own search, with safe search off, and didn’t get quite the same shocking results. Of the 24 on the front page, none were pornographic, about 8 or so were in bikinis or obviously suggestive, five were men, another may have been, two were memes, one was a woman with four eyes and two mouths, and the rest were pictures of attractive young ladies – hidden, sadly behind layers of clothes and photoshop.

    Let’s just say the search for redhead had rather more jism in it.

    The points have been made above — white and black are colours, asian is a porn category, latino/latina have gender meaning built in, no one searches for African porn etc. I’d add that it depends on what com[puter you use and which google you use.

    It might work better with nationalities, if you are thinking of actually doing a class activity.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s