?

Log in

No account? Create an account
 
 
06 August 2004 @ 13:24
Barack Obama's father taught me how to read  

I've been seeing a lot lately about Barack Obama, the young Illinois politician who reportedly gave a very impressive speech at the Democratic National Convention. (I say reportedly because I haven't read or heard the speech itself.) Many of the articles I've encountered, such as this one by William Finnegan in The Beau and the Butterfly, mention the fact that Obama is the son of a Kenyan economist.

I invariably have to read that phrase twice. What it means, of course, is that Barack Obama, Sr., was a practitioner of the Dismal Science, and that he was born in the country of Kenya, in eastern Africa. What I think it says, the first time my eyes pass over any given instance of it, is that the elder Mr. Obama was a Keynesian economist, a follower of John Maynard Keynes. (For all I know, he may have been that, too; none of the articles I've seen actually says what kind of economics he did.) Then I go back and look again, and, nope, it says Kenyan quite clearly.

This little repeated glitch in comprehension actually reveals quite a lot about how reading works:

  1. Fluent readers of English, or of any other language that uses an alphabetic writing system, do not read letter-by-letter, translating letters into sounds and then sounds into meanings, as neophyte readers are taught to do on Sesame Street. Rather, we recognize words as wholes, jumping directly from the larger graphic image to the lexical entry. (This means, by the way, that while alphabetic writing systems are kinder to learners, logographic systems, like that of Chinese, are more efficient for the grown-up power user of the language.) If I were sounding the article out one letter at a time, there's no way I would mistake Kenyan for Keynesian.
  2. Given that we read word-by-word rather than letter-by-letter, what makes one word pop to mind more readily than another that has a similar shape on the page? Well, frequency is one factor: we're quicker to recognize words we see more often, and correspondingly more likely to mistake an infrequent word for a frequent one than vice versa. But raw frequency doesn't explain this example, at least not for me. I encounter the word Kenyan at least as often as I encounter the word Keynesian. (Probably more often, given that I'm more interested in things like coffee and the tone system of Dho Luo than I am in things like the propensity to consume and the inducement to invest.)
  3. So the obvious explanation is semantic priming, or some sort of relativized frequency—I read Kenyan as Keynesian because I expect Keynesian to go with economist. But that appears to mean that by the time I get around to (mis)identifying Kenyan, I've already successfully read the word economist. (Another possibility is that I'm finding Keynesian economist as a single entry in my mental lexicon, and matching the pair of words on the page to it.) So I don't read strictly from left to right; I'm processing at least two words at a time, and sometimes the second one goes down faster than the first.
  4. And yet this seems to have no effect on my reading of the sentence as a whole: I don't stop and think, Huh? The son of an economist Keynesian? What's that supposed to mean? In no conscious way am I at all aware of the fact that I've interpreted the word on the right before the word on the left. So the syntactic component of my comprehension blithely parses away, assigning a structure on the basis of the order in which the words appear, not the order in which they're recognized. The whole system is quite remarkably robust.
 
 
 
acw on 30. August, 2004 14:19 (UTC)
A possible model
Perhaps we have a pipeline of word-interpreting agents. Hypothesis: Each time we scan a word, an agent is assigned to interpret it. The agents make progress at varying rates; an agent assigned to interpret "the" will recognize the word shape almost instantly; an agent assigned to "Kenyan" will take longer. While an agent is working, it may be informed by other active agents. So the "Kenyan" agent might well be still working when the "economist" agent reports in: "Hey, I'm 95% sure this word is 'economist'." That report could predispose the "Kenyan" agent to misreport "Keynesian". While older agents (corresponding to words occuring earlier) are on average closer to an answer than newer ones, in general some number of agents are usually working roughly in parallel, and the judgements of newer ones can affect the older ones.