Common Core and 'Close Reading': Effectiveness Questionable without Outside Knowledge

RCEd Commentary

“Close reading” has become strongly associated with the Common Core State Standards, as it’s touted as the reading technique that will allow students to get out of texts what they are meant to (and hence, score well on Common Core-aligned assessments). We might ask whether psychologists know anything about the process of reading that could influence how we think about close reading.

As I’ve seen it described, close reading has three critical features. First, we assume we will spend a good deal of time with a text. We will not simply read, but reread, and likely reread again. The first reading may be devoted to straightforward comprehension, but further readings will uncover other layers of meaning, allusions, techniques of authorship, and so on.

Second, the extended time spent on a text will be devoted mostly to the author’s words. We will pay close attention to the particular words used, to the structure of the argument, and so on.

Third, we will view a text as being self-contained. We will only draw conclusions that are defensible via the author’s words. In fact, we will read the text as though we know nothing about the subject at hand; the author’s words will be not only necessary for our interpretation, we’ll consider them sufficient.

It’s that last bit that seems crazy to me.

Rereading? Sure. Paying close attention to the author’s words? Great idea. But pretending that one’s knowledge is not relevant to interpreting a text conflicts with how writers write and with how readers read.

Writers count on their audience to bring knowledge to bear on the text. This principle is easiest to see in extreme cases, as when I wrote, “Researchers have utilized the savings in relearning paradigm in a variety of settings since Ebbinghaus developed the tool over a century ago,” as the opening sentence of an article (Willingham & Keisler, 2007). The editor did not protest that I should define the paradigm or identify Ebbinghaus because it was a professional journal; he knew that the audience had this information and definitions would be redundant.

Things get trickier when it’s not plain to the reader that he lacks information that is important to understanding the text. Researchers Eli Gottlieb and Sam Wineburg offered a wonderful example in a paper several years ago (Gottlieb & Wineburg, 2011). They asked Clergy, scientists, and historians to read George Washington’s Thanksgiving Proclamation of 1789. Clergy and scientists focused on Washington’s invoking the “providence of Almighty God,” and other religious phrasing, with clergy applauding the Christian tone, and scientists upset by it.

Historians, in contrast, focused on what the document did not say; it did not mention Jesus, nor salvation, nor Christianity. They saw the document as Washington’s self-conscious attempt to craft a statement that would be acceptable to the diversity of religions practiced in the United States, and in so doing send a message of religious tolerance and separation of church and state. That Washington knew his audience may be adduced from the fact that clergy at the time protested the lack of overt Christian references.

No amount of close reading restricted to the text would lead present-day students to this interpretation.

Writers’ expectation that readers will deploy their background knowledge in interpreting a text is well founded. It’s pretty well impossible not to bring your knowledge to bear on a text. For example, suppose you read:

David: Weren’t you and Linda going to take one of those cruises to Alaska?

Michael: Yeah, but she could only get time off in January.

To properly interpret the meaning you must coordinate two facts: that Alaska is unpleasantly cold in January, and that the weather matters to the pleasure afforded by a cruise. Note that neither fact is mentioned in either sentence. Note, too, that you don’t need to stop reading and wrack your brain, trying to find a connection between the two sentences. It just seems obvious.

The feature of cognition behind the scenes is called semantic priming. When you think about a concept, related ideas are primed, that is, they are more readily available for use. One of the first demonstrations of priming (Meyer & Schvaneveldt, 1971) used a very simple task: two letter strings appeared on a screen and subjects were to push one button if both formed words, and another button if either was not a word. The interesting finding was that people were faster to push the “yes” button if the words were related in meaning (e.g. “nurse” and “doctor”) than if they were not (“shell” and “doctor”). The interpretation is that reading “nurse” primes a large number of related concepts: doctors, hospital, vaccines, white shoes, and so on. So when “doctor” appears as the second word, you’re a little faster to verify that it is indeed a word. Background knowledge is invoked as a matter of course when we read.

So the nature of writing and the nature of reading seem contrary to one aspect of close reading, namely, the idea that we can put a fence around a text, and read it only with reference to the contents of the text.

That’s not to say that everything we do in schools ought to be natural. It’s natural to think about numerosity on a logarithmic scale, but that doesn’t mean we should teach math that way. Still, it seems a valid question to ask whether this artificial type of reading is likely to be useful in students’ ongoing work in and out of the classroom. Except in very restricted academic settings -- that is, among people who like close reading -- it’s not obvious to me how this sort of reading will serve students well.

Careful study of language, focus on the author’s words, assumption that rereading pays off: yes. Excluding knowledge outside of the text: no.

References

Gottlieb, E. & Wineburg, S. (2011). Between Veritas and Communitas: Epistemic Switching in the Reading of Academic and Sacred History, Journal of the Learning Sciences 20, 1–46.

Keisler, A. & Willingham, D. T. (2007). Non-declarative sequence learning does not show savings in relearning. Human Movement Science, 26, 247-265

Meyer, D. E. & Schvaneveldt,R.W (197). Facilitation in recognizing pairs of words: Evidence of a dependence between retrieval operations.  Journal of Experimental Psychology, 90, 227-234.