A few months ago I began reading a packet of letters from the Great Depression. They were handwritten in cursive by Ruth Cross, a native Texan who borrowed and bartered her way into building a country house in Litchfield County, Conn. Or perhaps I should say that I began trying to read them, because deciphering her letters at first seemed a hopeless task.
I was facing the same problems that jam up computers—which are notoriously poor at reading cursive. First, the individual letters that make up each word are often connected to each other, and it is not clear where one letter stops and another begins. A space between letters usually indicates the start of a new word but may also just be stylistic. Second, each writer has his or her own style of writing, usually pretty consistent within and across writing samples, but not always. And third, writers tend to get sloppy. Even they may not recognize what they have written.
Bit by bit, I figured out the missives of Ruth Cross. Now I can breeze through them.
But my initial difficulties understanding her correspondence reflect a real difference in the cognitive processes involved in reading print and reading cursive.
Anthony Barnhart, a lecturer in psychology at Northern Arizona University, told me this in a recent e-mail: “Everything we think we know about reading is based on the simplest case of pristine, unambiguous printed words. Thus, theories of word perception drastically underestimate the capabilities of the human perceptual system (and miss a big part of the picture—how we resolve ambiguous perceptual information). As a consequence, most models of word perception are incapable of handling the complexity of written words.”
When Barnhart was a graduate student at Arizona State University, he and a colleague, professor Stephen Goldinger, began studying cursive reading. They had observed that the small scope of research done on cursive reading stood in stark contrast to the prodigious efforts that have been made to figure out what happens when we are confronted with a single printed word. Since printed words are so controllable and precise, they permit researchers to focus on the factors that come into play when we translate letters into sounds.
In a sort of thousand-millisecond dash, we read some words quickly and others more slowly. Why? What is the difference between high-frequency words and low-frequency words? Regular spelling and irregular? Words with high imageability and low imageability?
Our speed at reading particular words can depend on where we’re from—one project found that British readers recognize such words as “nightie” and “yachtsman” much faster than Americans, while Americans kill when it comes to “Boston,” “homer,” and “butts.” The really illuminating work, however, is being done with cursive script.
In a study forthcoming in Psychonomic Bulletin & Review and published on the journal’s website in April, Barnhart and Goldinger set out to explore how cursive writing affects our ability to read. To do so, the researchers rotated printed and cursive words 90 degrees (both clockwise and counterclockwise). They presented these words, one at a time, for exposures of 500 milliseconds, to undergraduates at Arizona State University. Then they asked the students to type the word they had just seen.
When the words were presented in the traditional horizontal position, the error rates for printed words averaged under 5 percent. Errors for cursive words, Dr. Barnhart said in an e-mail, were significantly higher, averaging close to 16 percent.
But rotating the words plus or minus 90 degrees elevated this contrast to a different stratum. The error rate for printed words remained modest (9 percent), while the error rate for cursive took a giant leap to 53 percent.
This trebling of the cursive error rate reflects something notable: When we read handwritten words, we depend to a great extent on whole-word perception. Like faces, handwritten words have a configural identity that is important to their being recognized, and that configuration can withstand some distortion (as when words are rotated), but not too much: Beyond a certain point, the identity is lost.
With printed words, meanwhile, it appears that we rely on this whole-word perception far less. The authors noted that even when students were not able to correctly read a rotated printed word, they often came up with a few letters and a guess. They had no such luck when they tried to read rotated, handwritten words. As the researchers wrote, “Stated simply, while performing this task, on numerous trials the stimulus appears and disappears, and the observer experiences no perceptual coherence at all.”Continued...