A couple of weeks ago, I went into the Smith-Kettlewell Eye Research Institute in San Francisco to hear a presentation by my friend Robert Englebretson. Robert is a professor of linguistics at Rice University. He is blind, and I met him through an email group of blind academics. In addition to his primary linguistic research on Indonesian languages, Robert has begun to research Braille reading using the tools of cognitive linguistics. It’s exciting research that points the way toward better Braille for everyone.

This particular presentation was about a study he had done on Braille contractions with his Rice colleague Simon Fischer-Baum. I haven’t said much about contracted Braille since my very first post, so I’ll recap: in order to save space and increase reading speed, there are a bunch of contractions that reduce the length of common letter combinations and words. Some of these contractions have to appear by themselves, but others can be used as parts of larger words. These include small groups of letters like [er], [ch], [ing], and [the], as well as words like [time], [ever}, and [less]. 

In English Braille, these contractions were determined largely by statistical frequency. The most common letter combinations and words got contracted, meaning the most possible space got saved on the page. In the days when huge tomes had to be hand-Brailled, this made a lot of sense, but it’s a practice that benefits the producers of Braille, and its effect on readers had never been studied. 

For decades, the prevailing assumption in Braille education has held that blind Braille readers expand all of the contracted words into their full print form before interpreting them. If this is true, then all contractions should be equally readable. But are they?

Granted, there are some contracted spellings that everybody agrees are a bad idea. My favorite is this gem:

⠡⠑⠐⠍⠁⠏⠽

In English characters, that’s 

[ch] [e] [mother] [a] [p] [y]

Chemotherapy. Sure, it’s space-efficient—it shortens a twelve-letter word to seven cells. But the two-cell (dot 5, m) contraction for [mother] is so ingrained to mean “mother” that everyone inevitably reads it something like “Key-mother-a-pee.” Is saving the space of five cells on the page worth losing a few seconds every time you have to make sense of this word? Not to me.

Robert set out to show that the same process happens in less egregious situations as well. You don’t often run into contractions as bad as “chemotherapy,” but there are many, many examples of contractions that cause momentary confusion and delays in comprehension. Take ”redo” for example, which can be written in Braille as [r] [ed] [o]. Most people can’t help but read it as “red-o.”

Basically, Robert theorized that reading problems occur when contractions conflict with our understanding of a word’s sublexical structure. We all intuitively understand that some words are made up of smaller units that have been smooshed together. “Redo” is the verb “do,” plus the prefix “re-.” We understand it as a two-part word, and when the [ed] contraction effectively erases the boundary between the two parts, we stumble over the reading.

And this is exactly what he found in a large number of similar cases. I won’t go into the methodology or technical details here, but Robert was able to show that fluent Braille readers took longer to comprehend words when contractions crossed sublexical boundaries than they did when the contractions did not cross boundaries. This implies that Braille readers are reading those contractions as single units, covering up any dividing lines that may exist within them. In other words, the theory that Braille readers uncontract words before they process them does not hold up.

To me, sitting in the audience, Robert’s results seemed right on. To be fair, I remember a time when I did have to uncontract every word I read. When I was first learning contracted Braille, I had to picture every word in my head, building each one letter by letter as I read each cell. But this was just a symptom of my inexperience, and I moved past it quickly. After a month or so, I no longer had to picture the words I was reading. Instead, the cells brought sounds into my mind—whether letters, syllables, or entire words. 

Then, maybe a month ago, something different happened. I brushed my finger over some Braille text and all of a sudden the meaning of the words came into my head. Not the sounds, the meanings. It felt surreal, how automatic it was. I took that for granted with print reading—for most print readers it becomes so automatic that it isn’t even a choice. Text is read as soon as it is seen, whether you like it or not. It was very strange to have that sensation in Braille. 

Right now that only happens with very simple language. Less common words take a little longer, and rare words or weird contracted spellings can trip me up pretty bad. As I gain experience, more will become automatic and everything will come faster, but Robert’s research suggests that some contracted spellings will never stop slowing me down.

The study shows that cognitive linguistics can provide valuable insights into Braille reading, and points the way toward further research. In time, we may be able to more precisely distinguish between contractions that help and contractions that hinder, We may also come to see which other features of the current Braille system improve speed and comprehension, and which do not. Eventually, research like this could influence best practices that optimize Braille for the human beings who read it rather than pages it is embossed on.