A personal statement by Chief Scientist, Julie A. Van Dyke.
I am so excited to be able to bring to life a tool that I believe will be a game changer for reading instruction!
The latest results from the National Assessment of Educational Progress (nationsreportcard.gov) reports that only 63% of 4th and 70% of 8th and 12th grade students read at what’s termed as a Basic level. The specifics of what that means is different for each grade (see here for a description). Beyond simply knowing word meanings, it means being able to find a specific detail and make a simple inference from at least one section of the text, identifying basic text elements like main idea, order of events, reasons or motivations for actions or conclusions—like who did what to who and why, and forming an opinion based on the specific content of the text.
To my mind, this boils down to not being able to understand—or use—the information being read.
Nearly 1/3 of our students, and 30% of high school graduates simply can’t comprehend—and the NAEP tests this on all sorts of texts, including novels, poems, non-fiction, and the kinds of texts you read in a science or social studies class, in a job-training manual, or on a medical after-care report.
This paints a really alarming prognosis for the future of our country—but what’s even worse—these percentages haven’t changed at all since the NAEP was first administered in 1992!
I believe Cascade Reading is uniquely positioned to move this needle in a way that no other instructional methods have been able to do for the last 30+ years—and here’s why:
When I came to Haskins Laboratories in 2002, the scientific community was just beginning to reap benefits from technological advances in neuroimaging that helped us discover how particular reading interventions affect a child’s brain. Haskins scientists like Isabel Lieberman, Don Shankweiler, Alvin Liberman, and Ignatius Mattingly had been talking about the primary role of phonemic awareness in reading since the 70s, but when Ken Pugh and colleagues from Yale University began using functional MRI to directly examine effects of phonological- vs. meaning-based interventions, we had the first hard evidence to show that phonics-based reading instruction actually changes the way the brain functions when we read.
And now, over 30 years later, both the scientific and educational community have a really solid hold on what it takes to teach kids to read words. Phonics-based programs abound, and professional development programs like LETRS (Language Essentials for Teachers of Reading and Spelling) are seeing broad adoption at the state and district level, which means that teachers are even more prepared than ever to teach word decoding in a way that the brain will respond to most naturally.
Indeed, states that have embraced professional development models that include LETRS (e.g., Louisiana, Mississippi, Alabama), and other evidence-based training programs, saw the greatest growth on the NAEP scores in 2022.
Yet even with those gains, these states are still reporting no better than 66% of their students reading at a Basic level. The reason for this is that our emphasis on phonics-based instruction—and what’s termed the “Science of Reading”—is most fundamentally directed towards learning to read (or decode) WORDS.
But reading isn’t just about recognizing words. It’s about comprehension—about recognizing details, reasons, and main ideas in a text, knowing how they relate to each other so that inferences can be made, lessons learned, and opinions formed. So, if we could have 100% adoption of phonics-based or science-based reading instruction in every district in the country, we still wouldn’t reach the holy grail of achieving higher competencies in comprehension ability.
For this we need more. Just like phonics instruction emphasizes the linguistic structure in words, so too we need instruction that emphasizes the linguistics structure in sentences—and that is syntax. Word learning is built on phonemic/phonological awareness; text learning is built on syntactic awareness.
I came to the world of reading research from an unusual direction—my training began in computer science and computational linguistics, and focused on the mechanisms of sentence parsing. This is a computational process through which words in a sentence are analyzed to identify phrasal units (aka syntactic constituents), which are then set in relation to one another within a hierarchical structure.
This field is burgeoning with formal linguistic descriptions of grammatical relationships, and highly specified processing theories that simulate how the brain creates syntactic structures with millisecond precision. Mounds of data have been martialed to debate such questions as:
- Are syntactic and semantic features considered in sequence or in parallel?
- Does the parser create all possible interpretations simultaneously, or only the most likely one?
- What type of memory mechanism is used to access previously processed information?
- How much information is available at any one time and for how long?
- What are the consequences when memory capacities are exceeded?
- What mechanisms are brought to bear when an initial understanding is discovered to be in correct?
- How quickly can mistakes be corrected and are the corrections complete, or do the initial misunderstandings linger?
This body of knowledge is crucially relevant to any theory of how comprehension occurs and what happens when it fails—yet it has been glaringly disconnected from the Science of Reading dialogue. There is a wealth of knowledge about how comprehension happens, but few have connected it to educational practice. Cascade Reading seeks to change this.
For the first time, we are offering a system that can make the syntactic structures of a sentence easily accessible to a reader—and neither the reader nor the teacher need to know any of the specifics about how these structures came about.
Rather, the reader simply has to read. The structure of the cascade does the rest of the work, by helping the reader to know where the syntactic phrases are, and how they are related to one another. This is the work that the brain of a skilled reader does on its own, but now the formatting can provide support for any reader to help correctly interpret the relationships within the sentence. Over time—we believe—exposure to Cascade Format will train the brain to more easily recognize these relationships, and ultimately make it easier to understand details within and across the sentences of a text.
Truth be told, the idea we bring forth in Cascade Reading is not new. A number of scientific studies were published as early as the 1970s that considered whether reading could be aided via various methods that changed the format of text, including limitations on how much was presented on a line, or where in vertical or horizontal space words were positioned.
But Cascade Reading is the first approach to be based entirely on automated computational linguistic parsers that embody both theoretical linguistics and cross-linguistic universals. Plus, our use of automated parsing means that we are positioned to transform any text, of any length, for immediate use. This transforms the ad hoc text formatting that was featured in specialized research studies of the past, to a system that can be immediately deployed in any learning setting, or individual reading situation.
In this age of personalization, we at Cascade Reading believe that readers deserve to have choices about how they consume text. Cascade Reading seeks to transform the reading experience so that comprehension becomes more accessible and enjoyable to all.
Please try it out and let me know your thoughts!
Dr. Julie Van Dyke
Chief Scientist, Cascade Reading
