Models of reading

Reading is a complex skill involving the orchestration of a number of components.  Researchers often talk about a “model of reading” when talking about only one aspect of the reading process (for example, models of word identification are often referred to as “models of reading”). 

Models of Word Identification

Various models of word identification have been proposed during the last 30 years. Two models that have received  most attention are the Dual Route Cascaded (DRC) model and various parallel distributed processing or connectionist versions of what have become known as triangle models These two models because they provide contrasting frameworks for explaining the two most prominent theoretical perspectives in the on-going debate about how words are identified and represented in the mental lexicon. This debate has specifically focused on whether word identification is guided by linguistic “rules” that are used to access a word’s pronunciation and/or meaning from its orthography, or whether this process is more accurately described as being one in which different types of lexical information provide mutual “soft” constraints on the pronunciations and/or meanings that are generated during word identification. 

Models of Syntactic Parsing

There are many existing models of sentence-level processing that explain how the linguistic structures and constraints (e.g., syntax) guide the construction of the representations that are necessary to understand individual sentences. Thus, these models presumably take as bottom-up input the meanings of individual words that are provided by the types of word-identification models .These models can be classified into three broad categories to include the various garden-path models; constraint-based models and various models that have been implemented using connectionist frameworks.

Models of Discourse Processing

In contrast to the models above, the models which explain discourse processing are more difficult to categorize into groups that define opposing positions on some central theoretical question.  The models instead tend to describe certain aspects of the processes and representations that are necessary to connect the meanings of individual sentences into more global representations that support text comprehension.

Examples of these discourse-processing models include the Construction-Integration Situation-Space; Landscape; Resonance and Distributed Situation Space models, along with several connectionist and production system models of discourse processing.

Models of Eye Movement Control in Reading

There are now a large number of models of eye movement control in reading. The development of such models was motivated by the appearance of the E-Z Reader model (Although there were prior verbal and implemented models  that attempted to document aspects of eye movements in reading, E-Z Reader clearly stimulated the development of a number of competing models, of which SWIFT is generally regarded as the main competitor. Other models include Mr. Chips; EMMA; SERIF; Glenmore; SHARE and the Competition-Interaction model.

Cognitive processes that support reading are complex. This is undoubtedly the reason why to date efforts to develop computational models of ”reading” have largely been directed towards explaining only one or two components of the reading processing (e.g., how printed words are identified), with little effort being directed towards explaining how these components interact with the remaining processes that are important in reading (e.g., attention allocation). Although this strategy may have proven effective in that it allowed cognitive scientists to focus on their own particular problems, the approach may be limited in that it tends to provide a narrow view of what actually transpires in the minds of readers.

The Simple View of Reading

See https://dera.ioe.ac.uk/5551/2/report.pdf

Learning to read consists of developing skills in two critical areas: (1) Reading each word in texts accurately and fluently and (2) Comprehending the meaning of texts being read. This is known as the Simple View of Reading. The Rose Report made a number of recommendations for the teaching of early reading and  replaced the ‘Searchlights’ model used in early literacy frameworks in schools. It makes clear that there are two dimensions to reading – ‘word recognition’ and ‘language comprehension’. These two dimensions are represented as the ‘simple view of reading’. It is a formula demonstrating the view that reading has two basic components: word recognition (decoding) and language comprehension. The Simple View formula shows that a student’s reading comprehension (RC) score can be predicted if decoding (D) skills and language comprehension (LC) abilities are known.

The Simple View formula presented by Gough and Tunmer in 1986 is:

Decoding (D) x Language Comprehension (LC) = Reading Comprehension (RC)

Decoding (D) is defined as “efficient word recognition” (Hoover & Gough, 1990). This definition goes beyond the traditional definition of decoding as the ability to sound out words based on phonics rules. The meaning of decoding expands to include fast and accurate reading of familiar and unfamiliar words in both lists and connected text (Gough & Tunmer, 1986).  Students read a list of pseudowords to assess decoding.

Language comprehension (LC) is called by several other names in various studies, including linguistic comprehension, listening comprehension, and comprehension. All of these terms are defined as the ability to derive meaning from spoken words when they are part of sentences or other discourse. Language comprehension abilities, at a minimum, encompass “receptive vocabulary, grammatical understanding, and discourse comprehension” (Catts, Adlof, & Weismer, 2006). Students listen to a passage read aloud then retell the passage combined with answering oral questions that were not addressed in the retell.

Reading comprehension (RC) differs from language comprehension because of the reliance on print, as opposed to oral language, to perceive the words and derive meaning (Hoover & Gough, 1990). In other words, language comprehension becomes reading comprehension when word meaning is derived from print. It is possible to have strong language comprehension and still be a poor reader if there is difficulty with decoding. Students read a passage then retell the passage combined with answering oral questions that were not addressed in the retell.

Kamhi (2007) eloquently describes the differences between decoding (word recognition) and comprehension. Decoding is “a teachable skill” compared to comprehension, which “is not a skill and is not easily taught.” Kamhi explains that word recognition is a teachable skill because it “involves a narrow scope of knowledge (e.g. letters, sounds, words) and processes (decoding) that, once acquired, will lead to fast, accurate word recognition.” Kamhi further writes that comprehension “is not a skill.  It is a complex of higher level mental processes that include thinking, reasoning, imagining, and interpreting.” The processes involved in comprehension are dependent on having specific knowledge in a content area. This makes comprehension largely knowledge-based, not skills-based.

Scarborough’s Reading Rope

Hollis Scarborough—creator of the famous Reading Rope spoke of skilled reading as resembling the “strands” of a rope, using pipe cleaners to illustrate the interconnectedness and interdependence of all the components. The Reading Rope consists of lower and upper strands. The word-recognition strands (phonological awareness, decoding, and sight recognition of familiar words) work together as the reader becomes accurate, fluent, and increasingly automatic with repetition and practice. Concurrently, the language-comprehension strands (background knowledge, vocabulary, language structures, verbal reasoning, and literacy knowledge) reinforce one another and then weave together with the word-recognition strands to produce a skilled reader. This does not happen overnight; it requires instruction and practice over time.

Reading rope image

Reading is also supported through parental involvement. Parents have a major role in helping their children develop reading in pre-school and during the school years. Peer learning may be added to the school curriculum to individualize and differentiate reading (Topping, 2019)

References

Catts, H.W., Adlof, S.M. & Weismer, S.E. (2006) Language Deficits in Poor Comprehenders: A Case for the Simple View of Reading. Journal of Speech Language and Hearing Research, 49(2) pp.278-93

Gough, P. and Tunmer, W. (1986). Decoding, reading, and reading disability. Remedial and Special Education, 7, 6–10.

Hoover, W. and Gough, P. (1990). The simple view of reading. Reading and Writing: An Interdisciplinary Journal, 2, 127–160.

Kamhi, A. (2007). Knowledge deficits: the true crisis in education. ASHA Leader, 12 (7), 28–29.

Rose, J. (2006) Independent review of the teaching of early reading, Nottingham: DFES Publications

Scarborough, H. S. (2001). Connecting early language and literacy to later reading (dis)abilities: Evidence, theory, and practice. In S. Neuman & D. Dickinson (Eds.), Handbook for research in early literacy (pp. 97–110). New York, NY: Guilford Press.

Topping, K. (2019) Children's Reading Development and Instruction.