Chapter 2 Theoretical Perspectives on Reading, Reading Comprehension
2.3 A Cognitive perspective: reading for meaning
2.3.1 Information-processing models
The cognitive perspective initiated the development of a number of information- processing models which received ―varying degrees of research support‖ (Sadoski, 2008:39) and were concerned with the processing, storage and retrieval of knowledge from the mind. Examples of information-processing models include descriptions of reading as top-down, bottom-up and integrated processes. The information-processing view of reading is predominantly a text-driven view which regards reading primarily as an individual act consisting of processing steps that are separate and measurable – the sum of the steps constitutes reading. Furthermore, since the information-processing view suggests that every reader ―reads‖ in this way, the assumption is made that the output will be the same for every reader (Bernhardt, 1991:8).
The bottom-up models generally imply that reading is initiated at the ‗bottom‘ level of text structure, namely at grapheme, morpheme and word level, and that in constructing meaning readers work their way ‗up‘ to larger text structures such as phrases, sentences and paragraphs (Hedgcock & Ferris, 2009:17). Gough is a proponent of the bottom-up account of reading and believes that reading takes place ―letter by letter, from left to right‖ (Gough, 1991:142). Gough is of the opinion that a reader should be able to find meaning irrespective of the context and its influences, and that the reader arrives at meaning through a process which starts with sound- symbol relationships, moves to vocabulary, then the rules of syntax before acquiring meaning. LaBerge and Samuels (1974) propose a refined bottom-up model which identifies memory as an important part of the reading process, and suggests that readers automate their reading skills by practicing specific processing skills (such as grapheme identification), which as they became more automatic, places less demand on working memory. Overall, according to the bottom-up process a set pattern is followed in a set order during the reading process and meaning is not derived until the print has been decoded (Cairney, 1990:17).
More recent research seems to confirm that that there is a link between memory, specifically working memory, and reading comprehension (Borella, Cornoldi & De Beni, 2009; Daneman & Carpenter, 1980; Baddeley et al., 1985). Skilled readers seem more able to use their working memory for constructing meaning during reading, compared to poor readers who use up much of their memory for basic skills
such as word recognition, text knowledge and syntax and are therefore less able to concentrate on finding the meaning in a text (Juel, Griffith & Gough, 1986).
Top-down views of reading contrast with bottom-up models by viewing readers as the initiators of the reading process; readers have expectations about the text and, while reading, check which expectations are correct and modify those that do not match the text information (Hedgcock & Ferris, 2009:23). Goodman (1967) describes reading as a ―psycholinguistic guessing game‖ and suggests that the reading process
starts with the search for meaning and that this directs everything else the reader
does. Goodman found that readers who read words in a story context as opposed to unrelated word lists were able to read and understand more words, thereby implying that using context aided comprehension and word identification. Proponents of the top-down model emphasise that, given the wide range of literacy events that confront readers, human memory is not sufficient for the amount of information that must be processed, stored and retrieved according to the bottom-up model. Smith (2004) argues against the bottom-up models by stating that comprehension is gained from its context and that the ―… meaning readers comprehend from text is always relative to what they already know and to what they want to know‖. In other words, top-down models highlight the potential interaction of processes and participants, the acknowledgement of reader and text, rather than the sequential activation of processes.
However, as Hedgcock & Ferris (2009:27) state, the bottom-up and top-down view of reading ―are not dichotomously related‖ and that an interactive view of reading allows one to draw from the strengths of both paradigms while keeping the multiplicity of practices, processes and participants in mind. Hedgcock & Ferris (2009:28) continue by stating that from a socio-cognitive, reader-based view, L2 reading involves five ―indespensible‖ components, namely the literate context, the text, the reader (and his/her purpose for reading), text processing operations and the reader‘s reconstructed message.
Graves, Juel & Graves (1998:11) discuss Rumelhart‘s (1985) Interactive Model which emphasises that ―both the reader and the text play a role in reading‖. Rumelhart proposes the possibility of parallel processing, i.e. the simultaneous instead of linear processing from more than one source, which sees the reader draw simultaneously, but selectively, on a number of resources, such as visual, orthographic, semantic, syntactic and schematic information. Rumelhart‘s model allows higher level
processing (such as comprehending the meaning of a sentence) to assist lower level processing (such as word identification). Davies (1995:66) regards the Interactive Model as important because it leaves room for reader differences (e.g. first language vs. second language readers) and lends support to the importance of beginner readers being sensitive to all sources of information rather than only one source, such as sound-letter correspondences. The latter point is emphasised by Emergent Literacy supporters who promote reading instruction which encourages readers to integrate different kinds of information and demonstrate alternative ways of using information.
A central component of the cognitive view of reading (as proposed by psycholinguists) is that readers rely on a cueing system to help them construct meaning from the text they are reading. The underlying assumption is that readers make predictions as they read about what the text will say based on their existing knowledge in these areas. In other words, the reading process is initiated with expectations about texts and the information represented by the texts; as they read, readers ―sample information to determine which expectations [predictions] were accurate‖ and modify expectations that do not match the text content (Hedgcock & Ferris, 2009:23).