Читать книгу The Science of Reading - Группа авторов - Страница 13

PART I Word Recognition

Word recognition is the foundation of reading; all other processes are dependent on it. If word recognition processes do not operate fluently and efficiently, reading will be at best highly inefficient. The study of word recognition processes is one of the oldest areas of research in the whole of experimental psychology (Cattell, 1886). In the opening chapter of this second edition of The Science of Reading, Perfetti and Helder provide a comprehensive overview of recent advances in the study of reading, with particular emphasis on skilled reading in adults. This chapter sets the stage for the chapters that follow. Guided by the analogy of reading as a rapidly moving stream, they use the Reading Systems Framework to show how different knowledge sources, including critical lexical knowledge, underpin the processes that take the reader from word identification to comprehension (and how these may be disrupted in reading disorders). In so doing they consider the range of methodologies that have been used to investigate reading processes, including neurophysiological measures and brain imaging, emphasizing that computational methods and language corpora together add precision to theoretical proposals.

The science of reading should not be specific to one language or one writing system. Perfetti and Helder highlight not only universals in reading processes, but also some of the patterns of variation found across the world’s orthographies. Together these suggest that writing systems evolve to accommodate properties of the languages they represent. Regarding reading acquisition they note that “…effective experience [in reading] in which children read words successfully and achieve comprehension—is the only certain path to establishing rapidly accessible orthographic representations.”

In the first edition of the book, two models that framed much subsequent research were considered: the dual‐route model and the connectionist framework, respectively, by Coltheart (2005) and Plaut (2005). In this edition, Seidenberg, Farry‐Thorn and Zevin consider theoretical models of word recognition with a critical appraisal of computational models of reading aloud. Reminding us that feedback between empirical data and theoretical models is a powerful way of investigating cognitive processes and generating hypotheses, they argue that the standard of proof required is high – that the behavior of the model should align with that of the reader. In addition, the model should accommodate data from multiple languages and different writing systems. Seidenberg et al. focus on learning to read in English where a significant challenge is to model reading across words that vary in the consistency of their spelling‐sound correspondences – an issue that can be addressed by including mappings between semantics, orthography and phonology.

One advantage of connectionist models is that as they learn, they become sensitive to orthographic consistencies inherent in the writing system and the knowledge that accrues is used to feedback and further refine representations of that knowledge. Echoing the sentiments of Perfetti and Helder, computational models can guide reading curricula and make recommendations for both explicit instruction and implicit learning via reading experience. Advances in computational models of learning to read are likely to provide an important future direction with important educational implications.

In the first edition of this book, Lupker (2005) and Van Orden and Kloos (2005) reviewed a large body of experimental evidence concerned with how adults recognize printed words. Any complete model of word recognition has many phenomena to explain: that people perceive letters more efficiently when they are embedded in words than presented in isolation, that high‐frequency (i.e., more familiar) words are recognized more easily than less familiar words, and various patterns of priming. One conclusion that emerged powerfully from these earlier reviews was the need for interactive models in which activation of orthographic and phonological information reciprocally influence each other. In this second edition, three chapters on skilled visual word recognition continue and expand on this theme. Grainger begins by discussing the orthographic processes that mediate between vision and language when translating letters into word representations. He argues that words are the basic units of reading, and that letters are the basic units of words. Careful experimentation has shown that skilled readers use abstract information about the identity and position of letters to recognize words, and that processing is cascaded and interactive. Beyond word identification, he suggests that sentence reading – or at least the interface between whole‐word representations and the sentence – is governed by mechanisms that share the same general characteristics, namely processing that is interactive, cascaded, and performed at least partly in parallel. These allow the syntactic and semantic representations required for sentence comprehension to be activated rapidly from orthography. Grainger notes that word recognition and sentence processing have tended to be investigated within two independent lines of research, a theme echoed in several chapters of the handbook. Future work needs to better integrate evidence across words, sentences and texts to fully understand how reading happens.

While Grainger focuses on orthographic processes in skilled reading, Brysbaert tackles the role of phonology. Taking a historical perspective, he describes how researchers have defended very different positions over the years, from full phonological mediation between print and meaning through to no phonological involvement at all. It is now widely accepted that phonology plays a central role in skilled word reading. In alphabetic writing systems, phonology is particularly important in the early stages of reading development, when the ability to assemble the phonological form of an unknown word is foundational. Brysbaert’s review makes clear that phonology continues to be engaged automatically in skilled word reading. At least in alphabetic writing systems, orthographic and phonological processes jointly contribute to visual word recognition and this is achieved via coding interactions in the brain. Brysbaert closes by considering how extant models of word recognition accommodate the central role of phonology.

Since the first edition of the handbook, word recognition research has expanded to include detailed consideration of morphology. This progress is reviewed by Rastle in the context of processing English. Morphemes are defined as the smallest unit of meaning and are either stems or affixes. The majority of words in English (and many other languages) are built from morphemes. Although there is a tendency to think of the relationship between print and meaning as largely arbitrary, morphological structure represents an important interface between orthography and meaning. Rastle provides many examples of graded systematicity in the mappings between spelling and meaning at the level of morphemes. Skilled readers are highly sensitive to these “islands of regularity,” many of which are preserved in the writing system, often at the expense of maintaining regularity between spelling and sound. This means that morphological information is highly visible in the writing system. Rastle reviews evidence showing that morphological information is activated by skilled readers during the course of visual word recognition and discusses how different models of word recognition can capture these influences; like earlier chapters in this section, she emphasizes the value of computational approaches. Rastle reminds us that the goal of reading words is to rapidly compute their meaning, and therefore that the goal of learning to read is to develop a system that maps orthography to meaning quickly, directly, and accurately. Although morphological effects in skilled word recognition are well documented, far less research has considered how morphemic knowledge becomes represented in the reading system as children learn to read. Rastle identifies this as an important direction for future work, highlighting the likely role that reading experience plays as the substrate for establishing probabilistic mappings between orthography and meaning.

References

1 Cattell, J. M. (1886). The time taken up by cerebral operations. Mind, 11, 220–242.

2 Coltheart, M. (2005). Modeling reading: The dual‐route approach. In M.J. Snowling & C. Hulme (Eds.), The science of reading: A handbook (pp. 6–23). Blackwell Publishing. https://doi.org/10.1002/9780470757642.ch1.

3 Lupker, S. J. (2005). Visual word recognition: Theories and findings. In M.J. Snowling & C. Hulme (Eds.), The science of reading: A handbook (pp. 39–60). Blackwell Publishing. https://doi.org/10.1002/9780470757642.ch3.

4 Plaut, D. C. (2005). Connectionist approaches to reading. In M.J. Snowling & C. Hulme (Eds.), The science of reading: A handbook (pp. 24–38). Blackwell Publishing. https://doi.org/10.1002/9780470757642.ch2.

5 Van Orden, G. C., & Kloos, H. (2005). The question of phonology and reading. In M.J. Snowling & C. Hulme (Eds.), The science of reading: A handbook (pp. 61–78). Blackwell Publishing. https://doi.org/10.1002/9780470757642.ch3.