Figure 118

Some of the processes involved in reading. Adapted from Ellis and Young (1988).

grapheme-phoneme conversion should mean they find it very hard to pronounce relatively unfamiliar words and non-words.

Phonological dyslexics fit this predicted pattern fairly well. Phonological dyslexia is a condition in which there are particular problems with reading unfamiliar words and non-words. The first case of phonological dyslexia reported systematically was RG (Beavois & Derouesne, 1979). In one experiment with 40 words and 40 non-words, RG successfully read 100% of the real words but only 10% of the non-words. Similar findings with patient AM were reported by Patterson (1982). AM had problems in reading function words (e.g., with, if, yet), but was very successful in reading content words (nouns, verbs, and adjectives). In contrast, he managed to read correctly only 8% of a list of non-words.

SPEECH OUTPUT GRAPHEME-PHONEME

LEXICON CONVERSION

(stores spoken {converts spellings forms of words) into sounds)

Deep dyslexia is a condition in which there are particular problems in reading unfamiliar words, and in which there are semantic reading errors (e.g., "ship" read as "boat"). It is more severe and mysterious than phonological dyslexia. Deep dyslexics resemble phonological dyslexics in finding it very hard to read unfamiliar words and non-words, suggesting they cannot use grapheme-phoneme conversion effectively. Deep dyslexics also make semantic errors, in which a word related in meaning to the printed word is read instead of the printed word. Deep dyslexics may mainly use Route 2, but damage within the semantic system itself or in the connections between the visual input lexicon and the semantic system makes this route error-prone.

Route 3 (lexicon only)

Route 3 resembles Route 2 in that the visual input lexicon and the speech output lexicon are involved in the reading process. However, the semantic system is bypassed in Route 3, so that printed words that are pronounced are not understood. Otherwise, the expectations about reading performance for users of Route 3 are the same as those for users of Route 2: familiar regular and irregular words should be pronounced correctly, whereas most unfamiliar words and non-words should not (see Figure 11.8).

Schwartz, Saffran, and Marin (1980) reported the case of WLP, a 62-year-old woman suffering from senile dementia. She showed a reasonable ability to read familiar words whether they were regular or irregular, but she often indicated that these words meant nothing to her. She was totally unable to relate the written names of animals to pictures of them, although she was fairly good at reading animal names aloud. These findings are consistent with the view that WLP was bypassing the semantic system when reading words. However, it is possible that processing occurred in WLP's semantic system, but she could not use such processing to aid performance on the tasks she was given.

Evaluation of the dual-route (or triple-route) model

This approach to reading can be regarded as the triple-route model. However, as the fundamental distinction is between reading based on a lexical or dictionary look-up procedure and reading based on a letter-to-sound procedure, the approach is often referred to as the dual-route model. It provides a reasonably good account of normal and brain-damaged reading.

Coltheart et al. (1993) put forward a dual-route computational model of reading based loosely on the processes and structures shown in Figure 11.8. Their focus was on non-lexical reading via Route 1, with the computational system learning the grapheme-phoneme rules embodied in an initial set of words. These rules were later applied to previously unseen letter strings.

Coltheart et al. (1993) trained their computational model on 2897 words. They then tested its performance when reading various non-words on which it had not been trained. It scored 90% correct, which is very close to the figure of 91.5% obtained by human participants.

Coltheart et al. (1993) did not implement the lexical route in reading in their dual-route connectionist model. However, they suggested that a modified version of McClelland and Rumelhart's (1981) interactive activation model (discussed earlier) might account for visual word recognition, with Dell's (1986) spreading activation model (see Chapter 13) being used to account for spoken word production.

According to the dual-route model, normal readers make use of both routes with familiar words, but the direct route will generally be much faster. It was also assumed originally that the main two routes to reading are independent of each other. Some evidence is inconsistent with the independence assumption. For example, non-words are supposed to be read by means of grapheme-phoneme correspondence rules without reference to the lexicon. Glushko (1979) tested this prediction by comparing naming times to two kinds of non-words: (1) those having irregular word neighbours (e.g., "have" is an irregular word neighbour of "mave", whereas "gave" and "save" are regular word neighbours); and (2) non-words having only regular word neighbours. Non-words of the former type were named more slowly, suggesting that the lexical route can affect the non-lexical route.

Additional evidence that non-words are not always pronounced according to grapheme-phoneme correspondence rules was reported by Kay and Marcel (1981) in a study mentioned earlier. They found that the pronunciation of non-words was biased by preceding irregular words (e.g., "caste" biasing the pronunciation of "raste").

Simple versions of the dual-route model predict that the naming of familiar words should not be influenced by the regularity of their spelling-to-sound correspondences. In fact, however, irregular words are generally named more slowly than regular ones (see Harley, 1995). Seidenberg, Waters, Barnes, and Tanenhaus (1984) found this regularity effect with low-frequency words but not with high-frequency ones. These findings can be explained by assuming that the direct route operates relatively slowly with low-frequency words, and so allows the indirect route to influence naming performance. As a result, phonological processing (which is crucial to the indirect route) plays an important role in the naming of low-frequency words and in the regularity effect.

Pugh et al. (1997) replicated the findings of Seidenberg et al. (1984) using lexical decision rather than naming as the task. They also obtained fMRI evidence from a different task that the inferior frontal gyrus is activated when people are engaged in phonological processing. Of particular interest, they found that those participants who showed the regularity effect had a relatively greater involvement of the right hemisphere in phonological processing than did those not showing the effect. These findings indicate that phonological processing is involved in the regularity effect, and they shed some light on the brain areas that are active during phonological processing.

Connectionist approaches

Within the dual-route approach, it is assumed that separate mechanisms are required to pronounce irregular words and non-words. This contrasts with the connectionist approach of Plaut et al. (1996). Their approach eschews [avoids] separate mechanisms for pronouncing nonwords and exception [irregular] words. Rather, all of the system's knowledge of spelling-sound correspondences is brought to bear in pronouncing all types of letter strings [words and non—words]. Conflicts among possible alternative pronunciations of a letter string are resolved, by cooperative and competitive interactions based on how the letter string relates to all known words and their pronunciations.

Thus, Plaut et al. (1996) assumed that pronunciation of words and non-words is based on a highly interactive system.

The two approaches can be contrasted by considering the distinction between regularity and consistency. Dual-route theorists divide words into two categories: regular, meaning their pronunciation can be generated by applying rules; and irregular, meaning their pronunciation is not rule-based. Regular words can generally be pronounced more rapidly. In contrast, Plaut et al. (1996) argued that words vary in consistency, meaning the extent to which their pronunciation agrees with those of similarly spelled words. Highly consistent words can generally be pronounced faster and more accurately than inconsistent words, because more of the available knowledge supports the correct pronunciation of such words. Word naming is generally predicted better by consistency than by regularity (e.g., Glushko, 1979).

Plaut et al. (1996) tried various simulations based on two crucial notions:

The architecture of the connectionist approach to word reading put forward by Plaut et al. (1996). Copyright © 1986 by the American Psychological Association. Reprinted with permission.

Adult Dyslexia

Adult Dyslexia

This is a comprehensive guide covering the basics of dyslexia to a wide range of diagnostic procedures and tips to help you manage with your symptoms. These tips and tricks have been used on people with dyslexia of every varying degree and with great success. People just like yourself that suffer with adult dyslexia now feel more comfortable and relaxed in social and work situations.

Get My Free Ebook


Post a comment