The phonological route is developing in three steps, which are not easy for a child of six years old to achieve. The first step is to be able for the child to SEGMENT the written word that he or she sees on the page into the correct graphemes. Sometimes a grapheme will be a single letter, so that's easy. Sometimes the grapheme will be several letters, (as) in 'chap', for example, 'ch', which is a single grapheme. So the segmentation is the first step. The second step is the transcoding, or recoding, which means that the child will have to assign the corresponding phoneme(s) to each of the grapheme(s) that he or she sees on the page. And the third step is the FUSION of the corresponding phoneme to the grapheme so that he or she can pronounce the word by assembling, concatenating, or fusing all the phonemes that he or she has derived from the graphemes that he or she sees on the page. So all the set-up of the mechanism of what identification and text comprehension consist in (is) establishing connections in the brain, and I insist again on the fact that these connections are not there beforehand. The child, when learning to read, is going to have to establish billions of new connections. Now, let's look at how it looks like for a fluent reader when reading. What is happening in the brain? So, as you can see from this graph(ic) borrowed from the book 'Reading in the Brain' by Stanislas Dehaene, you have first transmission of information from the printed word to the visual areas at the back of the brain in blue. And then you have a transmission of information to the 'letterbox' in red, which is going to analyze the words into letters and graphemes. You have transmission of information to the auditory area of the brain, where articulation and pronunciation are involved - that's in orange. And then you have, obviously, access to the meaning of the word - this is the green path. And do not forget that reading and spelling require attention, so you have in yellow the attentional processes top-down. Developing these new connections between the visual part of the brain, the visual area of the brain dealing with visual stimuli, and the area of the brain dealing with auditory stimuli will be to establish these connections between the two. So, as you can see from this graph(ic), the development of the phonological recoding root will consist in three steps. The first step is to segment the graphemes. Here, we have an example in Greek, because we are going to put ourself in the shoes of a six-year old learner who is reading a word for the first time. Suppose that we all know the grapheme- phoneme correspondences of Greek, but we don't have any vocabulary in that language. What are we going to do if we see a word like this one? We are going to segment that word into its constituent graphemes. Then we are going to transcode, or to recode, each of these graphemes into the corresponding phonemes, and then we're going to assemble, fusing these phonemes together to be able to pronounce the word, which is 'kalispera', meaning good evening in Greek. As you can see, on this graph(ic) showing the second path to reading, the direct access route: with repetition, the words are going to be stored in the phonological and orthographic lexicons of the child. And so this will allow the development of the second route to reading, the direct access route, which is much more competent, so to speak, much faster, and induces less mistakes compared to the phonological recoding group. This process is called LEXICALIZATION. So, the words are progressively lexicalized so that, after a certain amount of repetition, the child has memorized a complete and structured representation of words. Let's turn now to the fluent reader. The fluent reader is actually nearly always using the direct access path for reading. He or she will be using the phonological decoding path only for rare words, where they don't have phonological and orthographic representation(s) stored in long-term memory, or for unknown words. For most of the other words, experience through life and the repetition of the words will have been sufficient so that we can store ... the fluent reader, the expert reader can store most of the phonological and orthographic representations that he or she will encounter in writing. The use of the phonological recording path can be sensed - Suppose you are reading a novel translated from a foreign language with quite exotic names which are difficult to pronounce in the language in which you are reading. You will read the novel using the direct access route for most of it. But when you will come to that specific name, you will realize that again you will use the phonological recoding route, because you will segment the word into constitutive graphemes, bind these graphemes to the corresponding phonemes, and fuse the phonemes to able to pronounce the word correctly. In fluent readers, the level of automatization of reading has been elegantly shown by Dehaene and his collaborators in a series of treatise that he describes in his book, 'Consciousness in the Brain, Deciphering How the Brain Codes Our Thoughts'. What they have shown is that ... they use a paradigm called 'subliminal repetition priming'. In this paradigm, in this experiment, the participants sees ... well, they don't really 'see', this remain(s) unconscious. They see a word, for example ... ... for a few dozen (milli)seconds, and then this is masked or this is replaced by another word, which comes to consciousness. So, the participant remains unconscious of the first word they have seen. And the question was, do they process it or not? And Dehaene and collaborators show that indeed, if you precede the word, say, 'piano', with the same word, although the participant remains unconscious that they have seen the word for the first time, they are faster at identifying the target word, compared to the situation when you would present as a prime, another word which is unrelated to piano. They went even further, because they showed that the coding of written word was very abstract. If you show, as a prime, piano in lowercase and as target PIANO in uppercase, you will have the same subliminal priming effect as when you present both the prime and the target in lower case, or in upper case. So this means that the coding is very quick, and very abstract. Another set of experiment made by Marcell showed that the meaning of words can also be processed even though participants do not have access to the prime, to the signification of the prime, to the meaning of the prime. For example, he flashed the word 'red' and the word 'blue', for example, and then the participants were asked, 'did you see a word?' 'No, no, no, I didn't see anything.' 'Well, we showed something. We showed you something. Can you tell me whether it was a word or a random string of consonants?' And the participants were unable to answer that question. So, clearly the processing of the color remained unconscious, and then the participants were asked to select in an array of colors one color, and they were more likely to pick the color which they have seen as a prime, the blue or the green or the red, compared to other colors.
