If utterances are sketched out in the way that was described earlier, there will be a stage during the production of a word when the physical form of a word (its sounds) will not yet have been chosen, but the concept associated with that word will have been. Only later will information about that word's physical form become available. This means that there will be an early stage during the preparation of the target's utterance when a priming word related in meaning could influence the time to produce the target, but a priming word with some of the same phonemes would not. Only at a later stage would the sound-related prime be effective. This is exactly what was found in a study conducted by Herbert Schriefers, Antje Meyer, and Willem Levelt. They asked people to name pictures of objects shown to them as quickly as possible, while playing over headphones a word that was related either in meaning or in sound to each object's name. And when they varied the delay between playing that word, and showing the picture, they found exactly the predicted pattern-to be effective, meaning-related words had to be played earlier (relative to when the picture appeared) than sound-related words.
For priming-by-sound to occur, as in this last study, the overlapping sounds must be presented at just the right time. Too soon and the activation they cause dies down completely before the target word is sketched out. Too late and the target word has been sketched out already and the relevant circuits activated. If the initial sounds of the target word are added to the sketch before the final sounds (as opposed to at the same time, which could in principle happen), something which will prime those initial sounds must be presented sooner than something that will prime those final sounds. And this was also found; primes sharing the initial sounds of the target had to be presented earlier than primes sharing the later sounds of the target.
The use of priming techniques to study the production process is still, at the time of writing, in its infancy. An even newer procedure for studying these same issues has been developed by Miranda van Turennout, also in The Netherlands, who has found that electrical activity within the brain can vary depending on when meaning-related or sound-related information is activated before naming a picture. Taken together, these new techniques mean we now no longer need to rely exclusively on detailed analyses of just the output of the language production process. We may not know much about what the input is to that process, but at least we can now tamper with it, and observe the effects of the tampering on the output. The study of language production is now an experimental science.
To produce language, and to understand it, are the most natural of adult human activities. But words are transient things, lasting little more than the moment it takes to speak them, or the moment it takes to comprehend them. At any one moment, the majority of words exist not in spoken form, but in written form. Such words usually exist for as long as the paper exists on which they are written, and yet reading those words is a surprisingly unnatural human activity. It is the activity to which we turn next.
The written word
Many of the world's languages exist not just in spoken form but in written form too. The advent of the written word must surely rank, together with fire and the wheel, as one of mankind's greatest achievements. John Maynard Smith, an evolutionist, has described the invention of writing as one of the major evolutionary transitions to have taken place since the creation of life itself-the only other time that evolution came up with a system for storing and transmitting information was when it came up with the genetic code. The practical consequences of this development are innumerable. Science and technology would hardly have progressed beyond the Dark Ages were it not for the written word-science simply relies on too much information for it to be passed down through the generations by word of mouth alone. Even within a single generation, or for a single individual, there is too much to commit to memory. Where would Einstein have been without his notes? But paradoxically, the earliest writings were not devised so as to enable scientific or cultural progress. They were neither religious manuscripts nor philosophical treatises. Those earliest writings were devised instead for the purposes of trade-they were, in effect, ledgers.
The precursor of the modern book-pages of goat or sheep skin sewn together and bound between wooden boards-dates from about 200 BC, and was developed in Ancient Greece, although papyrus scrolls had existed in Egypt since around 2500 BC. But some of the earliest writing systems were hardly writing systems at all-they were based on small clay tokens whose different shapes and markings represented a variety of different things (animals, body parts, substances, and so on). These tokens were used between around 9000 BC and 3000 BC. Some of the same shapes and patterns can be found in written scripts which developed in southern Mesopotamia (now Iraq). These scripts (called cuneiform) were originally derived from a hieroglyphic writing system, although not from the better known Egyptian variety. Egyptian hieroglyphics date from about 3500 BC, and were amongst the earliest and longest lasting of the hieroglyphic systems. Hieroglyphics were not particularly suited to mass production, and so more convenient versions were developed. The first of these were the Mesopotamian cuneiform scripts and the Egyptian demotic script.
The evolution of writing systems
The earliest writers suffered from a dearth of readers. There was no alternative in pre-hieroglyphic times but to use signs that fairly directly represented whatever it was that was being conveyed. As these signs became more widely used, they gradually changed and became more stylized, until they hardly resembled any more the things that they stood for. This process can be seen in cave paintings from as early as 20 000 BC. Of course, as writers became more expressive and wanted to refer to things which could not be pictured directly, it was inevitable that symbols would be introduced which bore less and less resemblance to the things being referred to ('horse' is easy, `summer' less so, and `fresh' a real challenge). But such writing systems are necessarily limited, because the more things you want to refer to, the more symbols you need to devise, and the more symbols you, as either the writer or the reader, need to learn. Fortunately, two developments occurred which relieved readers and writers alike from an overdose of symbols.
First, symbols started to be used to refer not to things out there in the world, but to the spoken words that were themselves used to refer to those things. This meant, for instance, that two words that sounded the same could have the same symbol. Many words sound alike, but are not identical (e.g. `peach' and `speech'), and the second development was that phonetic symbols were introduced to represent individual sounds (a symbol for the /s/ sound could be tacked onto the symbol for `peach'). The hieroglyphic and cuneiform scripts used combinations of symbols in exactly this way.
It is not hard to see how someone could have figured out that the next step should be the development of a fully sound-based writing system. The first such systems were in fact syllabaries-one symbol for each syllable. Japanese is an example of a language that uses a syllabary. In fact, it uses two-hiragana and katakana. Both these syllabaries, dating from around the ninth century, can represent all the syllables of the language, although katakana is more often used to represent words imported from other languages. Hiragana is the first scnpt taught in school.
Modern alphabets-one symbol per phoneme in the language-owe their origins to the semitic alphabet developed in Syria and Palestine between 1500 Bc and 2000 BC. It was itself based, probably, on a syllabary that had been developed from the Egyptian hieroglyphics. This Semitic alphabet contained only consonants (and to this day, vowels are only optionally shown in Hebrew and Arabic scripts). It was not until around 1000 BC that the Ancient Greeks added the vowels to create a full alphabet. It was from the Greek alphabet that the Roman alphabet (used in English and other European languages) was derived.
Alphabetic systems are a great improvement on the earlier ones, because no more symbols are required than there are individual phonemes in the spoken language. Often, fewer symbols are required. In English, for instance, there are about 44 phonemes, but just 26 letters. Combina
tions of letters are used to represent some of the individual phonemes, for example in `chatter', `shout', `thunder', `sing'. The number of symbols can increase if different pitch pattern:; need to be marked-in the pinyin version of Mandarin Chinese (which transcribes spoken Chinese using the Roman alphabet), `boli' means `glass', whilst `boll' means `to peel off (e.g. skin)' and `boll' means `small profits'-but even so only a relatively small number of symbols are required. Syllabaries are a nice idea in principle, but in a language like English, with around 12 000 syllables, many of which are rarely used, a syllabary would be rather unwieldy. By comparison, Japanese uses fewer than 120 syllables, and like other languages with small numbers of syllables, it relies on having lots of syllables in each word. In English, by contrast, many words are just single syllables.
The development of the alphabet was a landmark in writing designif you knew the mapping between the written symbols and the spoken sounds of your language, you could in principle read any printed word, even one that you had never seen before. Simple really-although the fact that the written version of the most spoken language in the world, Mandarin Chinese, does not use an alphabetic script is somewhat paradoxical. Chinese consists of logograms-a single character represents, in effect, a whole word. English has logograms too: $, k, @, and % are examples, as are many road signs, and the numerals 1, 2, 3, and so on. English can also combine some of its logograms with phonetic symbols, something which is common in Chinese and hieroglyphics: `1st', `2nd', and `3rd', for instance. But what have ancient history and modern Chinese got to do with how we read?
The characters you see on the page in front of you are just arbitrary shapes based on just one of the different kinds of writing system that have developed through the ages. Chinese has very much more in common with hieroglyphics than it does with the Roman alphabet. The Japanese syllabic scripts are different again, and have more in common with the written form of Cherokee (another syllabary) than they do with either English or Chinese. But does the existence of these different writing systems mean that their respective readers do profoundly different things when reading? Japanese writing commonly uses different writing systems in the space of a single line (kanji symbols, borrowed from the Chinese, and one or other or both of the syllabic scripts)-but does this mean that Japanese readers switch between one kind of reading and another in the space of a single line? The fact that some languages read top-to-bottom, some read right-to-left, and others left-to-right (although none, apparently, read bottom-to-top) seems like an almost trivial difference by comparison with the different kinds of script that are possible-alphabetic, syllabic, or logographic. What are the implications of these differences for the ways in which we read? Surprisingly, there are very few. And to see why, we need to look at not just how we, as adults, read, but also how children learn to read. For convenience, we shall start with how children learn just one of the many possible scripts in use today-this one.
Learning to read
How children learn to read, and how best to teach them to read, has been the subject of much controversy over the last 20 years or so. How children read, and what factors contribute to their reading skill, has often been overlooked. There is no doubt that the child's home environment plays an important part; even simplistic measures such as how many books there are in the household have been found to correlate with how well the child reads in later years. Of course, correlations do not imply causes-the sheer number of books in a household does not itself cause good reading. It probably means that parents who are keen readers are more likely to read with their children, and that this gives the children the advantage. Recently, it has even been claimed that children under one year of age who are given books to play with are likely to read better, later on, than children who have not been given books to play with at this age. But home environment is just one factor. A more complete picture of what influences the child's development of reading skills (which includes not just how well they read but how they read) would of course have to include the way in which the child is taught to read, not to mention the time spent learning to read, and the time spent practising what has been learned.
As adults, we forget how much we needed to be taught about reading. Some of it is so basic that we probably cannot imagine ever not knowing it. For instance, one consequence of an alphabetic system is that, generally, the more sounds in a spoken word, the more characters in the written version of that word. But this is something we need to learn. If you show pre-literate children the words TWo and TOOTHBRUSH, and ask them to point to which one corresponds to the spoken word `two', or to the spoken word `toothbrush', they can do no better than guess-they do not naturally assume that the longer sounding word must go with the longer looking word.' _link_ Even children who are just starting to read, as well as adults who have never learned to read, will often make mistakes. So the printed word starts off as a meaningless jumble.
The first task for the young child is to realize that written words correspond to spoken words. But the child also needs to know the nature of that correspondence-whether the characters it sees correspond to phonemes, syllables, whole words, or some combination of these. To a certain extent, the way in which the child is explicitly taught to read determines the way in which the child approaches the correspondence problem. The phonics method of teaching introduces the child to the notion of sounding out. When the child encounters a new word, he or she can sound it out and in this way retrieve the spoken version. This method explicitly teaches the correspondence between letters and sounds, something which, in principle, should be ideal for an alphabetic system. In practice, though, it may not be. English is a very irregular language-the same sounds can be written using different letters (e.g. `see'/`sea'/`ceiling'), and the same letters can stand for different sounds (e.g. 'gist'/'gift'). The English-speaking child has somehow to be able to cope with this irregularity. Speakers of Finnish or Arabic, for example, do not; these languages are almost totally regular.
An alternative to phonics is the whole-word (or look-and-say) method. Here, the aim is to get children to recognize written words without having to first break them down into letters and then figure out the sounds of those letters. Often, this method is taught by labelling objects around the classroom, or using pictures of objects with their labels underneath. There are two advantages to this approach: it avoids the problem, in English at least, of the vast irregularity of the written/spoken correspondence, and it also teaches children to concentrate not on how a word sounds, but on what it means. But is one method better than the other?
It is this last question that has caused so much argument and controversy. The arguments are complex. On the one hand, alphabetic writing systems seem designed for a phonics-style approach. On the other hand, and as we shall see below, there is evidence suggesting that proficient adult readers do not break down each written word into the individual sounds that the different letters represent--instead they appear to access the meaning directly from the visual form in much the same way as the whole-word approach to teaching encourages. Moreover, they appear to do this even if they had been taught by the phonics method. But just because adults do not necessarily take advantage of letter-to-sound correspondences when reading, does not mean that children should not need to acquire knowledge of those correspondences when learning to read. The evidence suggests, in fact, that the phonics method leads to better reading by the time the child is nine or ten. We shall return to this shortly.
But which way children are taught to read fails to address the psycholinguistic issue of how these children actually read, and how they extract meaning from the characters on the page. Do they extract meaning directly? Or do they first translate the letters into sounds, and get at the meaning that way? In fact, it seems that they do a bit of both, but at different stages in the development of their reading skill, and only partly determined by the way they have been taught.
To begin with, children develop what is called a sight vocabulary. They associate, for instance, the
sequence of letters D-O-G with the word `dog' directly, and not by having associated the letter o with the phoneme /d/, o with /o/, and G with /g/. So at this stage, the written form of `dog' could just as well be CAT. If you cannot imagine what it would be like not to use letter-to-sound correspondences when recognizing written words, it is a little like recognizing $5 as `five dollars', or 204 as `two hundred and four', or `I V New York' as `I love New York', or, finally, YACHT as `yacht'. This word is so irregular that only the letter-to-sound correspondences of the first and last letters are of much use.
There is some evidence that children taught by the phonics method never develop a pure sight vocabulary, but do use some information about letter-to-sound correspondences when reading certain words. But the information they use is very limited-in these earliest stages of reading development almost all children focus their attention on just parts of the written word (mainly the first letter, and to a certain extent, the last letter also). Children at this stage thus ignore much of the letter information contained within the written word. If they do have any letter-to-sound awareness, they might use this awareness as a clue regarding which word is intended. But without that awareness, the shape alone (of certain letters) would be the main clue.
The Ascent of Babel: An Exploration of Language, Mind, and Understanding Page 20