Monday, January 16, 2017

Zhou Youguang, inventor of Pinyin

The BBC has reported that Zhou Youguang, the man who brought literacy to millions in China after the communist victory in 1949, died on Saturday 14th January, a day after celebrating his 111th birthday.

After working as a Wall Street banker, Zhou decided, in 1949, to return to China to help with rebuilding the country. In 1955, he abandoned a professorship in economics and began studying writing systems. It took three years to create Pinyin – literally ‘spelled sounds’ (拼音) – a writing system for Chinese based on the Roman alphabet, and his achievement of developing a romanised Mandarin Chinese script brought literacy to millions. It has been estimated that prior to the victory of the communists in 1949, as many as 85% of the population were illiterate.

Many people believe that the traditional Chinese script is not based on sounds. In actual fact, it is! The correct description for the Chinese writing system is ‘morphosyllabic’ but, because of the large number of sinograms (Chinese characters) needed to represent the words in Chinese, it is very hard to learn - hence the invention of Pinyin.

One of the most recent estimates of the number of single graphs lists about 60,000 sinograms (Mair, V.H., ‘Modern Chinese Writing’ in Daniels, P.T. and Bright, W., (1996), The World’s Writing systems, p.200). Sixty thousand sinograms sounds like far too much for anyone to remember but, according to studies done in the late twentieth century, a mere (!)1000 sinograms cover 90% of all occurrences in texts and 2,400 sinograms cover 99%. This is about the upper limit of human memory: as Victor Mair explains, ‘It would appear that there is a natural upper limit to the number of unique forms that can be tolerated in a functioning script. For most individuals, this amount seems to lie within the upper range of approximately 2,000 to 2,500’. The problem is that even 2,400 sinograms is extremely hard to learn and takes years to achieve, which is why Mao’s government thought it worthwhile to invest in promoting Pinyin.

The difficulties of learning the traditional Chinese writing script give us a keen insight into why Whole Language approaches don’t work. Just as it is extremely hard to learn 2,400 sinograms, it is impossible to remember all the words in the English language as if they were logographs: a bit like trying to remember all the words in the Oxford dictionary. This is why teaching children to read using phonics is the only way to teach all children to read and spell effectively.

There are some fears that the conservatism of the traditional Chinese writing system will hold back the development of the country because of the pressures brought to bear on the use of sinograms by the advances in technology and information. The problem with such a system is that new sinograms need to be invented and added when new morphemes arise in the Chinese language or enter the language through borrowings. This makes the system open-ended and even harder to learn. Imagine learning a new character for every new invented word in a language! With Pinyin or English, for that matter, it's easy: all words are comprised of sounds and all we have to do is to assign already existing spellings to those sounds and, bingo, we have a new words.

After the killings in Tiananmen Square in 1989, as you can see from an interview he did with the US npr (National Public Radio) in 2011, Zhou became a fierce critic of the Chinese regime and a strong advocate of democracy. One of his last wishes was that he would live long enough to see democracy emerge in modern-day China. Amazingly, after his hundredth birthday he published no fewer than ten books, many of which are banned by the Chinese authorities, and he was an enthusiastic blogger right up to the time of his death.

Monday, January 02, 2017

Threshold concepts and the idea of sound to print

I’ve been thinking quite a bit recently about Glynis Cousin’s short introduction to ‘threshold concepts’, an idea developed by Erik Meyer and Ray Land and, although I have reservations about some of the things she argues, I’m finding her central theme quite helpful. She describes a threshold concept as one that is ‘central to the mastery of [one’s] subject’ and has ‘distinctive value ... for curriculum design’ and her contention is that ‘a focus on threshold concepts enables teachers to make refined decisions about what is fundamental to a grasp of the subject they are teaching’. 

However, knowing what the focus is can be problematical when, as she points out later in the introduction, teachers often have some difficulty in ‘retracing the journey back to their own days of “innocence”, when understandings of threshold concepts escaped them in the early stages of their own learning’.

Given the parlous state of initial teacher training in the domain of beginning reading instruction, it’s highly likely that – even though they may be teaching phonics on a daily basis –  many teachers have never been taught how the alphabet code they are teaching is structured. So, let’s start with that.

The threshold concept or first and overarching principle in the domain of teaching children (or older students) how to read and write is that the symbols we refer to as letters, combinations of letters or spellings are representations of sounds in our speech. The printed word is, as Diane McGuinness would have it, 'oral language written down'. This is what all writing systems were invented for and, as the psychologist Mark Seidenberg has argued in his book Language at the Speed of Sight, we need to go back to this fundamental principle and teach reading and writing as they were designed to be taught.

In countries, such as Spain, Italy and Germany, this idea is uncontroversial because it is starkly obvious: sounds and spellings are pretty much mutually implied. That is to say that there is, mostly, one way of spelling a sound and most sounds are spelled with single-letter spellings. This makes the writing system in these languages very easy to learn. English, on the other hand, is more complex – very much more complex. As I’ve stated a number of times on this blog, there are between 22 and 24 sounds in Spanish, depending on accent, and somewhere around the mid-thirties ways of spelling those sounds. In English, on the other hand, there are forty-four or forty-five sound for Scottish and Liverpudlian speakers of English. This fact in itself wouldn’t be a problem in teaching it; the problem is that there are multiple ways of spellings those sound.

So, going back to this idea of threshold concepts, the central idea that letters are symbols for sounds and that writing is speech written down subsumes a number of other concepts. The first of these is that we can spell sounds with a single letter, as in the word 'mat': three sounds, /m/ /a/ /t/, and three single letter spellings [m] [a] [t]. We can also spell sounds with two-letter spellings (digraphs), as in the word 'brush': /b/ /r/ /u/ /sh/, [b] [r] [u] [sh]; or 'steep': /s/ /t/ /ee/ /p/, [s] [t] [ee] [p]. In addition, we can spell sounds with three letters (trigraphs) in words such as 'light': /l/ /ie/ /t/, [l] [igh] [t]. And, in some words we can even spell sounds with four-letter spelling, such as 'weight': /w/ ae/ /t/, [w] [eigh] [t].

So, that’s quite a bit of complexity that isn’t evident in other alphabetic languages and many children will not work out this concept for themselves. In fact, although many children will be able to spell the sound /sh/ in 'brush' quite easily, they will probably not have realised explicitly that the sound /sh/ is spelled with two letters and that this idea generalises across the whole domain of reading and writing.

The next idea is the one-to-many principle: sounds can be spelled in different ways. This is the hard bit. It’s why it's virtually impossible to become a perfect speller in English – because if you have never come across a word containing a highly infrequent spelling of a sound, you probably wouldn’t be able to guess how to spell it. Moreover, for beginning readers it is hard work learning the common spellings for the sounds in English. You only have to look at the different ways of spelling the sound /ee/ to see that this is the case: she, meet, seat, happy, key, brief, taxi, amoeba, conceit. There’s a lot to learn – around 175 common spellings for the forty-four or so sounds!

The last concept (and this isn’t a hierarchy) is that many spellings represent more than one sound. So, the spelling [o] can be /o/ in 'hot', /oe/ in 'no', /oo/ in 'too' and /u/ in 'son'. As a symbol, the [o] could also be a ball, a moon, a sun, a pizza, etc, and if you draw a circle and ask even a young child of four what it can be, they’ll be able to give you all sorts of possibilities.

None of these concepts are difficult to understand but they are central to understanding how the sounds of the language and the writing system are connected. They should provide the framework for any phonics programme around which code knowledge (which spellings represent the sounds) and the skills of segmenting, blending and phoneme manipulation need to be taught.

The complexity of the above may appear to be counter-intuitive: last year, Daisy Christodoulou gave a very interesting talk in which she purposely cited the example of teaching reading. In her talk, she made the point that ‘literacy is the end point of education, but that this doesn’t mean that pupils will be practising this skill in its final format..’ To paraphrase Feltovich, Prietula and K. Anders Ericsson in their essay ‘Studies of Expertise from Psychological Perspectives’, expert readers look at the end point of literacy – automaticity, speed, understanding – and believe that this where novice readers begin. However, expertise in reading and writing is a long-term process. What Christodoulou argues is that 'as in many other skilled behaviours, the final product needs to broken down into its ‘component parts and ... lessons may look very different from the final skill a teacher is hoping to instil’.


What’s the message? Ground the teaching of literacy in the sounds of the language, which all L1 speakers learn naturally. Teach children the code knowledge and the skills they need to use that knowledge, and, especially, make explicit to them the elements of conceptual understanding which are transferable across the whole domain of reading and writing.