Saturday, July 02, 2016

Castles in the air

This morning, I'm posting a reply I made to this post on Read Oxford. It’s by Anne Castles, a professor at Macquarie University in Australia and it’s yet another egregious
example of how professors responsible for teaching teachers how to teach literacy come unstuck: they are so rooted in graphemic phonics approaches, they can’t see that we need to teach from sound to print and not from print to sound. You’ll probably need to read the post before you read the reply.

In her blog, Anne Castles asks the question 'Are sight words unjustly slighted’? Here’s the answer, Anne: No! And here’s why:

I will say at the outset that there are so many ideas and assertions in this blog post that simply cannot be justified, it’s difficult to know where to begin. I won’t, therefore, try and deal with them all, only two or three.

Firstly, I would question the mantra of ‘phonics first and phonics fast’. It’s only half correct. ‘Phonics first and only’ should be the mantra; ‘fast’ is not possible because the English alphabet code is highly complex. It takes about three years for most children to learn the (roughly) 175 common spellings of the 44 or so sounds in English. It then takes a further four years of exposure and explicit teaching for the 50% of children who are likely to need this kind of explicit teaching if they are to become properly literate and to cope with secondary education (11-18 years). After that, we are further refining and building our understanding and knowledge of the code for the rest of our lives, especially when dealing with new ways of spelling sounds (e.g. the < bh > and < dh > spelling s that have come into the language through Indian English.

My second point of disagreement is the implied acceptance by Castles of Coltheart et al’s contention that reading is a dual-route process. The idea that there need to be separate processes is not supported by plenty of other research (see McGuinness, D., Beginning Reading Instruction for chapter and verse). But let’s take the examples Castles cites of ‘sail’ and ‘sale’ and claims ‘we would not be able to distinguish the two by sounding them out’. Why not? The first thing you need to be able to do is precisely to ‘sound them out’. As they are being decoded, the brain is searching the mental lexicon for meaning. Why wouldn't the two processes take place simultaneously as the word is being decoded? The two homonyms do sound the same – when you’ve decoded them – but the sound /ae/ in the words is spelled differently. If children are taught to segment and blend sounds in words to automaticity and they are taught that < ai > and the split spelling < a-e > represent the sound /ae/, they get to ‘sail’ or ‘sale’ without trouble.  At this point, context does the rest.

Chomsky once implied that the English spelling system was well suited to the language partly because of this feature: there are thousands of homonyms in English and spelling sounds in different ways is one means by which they are ‘distinguished’. Teaching children the different ways of spelling sounds is also generative; teaching individual words, one at a time, is very, very time consuming, it is not generative and many children can’t do it (paired associate learning!). So, the way we answer Castles’s dilemma is to teach all the common ways of spelling the sounds in English over the first three years in school.

Next, and central to Castles’s argument, is her assertion that ‘phonological decoding... doesn’t always produce the right pronunciation’. Ah, the rock on which so many professors founder! In support, she offers us the words ‘the’, ‘I’, ‘said’ and ‘come’. Again, I ask why? The < e > in ‘the’ is a schwa. If a child is reading, they say /th/ /e/ or /th/ /ee/ and then normalise it. If they are writing, they need to be taught how and when schwas are likely to be a problem and how to overcome the problem. The professor obviously has no idea. I taught my five-year-old grandson how to deal with schwas and he then proceeded to read lots of words on the London tube and to tell me where the schwas occurred! It then took me about five minutes to teach him how to spell them - using a spelling voice when he is writing. Now here’s the thing: if you also teach children that many spellings can represent different sounds in the language and teach them which sounds they represent in a coherent and structured system, the problem evaporates. If you don’t believe that children can understand this idea, draw a circle and ask any four or five-year-old what it can be. They’ll tell you that it can be a circle, a moon, a pizza, a ball... If the spelling < o > can be /o/ in ‘hot’, it can also be /oe/ in ‘go’, /oo/ in ‘to’, or /u/ in ‘come’. The difficulty comes in teaching exactly which sounds it can be. So, the spelling < I > can be /i/. It can also be /ie/ as in ‘tie’. In ‘come’, the spelling < o > (for historical reasons) is spelled < o > and consonant plus < e > is a common way of spelling many sounds at the ends of words (sleeve, some, borne, engine, gauche, route – I could go on.).

And I could go on! Teaching ‘sight words’ is very dangerous because most teachers are not taught to teach phonics properly. Learning to teach phonics properly enables a teacher to dispense with all the nonsense of ‘silent letters’, ‘magic e’, ‘sight words’, ‘hard sounds and soft sounds’ and so on. Our orientation should be to teach from sound to print and NOT print to sound, to teach the essential skills, to teach children to understand how the code works, and to teach all the common sound-spelling correspondences (for starters).

If you teach phonics as it should be taught, even though it’s a complex business, you’ll never need to teach all these so-called ‘sight words’. And herein lies the danger: when teachers don’t understand the code, everything quickly becomes an excuse to teach ’sight words’, as well as all the 'cute' little tricks that don't work, and teachers quickly fall back into teaching Whole Language/Look and Say.

The real problem lies with the professoriate, many of whom also have no idea about how to teach children to read and spell and, furthermore, don’t actually get into a classroom and do it. It’s in the classroom that ideas such as teaching children lots of ‘sight words’ are put to the sword.

7 comments:

Juliet Vanyai said...

Hi John,
Great post but I think in your intro you mean "No! Sight words have not been unjustly slighted" (They have been justly slighted!)
Very interesting and I need more time to think and read more about it all!

On another topic, I asked this question on Twitter but I think it was missed: How many students do you think enter St Georges Primary with speech/language difficulties that should make them more resistant to instruction only this doesn't seem to be the case? (Prevalence estimates at around 7%?)

Regards,
Juliet

Max Coltheart said...

Two questions, John.

1. You can understand and correctly reply to the printed question "Is SAIL about ships or shops?" You couldn't do this if you were responding just to the pronunciations of the words since SAIL and SALE have the same pronunciation. And context cannot help here since context does not favour one of these interpretations. So how could you do this if reading comprehension depends on phonology?

2. Your view is that all written English words obey letter-sound rules and so all can be correctly translated from print to speech by rule. Can you list the rule for OO that when applied to GOOD MOOD and GOON gets the pronunciation of the OO correct for all three, even when they are presented as single words i.e. no context?

Max

John said...

Hi Juliet,
You're right! I've corrected it and thanks for spotting that and for letting me know.
As to the question about St George's, I'll have to ask but it has a very poor catchment area, the poorest in south London, so I imagine that they have quite a few children with delayed language. However, having said that, I know that they do sterling work in the nursery.
As soon as I've had chance to ask, I'll write to you and let you know.
Best,
John

John said...

Hi Max,
The question, to me, seems nonsensical. You give the example of 'sail' and 'sale' and then situate one of the words in the context of a sentence. As they are homonyms, the only thing that would differentiate their meaning is context. If the words are to be read, as in the UK government's Phonics Screening Check, as single words and thus decontextualised, to be able to read the words, children would have to have been taught that [ ai ] and [ a-e ] are alternative spellings of the sound /ae/. What's the problem, because that's exactly what they do? I suppose a child could render 'sail' as 'sel', given that [ ai ] is also a very infrequent spelling in /e/ but then the government does allow alternative pronunciations where a spelling can represent more than one sound.
Accurate decoding depends on children being taught explicitly the alternatives spellings of sounds. That is what we do and everything we see happening in schools demonstrates that it works: not only do children at the end of the first three years of school end up being able to read almost anything, they can spell almost anything, too.
On your second question, you are simply mistaken: I do not believe that there is any point at all in teaching rules, for the reason that I've never met a teacher that knows all the rules, not to mention all the exceptions to the rules. What I do agree with is McGuinness's contention that the human brain is remarkably good at spotting patterns and, with lots of exposure (reading and writing), patterns there are in English spelling aplenty.
As for your question about < oo >, we do NOT teach from print to sound. We go from sounds, which all children learn naturally, to print, an invented system. So, we don't teach that [ oo ] can be this, that or the other first. That is putting the cart before the horse and it is typical of all graphemic phonics approaches, which teach the code backwards. We teach that the sound /oo/ can be spelled [ oo ], [ ew ], [ ue ], etc. Shortly afterwards, we'll teach that < oo > is a way of spelling the sound /oo/, as in [ oul ] ('could'), [ u ] ('bush'), and [ oo ] ('book', though this is a regional variation).
When reading, if pupils aren't sure, they try the alternatives. However, someone has to teach them the alternatives before they can do this. Here's an example: 'Last night I ate a tasty steak.' which the child reads first as 'steek', then as 'stek' (possible in Yorkshire!), and finally as 'steak', at which point the child declares for 'steak'. Why? because the human brain is decoding AND searching for meaning simultaneously and it does this so fast that, for fluent readers, it takes place under the level of conscious attention - until that is the reader is faced with, say, a place name with which the reader is unfamiliar. What happens then? The reader ask someone and remembers the pronunciation of the word. Whether, however, they remember the spelling (for writing) is a different matter and that depends partly on how less frequently encountered the particularly problematical spelling in the word is and how often they read and write the word.

Dick Schutz said...

I understand that you "do NOT teach from print to sound," John. What I don't understand is your example. When a child encounters the word "steak" s/he doesn't hear any sounds. What s/he sees are letters. How does the child know to "first read as /steek/? Does Sounds~Right teach this as the "first choice" Correspondence for "ea," with the second choice /stek/?

The crux of the Code is in the Correspondence between the sounds/symbols, (graphemes/phonemes), rather than either separately. Or am I wrong about that? I understand that the Alphabetic Code applies to both reading and spelling, and that the Code was invented to convert spoken language into a graphic/written form, and the history that made for "spelling alternatives," but that history also made for "reading alternatives," didn't it?

While I'm asking-one more question. Do you subscribe to the doctrine that "spelling is the reverse of reading"? I concur with your last three sentences, which seem to conclude otherwise?

Just curious about these questions.

John said...

Hi Dick,
Thanks for the questions - they are always useful in providing opportunities to explain further what my take is on a complex business.
You ask: 'How does the child know to "first read as /steek/?' The child knows because we teach children that sounds can be spelled in different ways. We start with the sound [ ae ] and teach the spelling [ ai ], [ ay ], [ ea ], and [ a-e ]. So, the child(ren) know and work with this idea in the context of words such as 'rain', 'play', 'break', and 'gate'. And, yes, I know that [ ea ] is a spelling alternative for /ae/ in only three words.
We then go on to teach a limited number of spelling alternatives for the sound /ee/, one of which includes the spelling [ ea ]. Once they've had time to absorb this, we teach explicitly that [ ea ] can be /ae/ and it can also be /ee/ and we practise this in the context of real words and in real sentences. If a pupil is uncertain, they try both and see which one makes sense (meaning).
Of course, the pupil will also have to be skilful in taking sounds out of words and dropping in alternatives to 'try them out'. To that end, we teach children the skill of manipulating phonemes right from the start of the Sounds-Write programme.
Later, when we get to different spellings of /e/, expressed as [ e ] and [ ea], the pupils will have another alternative to try. This isn't a big deal. After all, we (fluent adults!) have to do exactly the same thing when, for example, we visit new places and are not sure how to pronounce a place name. My favourite is the time I pronounced the place 'Lowton' with an /ow/, as in 'cow', instead of an /oe/ as in 'grow'. So, in short, the child knows because we've taught the possibilities and the learner tries them until they make sense.
On the second point, I completely agree: we share the same understanding: the code does apply to both reading and spelling. The history of the language did, as you say, make for spelling alternatives, which is covered by the concept we teach children explicitly: that sounds can be spelled in more than one way. Again as you say, the history also made for 'reading alternatives': a spelling can represent different sounds.
Finally, I do think that reading and spelling are two sides of the same coin. However, from a psychological point of view, they are not the same. Reading is to do with recognition memory - the cue is there in front of us. Spelling, on the other hand, isn't in front us: we have to dredge it up from our memories and recall memory is a deeper kind of memory. This has always been why David Philpot and I went for testing children on their spelling - because we felt that it revealed much more than reading tests, which often masked lack of knowledge of the code in children with very good visual memories for whole words.
In the meantime, I hope you are well and wish you the best,
John

Dick Schutz said...

Thanks much, John. That clears the matters up for me. The Sounds-Right instructional architecture, now that you have explained it, indeed differs from the architecture of the programme class that starts instruction with short-vowel Correspondences rather than long-vowel Correspondences as S-R does. There is a rationale for both architectures, and my reading of the PSC data is that both are reliable.

There are two other reliable reading instructional architectures. One is the tightly-scripted programmes that Zig Engelmann popularized as "Direct Instruction." These programmes concretize a scenario that a "gifted instructor" has constructed--with or without influence of the Alphabetic Code. Engelmann's is a an example of "without." Tom and Hilsey Burkard's programme is an example of "with."

The fourth architecture is one that is "only and always 'decodable.'" That is, it teaches no Correspondences out of the context of text. Children are taught how to handle the Code by reading a "real book" from the get-go. The book entails only five Correspondences, but it taps the full complexity reading--responding to text with understanding equal to that were the text read to the child. Additional Correspondences and other linguistic considerations are then spoon-fed in subsequent texts. With this architecture the child is a "fluent reader" from the beginning and the boundaries of the "fluency" transparent to the child, the instructor, and everyone else. Piper Book's BRI/ARI is an example of this architecture.

Each of these architectures has been shown to be reliable. Whether they are equally reliable remains to be investigated. The programmes also differ in terms of personnel training requirements and cost; important considerations that receive little attention.

On the rest of your explanation, we're exactly on the same page--with only a few font differences that aren't consequential.