'Other strategies' for word identification

Moderators: Debbie Hepplewhite, maizie, Lesley Drake, Susan Godsland

volunteer
Posts: 755
Joined: Wed Nov 16, 2011 12:46 pm

Re: 'Other strategies' for word identification

Post by volunteer » Wed Jan 07, 2015 5:12 am

Please can you explain the reference to habituation to words in upper and lower case forms?

elsiep
Posts: 548
Joined: Tue Nov 02, 2010 10:23 pm

Re: 'Other strategies' for word identification

Post by elsiep » Wed Jan 07, 2015 7:15 am

Habituation is a drop in neural activity in response to a stimulus that’s been presented repeatedly. In this case, Dehaene calls it an ‘adaptation process’ because it happened after one presentation only.

What Dehaene & colleagues found was that in the first brain area that responds to the visual presentation of words, the adaptation process occurred only if the letters of a word were in exactly the same location - it didn’t matter what order the letters were in.

In the second area, the response was to identical groups of letters, even if they weren’t in exactly the same position - e.g. ‘anger’ and ‘range’ were treated as the same.

In the third area, the brain processed at the whole word level - it would read ‘anger’ as different to ‘range’. It also treated words as identical whether they were in upper or lower case - even if that meant the letters were a different shape in the two cases. Dehaene et al explored this further and found that this was true only for the third brain area in the left hemisphere. The same area in the right hemisphere treated e.g. ‘GET’ and ‘get’ as different words.

In other words, the brain processes words in exactly the same way as it processes any other visual stimulus, but has learned to make associations between upper and lower case letters.

Dehaene et al have mapped out a model for visual word recognition based on their observations http://readinginthebrain.pagesperso-ora ... tive35.jpg

elsiep

chew8
Posts: 4171
Joined: Sun Nov 02, 2003 6:26 pm

Re: 'Other strategies' for word identification

Post by chew8 » Wed Jan 07, 2015 9:11 am

elsiep wrote:Similarly with pictures. Work investigating the semantic priming effects of words and pictures found that the pictures had a stronger effect.

Researchers are not claiming that skilled readers actively choose to use these cues in order to guess words they don't recognise - these are sub-conscious processes used in word recognition that people are unaware of. We only know about them because they can be manipulated experimentally. They happen whether people want them to or not.
I'll stick just with pictures as there's a point which seems obvious. I accept that pictures might have a strong priming effect on me, but as virtually none of what I read contains any pictures, I know that I can identify printed words perfectly well without the help of pictures. I can't be using them, even sub-consciously, if they aren't there.

Jenny C.

elsiep
Posts: 548
Joined: Tue Nov 02, 2010 10:23 pm

Re: 'Other strategies' for word identification

Post by elsiep » Wed Jan 07, 2015 10:14 am

Of course you can't. Is anyone saying that skilled readers are primed by pictures that aren't there?

Now I'm puzzled.

elsiep

chew8
Posts: 4171
Joined: Sun Nov 02, 2003 6:26 pm

Re: 'Other strategies' for word identification

Post by chew8 » Wed Jan 07, 2015 11:54 am

I wasn't intending to cause puzzlement, elsiep.

What was going through my mind was that behaviour which can be produced and monitored when conditions are manipulated in certain ways needs to be weighed up against what happens in real life, so to speak. My own non-use pf picture cues was just a particularly clear example, but a similar point is made about context in the following:
Stanovich wrote:Specifically, the issue at hand is whether good readers have a greater tendency to use contextual redundancy to facilitate ongoing word recognition, not whether given virtually unlimited time good readers can make better predictions. The question is not whether the good readers actually have better predictive abilities, but whether they are actually prone to rely on such abilities to speed word recognition. (Progress in Understanding Reading, p.27)
Jenny C.

volunteer
Posts: 755
Joined: Wed Nov 16, 2011 12:46 pm

Re: 'Other strategies' for word identification

Post by volunteer » Wed Jan 07, 2015 12:03 pm

I wonder though if you might make use of picture cues - without realising you are doing so - when reading signs or posters in the distance that are a mix of pictures and words and that are just a bit too far away for you to read the words clearly without a picture cue e.g. if you saw a circus poster with a picture of a big top and the word circus on it you would probably be able to make out the word circus at a greater distance (from the number of letters and the approximate letter shapes that you could see) than if it didn't have a picture of a big top on it.

volunteer
Posts: 755
Joined: Wed Nov 16, 2011 12:46 pm

Re: 'Other strategies' for word identification

Post by volunteer » Wed Jan 07, 2015 12:12 pm

frances5 wrote:There is a lot of truth in the saying that "A picture is worth a thousand words". If you go on holiday to Greece then you are very dependent on the pictures on signs. The Greeks realise that most tourists can not read their signs so there are plenty of pictures as well as a lot of English/ Greek signs. Inspite of seeing plenty of Greek/ English signs I still cannot decode the Greek alphabet inspite of seeing several Greek signs with a picture and no English.
No, but if you learned the Greek alphabet and the letter sounds you could deliberately learn some Greek if you chose to by reading the words out loud (it's easy to pronounce once you know the sounds) and gathering the meaning from the pictures.

I once went to some Greek evening classes because the German class was full. It was impressive how quickly I could read Greek words. I can't now though. I forgot it all as quickly as I learned it.

I never reached the "skilled reader" level in Greek as I had a miniscule vocabulary and I didn't put in the hours. All gone. But I very much enjoyed doing it as it was so like learning to read English all over again- first learning the letters and sounds, then working out words slowly, then frequently recurring words starting to be recognisable pretty much instantly.

elsiep
Posts: 548
Joined: Tue Nov 02, 2010 10:23 pm

Re: 'Other strategies' for word identification

Post by elsiep » Wed Jan 07, 2015 12:56 pm

frances5 wrote:There is a lot of truth in the saying that "A picture is worth a thousand words". If you go on holiday to Greece then you are very dependent on the pictures on signs. The Greeks realise that most tourists can not read their signs so there are plenty of pictures as well as a lot of English/ Greek signs. Inspite of seeing plenty of Greek/ English signs I still cannot decode the Greek alphabet inspite of seeing several Greek signs with a picture and no English.
I wasn't suggesting that people use pictures instead of decoding, but as well as decoding. Greek isn't that difficult to read once you've learned the alphabet. Practice makes perfect.

elsiep

elsiep
Posts: 548
Joined: Tue Nov 02, 2010 10:23 pm

Re: 'Other strategies' for word identification

Post by elsiep » Wed Jan 07, 2015 1:04 pm

chew8 wrote:I wasn't intending to cause puzzlement, elsiep.

What was going through my mind was that behaviour which can be produced and monitored when conditions are manipulated in certain ways needs to be weighed up against what happens in real life, so to speak. My own non-use pf picture cues was just a particularly clear example,
There are two issues here;

1. Is a picture available?
2. Are you aware of using it in reading?

If there's no picture, then obviously you can't use it.

If there is a picture and you can see it, you will be using the information in it, even if you are not aware of doing so, because your brain processes a lot of visual information pre-consciously.
chew8 wrote:...but a similar point is made about context in the following:
Stanovich wrote:Specifically, the issue at hand is whether good readers have a greater tendency to use contextual redundancy to facilitate ongoing word recognition, not whether given virtually unlimited time good readers can make better predictions. The question is not whether the good readers actually have better predictive abilities, but whether they are actually prone to rely on such abilities to speed word recognition. (Progress in Understanding Reading, p.27)
Jenny C.
Not disputing that. Most of the experimental work involving contextual priming involves readers being shown words/pictures for very short periods of time indeed - much less than they'd have if they were reading in real life - and certainly not 'virtually unlimited time'.

elsiep

kenm
Posts: 1495
Joined: Sat Dec 17, 2005 5:19 pm
Location: Berkshire

Re: 'Other strategies' for word identification

Post by kenm » Wed Jan 07, 2015 2:02 pm

elsiep wrote:Skilled readers use multiple visual & auditory cues to decode words. Cues include letters that look similar, words that are a similar shape or begin with the same sound pattern, sounds that sound similar, context, illustrations etc.
My italics.
What are these sounds, how am I supposed to hear them, and if they are similar to something that is not the right word, how can they be cues?
"... the innovator has as enemies all those who have done well under the old regime, and only lukewarm allies among those who may do well under the new." Niccolo Macchiavelli, "The Prince", Chapter 6

elsiep
Posts: 548
Joined: Tue Nov 02, 2010 10:23 pm

Re: 'Other strategies' for word identification

Post by elsiep » Wed Jan 07, 2015 2:36 pm

kenm wrote:
elsiep wrote:Skilled readers use multiple visual & auditory cues to decode words. Cues include letters that look similar, words that are a similar shape or begin with the same sound pattern, sounds that sound similar, context, illustrations etc.
My italics.
What are these sounds, how am I supposed to hear them, and if they are similar to something that is not the right word, how can they be cues?
The sounds are the speech sounds represented by the letters; the spoken words represented by the written ones. If you've just read a word, all the neural pathways that process the letters, letter combinations, individual sounds and sound combinations associated with that word will be activated. This will include sounds that are the same but represented by different letters, such as 's' and soft 'c', 'j' and soft 'g'. A cue is a stimulus that prompts a particular kind of response, but not necessarily the correct one.

elsiep

User avatar
maizie
Administrator
Posts: 3121
Joined: Sun Mar 14, 2004 10:38 pm
Location: N.E England

Re: 'Other strategies' for word identification

Post by maizie » Thu Jan 08, 2015 12:57 pm

elsiep wrote:maizie wrote:
I have, believe me, read a great deal of the research literature. Apart from in the Whole Word/Whole Language literature, I have read nothing which has proven by research that skilled readers use word shape and pictures for word recognition. I have seen it asserted many times but not evidenced.


There's an entire chapter in Dehaene's book "Reading in the Brain" about visual processing in reading. It's clear that the first thing that skilled readers process is the visual features of the written word. The basic features are then chunked up into word units and then associated with the relevant speech sounds and meanings. If readers habituate to identical words in which the upper & lower case forms are visually the same, but don't habituate to identical words in which the upper & lower case forms are different, that suggests that they are using the visual features of the words in word recognition. I don't understand why anyone would think they didn't. How else would people be able to decode the orthography?
This is not supporting your contention that word shape is used for word recognition. Dehaene's account, and his diagram, clearly show that recognition starts with analysis of the letters within the word.

The 'habituation' is interesting. What Dehaene does not explore is the duration of the habituation effect. I would assume that the words he used with his subjects were already familiar to them (anger, range, get, hotel, etc.) yet on first exposure the word is analysed. Does this tell us the habituation effect is transitory? That if the same word was to be presented after a period of time the neuronal activity would be the same as in the first presentation x minutes/hours ago?

Also interested in the 'cultural training' which allows for recognition of different cases and fonts. Dehaene glosses over that by saying that it happens. But it would be interesting to know 'how' it happens.
elsiep wrote:Similarly with pictures. Work investigating the semantic priming effects of words and pictures found that the pictures had a stronger effect.
You've now sent me off exploring the literature on priming effects of pictures!

I can fully understand that a picture of a lion would prime for the word 'lion' but would it equally prime for the word 'nictitating'?
elsiep wrote:Researchers are not claiming that skilled readers actively choose to use these cues in order to guess words they don't recognise - these are sub-conscious processes used in word recognition that people are unaware of. We only know about them because they can be manipulated experimentally. They happen whether people want them to or not.
I'm not arguing with sub-conscious processes per se, just with some of the sub-conscious cues you are proposing.

kenm
Posts: 1495
Joined: Sat Dec 17, 2005 5:19 pm
Location: Berkshire

Re: 'Other strategies' for word identification

Post by kenm » Thu Jan 08, 2015 8:44 pm

In "Words and Rules", Pinker describes parallel processes to derive the plural form of a root. One is to form it from the most common rule - add "s" - the other looks up an internal table of words with exceptional forms of the plural - e.g. "en". If the look-up fails to deliver a result, the rule is used. I can easily believe that there are similar parallel processes in reading, with brain activity that is not needed most of the time but occasionally comes into play, e.g. to distinguish homographs, or possibly a taught activity (word outline? initial and rhyme?) that the reader has learnt to ignore but not yet to suppress.* One might describe the examination of a potentially useful characteristic of a word a potential cue; I would only call it a cue if it were used.

* Considering the amount of bad teaching that education departments seem to have been promulgating over the last 40 or so years, throughout much of the English-speaking world, where do experimenters get subjects who have been taught to identify words by SP and no other method? Others could be giving misleading responses.
"... the innovator has as enemies all those who have done well under the old regime, and only lukewarm allies among those who may do well under the new." Niccolo Macchiavelli, "The Prince", Chapter 6

elsiep
Posts: 548
Joined: Tue Nov 02, 2010 10:23 pm

Re: 'Other strategies' for word identification

Post by elsiep » Fri Jan 09, 2015 6:47 am

maizie wrote:
elsiep wrote:maizie wrote:
I have, believe me, read a great deal of the research literature. Apart from in the Whole Word/Whole Language literature, I have read nothing which has proven by research that skilled readers use word shape and pictures for word recognition. I have seen it asserted many times but not evidenced.

There's an entire chapter in Dehaene's book "Reading in the Brain" about visual processing in reading. It's clear that the first thing that skilled readers process is the visual features of the written word. The basic features are then chunked up into word units and then associated with the relevant speech sounds and meanings. If readers habituate to identical words in which the upper & lower case forms are visually the same, but don't habituate to identical words in which the upper & lower case forms are different, that suggests that they are using the visual features of the words in word recognition. I don't understand why anyone would think they didn't. How else would people be able to decode the orthography?
maizie wrote:This is not supporting your contention that word shape is used for word recognition. Dehaene's account, and his diagram, clearly show that recognition starts with analysis of the letters within the word.
First, I’m not sure ‘analysis’ is the best way of describing what the brain does because ‘analysis’ suggests a systematic, exhaustive and conscious process. What the brain does is detect, chunk and match visual features, pre-consciously. The visual features of a word are processed in parallel, but some features are processed faster than others, meaning that we rely on them more heavily for indicating similarities and differences.

There was a fair amount of research, up until the 1980s, suggesting that the overall shape of a word - its envelope, or ‘bouma’ - was involved in word identification. Since then, research has indicated that it’s actually the shape of the letters that acts as a cue. But since the shape of the word depends on the shape of the letters, it’s not entirely clear where letter shape ends and word shape begins. What’s also not clear is what features of the letters people are using to identify them.

The research suggests they use some features rather than others; in the example given in Fig. 10 here http://www.microsoft.com/typography/ctf ... ition.aspx similarities in the target letter shape resulted in more misses in a proof-reading task, but similarities in word shape resulted in fewer misses if the letter similarity condition was constant. I can’t access the Paap et al paper this example was taken from, so I don’t know if they looked at which features of letters their participants were relying on to discriminate. Work on how people process letter features suggests that some features are accessed faster than others and are therefore more important in letter discrimination - and in letter confusion e.g. Fiset et al http://tdlc.ucsd.edu/publications/2008- ... namics.pdf

Maybe I shouldn’t have referred to the processing of the visual features of words as ‘words that are a similar shape’ but rather ‘key visual features of the words’.
maizie wrote: The 'habituation' is interesting. What Dehaene does not explore is the duration of the habituation effect. I would assume that the words he used with his subjects were already familiar to them (anger, range, get, hotel, etc.) yet on first exposure the word is analysed. Does this tell us the habituation effect is transitory? That if the same word was to be presented after a period of time the neuronal activity would be the same as in the first presentation x minutes/hours ago?
Good question. Habituation can be short or long-term. Since the first presentation of a different familiar word didn’t show the adaptation process, it looks as if we’re talking about short-term habituation. Someone, somewhere has probably found out how long it lasts.
maizie wrote: Also interested in the 'cultural training' which allows for recognition of different cases and fonts. Dehaene glosses over that by saying that it happens. But it would be interesting to know 'how' it happens.
I think he calls it ‘cultural training’ because the exposure people get to different cases and fonts varies enormously. If by ‘how’ you mean the biological mechanisms involved, they are the same ones that Dehaene describes in his book. If you mean how do people encounter the different cases and fonts that would mean writing another book - but it wouldn’t make any difference to the biological mechanism.
maizie wrote:
elsiep wrote:Similarly with pictures. Work investigating the semantic priming effects of words and pictures found that the pictures had a stronger effect.
You've now sent me off exploring the literature on priming effects of pictures!

I can fully understand that a picture of a lion would prime for the word 'lion' but would it equally prime for the word 'nictitating'?
It would depend on the extent to which you associated the word ‘nictitating’ with images of lions. If you were researching vision in lions, it might have a priming effect , but not as much as it would on the word ‘lion’. If you’d never come across the word ‘nictitating’ before, a picture of a lion would have no effect. It’s all about frequency of exposure.

elsiep

elsiep
Posts: 548
Joined: Tue Nov 02, 2010 10:23 pm

Re: 'Other strategies' for word identification

Post by elsiep » Fri Jan 09, 2015 8:07 am

maizie wrote:
elsiep wrote:Researchers are not claiming that skilled readers actively choose to use these cues in order to guess words they don't recognise - these are sub-conscious processes used in word recognition that people are unaware of. We only know about them because they can be manipulated experimentally. They happen whether people want them to or not.
I'm not arguing with sub-conscious processes per se, just with some of the sub-conscious cues you are proposing.
Interesting discussion, but I just want to recap on why I commented on Dykstra's remarks.

As I understand it, the multi-method approach to teaching children to read has its origins in the research findings about how skilled readers read that emerged during the first half of the 20th century. As far as I can see, there are two problems with the reasoning involved, both involving assumptions.

One assumption is that novice readers would learn best by mimicking what skilled readers do. Novices often don’t learn best by doing this - it depends on the complexity of the task. Few people would attempt to teach a child to sew or cook by getting them to simply mimic what tailors or chefs do, because they are aware of the complexity of the tasks involved and know that the tasks need to be broken down into components and mastered one by one. But reading isn’t self-evidently complex like tailoring or cordon bleu. It’s deceptively simple; some children learn to read apparently effortlessly - all they need is the letter-sound correspondences or even just to be read to repeatedly. So although the assumption that all children could learn to read by mimicking what skilled readers did is a false assumption, it’s easy to see why people made it.

The second assumption is that the research findings about what skilled readers did were entirely valid and reliable. Not all of them were, of course, but there’s no way anyone could have known that at the time. For example, the word superiority effect was a focus in reading research for the best part of a century and we’re still trying to figure out the exact relationship between global and local feature processing in word recognition. So although the early ‘multi-methods’ proponents might have got it wrong, it’s easy to see why they did, and difficult, in the light of the reading research findings, to see how they would have come to any other conclusions.

The problem lies in the fact that people are still advocating multi-methods approaches to teaching reading even though the research shows this isn’t an effective way to teach reading.

What concerned me about Dykstra’s comments quoted in the original post on this thread was that he appears to dismiss the role that reading research had in the development of multi-methods approaches when he says;
It appears to me that the dominant approach to reading instruction does not rest not upon any established science of reading, but rather floats over a great, gaping hole with no science or data to support it.
He would be right to criticise the dominant approach to reading instruction for resting upon outdated findings in the science of reading, but it is completely inaccurate to say that it has no science or data to support it.

Having agreed with calls for a ‘scholarly discussion’, he says:
In order to support the practice of guessing, professors and various well known reading gurus have taught a series of well known falsehoods about reading which, if they were true, would make the use of guessing strategies both rational and necessary. These include the claim that only 50% of English is decodeable using phonics, that skilled readers don't look at all the letters, and skilled readers often use guessing strategies to identify words. There are many more.
It’s quite likely that Dykstra went on to be more specific, but in the extract quoted he is waving his hand in the general direction of reading research in a way that doesn’t, to me, appear to even approach a ‘scholarly discussion’. It’s possible that Professor X said “only 50% of English is decodeable using phonics”, when actually it’s 95%. Or that Guru Y said “skilled readers don't look at all the letters” when actually they process letters in parallel but not all features equally and they also skip words. Or that self-appointed reading expert Z said “skilled readers often use guessing strategies to identify words” when what they meant was that skilled readers don’t identify words systematically, exhaustively or consciously. But we don’t know because Dykstra doesn’t tell us who actually said what.

People making inaccurate generalisations about the skilled readers research is a major problem. But so is people alleging that the dominant approach to reading instruction ‘floats over a great, gaping hole with no science or data to support it’, when it largely originated in some very rigorous research. And so is calling for a ‘scholarly discussion’ then making generalisations about some quite complex issues and calling them ‘falsehoods’.



elsiep

Post Reply

Who is online

Users browsing this forum: No registered users and 9 guests