Tesing the ability to decode
Tesing the ability to decode
I am developing a test for a beginning reader's ability to decode words by showing a computer screen with five possible choices from three words of different spellings, each one with only one pronunciation. The choices are:
All words sound the same;
All words sound different;
Two words sound the same, choose the one that is different.
If I can make all this work reliably, it will have the merit of testing decoding with no reference to comprehension.
I can generate the test triplets from a list of about 400 homophone pairs and triplets that I have found, being careful to eliminate those which are not homophones in some dialects (e.g. "witch", "which", different in Scotland and N. England). What I want to do next is put them in some order of difficulty, with those involving only early correspondences first and unusual correspondences at the end, so as to present test sets of greater difficulty to more experienced readers. My present strategy is to make use of the 12 sets of correspondeces in the Phonics International 12 units. Can anyone think of a better way? I have considered using a word frequency chart, but too many of the homophones don't appear in the 7700 commonest words of English, and a good decoder can tell the pronunciation of a substantial proportion of English words without knowing what they mean.
All words sound the same;
All words sound different;
Two words sound the same, choose the one that is different.
If I can make all this work reliably, it will have the merit of testing decoding with no reference to comprehension.
I can generate the test triplets from a list of about 400 homophone pairs and triplets that I have found, being careful to eliminate those which are not homophones in some dialects (e.g. "witch", "which", different in Scotland and N. England). What I want to do next is put them in some order of difficulty, with those involving only early correspondences first and unusual correspondences at the end, so as to present test sets of greater difficulty to more experienced readers. My present strategy is to make use of the 12 sets of correspondeces in the Phonics International 12 units. Can anyone think of a better way? I have considered using a word frequency chart, but too many of the homophones don't appear in the 7700 commonest words of English, and a good decoder can tell the pronunciation of a substantial proportion of English words without knowing what they mean.
"... 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
It seems to me, Ken, that you may have to be very careful to avoid problems arising from the fact that an order of difficulty based on one programme may give a slightly inaccurate picture of the decoding ability of children taught by other programmes. In the first year or so of school, what is decodable for children taught by one programme may not be decodable for children taught by another programme.kenm wrote:What I want to do next is put them in some order of difficulty, with those involving only early correspondences first and unusual correspondences at the end, so as to present test sets of greater difficulty to more experienced readers. My present strategy is to make use of the 12 sets of correspondeces in the Phonics International 12 units
I'll be interested in progress reports on this.
Jenny C.
Yes, I had that factor in mind. I would like to get implemention sequences for correspondences* of several different programmes, to see how much they differ. I am also beginning to think about the information to be recorded and delivered to teachers and parents. Ideally, I would have a level for each test line/programme combination, and a computer program that would construct new test files with the programme and the level as input. However, I don't see that including correspondences not yet taught compromises the comparison of children taught by the same programme. Comparison of programmes would only be valid on the basis of test files including all taught levels, though perhaps not the rarest correspondences, which are likely to be known only by children from intellectual families.chew8 wrote:It seems to me, Ken, that you may have to be very careful to avoid problems arising from the fact that an order of difficulty based on one programme may give a slightly inaccurate picture of the decoding ability of children taught by other programmes. In the first year or so of school, what is decodable for children taught by one programme may not be decodable for children taught by another programme.
* I also need to know when constructs such as "VC" -> "VCCed", which are not strictly G/P correspondences, are taught.
"... 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
It seems to me that these need to be taught more for encoding than for decoding, though children do need to know, for decoding purposes, that final 'ed' is likely to be a past tense ending, that it has three possible pronunciations (/d/, /t/ and /id/) and what 'C' and 'CC' before it signal about the preceding vowel sound - e.g. 'hoped'/'hopped', 'taped'/'tapped', 'dined'/'dinned'. Is this what you had in mind, Ken?kenm wrote:I also need to know when constructs such as "VC" -> "VCCed", which are not strictly G/P correspondences, are taught.
Jenny C.
I hadn't thought of it in those terms, but a possible triplet is "guessed guest gest", and knowing its right answer requires this knowledge. Considering similar ones, of which there are several in my big homophone list, led me to consider the position of the "VCCed" construction in the various programme sequences.chew8 wrote:It seems to me that these need to be taught more for encoding than for decoding, though children do need to know, for decoding purposes, that final 'ed' is likely to be a past tense ending, that it has three possible pronunciations (/d/, /t/ and /id/) and what 'C' and 'CC' before it signal about the preceding vowel sound - e.g. 'hoped'/'hopped', 'taped'/'tapped', 'dined'/'dinned'. Is this what you had in mind, Ken?
"... 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
If you are really talking about the decoding ability of beginning readers, as you said in the first message in this thread, 'guessed/guest/gest' is probably too difficult. OK, 'g' before 'e' is usually soft, but there are such words as 'get'. Another point is that even proficient adult readers are often swayed by letter-knowledge in making judgements about sounds - many adults would probably say that 'guessed' ends with a /d/ sound, and I would say that this doesn't matter as long as they can read and spell the word correctly.
I wonder whether the type pf test you are envisaging will actually achieve anything that isn't already achieved by some existing tests which offer children choices between alternatives differing by just one phoneme and which could easily be computerised. I haven't really thought this through, though.
Jenny C.
Jenny C.
I wonder whether the type pf test you are envisaging will actually achieve anything that isn't already achieved by some existing tests which offer children choices between alternatives differing by just one phoneme and which could easily be computerised. I haven't really thought this through, though.
Jenny C.
Jenny C.
1 (Shakespeare) time fixed for a stay in a place.maizie wrote:This is clearly a chance for me to extend my vocabulary. What is 'gest', please?
2 (obs) An exploit; a tale of adventure, a romance.
3 (obs) bearing; gesture.
"... 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
The actual test files will need a lot of weeding for this sort of problem. Some dialects make distinctions that others do not, e.g. "wh/"w", "chew8 wrote:If you are really talking about the decoding ability of beginning readers, as you said in the first message in this thread, 'guessed/guest/gest' is probably too difficult. OK, 'g' before 'e' is usually soft, but there are such words as 'get'. Another point is that even proficient adult readers are often swayed by letter-knowledge in making judgements about sounds - many adults would probably say that 'guessed' ends with a /d/ sound, and I would say that this doesn't matter as long as they can read and spell the word correctly.
I was encouraged to continue working on this by reading about the failure of the "Lost for Words" project at Monteagle School to find such a test, even though they knew that it was what they needed. Have you a reference to a test such as you describe?I wonder whether the type pf test you are envisaging will actually achieve anything that isn't already achieved by some existing tests which offer children choices between alternatives differing by just one phoneme and which could easily be computerised. I haven't really thought this through, though.
"... 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
I don't have copies of a lot of tests, but have found the following. All three start with questions where the children have to look at a picture and choose the right answer from four offered. In most cases the words differ by more than one phoneme.
SPAR: picture of a bag - the words offered are 'log', 'big', 'bag' and 'dip'; picture of a fish - words are 'fit', 'fish', 'fill' and 'fist'.
Young's Group Reading Test: picture of a goat - words are ' 'get', 'goat', 'good', 'garden', 'got'.
NFER Primary Reading Test: picture of a boat - words are 'cake', 'boat', 'horse', 'candle', 'boot'; picture of a pear - words are 'lamp', 'boat', 'pear', 'bear', 'shoe'; picture of a 'hand' - words are 'shop', 'chin', 'hand', 'mouse', 'band'.
There are also cloze-type sentence questions - e.g. Primary Reading Test - 'He pulled up a ------ and sat down (chair, chap, chain, card, claim); Joan put on her --------- uniform (grown, drawn, brown, broad, brow); Suffolk Reading Scale - Chess is a game that is played on a ------------ (beard, board, bread, daub, broad).
I'm not suggesting that all these would be suitable for relative beginners or that they are better than what you have in mind, Ken - they just illustrate the type of thing that's in existing tests. Something like this may be needed at the point at which children have been taught few if any alternatives for sounds, because few if any homophones are possible at that stage.
Jenny C.
SPAR: picture of a bag - the words offered are 'log', 'big', 'bag' and 'dip'; picture of a fish - words are 'fit', 'fish', 'fill' and 'fist'.
Young's Group Reading Test: picture of a goat - words are ' 'get', 'goat', 'good', 'garden', 'got'.
NFER Primary Reading Test: picture of a boat - words are 'cake', 'boat', 'horse', 'candle', 'boot'; picture of a pear - words are 'lamp', 'boat', 'pear', 'bear', 'shoe'; picture of a 'hand' - words are 'shop', 'chin', 'hand', 'mouse', 'band'.
There are also cloze-type sentence questions - e.g. Primary Reading Test - 'He pulled up a ------ and sat down (chair, chap, chain, card, claim); Joan put on her --------- uniform (grown, drawn, brown, broad, brow); Suffolk Reading Scale - Chess is a game that is played on a ------------ (beard, board, bread, daub, broad).
I'm not suggesting that all these would be suitable for relative beginners or that they are better than what you have in mind, Ken - they just illustrate the type of thing that's in existing tests. Something like this may be needed at the point at which children have been taught few if any alternatives for sounds, because few if any homophones are possible at that stage.
Jenny C.
Yes, the Lost for Words exercise used Suffolk, IIRC, and it was computer administered, but they acknowledged that it was not ideal because of the comprehension element, from which all these suffer. I have ideas for mine having spoken instructions delivered by the computer, though that will be much more work than I have done so far.chew8 wrote:I don't have copies of a lot of tests, but have found the following. All three start with questions where the children have to look at a picture and choose the right answer from four offered.[...]
I can think of only one homophone that I can make from the correspondences in Debbies Unit 1 - "sac" "sack" - but there are several if Units 1 and 2 are combined, so a short, spoken-instruction version of the test would be applicable fairly soon after the start of her programme.
Thanks for your help.
"... 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
I agree that the comprehension element is a disadvantage - but then using homophones is also a disadvantage in the very early stages, because children simply don't know enough alternative spellings for homophones to be viable. And these very early stages are a vital time to check that children are developing a decoding mindset - that they read words by looking at the letters from left to right, saying sounds for them, and blending the sounds. Regular one-to-one monitoring needs to be done by teachers and teaching assistants - and parents, too, if they can be got on board (for example in the way that we see on the 'Jolly Phonics' video).
I think your kind of homophone test could work well after that, Ken. I just wouldn't want teachers to get the idea that the testing of decoding should start only when homophones become viable.
Jenny C.
I think your kind of homophone test could work well after that, Ken. I just wouldn't want teachers to get the idea that the testing of decoding should start only when homophones become viable.
Jenny C.
Yes, if you can afford one-to-one administration of e.g. a pseudo-words test, that is far better. The place for a computer administered test of decoding is comparison of schools and, possibly, reading prgrammes.chew8 wrote:[..] these very early stages are a vital time to check that children are developing a decoding mindset - that they read words by looking at the letters from left to right, saying sounds for them, and blending the sounds. Regular one-to-one monitoring needs to be done by teachers and teaching assistants - and parents, too, if they can be got on board (for example in the way that we see on the 'Jolly Phonics' video).
"... 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
I think it's extremely important for schools to realise that there's no substitute for one-to-one monitoring of children's decoding ability in the early stages. After that, computer-administered tests may well be a good thing. Because of the variation in the order in which different programmes cover grapheme-phoneme correspondences, however, I still wonder how early objective comparisons are possible - towards the end of the second year of instruction, perhaps?kenm wrote:Yes, if you can afford one-to-one administration of e.g. a pseudo-words test, that is far better. The place for a computer administered test of decoding is comparison of schools and, possibly, reading programmes.
By the way, I think that Durham University's Performance Indicators in Primary Schools tests are computer-administered. I haven't seen the reading test, but think it may be based on the assumption that children have been taught by mixed methods - i.e. that they should know 'sight' words as well as being able to decode. I'll try to find out.
Jenny C.
I've been using Phonics International to teach my 7 yo to read for about six months now and we currently reviewing unit 6 and starting unit 7.What I want to do next is put them in some order of difficulty, with those involving only early correspondences first and unusual correspondences at the end, so as to present test sets of greater difficulty to more experienced readers.
These are my observations regarding homophones:
They are easier if 1. the GPC is familiar 2. the word is part of the student's spoken vocabulary and 3. if the grapheme only represents one plausible phoneme.
eg. see and sea are relatively easy
When the grapheme represents a number of phonemes the words become more difficult to decode:
eg. bear could be pronounced as beer, stare could be pronounced as star (st + the word 'are'), ware as war (a sometime says 'oh' and re often says 'r', and war at the start of a word can be pronounced as 'wor')
So bear and bare, stairs and stares and wear and ware are more difficult.
If the word is beyond the child's spoken vocabulary it is effectively a nonsense word and the child has to make the decision to try other plausible phonemes until the word makes sense or accept that it could be a real word that they haven't heard before. Assume 'where' and 'wear' are familiar words but 'ware' isn't. You may be testing vocabulary just as much as decoding ability here and vocabulary like pronunciation, varies from region to region.
Not sure if that helps - I've found that homophones can become difficult very quickly. Even today when we were reviewing the /air/ sound using the Unit 6 Sounds Book activity sheet on homophones, I only taught the three easiest ones as homophones: bear and bare, pair and pear, and hair and hare. Reviewing GPCs, decoding new words, learning new vocabulary and learning about homophones all at once is a lot for my 'average' 7 yo to take in.
Your project sounds very interesting and I hope you are able to achieve your goal.
Who is online
Users browsing this forum: No registered users and 3 guests