TRT Podcast #116: What is set for variability? A conversation with Dr. Marnie Ginsberg
Set for variability is a skill that readers use to transform a pronunciation error into the correctly decoded form of the word. If you think it sounds important, you’re right! Today we learn all about set for variability from Dr. Marnie Ginsberg, creator of Reading Simplified.
Full episode transcript
Hello! It's Anna Geiger from The Measured Mom, and today we're going to talk about set for variability. It's an oral language task. It's when kids are reading a word using their decoding skills, but they don't quite land on the correct pronunciation, maybe the word's irregular or it has a schwa, so they adjust the word's pronunciation based on how the word is pronounced in real life. We're going to talk a lot about this today with my friend Dr. Marnie Ginsberg, who is the creator of Reading Simplified, a structured linguistic literacy approach. Here we go!
Anna Geiger: Welcome Marnie!
Marnie Ginsberg: Thank you for having me, Anna.
Anna Geiger: Marnie is a friend of mine. We love to talk business and family and work. Today, we're going to talk about set for variability because I think that's something a lot of us don't have a firm grasp on, and maybe that can help us think a little bit about the texts that we give students as well as maybe thinking about a different way to have them approach multisyllable words. So can you define set for variability for us?
Marnie Ginsberg: Well, it is a mouthful, isn't it? And another easy way to label it would be mispronunciation correction.
Anna Geiger: Yes, I like that.
Marnie Ginsberg: Yeah. So set for variability is a skill that a reader uses to transform a pronunciation error into the correctly decoded form. If we've been teaching reading, we've all seen this happen. The child looks at the word "down" and says, /d/ /ō/ /n/, but they recognize that's not a word, so they adjust and they come up with /d/ /ow/ /n/. So they went from the mispronunciation of /d/ /ō/ /n/ to the correctly decoded form /d/ /ow/ /n/, and then they're off to the races.
So that little cognitive flip is a strategy that kids need to apply, and it turns out we haven't really been studying it until recently. Some really exciting things are being discovered by researchers, just in the last few years especially, about the importance of this strategy.
One study figured out how to actually operationalize it, which is how researchers talk about how to actually demonstrate this skill apart from all the other things that happen when you read. A test for set for variability that, for researchers, would be taking it outside of the written form and just playing a game orally. So if I say, "breek-fast," can you transform that into a word that is a real word?
Anna Geiger: Yeah. Breakfast.
Marnie Ginsberg: Breakfast. Yeah? Breakfast. Or if I say, "Mow-there," you might transform that into "mother." So that's an oral task.
Researchers have discovered that when they give kids that test, it's correlated to a lot of reading measures, particularly word identification, being able to recognize words immediately. Researchers have also noticed when they use that kind of test that it's a second important decoding strategy.
I think this is really important for us to basically change our model, especially if we come from a phonics background. Most of us have this model where to get to a decoded word, you need letter sound knowledge, blend those sounds together, and then you come up with the word, right? Those are basically the two steps. You have to recognize the letter sounds, and you have to blend those sounds together, especially for a single-syllable word. With multisyllable words, you might add some other strategies.
Those are just two steps, and so that's what we focus on, that's what we're coaching for, and we're planning for that.
But there really should be a third step. It should be letter sound knowledge, it should be blending, and it should be then mispronunciation correction, aka set for variability. That is a second strategy.
The other thing that's exciting from research is not only that it's there and that good readers have that skill, but also that we can teach it.
Now, this is preliminary. There's just been a handful of studies showing that you can actually intervene to develop the skill, and that it will benefit kids in their reading. But from my personal experience with all the programs that I've done for almost twenty years, I have always used that second strategy with Reading Simplified. I call it "flex it." You're flexing the vowel of the O in /d/ /ō/ /n/ and plugging in the /ow/.
Anna Geiger: So with beginning readers, if we're giving them one hundred percent decodable text, as in there wouldn't even be the word "the" for example, they're not going to get any chance to practice this. Correct?
Marnie Ginsberg: They won't get much chance, I guess, because as you point out, they're not encountering anything that they haven't been taught. I think there might be some chance, because even if they've been taught, it doesn't mean they've mastered it.
So I think this is the important point, Anna, you could actually bring this strategy into your teaching from day one.
Say that they're reading something that's highly decodable, which we could get into that in a minute, like at Reading Simplified, the program that I've been working on for about almost ten years now, I advise mostly decodable texts for this reason. But even if you have a highly, highly decodable text, and they get to the word "sat" and they just kind of whiff on maybe the /ă/ sound, and they say, /s/ /ā/ /t/, you can employ this simple strategy by tapping on the A and saying, "What else could this be?"
And that's it! You're prompting them to use their own cognitive flexibility to play around with the sounds and try to figure it out. By the way, if they can't figure it out, there's certainly no problem in then saying, "Try /ă/." It's not like you have to always withhold the information and put all the burden on them, but you can have a mindset of first seeing what they can accomplish without your intervention or with the least intervention possible. And then if there's frustration or they're not going anywhere, then you can always give them the bit of phonics information that they need.
Anna Geiger: Any other phrases that you would use for coaching besides "What other sound could that be?"
Marnie Ginsberg: "Try another." "What else?" I like to say some of the time to affirm, "Yes, it could be, what else could it be?" That's really what I do the most.
Anna Geiger: Maybe you can talk to us a little bit about how this applies to multisyllable word reading. So there are different approaches to that; some are very, very structured, some are a little structured with more flexibility, and some are quite flexible, which is more what you would do.
So talk to us about a sample long word, and how you could use set for variability to help someone read it.
Marnie Ginsberg: When kids are ready, when they can blend a single syllable word pretty consistently, and they've been learning a lot of advanced phonics, like the long vowels or the /er/ sound, that could be /er/ in "her" or "girl" or "earth" or "fur." Once that's coming online for the child, they're probably already ready for the easiest level of multisyllable words, like two syllable words.
We would show them words already chunked, like the word "funny." We would write F-U-N-N because that's the first chunk of sounds, and then we'd write Y, because that's the second chunk. Notice that we're organizing it by phonology instead of the dictionary, because you don't say fun-ny with two /n/ sounds in funny, you have one.
So because we're focused on speech first, language first, that's how we organize it. So the child sees an example of a word by chunk, and then she would write it by chunks after she's decoded it. She would write it, and she would say, "funn," leave a little space, "y." So she gets that modeling and explicit examples of several simple two-chunk words where it's already displayed and she practices writing it.
We know writing and connecting the sounds is super helpful for building orthographic mapping or getting words to stick automatically and helping learn phonics information. So this is the beginning of the process of internalizing how to attack multisyllable words, how they function. And it becomes easier and easier, I should say, for this child to start to pick up the patterns unconsciously of our language. But that's step one.
The step two would be then you move her into a book, whether decodable or transitional text, or if I have an older reader, they're immediately into a novel that's at their level. Then they come to a word, maybe they come to the word "alternate" and they chunk it wrong, so we would just cover up the chunks and reveal it chunk by chunk, with a little card, maybe our finger. So they would see first AL, and they would say, "Al." Then you would reveal the next chunk and they would say, "ter." And so they would say, "Al-ter," and then you would reveal the whole thing, or maybe cover up the beginning now, and just let them see N-A-T-E. They say, "nate," and then they say, "Al-ter-nate ... oh, alternate!"
Which basically right there was a little bit of set for variability, because they decoded it, "nate," but then they realized that this doesn't really kind of jive with how they've heard that word. So then they flex into the right sound, which is kind of the schwa. I think that's the schwa, right? Alternate. It's almost the /ĕ/ sound.
Anna Geiger: Mm-hmm.
Marnie Ginsberg: So we move from a word work activity then into print. And then of course we're doing a lot of writing, and they're writing in chunks. They can also look at words that are already written that aren't chunked.
Now this would be kind of phase two. We just show them a word like "difficulties," and then they have to mark where the chunks are, like a slash between DIFF and I and CUL and TIES. And then also maybe, again, they would map it or say the sounds as they write it, diff-i-cul-ties.
So those are the main activities that we use with Reading Simplified, and what we see is that kids internalize more and more about the patterns of syllabication, and so we don't need to give them a lot of rules about the types, open and closed for instance. That starts to become internalized, because the brain is amazing at connecting the language system with the orthographic system and doing this work, a lot of it subconsciously.
Now we're going to have our really truly dyslexic kids that are less skilled at statistical learning, but we can still use the same process, and then just have more practice with the patterns that we think they're not noticing. So if the pattern is open-closed type syllables, we could have them sort words by that, but they don't have to label it open or closed, long or short. They could just read words and then decide whether it fits with the sound of /ă/ or the sound of /ā/. So we're focused mostly on the sounds that they hear. They do the work and sort those words, and then these patterns do get picked up by them.
Anna Geiger: Can you go back to statistical learning and explain more about that?
Marnie Ginsberg: The big picture is all of us learn oodles of things about the world through subconscious observation of patterns. We observe patterns of people's faces, and there's just so many things that we're learning all throughout our life, particularly in childhood, where we group things together into categories that help us process information in a split second.
We absolutely have to be told how the code works. There's very, very, very few people in this world that will deduce it just by looking at those squiggles. We have to be told how it works. We have to be coached into the letter sound knowledge. We have to be coached into these strategies that we've been talking about, blending and set for variability, and we have to practice it.
But at some point, and the point varies for each individual learner, more and more of the learning becomes implicit, because it's observed. We observed patterns that we see from reading, and then we're not even aware of it, but we know how to apply the rule.
For instance, Rebecca Treiman and her colleagues have done a study where they showed that some first grade readers come to a nonsense word that ends in E-A-D, and they don't say /ē/ /d/, they say /ĕ/ /d/, which is a variation on what you would expect. These are the better readers. The better readers know that E-A-M would be /ē/ /m/, but E-A-D would be /ĕ/ /d/ because they've seen it in the words "bread," "dead," and "instead."
Anna Geiger: Interesting. Yeah.
Marnie Ginsberg: So it's amazing that they could be that young where they've not been explicitly taught, most likely, that sometimes EA is /ē/, sometimes it's /ĕ/, and know how to apply it, because we're talking about a nonsense word. Because it could be /ē/ /d/, it could be /ĕ/ /d/, it could be /ā/ /d/, as in the EA in "great."
Why are they applying that? They read "dead" and "bread" enough times that they're subconsciously, through statistical learning, realizing there's something associated with this ending that's going to trigger me to say /ĕ/. But it's subconscious. That's implicit learning through observations of patterns. So that's one way for me to explain statistical learning.
Anna Geiger: How is this connected to David Share's self-teaching hypothesis? Can you walk us through that?
Marnie Ginsberg: David Share had a very influential article in 1995 called The Self-Teaching Theory Hypothesis, and it has since gone on to be validated with lots of studies. His idea is that we can't possibly be taught every single phonics spelling, and we can't be taught every single word because by the time a kid leaves high school, they know 20,000 to 40,000 words.
So he says, and there's research behind this, that at a certain point in the process as a reader develops, because they have, as Share says, the concept of the alphabetic principle, they have sufficient phonics knowledge, they have sufficient phonemic awareness, and then they have a decoding strategy. So you couple those things together.
You're probably seeing kids like this where they're reading a text and maybe they figure out a word that you haven't taught them, or maybe they figure out a phonics pattern that you haven't taught them, because this system is coming online. They're putting all those elements of the triangle, the semantics or meaning, the phonology or sound, and the orthography or spelling. They're putting all those systems together in such a way that they're processing and deducing.
So I remember somewhat, to some extent, learning how to read the word "pterodactyl." It starts with PT, which I had never been taught that PT says /t/. And so I looked at PT, and I was thinking, "I've never seen PT at the beginning of the word." You actually can see PT right at the end of the word, but it would be two sounds, like in "kept," but you don't see PT ever together at the beginning of the word except for these really odd words. So I probably was like, "/p/ /t/," and I had no idea what to do with the beginning of the word, but the rest of the word was fairly easy, "erodactyl." That was my orthography kind of kicking in with sounds. And then I thought, "Ah, erodactyl, that kind of sounds like 'pterodactyl,' which I have heard that said aloud maybe once or twice."
So then my brain's like, "Oh, pterodactyl, that PT must be /t/." Maybe I wasn't conscious of it because I probably just wanted to get on with the story knowing me and most readers, but maybe I read it again, and this time I got a little bit more observation data. My brain is taking in data of these patterns, and then the next time I saw "pteranodon," I might've just gone straight into it and said, "Pteranodon." So that's statistical learning enabling me to do the self-teaching.
Anna Geiger: Okay.
Marnie Ginsberg: So it's self-teaching of phonics, but also I just taught myself not only PT, I taught myself "pterodactyl" and "pteranodon."
So we want to get our kids as quickly as possible into challenging texts with the strong sound-based decoding skills that they need to be able to do that processing that I just talked about to uncork the trickiness of a word like "pterodactyl."
You prepare the way through explicit instruction, giving them good coaching on how to be strategic, giving the bits of information they didn't know along the way, and now they're off to the races.
Anna Geiger: So maybe you can talk about our last point we were going to cover today, which is how does this work with decodable text and nondecodable text? We know that with balanced literacy, myself included, and many other teachers were having beginning readers "read" leveled predictable books. So they weren't actually decoding the words. They were using cues. They were getting to the words, they were understanding the text, but they weren't technically reading. Many of us understand that now, so now we're not doing that, and we're having beginners read decodable text.
I think that's very useful within a phonics lesson for a long period of time, but there's so many questions out there, and I know there's not research that tells us. How do we decide that these kids are ready to read other types of text? How does that work? How do we make sure that they have enough knowledge to be successful? It's just kind of very sticky.
Marnie Ginsberg: It totally is sticky. And you're right, we don't really have a clear-cut answer, and I don't think we'll ever actually have a clear-cut answer for X percentage on day five of instruction versus X percentage on day 105, because each child is so different.
If we begin with the understanding of how word reading develops, and that set for variability is part of it, we will be preparing the way for an earlier entrance into transitional text.
I think that's the big idea that I would like to send out there into the world. The highly decodable texts, if we camp out in them, they could limit the child's opportunity to practice set for variability. They could also limit the child's opportunity to observe the patterns of our language. They can't really do as much statistical learning if they're spending a whole year in CVC text.
Our code is much more complex than that, and they need to understand, first of all, the concept that one sound can have multiple spellings; the /ō/ sound can be the /ō/ in boat, the /ō/ in snow, the /ō/ in home, the /ō/ in go, or the /ō/ in show. When we not only move to decodable texts that have that more sophisticated understanding about how our code works, then they can actually have an earlier entree into playing around with sounds and words, which is really what set for variability is.
They can also have an earlier entree into that information gathering that needs to happen if you're going to be able to do statistical learning, because when you camp out in one vowel letter, one sound, it's not how the code works. We need to prepare them for being flexible.
Also they just don't have enough time to observe the patterns, because if they don't see the OA until they're in their tenth month of school, the beginning of first grade or the second half of first grade, they can't do that sophisticated stuff that Rebecca Treiman's research says is happening in the middle to end of first grade, that a kid could read the E-A-D and say /ĕ/ /d/.
With Reading Simplified and all of my tutoring, it's not as if I go straight from a decodable text into a transitional text and we never see decodable texts again. We like to get to about the fourth advanced phonics sound, third to fourth or fifth advanced phonics sound, it varies based on the child. So they've learned the basic phonics, so the short vowels. They've learned /ō/ and its various spellings, /ē/ and its various spellings, /ā/. About then, they know a lot of the phonics that will help them with a book like Frog and Toad or Little Bear or Messy Bessy.
So we will continue to choose a decodable text to teach them the /ī/ sound, and/or maybe next week, the /er/ sound, but we're also going to be reading a text like Henry and Mudge.
Your first and relentless priority is developing that sound-based decoding approach that the child has an attack approach to an unfamiliar word, not to look at the picture, not to make a guess based on the first letter, but to look at every sound, every symbol, read them left to right. If that's their first approach, then the next thing they need to do is to have those words that they've decoded become automatically recognized, or become so-called sight words. They need to be orthographically mapped, and that takes repetition.
That's what's so great about these early transitional texts. They're really designed for these words that they're not automatic with yet, to become automatic because it's constrained to how many words are in a text, and that many of them are repeated, and they have a lot of high frequency words.
What is important to know is that the top three hundred words represent about 65% of written English. So let's teach our kids to decode those words and see them enough so that they can quickly get into Henry and Mudge. Then they reread Henry and Mudge, and those words become solid.
Once those words are automatic, then they can start to ... Really, that's again part of that hockey stick. That's when you start to see an explosion in their self-teaching and their recognition of lots of words, because, to some extent, they're kind of blowing by all these high-frequency words. Those are not a hard all. They read everything really rapidly, and when they get to a hard word like "challenging," they might slow down and work on that, but then they see words like "said," and "the," and "into," and "from," and "over," and "mother," and those take no mental energy. So if we could release more phonics information to kids earlier, those words would be easier.
Anna Geiger: Well, thank you. This has given us a lot of things to think about. Are there any specific things you want me to share in the episode?
Marnie Ginsberg: I always recommend our activity Switch It as a preparatory activity, and so we have that resource at our website. Your goal is to get the kids obviously to learn the letter sounds, but then also to push them into higher and higher levels of phonemic difficulty. To keep raising the bar with the words and those word chains. Don't just stay at three sounds. As soon as they can, to some extent, push them to four sounds, and then push them to five sounds. At the end of that, we even do Switch It with nonsense words so we're really still pushing that cognitive flexibility of sounds and symbols, which I think prepares the way for set for variability. A lot of people, no matter what curriculum they're using, can fold that activity within because it just takes about five minutes once you get the hang of it, and it hits a lot of skills and strategies that kids need.
Anna Geiger: Wonderful. I will link to any workshops that I find that you've got free online-
Marnie Ginsberg: Thank you.
Anna Geiger: ... as well as your website. Thank you again so much for taking time to talk to us today.
Marnie Ginsberg: It's been my pleasure, Anna. Thank you.
Sign up to receive email updates
Enter your name and email address below and I'll send you periodic updates about the podcast.
Links related to this episode
- Marnie’s Reading Simplified YouTube channel
- Video: Learning Phonics Quickly with Switch It
- Reading Simplified website
To find articles on set for variability, go to Google Scholar and search “set for variability.”
Anna we had connected a while back (Google Meet) regarding the Enriched Read Aloud. Just wondering if you could give me some feedback.
The podcast on Variability gave me a strategy name that I use often, thank-you
Shelley, I do remember having a conversation … but I’m sorry that it’s been so long I don’t remember the particulars! Feel free to send me an email and refresh my memory!
Enjoyed the interview. Thank you. Helpful. You speak of ead as a short e sound. What about “read” as in “to read a book?” I believe it is a long e sound, unlike “dead” and “bread.” Please clarify.
That word would not follow the pattern, but “read” as in “I have read a book” would.
The variability factor has been a strategy used after mapping doesn’t work in Reading Recovery. It’s nothing new to strongly trained RR teachers. It’s an integration of M (meaning/comprehension), S (structure/syntax), and V (phonics). And just as Marnie said, it depends on the child. In RR teaching, we are revisiting those patterns (which have already been explicitly taught) over and over depending on the needs of each individual student. RR is just so misunderstood. The actual SoR methodology just keeps confirming that RR is such a fantastic program! Yay! Great podcast!
Thanks for sharing your perspective, Taryn! I’m afraid that I can’ t agree with you about the value of Reading Recovery, because I don’t support the use of three-cueing.
I have to agree with you. I was trained in RR and explicit phonics instruction was not a part of the training or the lessons. Students were taught to look at the picture and check that the first letter matched the word they were predicting. The books used for RR and lesson components do not follow the science of reading at all. If anything they are polar opposites. I’ve had much more success with teaching children through a structured literacy approach and using decodable texts.
Thank you, Ell, I appreciate this feedback!
When the “visual cue” of phonics doesn’t work and the student mispronounces a word, my understanding is that we should then be teaching/modeling a skill Dr. Ginsberg refers to as “set for variability,” which seems to be encouraging students to use their background knowledge of meaning and syntax to correct their mispronunciation. How is this different than the three-cueing system aside from first and foremost, encouraging students to mostly rely on their visual cues (or phonics information)?
Also, I’d love to know if there is research on the timing Dr. Ginsburg suggests of moving students away from strictly using decodable texts. Thanks! I’m really getting a lot out of thought-provoking info from your Podcasts.
Hi Abigail! Students should always use the word itself when reading it – by attempting to sound it out if it’s not a word they’ve already mapped. After coming up with a pronunciation, they can check that using context and syntax, since phonics doesn’t always land us right on the correct word. The difference is that with three-cueing, the visual cue (phonics) is typically considered less valuable and is subordinate to context and syntax. Also, three-cueing theory says that skilled readers don’t look at all the letters of the word when reading; they mostly use context and syntax. However, we know from eye movement studies that this is not the case. There is very little research on decodable books at this time; I would reach out to Dr. Ginsburg to see if she knows of something related to when to move children out of decodable text. However, I highly doubt that research has been done yet.