A Question I Hate: Should We Use Pictures (Embedded Mnemonics) When Teaching Phonics?

  • decoding phonics
  • 12 June, 2021

Teacher question:

I teach kindergarten. Our school recently purchased the XXXX program for teaching decoding. I don’t like it as much as the program we had. One of the ways our previous program was better than XXXX is that it included pictures for each of the phonemes. The new program does not have those pictures and I think that is a real problem. Is there any evidence that I’m right that I could take to my principal? The other teachers agree with me.

Shanahan response:

I hate that question and I wish you hadn’t asked it.

Oh, sure it’s a practical question. And as a former first-grade teacher I get why you’d ask.

But it points out an error I have made in the past.

When I used to prepare primary grade teachers that question came up sometimes. And, I answered it… incorrectly.

I answered it on the basis of logic, a dangerous approach.

“Professor Shanahan is it a good idea to use pictures when you are teaching letter sounds?”

“No, impressionable young preservice teacher. It is a bad idea. Children have to learn the letters and the sounds. Adding a picture to that equation means that there is just one more thing the kids have to memorize. It is hard enough to memorize 52 letters, and 44 sounds, without having to memorize 44 pictures to go with those sounds.”

I love that answer. It makes so much logical sense. I sure sounded wise to those young women (and the occasional guy).

And, yet, I was wrong.

The research these days shows just the opposite. Using what are referred to as “embedded mnemonics,” that is pictures that remind the children of the letter sound, actually improve learning. Across various studies (Ehri, 2014; Ehri, Deffner, & Wilce, 1984; McNamara, 2012; Schmidman & Ehri, 2010) it has been found that such embedded mnemonic pictures can reduce the amount of repetition needed for kids to learn the letters and sounds, with less confusion, better long-term memory, and greater ability to transfer or apply this knowledge in reading and spelling.

Art Credit: The example of a visual mnemonic for teaching decoding was provided by the artist Cat MacInnes, from www.spelfabet.com.au/materials” 


If one relies on data – rather than reasoning – the answer is kind of a no-brainer — it is a good idea to use embedded mnemonics. It looks like, at least with regard to this feature, your previous program was better than the new one.

But let’s be careful with this. What’s good in this case (the use of pictures) is not such a great idea when you are teaching words. Research has long shown that when teaching words kids do better with just the letters without any pictures.

Why would that be?

The pictures have been found to distract from the information that you need to remember a word. In that case, students don’t need a visual mnemonic -- they need to focus their attention on the combination of letters. That’s what you want them to look at. If they spend time examining the picture instead of the letters, they are less likely to learn the word.

With something as specific or one-dimensional as a letter or a phoneme, an embedded picture provides a useful mnemonic (memory support), neither distracting students from what needs to be learned or overwhelming memory. With more complex or multidimensional items it is better to focus student attention on analyzing the parts. That’s why, when I’m teaching kids to read some high frequency words, I have them look at all the letters and spell the word and try to write it from memory.

But when it comes to teaching letters and sounds, no question about it, use embedded mnemonics. They work.

And, as for answering questions about what works in reading? An instructional approach is not "best practice" just because it makes sense. That's why we use research. A lot of times, those things that are sensible are, well, not a great idea!


Ehri, L.C. (2014) Orthographic Mapping in the Acquisition of Sight Word Reading, Spelling Memory, and Vocabulary Learning, Scientific Studies of Reading, 18:1, 5-21, DOI: 10.1080/10888438.2013.819356

Ehri, L. C., Deffner, N. D., & Wilce, L. S. (1984). Pictorial mnemonics for phonics. Journal of Educational Psychology, 76(5), 880–893. https://doi.org/10.1037/0022-0663.76.5.880

McNamara, G. (2012). The effectiveness of embedded picture mnemonic alphabet cards on letter recognition and letter sound knowledge. Unpublished Master’s thesis, Rowan University.

Shmidman, A., & Ehri, L. (2010) Embedded picture mnemonics to learn letters. Scientific Studies of Reading, 14:2, 159-182, DOI: 10.1080/10888430903117492


See what others have to say about this topic.

Harriett Jun 12, 2021 05:35 PM

Two thoughts. First, I was super lucky to have discovered Ehri's article on letter-embedded picture mnemonics before my one year teaching kindergarten, and it made a world of difference. I used a program that also involved songs and gestures. I'm not saying there's research behind those two features--I'm saying it was fun. Second, I've just finished reading Districts that Succeed: breaking the correlation between race, poverty, and achievement, and one of the superintendents featured would applaud your admission of error. He says: You have to check your ego at the door and admit there are some things you don't know. Thanks, Tim, for a very important post. I really believe we need picture mnemonics in our beginning reading programs.

Joan Sedita Jun 12, 2021 06:09 PM

You make a very good point ... that the picture clues are a great help as students are trying to learn and remember the sound (phoneme) associated with the letter. By saying the word that goes with the picture and then isolating the first sound, it provides a scaffold. However, once students have become automatic with the letter-sound correspondence the picture clue can be removed. In some phonics programs the picture is placed next to or below the letter on a letter-sound card. There are a few phonics programs, such as Telian's Lively Letters and the Letterland program that actually embed the picture into the letter shape, which seems to be added benefit, as noted in the National Reading Panel report: "The value of mnemonics for teaching letter-sound relations to kindergartners is supported by evidence. In a study by Ehri, Deffner, and Wilce (1984), children were shown letters drawn to assume the shape of a familiar object, for example, s drawn as a snake, h drawn as a house (with a chimney). Memory for the letter-sound relations was mediated by the name of the object. Children were taught to look at the letter, be reminded of the object, say its name, and isolate the first sound of the name to identify the sound (i.e., s ­ snake - /s/). With practice they were able to look at the letters and promptly say their sounds. Children who were taught letters in this way learned them better than children who were taught letters by rehearsing the relations with pictures unrelated to the letter shapes (e.g., house drawn with a flat roof and no chimney) and also better than children who simply rehearsed the associations without any pictures. Application of this principle can be found in Letterland (Wendon, 1992), a program that teaches kindergartners letter-sound associations. In this program, all the letters are animate characters that assume the shape of the letters and have names prompting the relevant sound,for example, Sammy Snake, Hairy Hat Man, Fireman Fred, Annie Apple. The task of learning the shapes and sounds of all the alphabet letters is difficult and time consuming, particularly for children who come to school knowing none. The relations are arbitrary and meaningless. Techniques to speed up the learning process are valuable in helping kindergartners prepare for formal reading instruction. The motivational value of associating letters with interesting characters or hand motions and incorporating this into activities and games that are fun is important for promoting young children’s learning. If the task of teaching letters is stripped bare to one of memorizing letter shapes and sounds, children will become bored and easily distracted and will take much longer to learn the associations."

A more recent study also found an advantage to having picture clues over plain letters (Theresa Roberts & Carol Sadler: Letter sound characters and imaginary narratives: Can they enhance motivation and letter sound learning? Early Childhood Research Quarterly, Vol. 46, 2019). This study compared two explicit teaching approaches for teaching alphabet letter sounds to preschool children: The Letterland program (i.e., integrated mnemonics) was used for the first treatment (utilizing letters with letters sound characters integrated into the letter shapes and short narratives about the letter sound characters). In the treated control, plain letters and alphabet books were the foundation of instruction. The study found significant effects in favor of the use of Letterland on identifying letter sounds (effect size 1.31), identifying initial consonants(effect size 0.61) and blending phonemes (effect size 0.62).
The final comment I'd like to make is related to this line in the NRP report "If the task of teaching letters is stripped bare to one of memorizing letter shapes and sounds, children will become bored and easily distracted." I have observed a number of classrooms that use Letterland or Lively Letters and have seen first hand how engaged young children are when they learn basic phonics instruction in this way.

Lauren Thompson Jun 12, 2021 06:17 PM

So glad Shanahan wrote about this so that this topic gets some reach! I'm surprised more phonics programs don't act on these findings. A few programs that feature letters embedded in mnemonic pictures are Letterland, ReadWriteInc, and Spelfabet. I'm sure there are others. Often you can buy just the letter cards apart from the whole program. (I've yet to find a program where I agree with all the choices for keywords -- e = egg is not the best, IMHO.)

Pam Edenfield Jun 12, 2021 07:49 PM

Harriet- what pictures , songs, gestures did you uae?

Jo-Anne Gross Jun 12, 2021 08:35 PM

I don’t believe in all that first sound stuff like the ones taught in very popular oral only PA programs right now.Segmenting from speech is applying the research after they learn grapheme phoneme connection.Ehri
I believe in every sound, blending to read and segmenting to spell.
The short vowels really need mnemonics so the kids can segment the sound from the word to get to the proper articulation, they can be so tough to learn for some kids.

Harriett Jun 12, 2021 08:51 PM

Pam, I really liked using Zoophonics.




Donna Jun 12, 2021 10:23 PM

52 letters? Ok, if you mean capitals and small letters.

Timothy Shanahan Jun 12, 2021 10:26 PM


Capital and lower case it is.


Mary Conner-Righter Jun 14, 2021 12:00 PM

Another example of letter-shaped pictures is Itchy’s Alphabet. There are short songs for each letter. One product has the picture with a plastic overlay of the letter that matches. An example is: ‘b’ is a bat and ball. We used this alphabet and other alphabet-related products in the Orton- Gillingham tutoring. My students really like the vowel characters that come as a set of plush toys.

Ann Coffeen Turner Jun 15, 2021 03:20 PM

Ehri, Deffner, and Wilce, 1984, didn't just make a comparison between letters with pictures versus letters without pictures. They also compared letters with the pictures BESIDE them (like Opencourt) versus letters with the pictures embedded mnemonically WITHIN the shapes of the letters. The "beside" letters didn't do nearly as well as the "embedded WITHIN" letters. Shanahan, above, advocated not using picture-letters when making words. I disagree, but only because there are intermediate steps. In the first place, I play games that involve matching picture-letters to plain letters so that both are automatic. Next, I have picture-letters on cards in beginning-middle-end positions, and my beginners build three-letter words from those cards.The cards are at the recognition level, not the retrieval level, so the children can choose the spellings quickly. They work with spelling this way until they can blend sounds (phoneme awareness). Then it's time to make words out of letters without pictures. The words are then practiced, through games, until they are automatic. Then and only then are they incorporated into stories. In other words, I agree with Shanahan that the words in stories should not have picture-letters, which would be a distraction, but I also find embedded picture-letters to be extremely helpful during the phoneme-awareness stage, which involves spelling with cards that have picture-letters on them. www.mnemonicpictures.com

Timothy Shanahan Jun 15, 2021 04:25 PM

I think you misread what I said. I definitely support the use of "picture letters" in the teaching of decoding.


Timothy Shanahan Jun 15, 2021 09:44 PM


You’re not wrong in your reasoning and given your reasoning it should not be surprising that some letters (and sounds) are easier to learn than others (Jones & Reutzel, 2012). The easier ones take less time to master, the harder ones (and the more confusable ones) tend to take a bit longer. That means that we don't need to teach all of them in the same way and we definitely don't need to devote as much attention to some as to others.


Eric Jun 15, 2021 10:25 PM

Hi Tim,

Thank you for writing this article. We use a resource, Fundations, with our K-2 students and are seeing great results with decoding. Our students do a great job recalling the letter, keyword, and sound (b, bat, /b/). However some students always go through the letter, keyword, sound progression when asked to identify only the letter or only the sound. Any advice on how to transition some of our readers from the embedded mnemonic support to identifying only one given element? Or will, with time, students naturally drop the support?

Emma Jun 16, 2021 09:26 AM

Thank you so much for this. I have written a response several times however struggle to do so succinctly:-) As an early years educator, former Inspector of Early Years Education, and now studying early literacy intervention at doctoral level, I am passionate about this topic as picture embedded mnemonics (in particular Letterland) has been used so widely for so many years. The idea behind it is great, as the picture clues prompt the child to remember the 'letter sounds' the program wants them to use in the early stages. However as someone who has run 2 large nursery schools for 8 years, where we tried this, I found there to be limitations and set about expanding on the concept. Im an action researcher and problem solver, as are so many who are passionate about eradicating illiteracy.

I wanted to think bigger, and ensure that the scaffolding was not short sighted. Most educational research, as variables needs to be controlled when undertaking qualitative research is, by it's very nature, limited. Researchers are only able to look at the potential effects in the very beginning stages of learning to map phonemes to graphemes, and this can ignore the issues children are going to then face as they navigate the whole code, and the learning cannot be applied. Letters do not 'make sounds', they represent speech sounds in word, and the phonemes (speech sounds) they represent DEPENDS ON THE WORD. So teaching a child to say 'æ' when they look at the letter a is all well and good, until they see a real word, and the letter a does not represent that sound at all, as demonstrated by Millie who is 'following the (phoneme) monster sounds to say the word in my blog about this topic - the word is 'any'. I wanted the picture clues to represent phonemes, as do phonetic symbols. So I created phonetic symbols for kids, called the Speech Sound Monsters. In the blog you can see pre-school children working out words like 'said, any, want, a...' and have no problem, even though the letter a represents a different phoneme each time! Why? The clue is just an alternative to a phonetic symbol. So the representation for that phoneme depends on the word. I teach very (very) young children to read, as an early intervention - to avoid them becoming instructional casualties. Schools in Australia and the US have also been using them for a few years now and the data collected is just wonderful - I would welcome contact by any scientists interested in quantitative research who wishes to explore an extension to the existing research relating to picture embedded mnemonics (phonetic symbol embedded mnemonics!)


Troy Jun 15, 2021 09:20 PM

I've wondered about the 52 symbols. Some are the same, just "lower": c o p s v w x y z. Some are such close cases: J/j; K/k; U/u. These examples are great for explicating spacial difference between upper and lowercase alongside conceptual teaching about lc and uc. We take these things for granted but for kids, this is significant and interesting as they learn to not just recognize but also form letters and gain spatial awareness/skills to foster writing.

That 52 thus reduces to 43 or less. Knowing up front that some letters are the same in both uc and lc (and some are not) is also useful. This is more direct teaching than the comments but we've found direct teaching to also improve memory and retrieval for letter knowledge. Not all letters need so much "noise" around their teaching. Elaboration--vs rote presentation-- helps recall.


Emma Jun 16, 2021 02:43 PM

Lauren T - yes, they are so useful for the children to prompt them to remember the speech sound to grapheme link. And yes, the choice of pictures is often tricky - and as many use a word and the intended phoneme link is not the first sound, so it can take adults a while to work out the association! And of course if they recall the intended phoneme they also need to blend them. But my question was why not just an alternative to a phonetic symbol - so it's universal. The child looks at the character and thinks of the sound. With Letterland and others they look at the letter a and see an apple, but that means the program has to put the elephant, or relevant picture clue, on the letter a when it doesn't represent that sound and the words presented must only have those - when do they move on from this? eg think of the word 'any' ...does an elephant get put on the letter a? Or when are they told what to do - the letter a can represent at least 8 sounds. When do the children see that it's not just æ. I have found that phonetic symbol/ speech sound (phoneme) monsters are great because from the very beginning they think sound, not ONE phoneme to graphem combination. It also addresses accents / dialect - when Aussie kids say 'pan' they do not use the sound æ in the middle of the word! So then when they see the æ monster it evokes discussions about accents and phonics. And they spell in monsters, which show us which sounds they are hearing. I initially created them for my non verbal kids.
I do hope scientists compare picture embedded mnemonics with phonetic symbol embedded mnemonics - of course kids aren't going to be interested in phonetic symbols, and why I used phoneme monsters instead - each with their own music, movement and personality. They even collect all the pictures of their sounds in their Spelling Cloud. It seems to align with the science, but is an assumption without empirical research - and I hope someone will step up and do that. I send kids to school reading, which is great proof for me- and schools send me their data to show huge growth. It is definitely worthy of investigation using the scientific method, and could build on existing findings.

Lisa Jun 16, 2021 09:37 PM

Thanks so much for your informative words. More research would certainly be helpful with respect to memory devices. Would you be able to discuss the distinction between 'teaching words' and 'teaching decoding'. You say..."But let’s be careful with this. What’s good in this case (the use of pictures) is not such a great idea when you are teaching words. Research has long shown that when teaching words kids do better with just the letters without any pictures."
And then in your response to Ann you say " I definitely support the use of "picture letters" in the teaching of decoding.".

Timothy Shanahan Jun 17, 2021 01:00 AM

Lisa-- Often teachers teach specific words to enable students to read text that they may not yet have the decoding skills to handle. Let's say that a teacher wants to teach students to read and spell the word monkey. Many schemes for this would have the teacher showing the student the word m-on-n-k-e-y along with a picture of a monkey, the idea being that this would show the students the meaning of the word and increase their ability to learn to read that word. Research on that approach show that it lowers rather than raises the amount of learning. The time students spend looking at the cute picture of the monkey distracts from the time used to look at the letters in the word, so it doesn't work. When teaching a word, it is essential that kids look at the letters (I often have them spell it out, and try to visualize it with the word removed, etc.), but pictures in that case are an impediment rather than a support to learning. (Whereas an illustration of a monkey that looks like an M, can be a really useful memory cue for helping kids to link the letter Mm and the phoneme /m/).



Timothy Shanahan Jun 17, 2021 02:15 PM


Yes and yes,


Troy Jun 18, 2021 10:36 AM

Do you have time to relate this post to one before last: What if no reading research on issue? Clearly some research on this letter-learning mnemonic topic but seems a low level of research? I followed your advice from prior post to read individual study. The SSR (2010) study is 36 English-speaking preschoolers taught 10 Hebrew letters. There IS value in this study for other populations in my opinion; the mechanisms at work are helpfully detailed. But I recall your advice: "read some of the individual studies to determine whether the approach worked in situations like mine and get clues about proper implementation." Not a situation like mine but absolutely clues about implementation. Research that applies to a different population is a sticking point for many teachers. I do worry about this relative to equity: research with typical middle-class kids applied to others, etc. To your point about research article vs practice article: one of the problems with the citation studies for applying to practice is segmentation as a prerequisite. This can be challenging skill for kids not knowing letters yet to learn, even if we try to teach it alongside letter learning, segmentation often emerges later. I'm going to review NELP but hoping to get this to you before this post goes cold in case you have comment. It's hard to translate research to practice.

Lisa Jun 17, 2021 11:31 AM

Thanks Tim.

Sorry to be pedantic, I'm a details person. So, using pictures letters for teaching decoding, do you mean just when teaching letter/sound correspondence (required for decoding) or when teaching actual decoding.? For example, I would teach the 'letter/sound's for s a t p i n d using the embedded mnemonics reaources but when I start to teach decoding the word 'sat' as /s/a/t/ I would switch to just letters.
I'm also hoping it would still be beneficial to show a picture of a monkey before the word activity if the child has no clue what a monkey is.

What Are your thoughts?

Leave me a comment and I would like to have a discussion with you!

Comment *

A Question I Hate: Should We Use Pictures (Embedded Mnemonics) When Teaching Phonics?


One of the world’s premier literacy educators.

He studies reading and writing across all ages and abilities. Feel free to contact him.