Deep Dives

The Issue

Much of the movement for the “Science of Reading” is centered around research showing that explicit instruction in grapheme-phoneme correspondences (GPCs) is a necessary foundation for proficient literacy development. It is no secret, however, that phonics instruction only takes students so far, even if we narrow our focus and only consider its impact on word reading. Students will inevitably encounter irregular, non-phonetic words that do not align with known GPCs, leaving teachers caught in the crossfire of needing students to read high-frequency words with irregular (or as-yet-untaught) spellings, without drawing their attention away from consistent grapheme-phoneme patterns– patterns essential for developing decoding skills and avoiding the habit of guessing. In response, one method has risen to prominence: Heart Word routines. With this approach, students "map" each grapheme to a phoneme and draw hearts above the irregular grapheme(s) to mark them as the portion they must “know by heart.” At the heart (pun intended) of this shift is a rejection of rote memorization (such as the use of flash cards) for word learning, based on the concern that memorizing whole words is inefficient and potentially even counterproductive. But whether this level of analysis is truly the most effective or efficient use of instructional time remains a question–especially when we know that repeated exposures to a word also contribute to automatic recognition.

Importantly, while Heart Word instruction is gaining ground as a “best practice,” no body of research has yet compared its effectiveness head-to-head with other methods.

The Research

At the risk of making a long article longer, I want to offer a brief, anecdotal example of the word hors d'oeuvres. Even though I was familiar with the word in my oral vocabulary, I did not recognize the written form when I first encountered it in a text. I needn’t burden you with the full account to convey the embarrassment I felt upon my pronunciation finally being corrected, and finally realizing that the letters <hors d'oeuvres> and the sounds /or | dervz/ were, in fact, one in the same. Research validates this anecdote in what can be thought of as somewhat of a hierarchy in instructional practices to support learning irregular words. The familiar but sometimes criticized technique of memorization (such as repetition-based flashcard practice) can lead to learning, and is certainly better than no instruction at all, or even incidental exposure during read-alouds (Stuart et al., 2000). Yet, while memorization does not appear to interfere with simultaneous phonics instruction (Castles et al., 2018; Shapiro & Solity, 2016), relying solely on memorization as a word-learning strategy can quickly become problematic.

I can successfully read the word hors d'oeuvres and never confuse it with another word because there aren't any other similar words in English. Because of my lack of knowledge of French GPCs, only a few letters in the word hors d'oeuvres are solidly anchored in my memory of the word. This means that not only did I need help spelling the word before typing it into this article, but I wouldn’t be able to catch a misspelling of the word. While this partial connection between graphemes and phonemes is sufficient for reading the word hors d'oeuvres, it would not be an effective strategy to help me read and differentiate a multitude of other words with only minimal differences distinguishing them, such as causal/casual, collage/college, desert/dessert. These ideas also align with both theoretical and empirical research. Learning irregular words can be understood as an extension of the theory of orthographic mapping, which explains how students form connections between phonemes, graphemes, and word meanings in memory (Ehri, 2024). Students need both knowledge of GPCs and phonemic awareness in order to successfully learn irregularly spelled words. Put another way, children in the pre- or partial alphabetic phases are generally not successful in learning irregularly spelled words (Castles et al., 2018; Miles 2015; Miles et al., 2018; Neuman et al., 2014; Stuart et al., 2000).

Research supports leveraging the fact that most irregular words actually only contain one unexpected GPC (Miles et al., 2024), and teaching irregular words through analytic methods rather than rote memorization alone (Colenbrander, et al. 2022; Steacy et al., 2020; Stuart, et al., 2000) seems to be beneficial. 

While these studies do not specifically examine the act of drawing a heart above irregular or unlearned GPCs, they share a commonality: directing students’ attention to how the words are spelled. This included techniques such as modeling attempts to sound out the words and then flex the pronunciation to a known word, as well as encouraging students to write the target words multiple times.

Further defining the instructional implications we can gather from current research is limited for several reasons. One major limitation is the inconsistent and often undefined terminology used across studies of irregular word learning. While educators commonly refer to the target words in these routines as “sight words,” this term is more precisely meant to indicate all words that readers have built an orthographic representation (spelling memory) of, regardless of whether or not that word is phonetic. Research literature may reference “high-frequency words” (Miles, 2024) or even “reading vocabulary” (Shapiro & Solity, 2016) to refer to words that students are expected to read automatically, without differentiating between the words that are regularly spelled or not. Moreover, research often fails to differentiate between temporarily irregular spellings (for early readers who have not yet encountered the GPC used) and permanently irregular words (Miles, 2024). Another variable with instructional implications that often goes undefined in studies is whether the target words are content or function words. We see from research that function words (such as of, been, does) are more difficult to read, use orally in a sentence, and spell than content words (e.g., friend, said) (Miles & Ehri, 2017; Stuart et al., 2000), and successfully learning them may depend heavily on students’ language proficiency (Ehri et al., 2018).

Finally, the outcomes measured in existing studies tend to emphasize immediate learning rather than long-term retention. Further scrutiny is warranted in how post-tests define “success.” Do they require students to read the words automatically, or merely accurately? Are the words assessed in context or in isolation? Are students asked to read or spell them? These variations in study design lead to differing–and potentially conflicting–research conclusions.

In Sum

Current research suggests that once students have a foundation of GPCs and phonemic awareness, they should be ready for instruction in irregularly spelled words, but additional language proficiency may be needed for success in learning irregular function words.

While rote memorization is not harmful for students, children learn words more efficiently and effectively when they are able to analyze the spelling of the words more carefully: by spelling the words themselves, sounding out and correcting the mispronunciations, or, theoretically, by having a teacher draw their attention to the irregular portion(s) of the word’s spelling. The Heart Words routine, specifically, hasn’t yet been researched, but it does have a theoretical and empirical basis. 

Teachers should be mindful that the routine may not be sufficient when introducing more difficult function words, which may require additional repetition, contextualized examples, and/or practice writing in isolation or sentences to fully cement their spellings and meanings.

Questions we just don’t have the answers to:

1. What is the ideal ratio of time to spend on irregular word instruction?
2. How many words should be introduced at once?
   1. Does age impact this? Skill level?
3. Which instructional method is most effective for teaching automatic word reading? For spelling?
   1. How should strategies be adapted by grade, reading abilities, and/or English proficiency?
4. How exactly does the difference in learning content and function words impact how educators should introduce and/or practice these words with students?

Nitty-gritty of some of the research:

Colenbrander, et al. (2022)

Participants

• 85 Australian students (including ELs, with a wide range of decoding abilities) in the 3rd term of Kindergarten

Study Design

• Students were randomly assigned to one of three conditions, receiving 10-minute sessions over 3 days, learning 12 irregular words. They were exposed to the written form of the word three times, and heard the correct spoken form of the word 12 times across the lessons in all conditions. One group’s teacher modeled sounding out the words using GCPs (although not necessarily GPCs a kindergartner would have been expected to know), such as reading the word heart as “heert” and then flexing the sounds to the familiar word. A second group copied each word twice after seeing it, while a third group repeated the word orally after seeing it.

Findings

• Children showed evidence of superior learning of trained words after either sounding out and correcting the pronunciation or spelling the word, as compared to the group who only repeated the word. Differences between the first two groups were not significant.

Notes/Limitations

• The target words were not pretested, so we don’t know for sure that students didn’t know them prior to the training. The study used a very short training session, so we don’t know how increased duration and/or frequency of these instructional techniques would have impacted the results. We also don’t know that this amount or type of instruction leads to long-term retention of these words, as they were tested immediately after training.

McArthur et al., (2013)

Participants

• 104 struggling readers, age 7-12

Study Design

• Students were broken into three groups: receiving phonics instruction for 8 weeks followed by sight word instruction for 8 weeks; a reversed order of this same training (sight word instruction before phonics instruction); or mixed with both phonics and sight words for 16 weeks. Students received 5 x 30-min. sessions per week outside of school hours. During sight word instruction, students were trained in a list of 24 irregular words on flashcards presented by their parent, followed by a bingo game training software on the computer.

Findings

• Students who received instruction in phonics before sight word training made greater gains than those who received the reverse order of instruction.

Notes/Limitations

• Due to the study design, it is possible that parents practiced these (and potentially other high-frequency words) outside of the research sessions. There are numerous other uncontrolled variables as well, including the instruction that the students received at school and any additional intervention they may have received. Importantly, the results of this study were not replicated when attempted again in 2015.

Steacy, et al., (2020)

Participants

• 93, at-risk 1st-grade students

Study Design

• Students received 63 x 30- to 45-min. one-on-one decoding and fluency sessions, with high-frequency word activities taking 1-3 minutes per session.

Findings

• Shorter words, as well as those whose meanings students could picture in their minds (i.e. content words) were faster to be mastered than longer words. The imageability feature in particular seemed to be especially impactful for the students struggling the most

Notes/Limitations

• This study did not control for how regular or irregular the word’s spelling was, only that it was considered a high-frequency word.

Miles & Ehri (2017)

Participants

• 81 kindergarten students (half of whom were English Learners)

Study Design

• Students received 2 x 15-min. sessions on the same day (with a short break in between) to train and then test their learning of high-frequency words after flashcard instruction, both in isolation and in the context of sentences.

Findings

• Students were better able to read and spell words after practicing the words when presented in isolation. However, they were better able to use the target words in sentences when they had been trained with contextualized frameworks, meaning that both presentations may be necessary to allow students the full range of necessary word knowledge. It was also discovered that function words are more difficult for students to learn than content words, even if presumably easier to read. Miles et al., (2018) further analyzed this data to discover that students needed baseline knowledge of GPCs and phonemic awareness in order to successfully learn the spelling of irregular words, and also that function words were additionally impacted by language proficiency.

Notes/Limitations

• It is unknown if the study outcomes would have been different had more time been allocated for this instruction. Furthermore, due to testing being performed immediately after training the target words, we are not able to assume that students will retain their learning in the future.