Deep Dives

The Issue

Using decodable texts to reinforce phonics patterns and create a scaffolded entryway to reading for young students is a widely held practice. This makes good, logical sense and is supported by theoretical perspectives seeking to explain the role automaticity plays in reading fluency. According to this framework, educators should explicitly teach a skill and then provide opportunities for students to practice it until it becomes effortless and automatic so that their cognitive resources can be directed towards comprehending the text rather than trying to read each word (Kuhn & Stahl, 2003; Mesmer, 2009). However, the practice of using decodable texts to this end is primarily based on theory, rather than actual, empirical research (Mesmer, 2000). That isn’t to say that decodable texts are harmful or should be avoided, but it does mean that a component commonly built into explicit phonics instruction lacks the level of scientific evidence typically expected of practices labeled as part of the “Science of Reading.”

The Research

One problem with finding solid research on the use of decodable texts is the lack of studies that control for this component as an isolated, independent variable.

Meaning, while there might be research to show that decodable texts are a part of a package of instructional or intervention strategies that are effective, it can’t be determined that that package would have been any less effective if the use of decodable texts had been removed altogether or replaced with a different reading experience, such as with authentic texts (Jenkins et al., 2004; Price-Mohr & Price, 2018). 

Another issue with demonstrating the effectiveness of decodable texts is defining what exactly is meant by the term. While often referred to as if a binary feature of text (it’s either decodable or not), decodability can actually be thought of as a scalable feature. Meaning, a text might have perfect, 100% decodability, but many stories considered decodable texts have slightly less “control,” and instead insert less phonetic, high-frequency words (such as ‘was’), both so that students have practice with these important words, as well as allowing for slightly more naturalistic language to be used.

Many structured literacy lesson plans that use decodable texts are designed to do two things: prioritize the specific sound-letter correspondences currently being taught and reinforce previously learned correspondences through cumulative review. Therefore, a story might be “decodable” to one student, but not to another student who hasn’t been provided effective instruction in the correspondences incorporated in the texts.

Many studies, despite aiming to investigate the effectiveness of decodable texts, fail to provide both the level of control (percent of decodability) as well as the alignment of the incorporated correspondences to those that the student has learned in the texts used in the study (Beverly et al., 2009; Juel & Roper-Schneider, 1985). Thus, it is difficult to draw many conclusions from the available research without having a good understanding of exactly how decodable the “decodable texts” used were for the student participants.

Lastly, there is the question of: what effect are we hoping to see from the use of decodable texts? Do we want research to be able to demonstrate that their usage aids in reading words (nonsense or real) that are aligned with the phonics patterns they have received instruction on? Do we want or expect the gains to be transferred to real words as a whole (phonetic or not)? Do we want or expect practice with decodable texts to lead to increased oral reading accuracy? Fluency? Overall comprehension? Within decodable texts alone, or do we expect to see transference to authentic texts? The amount of valid research we have on decodable texts is further reduced by the myriad of different ways the researchers may think to measure apparent success/effectiveness.

In Sum

Overall, based on the research that we have, there might be some benefit to using decodable texts to reinforce and practice word-reading skills through the middle of first grade, but there does not appear to be evidence for their use beyond then. 

This might be an instructional strategy that is more beneficial for at-risk and struggling readers, but again, there is not sufficient evidence to make this claim outright. It is important to note that we have almost nothing to suggest that this is or isn’t an effective strategy for English Learners, as this population has been almost completely excluded from research in this area to date.

Questions we just don’t have the answers to:

1. Do all students benefit from decodable texts?
   1. At what grade(s)?
   2. Does this answer change depending on whether or not the students are at-risk for reading difficulties?
   3. Does this answer change depending on whether or not students are native English speakers?
2. Is there a difference in effectiveness in how controlled and aligned a decodable text is, meaning, how close in alignment with taught/mastered sound/letter correspondences, and what percentage of words are non-decodable?
3. Do the benefits of decodable texts outweigh the limitations (Castles et al., 2018) when considering:
   1. The statistical learning of high frequency words and frequency of grapheme-phoneme correspondences (Share, 1995)
   2. Motivating factors associated with authentic texts (Cole, 2002; Solity & Vousden, 2009)
   3. Restriction of vocabulary to words that adhere to students’ developing phonics knowledge (Price-Mohr & Price 2020)
   4. The contrived and non-naturalistic vocabulary and language patterns often used, as opposed to the more natural language used in more complex texts (Martens, 2016)?
4. Would another means of simplifying text (such as predictable texts, leveled readers, etc.) be as or more effective if paired with an otherwise effective, explicit phonics lesson plan?
5. Are we setting an unrealistic standard that students should expect consistent decodability by using highly decodable texts, when students will inevitably need to learn many irregular words due to English’s deep orthography (Gibson & Levin, 1975)?

Nitty-gritty of some of the research:

Jenkins et al., 2004

Participants

• 99, 1st graders identified as “at-risk”

Study Design

• Participants were broken into two intervention groups to receive tutoring 4 times per week for 25 weeks. One group (39 students) read texts with 85% lesson-to-text decodability; the second group (40 students) read texts with 11% lesson-to-text decodability; both groups were compared to students who received no text-reading tutoring (20 students), but all children were provided the same in-class phonics instruction.

Findings

• Both tutoring groups showed positive effects in the areas of: decoding, word reading, passage reading, and comprehension. But there weren’t statistically significant differences in how effective the varying levels of text decodability were. Meaning, the students benefit from the time reading connected texts, but it didn’t appear to matter how decodable those texts were.

Notes/Limitations

• As a Randomized Controlled Trial (RCT) that defined decodability, lesson-to-text percentage match, and isolated the decodability of texts used, this is arguably the best research that we have on the effects (or lack thereof) of using decodable texts.

Mesmer, 2009

Participants

• 74, 1st graders identified as lower-performing readers, specifically those who could successfully read at least 8 out of 10 words on a pre-primer word list. While this group represented students with limited early reading skills, the inclusion criteria excluded the most severely struggling readers— those unable to meet the 8-word benchmark.

Study Design

• Using two different core curricula depending on school, students read leveled readers and decodable texts twice at 3 different occasions (Oct, Jan, May). The same books were read at each of these 3 points to measure growth absent of any text variations. On each occasion, the students read, listened to a teacher model, then reread both the leveled and decodable text, while researchers tracked for accuracy and fluency. Students were assigned 1 of 2 different levels of difficulty for both the leveled reader and decodable text based on a “word list proxy for reading level” and a short spelling test to demonstrate phonics knowledge provided at the beginning of the school year.

Findings

• The study yielded inconclusive results about students’ accuracy reading decodable texts compared to authentic texts at different points during their 1st grade year, possibly due to limitations with the study design (further explanation below). The study did show, however, that participants read leveled texts more fluently than decodables, with and without practice, at all points of the year.

Notes/Limitations

• The students in this research study had to do relatively little to prove “mastery” of phonics knowledge for the purposes of this study. For example, correctly spelling 1 CVC word with the short /ă/ sound to demonstrate knowledge of this sound/letter correspondence. Furthermore, the researchers acknowledged that not all phonics patterns in the more difficult of the two decodable books were represented on the spelling pretest. Therefore, we are unable to say that the words in this “decodable” book were decodable to the students reading them, as the study lacked any verification that the students had ever been taught, much less learned, the patterns in this text. Furthermore, no details about the level of control (inclusion or exclusion of any non-decodable words), nor alignment of the targeted sound-letter correspondences to the phonics instruction (if any) provided in the classroom, were detailed in this study.

Beverly et al., 2009

Participants

• 32, 1st-grade students (14 of whom were considered at-risk)

Study Design

• Students were broken into 3 groups: 10 minutes of phonics instruction with 20 minutes reading decodable texts, phonics instruction with authentic texts read aloud (same time allocation), or 30 minutes of read-alouds only (without any phonics instruction). Intervention sessions occurred for 16 sessions during the second half of 1st grade (2x 30-min sessions for 8 weeks).

Findings

• All 3 groups showed gains as compared to the untreated group, with the decodables group benefiting the least.

Notes/Limitations

• As the study did not control for reading abilities, and it is unclear both what scope the phonics treatment covered as well as the scope of sound-letter correspondences captured in the decodables texts (and the percent of decodability), we don’t know the extent to which students benefited from this instruction (meaning: were the phonics skills covered too easy or too advanced?).

Price-Mohr & Price, 2018

Participants

• 12, 5-6 year-old children identified by teachers as falling behind peers

Study Design

• Students split into two groups: either reading high-decodable (87.8% decodable based on the scope and sequence used in classwide instruction) or low-decodable (35.8%) texts alongside analytic phonics instruction consistent among both intervention groups, during biweekly, hour-long tutoring sessions for 26 weeks, with all students also receiving synthetic explicit phonics instruction in the classroom

Findings

• The low-decodable group demonstrated more significant word-reading skill improvement from pre- to post-test than the high-decodable group. The low-decodable group also showed better reading comprehension skills than the high-decodable group.

Notes/Limitations

• As a very small study, caution should be used when interpreting the generalizability of this study. Also, as students were too young to perform the reading comprehension as a pre-test measure, it is noted that perceived growth in this component was based on students' performance on their pre-test early word reading abilities.

Price-Mohr & Price, 2020

Participants

• 34, 4-5 year-old children

Study Design

• Students were split into one of two intervention groups running for 90 minutes per week for approximately 36 weeks. The only difference between the teaching, classroom instruction, and intervention was the decodability of texts used in the intervention (87.8% versus 35.8% decodability based on the scope and sequence used in classwide instruction) across 2 parallel sets of 12 books.

Findings

• The low-decodable group demonstrated large effect sizes for reading comprehension improvement as compared to the high-decodable group.

Notes/Limitations

• This again is a very small study with rather young children. Also, as students were too young to perform the reading comprehension as a pre-test measure, it is noted that perceived growth in this component was based on students' performance on their pre-test early word reading abilities. There were no other statistically significant differences in the progress the groups made.

Juel & Roper-Schneider, 1985

Participants

• 93, 1st graders

Study Design

• Students received the same class-wide phonics instruction but varied in the type of texts read throughout the school year (decodable versus high-frequency texts).

Findings

• The students reading decodable texts outperformed those reading high-frequency texts in real and nonsense word reading up until February, but this difference was diminished by May, and didn’t translate into any statistically significant difference between the two groups at the end of the year with overall reading proficiency (based on Iowa Test of Basic Skills).

Notes/Limitations

• The study did not define to what extent the words in the decodable texts were aligned with the phonics instruction being provided in the class.