Examining the Effects of Tier 1 Small-Group Reading Instruction on Early-Elementary Students’ Reading Outcomes

Abstract

The purpose of this systematic observation study was to explore kindergarten and first-grade teachers’ Tier 1 small-group reading instruction for students with or at risk for reading difficulties and examine the effects of instruction on students’ literacy outcomes. We focused on time spent providing instruction in different literacy domains, instructional practices and materials used during text reading instruction, and impacts on student literacy outcomes. We conducted 42 observations of 15 early-elementary teachers’ small-group reading instruction for students with or at risk for reading difficulties. We also conducted semi-structured interviews with teachers to learn more about their instruction and collected student literacy assessment data at the beginning and end of the school year. Results indicated that teachers’ instruction typically aligned with evidence-based practices. Although instructional time was not significantly correlated with improvements in students’ skills for most expected relations, there were three statistically significant relations: phonological awareness instruction and encoding outcomes (r = .39; p < .01), encoding instruction and encoding outcomes (r = .30; p = .04), and comprehension instruction and expressive comprehension outcomes (r = −.32; p = .03). Further research investigating teachers’ instructional practices during small-group reading instruction and their impacts on students’ literacy outcomes is warranted.

Based on decades of research, there is broad consensus regarding the components of effective reading instruction for elementary students (Foorman et al., 2016; Shanahan et al., 2010). There is also a large body of rigorous observation research examining the degree to which evidence-based practices occur during instruction, especially during supplemental reading instruction (i.e., intervention) for upper-elementary students (e.g., Ciullo et al., 2019; Hall et al., 2022; Kent et al., 2017). However, there is a lack of research examining typical core reading instruction for early-elementary students, especially for students with or at risk for reading difficulties (RDs). Additionally, although previous observation studies have provided information on the broad content foci of reading instruction, most have not explored specific practices (e.g., strategies students are taught to use when reading words, or ways teachers scaffold students’ word reading efforts). This study aims to improve our understanding of the extent to which typical core reading instruction for early-elementary students with or at risk for RDs aligns with evidence-based practices, with a focus on text reading instruction.

Evidence-Based Reading Instruction

Research shows that the acquisition of early reading skills is crucial for later academic success (Schatschneider et al., 2004). Students with RDs are likely to experience difficulties throughout their academic careers; there is a higher probability they will drop out of school before high school graduation and a lower probability they will enroll in postsecondary education programs relative to their typically developing peers (Boscardin et al., 2008; Daniel et al., 2006). They are also at greater risk for emotional issues, such as anxiety and depression, and behavioral disorders (Dahle & Knivsberg, 2014; Mugnaini et al., 2009; Willcutt et al., 2010). Fortunately, early access to evidence-based core reading instruction and intensive interventions within multi-tiered systems of support (MTSS) can reduce the incidence and severity of RDs experienced by students (Al Otaiba et al., 2009; VanDerHeydan et al., 2007). In a three-tiered MTSS framework, Tier 1 involves universal core instruction (i.e., general classroom instruction) and Tiers 2 and 3 involve more intensive, targeted interventions. According to Al Otaiba et al. (2009), “The goal of Tier 1 is for all students to receive evidence-based and well-implemented reading instruction for about 90 minutes per day. Generally speaking, Tier 1 should help the majority of children read on grade level” (p. 14). As such, providing evidence-based Tier 1 reading instruction is vital not only to facilitate students' development of proficient reading but also to support their overall academic performance and well-being.

Decades of reading research have yielded broad consensus as to the skills that teachers should address to facilitate students' acquisition of reading skills (Castles et al., 2018; Foorman et al., 2016; Shanahan et al., 2010). Empirically validated theoretical frameworks of reading development, such as the simple view of reading (Gough & Tunmer, 1986) and the direct and indirect effects model of reading (Kim, 2017), posit that reading comprehension is the product of word recognition and language comprehension. Both components comprise multiple subcomponents that contribute to skilled reading and reading comprehension. Word recognition depends on phonological awareness (PA), phonics knowledge, orthographic knowledge, and decoding skill, whereas language comprehension depends on background and vocabulary knowledge, syntactical knowledge, text genre knowledge, and verbal reasoning skill (Scarborough, 2001). Research indicates that effective reading instruction addresses skills within both components (i.e., word recognition and language comprehension) as well as skills specific to reading comprehension (Forman et al., 2016; Shanahan et al., 2010). For example, recommendations from the Foorman et al. (2016) practice guide focused on developing word recognition skills, which includes teaching students to decode words and analyze word parts as well as building students’ language comprehension, including vocabulary knowledge.

In relation to text reading instruction, some specific evidence-based practices suggested by Foorman et al. (2016) include, “have students read decodable words in isolation and in text,” “teach students to blend letter sounds and sound-spelling patterns from left to right within a word to produce a recognizable pronunciation,” and “provide opportunities for oral reading practice with feedback to develop fluent and accurate reading with expression” (p. 2). These suggestions align with more recent research demonstrating that efficient word reading is accomplished by attending to graphemes and using knowledge of grapheme-phoneme correspondences (GPCs) to decode words (rather than by relying on memorization of whole words or guessing words based on context cues; Ehri, 2017, 2020). Relatedly, these practices also rely on the use of texts that feature words that students can decode using their knowledge of taught GPCs (i.e., decodable texts) rather than texts that follow a predictable pattern, such as each sentence starting with the same few words, for which words can be identified using context cues rather than phonics knowledge (i.e., predictable texts; Davis et al., 2021; Ehri, 2015).

Typical Reading Instruction

In the field of education, the often-discussed “research to practice gap” (e.g., Farley-Ripple et al., 2018; Joyce & Cartwright, 2020; Solari et al., 2020) implies that we know (a) what research says works and (b) what instruction looks like in practice (and that they do not match). However, this is not true for all contexts. Although there is consensus regarding the importance of providing students with evidence-based reading instruction and what constitutes evidence-based reading instruction, there is a lack of research examining typical instruction when it comes to early-elementary classroom teachers’ provision of Tier 1 reading instruction. Previous observation studies (e.g., Ciullo et al., 2019; Hall et al., 2022; Kent et al., 2017; Swanson & Vaughn, 2010; Swanson et al., 2012) have primarily described Tier 2, or supplemental, reading instruction provided by a reading specialist or special education teacher to students with RDs. In addition, they have mostly focused on supplemental instruction provided to students in the upper-elementary grades. In other words, rather than a “research to practice gap,” there is a gap in our knowledge related to the reading instruction that early-elementary classroom teachers implement, including the extent to which they use elements of reading instruction that research shows to be consistently associated with positive effects.

As a first step toward improving our understanding of typical-practice Tier 1 reading instruction, members of our research team recently conducted a systematic observation study, along with teacher interviews, focused on small-group reading instruction for kindergarten students at risk for RDs (i.e., students scoring at or below the 40th percentile on two measures of foundational reading skills; Dahl-Leonard et al., 2024). We conducted 31 observations across 11 kindergarten teachers for a total of 640 minutes of instruction. Observations took place during times when teachers indicated they would be providing a typical Tier 1 small-group reading lesson to participating students at risk for RDs.

Observations were video recorded and coded using a researcher-developed observation measure adapted from the Snapshot of Dyslexia Instruction (SDI) measure. The SDI measure was used in previous studies of reading instruction (Hall et al., 2022) and was itself adapted from the Non-Evaluative Snapshot of Literacy Instruction measure (Denton et al., 2021) and the Instructional Content Emphasis instrument (Edmonds & Briggs, 2003). In alignment with previous observation studies, we coded in one-minute intervals the amount of time dedicated to (a) each activity type (e.g., PA, phonics/decoding, vocabulary text reading), (b) instructional practices employed by the teacher (e.g., explanation or modeling, prompting students to use their knowledge of GPCs to decode words), and (c) materials used (e.g., flashcards, sound boxes). The first author, a researcher with extensive experience coding instructional observations, served as the gold standard for coding. Prior to coding, an undergraduate research assistant was trained and achieved 93% interrater agreement on a set of two independently coded observations. All observations were then coded by the first author with the undergraduate student double coding approximately 20% of observations. Coders achieved 94% overall agreement across double-coded observations (with 95% agreement within the activity type category, 92% agreement in the instructional practices category, and 95% agreement in the materials category).

We found that kindergarten teachers’ small-group reading instruction often reflected evidence-based recommendations regarding effective instructional practices (e.g., teachers frequently employed explicit instructional practices that included explanation and modeling as well as opportunities for student practice; Gersten et al., 2009; Vaughn et al., 2012). At the same time, teachers’ instruction also included practices that are associated with less evidence of effectiveness. For example, teachers often used predictable texts and supported students’ text reading by prompting students to use memorization of whole words or context cues rather than their knowledge of GPCs to read words (Ehri, 2017; Ehri, 2020; Foorman et al., 2016). That said, the study had some limitations. For example, our interview protocol lacked questions about teachers’ decisions around their use of instructional time as well as their use of specific practices and materials. Perhaps more importantly, we only focused on kindergarten classrooms and did not measure student outcomes. As such, we were unable to examine how teacher-level instructional practices impacted student-level outcomes.

Reading Curricula

Even when teachers are using an evidence-based reading curriculum, successful implementation can be challenging. Implementation science research has identified factors that influence the feasibility of implementing evidence-based reading practices in authentic school contexts (Damschroder et al., 2022). For example, insufficient training on the curriculum, lack of compatibility with existing instructional practices, and lack of alignment with goals for student performance can make it less likely that teachers will implement a curriculum as intended (Dahl-Leonard et al., 2025). It follows that research has often identified a lack of teacher adherence to reading curricula. For example, in a survey study on early-elementary teacher implementation of evidence-based reading practices in Tier 1, Kretlow and Helf (2013) found that fewer than half of teachers reported using all components of their core reading curricula daily. They also noted that teachers reported frequently supplanting their core reading curricula with other materials. As a result, even implementation of an evidence-based reading curriculum will look different for different teachers influenced by a different set of barriers and facilitators.

Study Purpose

The current study is designed to expand our knowledge of typical Tier 1 reading instruction for early-elementary students with or at risk for RDs through observations of instruction and interviews with teachers. It is intended to serve as a conceptual replication of our previous work, exploring the extent to which findings vary across different contexts (i.e., teachers providing instruction in different districts; teachers providing instruction in different grades). A report by the National Science Foundation and the Institute of Education Sciences (2018) acknowledges the importance of such replication studies: “Efforts to reproduce and replicate research findings are central to the accumulation of scientific knowledge that helps inform evidence-based decision making and policies” (p. 1). The current study also extends our previous work by addressing two key limitations (i.e., our previous study only examined instruction in kindergarten classrooms and did not analyze student outcomes). As such, in the current study, we explored the content of kindergarten and first-grade teachers’ Tier 1 small-group reading instruction for students with or at risk for RDs as well as the relation between instruction and students’ literacy outcomes. Our research questions were:

1. How much time during Tier 1 small-group reading instruction for early-elementary students with or at risk for RDs is dedicated to particular literacy skills? How do teachers describe the literacy skills addressed during this instruction?

2. During text reading instruction:

   1. To what extent do teachers instruct students to read words using (a) knowledge of GPCs, (b) memorization of whole words, and (c) context cues? How do teachers describe their word reading instruction?
   2. What types of texts (e.g., predictable texts, decodable texts) are used?
   3. To what extent are students reading text aloud?

3. To what extent does teacher time spent providing instruction in specific literacy domains predict student gains on theoretically-related literacy subtests?

Method

Study Context and Participants

This study was conducted in five schools in two Virginia school districts during the 2023–2024 academic year. One district was primarily White (58%), with 30% of students eligible for free or reduced-price lunch; the other district was 46% White and 31% Hispanic, with 45% of students eligible for free or reduced-price lunch. We recruited 15 general education teachers to participate in the study. Of these 15 teachers, seven taught kindergarten and eight taught first grade. Teacher demographics are provided in Table 1. Teachers in the state of Virginia are required to participate in training focused on evidence-based literacy instruction, such as Language Essentials for Teachers of Reading and Spelling (LETRS). LETRS training includes two volumes: one focused on word recognition (e.g., PA, phonics, decoding, encoding) and the other focused on language and reading comprehension (e.g., vocabulary, text-based comprehension instruction, writing). Virginia teachers are also required to implement a state-approved evidence-based core literacy curriculum in their classrooms. Curricula are evaluated in seven areas: phonological and phonemic awareness, phonics and word study, vocabulary, text reading and fluency, developing comprehension and background knowledge, small group instruction and independent practice, and writing. Examples of state-approved curricula include Open Court Reading, Benchmark Advance, and Into Reading.

Table 1.

Teacher Demographic Information

	N	%
Gender
Female	14	93.33
Male	1	6.67
Race
White	15	100.00
Highest Level of Education
Associate’s Degree	1	6.67
Bachelor’s Degree	8	53.33
Master’s Degree or Beyond	6	40.00
Certification Type
Regular	13	86.67
Temporary	2	13.33
Grade Level Taught
Kindergarten	7	46.67
Grade 1	8	53.33
Years of Teaching Experience
1–5	8	53.33
6–10	3	20.00
11–15	2	13.33
16–20	2	13.33

Note. Total number of teachers = 15.

One of the districts used a tiering system based on students' reading abilities to assign students to classes. Of the ten participating teachers from this district, two had a Tier 1 class (i.e., students reading at or above grade level), three had a Tier 2 class (i.e., approaching grade-level reading proficiency), and five had a Tier 3 class (i.e., reading below grade level). In all instances, teachers were asked to allow us to observe their lowest-performing reading group. The students in this group were not required to have a formal identification (e.g., dyslexia, learning disability in reading) to participate in the study. Rather, this group of students were deemed by their teacher to be with or at risk for RDs relative to the rest of the students in their class. Consent forms were sent home with this group of students. The forms provided parents with the option to agree to (a) have their child be video recorded during their teacher’s small-group reading instruction and (b) have their child’s state literacy assessment data accessed by the research team. In other words, parents could not provide consent, provide consent for one activity or the other, or provide consent for both activities. We received consent to video record 60 students and to access assessment data for 54 students. Student demographic information is provided in Table 2.

Table 2.

Student Demographic Information

	Kindergarten		Grade 1
	N	%	N	%
Gender
Female	19	63.33	14	58.33
Male	11	36.67	10	41.67
Hispanic or Latino
Yes	5	16.67	7	29.17
No	25	83.33	17	70.83
Race
White	18	60.00	15	62.50
Black/African American	4	13.33	4	16.67
Asian	4	13.33	1	4.17
American Indian/Alaska Native	3	10.00	2	8.33
Black/African American and White	1	3.33	1	4.17
Native Hawaiian/Other Pacific Islander	-	-	1	4.17
Native English-Speaker
Yes	17	56.67	15	62.50
No	10	33.33	3	12.50
Unspecified	3	10.00	6	25.00
Services Received
English Language Service	3	10.00	2	8.33
Title I Service	4	13.33	3	12.50
EIRI Service	-	-	1	4.17
Other Services Not Listed	2	6.67	3	12.50
No Service	21	70.00	15	62.50

Note. Total number of kindergarten students = 30; Total number of Grade 1 students = 24; EIRI = Early Intervention Reading Initiative.

Data Collection and Analysis

Observations of Instruction

The video recorded observations of instruction were scheduled at times when teachers indicated they would be providing a typical Tier 1 small-group reading lesson to the target group of students. The entire small-group reading lesson was video recorded by a research assistant. We aimed to video record three observations of small-group reading instruction per teacher (i.e., one in the fall, one in the winter, and one in the spring). However, due to scheduling conflicts and time constraints, one of the teachers was only video recorded on two occasions. Additionally, two of our video files were unable to be coded due to technical issues (e.g., no audio). Thus, we analyzed a total of 42 video recorded observations.

All observations were video recorded and the video observations were coded using a researcher-developed observation measure. The measure was adapted from other observation tools (Denton et al., 2021; Edmonds & Briggs, 2003; Hall et al., 2022). For example, Hall et al., (2022) used the SDI measure to identify similarities and differences between the two reading interventions across five categories (i.e., literacy domain, format of instruction, rules versus associations, procedures for encoding/decoding, and materials) with an overall inter-rater agreement of 83%. Data obtained from the measure was paired with data obtained from reviewing intervention materials and from a teacher-completed survey about the intervention. Results from the observations, material review, and teacher surveys were well aligned, providing construct validity evidence for the observation measure. A revised version of this measure was then used in the previously-described study of small-group reading instruction (Dahl-Leonard et al., 2024). The primary revisions to the measure were the removal of categories that were not of interest (i.e., rules versus associations and procedures for encoding/decoding) and the addition of variables related to a category of interest (i.e., text reading instruction). This measure demonstrated an inter-rater agreement of 94%. Additionally, Dahl-Leonard et al. (2024) conducted interviews with teachers to learn more about their instruction and compared those findings to the observations. Results indicated alignment between the two, suggesting evidence of construct validity for the observation measure.

Like in Dahl-Leonard et al. (2024), in the current study, we coded in one-minute intervals the amount of time dedicated to each literacy domain (e.g., PA, phonics/decoding, vocabulary, text reading). Because we were particularly interested in understanding more about the ways teachers provided instruction and scaffolding during text reading instruction, we also coded for variables specially related to text reading instruction (i.e., use of knowledge of GPCs, memorization of whole words, context cues, reader). Coded variables were not mutually exclusive. More information about the definitions used to code each variable is provided in Supplemental Appendix A.

The first author, a researcher with extensive experience coding instructional observations, served as the gold standard for all coding. Prior to coding for literacy domains, three research assistants participated in a one-hour training, which included review of the variables, instructions on coding procedures, and practice coding the video observations. For example, coders were trained to watch each video in one-minute segments, pausing to mark the appropriate codes and re-watching as needed to ensure accuracy. As additional examples of coding procedures, coders were also trained that (a) more than one literacy domain may be marked as present during a one-minute interval and (b) at least one literacy domain or “not literacy” should be marked for each one-minute interval.

After training, coders achieved ≥90% interrater agreement on an independently coded video observation. All observations were then independently coded with approximately one-third of observations being double coded. For literacy domains, coders achieved 97% overall agreement across double-coded observations. Prior to coding for variables specially related to text reading instruction, one research assistant attended a one-hour training, including review of the variables and practice coding. The research assistant then achieved 96% interrater agreement on a set of two independently coded observations of text reading instruction. Approximately one-third of the observations with text reading instruction were double coded, with coders achieving 97% overall agreement for text reading variables across double-coded observations. All disagreements in coding were resolved via discussion and consensus.

All observation coding data was entered into Excel spreadsheets. Initially, data for each observation was entered in a separate tab, with each row representing a minute of instruction for that observation; all data was combined into a single spreadsheet for analysis. We calculated descriptive statistics for the lesson duration (e.g., total lesson length for each observation, average lesson length across observations). We also calculated descriptive statistics to determine the number of minutes and percentage of time that each literacy domain was observed. Finally, we calculated descriptive statistics to determine the number of minutes and percentage of time specific practices were present during text reading instruction.

Teacher Interviews

Semi-structured interviews are commonly used as a flexible approach to gather in-depth information about participants’ experiences and perspectives (Creswell & Poth, 2017). All 15 teachers participated in semi-structured virtual interviews conducted by the first author at the end of the academic year (i.e., after the observations of instruction had occurred). The interview questions analyzed for the present study were primarily related to their typical reading instructional practices (see Supplemental Appendix B). For example, we asked questions such as, “What does a typical small-group reading lesson look like?” and “How do you decide what skills to target during small-group reading instruction?” If needed, additional probes (e.g., “Can you tell me more about that?”) were used to encourage teachers to elaborate on their responses.

The teacher interviews were audio recorded and transcribed. After the interviews were transcribed, the first author and a research assistant independently read the transcripts and conducted inductive thematic analysis to identify patterns in the teachers’ responses (Braun & Clarke, 2006; Miles et al., 2014). This process involved generating initial codes and identifying themes across responses to each interview question. For example, when looking across teachers’ responses to the question, “What program do you use for your small group reading instruction?,” we identified that most teachers used one of three programs. At the same time, we noticed that some teachers also described supplementing the program with other resources. As such, we were able to check this potential theme across the rest of the interview transcripts to assess its consistency and relevance. To enhance the credibility and dependability of our analysis, after independent coding, the coders engaged in iterative discussions to refine themes and resolve discrepancies until consensus was reached (Braun & Clarke, 2006; Miles et al., 2014). This systematic coding procedure supported a rigorous examination of teachers’ perceptions regarding their small-group reading instruction and allowed us to explore how these perceptions aligned with our observations of their instruction.

Student Assessments

The Virginia Language and Literacy Screening System (VALLSS; 2024) is a literacy screening assessment designed to identify students at risk for RDs and provide teachers with information they can use to inform their instruction. VALLSS involves multiple subtests focused on word recognition and language comprehension skills. It is administered to students in kindergarten through third grade in the state of Virginia in fall and spring, with the use of specific subtests varying by grade. For the purpose of this study, we focused on the following subtests because they were administered to both kindergarten and first-grade students: Phoneme Segmenting, Letter Sounds, Real Word Decoding, Pseudoword Decoding, Encoding, Vocabulary Fluency, Passage Retell, Expressive Comprehension, and Nonsense Sentences. See Supplemental Appendix C for more information about each subtest. More specifically, we focused on subtests measuring constructs that theory suggests would improve as a result of instruction in the predictor domain. We explored the following relations:

• Does teacher time spent providing PA instruction predict student performance on the Phoneme Segmenting, Letter Sounds, Real Word Decoding, Pseudoword Decoding, or Encoding subtest?

• Does teacher time spent providing phonics/decoding instruction predict student performance on the Letter Sounds, Real Word Decoding, or Pseudoword Decoding subtest?

• Does teacher time spent providing encoding instruction predict student performance on the Encoding subtest?

• Does teacher time spent providing vocabulary instruction predict student performance on the Vocabulary Fluency subtest?

• Does teacher time spent providing comprehension instruction predict student performance on the Passage Retell, Expressive Comprehension, or Nonsense Sentences subtest?

We calculated the difference between fall and spring scores for each subtest to explore the correlations between the time spent providing instruction within a given literacy domain and score change on a given subtest. For significant correlations, we also used a linear regression analysis to explore the extent to which time spent on a specific literacy domain could predict students’ score change on a specific subtest. We also conducted exploratory hierarchical linear modeling (HLM) analyses to reflect the nested design of students being clustered into classrooms (i.e., by teacher). By separating the clustering effect, this model determines whether the impact of instructional time on students’ score change is influenced by students being taught by a specific teacher in a specific classroom.

Results

Across 42 observations of 15 classroom teachers, we observed a total of 612 minutes of Tier 1 small-group reading instruction for kindergarten and first-grade students identified by their teacher as being with or at risk for RDs relative to the rest of the students in their class. The average observation length was 14.57 minutes (SD = 5.21; range: 5–26).

Literacy Domains

As shown in Table 3, the observed instructional time was primarily devoted to phonics/decoding (59.48%), with all 15 teachers engaging in this literacy domain during our observations. The next most commonly-addressed literacy domains were text reading (29.90%) and encoding (29.41%). That said, only nine teachers addressed text reading (i.e., six did not engage students in text reading during any observed instruction) and 12 teachers addressed encoding (i.e., three did not engage students in encoding work during any observed instruction). All 15 teachers engaged in vocabulary instruction, which occurred during 18.14% of observed instructional time. Instruction involving high-frequency words occurred during 10.95% of observed time. Less than 10% of time was spent on comprehension, PA, writing, print concepts, handwriting, and letter-name knowledge (7.19%, 6.86%, 5.07%, 4.90%, 4.74%, and 3.27%, respectively). Lastly, 4.90% of the observed time was not spent on reading instruction.

Table 3.

Literacy Domain Information

Literacy Domain	Number of Minutes	Percentage of Total Minutes	Number of Teachers	Percentage of Total Teachers
Phonics/decoding	364	59.48	15	100.00
Text reading	183	29.90	9	60.00
Encoding	180	29.41	12	80.00
Vocabulary	111	18.14	15	100.00
High-frequency words	67	10.95	12	80.00
Comprehension	44	7.19	8	53.33
Phonological awareness	42	6.86	7	46.67
Writing	31	5.07	4	26.67
Print concepts	30	4.90	6	40.00
Handwriting	29	4.74	7	46.67
Letter-name knowledge	20	3.27	7	46.67
Not literacy	30	4.90	11	73.33

Note. Total minutes = 612 minutes. Total teachers = 15 teachers.

Interviews with teachers revealed that all teachers implemented a reading program during their small-group instruction: eight used Reading Mastery, four used University of Florida Literacy Institute, and three used Being a Reader. Six teachers noted that they supplement their main program with additional resources (e.g., decodable books from Reading A-Z). When asked about the skills they address during their small group reading instruction, teachers frequently reported addressing word recognition skills, such as PA, phonics, decoding, and encoding. One teacher noted that she provides instruction in more language-comprehension-focused skills, such as vocabulary and written composition, during whole-class instructional time and that her small groups focus more on phonics. About one-third of teachers noted addressing high-frequency words. Only one teacher described including vocabulary instruction during her work with small groups. She said, “sometimes I still need to supplement with things that I know my kids need. For example, I have a large English learner population, and a lot of them speak Spanish and English, so I feel like a lot of times I have to add in vocabulary concepts.” None of the 15 teachers mentioned providing reading comprehension instruction in small groups. All 15 teachers mentioned that text reading was typically addressed during their small-group reading instruction.

Text Reading Instruction

A total of 183 minutes of text reading instruction were observed across nine teachers. As shown in Table 4, teachers instructed students to use their knowledge of GPCs to decode words during 50.27% of text reading instruction. During 9.27% of observed text reading instruction, teachers instructed students to read words via memorization of whole words. In our observations, teachers never instructed students to use context cues to identify a word they were trying to read. When asked during interviews about how they scaffold student word reading when a student has trouble reading a word, all 15 teachers mentioned that they would prompt students to use their knowledge of GPCs to sound out the word. About half of the teachers described specific steps, such as “read the first sound, read the next sound, read the last sound, blend the sounds together, now read it fast.” A couple teachers noted that the reading program they use instructs teachers to simply tell the student the word. One of these teachers said, “I like the sounding it out better, because then I can hear if they're sounding at it out correctly.”

Table 4.

Text Reading Instruction Information

	Number of Minutes	Percentage of Total Minutes	Number of Teachers	Percentage of Total Teachers
Practice
Knowledge of GPCs	92	50.27	9	100.00
Memorization	17	9.27	7	77.78
Context cues	0	0.00	0	0.00
Text Type
Non-predictable text	183	100.00	9	100.00
Predictable text	0	0.00	0	0.00
Reader
Student(s) only	145	79.23	8	88.89
Student(s) and teacher together	29	15.85	7	77.78
Teacher only	9	4.92	3	33.33

Note. Total minutes = 183 minutes. Total teachers = 9 teachers. GPCs = grapheme-phoneme correspondences.

We also did not observe any reading of predictable texts. During teacher interviews, 14 of the 15 teachers indicated that they used decodable texts during their small-group instruction. The one teacher who did not mention using decodable texts only mentioned using sentences, usually with high-frequency words, during her small-group reading instruction. Most text reading instruction we observed involved student reading of text, with students reading independently without teacher assistance (79.23%) or with students and teacher reading together (15.85%). Less than 5% of observed text reading time involved only the teacher reading (4.92%).

Student Literacy Outcomes

Supplemental Appendix D shows descriptive statistics for students’ scores from the VALLSS subtests. The magnitude of score changes from Fall to Spring ranged from an average of 0.00 points (Passage Retell) to 9.40 points (Letter Sounds). Table 5 shows the summary of correlations between instructional time in specific literacy domains and VALLSS subtest score changes from Fall and Spring. Note that the sample sizes used for correlation analyses are smaller than the total student sample because some students did not have scores for the spring semester and were excluded from the analysis. For most hypothesized relations, instructional time was not significantly correlated with students’ score changes. However, a few results showed significant correlations: (1) PA instruction and Encoding subtest scores, (2) encoding instruction and Encoding subtest scores, and (3) comprehension instruction and Expressive Comprehension subtest scores. As shown in Supplemental Appendix E, HLM analyses with students nested in classrooms (i.e., by teacher) revealed only one significant association between the instructional time and students’ score changes: comprehension instruction and Expressive Comprehension subtest scores.

Table 5.

Correlation Results

Literacy Domain	VALLSS Subtest	N	Correlation	p
Phonological awareness	Phoneme Segmenting	49	−.01	.92
	Letter Sounds	26	.23	.26
	Real Word Decoding	49	−.09	.52
	Pseudoword Decoding	49	−.07	.66
	Encoding	49	.39*	.005
Phonics/decoding	Letter Sounds	26	.15	.45
	Real Word Decoding	50	−.05	.74
	Pseudoword Decoding	50	−.15	.29
Encoding	Encoding	49	.30*	.04
Vocabulary	Vocabulary Fluency	48	.01	.97
Comprehension	Passage Retell	20	.34	.14
	Expressive Comprehension	45	−.32*	.03
	Nonsense Sentences	50	−.22	.13

PA and Encoding

The correlation analysis showed a significant, positive correlation between the time spent on PA instruction and students’ score change on the Encoding subtest (r = .39; p = .005). The regression analysis showed that one minute increase in PA instruction could predict 0.31 points increase in Encoding scores, where about 14% of variance in Encoding score change can be explained by instructional time in this domain (y = 0.31x + 4.31; R² = .14).

Encoding and Encoding

The correlation analysis showed a significant, positive correlation between the time spent on encoding instruction and students’ score change on the Encoding subtest (r = .30; p = .04). The regression analysis showed that one minute increase in encoding instruction could predict 0.09 points increase in Encoding scores, where about 7% of variance in score change can be explained by instructional time in this domain (y = 0.09x + 4.27; R² = .07).

Comprehension and Expressive Comprehension

Different from previous relations, the correlation analysis showed that the time spent on comprehension instruction had a significant, negative correlation with students’ score change on the Expressive Comprehension subtest (r = −.32; p = .03). The regression analysis showed that one minute increase in comprehension instruction could predict 0.14 points decrease in Expressive Comprehension scores, where about 8% of variance in score change can be explained by instruction time (y = −0.14x + 1.35; R² = .08). HLM results similarly revealed a significant, negative association between the comprehension instructional time and students’ scores for the Expressive Comprehension subtest (r = −.14; p = .02), indicating that gains in Expressive Comprehension were somewhat smaller in classrooms with more comprehension instructional time. The random effects model indicated substantial variance at the classroom/teacher level, with an intraclass correlation coefficient (ICC) of .61, indicating that students being in the same classroom can explain about 61% of the variance in change of students’ Expressive Comprehension scores across semesters.

Discussion

This study served as a conceptual replication and extension of our previous systematic observation study on small-group reading instruction for kindergarten students with or at risk for RDs (Dahl-Leonard et al., 2024). More specifically, the purpose of this study was to explore kindergarten and first-grade teachers’ Tier 1, small-group reading instruction for students determined by their teacher to be with or at risk for RDs and explore the effects of instruction on students’ literacy outcomes. We focused on instructional time within different literacy domains, instructional practices and materials used during text reading instruction, and impacts on students’ literacy outcomes.

Literacy Domains

Our observations revealed that word recognition skills were the primary focus of small-group instruction, with over half of instructional time devoted to the phonics/decoding domain. Instruction involving encoding and text reading was also common during our observations. For the most part, these observational findings align with teachers’ descriptions of their small-group reading instruction. Given that early-elementary students with or at risk for RDs tend to have difficulty with word recognition skills (Jones et al., 2016), and given that these skills are often best addressed in small groups, it is perhaps unsurprising that teachers’ instruction focused on building phonics knowledge, decoding/encoding, and text reading skills (Ehri, 2020; Foorman et al., 2016). That said, it is important to note that many students with or at risk for RDs also have difficulty with language comprehension and may benefit from language-focused (e.g., vocabulary) instruction. At the same time, research does not furnish clear answers when it comes to the degree to which these skills are best addressed during whole-class versus small-group instruction.

One discrepancy between our observations and interviews with teachers was related to PA. Instruction in PA was frequently mentioned during interviews but not frequently observed during instruction. This difference is likely due to how we operationally defined PA. We only coded for the presence of PA instruction in instances in which students were expected to identify, blend, segment, or manipulate sounds in words without reference to print. As such, instances in which students were segmenting or blending phonemes while associating them with graphemes were not coded as PA instruction. Because there is some evidence that it is more beneficial for students to work on PA in the context of activities that provide opportunities to associate phonemes with graphemes than to do so during purely auditory activities (Clemens et al., 2021), it is not problematic that PA instruction for students typically also involved mapping phonemes to graphemes.

Another noteworthy discrepancy between our observations and interviews was related to comprehension instruction. Although none of the teachers mentioned providing comprehension instruction during their small-group time, we observed eight teachers address comprehension. Descriptively speaking, the instances of comprehension instruction we observed were often brief and not necessarily the primary focus of instruction. For example, before reading, a teacher would occasionally provide a purpose for reading (e.g., “We need to see if Walter kicks the ball today” or “We’re going to see what the goats do”); during reading, a teacher would occasionally check for students’ understanding (e.g., “Why were they saying don’t let Walter kick?” or “Were the goats supposed to jump out and start eating everything?”). We did not see any instances of comprehension strategy instruction. It is possible that, when describing the skills addressed during their small-group reading instruction, teachers did not consider comprehension to be a focal skill and, as such, did not report providing instruction in this domain.

Comprehension instruction was provided more frequently in our previous observation study (Dahl-Leonard et al., 2024), with 17.03% of instructional time involving comprehension instruction (compared to 7.19% in the current study; see Supplemental Appendix F for more comparisons between the two studies). Given that research suggests the importance of addressing language and reading comprehension in the early grades (Foorman et al., 2016; Shanahan et al., 2010), it is interesting that teachers in the current study chose not to spend more time addressing comprehension. Above, we discuss reasons why small-group instruction might typically focus more on word recognition skills than on language comprehension. Still, it is unclear why teachers in the current study spent so many fewer instructional minutes on comprehension than did teachers in the previous study. Another area of literacy instruction that was addressed more frequently in our previous study was print concepts, with 30.31% of instructional time devoted to this domain in the previous study and 4.90% of instructional time devoted to it in the current study. It makes sense that teachers in the current study did not spend much time on print concepts, which are typically considered pre-kindergarten skills (i.e., one mastered by the majority of students before entering kindergarten; Justice et al., 2006).

Two other literacy skills were addressed more frequently in the current study than in our previous study: encoding and vocabulary. Encoding was addressed during 29.41% of instructional time in the current study and 13.44% in the previous study. In the current study, teachers described their curricula as being more phonics-based than teachers did in the previous study. As such, it follows that their curricula addressed both decoding and encoding. That the current study included first-grade teachers in addition to kindergarten teachers may also provide some explanation. Teachers of first-grade students may spend more time on encoding than do kindergarten teachers, who, especially with their students with or at risk for RDs, may target simpler skills that are considered to be prerequisites for word reading and spelling, such as developing PA and knowledge of GPCs (Ehri, 2020; Foorman et al., 2016).

Vocabulary, which is a language comprehension skill, was addressed during 18.14% of instructional time in the current study and only 2.34% in the previous study. This increased focus on vocabulary instruction reflects evidence-based practice, as extensive evidence demonstrates the importance of vocabulary development for later reading achievement (Oakhill & Cain, 2012; Ouellette, 2006). Based on their large-scale meta-analysis of reading research on skills to support reading comprehension for students in kindergarten through Grade 3, Foorman et al. (2016) recommend to “teach academic vocabulary in the context of other reading activities.”

Overall, the literacy domains we observed teachers addressing during instruction in the current study primarily align with evidence-based recommendations for providing literacy instruction to early-elementary students with or at risk for RDs. This finding was somewhat different from our previous work, but not unexpected given the different context. More specifically, teachers in the current study were teaching in a state that required teachers to (a) participate in training on evidence-based literacy instruction and (b) implement an evidence-based state-approved core literacy curriculum. Thus, it follows that their instructional time was spent on these areas.

Text Reading Instruction

Although during interviews all 15 teachers mentioned including text reading instruction as part of their small-group reading instruction, we only observed nine teachers provide text reading instruction. It is unclear whether this lack of text reading instruction is related to teachers following a curriculum that does not emphasize reading texts or because teachers are choosing not to read texts during their small-group time. Regardless, the lack of text reading instruction in several classrooms is concerning. Providing students with opportunities to practice reading connected text is important for the development of accurate, fluent reading as well as the development of reading comprehension (Foorman et al., 2016; Shanahan et al., 2010).

During the text reading instruction we observed in the current study, teachers did not use predictable texts. Teachers’ interview responses corroborated this finding. When asked about the types of texts they used during text reading instruction, almost all teachers stated that they used decodable texts. Because decodable texts are designed to provide practice for students in using taught GPCs to sound out words, it is not surprising that teachers were frequently observed instructing students to use their knowledge of GPCs to decode words. They occasionally instructed students to use memorization of intact words while reading (e.g., in the case of a high-frequency word for which students had not yet learned a constituent GPC) but never prompted students to use context cues to identify a word they were trying to read. In other words, teachers provided students with opportunities to practice reading decodable words in connected text, an evidence-based instructional practice (Foorman et al., 2016). These findings are very different from our previous findings. In our previous study (Dahl-Leonard et al., 2024), the majority of text reading instruction involved reading predictable texts (77.16%) and frequently included prompts to use memorization of intact words while reading (95.26%). See Supplemental Appendix G for more comparisons between text reading instruction in the two studies.

The dramatic differences in the text reading instruction observed in each study are, upon further reflection, not as surprising as they may initially seem. Each study took place in a different state, and, in the United States (U.S.), individual states are given wide leeway to establish curriculum standards, manage teacher certification, and oversee the operation of school districts. Different U.S. states have in place different laws to ensure that teachers are trained to provide evidence-based reading instruction (Neuman et al., 2023). At the time when the original study was conducted, only 9% of participating teachers had completed state-mandated trainings to support their knowledge and implementation of evidence-based reading practices (i.e., Texas Reading Academies). In the current study, conducted in Virginia, 93% of participating teachers reported receiving district-provided training in reading instruction and/or other training focused on evidence-based reading instruction (e.g., LETRS). It is perhaps not surprising then that when the kindergarten and first-grade teachers in the current study provided text reading instruction to their students who they believed were with or at risk for RDs, it was better aligned with evidence-based practices.

Student Literacy Outcomes

In this study, instructional emphasis (i.e., time spent providing instruction within a particular literacy domain) was statistically significantly correlated with student literacy gains in three of the 13 associations we examined. However, two of these associations (i.e., PA instruction with encoding outcomes and encoding instruction with encoding outcomes) were not statistically significant when accounting for nesting within classrooms (i.e., by teacher).

That there was a significant, positive correlation between time spent on PA instruction and students’ gains in encoding (when not accounting for nesting within classrooms) seems, at first glance, as though it would be straightforward to explain. PA is critical for encoding. To encode a word successfully, a student must first segment the word into constituent phonemes and then associate each phoneme with a grapheme. Students with better PA are generally better able to develop knowledge of GPCs, although the relations between PA and phonics knowledge are bidirectional (Foy & Mann, 2006). It becomes more complicated to explain this positive correlation between time spent on PA instruction and gains in encoding when reflecting on the operational definition for PA instruction used in this study. As noted previously, an activity coded as PA instruction needed to be purely auditory (i.e., not involve reference to printed letters). This operational definition of PA instruction matters, because there is research to suggest it is more efficient to provide instruction in PA alongside of phonics instruction (i.e., to reference printed letters while helping students segment and blend sounds in spoken language; Clemens et al., 2021). For example, Stalega et al. (2024) found that purely auditory PA instruction had a smaller effect on reading-related outcomes than instruction that included a phonics component. In a meta-analysis of reading intervention research for students in Grades 1–3 (Gersten et al., 2020), interventions that included a stand-alone PA instructional component tended to result in smaller effects on reading fluency outcomes than interventions that did not include a stand-alone PA component. When considering this prior research, we cannot fully explain why it was time spent on PA instruction, rather than phonics/decoding instruction (i.e., instruction that built PA while referencing print), was the significant predictor of encoding gains.

There was also a significant positive correlation between time spent on encoding instruction and students’ gains in encoding (when not accounting for nesting within classrooms). It is logical that providing encoding instruction leads to gains in encoding. Still, encoding instruction sometimes falls by the wayside in early-elementary classrooms, at least partially as a result of the fact that state screening assessments tend to prioritize early reading skills over early encoding skills. It is important, then, that this study suggests that there may be some value in providing encoding instruction to support early encoding skills. In addition, although the current study did not find a statistically significant positive association between encoding instruction and gains in decoding, meta-analyses of research on early reading interventions suggest that including a encoding instructional component tends to improve outcomes on measures of early reading as well as on encoding measures (Gersten et al., 2020; Hall et al., 2023).

The statistically significant, negative association between time spent providing comprehension instruction and students’ gains on the VALLSS Expressive Comprehension subtest (when accounting for nesting within classrooms) is, perhaps, the most difficult association to explain. The VALLSS Expressive Comprehension subtest measures students’ ability to accurately answer, using oral language, a series of orally-presented literal and inferential questions about a passage read aloud. In this study, we defined comprehension instruction as any instruction that that supported students in interpreting the meaning of sentences or longer text segments and/or practicing strategies and skills that aim to improve understanding and recall of text. In practice, comprehension instruction often entailed teachers providing a purpose for reading before or checking for student understanding (e.g., asking questions or having a discussion about what was read) during or after students read decodable text with teacher support. As such, there was some misalignment between the comprehension instruction we observed (which supported understanding of student-read decodable text) and the Expressive Comprehension task (which tested understanding of a more sophisticated text read aloud by an adult). This may explain why instructional time in this area did not improve students’ Expressive Comprehension scores. Engaging students in read-alouds and discussion of content-knowledge-rich texts may be more likely to lead to improvement in the skills measured in the Expressive Comprehension subtest than asking and answering questions in the context of simple, decodable texts.

As mentioned previously, empirically-validated theories of reading comprehension (e.g., Gough & Tunmer, 1986; Kim et al., 2017) and comprehensive reviews of reading research (e.g., Foorman et al., 2016; Shanahan et al., 2010) support the provision of both word recognition and language comprehension instruction in early-elementary classrooms. Evidence is strangely mixed, though, as to the relative effects, in the short term, of instruction only focused on word recognition compared to instruction that addresses both word recognition and language comprehension. For example, in a meta-analysis of Tier 2 reading interventions for students in Grades K-3, Wanzek et al. (2016) investigated whether intervention effects on standardized measures of language and comprehension were moderated by the emphasis of the intervention (i.e., comparing word recognition-focused and multicomponent interventions). The difference in mean effect size for word recognition interventions and multicomponent interventions was not statistically significant. In a meta-analysis examining the effects of word recognition and multicomponent reading interventions on reading comprehension outcomes for K-3 students with or at risk for RDs, a sort of replication of the Wanzek et al. (2016) moderator analysis, Denton et al. (2022) again found that effects did not significantly differ for interventions focused only on word recognition skills and those that provided both word recognition and comprehension instruction. These findings suggest that, at least for early-elementary students with or at risk for RDs, providing comprehension instruction in addition to word recognition instruction did not improve short-term reading comprehension outcomes beyond providing word recognition instruction alone. Our finding that comprehension instruction is not positively associated with score increase on a measure of listening comprehension supports the same potential conclusion: For early elementary students with or at risk for RDs, providing comprehension instruction may not be the most valuable use of instructional time.

That said, Denton et al. (2022) interpreted their finding with caution. They pointed out that, even if their findings were true when it came to reading comprehension measured in the primary grades, it was important to consider long-term effects of comprehension instruction. Cross-sectional research demonstrates that, although word recognition tends to make a larger contribution to reading comprehension for younger readers than it does for older readers (e.g., García & Cain, 2014; Keenan et al., 2008), language comprehension takes on increased importance when readers progress beyond the primary grades (e.g., Tighe & Schatschneider, 2014). The word and world knowledge that is crucial for students' comprehension in the upper elementary and secondary grades must be built over time. Primary-grade instructional components emphasizing language comprehension may yield benefits that are realized only in later grades.

Limitations, Future Directions, and Implications

This study has a few limitations. First, despite our efforts to recruit a larger number of participants, our sample of kindergarten and first-grade teachers and students was relatively small. As such, the teacher and student data we gathered may not be representative of this population, and the generalizability of our findings may be limited to teachers and students with similar characteristics. The small sample size may have also limited our ability to detect statistically significant relations between instructional time and outcomes (e.g., given our small sample size, there is a greater possibility of Type 2 error, or failure to reject a false null hypothesis). Relatedly, due to our small sample, we were not able to conduct additional analyses, such as examining differences between kindergarten and first grade. Future systematic observation studies of this sort are encouraged to involve larger, more representative samples of participants and conduct more in-depth analyses.

Additionally, the primary data collection method utilized in this study was video observations, which has strengths and weaknesses. We were able to pause and re-watch the videos, which allowed us to carefully code each minute of instruction for several variables. However, one potential threat to reliability and validity of data collected through video recordings is observer effects. Although we sought to conduct three observations for each teacher to allow for habituation to being recorded, it is possible that teachers prepared and provided instruction differently because they knew their instruction was being recorded. Further, the videos were recorded in active early-elementary classrooms. Therefore, it was occasionally difficult to clearly hear what the teacher or students were saying on the recordings. It is also perhaps a limitation that, given our focus on text reading instruction, we did not code indicators of delivery style (e.g., presence of lesson goals, use of examples/non-examples) across instructional domains. Finally, during interviews, we did not ask all participants the exact same questions or prompt them to elaborate on their responses in the exact same way. Although semi-structured interviews are commonly used as a flexible approach to gathering information from participants, this data collection method inherently lacks the rigorousness of structured interviews (Creswell & Poth, 2017).

Despite these limitations, the findings from this study have important implications. First, compared to our previous study in which very few teachers had participated in training related to evidence-based reading instruction, teachers in the current study, almost all of whom had participated in evidence-based reading instruction training, typically provided instruction that aligned with evidence-based reading practices. This finding suggests that early-elementary teachers’ Tier 1 small-group reading instruction for students with or at risk for RDs, especially text reading instruction, may benefit from their participation in training focused on evidence-based reading instruction, such as LETRS training. Given the recent increase in states enacting legislation around in-service teacher training, this finding may be important for policymakers, as well as school leaders, to consider when developing training policies.

Our findings also provide some evidence suggesting that spending time addressing PA and encoding during Tier 1, small-group reading instruction may improve early-elementary students’ encoding skills. As such, it may be worthwhile for teachers to ensure that their small-group reading instructional time involves activities that focus on these skills. At the same time, our findings suggest that comprehension instruction in the context of decodable texts may have a negative effect on students’ language comprehension outcomes. It is important to note that these analyses were conducted with a very small sample of participants and should be considered exploratory. It will be valuable for future research to further investigate teachers’ instructional practices during Tier 1 small-group reading instruction and their impacts on students’ literacy outcomes.

Data Availability:

Data is available upon reasonable request.