Time in Text: Differentiating Instruction for Intermediate Students Struggling with Word Recognition

When students read below grade level, teachers are caught between two goals: (a) keeping text complexity high for comprehension instruction and (b) ensuring students are spending time in text, recognizing words themselves. This article informs the problem with research-based recommendations. Recently I was in a K–5 elementary school, one in which many intermediate-grade students had not passed their state outcome measure for reading. Following a state plan, the teachers were doing high-intensity tutoring in small groups with these intermediate-grade students struggling with reading. One fourth-grade teacher shared a concern that went something like this: "This tutoring requires us to do the interventions using 'on grade-level content' but I have fourth graders who are reading on a first-grade reading level. Are they supposed to read grade level texts or am I supposed to just read it to them? That's fine if I am supposed to read it to them but at what point do they practice reading?" Boy, I felt her pain. On the one hand, I had seen the devastating effects of limiting third–fifth-grade students with below-grade level word recognition skills, to simple, decodable books of 100–200 words. It meant they did not get the rich vocabulary, content knowledge, and advanced comprehension from grade-level materials. On the other hand, I worried about "exposure to grade-level" materials replacing time in text, the kind that builds accurate, fluent decoding and eventual automaticity. In fact, based on my observations, in many classrooms, exposure to grade-level text for comprehension was coming at the cost of systematic, explicit, decoding and morphology instruction. There are beginning readers in grades 3–5, readers who do not have mastery of the most common graphemes and morphemes that children typically solidify in K–1. There are also readers who have partial knowledge with decoding gaps that impede their ability to recognize words in text. I started to imagine an impending tsunami of sorts when students arrived in middle schools thinking like champs but unable to recognize words. Although reading aloud across K–12 is a critical practice, it should not supplant student reading. Read-alouds in the secondary school should not be the primary access to text for students. Many teachers are caught between keeping the challenge and complexity of text content high for comprehension and ensuring that students are spending time in text, recognizing words themselves so that they will become fluent, accurate, and expressive readers. In this article, I address four issues that classroom teachers face in dealing with below-grade decoders: (1) When to mediate student word recognition in grade-level texts; (2) How to mediate word recognition in grade-level text; (3) How to differentiate texts, when necessary, to ensure time-in-text; and (4) How to close decoding gaps. This article focuses on students in a given grade who find it challenging to recognize words in grade-level texts and it uses several terms, reading, grade-level text, text access, and word recognition scaffolding. Because these terms can have different meanings, I provide research-based definitions to each in the Table 1. Reading = Word Recognition × Comprehension Reading involves two fundamental processes: word recognition and comprehension, a model called the Simple View (Gough & Tunmer, 1986). From this theoretical model, reading happens only when students are engaged in both word recognition and comprehension. Word recognition refers to the process of identifying a word's oral pronunciation using a variety of mechanisms (e.g., decoding, structural analysis, and automatic recognition). According to Gough and Tunmer (1986), word recognition is fast, automatic, and importantly, in their model, it precedes language comprehension. Decoding is using the relationships between phonemes and graphemes, the alphabetic system, to translate words into their oral pronunciations. Decoding is sounding out words, a behavior that orthographically maps words and enables cognitive storage and retrieval. The Active View of reading further unpacks word recognition and language comprehension; identifying bridging processes such as fluency, morphology, and cognitive flexibility that reflect an overlap of word recognition and comprehension (Duke & Cartwright, 2021) In this article, text exposure refers to a strategy to give readers access to the words in text because they cannot recognize a large proportion of words. In short, it is someone else recognizing the words. Shared reading activities such as teacher read-alouds and computer-generated output can compensate for deficits or nascent skills in decoding. This access provides a temporary entrée to the meaning, vocabulary, and ideas in a text so that students can participate in vocabulary, content, and comprehension instruction. Exposure to grade-level texts and their content is a critical component of instruction to maintain high-level comprehension and vocabulary when students are unable to recognition a minimum threshold of words in a text (Montag & MacDonald, 2015). One of the pivotal questions that teachers must answer with students is when to scaffold word recognition in challenging texts and when the word recognition demands are too high for any benefit to be derived from scaffolding. In this section, I provide guidance in the following areas: (a) studies that used word accuracy to mediate word recognition, (b) studies that used reading level, and (c) caveats about mediating word recognition supports. Although flawed and perhaps overvalued, the approaches to reader-text matching provide a place to start (Mesmer, 2008; Reutzel & Fawson, 2022). The two approaches are (a) word accuracy levels and (b) reading level matches. In the word accuracy approach, also called "individual reader-text match," the percentage of words read accurately is used to determine a match (Amendum et al., 2018) (Note. All attempts to connect readers with texts will be estimates. As a student reads and a teacher responds, reader-text matches are adjusted). The word accuracy levels that are widely used today emanated from Betts' 1946 study, which suggested 95–99% for instructional (INS) level text, 99–100% for independent (IND) level, and 90% for frustrational (FRUS). These accuracy levels have shifted a bit to 90–95% for INS and 96–100% for IND. Educators have rightly questioned these levels due to the lack of strong research to support them and several studies suggest that, under the right conditions they can be flexed (See Halladay, 2012; Strong et al., 2018). A 2010 study tested the influence of initial accuracy levels on the word recognition, decoding, vocabulary, and comprehension of second to fourth graders struggling with reading (O'Connor et al., 2010). Students were randomly assigned to one of four groups in which they practiced reading texts that were (a) at an IND level initially (92–100%), (b) difficult level initially (80–90% accuracy), or (c) an untreated control. For a period of 20 weeks, students in the independent and difficult groups practiced reading to an adult listener who provided difficult words for them when the students stumbled. Both groups reading text had better reading rates, word recognition, and comprehension skills than the untreated control group. Interestingly, there were no significant differences in growth between the different text levels. Both groups of students, even those reading texts at 80% initially, benefited. However, second-grade and fourth-grade texts were different. Researchers found that second-grade texts had more overlapping or repeated words across passages meaning that second graders were rereading more previously seen words. The study suggested that, with an adult nearby, students can handle texts that they originally read with as low as 80% word accuracy but in second grade, the texts should have some word-level controls and repetition across passages. A 2007 study examined comprehension and word accuracy of 64 "language minority1" or multilingual (ML) first graders, participating in one-on-one tutoring (Ehri et al., 2007). Researchers analyzed progress based on if the students initially read the books at IND (98–100%), INS (90–97%), or FRUS (<90%) accuracy levels, and found that multilingual learners appeared to make the greatest gains in word reading and comprehension in IND-level texts, ones that they initially read at 98–100% accuracy. This suggested that higher levels of accurate word recognition were necessary for comprehension with students learning English. This makes sense, the demands of both recognizing words and then connecting those L2 words to meanings may tax reading comprehension of MLs. With the second approach, the reading level match, also called "group or grade/text match," a student's reading level is obtained in grades or Lexiles and matched to a text in the same units. With mixed results, researchers have tested the impact of having students read one, two, three, or even four levels above their levels (Brown et al., 2018; Morgan et al., 2000; O'Connor et al., 2002; Stahl & Heubach, 2005). Two studies investigated the impact of a partner reading invention on second graders (Brown et al., 2018; Morgan et al., 2000). In both these studies, students were paired with more capable peers and asked to read 15 minutes per day for 95 days. The students read together sitting side by side with one smooth finger tracking the print. They were told to keep their eyes on the words, read together, and identify difficult words for help later. In the first study, second graders who read at a pre-primer level initially were assigned texts at the following levels (a) on-level (PrePrimer); (b) two grades above level (first); and (c) four grades above level (second). The most improvement was for students reading two grades above—second graders reading first-grade materials (Morgan et al., 2000). A modified replication was conducted in 2017 with second graders reading texts at two, three, or four "levels" about their measured reading level. There were no effects for the more capable peers, but the readers being assisted, those working at two "levels" above their instructional levels, had the best gains in fluency and comprehension (Brown et al., 2018). Another study manipulated text with third through fifth graders struggling with reading. Within a tutoring program students were randomly assigned to read texts either (a) at their assigned to reading level; or (b) at their grade levels (~2 levels above). Fluency was best with texts matched to actual reading levels not grade-level texts (O'Connor et al., 2002). The fourth study, a robust, 2-year project, found positive effects of increasing text difficulty on the comprehension of second graders (Stahl & Heubach, 2005). The Fluency Oriented Reading Instruction (FORi) had students rereading the same text four to seven times over a week. Teachers presented texts using many different techniques and provided a series of supports (see below) and the intervention improved comprehension and reading levels. A final study with fourth graders functioning below level used sets of texts with different levels to build comprehension and vocabulary around social studies concepts (Gelzheiser, 2005). Students chose books and read them with volunteers over 50 sessions. Book choices were adjusted for ability and interest. The exploratory study found that readers' entering decoding skills, experience of the volunteer, and text difficulty levels were predictive of gains in reading achievement. One final consideration in using the reading level approach is to think about the standard error of measure with readability formulas. A standard error of measurement (SEM) is the range around which the true difficulty of a text is. Most formulas have a SEM of 0.5–1 grades (Mesmer, 2008). This means that if a text is estimated to be at a 3.3 grade level, its true difficulty may be as high as 4.3 or as low as 2.3. What this suggests is that if a text is within one grade level of a student's reading level, mediating word recognition might be productive, particularly if the student is typically developing. The literature provides additional insights about mediating word recognition in grade-level text. Many studies, for example, confirm that increases above one grade level, without support are not beneficial (Amendum et al., 2016; Mesmer & Hiebert, 2015). A literature review examined the relationship between text difficulty, comprehension, and fluency (Amendum et al., 2018). Overall, the findings showed negative or nonsignificant relationships between increases in text difficulty and comprehension and fluency. Two additional studies tested the impact of students reading above level without support (Amendum et al., 2016; Mesmer & Hiebert, 2015). Researchers examined the comprehension of over 600 students in grades 1–3 independently reading materials at and well above grade level (Amendum et al., 2016). Students reading materials well above grade level, even if reading with high levels of accuracy, had significantly lower levels of comprehension. In another study, researchers tested the comprehension of third graders at the end of the school year. Students independently read texts at two lengths (200 vs.1000 words) and three levels: second (400 L), third–fourth (600 L), and fifth (800 L) (Mesmer & Hiebert, 2015). When third graders read fifth grade-level materials, they achieved 50–57% comprehension. When they read third-to fourth-grade materials, they comprehended 64–66% and 71–84% in second-grade materials. Make sure that children have solid foundational skills in decoding—say through a high first-grade or a beginning second-grade level— before increasing the complexity of the texts used to teach reading… The caution given here for young readers also would make sense for older readers who still decode like kindergartners or first graders. Don't worry about taking students beyond "instructional level" texts until they are able to decode as well as a successful first-grader (p. 22). This perspective was also reflected in a piece by Reutzel and Fawson (2022) who pointed out the developmental importance of decoding accuracy and self-confidence. They wrote, "In the earliest stages of reading acquisition, teachers do well to select texts that act as scaffolds for developing young students' reading foundation skills as well as to bolster their collective sense of self-efficacy as early readers. (p. 498)." An experimental study with second and third graders struggling with reading, supported this perspective. Researchers analyzed all the words read by students and found that texts with more complex vowel and morphemic patterns were associated with compromised letter sound, fluency, and comprehension performance (Vadasy & Sanders, 2009). Incomplete decoding knowledge affects fluency and comprehension. Essentially text levels are less flexible with beginning readers who do not have a command of the most common graphemes. Educators can use a variety of research-based techniques to help below-grade decoders to engage in both parts of the reading equation themselves. I will highlight three of them here. Fluency Oriented Reading Instruction (Stahl & Heubach, 2005) and Wide-Reading FORi (Kuhn et al., 2006) are research-tested instructional routines that scaffold not only word recognition but also fluency and comprehension in grade-level texts. In their 2-year study, Stahl and Heubach (2005) reorganized a basal reading program, to include whole-group repeated readings, partner reading, and silent reading opportunities with self-selected texts. FORi resulted in significant gains in reading levels for all but two second-grade students who began the year at the primer level or higher, and half of those students who started below the primer level. These students were reading on grade level by the end of the year. Students made 1.88 year's growth in Year 1 and 1.77 year's growth in Year 2. The program's largest gains were on measures of rate and accuracy, factors that free up working memory so readers can focus on comprehension. Wide-Reading FORi, a modified replication of FORi, substituted wide reading, increased reading volume, for repeated reading. In Table 2, shown below, FORi components are summarized (Stahl & Heubach, 2005). Read along Story discussion Echo read Partner reading Read at home Whole-class choral read Partner reading A second technique for scaffolding word recognition, as well as comprehension is the text set (Gelzheiser, 2005; Lupo et al., 2020). Researchers applied this technique within content areas such as social studies or science to build knowledge. Difficult texts with rich, advanced vocabulary build knowledge and are paired with more accessible texts that build reader capacity and efficacy—the sense that I can read this and learned something. In one study, the set was built with following four types: (a) controlled vocabulary texts (INS or IND); (b) picture books (FRUS); (c) chapter books (FRUS); and (d) informational texts (FRUS) (Gelzheiser, 2005). Another study used the following types: (a) target text, a dense, challenging textbook chapter; (b) a visual text to convey information graphically; (c) a hook text, to show relevance and create excitement; and d) an informational text with word-level controls (e.g., Newsela). In both studies, students read all types of texts within the set. Often, the more challenging texts were after the easier texts and were sometimes read aloud. The approach worked because a diversity of materials with the same content presented vocabulary and information repeatedly and at different levels of depth. Students may read easier materials that have less depth of concept, but those materials produce background to access more challenging texts. A third approach to scaffolding may seem rather odd, even counter-productive, but under certain circumstances, it appears to have some benefits (O'Connor et al., 2010). The approach involves simply pronouncing a challenging word for a student. In a study examining fluency, second through fourth graders struggling with reading were able to handle materials at 80–90% if, when they encountered a word that they could not pronounce, an adult provided it for them. Thus, if students can initially pronounce about 80% of words, providing words for them may be useful. As I have described, under certain conditions, students' word recognition can be scaffolded in grade-level texts but sometimes it cannot be. Because students' literacy instruction should never be exclusively in texts that someone else must read, a plan must be made to ensure substantive, time-in-text. Quite simply, teachers will know that they need to do so when students are spending most of their time during reading instruction listening to someone else read texts. There are several sources for texts that students can read. Building a text set, as described in the previous section is a great way to keep the content focus and time-in-text (Gelzheiser, 2005; Lupo et al., 2020). Other techniques include a paired teacher/student text, in which the teacher provides in-depth content and the students read complementary text about the content within the same passage. In Figure 1, the teacher reads the smaller print sections that precede the student readings in larger, bold print. One of the patterns I see in schools, a pattern that causes much concern, is a lack of on-level, systematic, accelerated phonics, and morphology instruction for students who are not reading during comprehension and vocabulary lessons. I see teachers enjoying the satisfaction of bridging the text access gap for below-grade decoders. The true capacity of these students is met as they contribute to discussions, share insights, advance their vocabulary, and demonstrate higher level thinking. Yet, a kind of amnesia kicks in and sometimes everyone forgets that only half of the reading equation was taking place. Students were not recognizing words. To advance, students need intense, assessment-informed instruction to close decoding gaps. What the research tells us, however, is that this gap-closing instruction does not look like what I think of with typically developing beginner readers; it has two distinct features. First, because words with multiple morphemes outnumber single-morpheme words by 4-to-1 in English (Moats & Brady, 2000), instruction includes both phonics for single-syllable decoding and instruction for words with multiple syllables and morphemes (Archer et al., 2003; Lovett, Lacerenza, Borden, Frijters, et al., 2000; Lovett, Lacerenza, & Borden, 2000; Shefelbine, 1990; Toste et al., 2017, 2019) (Figure 2). A series of studies with third through fifth graders who participated in over 70 hours of instruction improved word recognition and decoding (Lovett, Lacerenza, & Borden, 2000; Lovett, Lacerenza, Borden, Frijters, et al., 2000). The instruction had both decoding instruction and word recognition strategies to flexibly handle variable vowels and polysyllabic words. Students were taught basic graphemes and decoding—not only how to blend words together but also the following four strategies for handling longer words: (a) vowel flexing (trying different sounds when there is a variable sound head -long e, short e, long a); (b) analogies (if I know long then I know strong); (c) peeling off prefixes and suffixes; and (d) seeking the parts you know (look for base words, vowel graphemes, syllables, prefixes, and suffixes). Another series of studies, by a different researcher included fourth–fifth graders and had similar effects on decoding, spelling, and comprehension instruction (Toste et al., 2017, 2019). The intervention in these studies consisted of the following parts: (a) vowel practice (naming sounds for vowel patterns); (b) affix reading; (c) blending and segmenting words into morphemes; (d) speeded word reading practice; (e) spelling; and (f) text reading with words. The second feature of instruction is flexibility in decoding. Whereas early decoding instruction focuses on blending through simple words where there are few alternate pronunciations (e.g. strip, bag, hit), advanced decoders confront more complex patterns with two-to-one correspondences (e.g. these, head). You do not decode a word like "head" using a left-to-right approach. Instead, you must group the ea together and map it to the correct pronunciation. A word like head, introduces another reason for flexibility, some vowel graphemes can represent more than one sound (e.g. oo, ea) and so students must try different sounds, a strategy called vowel flexing (Meese, 2016). Finally, as advanced decoders encounter less regular words, such as stomach, they must apply a kind of cross-checking against words in their vocabulary. Say a student pronounced this as "st ah mach," they might think, "Hmmm I don't know that word, but I do know stomach. That must be stomach." This cross-checking of pronunciations of less regular words against known words has been called "set for variability (SfV) or SPAAR, Semantic and Phonological Ability to Adjust Recoding (Kearns et al., 2016)". Look at the readers in your classroom who are decoding below-level and analyze their assessment data. Identify their strengths and weaknesses and consider if word recognition scaffolds should be applied to support text access. For readers who need different materials to support time-in-text, identify a collection of materials that would be suitable and motivating. Find a time when you can support and hold accountable these readers as they connect with text for at least 20 minutes. Challenge yourself to keep a record of the reading that these students are doing to ensure sufficient time-in-text. None. Heidi Anne Mesmer, Ph.D., is a Professor Literacy Education at Virginia Tech – Reading, 318 War Memorial Hall, Blacksburg, Virginia 24061, USA; email [email protected].