**4. Discussion**

The primary objective of the present systematic review was to examine the e ffectiveness of oral reading fluency interventions on aspects of oral reading fluency (i.e., accuracy, rate, and prosody) and reading comprehension for children in first through fifth grades with reading di fficulties. Additionally, we identified characteristics pertaining to the intervention format for studies where positive gains in reading outcomes were made after intervention in reading fluency was provided.

With regard to the first question on the e ffectiveness of oral language interventions on reading outcomes, out of the 16 studies examined in this review, 12 (75%) were found to produce small to large effects on the oral reading fluency of students with reading di fficulties. Further, eight (50%) studies had small to large e ffects on reading comprehension outcomes for these students, with the majority of interventions yielding moderate to large e ffect sizes across a variety of measures. Broadly, these findings point to the e fficacy of targeted fluency instruction in promoting reading comprehension, and to a lesser extent, oral reading fluency outcomes, among first to fifth graders with reading di fficulties.

In considering these results, we noted four interesting trends. First, the findings demonstrated that benefits of fluency interventions were not limited only to aspects of fluency that were the target of the interventions but also extended to reading comprehension outcomes. This finding is not surprising for two reasons. The first is that it is logical when we explain them within reading models that consider the role of fluency. Chall's stages of reading development outline the importance of fluency as a bridge that moves students from focusing on decoding to extracting meaning from connected texts [3,37–39]. Similarly, when we consider the theory of automatic word processing, oral reading fluency interventions provide students with practice in reading connected texts that help students gain automaticity with word recognition and thus devote more cognitive resources to comprehension [11]. Second, many of the studies reviewed reported having students engage in summarizing, generating questions, and retelling texts that they have read, which are activities included in reading comprehension interventions that help struggling students [68,69]. Further, since many of the interventions reviewed also included re-reading of texts (a fluency building activity) as a central element, our findings also sugges<sup>t</sup> the utility of including fluency-building elements in interventions aimed at improving reading comprehension. This is aligned with research suggesting that multifaceted interventions that incorporate the teaching of a variety of reading-related skills are more e ffective than those that target only one skill [70,71]. Future research could investigate oral reading fluency interventions in comparison to interventions directly targeting reading comprehension for both fluency and reading comprehension outcomes. This may help to provide teachers with greater insight on the most e ffective interventions for students with reading di fficulties.

The second trend is that e ffect sizes yielded for reading comprehension outcomes were on average larger than that for oral reading fluency. Additionally, there was a certain degree of heterogeneity of magnitude of e ffect sizes across studies on both outcomes, although the heterogeneity appeared to be more salient in oral reading fluency outcomes. Considering that the studies reviewed in this present synthesis examined reading fluency interventions, outcome measures of oral reading fluency would be considered more proximal outcomes as compared to reading comprehension. Therefore, this trend is somewhat surprising since it has been shown that e ffects of reading interventions on proximal outcomes are often greater than distal outcomes [72,73]. We put forth two possible explanations for this trend, both attributable to the method of measurement. Current conceptualizations of reading fluency favor the view that fluency encompasses both decoding and comprehension, as opposed to just decoding. Most studies reviewed assessed oral reading fluency using the Dynamic Indicators of Basic Early Literacy Skills (DIBELS), which functions more as a screening tool that focuses primarily on accuracy and speed of decoding rather than a comprehensive measure of reading skills [74]. Because inconsistencies in the association between fluency subtests of the DIBELS have been reported in some studies [75], researchers have suggested that the use of one-minute probes in DIBELS might not adequately capture all aspects of reading fluency [76]. This could explain why reading comprehension outcomes showed greater e ffect sizes as compared to oral reading fluency. On a related note, it was observed that there were more instances of the use of informal measures of oral reading fluency as compared to reading comprehension in the studies reviewed and most of these measures yielded increases of negligible e ffects for rate and accuracy outcomes. A major concern with the use of informal reading measures relates to the reliability of these measures [77], therefore, this trend calls for future studies to examine the e ffects of interventions on both informal and formal measures of fluency in order to gain a better picture of the extent of impact of fluency interventions.

The third trend is that outcome measures of fluency used to assess the e fficacy of the interventions focused on accuracy and rate aspects in most of the studies reviewed. This is predictable considering that these two components are widely regarded as representative components of fluency [10]. In contrast, only three studies documented interventions that examined or included prosody measures. This is despite the fact that prosody is also highlighted as a component of fluency in both the NRP and the Common Core State Standards (CCSS) as well as by prominent scholars in the field [5,78,79]. The lack of attention to prosody has been highlighted by researchers [22]. We put forth two reasons for why this third component is relatively neglected as compared to the other two aspects. First, there have been issues surrounding the accuracy of measurement of this component. Prosody encompasses several subcomponents such as pitch, intonation, and phrasing and is typically assessed using rating scales [80]. However, researchers have pointed out that rating scales might not adequately di fferentiate between readers of varying levels of fluency [81]. Second, the assessment of prosody is often viewed as being subjective based on the views of the person completing the assessment, which raises questions about the validity and reliability issues in assessing prosody [82]. In the present study, all three studies reported in this study showed positive increases in prosody following intervention, although di fferent measures and aspects of prosody were measured across studies. However, the dearth of literature on prosody and variability in measurement of prosody makes it di fficult to draw definite conclusions on the e ffects of interventions on prosody. This gap points to the need for future research to assess this aspect more systematically and comprehensively.

The fourth trend is that close to 90% of the studies reviewed in this search examined the same type of intervention, repeated reading. This focus on repeated reading in the literature is not surprising considering the strong theoretical underpinnings of this strategy [83,84], leading to researchers and educators alike viewing it favorably as an evidence-based approach to improving fluency [43]. This trend concurs with that in the synthesis conducted by Kim, Bryant, Bryant, and Park who found that 75% of the studies reviewed focused on repeated reading interventions for students with learning disabilities [45]. Our findings lend further support to the e fficacy of this type of reading intervention for children with reading di fficulties in promoting reading outcomes (see [39] for a review). They also provide further justification for the continued use of repeated reading in building up fluency. Repeated reading is e ffective because it addresses at least two of the three main problems that lead to dysfluent reading as identified in theoretical frameworks. These three main problems pertain to (1) di fficulty with identifying prosodic cues, (2) labored word recognition, and (3) di fficulty making associations between meanings and main ideas in text [85]. Repeated reading of text promotes practice with reading which facilitates speed and ease in the recognition of words/texts. This in turn, leads to more working memory resources being freed up, which according to Samuels and Laberge's model of automaticity in reading, allows one to focus on higher-order processes pertaining to comprehension such as accessing meaning of words and building up a situation model of the text [15]. Furthermore, at the level of higher-order processing, successive readings of the same text allow readers to focus on di fferent pieces of information in the text across readings and improve memory of central ideas that facilitate comprehension [86]. However, despite the success of the repeated reading method in promoting fluency and reading outcomes across studies, researchers also point out that specific characteristics of interventions impact the extent in which fluency interventions are e ffective [38,39,45,87,88]. Our second research objective thus focused on intervention characteristics pertaining to (1) instructional variables (e.g., pre-instructional support, the di fficulty level of text used, the number of times the text is read, and the amount of feedback provided), and (2) type of interventionist (e.g., teacher, clinician, researcher, parent, or peer coach) and how they influenced intervention outcomes.

It would appear that by examining e ffect sizes only across studies that there was no clear pattern of how type of interventionist relates to e ffectiveness of fluency interventions. This is as interventions across studies that were facilitated by either an adult or a peer coach yielded similarly positive e ffects on oral reading fluency outcomes as interventions that were facilitated by both an adult or a peer coach during the school day tended to yield positive e ffects on oral reading fluency outcomes. However, a closer examination of characteristics of studies suggests the need to consider a combination of other factors when evaluating e ffect sizes relating to type of interventionist. The first factor is whether the interventionist received appropriate and adequate training. We note that across studies, training was always provided to both the adults and students prior to the start of the intervention. For example, in Begeny and colleague's study which yielded sizeable e ffect sizes, it was highlighted that the teachers participating in the repeated reading intervention were provided with two 3-hour training sessions. In these sessions, they observed a trainer modeling the intervention procedures and were provided with time to practice implementing the intervention with one another under guidance from the research sta ff [47,48]. Training such as this may be key to the success of the oral reading fluency intervention since other studies show reading interventions conducted by trained tutors have been shown to be more e ffective than untrained tutors [89].

The second factor to consider is the close association between decoding (i.e., phonemic awareness, letter–sound correspondences) and fluency skills. It was observed that in three of the studies where a trained teacher facilitated repeated reading intervention, no e ffects on students' rate and accuracy outcomes were found. The authors across these studies made mention of issues pertaining to decoding skills in accounting for these null findings. For instance, Vadasy and Sanders attributed their non-significant findings to the use of a vocabulary extension activity rather than explicit instruction in alphabetic knowledge and decoding strategies in the intervention [57], where the latter was considered as being more crucial to developing automaticity in reading [11]. Similarly, Martens et al. acknowledged that the second-grade students in their sample may have ye<sup>t</sup> to master the foundational reading skills (i.e., phonemic awareness, sight word vocabulary) necessary for fluency-building interventions to be e ffective [53]. The importance of considering decoding elements in addition to fluency building activities in fluency interventions echoes the views of researchers in the field [85,90]. Future research that directly compares interventions with and without a focus on decoding and/or considers decoding abilities of children before and after interventions would provide insights into the significance of including decoding instruction to fluency interventions.

In thinking about group size, intervention provided by an adult in a one-on-one setting appeared to be more e ffective than interventions utilizing a small-group format. Nine of the 16 studies reviewed provided an oral reading fluency intervention in a one-on-one setting with an adult [47,48,53,54,56,60,61]. The majority (56%) of studies that employed a one-on-one student-to-teacher ratio for interventions were found to produce moderate to large e ffect sizes (ES = 0.53–1.18 for oral reading fluency and ES = 0.70–1.47 for reading comprehension) whereas the remainder of studies yielded small or negligible effects for oral reading fluency rate and accuracy. The variability in e ffect sizes across studies should be discussed in the context of other intervention characteristics. For one, previous research has indicated that the inclusion of more fluency-building features in instruction promoted better outcomes [39]. An examination of the magnitude of e ffect sizes across studies with one-on-one formats appear to support this. For instance, Begeny and colleagues' two studies which yielded large e ffect sizes on oral reading fluency measures [47,48], incorporated a total of seven additional features—six of which targeted aspects of fluency including prosody (i.e., teacher modeling, performance feedback, verbal cueing), accuracy (i.e., phrase-drill error correction), and motivation (i.e., goal setting, reward system). Conversely, in Therrien et al. [56], where a small e ffect was found, only three of the five features in the intervention focused on building fluency (i.e., verbal cueing, goal setting, error correction).

Interestingly, across studies, the repeated reading intervention often included a model of fluent reading or accurate word pronunciation (e.g., error correction, phrase drill). Chard et al. highlighted that repeated reading with a model is often more e ffective at increasing students' oral reading fluency and reading comprehension [38]. However, our review showed that some studies with negligible or small e ffects on oral reading fluency also included modeling procedures. For example, Martens et al. examined an after-school intervention program which included modeling through listening passage preview and phrase drill error correction [53]. While negligible e ffects were found for students oral reading fluency, it is unclear if absenteeism and attrition impacted the results as these issues have plagued other after school intervention studies [91]. This suggests the need for future research to examine the context in which these modeling strategies are most e ffective.

In addition, intervention duration also appears to a ffect the e ffectiveness of di fferent types of one-on-one interventions. To illustrate, while Esteves and Whitten found a moderate e ffect for a listening while reading intervention (ES = 0.64) [49], Friedland et al.'s intervention of the same nature was found to have no e ffect [50]. When examining the two studies, the duration of intervention was vastly di fferent, with students in Esteves and Whitten's study spending three to six times longer in the

intervention than students in Friedland et al.'s study. Conversely, the duration of the intervention did not appear to contribute to the e ffectiveness of the repeated reading interventions examined. This suggests that the success of a listening while reading intervention, but not repeated reading, might be partly dependent on duration of intervention. However, we recognize that the available sample of studies for such a comparison is small in the present review, and thus this hypothesis necessitates further validation.

In contrast to one-on-one interventions, three of the four studies with small-group format interventions yielded negligible or small e ffects in fluency and/or reading comprehension [57–59]. The only exception was the study by Kuhn [51] where a large e ffect size was found for reading comprehension outcomes. Although the findings appear to support one-on-one interventions, a practical constraint to consider is that this format is time consuming and classroom teachers may not be able to devote the time needed to provide one-on-one interventions. This is especially true if the classroom has a high percentage of students with or at risk for reading di fficulties. One viable option may be the use of peer tutors, as two studies of peer coaching in this review produced large and significant e ffects for oral reading fluency [46,52]. However, these findings should also be interpreted with caution since three of the four studies on small groups and two of the three studies on peer coaches came from the same research teams, respectively. In addition, we note that there was substantial heterogeneity in the features of the intervention between studies that conducted small-group interventions as opposed to those that carried out interventions in one-on-one formats. Therefore, future research examining small-group oral reading fluency interventions, especially in direct comparison to one-on-one interventions, may be beneficial in informing classroom practices.

#### *Limitations and Future Research*

This review should be considered in the context of its limitations. One limitation is that conclusions about reading comprehension should be interpreted with caution since a wide variety of reading comprehension measures were used across studies. Future research is encouraged to utilize common standardized measures of reading comprehension so that results across studies may be compared in a more meaningful manner.

Furthermore, one important question posed by Chard et al. is still left unanswered [38]. Chard et al. suggested that future researchers investigate whether the e ffects of fluency building interventions are sustainable over time. In the present synthesis, only one study included a measure of retention which assessed students' oral reading fluency two days post-intervention. Future research may also wish to examine the long-term maintenance of oral reading fluency gains found in students who received a successful fluency intervention, a limitation acknowledged by Martens and colleagues [53]. In addition, the findings of this synthesis also point to gaps in our understanding of the e fficacy of interventions other than repeated reading and on prosody outcomes. Future studies should examine other commonly used interventions such as choral and echo reading or readers' theater and examine prosody outcomes in the assessment of fluency in order to provide a more comprehensive understanding of fluency instruction and reading. Finally, researchers have acknowledged that the increasing attention to fluency in the US context can be attributed in large part to its inclusion in the National Reading Panel report published in 2000 [5]. This could explain why studies yielded for this and previous reviews were conducted primarily in the U.S. context despite no country filters being imposed in the search. The lack of diversity in this respect points to the need for more studies in other contexts where di fferences in instructional curriculum might exist in order to understand whether and how instructional curriculum might moderate the impact of fluency interventions. This is especially important considering that fluency is a central component in curriculum standards in many countries and languages [6].

#### **5. Implications and Conclusions**

Despite its limitations, the present synthesis contributes to the literature in three ways. First, our findings support previous syntheses—the last of which was conducted by Stevens et al. with studies

between 2001 and 2014 [40]. Generally, fluency interventions examined, which constituted mostly repeated reading procedures, made positive contributions to gains in reading fluency and reading comprehension among children with reading di fficulties. Considering that failure to replicate findings presents as a salient issue in research [92,93], the replication of findings of previous syntheses despite not including single-case studies and only focusing on those studies that included a control group provides convergen<sup>t</sup> evidence for the importance of fluency on reading outcomes. Second, extending from previous syntheses, our finding compared relative gains of fluency interventions on fluency and reading comprehension outcomes. The observation that fluency interventions produced greater gains in comprehension as compared to fluency in general adds credence to the current view that reading fluency relates to both decoding and comprehension rather than the earlier conceptualization that fluency was confined mainly to word recognition [5,94]. Third, the current study also examined characteristics of interventions and their impact on reading outcomes, including the impact of group size in intervention, a characteristic that had not been systematically investigated in previous studies. Our findings thus have implications on instructional practices in the classroom.

The findings from this synthesis sugges<sup>t</sup> that elementary students with reading di fficulties would benefit from one-on-one interventions which include a model of fluent reading who has received adequate training. The findings support earlier research which suggests that the repeated reading of text is one way to build the oral reading fluency of students with reading di fficulties. However, continuous reading with teacher support may also be an e ffective method. Moreover, although listening while reading and small-group interventions produced only small to moderate e ffect sizes, they may be more e fficient means of improving oral reading fluency since they require less instructional time of the classroom teacher. However, an important fact to remember about any intervention, and one Young et al. also mentioned, is that there is not one intervention that is e ffective for every student and if not responding to one type of fluency intervention, some students may need an alternate form of instruction [60].

Research on the e fficacy of certain approaches to teaching fluency has led to changes in instructional practices and should continue to do so [95]. With more recent focus on Response to Intervention (RTI) and Multi-Tiered Systems of Support (MTSS), evidence-based fluency strategies should be used to target students with and at risk for reading di fficulties in classroom instruction, as well as in small-group intervention outside of the classroom. Listening while reading, repeated reading with multiple features, and continuous reading with teacher support are methods that can be e ffectively used across a variety of contexts for increasing the oral reading fluency of elementary students who need more targeted instruction.

**Author Contributions:** Conceptualization, A.H. and E.B.-C.; Methodology, A.H., P.W.K., and K.A.M.; Investigation, A.H., P.W.K., and K.A.M.; Writing—Original Draft Preparation, A.H., P.W.K., and K.A.M.; Writing—Reviewing & Editing: A.H., P.W.K., K.A.M., and E.B.-C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.
