Associations Among Teacher–Child Interactions, Teacher Curriculum Emphases, and Reading Skills in Grade 1

ABSTRACT Research Findings: The purpose of the present study was to examine the extent to which the quality of teacher–child interactions and teachers’ self-reported curriculum emphases are related to children’s reading skill development during their 1st school year. To accomplish this, we assessed the reading skills of 1,029 Finnish children (M age = 85.77 months) twice during Grade 1, and the children’s teachers (n = 91) completed questionnaires concerning their literacy-related curriculum emphases. In addition, teacher–child interactions in terms of emotional support, classroom organization, and instructional support were observed in 29 classrooms. The results of multilevel modeling showed that a high global quality of teacher–child interactions was positively associated with improved children’s reading skills at the end of Grade 1. In addition, a teacher emphasis on comprehension and production skills was related to better reading skills via teacher–child interactions. Domain-specific analyses revealed that emotional support and classroom organization in particular were related to better reading skills. Practice or Policy: The present study adds to previous research by showing that children had better reading skills at the end of their 1st school year in classrooms in which the teachers were warm, responsive, and sensitive to children’s needs and provided well-planned activities, clear rules, and expectations for behavior.

A considerable evidence base attests to the powerful role of teachers in fostering children's reading and spelling skills in early elementary classrooms (e.g., Connor, Son, Hindman, & Morrison, 2005;National Early Literacy Panel, 2009). For example, the quality of teacher-child interactions in the classroom has been linked with children's reading skills (Curby, Rimm-Kaufman, & Ponitz, 2009;Hamre & Pianta, 2005;Ponitz, Rimm-Kaufman, Brock, & Nathanson, 2009) and vocabulary and print concepts (Cadima, Leal, & Burchinal, 2010). In addition, it has been suggested that teacher beliefs regarding language and literacy development are linked to children's outcomes (Cash, Cabell, Hamre, DeCoster, & Pianta, 2015). Even when implementing the same curriculum, different teachers may emphasize different curriculum goals or different priorities (Massetti & Bracken, 2010;Sonnenschein, Stapleton, & Benson, 2010), and these emphases may impact their practices and their interactions with the children in the classroom. For instance, depending on their goals and emphases, teachers may allocate their time and choice of pedagogical practices differently (Cunningham, Zibulsky, Stanovich, & Stanovich, 2009), thereby shaping children's daily learning experiences in diverse ways. Because the first school year is essential for later academic success (Entwisle & Alexander, 1998;Jimerson, Egeland, & Teo, 1999), there is a clear need to investigate the factors influencing children's reading skill development during the first school year.
The extant research in this field has some limitations. First, although some studies have focused on the association between teachers' beliefs and practices (Fang, 1996;Hamre et al., 2012;Schachter, Spear, Piasta, Justice, & Logan, 2016;Stipek & Byler, 1997, 2004Wilcox-Herzog, 2002), teachers' self-reported curriculum emphases and actual observed classroom interactions have seldom been investigated within the same study (for an exception, see Wen, Elicker, & McMullen, 2011). A cornerstone of the present study is its simultaneous inclusion of both observations of teacher-child interactions and teachers' reports of curriculum emphases. Second, only a few studies have investigated the connection between teachers' curriculum-related emphases and beliefs and children's reading skills (e.g., Cash et al., 2015;Massetti & Bracken, 2010), especially in transparent languages like Finnish, which children typically learn to decode during the first months of school and for which reading skills develop much faster during the first school year than for more complex languages, such as English (Seymour, Aro, & Erskine, 2003;Soodla et al., 2015). Third, observational research on the effects of teacher-child interactions on reading skill development in cultural and educational contexts outside of the United States is lacking, even though empirical studies in non-U.S. countries would provide important information on the generalizability of findings. Because of the repeated success of Finnish students in the Programme for International Student Assessment studies, Finland is an interesting context in which to examine whether similar kinds of teacher-child interactions can be found in Finnish and U.S. classrooms and to observe how Finnish teachers emphasize different curriculum aims. Consequently, the present study contributes to the existing literature by investigating the associations among teacher-child interactions, curriculum emphases, and reading skills in Finland. Learning more about these associations is important because the studied areas represent malleable aspects of teacher-child interactions that could serve as the focus of future interventions in teacher preparation and training.

Teacher-Child Interactions
Previous research has shown that the social-emotional, organizational, and instructional aspects of teaching contribute positively to students' successful learning and adjustment (Connor, Morrison, & Katch, 2004;Perry, Donohue, & Weinstein, 2007;Pianta, La Paro, & Hamre, 2008;Van de Grift, 2007). The teaching through interactions framework (Hamre et al., 2013) provides a theoretical and empirically tested model that conceptualizes teacher-child interactions in terms of three domains: emotional support, classroom organization, and instructional support . Emotional support refers to a classroom's supportive climate and positive tone of interactions. Emotionally supportive teachers are warm, sensitive, and responsive to children's needs, and they provide children with appropriate levels of autonomy . Classroom organization refers to the effective setting of rules and routines as well as to the teacher's management of time and attention (Yates & Yates, 1990), the promotion of student motivation, and the provision of inherently interesting activities . Unlike many other conceptualizations of classroom management or organization, in addition to providing structure, high-quality classroom organization may also promote children's interest and engagement in the classroom. Instructional support focuses on the quality of feedback, the teacher's ability to stimulate thinking skills and reasoning among students, and the teacher's ability to present content knowledge in meaningful contexts (Hamre et al., 2013;Pianta et al., 2008).
Teacher-child interactions can be seen as influential for child outcomes from many theoretical perspectives. For example, according to Bronfenbrenner and Morris's (2007) bioecological model of human development, interactions between adults and children form the key proximal processes influencing child development (Pianta, 1999). Thus, consistent, adaptively orchestrated teacher-child interactions of high quality can be assumed to be the central drivers of children's academic development. For instance, high-quality language modeling and sensitively timed process-oriented feedback from teachers provide children with rich learning opportunities and boost their literacy skills (Hamre et al., 2013. Existing research on the role of emotional support in classrooms has been guided by two broad theories: attachment theory and self-determination theory. The extended attachment theory posits that if children feel respected by and emotionally secure with their teachers, they are optimally able to focus their attention and engagement on learning (Bergin & Bergin, 2009;Pianta, 1999). Emotionally supportive and respectful interactions with a teacher thus provide a safe environment that allows for and encourages student exploration and inquiry without fear of embarrassment. In turn, this high emotional investment in learning is likely to lead to better learning outcomes. Self-determination theory (Ryan & Deci, 2000) explains the relation between teacher-child interactions and child outcomes slightly differently. It argues that students' engagement in learning activities will increase if their basic psychological needs for autonomy, relatedness, and competence are met in the classroom. Teachers can promote the fulfillment of these needs by being sensitive and responsive to students' needs, taking children's initiatives into account, and providing process-oriented feedback through high-quality emotional and instructional support (Hamre et al., 2013).
Empirical findings concerning the role of environmental support in children's outcomes have shown positive associations between instructional and organizational support and the development of children's self-regulatory (Paris & Paris, 2001;Raver et al., 2009) and cognitive and linguistic (Wharton-McDonald, Pressley, & Mistretta-Hampston, 1998) skills. High-quality classroom organization also helps children better regulate their behavior, enabling them to benefit more optimally from the available instruction (Cameron, Connor, & Morrison, 2005;Rimm-Kaufman et al., 2009). The concept of instructional support is drawn in part from cognitive learning theories (Yilmaz, 2011), which view students as active participants whose learning is supported by effective teaching that focuses on the understanding of broad concepts and ideas. Instructional support involving process-oriented feedback, extending, and language modeling promotes the development of higher order thinking skills and conceptual understanding (Hamre et al., 2013;Pianta et al., 2008). Drawing on these theoretical views and empirical results, the present study set out to investigate the associations between the quality of teacher-child interactions and children's reading skills in Grade 1 Finnish classrooms.

Teachers' Curriculum Emphases
A curriculum includes the learning objectives, knowledge, and skills children are expected to learn in school. In turn, pedagogy refers to how teachers engage with children to achieve their curriculum aims as well as what directs teachers' practices and interactions with the children. Although many factors influence classroom instruction and how time is spent in the classroom, literature indicates that teachers' beliefs play a critical role in their decision-making and classroom practices (Stipek & Byler, 1997). Therefore, what teachers think (i.e., their emphases and beliefs) might be equally as important as how they support children's learning through teacher-child interactions. The somewhat nebulous concept of curriculum emphases (cf. beliefs) has been measured in a variety of ways, but researchers have revealed mixed findings concerning whether teachers' beliefs are associated with their practices and instruction (e.g., Justice, Mashburn, Hamre, & Pianta, 2008;Wen et al., 2011).
Despite implementing the same curriculum in their literacy instruction, teachers may vary in their focus of emphasis, which may lead them to prioritize some areas over others. Taking into account teachers' curriculum emphases is particularly important in the Finnish educational context, in which teachers are granted a great deal of pedagogical autonomy (National Board of Education, 2014). Previous literature has shown that teachers differ in their instructional emphases (Sonnenschein, Stapleton, & Benson, 2010), pedagogical goals (Aunola, Leskinen, & Nurmi, 2006), curriculum beliefs (Wen et al., 2011), and curriculum goals (Massetti & Bracken, 2010;Stipek & Byler, 2004). For example, some teachers emphasize the social and motivational aspects of learning, such as self-conceptualizations of ability and interest in activities (Aunola et al., 2006;Massetti & Bracken, 2010), and the acquisition of higher order thinking skills, such as problem solving and comprehension. By contrast, other teachers stress the achievement of basic skills through drills and practice (Stipek & Byler, 2004).
Although many studies have focused on teachers' developmentally appropriate beliefs and the extent to which teachers' beliefs and practices are in congruence (Wen et al., 2011), empirical studies investigating the roles of curriculum emphases and beliefs in influencing teaching practices-and in turn child outcomes-through teacher-child interactions are rare (Sonnenschein et al., 2010;see Cash et al., 2015, andMassetti &Bracken, 2010, for exceptions). In their study, Massetti and Bracken (2010) reported that kindergartners in classrooms in which teachers emphasized emergent literacy skills demonstrated greater mastery of such skills than children in classrooms emphasizing social development. In a recent study by Cash et al. (2015), teachers' beliefs about language-and literacy-related skills were not related to children's literacy outcomes.
There are some theoretical grounds from which to assume a pathway from teachers' curriculum emphases and beliefs to their classroom practices and interactions and ultimately to children's learning (Desimone, 2009;Hamre et al., 2012). For example, Hamre et al. (2012) suggested that teacher beliefs and knowledge are important antecedents of teacher-child interactions (see also McMullen et al., 2006;Stipek & Byler, 1997), which in turn are related to child outcomes. Teacher instructional emphases serve as a foundation for setting goals and standards for learning by framing; emphasized areas are viewed as important by teachers and serve as the focuses of their attention and energy (Massetti & Bracken, 2010). Thus, instead of having a direct influence on reading outcomes, teachers' curriculum emphases and beliefs may affect their instructional choices, their planning of day-to-day activities, and their use of time (Cunningham et al., 2009), which may in turn affect children's engagement in learning (Sonnenschein et al., 2010). For instance, Guo, Connor, Yang, Roehrig, and Morrison (2012) found that classroom practices mediated the role of teacher beliefs on fifth graders' literacy outcomes. Furthermore, Aunola et al. (2006) showed that children's task motivation in math developed more positively in classrooms where teachers mentioned motivation and self-concept as their instructional goals compared with classes in which the teacher did not mention either of these as a pedagogical goal. Based on the postulations of self-determination theory (Ryan & Deci, 2000), it can be suggested that self-concept and motivation as a curriculum goal would be linked to high-quality emotional support.
Teachers' goals are not always consistent with their observed instructional practices (Justice et al., 2008;Wilcox-Herzog, 2002). Wen et al. (2011), for example, found that preschool teachers' curriculum beliefs and observed classroom practices were only weakly correlated (rs = −.03 to .22). In another study, Justice et al. (2008) showed that although preschool teachers reported implementing the prescribed components set down in a curriculum, the quality of their instructional practices was lower than that specified in the set criteria. Such results highlight the importance of examining teacher-child interactions through observation as well as through self-reported curriculum emphases. Thus, the present study adds to the current literature by being among the first to investigate the associations among curriculum emphases, teacher-child interactions, and children's reading skills. We hypothesized that curriculum emphases would not be directly linked to children's reading skills but would be associated with these skills via the quality of teacher-child interactions (Guo et al., 2012;Hamre et al., 2012;Sonnenschein et al., 2010). For example, based on empirical (Aunola et al., 2006) and theoretical (Downer, Sabol, & Hamre, 2010;Ryan & Deci, 2000) postulations it can be assumed that teacher emphases on motivation and self-concept would be linked to high-quality emotional support. In addition, because classroom organization of high quality is described by efficient and productive interactions and by children being engaged in activities , we assumed that emphases on decoding and spelling skills would be linked to greater classroom organization. Furthermore, because instructional support of high quality is described by interactions that facilitate children's cognitive and language development through meaningful instructional discussions, process-oriented feedback that expands learning, and language stimulation and facilitation techniques (Downer et al., 2010;Pianta et al., 2008), we assumed that teacher emphases on comprehension and production skills would be linked to higher quality instructional support.

Reading Skills and Teacher-Child Interactions
Language and literacy skills in early childhood and the early school years represent two interrelated constructs that are highly predictive of children's future achievements in reading (Storch & Whitehurst, 2002). Learning to read requires acquiring the foundational skills for phonological recoding: mapping between visual symbols and units of sounds (Ziegler & Goswami, 2005). The strongest proximal predictors of word reading are phonological awareness, letter knowledge, and naming speed in languages with both inconsistent (e.g., English; see Lonigan, Burgess, & Anthony, 2000) and consistent (e.g., Finnish; see Seymour et al., 2003) orthographies. In addition, vocabulary has been found to predict early word decoding (Verhoeven, Reitsma, & Siegel, 2011). The Finnish language has a highly regular orthography consisting of only 29 grapheme-phoneme combinations that are wholly consistent in both reading and spelling. Because of this, between a quarter and a third of Finnish children can read before they enter formal education at the age of 7 (Soodla et al., 2015), and the majority of nonreaders learn to read during the first semester of Grade 1 (Lerkkanen, Rasku-Puttonen, Aunola, & Nurmi, 2004). However, although most Finnish children are able to quickly acquire the basic decoding skills needed for accurate word-level reading, they may struggle with reading fluency and reading comprehension skills (Holopainen, Ahonen, & Lyytinen, 2001).
In addition to children's prereading skills and their underlying cognitive antecedents (Parrila, Kirby, & McQuarrie, 2004), teaching practices (e.g., Carlisle, Kelcey, Berebitsky, & Phelps, 2011;Connor et al., 2005) and the quality of teacher-child interactions  have also been linked with children's language and literacy skills in several studies. For example, emotional support has been found to contribute to the growth of phonological awareness in kindergarten and Grade 1  as well as to gains in expressive and receptive language (Burchinal, Vandergrift, Pianta, & Mashburn, 2010). Teachers' warmth and sensitivity have also been shown to be related to children's vocabulary and decoding at the end of Grade 1 . High-quality classroom organization in turn has been shown to be positively related to first graders' print concepts, vocabulary (Cadima et al., 2010), and literacy gains (Cameron, Connor, Morrison, & Jewkes, 2008;Ponitz et al., 2009). Instructional support in particular has been found to contribute to both children's preliteracy skills Mashburn et al., 2008) and their growth in word reading at school . Teachers' efforts to show more support and provide more positive classroom environments have been shown to correlate with stronger literacy skills among fifth-grade students (Guo et al., 2012). On a related note, Guo and colleagues (2011) have documented that classroom emotional and instructional supports contribute to third graders' reading skills via increased student engagement. In the present study, we seek to investigate the role of the quality of teacher-child interactions in first-grade children's reading skills. Thus, to summarize, each of the three domains of teacher-child interactions has either direct or indirect effects on children's language and literacy development (Downer et al., 2010).

Control Factors
In addition to curriculum emphases and teacher-child interactions, structural classroom factors, family backgrounds, and children's characteristics may also shape children's learning to read and teachers' efforts to support their literacy skills. For instance, there is evidence that a small class size has positive effects on students' reading performance, especially during the early school years (e.g., Blatchford, Bassett, Goldstein, & Martin, 2003). Teachers' work experience may also play a role in children's skill development. For example, teachers with fewer years of teaching have been found to have higher quality teacher-child interactions (e.g., Connor et al., 2005;Guo et al., 2012;Mashburn et al., 2008) and to be more sensitive in adapting their instructional practices to children's academic skills (Nurmi, Viljaranta, Tolvanen, & Aunola, 2012). In addition, parents' levels of educational qualifications have been found to predict children's prereading skills and reading performance (e.g., McClelland & Morrison, 2003). Numerous studies have documented higher levels of reading skills for girls than boys during the early school years (e.g., Logan & Johnson, 2009;Robinson & Lubienski, 2011). A child's age at school entry may also contribute to the development of reading skills (Stipek & Byler, 2001). Consequently, in our analyses we controlled for class size, teacher's work experience, maternal level of education, and child's gender and age.

Research Questions
The present study tested a broad theoretical model, which is presented in Figure 1. In this model, children's reading skills that are typical of the classroom are predicted by teacher curriculum emphases and teacher-child interactions, and children's initial reading skills and several background variables (i.e., age, gender, mother's education, class size, and teaching experience) are controlled.
The detailed research questions were as follows: (1) To what extent is the quality of teacher-child interactions related to the development of children's reading skills? We hypothesized that the global quality of teacher-child interactions would be linked to better reading skills (Hypothesis 1a; Cadima et al., 2010). In addition, we expected that high-quality emotional support (Hypothesis 1b; Burchinal et al., 2010), classroom organization (Hypothesis 1c; Cadima et al., 2010;Ponitz et al., 2009), and instructional support (Hypothesis 1d; Mashburn et al., 2008) would be associated with improved reading skills at the end of the first grade Guo et al., 2011Guo et al., , 2012. (2) To what extent are teachers' curriculum emphases on reading and writing (i.e., emphases on motivation and self-concept, decoding and spelling skills, and comprehension and production skills) related to the development of children's reading skills? We hypothesized that curriculum emphases would not be directly related to reading skill development but that they would contribute to reading skills via the quality of teacher-child interactions (Hypothesis 2a; Sonnenschein et al., 2010). More specifically, we hypothesized that teacher emphases on motivation and self-concept would be linked to greater emotional support (Hypothesis 2b; Aunola et al., 2006), emphases on decoding and spelling skills would be linked to greater classroom organization (Hypothesis 2c), and emphases on comprehension and production skills would be linked to higher quality instructional support (Hypothesis 2d). Finally, we hypothesized that these teacher emphases would ultimately-via the various teacher-child interactions described previously-be linked to improved student reading skills (Guo et al., 2012;Hamre et al., 2012).

Children
The present study was part of a larger follow-up study (Lerkkanen et al., 2006). The sample consisted of 1,029 Finnish children (523 boys) from 91 first-grade classrooms. Most of the children (99.1%) spoke Finnish as their mother tongue; however, there a small minority of children (0.9%) spoke English, Russian, Vietnamese, Albanian, or French as their native language. We excluded special education classrooms from the analyses; thus, none of the children in the present sample had any severe behavioral or emotional disabilities. Half of the teachers (50.5%) reported that there were no children with individualized education programs (IEPs) in their classrooms, 10.8% of teachers reported that they had one student with an IEP, and 17.2% reported that they had more than one student with an IEP. Only children who had the same teacher for the entirety of first grade were included in the sample. However, the testing indicated that children whose teacher had changed during the first grade (e.g., because of maternity leave) did not differ significantly from the other children. The mean age of the children was 85.77 months (SD = 3.44), which means that the children either were 7 years of age at school entry or turned 7 during the fall semester of Grade 1. Parents were asked for written consent for their children's participation in the study. The vast majority of the children (80%) came from nuclear families, 10% came from single-parent families, 8% came from blended families, and 2% came from families in which the parents were divorced and the children split their time between two homes. The sample was fairly representative of the Finnish population (Statistics Finland, 2007).

Teachers and Classrooms
All 91 teachers in our sample provided their written consent before the study. The participating classes were selected from mainstream schools in two medium-size towns and one smaller municipality located in Central and Eastern Finland. Finnish was the principal language used in the schools. The mean class size was 16.66 children (SD = 6.25, range = 3-27). Nearly all (97.5%) of the 91 teachers had at least a master's degree in education. The teachers' teaching experience ranged from less than a year to more than 15 years (mode = more than 15 years). Of the 91 teachers, 29 teachers (three males) participated in classroom observations on a voluntary basis. The remaining 62 teachers chose not to participate in the observations. However, data on the children's reading skills, class size, and other background variables were available for all children from all 91 classrooms.

Missing Data Analysis
Because only a subsample of 29 of the 91 teachers participated in classroom observations, we needed to address the issue of missingness of the design. To use all available data in cases of missing by design, the missingness should be random (Muthén & Muthén, 1998-2012. Consequently, the missingness of the data was tested in two ways. First, we conducted a set of independent-samples t tests to compare the 29 teachers who participated in observations to those who did not participate (n = 62). The results showed no statistically significant differences between the compared teacher groups in terms of age, educational background, work experience, number of children in the classroom, mean age of the children, or number of personnel available (i.e., the number of teachers and teacher's aides in the classroom). Furthermore, no differences were found in relation to teachers' self-reported stress, classroom management strategies, or efficacy beliefs. However, teacher-reported affection toward students was slightly higher among the observed teachers (M = 4.34, SD = 0.41) than among the teachers who chose not to participate in observations (M = 4.13, SD = 0.38), t(70) = 2.16, p < .05. Thus, overall there were very minimal differences between the observed teachers and classrooms and the teachers and classrooms that were not observed. As a second step, we tested the missing-completely-at-random (MCAR) assumption in relation to the study variables. To accomplish this, we ran Little's (1988) MCAR test. Little's MCAR test indicated that the data were not missing completely at random, n = 91, χ 2 (178) = 211.17, p = .045 (with ps >.05 indicating that the data fulfilled the MCAR assumption). The problem of missing data in between-level predictors (in our case, observed teacher-child interactions at the classroom level) has received relatively little attention in prior research (van Buuren, 2010). It has, however, been suggested that removing all of the observations in a class when there is missingness in one classroom-level predictor not only is wasteful but can lead to selection effects at the between level (van Buuren, 2010). Two recommended alternative methods for handling missingness are the use of full information maximum likelihood (FIML) estimation and multiple imputation (Enders, 2010;Schafer & Graham, 2002). Although researchers have indicated greater confidence in imputing data, there is still no consensus about the maximum amount of missingness in multilevel data that can be safely imputed or handled using FIML (van Buuren, 2010). FIML performs well even when the proportion missing is substantially high (Johnson & Young, 2011). Many simulation studies have tested missing data approaches with 50% or more missing values in model variables (e.g., Allison, 2001;Collins, Schafer, & Kam, 2001) and shown little change in findings based on missing-at-random (MAR) assumptions for missing data levels of up to 50% or more; however, beyond this level, there may be some differences among the results of estimators using different missing data strategies (Johnson & Young, 2011). Using FIML has been recommended as one of the most appropriate ways of dealing with possible selective attrition (e.g., Enders, 2010), as was the case in our study. In addition, simulation studies have shown that FIML yields less biased regression parameter estimates than other missing data procedures (e.g., Enders, 2001).
Consequently, we used the standard MAR approach to FIML estimation for missingness (Muthén & Muthén, 1998. On this basis, we were able to use the data for all teachers (N = 91) and children (N = 1,029) in further analyses. However, to make sure that the results would be similar in the smaller sample, we carried out additional analyses using only the subsample of 29 observed teachers. The pattern of results using this subsample was highly similar to the results for the whole sample, although the power to detect significant results decreased somewhat with the decrease in sample size.

Reading Skills
A group-administered subtest of the nationally normed reading test battery (reading test for primary school [ALLU]; Lindeman, 1998) was used to assess word-level reading. In this test, a maximum of 80 items can be attempted within a 2-min time limit. On each item, the child is asked to read four phonologically similar words and to draw a line connecting a picture to the word that matches it semantically. Given the nature of this speed test, the score reflects both the child's fluency in reading the stimulus words and his or her accuracy in making the correct choice from among alternatives. In the highly transparent Finnish language, this kind of reading speed test has frequently been used to identify differences among children (see Holopainen et al., 2001;Lerkkanen et al., 2004). The mean score for the test was 20.01 (SD = 9.20). According to the test manual (Lindeman, 1998), the Kuder-Richardson reliability coefficient for Grade 1 was 0.97.

Quality of Teacher-Child Interactions
The classrooms were observed using the Classroom Assessment Scoring System (CLASS K-3; Pianta et al., 2008). The CLASS consists of 10 dimensions measuring three domains of classroom quality: (a) Emotional Support (four dimensions: Positive Climate, Negative Climate, Teacher Sensitivity, and Regard for Student Perspectives), (b) Classroom Organization (three dimensions: Behavior Management, Productivity, and Instructional Learning Formats), and (c) Instructional Support (three dimensions: Concept Development, Language Modeling, and Quality of Feedback). Each dimension was rated on a 7-point scale: low (1, 2), moderate (3-5), or high (6, 7). The manual  provides detailed indicators of each dimension as well as examples of teacher behaviors and classroom interactions for these ratings. The ratings provide scores for the global quality of teacher-child interactions in the classroom, with a primary focus on the teacher. The CLASS measure has been validated in many countries, such as in Finland (Pakarinen et al., 2010), China (Hu, Fan, Gu, & Yang, 2016), Chile (Leyva et al., 2015), and Germany (von Suchodoletz, Fäsche, Guntzenhauser, & Hamre, 2014). Previous studies have provided evidence of both the structural validity and the predictive validity of the CLASS instrument. Overall, there is evidence of the applicability of the CLASS instrument to educational and cultural contexts different from those of the United States.
In the present study, the 12 observers (two male), all graduate or doctoral students in education or psychology, were carefully prepared with 10 hr of training and 3 hr of live observation practice over a 2-week period. In cases in which the ratings of a pair of observers showed a discrepancy of more than 1 point, extra rating practice in a live classroom situation was required, after which interrater agreement was monitored again. This extra practice was necessary for two of the observer pairs. At the end of the training, the observers' pairwise interrater reliability was .81. Subsequently, all observers who had completed the training were allowed to proceed with the classroom observations (see Pianta et al., 2008).
Each observation session lasted three lessons (3 hr) and began when the school day started (in Finland, the school day in Grade 1 typically begins at 8 a.m. or 9 a.m. and lasts 3 to 5 hr). There were always two observers present in the classroom, and each conducted independent ratings. The observations followed the CLASS manual, meaning that they were completed in 30-min cycles. In other words, the observers first observed for a 20-min period while making notes on indicators on a separate sheet of paper. In the subsequent 10-min period (before beginning the next observation cycle), the observers recorded their coding on the proper scoring sheet. The number of CLASS observation cycles ranged from four to six (M = 5.10, SD = 0.54) according to the core instructional components of the day. The number of observed cycles varied depending on the class timetables, as we mainly observed lessons in literacy, math, and science (whereas lessons in physical education, music, and arts and crafts were typically not observed). For each CLASS dimension, the ratings were averaged across all cycles. The observers' interrater reliability was calculated using intraclass correlation coefficients (ICCs) and was found to be between .76 and .96 (the only exception being for Behavior Management: ICC = .66). In subsequent analyses, a mean score for each dimension was calculated from the ratings of the two observers.

Curriculum Emphases
Teachers were asked to rate 11 items on a scale from 1 to 5 (1 = not at all important, 5 = very important) based on the level of importance they placed on the provided goals of literacy skills instruction. The items were adapted from Byler (1997, 2004), who reported two factors: 1) basic skills beliefs/basic activities, isolated skills and 2) child-centered beliefs/contextualized activities. In the present data, a factor analysis (principal axis factoring with oblimin rotation with a Kaiser normalization) on the 11 items produced the following three factors with eigenvalues greater than 1.0, accounting for 67.29% of the variance (see Table 1): Comprehension and Production Skills (three items), Motivation and Self-Concept (two items), and Decoding and Spelling Skills (four items). The productive writing item had almost equal factor loadings on two factors (see Table 1), but we decided to place it on the Comprehension and Production Skills factor because it made more sense conceptually. Two items (letter-sound correspondence and working habits) were excluded from the final solution because of their low factor loadings. Cronbach's alphas for the summary scores were .68, .68, and .68 for Comprehension and Production Skills, Motivation and Self-Concept, and Decoding and Spelling Skills, respectively.

Work Experience
Teachers were asked to rate their work experience at school on a 6-point scale (0 = none at all, 5 = more than 15 years). Teachers' work experience ranged from less than a year to more than 15 years. Teachers typically had more than 15 years of teaching experience (mode = more than 15 years).

Mothers' Education
Mothers were asked to report their education using questionnaires (T1). The measure of the vocational education level was used in the analyses as a control variable. A total of 6.1% (general population = 6%) of the mothers in the sample had a basic education (9 years of formal education, Grades 1-9); 25.5% (general population = 30%) had a secondary education (high school or vocational school degree, Grades 10-12); 37% (general population = 35%) had a vocational college degree, a polytechnic or bachelor's degree (3 years of education at a college or university); and 35% (general population = 29%) had a master's degree (4 to 5 years of education at s university) or a higher university degree (i.e., a licentiate or doctoral degree). The sample was fairly representative of the Finnish population, although the mothers showed a somewhat higher level of education than the general population (Statistics Finland, 2007). Although they were representative of Finland, the mothers in our sample typically had high levels of education from an international perspective.

Analysis Strategy
The multilevel modeling technique (Heck & Thomas, 2009) was used in the analyses because it enabled variance in the reading skill variable to be differentiated into two components: (a) variation due to differences between classrooms (between-classroom variation) and (b) variation due to individual differences in academic skills (within-classroom variation). Furthermore, it enabled us to enter various predictors at both the classroom level (between level) and the level of the individual children (within level). The analyses were conducted using the following three steps. First, ICCs were calculated at both the beginning (T1) and the end (T3) of Grade 1 to determine the proportion of variance in children's reading skills due to classroom differences (between-classroom variation; Heck & Thomas, 2009). Second, classroom-level correlations between teacher-child interactions and reading skills were calculated. Third, multilevel models, which are represented in Figure 1, were constructed to include teacher-child interactions and curriculum emphases as predictors of reading skills. CLASS scores were included in the models as three latent factors (Emotional Support, Classroom Organization, and Instructional Support) consisting of 10 observable dimensions (see Figure 1). Because the three factors correlated so highly with one another, a second-order factor called Teacher-Child Interactions (i.e., the global quality of teacherchild interactions) was constructed. The final set of analyses examined the associations at the level of domains; that is, the three domains of teacher-child interactions (Emotional Support, Classroom Organization, and Instructional Support) replaced the factor of the global quality of teacher-child interactions in the models described previously.
In the tested model (see Figure 1), children's reading skills at the beginning (T1) and end (T3) of Grade 1, measured at the individual level, were allowed to vary randomly between classrooms (the small filled ovals in the reading skills variables at the individual level, below the dashed line). The random intercepts of the children's reading skills are indicated by large ovals at the classroom level (in Figure 1, see between level, above the dashed line), as these intercepts were continuous latent variables that varied across classrooms. Maternal education level (ICC = .10, p < .01) was analyzed at both the between and within levels. The children's gender and age, however, were only analyzed at the individual level because of a lack of classroom differences (ICCs = .001 and .005 and ps = .98 and .76 for gender and age, respectively). The global quality of teacher-child interactions, teachers' curriculum emphases, work experience, and class size in turn were treated as classroom-level variables. In the tested model (see Figure 1), reading skills at the end of Grade 1 (T3) were regressed on reading skills at the beginning of Grade 1 (T1) at both the individual level and the classroom level. A classroom's typical level of maternal education was entered as a predictor of classroom differences in children's reading skills at the classroom level, whereas gender, age, and maternal education were entered as predictors of individual differences in children's reading skills at the beginning of Grade 1 (T1). Finally, teacher-child interactions and teachers' curriculum emphases at T2 were predicted by class size, work experience, and classroom differences in reading skills at T1, whereas classroom differences in reading skills at the end of the academic year (T3) were predicted by teacher-child interactions, work experience, and teachers' curriculum emphases at T2 when classroom differences at the beginning of the academic year were controlled. In addition, teacher-child interactions at T2 were predicted by teachers' curriculum emphases.

Intraclass Correlations and Descriptive Statistics
The results of the intraclass correlations for reading skills showed that differences between classrooms were statistically significant: In reading skills at school entry 6% (p < .01; T1) of the total variance was due to classroom differences, and in reading skills at the end of first grade 7% (p < .01; T3) of the total variance was due to classroom differences. The rest of the variance in reading skills was due to individual differences within classrooms. In addition, 10% (p < .01) of the total variance in maternal education levels was due to classroom differences. Children's gender and age in months had no statistically significant between-level variance.
Descriptive statistics and correlations between the study variables are shown in Table 2. The means for emotional support dimensions were somewhat lower (M = 5.19) and the means for classroom organization (M = 5.28) and instructional support (M = 4.17) dimensions were considerably higher than those reported in prior literature in the United States (approximately M = 5.50 for emotional support, M = 4.20 for classroom organization, and M = 2.00 for instructional support; Burchinal et al., 2010;Howes et al., 2008;Mashburn et al., 2008;Rimm-Kaufman et al., 2009;Weiland, Ulvestad, Sachs, & Yoshikawa, 2013). Items measuring teacher-child interactions correlated positively with children's reading skills at T3. In addition, teachers' curriculum emphases with respect to comprehension and production skills correlated positively with reading skills (T3).

The Role of Global Teacher-Child Interactions in Children's Reading Skills
The tested model (see Figure 1) did not fit the data well, χ 2 (138, N within = 1,029, N between = 91) = 232.764, p < .01, CFI = .907, RMSEA = .026, SRMR between = .161, SRMR within = .007. The modification indices suggested that the model fit would be improved if the residual correlations between (a) Productivity and Regard for Student Perspectives, (b) Quality of Feedback and Negative Climate, and (c) Concept Development and Regard for Student Perspectives were to be allowed. Following these modifications, the final model (see Figure 2) including only statistically significant predictors had a reasonable fit, although the chi-square test, the SRMR at the between level, and the CFI were somewhat below acceptable values, χ 2 (135, N within = 1,029, N between = 91) = 196.140, p < .01, CFI = .940, RMSEA = .021, SRMR between = .161, SRMR within = .007 (Byrne, 2012). The results (see Figure 2) showed that the higher the global quality of teacher-child interactions in the class, the better the children's reading skills at the end of Grade 1 when we controlled for previous reading skills, curriculum emphases, class size, teachers' work experience, and maternal education levels at the classroom level and previous reading skills, gender, child's age, and maternal education levels at the individual level. A teacher emphasis on comprehension and production skills was associated with the global quality of teacherchild interactions but not with reading skills.
At the within level (the level of individual differences), reading skills at school entry contributed strongly to subsequent skills. Girls exhibited better reading skills at school entry than boys. In addition, children's age and maternal education levels were related to reading skills at school entry, such that older children and children of highly educated mothers had better end-of-grade reading skills than their peers.

CLASS Domain-Specific Effects on Children's Reading Skills
The next step consisted of separate analyses of the three CLASS domains (i.e., Emotional Support, Classroom Organization, and Instructional Support) as predictors of end-of-Grade 1 reading skills in separate models.

Classroom Organization
The model fit the data well, χ 2 (52, N within = 1,029, N between = 91) = 58.276, p = .256, CFI = .992, RMSEA = .011, SRMR between = .177, SRMR within = .021. Furthermore, a latent factor of classroom organization (β = .401, p < .05) was significantly associated with reading skills at the end of Grade 1. A Table 2. Descriptive statistics and correlations between the study variables at the classroom level.  teacher emphasis on comprehension and production skills was associated with the observed quality of classroom organization (β = .43, p < .05) but not with reading skills.

Instructional Support
The model fit the data well, χ 2 (52, N within = 1,029, N between = 91) = 68.152, p = .066, CFI = .980, RMSEA = .017, SRMR between = .166, SRMR within = .032. Instructional support, however, was not significantly related to reading skills at the end of the first school year (β = .14, p > .10). A teacher emphasis on comprehension and production skills was associated with the observed quality of instructional support (β = .55, p < .001) but not with children's subsequent reading skills.

The Role of Curriculum Emphases in Reading Skills
Children's reading skills in the fall were associated with curriculum emphases, such that the higher children's reading skills typical of a classroom at the beginning of Grade 1, the more teachers emphasized comprehension and production skills in their literacy instruction in the spring. Teachers' curriculum emphases were not significantly associated with children's end-of-Grade 1 reading skills. However, a teacher emphasis on comprehension and production skills was related to the observed quality of teacherchild interactions as well as to observed classroom organization and instructional support. The quality of teacher-child interactions and classroom organization in turn were linked to children's end-of-Grade 1 reading skills (standardized estimate of indirect effect = .22, p < .01).

Discussion
The present study contributes to the previous literature by examining whether the quality of teacher-child interactions and teachers' curriculum emphases play a role in the development of children's reading skills during the first school year. The results showed that a high global quality of teacher-child interactions was positively associated with improvements in children's reading skills at the end of the first school year. Investigation of the domain-specific effects further showed that warm and responsive interactions (emotional support) and clear rules and expectations for behavior (classroom organization) were associated with better reading skills at the end of Grade 1. The results also revealed that teacher curriculum emphases were not directly related to children's end-of-year reading skills. However, a teacher emphasis on comprehension and production was associated with the quality of teacher-child interactions, which in turn was linked to reading skills. The present study is among the first to investigate the associations between observed teacherchild interactions, teacher-rated curriculum emphases, and children's reading skills within the same study.
In addition, our study is among the few studies that have so far investigated the links between the observed quality of teacher-child interactions and child outcomes in a European sample of school-age children (see also Cadima et al., 2010).

The Role of Teacher-Child Interactions in Children's Reading Skills
In accordance with Hypothesis 1a, the results showed that a high global quality of teacher-child interactions was associated with high levels of reading skills among children at the end of their first school year, after the children's previous skills were controlled. This result aligns with studies showing that child-adult interactions in their immediate environment are key proximal processes driving child development (Bronfenbrenner & Morris, 2007;Pianta, 1999). Our results are also in accordance with previous findings suggesting that emotionally supportive and sensitive interactions Curby et al., 2009;Guo et al., 2012), well-organized activities and routines, and clear expectations for behavior (Cadima et al., 2010;Ponitz et al., 2009) provide a basis for the beneficial development of children's reading skills . Previous literature has emphasized the importance of wellorganized classrooms (Cadima et al., 2010;Ponitz et al., 2009) and instructional support and language exposure in preschool (Mashburn et al., 2008;Pianta et al., 2008); however, our findings suggest, in line with those of Curby and colleagues (2009) and Guo and colleagues (2012), that a warm and supportive climate and teacher sensitivity are also important for reading skill development. Investigation of the domain-specific effects showed that emotional support and classroom organization, but not instructional support, were significantly associated with reading skills at the end of Grade 1. The results emphasizing the importance of emotional support for reading skill development can be interpreted through at least two theories: attachment theory and self-determination theory. Attachment theory posits that when adults provide emotional support and respond to children's needs contingently, children are more self-reliant and better able to profit from their environment (Bergin & Bergin, 2009;Pianta, 1999). Self-determination theory suggests that teachers can support students' needs for autonomy, competence, and relatedness by, for example, showing caring for and interest in the students, creating a warm and supportive classroom atmosphere, and giving students the freedom to make their own choices (Ryan & Deci, 2000). Recent studies using the self-determination framework have also shown that both teacher autonomy support and structure play an important role in student outcomes (Jang, Reeve, & Deci, 2010;Reeve, 2006). Earlier findings indicate that experiences of comfort and emotional safety in classrooms are important for maintaining interest in reading, especially among students with lower skills (Hamre & Pianta, 2005). Overall, our findings and the extant research in the field suggest that emotionally supportive classroom interactions are critical for maximizing students' motivation and initiative to learn to read in the first school year.
The results showing that reading skills develop better in classrooms with high-quality classroom organization can be interpreted, for example, in the context of the empirical and theoretical work on children's self-regulatory skills (Paris & Paris, 2001;Raver, 2004). Self-regulatory skills have been found to be important for successful learning (Morrison, Ponitz, & McClelland, 2010), and the development of such skills is seen as being highly dependent on external regulations imposed by parents and teachers (Raver, 2004;Rimm-Kaufman et al., 2002). By providing clear rules and expectations for behavior, teachers ensure that time is allocated for learning and provide structure for engaging children in learning effectively . Our findings parallel those of Ponitz et al. (2009), who found a significant relation between well-organized classrooms and reading skills gains among first graders.
In contrast to our expectation (Hypothesis 1d) and to many previous studies (e.g., Burchinal et al., 2010;Mashburn et al., 2008), our results did not show instructional support to be significantly associated with children's reading skills. Although instructionally supportive teachers typically provide children with rich learning opportunities through scaffolding; extending; and consistent, process-oriented feedback , instructional support was not a significant predictor in the present study. One explanation for this result may be that the reading skill test used in the present study measured reading accuracy and speed (i.e., reading fluency), whereas instructional support might be associated with more complicated skills, such as reading comprehension. Another possible reason for the nonsignificant link between instructional support and reading skills may be the lack of a linear association between the constructs; however, there is a threshold after which effective instructional support begins to play a role. Recent studies have provided evidence of curvilinear associations between teacher-child interactions and children's outcomes (Burchinal, Kainz, & Cai, 2011;Burchinal et al., 2010;Weiland et al., 2013). However, it should be noted that the standardized estimate (β = .14) of the association between instructional support and reading skills in the present study was similar in size to those found in previous studies (e.g., Burchinal et al., 2010;Howes et al., 2008;Mashburn et al., 2008), despite not reaching statistical significance.
The results further showed that class size was negatively related to the global quality of teacher-child interactions, whereas teacher experience was not. This suggests that a small class size in first grade is favorable for high global quality of teacher-child interactions. Teachers engage in more individual interactions with students and take more time for individual tutoring in small classes than they do in large classes (Blatchford et al., 2003). One possible explanation for the nonsignificant link between teacher experience and teacher-child interactions is that in the present sample work experience had very limited variance, as most of the teachers had more than 15 years of experience.

The Role of Curriculum Emphases in Reading Skills
We expected that teacher curriculum emphases would predict children's reading skills via high-quality teacher-child interactions (Hypothesis 2a). In line with the theoretical underpinnings of Desimone (2009) and Hamre et al. (2012), the results of the present study showed that a teacher emphasis on comprehension and production skills was indirectly related to children's reading skills: In other words, teacher curriculum emphases were relevant to reading skills only to the extent that they were related to teacher-child interactions. This finding is in line with the results of Sonnenschein et al. (2010), who suggested that teachers' curriculum emphases are related to the quality of interactions in the classroom and, via this correlation, to reading outcomes.
Partly in line with our expectations (Hypothesis 2d), we found that a teacher emphasis on comprehension and production skills was related to a high global quality of teacher-child interactions as well as to high levels of classroom organization and instructional support. This result suggests that the observed quality of teacher-child interactions and teachers' self-rated curriculum emphases go partly hand in hand: Teachers who emphasize higher order literacy skills (i.e., comprehension and production skills) in their instruction seem to be able to provide cognitively stimulating instruction and create classroom environments with well-organized and motivating activities. Teachers who placed more value on listening and reading comprehension and the production of one's own texts than the correctness of reading and spelling were also observed to score highly in teacher-child interaction quality. It may be that these teachers are better able to promote children's thinking skills and conceptual understanding. In a similar vein, Stipek and Byler (2004) found that teachers who reported stronger emphases on social development and higher order thinking skills scored higher on child-centered practices than other teachers. However, the results of the present study showed that teacher emphases on motivation and self-concept skills and decoding and spelling skills were not associated with the observed quality of interactions. One possible reason for these nonsignificant results is that a teacher emphasis on motivation and self-concept, for example, is rather narrow, including only two items. Another explanation can also be the fact that all three curriculum emphases were investigated in the same model, which may have resulted in only the most powerful associations being significant. The correlation analyses, however, showed that an emphasis on motivation and self-concept was significantly related to some of the CLASS dimensions, whereas an emphasis on decoding and spelling skills was not. It may be that a teacher emphasis on decoding and spelling skills reflects more drill-and-practice teaching, which is not reflected in CLASS scores. In addition, a teacher emphasis on decoding and spelling skills may be more relevant at the beginning of the first school year. It should also be noted that the measure of teacher-child interactions was one of global interaction quality within the classroom, not curriculum actions or the instructional quality of literacy as content. Further research is needed to examine the associations between curriculum emphases and teacher-child interactions in more detail.
The results of the present study also showed that a classroom's typical initial level of reading skills was associated with teachers' curriculum emphases: When the children in a classroom had a high typical level of reading skills at the beginning of the first grade, their teachers emphasized comprehension and productive skills more heavily. This finding suggests that teachers actively monitor children's progress and adapt their instructional goals and practices according to the children's academic skills (Nurmi et al., 2012). Such behavior is particularly important in the context of transparent languages, such as Finnish, in which the learning of basic decoding skills tends to happen very quickly during the first school year (Lerkkanen et al., 2004).
Neither teachers' work experience nor class size was related to the teachers' curriculum emphases. This finding suggests that Finnish teachers emphasize curriculum goals in a similar way despite their years of experience and the number of children in the classroom. This finding aligns with the Finnish educational context, in which teachers have highly homogenous and high-quality teacher training. Teachers follow the core curriculum and differences in their emphases seem to be related to children and their skill level. Another possible explanation involves the very limited variance in teacher work experience.
Overall, the results of the present study contribute to the literature by showing that high global quality teacher-child interactions provide a basis for reading skill development when teachers' curriculum emphases and other background factors are taken into account. For policymakers, this result highlights the importance of quality assurance in teacher-child interactions. The implementation of the curriculum within classroom instructional practices (how) appears to have a stronger impact on child outcomes than the content of the curriculum (what; Pianta et al., 2008). Our results also suggest that both preservice and in-service teacher training may benefit from an emphasis on training and reflection as a way to improve future teacher-child interactions. Observations of the quality of teacher-child interactions can be an excellent tool for such training and interventions. In addition to professional development, which involves the acquisition of knowledge and practical skills, analytic skills related to reflecting on one's own teacher-child interactions should be emphasized (Hamre et al., 2012).

Limitations and Future Directions
The present study has certain limitations that need to be taken into account when trying to generalize the findings. First, there was a great deal of missing data regarding classroom observations, and the sample size of the observed teachers was small. This naturally decreases the power of the statistical testing. In addition, although the observed teachers were not found to differ from those who chose not to participate in classroom observations for many of the variables, we were not able to rule out the possibility that these teacher groups differed, for example, in their motivation or teaching practices. We found that the observed teachers had somewhat higher self-reported affection than the other teachers, which may have led to some bias in our findings. This possibility warrants further examination in future studies. On a related note, we do not really know whether the observed teachers fully reflected the 62 who did not agree to participate in observations. Second, although the present study was longitudinal, the analyses were correlational. Therefore, causal inferences cannot be drawn. Third, the experiences children have at home, both prior to and after school entry, also predict their literacy skills (Morrison, Connor, & Bachman, 2006); this relationship should be taken into consideration. Although we controlled for maternal education levels, this measure provides only a very limited estimate of the home environment. Thus, future studies should include a wider range of measures of home literacy environments, such as parental teaching of reading, literacy-related materials available at home, parental sensitivity and autonomy support, and the home-school relationship. Fourth, teacher-child interactions were operationalized via global quality, and we did not have data on children's individual experiences or a detailed analysis of the quality of the literacy instruction. Recent studies have emphasized the need to consider the interactions among children's skills, instruction types, and teacher practices (Cadima et al., 2010;Connor et al., 2009;Nurmi et al., 2012). A particularly important topic for future study is comparing the influence of teacher instruction on the learning of reading among children with diverse skills. Fifth, we assessed teacher curriculum emphases using only a few measurement items, which may have caused certain measurement problems, such as low internal consistencies among variables. Thus, future studies utilizing multidimensional measures for teachers' curriculum emphases are warranted. Moreover, the teacher ratings were collected during the spring term, when most children can already decode accurately in the Finnish language. Furthermore, the constructs used in the present study (i.e., curriculum emphases rather than beliefs or goals) were somewhat different from those used in several previous studies (e.g., Stipek & Byler, 1997, 2004; this warrants caution in the interpretation of the results. Sixth, the work experience variable was very limited, as most of the teachers had worked for longer than 15 years. This may have led to restricted variance and a lack of predictive power regarding this variable and the nonsignificant link between teacher education and teacher-child interactions. Seventh, although rigorous and carefully conducted, the training and the method of ensuring observational reliability in the present study differed somewhat from the typical CLASS training, in which observers score against master codes of videos. Finally, the CLASS instrument focuses on content-neutral teacher-child interactions, not literacy-related teaching quality; thus, future studies might benefit from the use of complementary content-specific observational methods, such as individualized student instruction (Connor et al., 2009), and from taking into account the books and other materials available in the classroom or teachers' use of specific instructional approaches when teaching reading.

Conclusions
Overall, the results of the present study broaden the previous understanding of effective teacher-child interactions and the impact of teachers' curriculum emphases on children's literacy outcomes. The results indicated that a high global quality of teacher-child interactions is associated with children's reading skill development during their first school year. Along with expectations for behavior and wellorganized activities, warm and sensitive teacher-child interactions were found to be influential in children's reading skills development in Grade 1. Teachers who emphasized production and comprehension in literacy instruction were more likely to engage in highly organized activities and cognitively stimulating instruction, which in turn had positive consequences for the development of children's reading skills. Teacher curriculum emphases, however, seemed to be important for children's reading skills only through their effects on high-quality interactions. Therefore, what teachers think might be important only when taking into account how teachers support children's learning through teacherchild interactions. Professional development programs focusing on supportive teacher-child interactions in particular may be important for enhancing teacher effectiveness in promoting students' learning.