View in EDS HTML Full Text PDF Full Text

Which Words Are Special? Identification of 'Sight' Words in Educational Resources

Saved in:
Bibliographic Details
Title: Which Words Are Special? Identification of 'Sight' Words in Educational Resources
Language: English
Authors: Matthew J. Cooper Borkenhagen (ORCID 0000-0002-8245-0130), Lauren P. Schilling (ORCID 0009-0004-3282-5653), Mark S. Seidenberg (ORCID 0000-0001-8519-3259)
Source: Reading Research Quarterly. 2026 61(1).
Availability: Wiley. Available from: John Wiley & Sons, Inc. 111 River Street, Hoboken, NJ 07030. Tel: 800-835-6770; e-mail: cs-journals@wiley.com; Web site: https://www.wiley.com/en-us
Peer Reviewed: Y
Page Count: 25
Publication Date: 2026
Sponsoring Agency: Institute of Education Sciences (ED)
Contract Number: R305B150003
Document Type: Journal Articles
Reports - Research
Descriptors: Sight Vocabulary, Sight Method, Reading Instruction, Word Lists, Reading Skills, Beginning Reading, Spelling, Phoneme Grapheme Correspondence, Word Frequency
DOI: 10.1002/rrq.70077
ISSN: 0034-0553
1936-2722
Abstract: Which words are important for early reading instruction? A standard view holds that certain words should be emphasized early in development because they are used with high frequency and/or contain atypical spelling-sound structure. Such words have been labeled "sight," "trick," "snap," or simply "high frequency" words; we refer to them as "special" words, which is intended to reflect their status as words that have been identified as being of particular instructional importance. The present study examined whether instructional resources such as commonly used curricula and word lists agree on their identity. Understanding the contents of these resources and those like them is important given their prevalence in instruction: teachers rely on wordlists to plan activities to support early word reading skills. We addressed this question using six such resources ranging from the classic Dolch list to modern commercial curricula. Results show substantial disagreement about the designated words and their properties. A total of 973 distinct words are identified in these materials. Only 28 words (3%) appear in all six resources, and over half appear in only a single one (560 words; 56%). Additional analyses demonstrate that the materials differ in terms of a number of word properties including frequency and spelling-sound consistency. Together the results indicate a surprising lack of agreement about which words should be treated as special for instructional purposes. These differences suggest that beginning readers' learning experiences may vary greatly. In the general discussion we describe an alternative method for identifying words on a more principled basis, which would also facilitate comparisons across curricula. This method is based on computational theories of word reading that specify how properties of words such as spelling-sound consistency and word frequency affect learning. Theories of this kind can provide a more systematic basis for identifying words to emphasize in early instruction.
Abstractor: As Provided
Notes: https://github.com/MCooperBorkenhagen/special_words
IES Funded: Yes
Entry Date: 2026
Accession Number: EJ1494630
Database: ERIC
Full text is not displayed to guests.
FullText Links:
  – Type: pdflink
    Url: https://content.ebscohost.com/cds/retrieve?content=AQICAHj0k_4E0hTGH8RJwT4gCJyBsGNe_WN95AvKlDbXJGqwxwFPWentDb_lXilyiL6iDPF9AAAA4zCB4AYJKoZIhvcNAQcGoIHSMIHPAgEAMIHJBgkqhkiG9w0BBwEwHgYJYIZIAWUDBAEuMBEEDAf5skV5La5su4eqeQIBEICBm8WHVSEaqQPWuKDdUmQJ5xb6_5BjSrM5AF2yP78Lg5ILkJf7CTCGIN_ZD4u_HJPeONL3HP813Y9o6h33LLdGKxxyJ3gBaGUlqT9hQcMDnE4XLwP6s9rTGScAkUlsnkFVk-pkyPPraa_dHpe0f4aq7RoFhHexzRYJQC3tU6EOYP7DAxgPnGw6TfpKBidW8bg2ujhjR-BXyCGwHTM7
Text:
  Availability: 1
  Value: <anid>AN0191105817;[nrnu]01jan.26;2026Jan28.02:53;v2.2.500</anid> <title id="AN0191105817-1">Which Words Are Special? Identification of "Sight" Words in Educational Resources </title> <p>Which words are important for early reading instruction? A standard view holds that certain words should be emphasized early in development because they are used with high frequency and/or contain atypical spelling‐sound structure. Such words have been labeled "sight," "trick," "snap," or simply "high frequency" words; we refer to them as "special" words, which is intended to reflect their status as words that have been identified as being of particular instructional importance. The present study examined whether instructional resources such as commonly used curricula and word lists agree on their identity. Understanding the contents of these resources and those like them is important given their prevalence in instruction: teachers rely on wordlists to plan activities to support early word reading skills. We addressed this question using six such resources ranging from the classic Dolch list to modern commercial curricula. Results show substantial disagreement about the designated words and their properties. A total of 973 distinct words are identified in these materials. Only 28 words (3%) appear in all six resources, and over half appear in only a single one (560 words; 56%). Additional analyses demonstrate that the materials differ in terms of a number of word properties including frequency and spelling‐sound consistency. Together the results indicate a surprising lack of agreement about which words should be treated as special for instructional purposes. These differences suggest that beginning readers' learning experiences may vary greatly. In the general discussion we describe an alternative method for identifying words on a more principled basis, which would also facilitate comparisons across curricula. This method is based on computational theories of word reading that specify how properties of words such as spelling‐sound consistency and word frequency affect learning. Theories of this kind can provide a more systematic basis for identifying words to emphasize in early instruction.</p> <p>Keywords: reading development; reading instruction; sight words</p> <p>We examined the overlap in words and properties in the contents of six commonly used word lists designed to support early word reading skills. The resources overlapped very little in their contents, including dramatic variation in terms of the properties of the words included in each. Variation of this kind is important to understand because it suggests that children's experiences differ in terms of their instructional experiences in the reading classroom.</p> <p> <img src="https://imageserver.ebscohost.com/img/embimages/rdk/NRNU/01jan26/rrq70077-toc-0001.jpg?ephost1=dGJyMNXb4kSepq84yOvqOLCmsE6epq5Srqa4SK6WxWXS" alt="rrq70077-toc-0001.jpg" title="." /> </p> <p></p> <hd id="AN0191105817-3">Introduction</hd> <p>A major step in becoming a reader is being able to recognize and understand sufficient words to comprehend simple texts. These include the very common, "high frequency" words that occur in most texts, as well as content words such as "dinosaur" and "skeleton" that are common in texts about specific topics (Adams [<reflink idref="bib1" id="ref1">1</reflink>]). Achieving this level of skill is often called "cracking" or "breaking" the orthographic code. The child understands what there is to learn about print (e.g., how letters combine and represent the sounds of words) and has acquired sufficient knowledge to start reading. Learners can be said to have achieved escape velocity: they can increase their reading skill and acquire other types of knowledge through reading itself, with decreasing reliance on external feedback.</p> <p>Although approaches to instruction vary, they generally assume that beginning readers will need to learn some number of common words through processes other than mere decoding ("sounding out") because of the nature of written English. Many common words can be decoded using knowledge of letter‐sound correspondences ("phonics"). However, teaching these correspondences is time‐intensive and the sheer number of them is daunting (Seidenberg and McClelland [<reflink idref="bib48" id="ref2">48</reflink>]; Venezky [<reflink idref="bib59" id="ref3">59</reflink>]). Rather than delaying reading until a sufficient number of spelling‐sound mappings are learned, instruction typically incorporates memorizing some words. Learning to read words by means other than decoding also seems necessary for the many high‐frequency words whose pronunciations differ from those specified by standard phonics rules (e.g., "have," "said," "was," "were," and "some"). Many of these are function words (prepositions, pronouns, auxiliary verbs, determiners) essential for constructing sentences and which occur in many different contexts in the language.</p> <p>Thus, beginning readers' progress depends on gaining facility with common words, including ones that exhibit atypical spelling‐sound patterns. Some number of words need to be learned through processes other than strict decoding in order to jump‐start reading while knowledge of spelling‐sound mappings develops. This typically requires some sort of memorization, which involves learning associations between the spelling of a word and its primary meaning and/or pronunciation. Historically, there have been a variety of proposals about which words should be treated this way and strategies for teaching them related to long‐standing assumptions about the status of certain words being "sight words" (Dolch [<reflink idref="bib16" id="ref4">16</reflink>]; Fry [<reflink idref="bib20" id="ref5">20</reflink>]; see Colenbrander et al. [<reflink idref="bib12" id="ref6">12</reflink>], for a summary). While this topic has been the subject of renewed focus in recent years (Colenbrander et al. [<reflink idref="bib12" id="ref7">12</reflink>]; Miles et al. [<reflink idref="bib35" id="ref8">35</reflink>]), there remains a lack of consensus about the status of these words in educational programming.</p> <p>Behavioral studies have not been conducted that causally investigate the learning benefits of different sets of common words that represent the natural scale at which children amass print knowledge. Evidence for the differential benefits of ensembles of words at large scale is based on corpus analysis. For example, Solity and Vousden ([<reflink idref="bib53" id="ref9">53</reflink>]) studied a large corpus of words derived from a variety of texts finding that the majority of monosyllabic words (almost 90%) could be named aloud from 62 established sound‐spelling patterns ("grapheme phoneme correspondences") along with two additional morphological "rules" and assuming that the most frequent 89 words could be learned through some other mechanism (such as memorization). This research is conducted within a tradition that assumes that the distributional properties of language (in this case the frequency of letters and sounds in text) influence human cognition directly based on the frequency of occurrence of language elements in the environment (see Vousden [<reflink idref="bib60" id="ref10">60</reflink>], and Anderson and Schooler [<reflink idref="bib2" id="ref11">2</reflink>], for review, discussion, and empirical support for this approach). In the finding described above, the Solity and Vousden ([<reflink idref="bib53" id="ref12">53</reflink>]) study quantified sound‐spelling patterns in a way that is consistent with phonics‐based instructional practices, though other ways of quantifying useful properties are possible. For example, Siegelman et al. ([<reflink idref="bib51" id="ref13">51</reflink>]) quantify distributional properties of letters and sounds using information‐theoretic measures, which have the benefit of measuring the likelihood of a given pattern conditional on the variable contexts in which the pattern occurs (i.e., relative to its uncertainty), which can be defined in different ways (e.g., relative to other letters and phonemes, or aspects of syllabic structure). They find that statistical models that include information‐theoretic measures outperform those based on frequency alone (see also Siegelman et al. [<reflink idref="bib50" id="ref14">50</reflink>], for an application of this approach).</p> <p>In instructional studies, treatment effects are typically investigated regarding the differential benefits of methods of instruction, rather than the relative merits of different ensembles of experiences with printed words. For example, Colenbrander et al. ([<reflink idref="bib11" id="ref15">11</reflink>]) compared instructional methods for teaching children common irregular words (also described as "sight word instruction" in their study; e.g., "heart," "color," "listen"), finding that spelling and mispronunciation correction each benefitted learning above and beyond an alternative method lacking these components (as well as outperforming a control condition). Generalization performance on a different set of common words did not differ across all groups. Their experiment focused on the manner of teaching print vocabulary rather than the contents (i.e., the specific words) utilized in treatment. Furthermore, as is usually the case in studies of word reading instruction of this sort, the words they studied were identified as "irregular" (i.e., rule‐violating, exceptional) as compared to words that are regular (i.e., rule governed).[<reflink idref="bib1" id="ref16">1</reflink>] The rules‐versus‐exceptions dichotomy is contrasted with decades of research on statistical learning that has demonstrated that words in English exist on a continuum of <emph>consistency</emph>: the extent to which a word is pronounced in a way that is similar to other similarly spelled words (Glushko [<reflink idref="bib23" id="ref17">23</reflink>]; Plaut et al. [<reflink idref="bib39" id="ref18">39</reflink>]; Seidenberg and McClelland [<reflink idref="bib48" id="ref19">48</reflink>]; Seidenberg et al. [<reflink idref="bib46" id="ref20">46</reflink>]). This description is compatible with recent findings by Miles et al. ([<reflink idref="bib34" id="ref21">34</reflink>]) that even words classified as "sight words" reliably contain predictable sound‐spelling patterns despite being "irregular" in the standard sense. This alternative theoretical orientation may help advance the educational science in this learning domain (see Seidenberg et al., 2020 for discussion). One important experiment to this end was Apfelbaum et al. ([<reflink idref="bib3" id="ref22">3</reflink>]), which taught first grade children about invariant sound‐spelling patterns in printed words (the pronunciation of the letters that represent the vowel) by systematically varying the consonantal context around the vowel. Their teaching approach did not include instruction about the "rules" that pertain to how letters map onto phonemes in English; rather the tasks required participants to interact with printed words in ways that supported inductive forms of learning about their spelling and pronunciation. They found that more variable sets of words (in terms of consonant contexts) led to more robust patterns of learning and generalization than less variable ones. This follows directly from statistical learning theories: we learn about stable patterns in our environment largely as a function of the variable contexts in which an invariant pattern is observed. This theory is described more fully later in this introduction.</p> <p>Regardless of the reason, lack of consensus about how to approach this learning problem in the classroom leads to a number of important questions about the differences in practices that are oriented towards the same (or very similar) educational outcomes: how to pronounce common words that contain atypical sound‐spelling patterns. These questions ought to be investigated in terms of the resources that support the practices, and should include how exactly approaches vary. This article is oriented towards filling this gap by examining variation in how lists of words are constructed in popular resources used in the reading classroom.</p> <p>In this article, we refer to these words as "special words" for several reasons. First, this name is intended to signal something common to all the resources studied here in that they all identify a set of words that they report to have a unique status among all printed words in English (i.e., they identify a set of "special words"). Second, this name is flexible enough to refer to the lists studied despite (sometimes subtle) differences in the types of words they contain and the rationale for including certain words and not others. This avoids imprecision and confusion associated with using terms that have become variably defined in the literature and in educational practice (e.g., "sight word," "high frequency word," etc.). This allows us to be both theory‐ and curriculum‐neutral in our description of the resources while maintaining the focus on understanding resources in terms of the empirical properties of the words they contain and, to some extent, in terms of the resources' stated reasons and criteria for doing so. Finally, the generic nature of the term is intended to invite thought and discussion in the reading science community in considering current conceptualizing—explicit or implicit—about which experiences with printed words matter when supporting children in early word reading development.</p> <p>In a classic article, Dolch ([<reflink idref="bib16" id="ref23">16</reflink>]) developed a list of 220 "sight words" that he thought essential for beginning readers to learn (for discussion disambiguating instruction and the underlying psychological processes often referred to with the same name, see Ehri [<reflink idref="bib17" id="ref24">17</reflink>], [<reflink idref="bib18" id="ref25">18</reflink>]). The Dolch list (with minor updates) remains a widely used resource. Dolch prescribed that all of these words can be learned by treating them as visual patterns associated with meanings and pronunciations. Treating words this way is potentially time‐consuming because of the substantial practice required by learning to read words by "rote" and without special attention paid to the many ways that printed words overlap in their structure and meaning. Additionally, the Dolch list also contains many words that are decodable if readers learn the relevant phonics rules (e.g., "can," "make," "fast"; see Miles et al. [<reflink idref="bib34" id="ref26">34</reflink>]). Learning rules also takes time, but it reduces the number of words that have to be learned via other forms of association (e.g., instruction about printed words including but not limited to plain memorization). Numerous alternatives to and modifications of the Dolch list have been proposed by scholars, most notably Fry ([<reflink idref="bib20" id="ref27">20</reflink>]). Commercial curricula also incorporate their own lists. The term "sight word" has fallen into disfavor with other labels such as "trick," "snap," and "heart" words employed instead,[<reflink idref="bib2" id="ref28">2</reflink>] each with its own rationale and corresponding instructional method. Although the terms vary, they reflect a shared view that some number of common words need special attention at the onset of reading instruction.</p> <hd id="AN0191105817-4">A Theoretical Framework for Learning to Read Words</hd> <p>Scientific theories of word reading development all agree that learning to read printed words involves, principally, gaining knowledge about the visual identity of the printed form of the word (orthography), associating that knowledge with the corresponding spoken form (phonology), and the word's meaning (semantics). These aspects of processing underlie all the major theories of word reading (Ehri [<reflink idref="bib19" id="ref29">19</reflink>]; Nation [<reflink idref="bib37" id="ref30">37</reflink>]; Perfetti and Hart [<reflink idref="bib38" id="ref31">38</reflink>]; Seidenberg and McClelland [<reflink idref="bib48" id="ref32">48</reflink>]; Share [<reflink idref="bib49" id="ref33">49</reflink>]). Some theories focus on whether, or the extent to which, learning involves developing knowledge about words one at a time (e.g., Share [<reflink idref="bib49" id="ref34">49</reflink>]) and others focus on learning vis‐à‐vis the shared properties across ensembles of printed words (Seidenberg and McClelland [<reflink idref="bib48" id="ref35">48</reflink>]). Regardless, there is general consensus that learning to read requires relating orthographic knowledge to our knowledge of phonology and semantics. Furthermore, there is consensus about our capacity to generalize knowledge within and across these knowledge domains, and do so as a function of experience. For example, even in the context of Ehri's phase theory of reading development (Ehri [<reflink idref="bib17" id="ref36">17</reflink>], [<reflink idref="bib18" id="ref37">18</reflink>], [<reflink idref="bib19" id="ref38">19</reflink>]), knowledge about printed words and the sound‐spelling patterns that comprise them form the basis for transfer to novel word reading experiences within and across phases of development.</p> <p>The connectionist paradigm, typically understood more specifically in terms of the Triangle Framework of reading development (Plaut et al. [<reflink idref="bib39" id="ref39">39</reflink>]; Seidenberg and McClelland [<reflink idref="bib48" id="ref40">48</reflink>]), offers a thorough and mechanistic account of these learning processes. In this framework, learning involves the development of knowledge of patterns of coherent covariation between the featural information that comprise printed words ‐ letters, phonemes, and the features that underlie a word's meaning. This takes place through various forms of perception (e.g., visual word recognition, spoken word recognition) and action (e.g., spelling, spoken word production) related to language that occur throughout development. Learning accumulates also as a function of knowledge that develops in other experiences with words and the world, as in the use of spoken language and the non‐linguistic perceptual experiences that come along with our daily experiences in the world (like seeing a car driving down the street and using the associated spoken label "car," or describing someone's actions of tossing a ball using the verb "throw"). As it does in humans, in this framework knowledge is influenced by the frequency of occurrence of the learning experiences that give rise to it; frequent (and more recent) experiences operate more strongly in memory than those that are less common (and more temporally distal).</p> <p>Importantly, these cognitive and experiential factors are embodied in the Triangle Framework in simulations of learning, which have been performed and studied numerous times in the preceding decades (Chang [<reflink idref="bib8" id="ref41">8</reflink>]; Chang et al. [<reflink idref="bib9" id="ref42">9</reflink>]; Cox et al. [<reflink idref="bib15" id="ref43">15</reflink>]; Harm and Seidenberg [<reflink idref="bib26" id="ref44">26</reflink>]; Monaghan et al. [<reflink idref="bib36" id="ref45">36</reflink>]; Plaut et al. [<reflink idref="bib39" id="ref46">39</reflink>]; Seidenberg and McClelland [<reflink idref="bib48" id="ref47">48</reflink>]; Yang et al. [<reflink idref="bib61" id="ref48">61</reflink>]). Simulation studies have successfully replicated behavioral effects associated with word frequency, orthography‐phonology consistency, and age of acquisition, among other canonical effects important to understanding reading development (Plaut et al. [<reflink idref="bib39" id="ref49">39</reflink>]; Chang et al. [<reflink idref="bib9" id="ref50">9</reflink>]; Seidenberg and McClelland [<reflink idref="bib48" id="ref51">48</reflink>]; see Seidenberg et al., 2023, for a review). Simulations have been utilized to understand information processing and underlying etiology associated with atypical reading development (Harm and Seidenberg [<reflink idref="bib25" id="ref52">25</reflink>]) and in learning associated with language variation (Brown et al. [<reflink idref="bib6" id="ref53">6</reflink>]). The theory and associated computational models have also been applied to questions in education (Powell et al. [<reflink idref="bib40" id="ref54">40</reflink>]; Seidenberg et al. [<reflink idref="bib45" id="ref55">45</reflink>]). It is also worth noting that the Triangle Framework, and connectionist cognitive science more generally, is broadly consistent with several other theories of learning and development, namely statistical learning (Sawi and Rueckl [<reflink idref="bib44" id="ref56">44</reflink>]; Seidenberg and MacDonald [<reflink idref="bib47" id="ref57">47</reflink>]) and dynamical systems theory (Smith and Thelen [<reflink idref="bib52" id="ref58">52</reflink>]; see McClelland et al. [<reflink idref="bib33" id="ref59">33</reflink>], for a discussion). For this reason in this article we will refer to the Triangle Framework and statistical learning in ways that suggest that the theories are largely overlapping.</p> <p>We will discuss our findings in the context of connectionist and statistical learning approaches (i.e., the Triangle Framework) to reading development, providing an interpretation of the data (and future directions) in the discussion vis‐à‐vis this learning theory. It follows from this that we study the educational resources in ways that align with the first principles of learning employed by the theory. Namely, learning to read takes place, at least in part, as a function of the strength of the representation of a word in the reader's mind, which is directly influenced by how frequently the word is encountered (word frequency), how early the word is learned (age of acquisition), the statistical properties of print and speech (consistency, length), and other perceptual (semantic) properties associated with printed and spoken words (e.g., imageability). However, we selected the resources studied in a theory‐agnostic way in the sense that the theoretical underpinnings of each program were not the basis for inclusion. Rather, we selected resources that are used commonly in practice that are representative of approaches to teaching word reading that consider the instructional words' frequency, sound‐spelling properties, or both.</p> <hd id="AN0191105817-5">Our Study of Six Resources</hd> <p>In order to understand variation in practice our study examined the treatment of special words in six commonly used instructional materials. These include popular commercial programs (Fountas & Pinnell's "Phonics, Spelling, and Word Study Guide," McGraw‐Hill's <emph>Wonders</emph>, and <emph>Fundations</emph> from Wilson Language Basics), the Dolch and Fry lists (Dolch [<reflink idref="bib16" id="ref60">16</reflink>]; Fry [<reflink idref="bib20" id="ref61">20</reflink>]), and "Equipped for Reading Success" (Kilpatrick [<reflink idref="bib30" id="ref62">30</reflink>]) a popular book that offers instructional guidance on early reading development (but is not generally considered a formal curriculum). We chose materials that are widely used and represent a variety of approaches and were developed at different times, including both commercial products and materials in the public domain. We will refer to them as "instructional resources" (or simply "resources").</p> <p>The number six is somewhat arbitrary, however we sought to identify a quantity of word lists that would represent a substantial and meaningful sampling from a practical standpoint, while focusing on those we knew to be influential in the field based on reporting from educators. The results of our study are not obtained under the assumption that we can generalize these findings to all possible resources. That is, the number we obtained was not determined based on an estimate statistical power. Rather, we sampled enough such that we could estimate variation across the sample to an extent that would be interesting and important from an expository standpoint.</p> <p>Our primary goal was to identify the points of convergence and divergence regarding the treatment of special words in this sample. We asked: How many words are identified as requiring special instruction, what was the basis for choosing them, and what are their properties? A high degree of convergence across the lists might be predicted because of their shared focus on teaching high‐frequency words and words with atypical structure. However, the words that are identified could differ for many reasons, such as pressure to limit the number of words that must be memorized, and different assumptions about which words contain decodable elements.</p> <p>These questions are important because of long‐standing assumptions about how some words will have to be treated as special in early reading instruction, despite there being little research addressing disagreement about which words are so designated. Agreement across resources would indicate that special words are being treated in a consistent manner, leading to commonalities in learners' experiences. A lack of agreement would suggest that beginning readers' early instructional experiences vary in ways that may affect progress, despite general agreement about the importance of focusing on high‐frequency and "irregular" words. Lack of agreement might also reflect differences in assumptions about early reading that could affect children's progress. To foreshadow the results, these resources exhibit a high degree of variability in the words identified and the properties used to identify them. Although word frequency is a factor identified in all resources studied, there is little agreement about which high‐frequency words are included, and similarly for words with irregular pronunciations. These disagreements reflect the fact that the special words were mainly identified based on intuitions about words and their properties rather than explicit theories of how properties of words affect learning. In the general discussion we describe how word properties relate to learning, and provide a simple method for identifying words in terms of their frequency and the extent to which they contain atypical structure. The method is flexible enough such that it can be used to understand the relationship of different words in terms of essential learning properties. It also provides a way to compare the special words included in the six resources studied here, and could also be used to select special words in the future.</p> <hd id="AN0191105817-6">Method</hd> <p></p> <hd id="AN0191105817-7">Descriptions of Instructional Resources</hd> <p>Six instructional resources were selected because of their popularity and because they address the development of early reading vocabulary, including the role of common words and the treatment of words with atypical pronunciations given their spellings. For curricula that specify words for several grades, we focused on those for 3rd grade students and younger. We only studied materials that are designed for these early elementary grades given the focus on word recognition skill in these early years of formal schooling. Grades 1 through 3 in the United States are often considered the grades in which children are "learning to read" in that the majority of code‐based instruction in primary school happens during these years (Chall [<reflink idref="bib7" id="ref63">7</reflink>]; Leach et al. [<reflink idref="bib32" id="ref64">32</reflink>]). The Dolch ([<reflink idref="bib16" id="ref65">16</reflink>]), Fry ([<reflink idref="bib21" id="ref66">21</reflink>]), and Kilpatrick ([<reflink idref="bib30" id="ref67">30</reflink>]) resources do not differentiate based on grade level, and so all words were included. Note that several of the resources studied here also teach decoding. We do not include in these analyses words taught during decoding instruction (e.g., the words "pat," "pick," and "pet" might be taught when teaching that the phoneme /p/ is associated with the letter "p") given that such instruction focuses on a different set of skills (related to pronouncing printed words exclusively based on the letter‐sound patterns they contain).</p> <p>The Dolch ([<reflink idref="bib16" id="ref68">16</reflink>]) and Fry ([<reflink idref="bib21" id="ref69">21</reflink>]) materials are lists of words included in journal articles that described issues about learning to read and approaches to instruction. The lists, which are widely available on the internet, are utilized in research and in developing educational programs and activities, but are not curricula themselves. The remaining four (Fundations, Wonders, Fountas & Pinnell, and Kilpatrick) are instructional programs that include both lists of words and methods for teaching them. These four programs vary in how comprehensive they are as instructional tools. On one end, Wonders is a large‐scale comprehensive reading curriculum and, at the other end, Kilpatrick ([<reflink idref="bib30" id="ref70">30</reflink>]) is a supplementary instructional manual lacking many of the elements of more formal programs.</p> <hd id="AN0191105817-8">Dolch "Basic Sight Vocabulary"</hd> <p>Dolch ([<reflink idref="bib16" id="ref71">16</reflink>]) presented "a basic sight vocabulary" of 220 words that he considered essential for reading development because they are "...'tool' words that are used in all writing, no matter what the subject" (p. 457). This set of what are often called "function" words was synthesized from several previous lists and supplemented with 95 common nouns. The "Dolch list" in wide circulation usually consists of these 315 words, although sometimes with minor changes. The list includes words with both typical and atypical spelling‐sound correspondences. To be precise, in words with typical correspondences, the pronunciation is one that is commonly associated with the spelling (e.g., "five," "paid"); in words with atypical correspondences, the pronunciation is less commonly associated with the spelling (e.g., "give," "said"). Dolch recommended that the words be printed on small cards that can be "flashed" before the pupil until they can be recognized "instantly and easily," which is a method that leverages rote memorization. However he also noted that if the learner's "sounding" (phonics) is weak additional instruction in that area may be required.</p> <hd id="AN0191105817-9">Fry "Instant Words"</hd> <p>Fry ([<reflink idref="bib20" id="ref72">20</reflink>]) presented a set of 600 high frequency words based on word frequency counts available at the time (Dolch [<reflink idref="bib16" id="ref73">16</reflink>]; Rinsland [<reflink idref="bib43" id="ref74">43</reflink>]; Thorndike and Lorge [<reflink idref="bib58" id="ref75">58</reflink>]) and additional words that the author deemed common based on personal classroom experience. The words were organized into sets of 100, ordered by frequency, and within each set into four groups of 25. The goal was to allow a reading teacher access to groups of words, decreasing in frequency, in a quantity that could be easily managed in an instructional unit. Fry ([<reflink idref="bib21" id="ref76">21</reflink>]) introduced "The New Instant Word List," a revised version of the original set that included fewer words (<reflink idref="bib300" id="ref77">300</reflink>) chosen using newer frequency counts. We have utilized the Fry ([<reflink idref="bib21" id="ref78">21</reflink>]) list in the analyses here.</p> <hd id="AN0191105817-10">Wilson's Fundations "Trick Words"</hd> <p>The Fundations curriculum incorporates extensive instruction about correspondences from spelling to sound and sound to spelling, using methods such as "tapping out" component sounds. The teacher's manual notes that common (i.e., high frequency) words need to be recognized and spelled quickly even if the relevant spelling‐sound correspondences have not yet been taught. These words include both "phonetically regular" and "phonetically irregular" words (i.e., words with typical and atypical mappings). The curriculum designates 193 words in K‐3 as "trick" words that are to be memorized (see Table 1 for the number of words per grade level). These words have irregular spelling‐sound mappings according to most theories. Students are told that these words "do not follow the 'system' of the language...[they] are phonetically irregular".</p> <p>1 TABLE Number of words in each instructional source.</p> <p> <ephtml> <table><thead valign="bottom"><tr><th align="left">Grade</th><th align="center">F&P</th><th align="center">Fundations</th><th align="center">Wonders</th><th align="center">Fry</th><th align="center">Dolch</th><th align="center">Kilpatrick</th></tr></thead><tbody valign="top"><tr><td align="left">K</td><td align="center">39</td><td align="center">27</td><td align="center">40</td><td align="center">–</td><td align="center">–</td><td align="center">–</td></tr><tr><td align="left">1</td><td align="center">42</td><td align="center">64</td><td align="center">155</td><td align="center">–</td><td align="center">–</td><td align="center">–</td></tr><tr><td align="left">2</td><td align="center">97</td><td align="center">84</td><td align="center">166</td><td align="center">–</td><td align="center">–</td><td align="center">–</td></tr><tr><td align="left">3</td><td align="center">134</td><td align="center">18</td><td align="center">79</td><td align="center">–</td><td align="center">–</td><td align="center">–</td></tr><tr><td align="left">Total</td><td align="center">312</td><td align="center">193</td><td align="center">440</td><td align="center">300</td><td align="center">315</td><td align="center">425</td></tr></tbody></table> </ephtml> </p> <p>1 <emph>Note:</emph> Number of unique words are shown. "–" values indicate that the source does not specify words at the level of individual grades. F&P = Fountas & Pinnell.</p> <hd id="AN0191105817-11">Fountas & Pinnell's "Words to Know"</hd> <p>Fountas & Pinnell identify 312 "unique" words they label "words to know". The words are meant to be taught throughout the elementary grades in the program, using specific instructional routines. The words are identified as important to reading success because they are encountered often in print, and the fact that the words are common is referenced explicitly when teaching. For example, students are told that "You see some words many times when you read. You need to learn words that you see many times because they help you read and write" (Grade 3; p. 34). Other properties of words, such as atypical spelling and pronunciation, are treated as secondary and not a focus of instruction. "Discussion points" for words sometimes draw attention to parts of words or similarities between them, but these are optional. For example, for the word "because," the "discussion points" include discussing that the word resembles "before" and "become," or that the word "has a silent 'e' at the end".</p> <hd id="AN0191105817-12">Wonders' High‐Frequency Words</hd> <p>The treatment of special words in the Wonders curriculum is similar to that in Fundations. The curriculum identifies a set of high frequency words that must be prioritized in instruction, many of which contain atypical structure and "cannot be decoded" (Grade 1; p. S7). There are 440 words across K‐3rd grade levels (see Table 1). Words are learned through reading aloud, spelling aloud, and writing. For instructional purposes, high frequency words are organized for each year with words that share spelling patterns grouped together.</p> <hd id="AN0191105817-13">Kilpatrick's "Irregular Words"</hd> <p>Kilpatrick ([<reflink idref="bib30" id="ref79">30</reflink>]) focuses on high frequency words that violate the "grapho‐phoneme regularities of printed English." The instructional materials (Appendix A) specify a set of words that comprise "over 300 of the most common irregular words" (p. 64). The list includes many of the Dolch words, and the author states that students should be able to spell the Dolch words correctly before moving on to the remaining words.</p> <p>Kilpatrick's approach to teaching these words is to enable readers to recognize which parts of the word are regularly or irregularly pronounced. The irregularity often arises from a single letter or letter pair; the examples provided in the text are the "ai" in "said" and the "s" in "island." The author explains that with instruction the child learns to identify the irregular part (p. 58). The remainder of the word is usually regular. Here the teacher's job is to "point out the regular elements within all words." Learning can be enhanced by focusing the child's attention on this part of the word, which the author terms its "phonological framework." The child can then "use the normally performing letter‐sound combinations to 'anchor' [the] irregular word in memory." The text does not specify the regular and irregular elements for all of the included words, but provides examples.</p> <p>Kilpatrick also provides a second technique for teaching some of these words: Irregular words should be taught using an alternative spelling that conforms to how the word would be spelled if it had a regular pronunciation. One example provided is "Wednesday," which should be taught with the alternative spelling "wed‐ness‐day," and the regular pronunciation based on that spelling. The text notes that only some words can be taught this way, but does not indicate which ones. This is similar to a concept sometimes referred to as "set for variability" (Steacy et al. [<reflink idref="bib56" id="ref80">56</reflink>]; Steacy et al. [<reflink idref="bib55" id="ref81">55</reflink>]; Venezky [<reflink idref="bib59" id="ref82">59</reflink>]).</p> <hd id="AN0191105817-14">Data Collection</hd> <p>The lists of special words and other information were taken from the original sources. The words are listed in tables in Dolch ([<reflink idref="bib16" id="ref83">16</reflink>]), Fry ([<reflink idref="bib21" id="ref84">21</reflink>]), and Kilpatrick ([<reflink idref="bib30" id="ref85">30</reflink>]). For the remaining resources, each word taught in each unit of instruction was identified and entered in a database for analysis. This required examining instructional materials in order to extract this information about words taught (i.e., the words weren't identified in a separate table within the resource). The resulting list is comprised of 973 unique words, and 2598 total words. Table 2 provides terms used by each program or author when referring to their special words.</p> <p>2 TABLE Terms for special words in each instructional source.</p> <p> <ephtml> <table><thead valign="bottom"><tr><th align="left">Resource</th><th align="center">Term</th></tr></thead><tbody valign="top"><tr><td align="left">Dolch</td><td align="center">"Sight Word"</td></tr><tr><td align="left">Fountas & Pinnell</td><td align="center">"Word to Know"</td></tr><tr><td align="left">Fry</td><td align="center">"Instant Word"</td></tr><tr><td align="left">Fundations</td><td align="center">"Trick Word"</td></tr><tr><td align="left">Kilpatrick</td><td align="center">"Irregular Word"</td></tr><tr><td align="left">Wonders</td><td align="center">"High Frequency Word"</td></tr></tbody></table> </ephtml> </p> <p>2 <emph>Note:</emph> See methods for descriptions of each construct and discussion of the term.</p> <hd id="AN0191105817-15">Results</hd> <p></p> <hd id="AN0191105817-16">Words Included in Instructional Resources and Amount of Overlap</hd> <p>Do these instructional resources agree about the words identified as needing special attention in early reading? The short answer to this question is no: only 28 of the 973 different words (3%) occur in all 6 resources (Table 3). All are high frequency words, ranking in the top 20% of words by frequency; all would generally be considered to contain atypical spelling‐sound correspondences (e.g., "said" does not rhyme with other "‐aid" words). This overall result indicates that there is not an obvious or consensus set of words treated as special.</p> <p>3 TABLE The 28 words that occur in all instructional sources (with rank frequency).</p> <p> <ephtml> <table><tbody valign="top"><tr><td align="left">the (1)</td><td align="center">what (27)</td><td align="center">into (66)</td><td align="center">has (124)</td></tr><tr><td align="left">to (2)</td><td align="center">are (29)</td><td align="center">know (71)</td><td align="center">many (125)</td></tr><tr><td align="left">of (8)</td><td align="center">do (33)</td><td align="center">come (83)</td><td align="center">been (127)</td></tr><tr><td align="left">said (11)</td><td align="center">one (36)</td><td align="center">put (92)</td><td align="center">again (130)</td></tr><tr><td align="left">was (12)</td><td align="center">were (47)</td><td align="center">where (100)</td><td align="center">only (136)</td></tr><tr><td align="left">is (13)</td><td align="center">from (49)</td><td align="center">who (107)</td><td align="center">does (192)</td></tr><tr><td align="left">have (26)</td><td align="center">would (55)</td><td align="center">water (112)</td><td align="center">both (368)</td></tr></tbody></table> </ephtml> </p> <p>3 <emph>Note:</emph> Rank frequency (in parentheses) is from the Educator's Word Frequency Guide (TASA; Zeno et al. [<reflink idref="bib62" id="ref86">62</reflink>]).</p> <hd id="AN0191105817-17">How Many Words Are in How Many Resources</hd> <p>In contrast to the small number of words common to all the resources, the number of words that are only included in a single resource is high, 526 (54%). These words vary greatly with respect to structural properties such as typicality of spelling or pronunciation, frequency, and meaning. They include relatively uncommon words like "yacht," "sovereign," "spatula," "whom," "etiquette," and "gnat" as well as very commonly used words such as "shoe," "snow," "dry," "last," "seem," and "easy." See Table 4 for a sample of 50 of these words and their frequencies and the supplemental materials for the complete list. The variability among these words speaks to a lack of common criteria for selecting words and word properties.</p> <p>4 TABLE An illustrative sample of 50 unique words that appear in only a single resource.</p> <p> <ephtml> <table><tbody valign="top"><tr><td align="left">next (157)</td><td align="center">reached (473)</td><td align="center">music (862)</td><td align="center">mark (2032)</td><td align="center">pattern (8414)</td></tr><tr><td align="left">cried (189)</td><td align="center">rock (475)</td><td align="center">mountains (1019)</td><td align="center">plant (2706)</td><td align="center">numeral (NP)</td></tr><tr><td align="left">dance (220)</td><td align="center">sea (504)</td><td align="center">plan (1162)</td><td align="center">neighbor (2995)</td><td align="center">products (NP)</td></tr><tr><td align="left">town (224)</td><td align="center">stand (588)</td><td align="center">north (1221)</td><td align="center">state (3088)</td><td align="center">unit (NP)</td></tr><tr><td align="left">told (266)</td><td align="center">covered (593)</td><td align="center">seem (1342)</td><td align="center">travel (3121)</td><td align="center">vowel (NP)</td></tr><tr><td align="left">room (338)</td><td align="center">library (691)</td><td align="center">daughter (1446)</td><td align="center">figure (3312)</td><td align="center">paste (NP)</td></tr><tr><td align="left">king (346)</td><td align="center">step (695)</td><td align="center">waves (1570)</td><td align="center">map (3437)</td><td align="center">acne (NP)</td></tr><tr><td align="left">pulled (375)</td><td align="center">real (723)</td><td align="center">cousin (1705)</td><td align="center">aisle (4539)</td><td align="center">acreage (NP)</td></tr><tr><td align="left">birds (430)</td><td align="center">easy (762)</td><td align="center">usually (1947)</td><td align="center">area (4556)</td><td align="center">actual (NP)</td></tr><tr><td align="left">himself (466)</td><td align="center">breakfast (824)</td><td align="center">hours (2015)</td><td align="center">ache (6030)</td><td align="center">algae (NP)</td></tr></tbody></table> </ephtml> </p> <p>4 <emph>Note:</emph> Rank frequency taken from the TASA is given in parentheses. Words that are not present in that corpus are marked with "NP" ("not present"), indicating that they are very infrequent in texts written for children.</p> <p>Turning to the occurrences of words across resources, Figure 1 (Panel A) shows how many words appear in how many of the resources. The bar on the far left shows that most words occur in only one resource (54%). Much smaller percentages occur in 2–6 resources. For 2–5 resources the percentages range from 7% to 15%. The value drops to 3% for all six, due to the impact of the Kilpatrick materials, which overlap least with other resources. The lack of agreement across resources is not due to a few idiosyncratic items. This is indicated by the fact that the percentages of words shared by either four or five resources are similar to those for two to three (A). Figure 1, Panel B shows these data excluding the Kilpatrick data. Overlap increases but is still quite low (12%–19%). The number of words appearing in all resources increases from 28 to 79 (3%–12%). With the Kilpatrick program excluded (B), levels of overlap are somewhat higher, but there still are not a large number of near‐misses (i.e., high percentages of words in the 4–5 overlap groups).</p> <p> <img src="https://imageserver.ebscohost.com/img/embimages/rdk/NRNU/01jan26/rrq70077-fig-0001.jpg?ephost1=dGJyMNXb4kSepq84yOvqOLCmsE6epq5Srqa4SK6WxWXS" alt="rrq70077-fig-0001.jpg" title="1 Number of words shared by number of resources. (A) Data from all six resources; (B) these data with Kilpatrick program excluded. The leftmost bar in each panel shows the number of words that only appear in a single resource. The rightmost bar shows the words shared by all resources. FP = Fountas & Pinnell." /> </p> <p></p> <p>Figure 2 shows for each resource the proportion of its special words that occur in other resources. Panel A shows the results using all six resources. A few main patterns are apparent. The Dolch, Fry, and Fountas & Pinnell results are very similar, likely because the two later resources relied heavily on the earlier Dolch article. Fundations contains the most words that are present in all other resources: 15% of the Fundations words appear in all six lists; 41% appear in all five lists when excluding Kilpatrick. Kilpatrick's list is the most idiosyncratic, with most of the words (<reflink idref="bib328" id="ref87">328</reflink>, 77%) included in no other resources.</p> <p> <img src="https://imageserver.ebscohost.com/img/embimages/rdk/NRNU/01jan26/rrq70077-fig-0002.jpg?ephost1=dGJyMNXb4kSepq84yOvqOLCmsE6epq5Srqa4SK6WxWXS" alt="rrq70077-fig-0002.jpg" title="2 The proportion of words that occur in 1–6 of the resources (for each resource). (1) indicates words that only occurred in that resource; (6) indicates words that occurred in all resources. For example, the portion of the bar labeled "3" in Panel A for "Dolch" depicts the proportion of words in the Dolch list that appeared in three lists in total (i.e., two other than the Dolch list itself). Panel A shows these data for all six lists, and Panel B shows these data with Kilpatrick words removed. See supplement for these data along with the values for each proportion and the number of words associated with each proportion shown in the graph. FP = Fountas & Pinnell." /> </p> <p></p> <p>Figure 3 provides a comprehensive summary of the distribution of words over resources and the amount of overlap using a Venn Diagram. The 28 words shared across all sources are seen in the center of the figure. The outermost regions show the words that were only found in one resource. Other regions of the figure show the number and percentage of words that overlap in two or more resources. The colors of the outer lines of each region indicate the resources that are used to calculate the region.</p> <p> <img src="https://imageserver.ebscohost.com/img/embimages/rdk/NRNU/01jan26/rrq70077-fig-0003.jpg?ephost1=dGJyMNXb4kSepq84yOvqOLCmsE6epq5Srqa4SK6WxWXS" alt="rrq70077-fig-0003.jpg" title="3 A diagram depicting the overlap in words across resources. The data are the number of words and percentages of all special words included in a source or in multiple sources. Percentages are calculated relative to all words across all resources. Values less than 6 represent less than 1% of all words and are shown as 0%. Sources are indicated by line color (Dolch = brown; Fountas & Pinnell = green; Fry = orange; Fundations = light blue; Kilpatrick = purple; Wonders = red). The color of the shade in an area represents the number of words in a given region (darker = more words). The outer colors of the lines demarcating a region indicate the resources involved. For example, the center of the plot is encircled by all six colors, therefore representing those words that are common to all six programs (28 words; 3% of all words). The figure was generated using the ggVennDiagram package in R (Gao et al. [22])." /> </p> <p></p> <hd id="AN0191105817-21">Summary of Words Included in Each Resource and Their Overlap</hd> <p>To summarize, whereas the concept of treating some words as special is common to all the resources, the words they identify as such are not. Few words appear in all six; over half appear in a single resource. With the Kilpatrick items removed, the percentage of unique words decreases to 39% and there is somewhat greater overlap among the remaining resources. Still, the striking aspect of the data is the degree to which the resources differ in words that are selected.</p> <p>We next consider whether there might be more commonality across resources than the above data suggest. Although they differ in the exact choice of words, perhaps they agree on which types of words should be identified in terms of their properties. In order to examine this possibility we looked at each set of words in terms of a range of properties known to be important in reading development: frequency, age of acquisition, consistency, imageability, and length (number of letters and number of syllables).</p> <hd id="AN0191105817-22">Word Frequency</hd> <p>Early instruction focuses on words that occur frequently in texts. Unlike properties such as number of letters, frequency can only be estimated. This is done by examining the number of times a word occurs in a relevant sample of texts (i.e., a corpus). Previous research has used several corpora for this purpose, which sometimes yield different results (Gries [<reflink idref="bib24" id="ref88">24</reflink>]). We conducted analyses using several measures of frequency, which yielded similar results. Here we present analyses using frequencies from the TASA norms (Zeno et al. [<reflink idref="bib62" id="ref89">62</reflink>]), which have been widely used in other research (analyses employing other frequency measures are included in the Supporting Information). TASA provides grade‐level estimates of word frequency, based on samples of books for different grade levels. Data for texts through 3rd grade were used in order to match the grade levels associated with the instructional resources.</p> <hd id="AN0191105817-23">Standardizing Rank Word Frequency</hd> <p>The following analyses employed rank frequencies, where numerically lower values indicate higher frequencies, which were then standardized using the 19,468 words from the TASA corpus through grade three. This results in a distribution of rank frequency for each word in TASA set in relation to the variability of rank frequency in that corpus (all words from the six resources were present in TASA). The standardized rank frequencies were used to examine how much the scores for words in each resource differ from the normative TASA distribution and to compare the resources to each other. Summary statistics for each resource are provided in Table 5.</p> <p>5 TABLE Frequencies of words in each resource.</p> <p> <ephtml> <table><thead valign="bottom"><tr><th align="left">Source</th><th align="center">Rank</th><th align="center">Rank (<italic>Z</italic>)</th><th align="center">Raw</th></tr></thead><tbody valign="top"><tr><td align="left">Dolch</td><td align="center">1171 (644)</td><td align="center">−1.67 (0.08)</td><td align="center">5311 (12500)</td></tr><tr><td align="left">FP</td><td align="center">1086 (675)</td><td align="center">−1.68 (0.09)</td><td align="center">6344 (13927)</td></tr><tr><td align="left">Fry</td><td align="center">1075 (595)</td><td align="center">−1.69 (0.06)</td><td align="center">5673 (12724)</td></tr><tr><td align="left">Fundations</td><td align="center">1186 (740)</td><td align="center">−1.64 (0.16)</td><td align="center">7175 (16854)</td></tr><tr><td align="left">Kilpatrick</td><td align="center">2157 (579)</td><td align="center">−0.61 (1.09)</td><td align="center">1714 (10024)</td></tr><tr><td align="left">Wonders</td><td align="center">1142 (646)</td><td align="center">−1.67 (0.09)</td><td align="center">6101 (14778)</td></tr></tbody></table> </ephtml> </p> <p>5 <emph>Note:</emph> Means and standard deviations (in parentheses) are shown for rank frequency, standardized rank frequency (Rank Z), and raw frequency values for all words in a given instructional resource for the TASA corpus. Frequencies are standardized based on all words in TASA before subsetting and calculating the mean and spread (SD) by instructional source. As a result, the mean rank frequency shown represents how far a given distribution is from the mean of all words in TASA. FP = Fountas + Pinnell.</p> <p>Figure 4 depicts every word in each of the six lists arranged by increasing frequency, where each word appears as a line. The presence of a black line indicates that the word is included in the program. If it is not in the program, the line is blank. The figure provides visual evidence of the inconsistencies in the selection of words and in the treatment of word frequency in the six resources, abstracting away from the identities of individual words (see supplement for the identities of the words). Inconsistencies across resources are visualized as horizontal discontinuities in lines. Differences in the treatment of word frequency are apparent from the density of black lines at different frequency levels. Dolch, Fry, and Wonders contain high concentrations of very high frequency words (dark upper sections of those columns). Fountas & Pinnell and Fundations contain fewer of these words. The Kilpatrick materials are the obvious outlier, with heavy representation of low frequency words (dark lower portion).</p> <p> <img src="https://imageserver.ebscohost.com/img/embimages/rdk/NRNU/01jan26/rrq70077-fig-0004.jpg?ephost1=dGJyMNXb4kSepq84yOvqOLCmsE6epq5Srqa4SK6WxWXS" alt="rrq70077-fig-0004.jpg" title="4 All words in all resources depicted as tileplot. The plot depicts all words in each of the six resources, arranged in terms of frequency (increasing vertically on the y‐axis). Each word is depicted as a black line within a given resource. A word is shared across all six resources if the black bar extends continuously across all six along the x‐axis. Any line that shows a horizontal discontinuity indicates that the word does not appear in all six resources (which is the case for most words). FP = Fountas & Pinnell." /> </p> <p></p> <hd id="AN0191105817-25">Frequencies of Special Words Compared to Words in General</hd> <p>All instructional resources contain words that are, on average, more frequent (lower standardized rank frequency) than the average word from TASA, as indicated by each mean value for standardized rank frequency being negative (Table 5). This is unsurprising given that all of the resources are intended to contain common words. We also examined a statistical model of each resource's frequency distribution against zero (the TASA mean). Table 6 provides descriptive data and Table 7 shows the model results from for the six resources.</p> <p>6 TABLE Descriptive data for mean frequency for instructional resources.</p> <p> <ephtml> <table><thead valign="bottom"><tr><th align="left">Source</th><th align="center"><italic>N</italic></th><th align="center">Raw</th><th align="center">Rank</th></tr></thead><tbody valign="top"><tr><td align="left">Dolch</td><td align="center">315</td><td align="center">5311 (12500)</td><td align="center">−1.67 (0.08)</td></tr><tr><td align="left">FP</td><td align="center">312</td><td align="center">4965 (12526)</td><td align="center">−1.68 (0.09)</td></tr><tr><td align="left">Fry</td><td align="center">300</td><td align="center">5673 (12724)</td><td align="center">−1.69 (0.06)</td></tr><tr><td align="left">Fundations</td><td align="center">193</td><td align="center">6113 (15291)</td><td align="center">−1.64 (0.16)</td></tr><tr><td align="left">Kilpatrick</td><td align="center">425</td><td align="center">1714 (10024)</td><td align="center">−0.61 (1.09)</td></tr><tr><td align="left">Wonders</td><td align="center">440</td><td align="center">4050 (10752)</td><td align="center">−1.67 (0.09)</td></tr></tbody></table> </ephtml> </p> <ulist> <item>6 <emph>Note:</emph><emph>N</emph> indicates the number of unique words in each wordlist. Frequencies taken from TASA (Zeno et al. [<reflink idref="bib62" id="ref90">62</reflink>]). Raw = raw frequencies. Rank = rank frequencies. Values in parentheses are standard deviations. FP = Fountas & Pinnell.</item> <item>7 TABLE Model outputs by resource for statistical test against mean rank frequency from the Educator's Word Frequency Guide (TASA).</item> </ulist> <p> <ephtml> <table><thead valign="bottom"><tr><th align="left">Source</th><th align="center"><italic>b</italic></th><th align="center"><italic>t</italic></th><th align="center">df</th><th align="center">SE</th><th align="center">95% CI</th><th align="center"><italic>p</italic></th></tr></thead><tbody valign="top"><tr><td align="left">Dolch</td><td align="center">−1.67</td><td align="center">−378.19</td><td align="center">313</td><td align="center">0.00</td><td align="center">[−1.68, −1.67]</td><td align="center">< 0.001</td></tr><tr><td align="left">FP</td><td align="center">−1.68</td><td align="center">−412.47</td><td align="center">441</td><td align="center">0.00</td><td align="center">[−1.68, −1.67]</td><td align="center">< 0.001</td></tr><tr><td align="left">Fry</td><td align="center">−1.69</td><td align="center">−509.33</td><td align="center">298</td><td align="center">0.00</td><td align="center">[−1.69, −1.68]</td><td align="center">< 0.001</td></tr><tr><td align="left">Fundations</td><td align="center">−1.64</td><td align="center">−184.19</td><td align="center">309</td><td align="center">0.01</td><td align="center">[−1.66, −1.62]</td><td align="center">< 0.001</td></tr><tr><td align="left">Kilpatrick</td><td align="center">−0.61</td><td align="center">−10.40</td><td align="center">343</td><td align="center">0.06</td><td align="center">[−0.73, −0.5]</td><td align="center">< 0.001</td></tr><tr><td align="left">Wonders</td><td align="center">−1.67</td><td align="center">−547.20</td><td align="center">800</td><td align="center">0.00</td><td align="center">[−1.68, −1.67]</td><td align="center">< 0.001</td></tr></tbody></table> </ephtml> </p> <p>7 <emph>Note:</emph> Statistical test used was t.test() from base R (R Core Team [<reflink idref="bib41" id="ref91">41</reflink>]). Rank frequencies were <emph>Z</emph> transformed (standardized) prior to test, so resulting coefficients express the difference in standard deviations of rank frequency from TASA between the source mean on rank frequency and the TASA mean (0 due to <emph>Z</emph> transformation). FP = Fountas & Pinnell.</p> <p>Fry tends to contain the highest frequency words as well as the least amount of variability. On the other end is Kilpatrick. This set of words is much lower frequency on average than any of the other resources. The variability in frequency for this set is noteworthy as well. Not only are the averages for Kilpatrick much lower frequency than any of the other instructional resources, but the spread of frequency values for that source is higher. Note that the Kilpatrick distribution is the only one to include words above the mean of words in TASA (i.e., very low frequency words).</p> <p>Figure 5 shows these trends for the six lists. The point estimate is shown for each resource in each panel, with deviations shown as boxes above and below the estimate in Panel A and standard errors in Panels B and C. The scale in Panel A is such that one can see each list's distribution relative to the TASA mean, which is shown as a dashed line (zero due to standardization). Standard deviations are shown in this panel because the small size of the standard errors renders them imperceptible at this scale. The data displayed in this way give a sense of the variation within and across sources. The idiosyncrasies of the Kilpatrick list are also apparent: That resource contains more variable word frequencies and the lowest frequency words are far towards the extreme (i.e., high on the <emph>y</emph>‐axis). Panel B is a zoomed‐in version of this plot, but showing standard errors from the statistical test of each distribution against zero (zero = the TASA mean). This view permits better comparison across the resources and relative to the mean for the six resources (dotted line). Panel C shows the same data shown in Panel B excluding the Kilpatrick data. This more clearly depicts the differentiation of Wonders at one end (reliably containing lower frequency words than the other four) and Fry at the other (tending to contain the highest frequency words). Dolch, Fountas & Pinnell, and Wonders exhibit comparable distributions for word frequency. See Tables 6 and 7 for means and estimates from the model to supplement the graphical depiction in this figure.</p> <p> <img src="https://imageserver.ebscohost.com/img/embimages/rdk/NRNU/01jan26/rrq70077-fig-0005.jpg?ephost1=dGJyMNXb4kSepq84yOvqOLCmsE6epq5Srqa4SK6WxWXS" alt="rrq70077-fig-0005.jpg" title="5 Three depictions of the distribution of word frequency for all resources. In Panel (A), the raw distribution of words for rank frequency (standardized) from the TASA corpus are shown as points (in gray) with point estimates from the statistical model shown as colored bars. Boxes around the point estimates show standard deviations because standard errors are imperceptibly small. All distributions are significantly different than the corpus mean for TASA (0), which is shown as the dashed line. Panel (B) shows the same plot but zoomed in so that the differences across resources can be seen, and Panel C shows the same plot but with the Kilpatrick set removed so that the other points (and bars) can be seen more clearly. (B) and (C) show SEs as error bars around the point estimates. In Panels (A) and (B) the dotted lines show the mean across the six resources, and in Panel C the dot‐dashed line shows the mean for only the five resources shown. FP = Fountas & Pinnell." /> </p> <p></p> <hd id="AN0191105817-27">Coverage of the Most Frequent Words</hd> <p>To what extent do these resources include the most frequent words in grade‐appropriate texts? We examined this by calculating the proportion of the most frequent words in TASA that a given resource contains. For this purpose frequencies were divided into bands: 50, 100, 500, and 1000 words. The results for all six resources follow a common pattern: they tend to exhibit more coverage for the most frequent 50 and 100 words, less so for the most frequent 500 and even less for the most frequent 1000. This general finding is not surprising. However, the resources vary in how they cover words in these bands.</p> <p>Wonders contains the most words, with 440, so it is perhaps unsurprising that it shows greater coverage in all frequency bands. It nonetheless contains only 65% of the most frequent 500 words. By contrast, the Dolch and Fry words obtain almost comparable coverage with fewer words (less than 3/4 of the number of words in Wonders). All three have greater than 75% coverage for TASA on the top 50 and top 100 words and around half of the top 500 words. Fountas & Pinnell exhibit slightly lower levels of coverage than those three, with Fundations demonstrating even less coverage than Fountas & Pinnell. The coverage for Kilpatrick is very low, even for the most frequent 50 words. This fact is notable given that Kilpatrick contains the second most number of words behind Wonders, and over 100 words more than the Dolch list.</p> <hd id="AN0191105817-28">Summary of Word Frequency</hd> <p>The data indicate (see Figure 5) that the words in Wonders, Fry, Dolch, and Fountas & Pinnell have similar frequencies even though they are largely different words. As other analyses have indicated, the Kilpatrick words differ the most from the others, but the Fundations words stand out as well because they include fewer of the words in each frequency band.</p> <p>All resources contain words more frequent than the average frequency from TASA. However, Kilpatrick's set contains many words of lower frequency than the mean frequency from TASA (shown with points falling above the zero line in Figure 5A), a trend that differentiates this program from the rest. Kilpatrick's resource aside, all other resources do in fact contain "high frequency" words when comparing against this normative distribution. This is good and expected given that they advertise themselves as such. Nonetheless, the extent to which they do so varies considerably. These trends are recapitulated when we examine the coverage of the most frequent words in TASA (e.g., Figure 6), though the relative distributions of each program when examining the data in this way shift a bit. Here, Wonders, Fry, and Dolch do very well in their coverage of the top 50 and top 100 words from TASA: they contain almost every one of these words. Wonders contains the best coverage of the top 500 and 1000 words from TASA (65% and 41% of these words, respectively). Analogous to the comparison of average frequency relative to the mean TASA frequency, Fundations and Kilpatrick do least well in coverage of the top 50, 100, 500, and 1000 words. These resources contain only 16% (Fundations) and 1% (Kilpatrick) of the most frequent 1000 words from TASA, which is much lower than the 41% contained by Wonders. It is important to keep in mind that Wonders contains the most words in general, with 440.</p> <p> <img src="https://imageserver.ebscohost.com/img/embimages/rdk/NRNU/01jan26/rrq70077-fig-0006.jpg?ephost1=dGJyMNXb4kSepq84yOvqOLCmsE6epq5Srqa4SK6WxWXS" alt="rrq70077-fig-0006.jpg" title="6 Coverage for all programs of most frequent words from TASA by frequency band. Coverage for each program for each level of most frequent 50, 100, 500, and 1000 words for TASA. Programs exhibit variable coverage at each level for the corpus, with Wonders, Fry, and Dolch showing comparable levels of coverage at one extreme, and Kilpatrick at the other. FP = Fountas & Pinnell." /> </p> <p></p> <p>To summarize, the concept of "high frequency words" is featured prominently in discussions of early reading instruction. The results here suggest that although the concept is familiar, it does not designate a specific set of words as implemented in instructional wordlists studied here. The six resources differ in terms of which words they contain and the frequency of those words.</p> <hd id="AN0191105817-30">Other Word Properties</hd> <p>The following analyses examined other properties of words known to affect reading. Orthography‐to‐phonology consistency refers to how consistently a spelling pattern is pronounced across words (Chee et al. [<reflink idref="bib10" id="ref92">10</reflink>]; Jared [<reflink idref="bib29" id="ref93">29</reflink>]). A word is consistent if its pronunciation is consistent with words with similar spellings. Consistency varies in degree. "Must," for example, is highly consistent with neighbors such as "dust," "gust," and "musk". "Have" is inconsistent because most "‐ave" words are pronounced as in "save." "Save" is less consistent than "must" because one of its neighbors is "have," which is even more influential on its neighbors because it is very common. For these analyses we used the average feedforward consistency using the norms from Chee et al. ([<reflink idref="bib10" id="ref94">10</reflink>]) for body‐rime units across syllables (using word types; see p. 2536), an aggregate measure of the consistency of a word regardless of length. Numerically lower scores indicate lower consistency. Age of acquisition (AoA) is a rating of the age at which words were learned (Kuperman et al. [<reflink idref="bib31" id="ref95">31</reflink>]). Numerically lower scores indicate earlier acquisition. AoA is correlated with word frequency (e.g., <ephtml> <math altimg="urn:x-wiley:00340553:media:rrq70077:rrq70077-math-0001" display="inline" overflow="scroll" xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>ρ</mi></mrow><annotation encoding="application/x-tex">$$ \rho $$</annotation></semantics></math> </ephtml>  = −0.47 in the TASA words included in the analyses here) but accounts for additional variance associated with early experience with words. Imageability (Stadthagen‐Gonzalez and Davis [<reflink idref="bib54" id="ref96">54</reflink>]) is a rating of how much a word evokes a visual image; it is correlated with concreteness but somewhat more sensitive (e.g., a word such as "red" rates low on concreteness but high on imageability). Lower scores indicate lower imageability. Number of letters and number of syllables were also counted and included in these analyses.</p> <p>As with word frequency, the measures of these additional word‐level variables were standardized in the TASA corpus (levels 1–3) in the same way that frequency was standardized (described previously in the section on Standardizing Rank Word Frequency). Due to distributional characteristics (namely that they are normally distributed rather than Zipfian distributed), raw values for these variables were standardized instead of rank values as was the case for frequency. Table 8 provides a description of what low values on each scale can be interpreted as, and example words. Low values are described for these variables because words in all distributions in the data described tend to be on the low end of the scale for each variable (and tend to be below the mean on the variable in the TASA corpus).</p> <p>8 TABLE Description and examples of words for low values for each variable.</p> <p> <ephtml> <table><thead valign="bottom"><tr><th align="left">Variable</th><th align="center">Description</th><th align="center">Example</th></tr></thead><tbody valign="top"><tr><td align="left">Frequency (rank)</td><td align="center">More common in print</td><td align="center">"the"</td></tr><tr><td align="left">AoA</td><td align="center">Learned early in life</td><td align="center">"my"</td></tr><tr><td align="left">Consistency</td><td align="center">Differs in pronunciation to similarly spelled words</td><td align="center">"shall"</td></tr><tr><td align="left">Letters</td><td align="center">Contains few letters</td><td align="center">"I"</td></tr><tr><td align="left">Syllables</td><td align="center">Contains few syllables</td><td align="center">"my"</td></tr><tr><td align="left">Imageability</td><td align="center">Is difficult to create a mental image of</td><td align="center">"is"</td></tr></tbody></table> </ephtml> </p> <p>8 <emph>Note:</emph> Each example provided comes from the lowest end of the distribution for each variable. For example, the word "my" is low on age of acquisition and the word "shall" is inconsistent in pronunciation to other words that share similar spellings (e.g., "ball" and "call").</p> <hd id="AN0191105817-31">Differences in Word Properties by Resource</hd> <p>Table 9 provides summary data and Figure 7 shows the point estimates from statistical tests of each resource against zero (TASA mean) for each variable. See Tables A1–A5 for results of the statistical tests corresponding to the figure. Colors represent the five different variables of interest within each panel and the dotted line refers to the average for the six sources for a given variable. The dashed line shows the mean for all variables in all words in TASA.</p> <p>9 TABLE Descriptive statistics of other word variables for each resource.</p> <p> <ephtml> <table><thead valign="bottom"><tr><th align="left">Source</th><th align="center">AoA</th><th align="center">Consistency</th><th align="center">Letters</th><th align="center">Syllables</th><th align="center">Imageability</th></tr></thead><tbody valign="top"><tr><td align="left">Dolch</td><td align="center">−0.66 (0.66)</td><td align="center">−0.01 (1.45)</td><td align="center">−1.37 (0.53)</td><td align="center">−1.06 (0.38)</td><td align="center">−0.44 (1.33)</td></tr><tr><td align="left">FP</td><td align="center">−0.70 (0.61)</td><td align="center">−0.24 (1.45)</td><td align="center">−1.16 (0.69)</td><td align="center">−0.88 (0.55)</td><td align="center">−1.09 (0.99)</td></tr><tr><td align="left">Fry</td><td align="center">−0.69 (0.63)</td><td align="center">−0.13 (1.47)</td><td align="center">−1.31 (0.58)</td><td align="center">−0.98 (0.50)</td><td align="center">−0.85 (1.10)</td></tr><tr><td align="left">Fundations</td><td align="center">−0.85 (0.61)</td><td align="center">−0.80 (1.47)</td><td align="center">−1.08 (0.77)</td><td align="center">−0.77 (0.68)</td><td align="center">−1.10 (1.05)</td></tr><tr><td align="left">Kilpatrick</td><td align="center">−0.29 (0.89)</td><td align="center">−0.69 (1.39)</td><td align="center">−0.76 (0.71)</td><td align="center">−0.62 (0.70)</td><td align="center">−0.43 (1.12)</td></tr><tr><td align="left">Wonders</td><td align="center">−0.74 (0.62)</td><td align="center">−0.44 (1.48)</td><td align="center">−1.20 (0.66)</td><td align="center">−0.91 (0.54)</td><td align="center">−0.94 (1.01)</td></tr></tbody></table> </ephtml> </p> <ulist> <item>9 <emph>Note:</emph> Means and SDs (in parentheses) are shown for standardized values for all words in a given instructional resource. Standardization takes place in TASA (see text). Therefore the mean value for a resource for a given variable indicates the average difference from the TASA mean.</item> <item>10 Abbreviations: FP = Fountas & Pinnell.</item> </ulist> <p> <img src="https://imageserver.ebscohost.com/img/embimages/rdk/NRNU/01jan26/rrq70077-fig-0007.jpg?ephost1=dGJyMNXb4kSepq84yOvqOLCmsE6epq5Srqa4SK6WxWXS" alt="rrq70077-fig-0007.jpg" title="7 Estimates for a range of word variables for all resources. Point estimates and SEs are shown. Data are calculated over standardized measures within TASA, therefore the position of the point for a source for a variable can be interpreted as the average difference from the mean of all words within TASA (0; red dashed line) for that variable in terms of standard deviation units. The mean on a given variable across all sources is shown as a gray dotted line. FP = Fountas & Pinnell. Estimates were derived using t.test() in base R." /> </p> <p></p> <p>Scores for the resources almost all land below the TASA mean for each variable (below the dashed line). All resources contain words that are learned earlier than the average word in TASA (lower in AoA), are less consistent (lower consistency; except for Dolch), are less imageable, have fewer letters, and fewer syllables. Additionally, each resource rarely falls on the mean for all resources (i.e., has a bar that falls on the dotted line), which suggests noteworthy variability across resources for each variable.</p> <p>Fundations, Kilpatrick, and Dolch are, in general, associated with more extreme positions relative to the group mean across each of the variables—though in different directions. Fundations favors words with lower age‐of‐acquisition, lower consistency, lower imageability, and more syllables than the rest. Kilpatrick tends towards words with higher AoA (i.e., learned later) and more letters. Dolch contains words that are more consistent and more imageable, though shorter (in terms of letters and syllables). Fry, Fountas & Pinnell, and Wonders tend to be more moderate than the others, exhibiting more instances where their distribution straddles the mean for all resources (and each showing no instance of being the extreme value on any given variable).</p> <p>Except for the Dolch words' distribution for consistency, differences between the source distribution on each word‐level variable relative to the normative distribution (TASA) are all statistically significant based on a statistical test of the mean against zero (paralleling the tests conducted previously on the frequency distributions).</p> <hd id="AN0191105817-33">Summary of Other Word Properties</hd> <p>In summary, we asked how the words in these resources compared in terms of other properties of words known to affect learning to read, again finding meaningful variation across lists. This was true along all variables studied, and was greater than the variation found with word frequency.</p> <p>Regarding spelling‐sound consistency, the resources split into two subgroups: those that contain words that are below the mean for the six resources (Fundations, Wonders, Kilpatrick) and those that are above that mean (Dolch, Fry, and Fountas & Pinnell). Fundations and Kilpatrick are most clearly differentiated at one extreme. These two programs contain words that are in general more inconsistent than the others. This tendency fits with the characterization of the Fundations and Kilpatrick programs, whose instruction tends to focus on the idiosyncratic print structure of words in English. Interestingly, Fundations targets words that are both irregular and frequent, where Kilpatrick's explanation focuses primarily on the irregularity of the words selected (see the Methods section on this program for further description). Nonetheless, Fundations contains words that tend to be more inconsistent than those in Kilpatrick.</p> <p>Fundations and Kilpatrick also occupy outlier positions for age of acquisition, though in this case they are on opposite ends: Fundations contains words that are on average the lowest AoA words, and Kilpatrick contains the highest AoA words. The other four programs cluster together, right around the average value for AoA for the six resources. Note that for Fundations, the tendency to contain words that are low in age of acquisition but contain inconsistent print‐speech structure characterizes the potential learnability of that program's words. Their words, while idiosyncratic from the perspective of print and speech, are more likely to be words that the child already knows from spoken language. By contrast, Kilpatrick's words are both inconsistent and are higher in age of acquisition. This will cause words in that set to be more difficult to learn. The words will have fewer neighbors and are less likely to be known from speech by the child. This tendency is compounded by their relative infrequency, as described above.</p> <hd id="AN0191105817-34">Frequency and Consistency</hd> <p>The final set of analyses examine the intersection of frequency and consistency. This is particularly relevant to decisions about early reading instruction because of the tendency to focus on words that are common and/or have atypical pronunciations given their spellings (i.e., words that are low in consistency). It is important to keep in mind that these properties are related (Seidenberg and McClelland [<reflink idref="bib48" id="ref97">48</reflink>]). Inconsistent words tend to be high frequency words (e.g., "said," "give," "have") but some are lower in frequency (e.g., "aisle," "pint," "plaid"). Words that are inconsistent but used frequently are easier to learn than inconsistent words that are used less often. The two concepts are also related in a more important sense: consistency refers to a type of frequency, just calculated at a different level of structure within a printed word (see more on this point in the discussion later). The following analyses provide a way to visualize this relationship, and how it plays out in the six resources.</p> <p>A simple approach that lends itself well to visual interpretation is placing a set of words in the two‐dimensional space defined by the two variables. The <emph>x</emph>‐axis is ordered in terms of frequency (further right indicates less frequent) and the <emph>y</emph>‐axis is ordered in terms of consistency (further up the axis indicates more typical structure). The space can be divided into quadrants (Figure 8A): inconsistent and frequent words are commonly the target of instructional emphasis; consistent and frequent words may be included less often because they can be pronounced using phonics rules; consistent and infrequent words are also correctly specified by phonics rules and also used less often; and inconsistent and infrequent, uncommon words with atypical structure that can be learned later, but only when necessary.</p> <p> <img src="https://imageserver.ebscohost.com/img/embimages/rdk/NRNU/01jan26/rrq70077-fig-0008.jpg?ephost1=dGJyMNXb4kSepq84yOvqOLCmsE6epq5Srqa4SK6WxWXS" alt="rrq70077-fig-0008.jpg" title="8 Set of words from all programs organized into quadrants defined by frequency and consistency. (A) A key for the quadrants defined by rank word frequency and rank spelling‐sound consistency. (B) The values for all words in the six programs (darker data points indicate more words around that value). The frequency and consistency variables are plotted using their index (rank) in the distribution rather than their raw values, such that frequency is plotted high to low (with the highest frequency word being furthest to the left on the x‐axis) and consistency is plotted low to high (with the lowest consistency word being furthest down on the y‐axis). Quadrants are defined by the median value for each variable (rank of 444). Higher frequency inconsistent words are highlighted in the lower left quadrant." /> </p> <p></p> <p>We examined the distribution of each resource's words in the space defined by these two variables. Figure 8B shows the locations of all words from the six resources. 187 of the words (19%) fall into the lower left <emph>frequent but inconsistent</emph> quadrant. The distribution of words is spread considerably across the four quadrants. Figure 9 shows where each of the resources falls in terms of these four quadrants, exhibiting wide variation across resources, with the darkness of points corresponding to the density of words. These differences are quantified in the next section which describes the amount of coverage of the target (lower left) quadrant of each figure for the six resources.</p> <p> <img src="https://imageserver.ebscohost.com/img/embimages/rdk/NRNU/01jan26/rrq70077-fig-0009.jpg?ephost1=dGJyMNXb4kSepq84yOvqOLCmsE6epq5Srqa4SK6WxWXS" alt="rrq70077-fig-0009.jpg" title="9 Words organized into quadrants defined by frequency and consistency broken out by program. Words (shown in hexagonal bins, which bin across proximal points) for all resources are shown in two‐dimensional space defined by frequency and consistency. Darker hexbins indicate more points. The dimensions of the space are defined by all words in the TASA sample (grade levels 1–3)." /> </p> <p></p> <hd id="AN0191105817-37">Words in the Lower Left Quadrant of Frequency and Consistency</hd> <p>All of the resources include some of the 187 higher frequency‐higher inconsistency words (words contained in the lower left quadrant). Table 10 shows the proportion of coverage of the quadrant for each resource (Column 1) as well as the proportion of the resource the words in the target quadrant represents for that resource (Column 2). Wonders has the highest coverage, covering the vast majority of the target quadrant (0.89), but it also has the highest number of special words overall (440 words). The distribution of words in that program spreads across all quadrants of the figure, including many words in the "infrequent and inconsistent" category. Compare Wonders to Fry, for example, which contains well more than half the target words (0.66), and whose words are less spread across the four quadrants in general. Similar to other results, Kilpatrick has the lowest coverage, despite containing a large number of words (2nd most, behind only Wonders).</p> <p>10 TABLE Coverage of programs for words in lower left quadrant.</p> <p> <ephtml> <table><thead valign="bottom"><tr><th align="left">Source</th><th align="center">Proportion of quadrant</th><th align="center">Prop. of resource total</th></tr></thead><tbody valign="top"><tr><td align="left">Dolch</td><td align="center">0.58</td><td align="center">0.35</td></tr><tr><td align="left">FP</td><td align="center">0.67</td><td align="center">0.40</td></tr><tr><td align="left">Fry</td><td align="center">0.66</td><td align="center">0.41</td></tr><tr><td align="left">Fundations</td><td align="center">0.49</td><td align="center">0.48</td></tr><tr><td align="left">Kilpatrick</td><td align="center">0.29</td><td align="center">0.13</td></tr><tr><td align="left">Wonders</td><td align="center">0.89</td><td align="center">0.38</td></tr></tbody></table> </ephtml> </p> <ulist> <item>11 <emph>Note:</emph> Words in lower left quadrant are those that fall below the median on both axes. The total number of words in that quadrant is 187. The first column shows the proportion of that quadrant that each resource covers. The second column shows the proportion of each resource represented by their words in the lower left quadrant. For example, Wonders has 440 words. 166 of them fall in the lower left quadrant, which equals 0.38.</item> <item>12 Abbreviations: FP = Fountas & Pinnell.</item> </ulist> <hd id="AN0191105817-38">Choosing Words Based on Frequency and Consistency</hd> <p>All of the evidence we have presented points to a general conclusion: The methods that are used to identify words that merit special instructional attention yield different results rather than converging on a consensus set of items. The six resources identify different numbers of words with differing properties. In this section we offer an alternative method for choosing such words based on quantifiable properties that are highly relevant for early reading.</p> <p>Modern theories of learning have identified frequency and consistency as properties that affect learning to read and skilled word reading (Houghton and Zorzi [<reflink idref="bib27" id="ref98">27</reflink>]; Plaut et al. [<reflink idref="bib39" id="ref99">39</reflink>]; Seidenberg and McClelland [<reflink idref="bib48" id="ref100">48</reflink>]). Words that are more frequent tend to be read more quickly (Balota et al. [<reflink idref="bib4" id="ref101">4</reflink>]; Jared et al. [<reflink idref="bib28" id="ref102">28</reflink>]). Consistency has a similar effect in that words that are higher in consistency (i.e., have more words that are similar in the letters they contain and their pronunciation) are also named faster (Glushko [<reflink idref="bib23" id="ref103">23</reflink>]; Jared et al. [<reflink idref="bib28" id="ref104">28</reflink>]). Both properties are graded, such that words fall on a continuum of both frequency and consistency (see Seidenberg et al. [<reflink idref="bib46" id="ref105">46</reflink>], 38–39 for discussion) and their effects on word naming are also graded. The relationship of these effects is the result of the properties being related; consistency is the result of the frequency with which orthographic and phonological patterns appear across words in the language. As a result, it seems important for children to be able to read very common words, including ones with atypical pronunciations given their spellings (low consistency). However, choosing words for instruction based on one of these properties lets the other property vary. It would therefore be helpful to have a way to take both factors into account in selecting words. That might also permit both dimensions to be covered using a smaller set of words, which would be desirable because instruction about printed words is typically time‐consuming.</p> <hd id="AN0191105817-39">A Graded Measure of Frequency and Consistency</hd> <p>We have provided a graphical analysis that demonstrates the organization of words in terms of frequency and consistency by arranging words in their respective quadrants based on these two properties. However, a method of quantifying each word with a single metric that considers both would be helpful. For this purpose, we used the distance (here using the Euclidean norm) of each word from the origin in the space defined by frequency ( <ephtml> <math altimg="urn:x-wiley:00340553:media:rrq70077:rrq70077-math-0002" display="inline" overflow="scroll" xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>f</mi></mrow><annotation encoding="application/x-tex">$$ f $$</annotation></semantics></math> </ephtml> ) and consistency ( <ephtml> <math altimg="urn:x-wiley:00340553:media:rrq70077:rrq70077-math-0003" display="inline" overflow="scroll" xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>c</mi></mrow><annotation encoding="application/x-tex">$$ c $$</annotation></semantics></math> </ephtml> ), using their ranks.[<reflink idref="bib3" id="ref106">3</reflink>] The distance between the two points in this space is denoted as <ephtml> <math altimg="urn:x-wiley:00340553:media:rrq70077:rrq70077-math-0004" display="inline" overflow="scroll" xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>d</mi><mfenced close=")" open="("><mrow><mi>c</mi><mo>,</mo><mi>f</mi></mrow></mfenced></mrow><annotation encoding="application/x-tex">$$ d\left(c,f\right) $$</annotation></semantics></math> </ephtml> . <ephtml> <math altimg="urn:x-wiley:00340553:media:rrq70077:rrq70077-math-0005" display="inline" overflow="scroll" xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>n</mi></mrow><annotation encoding="application/x-tex">$$ n $$</annotation></semantics></math> </ephtml> is defined as 2 given that we are working in 2 dimensions, where <ephtml> <math altimg="urn:x-wiley:00340553:media:rrq70077:rrq70077-math-0006" display="inline" overflow="scroll" xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>i</mi></mrow><annotation encoding="application/x-tex">$$ i $$</annotation></semantics></math> </ephtml> is summed over the two dimensions and then a square root is applied. <ephtml> <math altimg="urn:x-wiley:00340553:media:rrq70077:rrq70077-math-0007" display="block" overflow="scroll" xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>d</mi><mfenced close=")" open="("><mrow><mi>c</mi><mo>,</mo><mi>f</mi></mrow></mfenced><mo linebreak="goodbreak">=</mo><msqrt><mrow><munderover><mo>∑</mo><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><mrow><mi>n</mi><mo>=</mo><mn>2</mn></mrow></munderover><msup><mfenced close=")" open="("><mrow><msub><mi>c</mi><mi>i</mi></msub><mo linebreak="goodbreak">−</mo><msub><mi>f</mi><mi>i</mi></msub></mrow></mfenced><mn>2</mn></msup></mrow></msqrt></mrow><annotation encoding="application/x-tex">$$ d\left(c,f\right)=\sqrt{\sum \limits_{i=1}^{n=2}{\left({c}_i-{f}_i\right)}^2} $$</annotation></semantics></math> </ephtml></p> <p>This method identifies a different set of words than if they are chosen on the basis of frequency or consistency alone, and provides a way of quantifying what is otherwise a visual pattern demonstrated in the quadrant‐based analyses described previously. Table 11 shows the top 10 words using this measure (see Table A6 for the top 50). Words with high frequency and low consistency are positioned towards the top of this distribution.</p> <p>11 TABLE Top ranked words based jointly on frequency and consistency.</p> <p> <ephtml> <table><thead valign="bottom"><tr><th align="left">Rank</th><th align="center">Word</th><th align="center">Dist. (Euclidean)</th><th align="center">Freq. (high‐low)</th><th align="center">Constncy. (low‐high)</th></tr></thead><tbody valign="top"><tr><td align="left">1</td><td align="center">to</td><td align="center">5.39</td><td align="center">2</td><td align="center">5</td></tr><tr><td align="left">2</td><td align="center">is</td><td align="center">20.62</td><td align="center">13</td><td align="center">16</td></tr><tr><td align="left">3</td><td align="center">his</td><td align="center">24.76</td><td align="center">18</td><td align="center">17</td></tr><tr><td align="left">4</td><td align="center">was</td><td align="center">29.55</td><td align="center">12</td><td align="center">27</td></tr><tr><td align="left">5</td><td align="center">are</td><td align="center">31.02</td><td align="center">29</td><td align="center">11</td></tr><tr><td align="left">6</td><td align="center">have</td><td align="center">32.80</td><td align="center">26</td><td align="center">20</td></tr><tr><td align="left">7</td><td align="center">do</td><td align="center">33.54</td><td align="center">33</td><td align="center">6</td></tr><tr><td align="left">8</td><td align="center">what</td><td align="center">39.62</td><td align="center">27</td><td align="center">29</td></tr><tr><td align="left">9</td><td align="center">said</td><td align="center">47.30</td><td align="center">11</td><td align="center">46</td></tr></tbody></table> </ephtml> </p> <p>13 <emph>Note:</emph> Rankings derived from Euclidean distance of frequency and consistency (ranks) relative to the origin (0, 0). The calculated distance is also provided. All words for all resources were considered when making the calculation. Consistency is arranged such that a low rank means low consistency. Frequency is arranged so that a low rank means high frequency.</p> <p>The quadrant‐based results limit the number of words to those that fall on the median of each axis. Discretizing the words into four groups (quadrants) aids a visual interpretation, but is somewhat arbitrary. Using the distance from point 0, 0 we can examine how each program falls within sets of words defined by this distance, using several different quantities. Here we consider <ephtml> <math altimg="urn:x-wiley:00340553:media:rrq70077:rrq70077-math-0008" display="inline" overflow="scroll" xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>n</mi><mo>∈</mo><mfenced close="}" open="{"><mrow><mn>20</mn><mo>,</mo><mn>50</mn><mo>,</mo><mn>100</mn><mo>,</mo><mn>250</mn></mrow></mfenced></mrow><annotation encoding="application/x-tex">$$ n\in \left\{20,50,100,250\right\} $$</annotation></semantics></math> </ephtml> . Figure 10 shows the top 50 words using this approach for reference, and Figure 11 shows the data for all values of <emph>n</emph>.</p> <p> <img src="https://imageserver.ebscohost.com/img/embimages/rdk/NRNU/01jan26/rrq70077-fig-0010.jpg?ephost1=dGJyMNXb4kSepq84yOvqOLCmsE6epq5Srqa4SK6WxWXS" alt="rrq70077-fig-0010.jpg" title="10 Top 50 words organized jointly by rank frequency and consistency using Euclidean distance measure. The top 50 words using the distance from origin (see list in Table 11 and expanded table in Appendix A) plotted in terms of frequency (high to low) and consistency (low to high). The rank according to this distance is provided before each word. The word closest to the origin is "to," representing the word that is most frequent and atypical when both properties are considered together." /> </p> <p></p> <p> <img src="https://imageserver.ebscohost.com/img/embimages/rdk/NRNU/01jan26/rrq70077-fig-0011.jpg?ephost1=dGJyMNXb4kSepq84yOvqOLCmsE6epq5Srqa4SK6WxWXS" alt="rrq70077-fig-0011.jpg" title="11 Coverage by program of top words defined jointly by frequency and consistency. The proportion of n furthest words from origin in the two‐dimensional space organized by frequency and consistency for values of n = 20, 50, 100, and 250. Panels are arranged left to right from greatest to least overall proportion covered (averaging across proportions of all values of n for a given program). FP = Fountas & Pinnell." /> </p> <p></p> <p>Wonders provides the best coverage across all values of <emph>n</emph>, covering all top 20 words by distance, 98% of the top 50 words, 94% of the top 100 words, and 88% of the top 250 words. Both the Fry and Dolch resources have perfect coverage of the top 20 words, and high coverage of the top 50 words (96% and 92% respectively). Fundations and Kilpatrick lists occupy the other end of the extreme, with Kilpatrick ranking last in all proportions across values for <emph>n</emph>. This is somewhat surprising given the focus of Kilpatrick's resources on irregular words and that Kilpatrick's list is quite large (425 total words). However, because the list contains relatively infrequent words, the coverage of the sets of top 20, 50, 100, and 250 words remains quite low (especially relative to the other programs). The Fundations list ranks next‐lowest overall. While Fundations contains 95% of the top 20 words defined by this variable, its coverage of the top 50, 100, and 250 words is low compared to the others.</p> <hd id="AN0191105817-42">Summary of Frequency and Consistency</hd> <p>The six programs studied here differ in the extent to which they jointly consider frequency and consistency in selecting instructional words. This can be seen in terms of examining the lower left quadrant of their bivariate distribution as well as examining the top 20, 50, 100, and 250 words in the joint rank distribution (defined by the distance from the origin, point 0, 0). Wonders shows strong coverage of the top ranking words the join frequency‐consistency distribution, as do Fry and Dolch. It is noteworthy that the Fry and Dolch wordlists have high coverage here because they also tend to contain more consistent words on average (e.g., see Figure 7). Relatively low coverage is achieved by Fundations and Kilpatrick, again demonstrating that these two resources differ from the others in terms of the properties of the words they contain. Wonders contains the largest number of words overall, and therefore holds an advantage over the other programs in terms of the likelihood of having strong coverage on these and other variables examined. At the other end of the extreme, Kilpatrick's resource contains a large number of atypical words in terms of spelling‐sound structure, but also tends to contain infrequent words. So these results aren't as surprising. Fundations' relatively low coverage here stands out, however. Their program focuses on words that are "tricky" due to their spelling‐sound structure in addition to targeting words based on their (high) frequency. However, Fundations contains only 45% of the top 250 words when considering frequency and inconsistency ranked jointly (by distance).</p> <hd id="AN0191105817-43">General Discussion</hd> <p>Words are fundamental units in language and in reading instruction. The instructional resources we have analyzed all have in common that certain words in English are prioritized for teaching, all focusing on some combination of (a) the words' structural properties in terms of letters and sounds and (b) the words' frequencies. There is noteworthy disagreement among programs both about which words should be included and about those properties that matter most in selecting words. Only 3% of the words across all unique words are present in all six resources, and most words belong to a single source and no others (56% of words). Programs vary considerably in terms of the properties of the words they contain. Some select the most frequent words in print, like the Fry ([<reflink idref="bib21" id="ref107">21</reflink>]) list, which contains words that are on average 1.69 standard deviations more frequent than the average of words from the TASA corpus. Kilpatrick, on the other hand, tends to select words that are far less common (much closer to the average frequency of words from TASA; 0.68 SDs in frequency).</p> <p>Considerable variation is also seen for other lexical variables: age of acquisition, consistency, imageability, number of letters, and number of syllables. If educational approaches agreed on the properties of the language that matter for instruction, we would likely not see such variability across the different measures considered in the analyses here.</p> <p>We have also demonstrated a method that considers word frequency and consistency jointly, given the relevance of these two variables in selecting words for early print vocabulary. The method is explained both graphically (see Figures 8–10) and quantitatively (see Figure 10, Table 11, and Appendix A). This method describes where a word fits in the frequency‐consistency distribution, and measures a word's position from the extreme point in the distribution (Figure 10 and Table 11). The approach offers a reproducible method with a concrete (visual) interpretation. When comparing programs on the distribution of words using this measure (Figure 11), we see additional variability. Again the Fry, Dolch, and Wonders lists tend toward one extreme (selecting words that are highly ranked in frequency‐consistency) where Fundations and Kilpatrick tend toward the other.</p> <hd id="AN0191105817-44">How Should We Assemble Words for Instruction?</hd> <p>What are the implications of these findings for research and instruction? The fact that there is little agreement about the words treated as special raises the empirical question as to whether the choice of such words has a major impact on children's progress. Does it matter which particular special words are chosen, or is it simply that some number of them must be taught in order to get reading off the ground? At present little research directly addresses these questions. The experimental evidence on learning to read words with irregularities is limited (Colenbrander et al. [<reflink idref="bib12" id="ref108">12</reflink>]). Colenbrander et al. ([<reflink idref="bib11" id="ref109">11</reflink>]) found evidence that mispronunciation correction and spelling irregular words outperformed passive reading. Their study included few words during training ( <ephtml> <math altimg="urn:x-wiley:00340553:media:rrq70077:rrq70077-math-0009" display="inline" overflow="scroll" xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>N</mi></mrow><annotation encoding="application/x-tex">$$ N $$</annotation></semantics></math> </ephtml>  = 12) and generalization outside the training items was not observed in any learning condition.</p> <p>It is clear from our study and from the variation seen in instructional practice more generally that popular approaches to reading instruction vary greatly in the number of words treated as special and the amount of time allocated to teaching them. It seems unlikely that all of these methods would be equally effective, but that question requires further research.</p> <hd id="AN0191105817-45">Can Theories of Word Reading Guide Us?</hd> <p>A broader question is whether current theories of reading and learning could be used to guide the selection of special words, the timing of when they are introduced, and the methods used in teaching them. How to accommodate the words with inconsistent (atypical, "exceptional") pronunciations is a major challenge for all phonics‐based instruction. At present, their identification is based on intuitions that yield highly variable results. Theories of word reading and learning may provide a more principled basis for what to teach, when, and how. In particular, theories dating from Seidenberg and McClelland ([<reflink idref="bib48" id="ref110">48</reflink>]) do not distinguish between rule‐governed words and irregular words. All words fall on a continuum of spelling‐sound consistency. At one end of this continuum are highly consistent mappings (e.g., "‐ist" is pronounced the same in "list," "mist," "gist," etc.), at the other there are seemingly irregular words such as "pint," "none," and "does". Other words exhibit intermediate degrees of consistency. Consistency provides a good account of numerous phenomena concerning beginning, skilled, and impaired reading (Bolger et al. [<reflink idref="bib5" id="ref111">5</reflink>]; Glushko [<reflink idref="bib23" id="ref112">23</reflink>]; Plaut et al. [<reflink idref="bib39" id="ref113">39</reflink>]; Taylor et al. [<reflink idref="bib57" id="ref114">57</reflink>]; Zevin and Seidenberg [<reflink idref="bib63" id="ref115">63</reflink>]).</p> <p>This account is based in part on the observation that even the most irregular words, including ones in the resources we studied, are not arbitrary; rather, they partially overlap with other words. "Have," for example, is standardly treated as irregular because the vowel would normally be long (as most educators tend to describe it), as in "save" and "gave". However, "have" overlaps with neighboring words such as "hive," "hate," and "had". Treating "have" as a word to be memorized ignores the fact that what is learned about the word builds on and applies to partially overlapping words. Numerous behavioral studies have established that learning and skilled reading are affected by degrees of consistency among neighborhoods of overlapping words such as "have," "hive," "hate," "save," and "gave". Rather than learning words as arbitrary, isolated patterns, instruction could focus on teaching such words in the context of overlapping neighbors. Apfelbaum et al. ([<reflink idref="bib3" id="ref116">3</reflink>]), for example, found that words with a particular vowel that appeared in more variable consonant contexts (e.g., "fan," "pat," "pal," "lap," "ram," "cab") promoted learning and generalization more than learning less variable sets of items (e.g., "bat," "hat," "pat," "cat," etc.). Assembling ensembles of words in this way could be a way to reduce the number of words to be memorized, which is time‐consuming and a distraction from learning the many systematic correspondences between spelling and sound in English reading and spelling. In line with our example "have" above, note also that this approach can apply to learning the invariant patterns in words that have been historically identified as "irregular." Take the highest ranking item on frequency and inconsistency in our method: "to." This word contains a pattern shared by many others, and by varying the context around this pattern (e.g., "tap," "tin," "ten") we would expect to enhance learning, as has been shown in learning studies within the statistical learning framework. Instruction need not solely focus on teaching the child a set of rules that putatively govern acceptable forms. The learning mechanism can be engaged also in this more implicit way as was shown in the Apfelbaum et al. ([<reflink idref="bib3" id="ref117">3</reflink>]) study. This instructional implication is quite different from a theory in which one assumes that the child's learning mechanism develops knowledge exclusively as a function of learning about rule‐governed and exceptional forms.[<reflink idref="bib4" id="ref118">4</reflink>]</p> <p>This possibility, supported empirically and explained by these established, convergent, and influential learning theories, is an important insight. We've provided a quantitative approach to aiding this instructional problem: identify words jointly based on how common they are while also placing words on the consistency continuum. This will accomplish two things at once. It ensures that the words being learned are useful (because they are common) while also quantifying their frequency relative to their consistency. This will also result in avoiding words that are infrequent, and thus less likely to be used (and useful) in the child's learning environment. It is also worth noting that the concepts of frequency and consistency themselves are integrally related, a fact that is accounted for in a statistical learning approach and the Triangle Framework (and connectionism more generally). A word is considered consistent precisely because it contains structure that is frequent across other words in the printed language. For example, the word "ball" has a high value for consistency because it contains the common orthographic pattern "all" which is associated with a frequent pattern of pronunciation.[<reflink idref="bib5" id="ref119">5</reflink>] Our hope is that an additional benefit of the account we provide here is that we can advance instructional theories in a way that reflects also these deeper insights about learning explained by the theory.</p> <hd id="AN0191105817-46">Other Existing Instructional Approaches</hd> <p>Recent discussions among researchers and practitioners have focused on reducing the number of words that must be memorized by building on such subregularities. For example, the alternative to treating "have," "give," and "love" as sight words is to build on the fact that they exhibit a pattern: the letters representing vowels are followed by "v" and "silent e". Whereas monosyllabic spoken words can end in /v/, their written forms do not end in the letter "v". Hence it could be argued that the final "e" is there to satisfy this condition, not to signal a long vowel (as is often taught in phonics programs). The child might be taught a phonics rule that covers these cases, which would allow them to "sound out" the words, obviating the need to memorize them by rote. The obvious challenge with such an approach is the additional burden on the child to learn about the contexts in which these constraints do and do not apply, which is not trivial especially when one considers all the additional content children need to learn in order to become proficient readers.</p> <p>Approaches such as "sounding out the sight words" (e.g., Miles et al. [<reflink idref="bib35" id="ref120">35</reflink>]) potentially extend this approach to numerous words that would otherwise be candidates to be memorized. It is possible to formulate rules that describe patterns that hold over small groups of words. Written English has innumerable subpatterns of this sort, each of which can be formulated as a teachable rule, added to the standard set used in phonics‐type approaches. Research is needed to determine whether teaching such rules facilitates reading acquisition compared to other approaches. Although the intent of rules such as the one for syllable‐final "v" and "silent e" is to reduce the amount of memorization in beginning reading, it is not clear that it does so. First, there is an additional rule to memorize, which itself requires attention, effort, and practice. Second, for the rule to be useful, the child has to learn where it does and does not apply. The rule about the "e" being there to avoid having the spelling end in "v" works for words such as "have" and "give," but not the many ones like "save" and "five." Learning which words the rule applies to is another type of memorization and thus not obviously easier than memorizing individual words. Given that explicit phonics instruction in many curricula is already very extensive, involving dozens of rules taught over several grades, the introduction of additional rules governed by complex conditions may not be wholly welcome or feasible.</p> <p>Another approach to facilitate learning irregular words relies on a task that has come to be known as "set for variability" (Steacy et al. [<reflink idref="bib56" id="ref121">56</reflink>]; Venezky [<reflink idref="bib59" id="ref122">59</reflink>]). In this procedure a word is sounded out and the spoken output is compared to words that the reader has already experienced via spoken language (a corresponding task for listening is also sometimes used). So, a word like "have," which is pronounced /hæv/, would be sounded out as /heIv/ (rhymes with "cave") and the reader, via introspection, would determine that the word they encountered in print is actually /hæv/. While there is little direct support for the effectiveness of the task in terms of supporting reading development, individuals who do well on the task tend to be better readers. For example, Steacy et al. ([<reflink idref="bib56" id="ref123">56</reflink>]) reported high correlations between the ability of elementary school children (grades 2–5 in the United States) to perform this task successfully and a range of characteristics known to be associated with strong reading skills, such as phonemic awareness ( <ephtml> <math altimg="urn:x-wiley:00340553:media:rrq70077:rrq70077-math-0010" display="inline" overflow="scroll" xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>r</mi></mrow><annotation encoding="application/x-tex">$$ r $$</annotation></semantics></math> </ephtml> = 0.47) and irregular word reading ( <ephtml> <math altimg="urn:x-wiley:00340553:media:rrq70077:rrq70077-math-0011" display="inline" overflow="scroll" xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>r</mi></mrow><annotation encoding="application/x-tex">$$ r $$</annotation></semantics></math> </ephtml> = 0.76). A current issue is whether or not teaching this skill directly results in gains in irregular word reading and word reading in general, though this outcome is plausible. Note, however, that how such a task would also leverage learning about shared properties with other words is also not clear and should be the topic of future research.</p> <hd id="AN0191105817-47">Ensembles of Words and Statistical Learning</hd> <p>Capitalizing on the patterns that hold among small neighborhoods of words is nonetheless an important idea. The alternative to explicitly teaching all patterns and treating them like phonics rules is to create conditions that promote learning the patterns without having to label them. Models such as Seidenberg and McClelland's focused on the statistical learning aspect of learning to read words. Such models illustrate how a statistical learning procedure picks up on the patterns that exist across words. The beauty of the procedure is that it expands basic ideas about the role of the environment and "teaching". The learner does not have to be explicitly taught (in the traditional sense) what all the patterns are or where to look for them. Learning in this view is influenced by the representation of coherent covariation (the patterns) uncovered when performing a task such as reading printed words aloud. The model discovers patterns inherent in sets of words automatically. From this perspective the main requirement is sufficient experience with words that exhibit learnable patterns. Such models can also be used to identify efficient ways to introduce words over time: groupings that optimize learning the desired inventory of words. This approach suggests instructional activities using groups of overlapping words, drawing attention to a pattern without treating them all as an explicit rule to be memorized. This idea has been suggested elsewhere but has been applied to reading in educational contexts in very limited ways (Compton et al. [<reflink idref="bib13" id="ref124">13</reflink>]; Compton et al. [<reflink idref="bib14" id="ref125">14</reflink>]; Seidenberg et al. [<reflink idref="bib45" id="ref126">45</reflink>]). Words that overlap in structure support each other in learning: What is learned about "have" is supported by neighbors such as "had" and "has" (see Compton et al. [<reflink idref="bib14" id="ref127">14</reflink>] for discussion about leveraging also statistical properties related to semantics). The instructional challenge, then, is to find the ways of grouping words, including "sight words," to promote mutually supportive learning alongside efficient explicit instruction about only the most useful rule‐like behavior of the writing system (also discussed in Compton et al. [<reflink idref="bib14" id="ref128">14</reflink>]).</p> <p>In summary, we have demonstrated a number of ways in which instructional resources differ in the words they select for instruction. This research is the first of its kind to look at resources like these and determine empirically the degree to which they agree on what should be taught in the reading classroom. While questions remain about the relative benefits of learning of these (and other similar) lists, our results provide convincing evidence of fundamental differences across resources that are intended to perform similar functions in reading instruction. The finding that there is little agreement about the treatment of special words raises important questions about current instructional practices. Our results suggest that both researchers and educators, perhaps especially curriculum authors, need to find more systematic ways to identify words for instructional attention. Looking to the future, efforts might turn from identifying and explicitly teaching phonics rules and exceptions towards structuring instructional activities to promote learning statistical patterns, with explicit guidance and feedback. Ideally words could be identified based on a model of how words are learned that shows how and why properties such as frequency and consistency, as well as imageability and age of acquisition, among others, affect learning and skilled performance. We've described several ideas of how this might be achieved. Lacking a fully realistic model we have also offered an intermediate step: choosing words based on the conjunction of two important factors, frequency and consistency.</p> <hd id="AN0191105817-48">Limitations</hd> <p>Only a subset of possible programs have been selected for analysis here. Those that participated in analyses were included due to their popularity, and a more comprehensive sample could no doubt be obtained. Likewise, the programs here were selected not because they are matched perfectly with respect to their instructional priorities and methods. Future work could consider programs that report to implement more specific shared theoretical perspectives about the development of word recognition skills in order to extend these findings to approaches that are more narrowly matched.</p> <hd id="AN0191105817-49">Funding</hd> <p>This work was supported by Institute of Education Sciences through award #R305B150003 to University of Wisconsin‐Madison. The opinions presented are those of the authors and do not reflect those of the US Department of Education.</p> <hd id="AN0191105817-50">Conflicts of Interest</hd> <p>The authors declare no conflicts of interest.</p> <hd id="AN0191105817-51">Data Availability Statement</hd> <p>All data for this project including the code that generated this manuscript can be found at: https://github.com/MCooperBorkenhagen/special%5fwords. Supporting Information can be found at the following location: https://osf.io/zqnd7.</p> <hd id="AN0191105817-52">A Appendix</hd> <p>A1 TABLE Parameter estimates for resource means on age‐of‐acquisition relative to normative sample.</p> <p> <ephtml> <table><thead valign="bottom"><tr><th align="left">Source</th><th align="center"><italic>b</italic></th><th align="center"><italic>t</italic></th><th align="center">df</th><th align="center">SE</th><th align="center">95% CI</th><th align="center"><italic>p</italic></th></tr></thead><tbody valign="top"><tr><td align="left"><p>Dolch</p></td><td align="center"><p><bold>−0.66</bold></p></td><td align="center"><p>−17.46</p></td><td align="center"><p>310</p></td><td align="center"><p>0.04</p></td><td align="center"><p>[−0.73, −0.58]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Fountas & Pinnell</p></td><td align="center"><p><bold>−0.70</bold></p></td><td align="center"><p>−23.96</p></td><td align="center"><p>433</p></td><td align="center"><p>0.03</p></td><td align="center"><p>[−0.75, −0.64]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Fry</p></td><td align="center"><p><bold>−0.69</bold></p></td><td align="center"><p>−18.78</p></td><td align="center"><p>297</p></td><td align="center"><p>0.04</p></td><td align="center"><p>[−0.76, −0.61]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Fundations</p></td><td align="center"><p><bold>−0.85</bold></p></td><td align="center"><p>−24.29</p></td><td align="center"><p>304</p></td><td align="center"><p>0.03</p></td><td align="center"><p>[−0.91, −0.78]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Kilpatrick</p></td><td align="center"><p><bold>−0.29</bold></p></td><td align="center"><p>−6.05</p></td><td align="center"><p>338</p></td><td align="center"><p>0.05</p></td><td align="center"><p>[−0.39, −0.2]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Wonders</p></td><td align="center"><p><bold>−0.74</bold></p></td><td align="center"><p>−33.06</p></td><td align="center"><p>780</p></td><td align="center"><p>0.02</p></td><td align="center"><p>[−0.78, −0.69]</p></td><td align="center"><p>< 0.001</p></td></tr></tbody></table> </ephtml> </p> <p>14 <emph>Note:</emph> Normative sample taken from TASA. Bolded coefficients are statistically significant.</p> <p>A2 TABLE Parameter estimates for resource means on consistency relative to normative sample.</p> <p> <ephtml> <table><thead valign="bottom"><tr><th align="left">Source</th><th align="center"><italic>b</italic></th><th align="center"><italic>t</italic></th><th align="center">df</th><th align="center">SE</th><th align="center">95% CI</th><th align="center"><italic>p</italic></th></tr></thead><tbody valign="top"><tr><td align="left"><p>Dolch</p></td><td align="center"><p>−0.01</p></td><td align="center"><p>−0.15</p></td><td align="center"><p>309</p></td><td align="center"><p>0.08</p></td><td align="center"><p>[−0.18, 0.15]</p></td><td align="center"><p>0.88</p></td></tr><tr><td align="left"><p>Fountas & Pinnell</p></td><td align="center"><p><bold>−0.024</bold></p></td><td align="center"><p>−3.48</p></td><td align="center"><p>434</p></td><td align="center"><p>0.07</p></td><td align="center"><p>[−0.38, −0.11]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Fry</p></td><td align="center"><p>−0.13</p></td><td align="center"><p>−1.51</p></td><td align="center"><p>294</p></td><td align="center"><p>0.09</p></td><td align="center"><p>[−0.3, 0.04]</p></td><td align="center"><p>0.13</p></td></tr><tr><td align="left"><p>Fundations</p></td><td align="center"><p><bold>−0.80</bold></p></td><td align="center"><p>−9.48</p></td><td align="center"><p>305</p></td><td align="center"><p>0.08</p></td><td align="center"><p>[−0.96, −0.63]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Kilpatrick</p></td><td align="center"><p><bold>−0.69</bold></p></td><td align="center"><p>−9.13</p></td><td align="center"><p>340</p></td><td align="center"><p>0.08</p></td><td align="center"><p>[−0.83, −0.54]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Wonders</p></td><td align="center"><p><bold>−0.44</bold></p></td><td align="center"><p>−8.40</p></td><td align="center"><p>792</p></td><td align="center"><p>0.05</p></td><td align="center"><p>[−0.54, −0.34]</p></td><td align="center"><p>< 0.001</p></td></tr></tbody></table> </ephtml> </p> <p>15 <emph>Note:</emph> Normative sample taken from TASA. Bolded coefficients are statistically significant.</p> <p>A3 TABLE Parameter estimates for resource means on imageability relative to normative sample.</p> <p> <ephtml> <table><thead valign="bottom"><tr><th align="left">Source</th><th align="center"><italic>b</italic></th><th align="center"><italic>t</italic></th><th align="center">df</th><th align="center">SE</th><th align="center">95% CI</th><th align="center"><italic>p</italic></th></tr></thead><tbody valign="top"><tr><td align="left"><p>Dolch</p></td><td align="center"><p><bold>−0.44</bold></p></td><td align="center"><p>−5.11</p></td><td align="center"><p>241</p></td><td align="center"><p>0.09</p></td><td align="center"><p>[−0.6, −0.27]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Fountas & Pinnell</p></td><td align="center"><p><bold>−1.09</bold></p></td><td align="center"><p>−18.65</p></td><td align="center"><p>282</p></td><td align="center"><p>0.06</p></td><td align="center"><p>[−1.21, −0.98]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Fry</p></td><td align="center"><p><bold>−0.85</bold></p></td><td align="center"><p>−11.48</p></td><td align="center"><p>219</p></td><td align="center"><p>0.07</p></td><td align="center"><p>[−1, −0.7]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Fundations</p></td><td align="center"><p><bold>−1.10</bold></p></td><td align="center"><p>−13.75</p></td><td align="center"><p>171</p></td><td align="center"><p>0.08</p></td><td align="center"><p>[−1.26, −0.95]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Kilpatrick</p></td><td align="center"><p><bold>−0.43</bold></p></td><td align="center"><p>−4.81</p></td><td align="center"><p>158</p></td><td align="center"><p>0.09</p></td><td align="center"><p>[−0.6, −0.25]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Wonders</p></td><td align="center"><p><bold>−0.94</bold></p></td><td align="center"><p>−21.29</p></td><td align="center"><p>518</p></td><td align="center"><p>0.04</p></td><td align="center"><p>[−1.03, −0.86]</p></td><td align="center"><p>< 0.001</p></td></tr></tbody></table> </ephtml> </p> <p>16 <emph>Note:</emph> Normative sample taken from TASA. Bolded coefficients are statistically significant.</p> <p>A4 TABLE Parameter estimates for resource means on letters per word relative to normative sample.</p> <p> <ephtml> <table><thead valign="bottom"><tr><th align="left">Source</th><th align="center"><italic>b</italic></th><th align="center"><italic>t</italic></th><th align="center">df</th><th align="center">SE</th><th align="center">95% CI</th><th align="center"><italic>p</italic></th></tr></thead><tbody valign="top"><tr><td align="left"><p>Dolch</p></td><td align="center"><p><bold>−1.37</bold></p></td><td align="center"><p>−46.01</p></td><td align="center"><p>313</p></td><td align="center"><p>0.03</p></td><td align="center"><p>[−1.43, −1.31]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Fountas & Pinnell</p></td><td align="center"><p><bold>−1.16</bold></p></td><td align="center"><p>−35.55</p></td><td align="center"><p>441</p></td><td align="center"><p>0.03</p></td><td align="center"><p>[−1.22, −1.09]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Fry</p></td><td align="center"><p><bold>−1.31</bold></p></td><td align="center"><p>−38.66</p></td><td align="center"><p>298</p></td><td align="center"><p>0.03</p></td><td align="center"><p>[−1.37, −1.24]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Fundations</p></td><td align="center"><p><bold>−1.08</bold></p></td><td align="center"><p>−24.74</p></td><td align="center"><p>309</p></td><td align="center"><p>0.04</p></td><td align="center"><p>[−1.17, −0.99]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Kilpatrick</p></td><td align="center"><p><bold>−0.76</bold></p></td><td align="center"><p>−19.99</p></td><td align="center"><p>343</p></td><td align="center"><p>0.04</p></td><td align="center"><p>[−0.84, −0.69]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Wonders</p></td><td align="center"><p><bold>−1.20</bold></p></td><td align="center"><p>−51.23</p></td><td align="center"><p>800</p></td><td align="center"><p>0.02</p></td><td align="center"><p>[−1.25, −1.16]</p></td><td align="center"><p>< 0.001</p></td></tr></tbody></table> </ephtml> </p> <p>17 <emph>Note:</emph> Normative sample taken from TASA. Bolded coefficients are statistically significant.</p> <p>A5 TABLE Parameter estimates for resource means on syllables per word relative to normative sample.</p> <p> <ephtml> <table><thead valign="bottom"><tr><th align="left">Source</th><th align="center"><italic>b</italic></th><th align="center"><italic>t</italic></th><th align="center">df</th><th align="center">SE</th><th align="center">95% CI</th><th align="center"><italic>p</italic></th></tr></thead><tbody valign="top"><tr><td align="left"><p>Dolch</p></td><td align="center"><p><bold>−1.06</bold></p></td><td align="center"><p>−48.81</p></td><td align="center"><p>313</p></td><td align="center"><p>0.02</p></td><td align="center"><p>[−1.1, −1.01]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Fountas & Pinnell</p></td><td align="center"><p><bold>−0.88</bold></p></td><td align="center"><p>−34.01</p></td><td align="center"><p>441</p></td><td align="center"><p>0.03</p></td><td align="center"><p>[−0.94, −0.83]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Fry</p></td><td align="center"><p><bold>−0.98</bold></p></td><td align="center"><p>−33.46</p></td><td align="center"><p>298</p></td><td align="center"><p>0.03</p></td><td align="center"><p>[−1.03, −0.92]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Fundations</p></td><td align="center"><p><bold>−0.77</bold></p></td><td align="center"><p>−19.88</p></td><td align="center"><p>309</p></td><td align="center"><p>0.04</p></td><td align="center"><p>[−0.84, −0.69]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Kilpatrick</p></td><td align="center"><p><bold>−0.62</bold></p></td><td align="center"><p>−16.51</p></td><td align="center"><p>343</p></td><td align="center"><p>0.04</p></td><td align="center"><p>[−0.7, −0.55]</p></td><td align="center"><p>< 0.001</p></td></tr><tr><td align="left"><p>Wonders</p></td><td align="center"><p><bold>−0.91</bold></p></td><td align="center"><p>−47.39</p></td><td align="center"><p>800</p></td><td align="center"><p>0.02</p></td><td align="center"><p>[−0.94, −0.87]</p></td><td align="center"><p>< 0.001</p></td></tr></tbody></table> </ephtml> </p> <p>18 <emph>Note:</emph> Normative sample taken from TASA. Bolded coefficients are statistically significant.</p> <p>A6 TABLE Top 50 ranked words based jointly on consistency and frequency.</p> <p> <ephtml> <table><thead valign="bottom"><tr><th align="left">Rank</th><th align="center">Word</th><th align="center">Dist. (Euclidean)</th><th align="center">Freq. (high‐low)</th><th align="center">Constncy. (low‐high)</th></tr></thead><tbody valign="top"><tr><td align="left"><p>1</p></td><td align="center"><p>to</p></td><td align="center"><p>5.39</p></td><td align="center"><p>2</p></td><td align="center"><p>5</p></td></tr><tr><td align="left"><p>2</p></td><td align="center"><p>is</p></td><td align="center"><p>20.62</p></td><td align="center"><p>13</p></td><td align="center"><p>16</p></td></tr><tr><td align="left"><p>3</p></td><td align="center"><p>his</p></td><td align="center"><p>24.76</p></td><td align="center"><p>18</p></td><td align="center"><p>17</p></td></tr><tr><td align="left"><p>4</p></td><td align="center"><p>was</p></td><td align="center"><p>29.55</p></td><td align="center"><p>12</p></td><td align="center"><p>27</p></td></tr><tr><td align="left"><p>5</p></td><td align="center"><p>are</p></td><td align="center"><p>31.02</p></td><td align="center"><p>29</p></td><td align="center"><p>11</p></td></tr><tr><td align="left"><p>6</p></td><td align="center"><p>have</p></td><td align="center"><p>32.80</p></td><td align="center"><p>26</p></td><td align="center"><p>20</p></td></tr><tr><td align="left"><p>7</p></td><td align="center"><p>do</p></td><td align="center"><p>33.54</p></td><td align="center"><p>33</p></td><td align="center"><p>6</p></td></tr><tr><td align="left"><p>8</p></td><td align="center"><p>what</p></td><td align="center"><p>39.62</p></td><td align="center"><p>27</p></td><td align="center"><p>29</p></td></tr><tr><td align="left"><p>9</p></td><td align="center"><p>said</p></td><td align="center"><p>47.30</p></td><td align="center"><p>11</p></td><td align="center"><p>46</p></td></tr><tr><td align="left"><p>10</p></td><td align="center"><p>that</p></td><td align="center"><p>51.24</p></td><td align="center"><p>15</p></td><td align="center"><p>49</p></td></tr><tr><td align="left"><p>11</p></td><td align="center"><p>were</p></td><td align="center"><p>51.48</p></td><td align="center"><p>47</p></td><td align="center"><p>21</p></td></tr><tr><td align="left"><p>12</p></td><td align="center"><p>for</p></td><td align="center"><p>56.30</p></td><td align="center"><p>19</p></td><td align="center"><p>53</p></td></tr><tr><td align="left"><p>13</p></td><td align="center"><p>of</p></td><td align="center"><p>62.51</p></td><td align="center"><p>8</p></td><td align="center"><p>62</p></td></tr><tr><td align="left"><p>14</p></td><td align="center"><p>your</p></td><td align="center"><p>63.01</p></td><td align="center"><p>51</p></td><td align="center"><p>37</p></td></tr><tr><td align="left"><p>15</p></td><td align="center"><p>a</p></td><td align="center"><p>63.07</p></td><td align="center"><p>3</p></td><td align="center"><p>63</p></td></tr><tr><td align="left"><p>16</p></td><td align="center"><p>one</p></td><td align="center"><p>76.94</p></td><td align="center"><p>36</p></td><td align="center"><p>68</p></td></tr><tr><td align="left"><p>17</p></td><td align="center"><p>the</p></td><td align="center"><p>89.01</p></td><td align="center"><p>1</p></td><td align="center"><p>89</p></td></tr><tr><td align="left"><p>18</p></td><td align="center"><p>he</p></td><td align="center"><p>90.20</p></td><td align="center"><p>6</p></td><td align="center"><p>90</p></td></tr><tr><td align="left"><p>19</p></td><td align="center"><p>she</p></td><td align="center"><p>92.07</p></td><td align="center"><p>14</p></td><td align="center"><p>91</p></td></tr><tr><td align="left"><p>20</p></td><td align="center"><p>want</p></td><td align="center"><p>96.21</p></td><td align="center"><p>90</p></td><td align="center"><p>34</p></td></tr><tr><td align="left"><p>21</p></td><td align="center"><p>we</p></td><td align="center"><p>97.08</p></td><td align="center"><p>31</p></td><td align="center"><p>92</p></td></tr><tr><td align="left"><p>22</p></td><td align="center"><p>be</p></td><td align="center"><p>98.35</p></td><td align="center"><p>32</p></td><td align="center"><p>93</p></td></tr><tr><td align="left"><p>23</p></td><td align="center"><p>an</p></td><td align="center"><p>100.00</p></td><td align="center"><p>100</p></td><td align="center"><p>1</p></td></tr><tr><td align="left"><p>24</p></td><td align="center"><p>me</p></td><td align="center"><p>102.96</p></td><td align="center"><p>42</p></td><td align="center"><p>94</p></td></tr><tr><td align="left"><p>25</p></td><td align="center"><p>put</p></td><td align="center"><p>104.81</p></td><td align="center"><p>91</p></td><td align="center"><p>52</p></td></tr><tr><td align="left"><p>26</p></td><td align="center"><p>who</p></td><td align="center"><p>106.23</p></td><td align="center"><p>106</p></td><td align="center"><p>7</p></td></tr><tr><td align="left"><p>27</p></td><td align="center"><p>two</p></td><td align="center"><p>108.30</p></td><td align="center"><p>108</p></td><td align="center"><p>8</p></td></tr><tr><td align="left"><p>28</p></td><td align="center"><p>i</p></td><td align="center"><p>127.10</p></td><td align="center"><p>5</p></td><td align="center"><p>127</p></td></tr><tr><td align="left"><p>29</p></td><td align="center"><p>been</p></td><td align="center"><p>128.10</p></td><td align="center"><p>125</p></td><td align="center"><p>28</p></td></tr><tr><td align="left"><p>30</p></td><td align="center"><p>long</p></td><td align="center"><p>133.28</p></td><td align="center"><p>120</p></td><td align="center"><p>58</p></td></tr><tr><td align="left"><p>31</p></td><td align="center"><p>i'll</p></td><td align="center"><p>137.52</p></td><td align="center"><p>137</p></td><td align="center"><p>12</p></td></tr><tr><td align="left"><p>32</p></td><td align="center"><p>don't</p></td><td align="center"><p>141.74</p></td><td align="center"><p>77</p></td><td align="center"><p>119</p></td></tr><tr><td align="left"><p>33</p></td><td align="center"><p>so</p></td><td align="center"><p>145.17</p></td><td align="center"><p>48</p></td><td align="center"><p>137</p></td></tr><tr><td align="left"><p>34</p></td><td align="center"><p>again</p></td><td align="center"><p>145.50</p></td><td align="center"><p>127</p></td><td align="center"><p>71</p></td></tr><tr><td align="left"><p>35</p></td><td align="center"><p>you</p></td><td align="center"><p>147.17</p></td><td align="center"><p>7</p></td><td align="center"><p>147</p></td></tr><tr><td align="left"><p>36</p></td><td align="center"><p>go</p></td><td align="center"><p>148.19</p></td><td align="center"><p>54</p></td><td align="center"><p>138</p></td></tr><tr><td align="left"><p>37</p></td><td align="center"><p>no</p></td><td align="center"><p>149.86</p></td><td align="center"><p>56</p></td><td align="center"><p>139</p></td></tr><tr><td align="left"><p>38</p></td><td align="center"><p>our</p></td><td align="center"><p>152.19</p></td><td align="center"><p>151</p></td><td align="center"><p>19</p></td></tr><tr><td align="left"><p>39</p></td><td align="center"><p>some</p></td><td align="center"><p>167.87</p></td><td align="center"><p>62</p></td><td align="center"><p>156</p></td></tr><tr><td align="left"><p>40</p></td><td align="center"><p>come</p></td><td align="center"><p>177.12</p></td><td align="center"><p>82</p></td><td align="center"><p>157</p></td></tr><tr><td align="left"><p>41</p></td><td align="center"><p>through</p></td><td align="center"><p>177.69</p></td><td align="center"><p>162</p></td><td align="center"><p>73</p></td></tr><tr><td align="left"><p>42</p></td><td align="center"><p>here</p></td><td align="center"><p>186.98</p></td><td align="center"><p>80</p></td><td align="center"><p>169</p></td></tr><tr><td align="left"><p>43</p></td><td align="center"><p>they</p></td><td align="center"><p>189.76</p></td><td align="center"><p>17</p></td><td align="center"><p>189</p></td></tr><tr><td align="left"><p>44</p></td><td align="center"><p>on</p></td><td align="center"><p>190.67</p></td><td align="center"><p>16</p></td><td align="center"><p>190</p></td></tr><tr><td align="left"><p>45</p></td><td align="center"><p>food</p></td><td align="center"><p>191.13</p></td><td align="center"><p>186</p></td><td align="center"><p>44</p></td></tr><tr><td align="left"><p>46</p></td><td align="center"><p>does</p></td><td align="center"><p>193.00</p></td><td align="center"><p>185</p></td><td align="center"><p>55</p></td></tr><tr><td align="left"><p>47</p></td><td align="center"><p>read</p></td><td align="center"><p>202.00</p></td><td align="center"><p>198</p></td><td align="center"><p>40</p></td></tr><tr><td align="left"><p>48</p></td><td align="center"><p>began</p></td><td align="center"><p>203.30</p></td><td align="center"><p>177</p></td><td align="center"><p>100</p></td></tr><tr><td align="left"><p>49</p></td><td align="center"><p>bear</p></td><td align="center"><p>218.33</p></td><td align="center"><p>215</p></td><td align="center"><p>38</p></td></tr><tr><td align="left"><p>50</p></td><td align="center"><p>into</p></td><td align="center"><p>222.04</p></td><td align="center"><p>66</p></td><td align="center"><p>212</p></td></tr></tbody></table> </ephtml> </p> <p>19 <emph>Note:</emph> This table extends the shorter table (top 10) provided in the text. Rankings derived from Euclidean distance of inconsistency and frequency ranking against the origin (0, 0). The calculated distance is also provided. All words for all resources were considered when making the calculation. Consistency is arranged such that a low rank means low consistency. Frequency is arranged so that a low rank means high frequency.</p> <ref id="AN0191105817-53"> <title> Footnotes </title> <blist> <bibl id="bib1" idref="ref1" type="bt">1</bibl> <bibtext> Note that "regular" words were not included in any treatment condition, including the waitlist control group. We are merely contrasting these two types of words for purposes of clarity.</bibtext> </blist> <blist> <bibl id="bib2" idref="ref11" type="bt">2</bibl> <bibtext> Different terms are used by different programs. For example, "trick word" is used in Wilson's Fundations, and "heart words" is a term used in the University of Florida Literacy Institute's "UFLI" program (which is not analyzed in the study here), and elsewhere. The term "snap word" is associated with an educational product by the same name, but also is used colloquially among practitioners. These terms, while associated with different specific instructional practices, share the purpose of drawing attention to the teacher and student that a word exhibits atypical structure and require special attention.</bibtext> </blist> <blist> <bibl id="bib3" idref="ref22" type="bt">3</bibl> <bibtext> The identification of words in this way is conceptually related to calculating consistency over word tokens (i.e., observed occurrences of words in a text database) rather than word types (unique words), but focuses on the frequency of the word itself rather than the observed letter‐sound pattern (see Chee et al. [10] for discussion of these two methods). Of additional importance here is that this method using the two variables and the distance of each word from the origin permits a clear visual interpretation of the utility of a word, not easily achieved when described only quantitatively.</bibtext> </blist> <blist> <bibl id="bib4" idref="ref101" type="bt">4</bibl> <bibtext> Note that we are not denying the benefits of explicit forms of instruction during reading development, but are describing other established learning mechanisms that might be considered in educational approaches to this learning problem. Explicit forms of instruction (typically those that directly benefit corresponding explicit forms of knowledge) have been shown to benefit learning to read (e.g., Rastle et al. [42]) though there is variation in how explicit instruction is approached.</bibtext> </blist> <blist> <bibl id="bib5" idref="ref111" type="bt">5</bibl> <bibtext> Leaving aside the issue of how consistency is measured, for which there are different units over which such calculations can be made (Chee et al. [10]; Plaut et al. [39]).</bibtext> </blist> </ref> <ref id="AN0191105817-54"> <title> References </title> <blist> <bibtext> Adams, M. J. 2011. " Advancing Our Students' Language and Literacy: The Challenge of Complex Texts." American Educator 34 : 3 – 12.</bibtext> </blist> <blist> <bibtext> Anderson, J. R., and L. J. Schooler. 1991. " Reflections of the Environment in Memory." Psychological Science 2, no. 6 : 396 – 408. https://doi.org/10.1111/j.1467‐9280.1991.tb00174.x.</bibtext> </blist> <blist> <bibtext> Apfelbaum, K. S., E. Hazeltine, and B. McMurray. 2013. " Statistical Learning in Reading: Variability in Irrelevant Letters Helps Children Learn Phonics Skills." Developmental Psychology 49 : 1348 – 1365. https://doi.org/10.1037/a0029839.</bibtext> </blist> <blist> <bibtext> Balota, D. A., M. J. Cortese, S. D. Sergent‐Marshall, D. H. Spieler, and M. J. Yap. 2004. " Visual Word Recognition of Single‐Syllable Words." Journal of Experimental Psychology: General 133 : 283 – 316. https://doi.org/10.1037/0096‐3445.133.2.283.</bibtext> </blist> <blist> <bibtext> Bolger, D. J., J. Hornickel, N. E. Cone, D. D. Burman, and J. R. Booth. 2008. " Neural Correlates of Orthographic and Phonological Consistency Effects in Children." Human Brain Mapping 29 : 1416 – 1429. https://doi.org/10.1002/hbm.20476.</bibtext> </blist> <blist> <bibl id="bib6" idref="ref53" type="bt">6</bibl> <bibtext> Brown, M. C., D. E. Sibley, J. A. Washington, et al. 2015. " Impact of Dialect Use on a Basic Component of Learning to Read." Frontiers in Psychology 6 : 196. https://doi.org/10.3389/fpsyg.2015.00196.</bibtext> </blist> <blist> <bibl id="bib7" idref="ref63" type="bt">7</bibl> <bibtext> Chall, J. S. 1983. Stages of Reading Development. McGraw‐Hill.</bibtext> </blist> <blist> <bibl id="bib8" idref="ref41" type="bt">8</bibl> <bibtext> Chang, Y. N. 2023. " The Influence of Oral Vocabulary Knowledge on Individual Differences in a Computational Model of Reading." Scientific Reports 13, no. 1 : 1680. https://doi.org/10.1038/s41598‐023‐28559‐3.</bibtext> </blist> <blist> <bibl id="bib9" idref="ref42" type="bt">9</bibl> <bibtext> Chang, Y. N., P. Monaghan, and S. Welbourne. 2019. " A Computational Model of Reading Across Development: Effects of Literacy Onset on Language Processing." Journal of Memory and Language 108 : 104025. https://doi.org/10.1016/j.jml.2019.05.003.</bibtext> </blist> <blist> <bibtext> Chee, Q. W., K. J. Chow, M. J. Yap, and W. D. Goh. 2020. " Consistency Norms for 37,677 English Words." Behavior Research Methods 52 : 2535 – 2555. https://doi.org/10.3758/s13428‐020‐01391‐7.</bibtext> </blist> <blist> <bibtext> Colenbrander, D., S. Kohnen, E. Beyersmann, et al. 2022. " Teaching Children to Read Irregular Words: A Comparison of Three Instructional Methods." Scientific Studies of Reading 26 : 545 – 564. https://doi.org/10.1080/10888438.2022.2077653.</bibtext> </blist> <blist> <bibtext> Colenbrander, D., H. Wang, T. Arrow, and A. Castles. 2020. " Teaching Irregular Words: What We Know, What We Don't Know, and Where We Can Go From Here." Educational and Developmental Psychologist 37 : 97 – 104. https://doi.org/10.1017/edp.2020.11.</bibtext> </blist> <blist> <bibtext> Compton, D. L., A. C. Miller, A. M. Elleman, and L. M. Steacy. 2014. " Have We Forsaken Reading Theory in the Name of "Quick Fix" Interventions for Children with Reading Disability? " Scientific Studies of Reading 18, no. 1 : 55 – 73.</bibtext> </blist> <blist> <bibtext> Compton, D. L., L. M. Steacy, M. J. Cooper Borkenhagen, V. M. Rigobon, C. E. Vazquez, and J. A. Smith. 2025. " Pressure Points on Representational Quality: Evolving the Traditional Instructional Model for Dyslexia." Mind, Brain, and Education 19, no. 4. https://doi.org/10.1111/mbe.70026.</bibtext> </blist> <blist> <bibtext> Cox, C. R., M. J. Cooper Borkenhagen, and M. S. Seidenberg. 2019. " Efficiency of Learning in Experience‐Limited Domains: Generalization Beyond the WUG Test." In Proceedings of the 41st Annual Meeting of the Cognitive Science Society, 1566 – 1571. Cognitive Science Society.</bibtext> </blist> <blist> <bibtext> Dolch, E. W. 1936. " A Basic Sight Vocabulary." Elementary School Journal 36, no. 6 : 456 – 460. https://doi.org/10.1086/457353.</bibtext> </blist> <blist> <bibtext> Ehri, L. C. 2005a. " The Development of Sight Word Reading: Phases and Findings." In The Science of Reading: A Handbook, edited by M. J. Snowling and C. Hulme. Blackwell Publishing.</bibtext> </blist> <blist> <bibtext> Ehri, L. C. 2005b. " Learning to Read Words: Theory, Findings, and Issues." Scientific Studies of Reading 9, no. 2 : 167 – 188. https://doi.org/10.1207/s1532799xssr0902_4.</bibtext> </blist> <blist> <bibtext> Ehri, L. C. 2014. " Orthographic Mapping in the Acquisition of Sight Word Reading, Spelling Memory, and Vocabulary Learning." Scientific Studies of Reading 18, no. 1 : 5 – 21. https://doi.org/10.1080/10888438.2013.819356.</bibtext> </blist> <blist> <bibtext> Fry, E. 1957. " Developing a Word List for Remedial Reading." Elementary English 34, no. 7 : 456 – 458.</bibtext> </blist> <blist> <bibtext> Fry, E. 1980. " The New Instant Word List." Reading Teacher 34, no. 3 : 284 – 289.</bibtext> </blist> <blist> <bibtext> Gao, C. H., C. Chen, T. Akyol, et al. 2024. " ggVennDiagram: Intuitive Venn Diagram Software Extended." iMeta 3, no. 1 : e177. https://doi.org/10.1002/imt2.177.</bibtext> </blist> <blist> <bibtext> Glushko, R. J. 1979. " The Organization and Activation of Orthographic Knowledge in Reading Aloud." Journal of Experimental Psychology: Human Perception and Performance 5 : 674 – 691. https://doi.org/10.1037/0096‐1523.5.4.674.</bibtext> </blist> <blist> <bibtext> Gries, S. T. 2006. " Exploring Variability Within and Between Corpora: Some Methodological Considerations." Corpora 1 : 109 – 151. https://doi.org/10.3366/cor.2006.1.2.109.</bibtext> </blist> <blist> <bibtext> Harm, M. W., and M. S. Seidenberg. 1999. " Phonology, Reading Acquisition, and Dyslexia: Insights From Connectionist Models." Psychological Review 106, no. 3 : 491 – 528.</bibtext> </blist> <blist> <bibtext> Harm, M. W., and M. S. Seidenberg. 2004. " Computing the Meanings of Words in Reading: Cooperative Division of Labor Between Visual and Phonological Processes." Psychological Review 111, no. 3 : 662 – 720. https://doi.org/10.1037/0033‐295X.111.3.662.</bibtext> </blist> <blist> <bibtext> Houghton, G., and M. Zorzi. 1999. " A Model of Sound‐Spelling Mapping in English and Its Role in Word and Nonword Spelling." In The Proceedings of the 20th Annual Conference of the Cognitive Science Society, edited by M. A. Gernsbacher and S. J. Derry, 490 – 495. Routledge.</bibtext> </blist> <blist> <bibtext> Jared, D., K. McRae, and M. S. Seidenberg. 1990. " The Basis of Consistency Effects in Word Naming." Journal of Memory and Language 29 : 687 – 715. https://doi.org/10.1016/0749‐596X(90)90044‐Z.</bibtext> </blist> <blist> <bibtext> Jared, D. 2002. " Spelling—Sound Consistency and Regularity Effects in Word Naming." Journal of Memory and Language 46, no. 4 : 723 – 750. https://doi.org/10.1006/jmla.2001.2827.</bibtext> </blist> <blist> <bibtext> Kilpatrick, D. A. 2016. Equipped for Reading Success. Casey & Kirsch.</bibtext> </blist> <blist> <bibtext> Kuperman, V., H. Stadthagen‐Gonzalez, and M. Brysbaert. 2012. " Age‐Of‐Acquisition Ratings for 30,000 English Words." Behavior Research Methods 44 : 978 – 990. https://doi.org/10.3758/s13428‐012‐0210‐4.</bibtext> </blist> <blist> <bibtext> Leach, J. M., H. S. Scarborough, and L. Rescorla. 2003. " Late‐Emerging Reading Disabilities." Journal of Educational Psychology 95, no. 2 : 211 – 224. https://doi.org/10.1037/0022‐0663.95.2.211.</bibtext> </blist> <blist> <bibtext> McClelland, J. L., M. M. Botvinick, D. C. Noelle, et al. 2010. " Letting Structure Emerge: Connectionist and Dynamical Systems Approaches to Cognition." Trends in Cognitive Sciences 14, no. 8 : 348 – 356.</bibtext> </blist> <blist> <bibtext> Miles, K. P., D. Eide, and J. R. Butler. 2024. " The Regularity of High‐Frequency Words (Sight Words): Teacher Phonetic Knowledge Is Key." Reading Psychology 45 : 832 – 852. https://doi.org/10.1080/02702711.2024.2379255.</bibtext> </blist> <blist> <bibtext> Miles, K. P., G. B. Rubin, and S. Gonzalez‐Frey. 2018. " Rethinking Sight Words." Reading Teacher 71, no. 6 : 715 – 726. https://doi.org/10.1002/trtr.1658.</bibtext> </blist> <blist> <bibtext> Monaghan, P., Y. N. Chang, S. Welbourne, and M. Brysbaert. 2017. " Exploring the Relations Between Word Frequency, Language Exposure, and Bilingualism in a Computational Model of Reading." Journal of Memory and Language 93 : 1 – 21.</bibtext> </blist> <blist> <bibtext> Nation, K. 2017. " Nurturing a Lexical Legacy: Reading Experience Is Critical for the Development of Word Reading Skill." npj Science of Learning 2, no. 1 : 3.</bibtext> </blist> <blist> <bibtext> Perfetti, C. A., and L. Hart. 2008. " The Lexical Quality Hypothesis." In Precursors of Functional Literacy, 189 – 213. John Benjamins Publishing Company.</bibtext> </blist> <blist> <bibtext> Plaut, D. C., J. L. McClelland, M. S. Seidenberg, and K. Patterson. 1996. " Understanding Normal and Impaired Word Reading: Computational Principles in Quasi‐Regular Domains." Psychological Review 103 : 56 – 115. https://doi.org/10.1037/0033‐295X.103.1.56.</bibtext> </blist> <blist> <bibtext> Powell, D., D. Plaut, and E. Funnell. 2006. " Does the PMSP Connectionist Model of Single Word Reading Learn to Read in the Same Way as a Child? " Journal of Research in Reading 29, no. 2 : 229 – 250.</bibtext> </blist> <blist> <bibtext> R Core Team. 2021. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. www.R‐project.org/.</bibtext> </blist> <blist> <bibtext> Rastle, K., C. Lally, M. H. Davis, and J. S. H. Taylor. 2021. " The Dramatic Impact of Explicit Instruction on Learning to Read in a New Writing System." Psychological Science 32, no. 4 : 471 – 484.</bibtext> </blist> <blist> <bibtext> Rinsland, H. D. 1945. A Basic Vocabulary of Elementary School Children. Macmillan Co.</bibtext> </blist> <blist> <bibtext> Sawi, O. M., and J. Rueckl. 2019. " Reading and the Neurocognitive Bases of Statistical Learning." Scientific Studies of Reading 23, no. 1 : 8 – 23.</bibtext> </blist> <blist> <bibtext> Seidenberg, M. S., M. J. Cooper Borkenhagen, and D. M. Kearns. 2020. " Lost in Translation? Challenges in Connecting Reading Science and Educational Practice." Reading Research Quarterly 55 : S119 – S130.</bibtext> </blist> <blist> <bibtext> Seidenberg, M. S., M. Farry‐Thorn, and J. D. Zevin. 2022. " Models of Word Reading: What Have We Learned? " In The Science of Reading, edited by M. J. Snowling, C. Hulme, and K. Nation, 2nd ed., 36 – 59. Blackwell Publishing.</bibtext> </blist> <blist> <bibtext> Seidenberg, M. S., and M. C. MacDonald. 2018. " The Impact of Language Experience on Language and Reading: A Statistical Learning Approach." Topics in Language Disorders 38, no. 1 : 66 – 83.</bibtext> </blist> <blist> <bibtext> Seidenberg, M. S., and J. L. McClelland. 1989. " A Distributed, Developmental Model of Word Recognition and Naming." Psychological Review 96, no. 4 : 523 – 568. https://doi.org/10.1037//0033‐295X.96.4.523.</bibtext> </blist> <blist> <bibtext> Share, D. L. 1995. " Phonological Recoding and Self‐Teaching: Sine Qua Non of Reading Acquisition." Cognition 55, no. 2 : 151 – 218.</bibtext> </blist> <blist> <bibtext> Siegelman, N., B. C. Armstrong, Y. Raz, and R. Frost. 2025. " The Statistical Reader: The Role of Orthographic Regularities in Reading." Journal of Experimental Psychology: General 154, no. 7 : 1800 – 1823. https://doi.org/10.1037/xge0001775.</bibtext> </blist> <blist> <bibtext> Siegelman, N., D. M. Kearns, and J. G. Rueckl. 2020. " Using Information‐Theoretic Measures to Characterize the Structure of the Writing System: The Case of Orthographic‐Phonological Regularities in English." Behavior Research Methods 52, no. 3 : 1292 – 1312.</bibtext> </blist> <blist> <bibtext> Smith, L. B., and E. Thelen. 2003. " Development as a Dynamic System." Trends in Cognitive Sciences 7, no. 8 : 343 – 348.</bibtext> </blist> <blist> <bibtext> Solity, J., and J. Vousden. 2009. " Real Books vs Reading Schemes: A New Perspective From Instructional Psychology." Educational Psychology 29, no. 4 : 469 – 511.</bibtext> </blist> <blist> <bibtext> Stadthagen‐Gonzalez, H., and C. J. Davis. 2006. " The Bristol Norms for Age of Acquisition, Imageability, and Familiarity." Behavior Research Methods 38 : 598 – 605.</bibtext> </blist> <blist> <bibtext> Steacy, L. M., A. A. Edwards, V. M. Rigobon, et al. 2023. " Set for Variability as a Critical Predictor of Word Reading: Potential Implications for Early Identification and Treatment of Dyslexia." Reading Research Quarterly 58, no. 2 : 254 – 267.</bibtext> </blist> <blist> <bibtext> Steacy, L. M., L. Wade‐Woolley, J. G. Rueckl, K. R. Pugh, J. D. Elliott, and D. L. Compton. 2019. " The Role of Set for Variability in Irregular Word Reading: Word and Child Predictors in Typically Developing Readers and Students At‐Risk for Reading Disabilities." Scientific Studies of Reading 23 : 523 – 532. https://doi.org/10.1080/10888438.2019.1620749.</bibtext> </blist> <blist> <bibtext> Taylor, J. S. H., K. Plunkett, and K. Nation. 2011. " The Influence of Consistency, Frequency, and Semantics on Learning to Read: An Artificial Orthography Paradigm." Journal of Experimental Psychology. Learning, Memory, and Cognition 37 : 60 – 76. https://doi.org/10.1037/a0020126.</bibtext> </blist> <blist> <bibtext> Thorndike, E. L., and I. Lorge. 1944. The Teacher's Word Book of 30,000 Words. Teacher's College Press.</bibtext> </blist> <blist> <bibtext> Venezky, R. L. 1999. The American Way of Spelling: The Structure and Origins of American English Orthography. Guilford Press.</bibtext> </blist> <blist> <bibtext> Vousden, J. I. 2008. " Units of English Spelling‐To‐Sound Mapping: A Rational Approach to Reading Instruction." Applied Cognitive Psychology 22, no. 2 : 247 – 272.</bibtext> </blist> <blist> <bibtext> Yang, J., H. Shu, B. D. McCandliss, and J. D. Zevin. 2013. " Orthographic Influences on Division of Labor in Learning to Read Chinese and English: Insights From Computational Modeling." Bilingualism: Language and Cognition 16, no. 2 : 354 – 366.</bibtext> </blist> <blist> <bibtext> Zeno, S. M., S. H. Ivens, R. T. Millard, and R. Duvvuri. 1995. The Educator's Word Frequency Guide. Touchstone Applied Science Associates.</bibtext> </blist> <blist> <bibtext> Zevin, J. D., and M. S. Seidenberg. 2006. " Simulating Consistency Effects and Individual Differences in Nonword Naming: A Comparison of Current Models." Journal of Memory and Language 54 : 145 – 160. https://doi.org/10.1016/j.jml.2005.08.002.</bibtext> </blist> </ref> <aug> <p>By Matthew J. Cooper Borkenhagen; Lauren P. Schilling and Mark S. Seidenberg</p> <p>Reported by Author; Author; Author</p> </aug> <nolink nlid="nl1" bibid="bib48" firstref="ref2"></nolink> <nolink nlid="nl2" bibid="bib59" firstref="ref3"></nolink> <nolink nlid="nl3" bibid="bib16" firstref="ref4"></nolink> <nolink nlid="nl4" bibid="bib20" firstref="ref5"></nolink> <nolink nlid="nl5" bibid="bib12" firstref="ref6"></nolink> <nolink nlid="nl6" bibid="bib35" firstref="ref8"></nolink> <nolink nlid="nl7" bibid="bib53" firstref="ref9"></nolink> <nolink nlid="nl8" bibid="bib60" firstref="ref10"></nolink> <nolink nlid="nl9" bibid="bib51" firstref="ref13"></nolink> <nolink nlid="nl10" bibid="bib50" firstref="ref14"></nolink> <nolink nlid="nl11" bibid="bib11" firstref="ref15"></nolink> <nolink nlid="nl12" bibid="bib23" firstref="ref17"></nolink> <nolink nlid="nl13" bibid="bib39" firstref="ref18"></nolink> <nolink nlid="nl14" bibid="bib46" firstref="ref20"></nolink> <nolink nlid="nl15" bibid="bib34" firstref="ref21"></nolink> <nolink nlid="nl16" bibid="bib17" firstref="ref24"></nolink> <nolink nlid="nl17" bibid="bib18" firstref="ref25"></nolink> <nolink nlid="nl18" bibid="bib19" firstref="ref29"></nolink> <nolink nlid="nl19" bibid="bib37" firstref="ref30"></nolink> <nolink nlid="nl20" bibid="bib38" firstref="ref31"></nolink> <nolink nlid="nl21" bibid="bib49" firstref="ref33"></nolink> <nolink nlid="nl22" bibid="bib15" firstref="ref43"></nolink> <nolink nlid="nl23" bibid="bib26" firstref="ref44"></nolink> <nolink nlid="nl24" bibid="bib36" firstref="ref45"></nolink> <nolink nlid="nl25" bibid="bib61" firstref="ref48"></nolink> <nolink nlid="nl26" bibid="bib25" firstref="ref52"></nolink> <nolink nlid="nl27" bibid="bib40" firstref="ref54"></nolink> <nolink nlid="nl28" bibid="bib45" firstref="ref55"></nolink> <nolink nlid="nl29" bibid="bib44" firstref="ref56"></nolink> <nolink nlid="nl30" bibid="bib47" firstref="ref57"></nolink> <nolink nlid="nl31" bibid="bib52" firstref="ref58"></nolink> <nolink nlid="nl32" bibid="bib33" firstref="ref59"></nolink> <nolink nlid="nl33" bibid="bib30" firstref="ref62"></nolink> <nolink nlid="nl34" bibid="bib32" firstref="ref64"></nolink> <nolink nlid="nl35" bibid="bib21" firstref="ref66"></nolink> <nolink nlid="nl36" bibid="bib43" firstref="ref74"></nolink> <nolink nlid="nl37" bibid="bib58" firstref="ref75"></nolink> <nolink nlid="nl38" bibid="bib300" firstref="ref77"></nolink> <nolink nlid="nl39" bibid="bib56" firstref="ref80"></nolink> <nolink nlid="nl40" bibid="bib55" firstref="ref81"></nolink> <nolink nlid="nl41" bibid="bib62" firstref="ref86"></nolink> <nolink nlid="nl42" bibid="bib328" firstref="ref87"></nolink> <nolink nlid="nl43" bibid="bib24" firstref="ref88"></nolink> <nolink nlid="nl44" bibid="bib41" firstref="ref91"></nolink> <nolink nlid="nl45" bibid="bib10" firstref="ref92"></nolink> <nolink nlid="nl46" bibid="bib29" firstref="ref93"></nolink> <nolink nlid="nl47" bibid="bib31" firstref="ref95"></nolink> <nolink nlid="nl48" bibid="bib54" firstref="ref96"></nolink> <nolink nlid="nl49" bibid="bib27" firstref="ref98"></nolink> <nolink nlid="nl50" bibid="bib28" firstref="ref102"></nolink> <nolink nlid="nl51" bibid="bib57" firstref="ref114"></nolink> <nolink nlid="nl52" bibid="bib63" firstref="ref115"></nolink> <nolink nlid="nl53" bibid="bib13" firstref="ref124"></nolink> <nolink nlid="nl54" bibid="bib14" firstref="ref125"></nolink>
Header DbId: eric
DbLabel: ERIC
An: EJ1494630
AccessLevel: 3
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Which Words Are Special? Identification of 'Sight' Words in Educational Resources
– Name: Language
  Label: Language
  Group: Lang
  Data: English
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Matthew+J%2E+Cooper+Borkenhagen%22">Matthew J. Cooper Borkenhagen</searchLink> (ORCID <externalLink term="https://orcid.org/0000-0002-8245-0130">0000-0002-8245-0130</externalLink>)<br /><searchLink fieldCode="AR" term="%22Lauren+P%2E+Schilling%22">Lauren P. Schilling</searchLink> (ORCID <externalLink term="https://orcid.org/0009-0004-3282-5653">0009-0004-3282-5653</externalLink>)<br /><searchLink fieldCode="AR" term="%22Mark+S%2E+Seidenberg%22">Mark S. Seidenberg</searchLink> (ORCID <externalLink term="https://orcid.org/0000-0001-8519-3259">0000-0001-8519-3259</externalLink>)
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="SO" term="%22Reading+Research+Quarterly%22"><i>Reading Research Quarterly</i></searchLink>. 2026 61(1).
– Name: Avail
  Label: Availability
  Group: Avail
  Data: Wiley. Available from: John Wiley & Sons, Inc. 111 River Street, Hoboken, NJ 07030. Tel: 800-835-6770; e-mail: cs-journals@wiley.com; Web site: https://www.wiley.com/en-us
– Name: PeerReviewed
  Label: Peer Reviewed
  Group: SrcInfo
  Data: Y
– Name: Pages
  Label: Page Count
  Group: Src
  Data: 25
– Name: DatePubCY
  Label: Publication Date
  Group: Date
  Data: 2026
– Name: SourceSuprt
  Label: Sponsoring Agency
  Group: SrcSuprt
  Data: Institute of Education Sciences (ED)
– Name: NumberContract
  Label: Contract Number
  Group: NumCntrct
  Data: R305B150003
– Name: TypeDocument
  Label: Document Type
  Group: TypDoc
  Data: Journal Articles<br />Reports - Research
– Name: Subject
  Label: Descriptors
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22Sight+Vocabulary%22">Sight Vocabulary</searchLink><br /><searchLink fieldCode="DE" term="%22Sight+Method%22">Sight Method</searchLink><br /><searchLink fieldCode="DE" term="%22Reading+Instruction%22">Reading Instruction</searchLink><br /><searchLink fieldCode="DE" term="%22Word+Lists%22">Word Lists</searchLink><br /><searchLink fieldCode="DE" term="%22Reading+Skills%22">Reading Skills</searchLink><br /><searchLink fieldCode="DE" term="%22Beginning+Reading%22">Beginning Reading</searchLink><br /><searchLink fieldCode="DE" term="%22Spelling%22">Spelling</searchLink><br /><searchLink fieldCode="DE" term="%22Phoneme+Grapheme+Correspondence%22">Phoneme Grapheme Correspondence</searchLink><br /><searchLink fieldCode="DE" term="%22Word+Frequency%22">Word Frequency</searchLink>
– Name: DOI
  Label: DOI
  Group: ID
  Data: 10.1002/rrq.70077
– Name: ISSN
  Label: ISSN
  Group: ISSN
  Data: 0034-0553<br />1936-2722
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: Which words are important for early reading instruction? A standard view holds that certain words should be emphasized early in development because they are used with high frequency and/or contain atypical spelling-sound structure. Such words have been labeled "sight," "trick," "snap," or simply "high frequency" words; we refer to them as "special" words, which is intended to reflect their status as words that have been identified as being of particular instructional importance. The present study examined whether instructional resources such as commonly used curricula and word lists agree on their identity. Understanding the contents of these resources and those like them is important given their prevalence in instruction: teachers rely on wordlists to plan activities to support early word reading skills. We addressed this question using six such resources ranging from the classic Dolch list to modern commercial curricula. Results show substantial disagreement about the designated words and their properties. A total of 973 distinct words are identified in these materials. Only 28 words (3%) appear in all six resources, and over half appear in only a single one (560 words; 56%). Additional analyses demonstrate that the materials differ in terms of a number of word properties including frequency and spelling-sound consistency. Together the results indicate a surprising lack of agreement about which words should be treated as special for instructional purposes. These differences suggest that beginning readers' learning experiences may vary greatly. In the general discussion we describe an alternative method for identifying words on a more principled basis, which would also facilitate comparisons across curricula. This method is based on computational theories of word reading that specify how properties of words such as spelling-sound consistency and word frequency affect learning. Theories of this kind can provide a more systematic basis for identifying words to emphasize in early instruction.
– Name: AbstractInfo
  Label: Abstractor
  Group: Ab
  Data: As Provided
– Name: Note
  Label: Notes
  Group: Note
  Data: https://github.com/MCooperBorkenhagen/special_words
– Name: CodeSource
  Label: IES Funded
  Group: SrcInfo
  Data: Yes
– Name: DateEntry
  Label: Entry Date
  Group: Date
  Data: 2026
– Name: AN
  Label: Accession Number
  Group: ID
  Data: EJ1494630
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=eric&AN=EJ1494630
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1002/rrq.70077
    Languages:
      – Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 25
    Subjects:
      – SubjectFull: Sight Vocabulary
        Type: general
      – SubjectFull: Sight Method
        Type: general
      – SubjectFull: Reading Instruction
        Type: general
      – SubjectFull: Word Lists
        Type: general
      – SubjectFull: Reading Skills
        Type: general
      – SubjectFull: Beginning Reading
        Type: general
      – SubjectFull: Spelling
        Type: general
      – SubjectFull: Phoneme Grapheme Correspondence
        Type: general
      – SubjectFull: Word Frequency
        Type: general
    Titles:
      – TitleFull: Which Words Are Special? Identification of 'Sight' Words in Educational Resources
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Matthew J. Cooper Borkenhagen
      – PersonEntity:
          Name:
            NameFull: Lauren P. Schilling
      – PersonEntity:
          Name:
            NameFull: Mark S. Seidenberg
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 01
              M: 01
              Type: published
              Y: 2026
          Identifiers:
            – Type: issn-print
              Value: 0034-0553
            – Type: issn-electronic
              Value: 1936-2722
          Numbering:
            – Type: volume
              Value: 61
            – Type: issue
              Value: 1
          Titles:
            – TitleFull: Reading Research Quarterly
              Type: main
ResultId 1