Cognitive and Linguistic Influences on EFL Real Word and Pseudoword Spelling: Predictors and Error Analysis

Abstract

The present study aimed to enhance the understanding of the spelling processes used by young German-speaking learners of English as a foreign language (EFL). Specifically, we sought to (1) compare the children’s accuracy in spelling English real words versus pseudowords to elucidate the role of lexical and sublexical knowledge, and (2) determine which cognitive (phonological awareness, phonological short-term memory, working memory, nonverbal intelligence) and linguistic skills (English receptive grammar and vocabulary) underlie learners’ spelling abilities and misspellings (phonological and orthographic). We followed participants (N = 101) for two years, from the beginning of grade 3 to the end of grade 4. Cognitive skills and linguistic abilities were determined using standardized assessment procedures. Our results indicate that the learners in our study demonstrated greater accuracy in spelling English real words compared to pseudowords. English grammar knowledge significantly predicted real word and pseudoword spelling accuracy, as well as the number of phonological errors, which was the predominant error category. English vocabulary knowledge was a predictor for real word spelling, while nonverbal intelligence predicted pseudoword spelling accuracy. Phonological short-term memory positively predicted the number of orthographic errors (phonologically plausible misspellings but lacking orthographic conventions).

1. Introduction

Although spelling is vital to the writing process (Abbott et al., 2010; Silliman et al., 2018), it remains understudied among young English as a foreign language (EFL) learners. Spelling proficiency is a major developmental challenge in writing English texts (Graham et al., 1997), especially for EFL learners (e.g., Schoonen et al., 2011). Thus, understanding the factors that affect spelling as a foundational skill for later EFL writing is essential to developing targeted instructional literacy strategies. Contrary to first-language (L1) spelling, learning to spell in EFL is more challenging as it involves the continual interplay of two languages (Sammour-Shehadeh et al., 2023), which can involve cross-language influences from the L1 in the initial stages of spelling acquisition, even more so without explicit instruction (Mlakar et al., 2024). Further, unlike learning to spell in an L1 environment, the young EFL learner lacks the same amount of oral input as well as print exposure in the foreign language (Kahn-Horwitz et al., 2012). Therefore, EFL learners will be less familiar with certain orthographic patterns, letter–sound and sound–letter correspondences, and meanings of words in English than L1 English speakers.

As much of the research in second-language (L21) English spelling has its roots in L1 spelling studies, it is worth investigating evidence that suggests that cognitive processes, such as phonological awareness (PA), rapid automatized naming (RAN), or phonological short-term memory (PSTM), underlie spelling development. These processes are commonly shared across languages during the acquisition of literacy skills (for a review, see Genesee et al., 2006). Several studies in North American English as a second-language (ESL) settings have identified these cognitive processes as significant predictors of learners’ English spelling abilities (e.g., Harrison et al., 2016; Jongejan et al., 2007; Keilty & Harrison, 2015; Zaretsky, 2020). These cognitive factors intricately interact with the linguistic components involved in spelling, as spelling is a language activity (Perfetti & Hart, 2002). Several leading theories on literacy development, for example, Ehri’s amalgamation theory (1992) and the repertoire theories (Apel et al., 2004; Apel, 2009; Masterson & Apel, 2007), propose that individuals use their linguistic knowledge stores of phonology, morphology, and orthography during the spelling process. Although these theories differ on whether these skills contribute to literacy development progressively or simultaneously, they all agree that an awareness of sounds, spelling patterns, and meaning is crucial for developing literacy abilities. The study at hand is premised on the linguistic repertoire theory approach in an attempt to shed light on the linguistic underpinnings contributing to young EFL learners’ spelling development. The linguistic repertoire theory (Apel et al., 2004; Apel, 2009; Masterson & Apel, 2007) takes a broad and non-linear perspective on spelling development, and contends that the spelling process draws concurrently on the individual’s linguistic repertoire from their phonological, orthographic, morphological, and semantic knowledge bases both early in spelling development and throughout the entire acquisition period (Apel et al., 2004; Treiman & Kessler, 2014). Further, it suggests that one aspect of linguistic awareness may dominate at certain times, influenced by the current linguistic proficiency level of the speller and the linguistic structure of the word-spelling. Despite the potential for over-reliance on one linguistic skill during the spelling process, other linguistic stores are not excluded (Apel et al., 2004; Werfel et al., 2021). For example, beginner spellers may rely more heavily on their phonological knowledge at the onset of spelling acquisition, whereas more experienced spellers may draw more heavily on their morphological knowledge (Apel et al., 2004). As it was assumed that the young EFL learners in this study had not yet developed sufficient morphological knowledge in English, the error analysis attempted solely to provide insight into the phonological and orthographic knowledge of the learners, and did not include morphological errors.

To explore young EFL learners’ spelling errors and gain insights into the cognitive processes they employ, it is worth taking a closer look at dual-route models of spelling (e.g., Barry, 1994; Cook & Bassetti, 2005; Ellis & Young, 2013). These models propose that two distinct processes are essential for skilled spelling. Individual spellers may take different strategies for successful spelling—either sublexical (based on sound–letter mappings) or lexical/semantic (e.g., Niolaki et al., 2020; Niolaki et al., 2023; Treiman, 2017). Pseudowords (‘trog’ for ‘frog’, ‘shirp’ for ‘shirt’), or low-frequency regularly spelled words (‘hush’), are spelled via the sublexical route, whereas familiar irregular words (‘knife’) as well as familiar regularly spelled words (‘cat’) go through the lexical spelling route. In the analysis of spelling outcomes of young EFL learners, these models might help us to understand the processes children employ during English spelling. Initially, beginner EFL spellers are expected to rely more heavily on sublexical rather than lexical processing, as many words they initially encounter will be unfamiliar. While the sublexical route may be necessary at the beginning of literacy development—helping learners apply phoneme–grapheme correspondences (PGCs) to spell unfamiliar words—it is not very efficient as a default strategy throughout literacy acquisition. In line with extant L1 spelling developmental models, it can be assumed that young EFL learners will gradually develop a wider range of high-quality orthographic representations and will rely less on the time-consuming and cognitively demanding sublexical route (e.g., Moats, 2010). This process involves coordinating the knowledge of a word’s phonological, orthographic, morphological, and semantic properties to ensure the long-term storage of word-specific representations (e.g., Ehri, 2014; Werfel et al., 2021). Via a two-level approach to orthographic knowledge (Zarić et al., 2021), proficient spellers use a robust orthographic lexicon (lexical knowledge store) and have an understanding of the high level of inconsistency in English orthography (general orthographic knowledge), which is crucial for efficient spelling (Niolaki et al., 2020). This knowledge is an integral component of learning to spell, as strong orthographic representations for words are essential to save cognitive resources for higher components of the writing process.

In line with the linguistic repertoire theory, we utilize spelling error analysis in our study to investigate patterns of spelling errors, focusing particularly on errors that can provide insights into the young EFL learners’ phonological and orthographic knowledge. The POMAS (Phonological, Orthographic, and Morphological Assessment of Spelling) framework (Bahr et al., 2012; Silliman et al., 2006) was selected and adapted to examine phonological and orthographic errors. Morphological knowledge was not investigated due to the young EFL learners’ basic level of proficiency in English. For example, misspellings can be categorized as phonologically plausible (e.g., spelling <wont> for <want>, <shi> for <she>, <taim> for <time>) and non-phonologically plausible (<dis> for <this>, <finn> for <thin>) in relation to the target word. A predominance of phonologically appropriate but orthographically inappropriate errors hints towards a reliance on the sublexical spelling route due to a problem with lexical processes, whereas phonologically inappropriate errors indicate difficulties with sublexical strategies. Advanced orthographic skills when becoming a more proficient L2 English speller will consequently lead to fewer phonological and more orthographic errors (Harrison, 2021; Martin et al., 2020; Russak & Kahn-Horwitz, 2015; Zaretsky, 2020). The spelling of phonologically transparent words is relatively easy (e.g., <bed>), as spellers can segment the word into phonemes and apply PGC rules to achieve correct spellings. For words that are less phonologically transparent (e.g., <thumb>), spellers may try to find analogies with similar spellings of known words (<numb>, <plumb>), relying on existing phonological and orthographic representations (Treiman, 2017).

In recent years, studies analyzing spelling errors have been conducted in ESL contexts with various participant groups. These include comparisons of L1 and L2 English in both elementary school children (Harrison, 2021) and adults (Martin et al., 2020), L1 Spanish and L2 English among elementary bilingual speakers (Fashola et al., 1996; Howard et al., 2012; Zaretsky, 2020), and L1 German with L2 French in fourth graders (Weth & Wollschläger, 2019). EFL error analyses in school settings have included learners from various linguistic backgrounds, like Arabic elementary and secondary school learners (Allaith & Joshi, 2011; Russak, 2020, 2022), Chinese elementary students (Dixon et al., 2012), Hebrew young (Russak, 2020) and secondary learners (Kahn-Horwitz et al., 2012; Russak & Kahn-Horwitz, 2015), Italian dyslexic elementary (Palladino et al., 2016) and secondary learners (Arfé & Danzak, 2020), and Spanish fourth-, fifth-, and sixth-grade learners (Hevia-Tuero et al., 2023). To date, only one error analysis study has been conducted with L1 German elementary school learners in an EFL setting (Mlakar et al., 2024). These studies provide valuable insights into the types of spelling errors made by ESL and EFL learners from various linguistic backgrounds and age groups. Understanding these errors is crucial because they reveal underlying language processing challenges and provide evidence of how learners transfer skills from their L1 to their L2. The current study aims to build on this evidence base by focusing on an under-researched group: elementary L1 German learners in an EFL setting.

1.1. Cognitive and Linguistic Predictors of Spelling

Phonological processing is a crucial cognitive and linguistic component to investigate, as it has been shown to predict spelling across various orthographies in both cross-sectional and longitudinal studies (Caravolas et al., 2012; Furnes & Samuelsson, 2011; Moll et al., 2014; Ziegler & Goswami, 2005), outlining the relevance of phonology especially for beginner spellers. Recognizing and manipulating sounds in spoken language is a fundamental skill for understanding how these sounds correspond to letters and letter patterns in written language. Numerous research findings have corroborated the importance of children’s PA skills for proficient spelling in L1 (e.g., Caravolas et al., 2001; Niolaki et al., 2020) as well as in L2 English spelling for ESL learners (Harrison, 2021; Harrison et al., 2016; Jongejan et al., 2007; Keilty & Harrison, 2015) and EFL (Yeung et al., 2013) pre-schoolers. In a study with 75 L1 German young EFL learners, Mlakar et al. (2025) found that cognitive abilities did not predict spelling directly, but PA had predictive effects on spelling mediated by reading comprehension and reading accuracy. Some studies, however, found no influence of PA on fifth-grade EFL learners’ spelling accuracy (Russak, 2020). In different orthographies, reliance on PA varies greatly depending on the PGC regularity of the orthography; as children grow older, PA makes a less significant contribution to spelling ability in opaque orthographies (e.g., Nielsen & Juul, 2016). Another cognitive skill that involves the processing of sound structures in language is PSTM. PSTM is the capacity to hold and manipulate verbal information over a brief period. This cognitive mechanism is crucial for spelling tasks that involve dictation, as it allows for the accurate retention of words and sentences. Additionally, PSTM plays a fundamental role when writing down novel words or pseudowords, as these unfamiliar sound sequences need to be temporarily stored and retrieved. Pseudowords (as used in our study) are typically spelled using the sublexical spelling route without the direct retrieval of word forms from the orthographic lexicon. Pseudowords are spelled using letter knowledge and PGC rules, and they require a more analytical, sound-by-sound approach to spelling. For young EFL learners, the task of spelling English pseudowords places substantial demands on verbal short-term memory. Keilty and Harrison (2015) found that young ESL kindergarten children relied on PSTM and phonemic processing when spelling, whereas L1 English children did not exhibit the same reliance on PSTM. A study by Zaretsky (2020) also highlighted PSTM as the most significant cognitive factor aiding English L2 spelling in a cohort of 60 grade five and six ESL learners from a Spanish-speaking background. Furthermore, recent research by Mlakar and Schilk (forthcoming) involving 49 German-speaking young EFL learners demonstrated that children’s PSTM skills showed notable improvements with phonics instruction during English lessons over a two-year span. In contrast, the whole word approach comparison group did not exhibit significant PSTM effects. However, another study by Mlakar et al. (2025), which examined how cognitive–linguistic factors and reading skills affect L2 spelling performance in 75 L1 German EFL learners, found no significant impact of PSTM on children’s spelling attainment. These results can be complemented by those of a longitudinal study by Sigmund et al. (2024), which examined predictors of spelling in 353 grade one German L1 children. The study found that phonological processing (PA and PSTM), letter knowledge, and intelligence significantly predicted spelling proficiency in grade 1. However, the findings focus on L1 German spelling, with German being more transparent than English on a scale of orthographic depth (Seymour et al., 2003). Building on the overall evidence from these studies, it can be argued that young EFL learners may employ their phonological processing skills when encoding English word-spellings.

The development of literacy skills in a foreign language is cognitively demanding for young EFL learners, requiring not only controlled attentional processing but also the coordination of cognitive processes orchestrated by WM (Kormos, 2023). These processes place substantial demands on WM, as both storage and processing skills are necessary. Similar to L1 spelling, L2 spelling heavily relies on the interplay between phonological memory (holding and manipulating sounds) and orthographic memory (visual representation of words), with WM skills playing an important role in integrating these aspects (Gathercole & Baddeley, 1993). A recent systematic review found that working memory skills are largely unrelated to overall L2 writing proficiency (Li, 2023), but to date only little is known about the impact of WM on L2 learners’ spelling processes. Findings on the influence of WM on L2 spelling are mixed. In L2 contexts involving majority languages, Jongejan et al. (2007) found that WM and syntactic knowledge were the most significant predictors of spelling skills in L1 English children. Yet, these factors were less influential for the ESL learners, as they relied more on phonological processing throughout their elementary school years. Similarly, a study by Czapka et al. (2019) revealed that WM did not predict spelling abilities in L2 German learners. In a study conducted in Germany with young EFL learners, Mlakar et al. (2025) reported that WM skills had no impact on children’s single-word spelling accuracy. However, a study conducted by Engel de Abreu and Gathercole (2012) involving 119 children who spoke L1 Luxembourgish, L2 German, and L3 French demonstrated a significant correlation between L2 German spelling proficiency and working memory abilities assessed via complex span tasks. In sum, these research findings are mixed and may result from methodological differences within the studies (Kormos, 2023). As Kormos (2023) notes, most studies examining the impact of WM on L2 writing have excluded spelling errors from their analyses of accuracy, despite the fact that spelling can be conceptualized as orthographic accuracy—a gap we address in our research.

To clarify these discrepancies and enhance our understanding of what influences EFL spelling, the current study incorporates nonverbal intelligence (NVI) as a cognitive measure. This inclusion is based on the recognition that general intelligence, including nonverbal reasoning abilities, is essential for learning to read and spell. Some studies have highlighted the significance of nonverbal reasoning abilities in early spelling performance in L1 English (e.g., Caravolas et al., 2001; Horn & Packard, 1985; Stanovich et al., 1984) and L1 German elementary school children (e.g., Ennemoser et al., 2012; Schneider & Niklas, 2017; Zarić et al., 2021). In a study with 353 first-grade L1 German participants, Sigmund et al. (2024) found that NVI was a significant predictor for spelling, using the same standardized assessment as in our study.

As spelling involves encoding linguistic units of spoken language into written language and is essentially considered a linguistic activity, the linguistic repertoire theory (Apel et al., 2004) highlights the crucial role that multiple linguistic factors play concurrently in English spelling development. This perspective facilitates an understanding of how phonological, morphological, syntactic, and semantic knowledge collectively influence the ability to spell effectively. Evidence suggests that semantic knowledge is a linguistic knowledge store that contributes to young children’s spelling abilities (Kim et al., 2013; Perfetti & Hart, 2002). The quantity (size) and quality (depth) of words in a child’s vocabulary and their capacity to pick up new words are both examples of semantic knowledge (Werfel et al., 2021). In this study at hand, only one measure of the learner’s vocabulary has been investigated, namely, receptive vocabulary (size), which tested the mapping of the young learner’s phonological and semantic knowledge. This combined knowledge is an integral linguistic ingredient of orthographic mapping (Ehri, 2014), which, together with morphological and orthographic knowledge, gradually leads to orthographic accuracy. Grammar knowledge is another key linguistic skill linked to spelling proficiency, although defining grammatical knowledge is challenging due to its broad scope, encompassing syntax, morphology, and metalinguistic awareness. Research indicates that young children’s spelling abilities are influenced by their grasp of morphological and syntactic elements, collectively referred to as morphosyntactic knowledge, which is crucial in the spelling process (e.g., Apel et al., 2012). This ambiguous definition, alongside varying methodological approaches, may explain inconsistent findings regarding the impact of grammar skills in predicting learners’ L2 English spelling skills. To our knowledge, no study has specifically focused on whether grammar is a predictor for pseudoword spelling, and only three studies from the Canadian ESL context have explored the role of syntactic knowledge or awareness in L2 real-word spelling. Keilty and Harrison’s (2015) research on 77 ESL and L1 English Canadian Kindergarten children suggests that sophisticated spelling2 is predicted by good syntactic knowledge and phonological processing skills, but also increased performance on early word reading irrespective of language status. However, using the same syntactic knowledge test, Harrison et al. (2016) found no effect of syntactic knowledge on ESL children’s spelling proficiency. In another study with elementary ESL learners in Canada, Jongejan et al. (2007) concluded that although syntactic awareness did not play a role in the prediction of children’s spelling abilities at the word level, it is likely that such factors would have influenced later writing ability. However, the relationship between grammar knowledge and spelling remains unexplored in young EFL classrooms in Germany.

1.2. German and English Spelling

One major aspect when analyzing spelling errors is L1’s influence on the mechanisms of spelling in EFL. Central to the spelling development of young EFL learners is the cross-linguistic influence between the two languages (Chung et al., 2019; Cook & Bassetti, 2005); in this study, German and English. These two alphabetic languages share the same alphabet but differ in their orthographic depth (Landerl, 2005; Seymour et al., 2003). As the young EFL learners in this study are already literate in their L1, it can be assumed from cross-linguistic research that the EFL children will share the same cognitive and linguistic spelling predictors (language-general, e.g., PA), which will vary in degree as a result of the linguistic proximity and orthographic depth between the two languages (Caravolas et al., 2012; Kahn-Horwitz et al., 2011; Moll et al., 2014). From the contrastive–typological perspective (language-specific), similarities and differences in the linguistic structure between the L1 and L2 word-spellings will influence L2 spelling ability. Therefore, it can be assumed that cross-linguistic influence involving language-general and language-specific factors will influence the word-spelling as well as the cognitive resources and linguistic knowledge bases of the EFL learner during the spelling process (script dependent hypothesis; Geva & Siegel, 2000). With regard to the structural similarities and differences between the two languages, English and German both use the same letter set (apart from ä, ö, ü and ß). Both German and English orthography use multi-letter graphemes (jacket—Jacke), silent letters (hate—Zahl), or onset and end clusters (green—grün), but with more complex clusters in German (e.g., sword—Schwert). In general, German EFL learners often struggle with English phonemes that do not exist in their phonetic inventory, such as the interdental fricatives /θ/ or /ð/. These sounds are absent in German, leading to difficulties in both perception and production. In research conducted with a sub-sample of 75 young L1 German EFL learners (Mlakar et al., 2024), using the spelling tests from the study at hand, error rates across phonologically and orthographically novel spelling patterns revealed a remarkably high occurrence of spelling mistakes with the /θ/ sound, as indicated by a 92.67% error rate. This was followed by a 67.33% misspelling rate for the /ð/ sound, and a 46.67% rate for /ʃ/ spelled as <sh>. In contrast, errors involving the silent <e> had a significantly lower rate of 27.33%. In summary, these error patterns highlight the specific challenges German EFL learners face due to their linguistic background, setting the stage for a deeper exploration of the linguistic factors at play in their spelling development.

This current study is anchored in the linguistic repertoire theory (Apel et al., 2004; Apel, 2009; Masterson & Apel, 2007) and aims to shed further light on young EFL learners’ spelling development by examining the linguistic underpinnings that influence their ability to spell both real words and pseudowords. Consistent with this theory, it is anticipated that our participants will draw concurrently on their linguistic knowledge during the spelling process, and not on one single knowledge store in isolation. Our focus will specifically be on analyzing phonological and orthographic knowledge within spelling errors. We selected a spelling test composed of monomorphemic words, which are particularly suited for assessing EFL learners’ spelling skills during the early stages of their English language and literacy development. Consequently, the choice of these words does not allow for the analysis of morphological errors.

2. The Present Study

This study compares children’s real word and pseudoword spelling accuracy to understand the processes involved in spelling, focusing on lexical versus sublexical knowledge. We analyze phonological and orthographic misspellings to explore whether L1 German-speaking EFL learners use the same cognitive (PA, PSTM, WM, NVI) and linguistic skills (L2 English receptive grammar and vocabulary) in spelling both real words and pseudowords. Additionally, we investigate how these processes influence children’s misspellings. To address our objectives, we pose the following research questions:

 

Research Question 1a

Do the German-speaking EFL learners in our study demonstrate greater accuracy in spelling English real words compared to pseudowords?

Hypotheses

We hypothesize that the participants in our study will spell real words more accurately than pseudowords, because real-word spelling involves both lexical–semantic and sublexical processes, whereas pseudoword spelling relies solely on sublexical processes.

 

Research Question 1b

To what extent do the learners in our study rely on phonological and orthographic knowledge for spelling in EFL?

Hypotheses

Contingent on the linguistic repertoire theory (Apel et al., 2004), we expect that young EFL learners will draw from both their phonological and orthographic knowledge stores and make both phonological and orthographic errors. However, we anticipate more phonological errors, as the participating children in our study are at the onset of L2 spelling acquisition and lack spelling practice and print exposure. Further, crosslinguistic influence (language-specific) is expected to increase the prevalence of phonological errors among participants due to lack of L2 English sound–letter knowledge.

 

Research Question 2a

Which cognitive and linguistic skills predict real word and pseudoword spelling proficiency in German-speaking EFL learners?

Hypotheses

We anticipate that according to the DR model, phonological processing skills will be strong predictors of pseudoword spelling (e.g., Niolaki et al., 2022). This is because spelling primarily involves phonological encoding—the process of converting phonemes into corresponding graphemes. As research findings have demonstrated the importance of children’s PA skills for proficient spelling in both ESL (Harrison, 2021; Harrison et al., 2016; Jongejan et al., 2007; Keilty & Harrison, 2015) and EFL (Yeung et al., 2013) learners, we anticipate that participants’ PA skills will predict real-word spelling accuracy.

 

Research Question 2b

Which cognitive and linguistic skills predict the types of phonological and orthographic errors L1 German EFL learners make?

Hypotheses

We anticipate that phonological errors in L1 German EFL learners may be influenced by phonological ability. PA and PSTM as sublexical cognitive skills could affect learners’ capacity to accurately process phonological information, potentially resulting in fewer phonological errors when these skills are strong. Even though exploratory in nature, we hypothesize that higher NVI will predict a lower number of orthographic errors made. This expectation arises because visual reasoning, a component of NVI, might tap into orthographic processing, as it involves symbolic processing and manipulation (Zarić et al., 2021).

3. Materials and Methods

3.1. Participants and English Learning Context

A total of 101 young EFL learners (49 boys, 52 girls) from three primary schools in Lower Saxony, Germany, attending five classes, were included in the study. The schools were located in predominantly middle-class, semi-urban areas with comparable socio-economic status, as indicated by parent questionnaires. The children had a mean age of 10.28 years at the end of grade 4 when the spelling tests were administered (SD = 0.83; range = 8.74 to 11.76 years). In total, 84 learners exclusively used German at home, while 17 children also spoke languages other than German within their families, including Arabic (5 children), Bulgarian (3), Polish (3), Kurdish (1), Russian (1), Hungarian (1), Armenian (1), Ukrainian (1), and Turkish (1). After beginning formal schooling at the age of six, most German schools employ a phonics approach to teach reading and spelling, which enables learners to simultaneously learn PGCs and practice sounding them out. Responses from parents in the questionnaires revealed that out of the 17 L2 German-speaking participants, 13 had started formal schooling in Germany in grade 1. Among the 101 participating children, 5 were identified as having special educational needs, which include dyslexia (2), general learning difficulties (2), and social-emotional challenges (1). Additionally, one student was a refugee who had arrived in Germany at the beginning of the grade 3 and, like the rest of the group, received EFL lessons for the first time. The EFL instruction for the five classes in our study was provided by three highly experienced teachers. Each of them held a university degree in Teaching English as a Foreign Language (TEFL) and had considerable expertise in the field, with 35, 26, and 16 years of teaching experience, respectively. Learners received EFL instruction through two 45-min sessions per week. The teaching approach in all classes was based on a communicative language teaching framework and instruction covered all four primary language skills: listening, speaking, reading, and writing. Within the participant group, children received one of the following types of literacy instruction over two school years: phonics, whole-word, combined, or no specialized instruction (Mlakar et al., 2024). Only the phonics group received explicit spelling instruction, while all other groups received no systematic, direct instruction. No group had spelling tests as part of their regular EFL lessons, making the spelling test a novel experience for the children. Parent questionnaires indicated that before entering grade 3, the participants of our study had very limited exposure to the English language. Most of the contact time with English was facilitated by the EFL teachers and reinforced through learning materials, EFL textbooks—all classes used Playway 3 and 4 (Gerngross et al., 2020a, 2020b)—and other written resources like children’s books and worksheets. All lesson contents were based on Lower Saxony’s core curriculum for English at primary schools (Years 3–4), the Common European Framework of Reference for Languages (CEFR; Council of Europe, 2020) and the schools’ syllabi for TEFL. The study received approval from a research ethics commission, and parental consent was obtained.

3.2. Data Collection and Analysis

Our study employed spelling, cognitive, and linguistic assessments for data collection. We monitored the participants over a two-year period, starting with the cognitive–linguistic assessments (PA, PSTM, WM, NVI, L2 grammar, and L2 vocabulary) at the beginning of grade 3 and concluding with the spelling tests (real words and pseudowords) at the end of grade 4. Individual tests in quiet surroundings were performed for measuring PA, PSTM, and WM. L2 spelling, L2 grammar/vocabulary, and NVI were tested in group settings. Instructions for all measures were provided in German. The data collection team consisted of researchers and graduate students who had undergone formal training in administering and scoring the specific measures.

3.2.1. Outcome Measures

L2 English real word spelling. In this task, we measured our participants’ ability to spell both familiar (non-novel) and unfamiliar (novel) phonological and orthographic properties of English. We assessed four categories. The first category targeted the spelling of eight words containing four novel phonemes for German L1 speakers (/æ/, /θ/, /ð/, and /ɒ/), such as ‘had’, ‘thin’, ‘with’, or ‘frog’. The second category focused on evaluating the spelling of four words with orthographic features that are unfamiliar in German orthography, such as the silent <e> (‘like’ and ‘time’) and /ʃ/ spelled <sh> (‘shirt’ and ‘shop’). The third category covered seven high-frequency words, including words with standard (e.g., ‘three’) and irregular PGCs (e.g., ‘eight’). The fourth category consisted of four words with two non-novel phonemes, /e/ and initial /h/, as in ‘went’ and ‘help’. Participants were presented with 23 audio-recorded words pronounced in British English (Appendix A). Each word was articulated three times; first in isolation, then within a sentence, and finally in isolation again. Learners transcribed these words onto paper. Using these 23 words, we calculated the number of correct spellings (marked as either correct or incorrect), along with the number of phonological and orthographic errors.

L2 English pseudoword spelling. We used the words from categories I, II, and IV of the real-word spelling test to create pseudowords (e.g., ‘hab’ for ‘had’, ‘trog’ for ‘frog’). We carefully restricted the number of orthographically ambiguous options by selecting pseudowords that closely adhere to common PGC in English and by avoiding homophones and English real words. Participants listened to 16 audio-recorded pseudowords (Appendix A) and transcribed these words onto paper. We marked the spellings as correct or incorrect.

3.2.2. Predictor Measures

L1 German phonological awareness. The Test of Basic Competencies for Reading and Spelling Skills (BAKO 1–4; Stock et al., 2017) was used to evaluate our participants’ phonological awareness skills at the phoneme level. Three sub-tests were administered: non-word segmentation, vowel substitution, and vowel length. The maximum score that can be reached is 30.

Phonological short-term memory. We measured short-term verbal memory by administering the Digit Span Forward (DSF) sub-test from the Wechsler Intelligence Scale for Children—4th Edition (Petermann & Petermann, 2011). The children repeated a list of pre-recorded digits pronounced in German, starting with 2 digits and progressively increasing in number.

Working memory. Children’s working memory skills were evaluated using the Wechsler Intelligence Scale for Children—4th Edition’s (Petermann & Petermann, 2011) Digit Span Backwards (DSB) and Letter-Number Sequencing (LNS) subtests, both administered in German. The DSB tested reversed digit recall, while the LNS tested letter and number arrangement and recall.

Nonverbal intelligence. Participants’ general cognitive abilities in logical thinking and reasoning were evaluated by using a subtest from the German school readiness test (BUEGA, Esser et al., 2008). Learners had to complete incomplete matrices by selecting the correct element from a range of five to eight options. The maximum score that can be reached is 41.

L2 English receptive grammar. Participants’ English receptive grammar knowledge was elicited with the ELIAS Grammar Test II (Kersten et al., 2012), which contains 72 items for 12 grammatical phenomena (e.g., possessive pronouns, negation, or passive). Learners listened to 72 grammatical prompts and were asked to choose the correct picture from three possible options. Koch et al. (2021) confirmed the test’s validity by using a Rasch model, demonstrating a strong model fit through a bootstrap goodness-of-fit test.

L2 English receptive vocabulary knowledge. The British Picture Vocabulary Scale III (Dunn et al., 2009) test was used to assess the participants’ receptive (hearing) vocabulary. The paper/pen test was administered in a group setting using a recorded format. Children listened to words from the recording and marked the correct image (from four options) that matched the spoken word. Sets A to J were used (120 items). In a study involving 388 primary school children (Ponto, 2024), the dataset for the BPVS demonstrated high internal consistency, with a Cronbach’s alpha of 0.94.

3.3. Analyzing Learners’ Spelling Errors

We used a modified version of the POMAS (Phonological, Orthographic, and Morphological Assessment of Spelling) framework (Bahr et al., 2012; Silliman et al., 2006) to analyze children’s spelling errors in our study. We scored phonological errors when children violated the phonological plausibility of a word (<wis> for <with>; <dis> for <this>), while orthographic errors were phonologically plausible spellings but lacked orthographic conventions (<onet> for <want>; <vather> for <father>; <whent> for <went>). The sample was scored by two authors and rescored by one trained graduate student. The inter-rater reliability, as analyzed by Cohen’s Kappa coefficient, was 0.87 (almost perfect agreement).

4. Results

First, we explored the extent to which young German EFL learners used lexical and/or sublexical knowledge when spelling English real words and pseudowords. We also investigated the number of phonological and orthographic errors. Out of the 101 children who spelled a total of 2323 real words, 1148 errors were found, accounting for 49.41% of the words produced. Errors were found in both POMAS categories used for error analysis—phonological and orthographic. Specifically, 897 errors were phonological in nature, representing 78.14% of all errors, while 251 errors were due to orthographic misspellings, making up 21.86%. A closer observation of the rate of errors made when spelling pseudowords found that the participants in our study spelled 1148 (= 71.04%) out of 1616 pseudowords incorrectly.

In a more refined error analysis, several consistent patterns emerged that highlight specific phonological and orthographic influences from L1 German. One significant error pattern involved the substitution of the English diphthong <ou> with the German <au> (‘south’ spelled ‘sauf’), reflecting that learners may not yet have internalized English orthographic rules. Additionally, learners frequently omitted the silent <e> at the end of words and represented <i> with <ei> (‘time’ spelled ‘teim’, ‘like’ spelled ‘leik’). The English /θ/ phoneme (as in ‘thin’ and ‘south’) and /ð/ phoneme (as in ‘with’ and ‘this’), which pose particular challenges due to their absence in German, were often replaced with <f>, <s>, or <d> by L1 German spellers. This substitution demonstrates the difficulties these learners have in producing these unfamiliar phonemes, resulting in a large number of phonological errors. Furthermore, the phenomenon of the final devoicing of consonants in L1 German influenced spelling errors among learners. Voiced consonants at the ends of words, such as <g> in ‘frog’, were often spelled with their voiceless counterparts, resulting in misspellings like ‘frok’ (or ‘had’ spelled as ‘hat’). Another error pattern observed was the substitution of the English <f> with the German <v>, reflecting learners’ strong reliance on phonological strategies influenced by L1 German.

Table 1. Examples of frequent misspellings.

Real Word	Error Category
	Phonological Error	Orthographic Error
thin	fin/finn	-
with	wif/wis	-
this	dis	-
frog	frok	-
around	eraund	-
had	-	hed/head
went	-	whent
shop	-	schop
want	-	wont/onet
she	-	shi
father	-	vather
does	-	das
time	-	teim/taim
south	-	sauf

Note. Phonological errors impact the phonetic representation of words, while orthographic errors preserve pronunciation but deviate from conventional spelling norms.

highlights common examples of phonological and orthographic errors identified across the sample.

Our next research question aimed to investigate whether children’s accuracy in spelling real words and pseudowords, as well as the number of phonological and orthographic errors they made, depended on different cognitive–linguistic predictors. First, we conducted correlations between early cognitive–linguistic variables (assessed at the beginning of grade 3) and spelling data from the end of grade 4 (almost two years later) to explore which early skills are associated with later literacy attainment. Our analyses were conducted using IBM SPSS Version 29, applying an alpha level of 0.05 to test the significance of all analyses. To test for bivariate relationships between the linguistic skills (L2 receptive grammar and vocabulary knowledge), cognitive skills (PSTM, WM, NVI, and PA), and the spelling variables, we calculated linear Pearson correlations between all variables (

Table 2. Pearson correlations.

Variables	1.	2.	3.	4.	5.	6.	7.	8.	9.
1. L2 Grammar
2. L2 Vocabulary	0.17
3. L1 PA	0.35 **	0.32 **
4. PSTM	0.15	0.18	0.13
5. WM	0.04	0.26 *	0.40 **	0.20 *
6. NVI	0.06	0.20 *	0.33 **	0.06	0.17
7. L2 Spelling (RW)	0.36 **	0.33 **	0.33 **	0.11	0.17	0.20 *
8. L2 Spelling (PW)	0.13	−0.02	0.26 **	0.02	0.16	0.22 *	0.30 **
9. L2 Phon. errors	−0.33 **	−0.31 **	−0.36 **	−0.19	−0.18	−0.27 **	−0.89 **	−0.30 **
10. L2 Orth. errors	−0.07	−0.11	0.05	0.18	0.02	0.11	−0.19	0.01	−0.19

Note. * p < 0.05, ** p < 0.01. L1 PA = L1 phonological awareness, PSTM = phonological short-term memory, WM = working memory, NVI = nonverbal intelligence, L2 Phon. Errors = L2 phonological errors, L2 Orth. errors = L2 orthographic errors.

).

We found several positive and negative significant correlations. L2 English receptive grammar and vocabulary, PA, and NVI were significantly correlated with L2 English real-word spelling. Higher values of these cognitive–linguistic variables were linked to better real-word spelling accuracy. These skills also showed significant correlations with the number of L2 English phonological errors, with higher values corresponding to fewer phonological errors. PA and NVI were significantly positively correlated with pseudoword spelling, with higher values resulting in more proficient pseudoword spelling. We did not find any correlations between these variables and the number of orthographic errors.

Next, we ran a linear mixed model analysis using JASP Version 0.18-30 (JASP TEAM, 2024) and employed the pre-test data collected at the onset of grade 3 (covering cognitive and linguistic measures) to predict L2 spelling skills at the end of grade 4. We included L2 English real-word and pseudoword spelling accuracy, and the number of phonological and orthographic errors, as dependent variables in four separate analyses. As fixed effects, we added PA, PSTM, NVI, WM, L2 grammar, and L2 vocabulary. We further incorporated the schools as a random effect-grouping variable accounting for variations between classes, including differences in schools and teaching methods. The data consisted of 101 observations across three clusters (i.e., schools). We did not report fit indices (e.g., deviance, AIC, BIC) because we did not compare different models. The models were fitted using restricted maximum likelihood estimation. Significance was calculated using the Satterthwaite method to estimate degrees of freedom and generate p-values for mixed models with square sums type III. We first tested the following model: real-word spelling ~ PA + PSTM + NVI + WM + grammar + vocabulary + (1|school). We found significant predictive effects of L2 English grammar and vocabulary (

Table 3. Effects of the linear mixed model on L2 English real word spelling and pseudoword spelling.

	L2 Real-Word Spelling					L2 Pseudoword Spelling
Fixed effects	β	SE	df	t	p	β	SE	df	t	p
Intercept	−6.24	4.34	77.22	−1.44	0.154	−1.96	1.92	65.77	−1.02	0.311
L1 PA	0.13	0.11	92.57	1.16	0.259	0.02	0.05	92.42	0.46	0.644
PSTM	0.02	0.29	91.99	0.08	0.939	−0.07	0.12	92.07	−0.59	0.555
WM	−0.11	0.31	92.94	−0.37	0.716	0.09	0.13	92.69	0.66	0.510
NVI	0.06	0.10	93.18	0.65	0.520	0.12	0.04	92.83	2.85	0.005
L2 Grammar	0.37	0.13	93.19	2.77	0.007	0.18	0.06	92.85	3.08	0.003
L2 Vocabulary	0.14	0.06	92.93	2.37	0.020	−0.01	0.03	92.66	−0.51	0.612
random effects	Var	SD				Var	SD
school (Intercept)	1.87	3.49				1.04	1.08

Note. L1 PA = L1 phonological awareness, PSTM = phonological short-term memory, WM = working memory, NVI = nonverbal intelligence.

). The other variables did not significantly predict real-word spelling. Second, we tested the following model: pseudoword spelling ~ PA + PSTM + NVI + WM + grammar + vocabulary + (1|school). Children’s pseudoword spelling accuracy was significantly predicted by NVI and grammar but not by the other variables.

To investigate the predictors of the number of phonological and orthographic spelling errors that the participants in our study made, we tested the following model: number of phonological errors ~ PA + PSTM + NVI + WM + grammar + vocabulary + (1|school). We found a significant predictive effect of L2 receptive grammar knowledge on the number of phonologically plausible (but orthographically inaccurate) misspellings. Lastly, we tested the following model: number of orthographic errors ~ PA + PSTM + NVI + WM + grammar + vocabulary + (1|school). The number of orthographic misspellings (that were phonologically plausible) was significantly predicted by PSTM, but not by the other variables (

Table 4. Effects of the linear mixed model on the number of L2 English phonological and orthographic errors.

	L2 Phonological Errors					L2 Orthographical Errors
Fixed Effects	β	SE	df	t	p	β	SE	df	t	p
Intercept	29.62	4.28	81.74	6.93	<0.001	0.84	1.73	80.53	0.49	0.629
L1 PA	−0.14	0.11	93.25	−1.20	0.234	0.02	0.05	92.75	0.42	0.678
PSTM	−0.29	0.29	91.93	−0.99	0.323	0.23	0.12	92.10	2.04	0.044
WM	0.06	0.31	93.65	0.18	0.854	0.03	0.12	93.12	0.23	0.819
NVI	−0.17	0.10	93.99	−1.72	0.088	0.07	0.04	93.39	1.76	0.082
L2 Grammar	−0.32	0.14	93.97	−2.37	0.020	−0.04	0.05	93.39	−0.79	0.433
L2 Vocabulary	−0.11	0.06	93.77	−1.88	0.064	−0.03	0.02	93.12	−1.37	0.175
random effects	Var	SD				Var	SD
school (Intercept)	1.14	1.31				0.69	0.47

Note. L1 PA = L1 phonological awareness, PSTM = phonological short-term memory, WM = working memory, NVI = nonverbal intelligence.

).

5. Discussion

The present study aimed to compare the performance of young L1 German EFL learners in accurately spelling English real words and pseudowords. We also examined the cognitive and linguistic contributors to EFL real-word and pseudoword spelling accuracy, and investigated the roles of these skills in making either phonological or orthographic errors.

Addressing research question 1a, which investigates whether the German-speaking EFL learners in our study demonstrate greater accuracy in spelling English real words compared to pseudowords, our cohort correctly spelled 50.59% of real words, while their accuracy for pseudowords was significantly lower, at just 28.96%. As pseudoword spelling relies solely on sublexical processes in the dual route model of spelling (e.g., Barry, 1994), our participants obviously could not master accurate spellings based on phonemic and graphemic patterns and rules. Processing pseudowords typically requires greater cognitive effort (e.g., Hershman et al., 2024), because learners must rely more heavily on phonological decoding and less on stored word representations. The expected difficulty of spelling pseudowords correctly might highlight gaps in the dual route model’s sublexical processing in the EFL learners.

Turning to research question 1b (To what extent do the learners in our study rely on phonological and orthographic knowledge for spelling in EFL?), our analysis using the POMAS categories (Bahr et al., 2012; Silliman et al., 2006) revealed a predominance of phonological errors, which constituted 78.14% of all errors, compared to 21.86% orthographic errors. Within the linguistic repertoire theory (Apel et al., 2004), most learners over-relied on phonological knowledge and produced spellings that were not phonologically plausible (e.g., ‘dis’ for ‘this’, ‘fin’ for ‘thin’, ‘wis’ for ‘with’). This pattern suggests that while many learners relied heavily on their phonological knowledge, it was not to the exclusion of their orthographic knowledge. Unlike stage theories of spelling, which suggest a sequential development, linguistic repertoire theory approaches suggest that children use their knowledge of phonology, orthography, morphology, and mental graphemic representations (MGRs) concurrently. The predominance of phonological errors in our spelling samples highlights differences in the phonological structures of English and German, reflecting their distinct typologies and suggesting L1 interference, as evidenced by theories of cross-linguistic influence (Chung et al., 2019; Cook & Bassetti, 2005; Geva & Siegel, 2000). A more detailed error-analysis (Mlakar et al., 2024) of a sub-sample of 75 L1 German EFL learners provided further evidence of the clearly observable influence of L1 German phonology and orthography on children’s misspellings. Additionally, learners from shallow, alphabetic L1 backgrounds, like German, often apply phonological skills when spelling in an L2 due to the transparent PGC in their L1 (Bhide, 2015). Considering the predictable PGC in German orthography and the emphasis on phonics in L1 German literacy education (Landerl, 2017), it is plausible that learners in our study relied heavily on their L1 phonological knowledge to spell English words, resulting in a majority of phonological misspellings. The prevalence of both phonological errors and orthographic misspellings observed in the sample suggests a need for instructional strategies that enhance the integration of both phonological and orthographic knowledge bases into children’s linguistic repertoire.

The results from research question 2a, which explored the cognitive and linguistic skills predicting real-word and pseudoword spelling proficiency among German-speaking EFL learners, reveal that L2 English receptive grammar knowledge and L2 receptive vocabulary size at the beginning of grade 3 were significant predictors of real-word spelling accuracy two years later. Meanwhile, L2 grammar and NVI predicted pseudoword spelling proficiency. We hypothesized, following the DR model of spelling (e.g., Barry, 1994), that phonological processing skills would be powerful predictors of pseudoword spelling ability (Niolaki et al., 2022). Several previous studies have consistently indicated that PA tested in English is a foundational skill for ESL learners’ real-word spelling accuracy (e.g., Harrison, 2021; Harrison et al., 2016; Jongejan et al., 2007; Keilty & Harrison, 2015). In our study, PSTM and L1 German PA predicted neither children’s EFL real word nor pseudoword spelling accuracy. Two theoretical considerations may help explain these unexpected results. First, while phonological skills are critical for spelling in L1 German (e.g., Pfost, 2015; Sigmund et al., 2024), their transfer to L2 English might be less straightforward due to significant differences in phonology and orthography between German and English (Landerl, 2005). Second, the prominence of L2 receptive grammar knowledge and vocabulary size as predictors suggests that linguistic skills may exert a stronger influence on English spelling proficiency for EFL learners than L1 German phonological skills. These findings indicate that L2 English spelling skills might rely more heavily on the acquisition of language-specific linguistic and cognitive abilities, such as understanding L2 English grammatical structures and developing a robust vocabulary. This reliance arises because spelling connects language and literacy in cognitive development by facilitating the creation of specific, stored mental orthographic representations of written words (Apel, 2009). This intricate relationship involves integrating a word’s distinct pronunciation, orthography, and meaning (i.e., morphology/syntax and vocabulary) at both the lexical and the sublexical level. If this bonding is successful, the mental lexicon can continue to expand in the number of stored orthographic representations, and as a result, this will facilitate fluent reading and spelling (e.g., Bahr et al., 2012; Ehri, 2014; Treiman, 2017). Children’s spelling abilities are significantly influenced by their understanding of morphological and syntactic components, collectively referred to as morphosyntactic knowledge. This integrated knowledge is crucial in the spelling process (e.g., Apel et al., 2012). Building on this understanding, we turn to L2 English vocabulary as a predictor of EFL learners’ real-word spelling skills. Participants with a larger receptive (listening) vocabulary of familiar word meanings were able to produce more accurate, word-specific spellings. This suggests that an enriched vocabulary bolsters the learners’ ability to access and utilize detailed lexical information during spelling tasks. The link between L2 vocabulary and L2 spelling has been anticipated, as some empirical studies have highlighted vocabulary knowledge as a significant factor in L2 English spelling proficiency (e.g., Zaretsky, 2020). Notably, the significant prediction of pseudoword spelling by children’s NVI skills also highlights the essential role of nonverbal reasoning abilities in spelling acquisition. Given that general intelligence, including nonverbal reasoning, is a crucial prerequisite for reading and spelling development (e.g., Stanovich et al., 1984), these cognitive abilities likely enable learners to employ problem-solving and analytical thinking strategies effectively when spelling novel word forms. NVI may act as a proxy for the cognitive flexibility and strategic application needed to support the sublexical route, thus aligning with the assumptions of the dual route model.

Turning to research question 2b, we initially hypothesized that phonological errors would primarily be influenced by learners’ phonological ability, as sublexical cognitive skills (PA, PSTM) were expected to play a critical role in accurately processing phonological information. However, our findings reveal that L2 English receptive grammar knowledge stood out as a significant predictor of phonologically plausible but orthographically inaccurate misspellings. This outcome suggests that our participants’ understanding of English grammatical structures not only underpinned their general real-word spelling accuracy—a finding consistent with studies on ESL learners (Keilty & Harrison, 2015) and L1 English learners (Muter & Snowling, 1997; Nunes et al., 1997)—but also contributed to reducing phonological errors. The robust correlations between L2 grammatical knowledge, PA, and real-word spelling accuracy (Table 2) suggest an integrated framework that highlights the necessity of nurturing a diverse linguistic repertoire to facilitate more effective spelling acquisition.

When examining the factors contributing to children’s orthographic errors, we found that PSTM positively predicted the number of orthographic errors. These errors are characterized as phonologically plausible misspellings that do not follow orthographic conventions (for example, ‘father’ spelled ‘vather’). Our initial hypothesis suggested that NVI would predict the number of orthographic errors, given the potential of visual reasoning to influence orthographic processing (Zarić et al., 2021). However, with NVI not emerging as a predictor, this outcome indicates that orthographic processing in our study may depend more on phonological knowledge than on general nonverbal reasoning skills. The impact of PSTM as a crucial cognitive skill for L2 spelling has been highlighted in previous studies involving pre-school learners (Keilty & Harrison, 2015) and grade five and six ESL students (Zaretsky, 2020), and across different orthographies (Moll et al., 2014). Our findings suggest that as participants enhance their phonological skills, which are essential for spelling proficiency, they may face new challenges in mastering orthographic conventions. This development could result in an increase in orthographic errors rather than a reduction in phonological errors, contrary to our initial hypothesis. The orthographic misspellings in the sample indicate that participants might have captured letter–sound knowledge without yet having established accurate MGRs (resulting in misspellings such as ‘hed’ for ‘had’ or ‘laik’ for ‘like’). A strong research base provides evidence that more advanced orthographic skills when becoming more proficient in L2 English spelling will consequently lead to fewer phonological and more orthographic errors (Harrison, 2021; Martin et al., 2020; Russak & Kahn-Horwitz, 2015; Zaretsky, 2020).

6. Limitations

A noted limitation in our study arises from the binary approach to error analysis for phonological and orthographic errors (correct/incorrect). This binary classification might oversimplify the complexities involved in EFL learners’ spelling development. A more fine-grained spelling error analysis using orthographic (e.g., letter distance) or phonological measures (e.g., phoneme representation) would probably provide more insight into how children gradually master increasingly accurate phonological content and phoneme–grapheme mappings. Second, the results concerning the grammar–spelling connection need more investigation. Some of the categories of the ELIAS grammar test we administered addressed morphological aspects (e.g., plural -s, 3rd person -s), while other items were more focused on syntax or functionality (e.g., subject–verb–object, negation, passive forms). It is therefore difficult to make explicit claims about the contributions of morphological or syntactical knowledge to the spelling–grammar relationship based on our findings. As knowledge of morphology and syntax plays a crucial role in spelling sophistication (e.g., Levesque et al., 2020) and the current study only measured the spelling of monomorphemic words, a more detailed analysis that considers both morphological and syntactic knowledge would be essential for understanding how these knowledge sources contribute to the spelling of polymorphemic words, both in isolation and in the context of sentences. Additionally, the administered ELIAS grammar test might not have been sensitive enough to differentiate more finely between grammar and lexical knowledge. As lexical–semantic processing is more important for EFL beginner learners (Clahsen & Felser, 2018), our results should be interpreted with certain caution. Additionally, while grammar, vocabulary, and spelling were assessed in L2 English, the PA measures were conducted in German. This difference might account for the lack of a strong association between PA and spelling in our study. Future studies should include an L2 English PA assessment to better understand respective contributions to spelling proficiency in EFL learners.

7. Conclusions and Implications for the Classroom

The results of our study provide important information on the acquisition of spelling skills with a unique population of L1 German EFL learners. Our research implies that young EFL learners might benefit from an instructional approach that connects spelling with the linguistic components that underlie word-specific knowledge, that is, phonological, semantic, morphological, and orthographic information. These sources are vital to promote the acquisition of robust MGRs to become proficient spellers (Apel et al., 2004). In the German EFL context, it is important to recognize that spelling is fundamentally a cognitive–linguistic process rather than rote visual memorization (Treiman, 2017). If young EFL learners are taught to spell English words through rote memorization rather than by explicitly learning PGC for complex patterns, they may not recognize the phonological structure of a word as a valuable tool for spelling (Saban & Kahn-Horwitz, 2022). Consistent with the linguistic repertoire theory (Apel et al., 2004; Apel, 2009; Masterson & Apel, 2007), which recommends the integrated and explicit spelling instruction of concurrent linguistic factors, our study supports a similar approach for young EFL spellers. Guided and supportive spelling instruction could introduce beginner EFL learners to the multiple linguistic components that underlie English word-spellings. In addition, our study suggests that young EFL learners may benefit from spelling instruction that highlights the typological differences between the two languages as a result of crosslinguistic influence. In sum, this method of integrated and explicit spelling instruction could play a critical role in facilitating the spelling development of EFL learners.