Content area
The relationship between lexical prosodic competence and word reading was explored, taking into consideration of phonemic awareness and morphological awareness. A total of 90 Chinese university freshmen learning English as a foreign language participated in the study. Two measures were administered to examine their English lexical prosodic competence, i.e., word stress perception and stress production in nonneutral derivatives (derived forms with stress-shifting suffixes, e.g., able + ity → ability). Two measures were administered to examine their English word reading ability, i.e., word identification and nonword decoding. A series of hierarchical regression and mediation analyses were conducted. Both lexical prosodic competences predicted word identification and nonword decoding, respectively, after controlling for nonverbal reasoning, oral vocabulary, and other metalinguistic skills. Word stress perception continued to independently explain unique variance in word identification. Despite the different prosodic systems between Mandarin and English, the adult EFL learners possessed a sensitivity to the primary stress in English multisyllabic words and may have implicit awareness of stress shift cued by nonneutral suffixes. Both types of lexical prosodic competence had direct effects on word identification and nonword decoding. Their relationships to word identification were partially mediated by phonemic awareness and morphological awareness, whereas their relationships to nonword decoding were partially mediated by phonemic awareness only. The combined results provide evidence for the important role of lexical prosodic competence in word reading development and also point to the relevance of this role for adult EFL learners.
Introduction
Researchers have long agreed that phonemic awareness and morphological awareness are crucial for later success at reading in English. Phonemic awareness, the ability to reflect on and manipulate individual sound units of speech, has a particular predictive power in reading development by serving as a foundation for the acquisition of the alphabetic principle (Ehri and Nunes 2002). Morphological awareness, the ability to reflect on and manipulate morphemes (the smallest meaning-bearing units) and the morphemic structure of words, supports word reading by providing clues not only to meaning but also to pronunciation (Kotzer, Kirby and Heggie 2021). Together with these two metalinguistic skills, prosodic competence has also been incorporated specifically as part of linguistic knowledge that is essential for the word reading process within the reading systems framework revised by Wade-Woolley, Wood, Chan, and Weidman (2022). In the general sense, prosodic competence refers to the ability to perceive and manipulate the rhythm of spoken language at the word, phrase, and discourse levels. At the lexical level, it involves knowledge of which syllable within a word is sounded out with greater intensity, longer duration, and higher pitch, compared with the other syllable(s).
As a stress-timed language, English has a rhythm characterized by roughly the same amount of time elapses between stressed syllables and by the alternations between stressed and unstressed syllables. Stressed syllables, which contain a fully articulated vowel, demonstrate greater duration compared to unstressed ones, whose vowels are often reduced to a schwa /ə/. For instance, the first vowel in toDAY is reduced to /ə/, whereas the first vowel in TOtal is fully articulated as /əʊ/. Crucially, in English, prosodic features, as part of the phonological representation in the mental lexicon, support the storage and access of lexical entries (Ashby and Clifton 2005). English word meanings can be distinguished by prosodic cues. Different stress locations can signify different meanings in the same spelling (e.g., DESert vs. deSERT), and discriminate between words (e.g., DESert vs. desSERT). But there is no diacritic indicating the word stress location in English orthography. This poses a great challenge to EFL (English as a Foreign Language) learners, especially those whose native languages contain a very different prosodic system from that of English.
Chinese, in contrast to English, is fundamentally a syllable-timed language and lacks the system of contrastive syllable stress that distinguishes meaning. In Chinese, each character corresponds to a single syllable, and these syllables are pronounced with relatively equal duration. The phenomenon of vowel reduction found in English is largely absent in Chinese. Vowels typically retain their full quality within Chinese words or phrases (Duanmu 2007). In Chinese, the primary prosodic feature carrying lexical meaning at the syllable level is tone. Mandarin Chinese uses distinct pitch contours (tones) on syllables to differentiate meanings of otherwise identical phonetic sequences, for example, mā (妈, mum), má (麻, hemp), mǎ (马, horse), and mà (骂, scold). The fundamental reliance on tone rather than stress for lexical distinction presents a significant contrast to the English prosodic system.
Such starkly different prosodic systems between English and Chinese may hinder Chinese speakers from developing sufficient awareness and knowledge of English lexical prosody. English lexical stress perception, as the foundational prosodic skill that develops in speech perception at a very early stage of language acquisition, is determined by the suprasegmental cues of pitch, duration, and intensity (Mattys 2000). Chinese speakers, however, have been found to rely particularly on pitch cues in the perception of English word stress whereas the prosodic cues of intensity and duration may often be missed out (Juffs 1990; Yu and Andruski 2010). Moreover, in terms of the more demanding and complex prosodic task that draws on knowledge of the rules governing lexical stress patterns cued by morphological suffixes, Chinese speakers may not be as capable as native English speakers of drawing on this knowledge when reading derived words (Chung and Jarmulowicz 2017).
Despite the accumulating evidence concerning the close relationship of lexical prosodic competence to English word reading among L1 (first language) children and adults (see the review by Wade-Woolley, Wood, Chan and Weidman 2022), what we know about this relationship among English language learners is far from sufficient, particularly for adult learners. This study therefore intends to examine the extent to which lexical stress perception and derived word stress production influence word reading among Chinese adult EFL learners, with other crucial metalinguistic skills taken into account at the same time.
Literature review
Word stress perception and word reading
Word stress perception, as a receptive prosodic skill, involves identifying, discriminating, matching, or judging the difference between stimuli presented aurally (Wade-Woolley et al. 2022). In English, stress perception may precede segmental phonological awareness and make a direct contribution to the development of phonemic awareness (Wood 2006; Wood and Terrell 1998; Wood, Wade-Woolley and Holliman 2009). This may be due to the facilitative effect of stress cues on syllable segmentation (Cutler, Dahan and van Donselaar 1997). The phonemes in stressed syllables tend to be perceived more easily than those in unstressed syllables, and thus sensitivity to phonemes is developed (Goswami et al. 2002). The increasing phonemic awareness may in turn enhance the ability of phonological decoding and word identification.
There is abundant evidence that word stress perception makes a unique contribution to native English-speaking children’s word identification beyond phonemic awareness (Enderby, Carroll, Tarczynski-Bowles and Breadmore 2021; Holliman, Wood and Sheehy 2010; Wade-Woolley 2016; Whalley and Hansen 2006), or beyond both phonemic awareness and morphological awareness (Chan, Wade-Woolley, Heggie and Kirby 2020; Holliman, Mundy, Wade-Woolley, Wood and Bird 2017). Some studies, however, suggest that word stress perception may support word identification at the beginning of learning to read in English, whereas its effect on word reading may ebb away in the later primary school years (Deacon, Holliman, Dobson and Harrison 2018; Lin, Wang, Newman and Li 2018). Based on the findings with L1 adults, who significantly outperformed 6- and 8-year-olds in stress sensitivity, Lin et al. (2018) further suggested that after primary education, with more reading experience, children’s lexical stress sensitivity might develop at a faster rate. Frequent encounters with multisyllabic words can activate the prosodic properties in the mental lexicon and thus enhance the effect of stress sensitivity on word reading. This was corroborated by Chan and Wade-Woolley’s (2018) findings that L1 adults’ prosodic competence, as a composite of word stress perception and word stress movement, made a unique contribution to word reading as a composite of word identification and nonword decoding.
The predictive power of word stress perception in nonword decoding beyond phonemic awareness has been reported in the entire group of L1 adults with and without dyslexia (Mundy and Carroll 2012). Nonetheless, this effect seems to be missing in children. L1 children’s stress perception failed to be a significant predictor of nonword reading beyond phonemic awareness in 4th and 5th graders (Wade-Woolley 2016), or beyond rhyme awareness in 4th graders (Whalley and Hansen 2006). Likewise, the Mandarin-speaking EFL peers’ word stress perception could not explain unique variance in nonword reading, whether after controlling for rhyme awareness or not (Chung, Jarmulowicz and Bidelman 2017). The role of word stress perception appears to be less prominent than segmental phonological awareness in decoding nonwords for L1 and EFL children in the upper grades of primary school. With the accumulation of reading experience, however, sensitivity to lexical stress may become increasingly prominent in decoding unfamiliar multisyllabic words.
Stress production in nonneutral derivatives and word reading
Most studies on English lexical prosodic competence have focused on receptive prosodic skills (see above), with considerably less attention to expressive prosodic skills, which, as Wade-Woolley et al. (2022) stated, involve producing or imitating the prosodic features of words (Chan and Wade-Woolley 2018; Chung et al. 2017; Chung, Jarmulowicz and Bidelman 2021). Recently, there has been an increasing interest in the relation of stress production in nonneutral derivatives to English word identification and nonword decoding (Chung and Jarmulowicz 2017; Clin, Wade-Woolley and Heggie 2009; Jarmulowicz, Hay, Taran and Ethington 2008; Jarmulowicz, Taran and Hay 2007; Park, Kuo, Dixon and Kim 2022; Wade-Woolley and Heggie 2015).
In English, the placement of word stress can be cued by suffixes (Wood et al. 2009). Neutral suffixes (e.g., -ment, -ness, -ful) do not affect word stress, keeping the primary stress of the derived word at the same position as in the base word. In contrast, nonneutral suffixes (e.g., -ity, -ic, -tion) shift the primary stress from its original location in the base word to the syllable immediately before the suffix in the derived word. The reassignment of stress then may lead to a change in vowel quality. For example, complex /ˈkɒmpleks/ to complexity /kəmˈpleksəti/, economy /ɪˈkɒnəmi/ to economic /ˌiːkəˈnɒmɪk/. The stress production task involving nonneutral suffixes has been found to be more challenging than that involving neutral suffixes, because the former necessitates stress shifts and often vowel quality changes, while the latter results in no such changes (Clin et al. 2009). Clin and colleagues further revealed that the competence for the nonneutral derivatives significantly contributed to reading ability beyond phonemic awareness and stress perception in 3rd, 5th, and 7th graders, whereas the competence for the neutral derivatives did not. Hence, the competence for stress production in English words with nonneutral suffixes, though more complex and demanding, appears to be more crucial for word reading than that with neutral suffixes. According to Jarmulowicz et al. (2007), the essential knowledge of English lexical stress cued by nonneutral suffixes may develop in the later primary school years. English-speaking children’s competence in stress production on nonneutral suffixes has been found to undergo improvements between the ages of 7 and 9 (Jarmulowicz 2002), unlike their word stress perception, which shows no sign of significant improvements between the ages of 7 and 10 (Lin et al. 2018).
Only a handful of studies have explored the relationship between stress production in English derived words with nonneutral suffixes and word reading, among monolingual children (Jarmulowicz et al. 2008; Jarmulowicz et al. 2007), bilingual children (Park et al. 2022), monolingual adults (Wade-Woolley and Heggie 2015), and Mandarin-speaking adults (Chung and Jarmulowicz 2017). Among L1 children in grade 3, stress production in derivatives with nonneutral suffixes not only explained unique variance in nonword decoding beyond phonemic awareness and morphological awareness but also had a direct effect on it (Jarmulowicz et al. 2008; Jarmulowicz et al. 2007). In L1 adults, stress production performance in nonneutral derivatives was found to be significantly correlated with their word and nonword reading abilities (Wade-Woolley and Heggie 2015). This strong correlation, however, was not found in Korean-English bilingual children (Park et al. 2022). The researchers ascribed the lack of significant correlation to the task design. The task involved circling the syllable that carries the primary stress in a pre-syllabified written word, and might not tap into the particular aspect of prosodic competence that could be correlated with word reading abilities. Another possibility is that the bilingual children did not develop sufficient knowledge of prosody in derivation at grades 3 to 5. Although a significant correlation was reported among adult EFL learners in Chung and Jarmulowicz’s (2017) study, performance on stress production in nonneutral derivation did not make a significant contribution to either word identification or nonword decoding beyond oral vocabulary and working memory, which, as the researchers indicated, might be due to the great difficulty adult learners encountered with stress production in the nonneutral condition (mean proportion of correct responses = 0.55). The adult EFL learners’ explicit knowledge of lexical prosody conditioned by nonneutral suffixes appeared to be insufficient and therefore not yet very conducive to word reading. Since Chung and Jarmulowicz did not include other metalinguistic skills in their study, it is not known whether the effect of stress production in nonneutral derivation on word reading could be mediated through phonemic awareness and morphological awareness.
Theoretical model of prosody-reading relation
Wood et al. (2009) proposed a theoretical model mapping and explaining the contribution of prosodic competence to word reading in typically developing readers (see Fig. 1). The pathway model underpinning the current study starts from the periodicity bias, sensitizing infants to the prosodic or rhythmic features of their native language (Cutler, Dahan and van Donselaar 1997). This prosodic sensitivity directly facilitates spoken word recognition processes, which in turn support vocabulary development, and vocabulary knowledge indirectly contributes to reading development through the mediation of phonemic awareness and rhyme awareness (Ashby and Clifton 2005; Wood and Terrell 1998). Further, the model hypothesizes indirect pathways from prosodic sensitivity to reading, mediated by phonemic awareness (via phonemic identification), rhyme awareness, and morphological awareness. In line with Goswami et al. (2002), prosodic sensitivity may help children identify phonemes within stressed syllables, thereby supporting the development of phonemic awareness. Sensitivity to stress may also draw children’s attention to vowel pitch and intensity, thus aiding recognition of the onset-rime boundary or the rime of words. In addition, prosodic sensitivity may support the development of morphological awareness by enabling children to detect morpheme boundaries, and inflectional and derivational morphemes within morphologically complex words (Clin et al. 2009). These three mediating factors, phonemic awareness, rhyme awareness, and morphological awareness, in turn, promote reading development (Ehri and Nunes 2002; Kotzer, Kirby and Heggie 2021).
[See PDF for image]
Fig. 1
The pathway model of prosody-reading relationship (Wood et al. 2009).
The lack of direct effect of prosodic competence on reading development in the Wood et al. (2009) pathway model was corroborated by the findings of Holliman et al. (2014) and Kim and Petscher (2016) in L1 early readers. In the study by Holliman et al. (2014) involving beginning readers aged 5-7 years, a holistic measure of prosodic competence was adopted, encompassing the prosodic features of stress, timing, and intonation at the levels of words, phrases, and sentences. These early readers’ prosodic competence had indirect effects on phonemic awareness and morphological awareness, and only morphological awareness directly influenced word identification. Similarly, among 6-8-year-olds in the study by Kim and Petscher (2016), the relationship between word stress perception and word identification was fully mediated by phonemic awareness and morphological awareness. Neither did Chung and Jarmulowicz (2017) find the unique contribution of adult EFL learners’ stress production in nonneutral derivatives to either word identification or nonword decoding after controlling for vocabulary and working memory.
However, the unique contribution of stress production in nonneutral derivatives to nonword decoding was reported by Jarmulowicz et al. (2007) in L1 children aged 7 years 11 months to 10 years after controlling for phonemic awareness and morphological awareness. Moreover, in the mediation analysis by Chan et al. (2020), a direct effect of stress perception was found on the word reading composite (comprising word identification and nonword decoding) among older L1 children aged 9 to 11, in addition to its indirect effects on word reading via phonemic awareness and morphological awareness. Among L1 adults, regression analyses also indicated the unique contribution of stress perception and production to the word reading composite, which comprised word identification and nonword decoding, beyond IQ, vocabulary and phonemic awareness (Chan and Wade-Woolley 2018), and to nonword decoding beyond IQ, vocabulary and phonemic awareness (Mundy and Carroll 2012). It appears that, with the accumulation of reading experience and more frequent encounters with multisyllabic words, the contribution of prosodic competence to word reading becomes more direct and stronger.
Given the mixed findings, more studies focusing on learners from older age groups and those with English as their second or foreign language are warranted to examine the effects of their prosodic competence on word reading. Further, considering the close links of prosodic competence to segmental phonological skills and morphological awareness, this study incorporates lexical prosodic competence, phonemic awareness, and morphological awareness in a model of reading, aiming to explore the prosody-reading relationships proposed in the Wood et al. (2009) model.
Research questions
Considering the crucial role of prosodic competence in word reading and the limited number of extant relevant studies among adult learners, particularly those whose native language adopts a strikingly different prosodic system than English, more research is needed to obtain a clearer picture. This study intends to investigate Chinese adult EFL learners’ lexical prosodic competence and to explore the extent to which their word stress perception and stress production in nonneutral derivatives contribute to English word reading. Given the ample evidence that word identification and nonword decoding often entail phonemic awareness and morphological awareness (Castles, Rastle and Nation 2018; Levesque, Breadmore and Deacon 2021), these two types of metalinguistic skills are also taken into account in the study. The current study intends to address the following research questions.
RQ1: Do adult EFL learners perceive lexical stress and produce stress in nonneutral derivatives above chance level? Based on previous studies (Chan et al. 2020; Chung and Jarmulowicz 2017; Wade-Woolley 2016; Wade-Woolley and Heggie 2015), it is expected that adult EFL learners show sensitivity to stress location in multisyllabic words and implicit knowledge of stress shift in nonneutral derivatives. It is hypothesized that if learners possess explicit knowledge of stress location conditioned by suffixes, there will be no difference between their performance on stress production in high-frequency, low-frequency, and nonsense derivatives.
RQ2: What contributions do foundational prosodic competence (i.e., word stress perception) and more demanding prosodic skill (i.e., stress production in nonneutral derivatives) make to word identification in Chinese adult EFL learners, after controlling for phonemic awareness and morphological awareness? Based on the results reported by Chan and Wade-Woolley (2018) among L1 adults, this study hypothesizes that learners’ word stress perception will make unique contribution to word identification. Based on the findings with L1 adults (Wade-Woolley and Heggie 2015) and Mandarin-speaking adult EFL learners (Chung and Jarmulowicz 2017), learners’ stress production in nonneutral derivatives will not be a significant predictor of word identification, despite their significant correlation. In addition, as reported by Chan et al. (2020) in a study with older elementary children, word stress perception is expected to have a direct effect on word identification, and also indirect effects through phonemic awareness and morphological awareness. Besides, it is hypothesized that adult EFL learners’ stress production in nonneutral derivatives will not directly affect word identification but will influence it indirectly through phonemic awareness and morphological awareness.
RQ3: What contributions do word stress perception and stress production in nonneutral derivatives make to nonword decoding, after controlling for phonemic awareness and morphological awareness? Drawing on Mundy and Carroll’s (2012) findings with L1 adults, it is hypothesized that adult EFL learners’ word stress perception will be a unique predictor of nonword decoding. Based on the study by Chung and Jarmulowicz (2017), Mandarin-speaking adult EFL learners’ stress production in nonneutral derivatives is expected not to make a unique contribution to nonword decoding, although the two will be significantly correlated. In addition, considering the primary role of phonemic awareness in decoding (Ehri and Nunes 2002), performance on both tasks of prosodic competence is expected to have an indirect effect on nonword decoding through phonemic awareness.
Methods
Participants
Ninety Mandarin-speaking college students from a university in Shanghai, China, were recruited in the study. They consisted of 77 females and 13 males, with an average age of 18.73 (SD = 0.64). They were from four parallel classes, all majoring in English. The data were collected at the beginning of the first semester of their freshman year. The participants all started formal English language learning during primary school in China, with 37 from first grade and 53 from third grade. None of them had experience of studying abroad in English-speaking countries. They could be considered as independent users at the B1-intermediate level within the Common European Framework of Reference for Languages (CEFR) (Council of Europe 2001) because senior high school graduates are required to reach level 4 within the framework of CSE (China’s Standards of English Language Ability, Ministry of Education & National Language Commission 2018), which can be aligned with the B1-intermediate level (Dunlea, Spiby, Wu, Zhang and Cheng 2019).
Measures
Nonverbal reasoning
Raven’s Standard Progressive Matrices (Raven, Raven and Court 2000) was used. Participants were asked to choose the missing piece to fill in an incomplete pattern. The maximum possible score was 60. The reported reliability coefficient of Cronbach’s alpha is 0.88 and for the current sample, it was 0.76.
Vocabulary
The Peabody Picture Vocabulary Test, Fourth Edition (PPVT-IV; Dunn and Dunn 2007) was used here as a measure of English receptive vocabulary. Four pictures for one stimuli were presented on the computer screen at a time. Participants were aurally presented with a word and then asked to choose a picture out of the four options to match the spoken word. All test items were completed. The maximum possible score was 228. The reported reliability coefficient of Cronbach’s alpha ranges from 0.92 to 0.94 in a sample older than 18 years of age and in the present sample, it was 0.96.
Phonemic awareness
The task of phoneme counting was used to measure phonemic awareness. The stimuli were prerecorded by a native English speaker. Participants listened to real words (e.g., many) and counted the phonemes in each word. The number of phonemes ranged from 3 to 7. Seven stimuli were equally assigned to each of the 5 categories of words that differ in phoneme number. There were 35 test items in total. Three practice items were given but no feedback was provided. A Cronbach’s alpha of 0.83 was calculated for this task.
Morphological awareness
Carlisle’s (2000) measure of derivational morphological awareness was adopted in the current study. This type of task was regarded as “a valid examination of college students’ abilities given the prevalence and nature of academic language in college text” (Wilson-Fowler and Apel 2015, p. 410). Participants were orally given a base word (e.g., revise) and a sentence with a word missing (This paper is his second ____________.). Then they were asked to provide a derivational form of the base word to complete the sentence (e.g., revision). There were 28 experimental trials in total, and three practice trials. A Cronbach’s alpha of 0.75 was calculated for this task.
Word stress perception
The word stress perception task was adapted from the stress identification task developed by Heggie and Wade-Woolley (2018), who administered the task among L1 adults. The test items are all either four- or five-syllable words. Besides the 7 words with primary stress on the first syllable (e.g., cannibalism, alabaster), 7 on the second syllable (e.g., kaleidoscope, mahogany), and 5 on the third syllable from the original task, another 2 words with primary stress on the third syllable (e.g., incidental, recapitulate) were added to the list. These words were confirmed as unfamiliar words to the participants by their reading teacher. The frequency of these word stimuli (Log10CD value, M = 1.37, SD = 0.54) was also obtained from SUBTLEX-US (Brysbaert and New 2009). There were 21 test items in total, and 3 practice items. Participants heard prerecorded stimuli and repeated the syllable that carried primary stress. A Cronbach’s alpha of 0.76 was calculated for this task.
Stress production in nonneutral derivatives
This task was adapted from the derived word stress production task developed by Jarmulowicz and Hay (2009). Three sets of derivatives involving two stress-shifting suffixes (-ic and -ity) were included in the list of stimuli for the current study. Each set consists of 8 stimuli. There were 24 test items in total. One set of stimuli comprised high-frequency derivatives (e.g., artistic) and another comprised low-frequency derivatives (e.g., systemic). The other set of stimuli were nonsense derivatives (e.g., umdelic). Lexical frequency was obtained from SUBTLEX-US (Brysbaert and New 2009) and Lg10CD value was used. The ANOVA test found significant difference between the means of high-frequency derived words (M = 2.41, SD = 0.38) and low-frequency derived words (M = 1.3, SD = 0.31), F (1, 14) = 41.55, p = 0.000. No statistically significant difference was found between base words for high-frequency derivatives (M = 2.69, SD = 0.52), F (1, 14) = 1.2, p = 0.29, and base words for low-frequency derivatives (M = 2.32, SD = 0.80). In addition, the number of syllables in the derived forms was the same in the three sets. Besides the 24 test items, 9 stimuli with neutral suffixes (e.g., embarrassment, gentleness) were included in the list as filler items.
The prerecorded prompts were presented aurally to individual participants, e.g., “Put -ic on the end of magnet”. Then participants combined the suffix with the base word and read aloud the derived form. Correct placement of stress in the derivative was given one point. The maximum possible score was 24. A Cronbach’s alpha of 0.87 was calculated for this task.
Word identification
Word identification was measured by the word identification subtest of the Woodcock Reading Mastery Test-Revised (Woodcock 1998). Each participant was asked to read aloud a list of 106 real words, which were presented one at a time. This task was administered according to the procedures outlined in the manual. One point was given for each word that had been read accurately. The reported split-half reliability score is 0.97. In this sample, its cronbach’s alpha was 0.87.
Nonword decoding
The silent phonological choice task (Olson, Forsberg, Wise and Rack 1994) was adopted in the current study to assess phonological decoding. Participants were presented with a trial of three nonwords at a time on the computer screen (e.g., grait, throut, bloot), and chose the one that sounded most like a real word (e.g., grait). As nonwords were used, participants must apply their knowledge of spelling-sound correspondences to decode the visually presented nonwords. The crucial step is then to judge which of the resulting phonological representations (i.e., how the nonword sounds when decoded) corresponds to the pronunciation of a real English word (e.g., choosing grait because its decoded form /greɪt/ sounds like great). According to Olson et al. (1994), the results of the silent phonological decoding task have shown a pattern similar to those of oral nonword reading. This task was also adopted by Yeong, Fletcher, and Bayliss (2017) to assess the phonological decoding skills of English monolingual and Chinese–English bilingual adults. Kotzer, Kirby, and Heggie (2021) also used this type of task to measure the phonological decoding ability of English monolingual adults.
The stimuli for the current study were taken from Olson et al. (1994), which included 5 practice trials and 60 experimental trials, with each trial consisting of 3 nonwords. Another 4 trials of multisyllabic nonwords (e.g., A. lettle, B. tapple, C. moddle) were added to the list of experimental trials. In total, the nonword decoding task consisted of 53 monosyllabic trials and 11 multisyllabic trials, ranging from one to three syllables. The maximum possible score was 64. A Cronbach’s alpha of 0.85 was calculated for this task.
Procedures
Nonverbal reasoning and nonword decoding were administered collectively in a group of 15 participants, who were presented with visual stimuli on computer screens and then were asked to write answers on the answer sheet. The remaining 6 tasks were administered individually, with participants’ oral responses recorded with a digital voice recorder and on the answer sheets by the researcher for later analyses. To ensure the scoring of stress production in nonneutral derivatives was reliable, 18 (20%) students’ voice recording files were sent to the second rater for scoring. The percentage of our agreement on the accuracy of stress placement in nonneutral derivatives was 98.5%.
Results
Prior to reporting the results related to the research questions, preliminary analyses were conducted to calculate the minimum, maximum, mean, standard deviation, skewness, and kurtosis values of each variable. Descriptive statistics for all measures are displayed in Table 1. The absolute values for skewness (statistics < 2) and kurtosis (statistics < 7) indicated that all measurements were acceptable pertaining to symmetricity and normality (Kline 2023). The matrix of Pearson correlations between variables is shown in Table 2. All the observed variables were significantly correlated (p < 0.05), except nonverbal reasoning, which was significantly associated only with vocabulary, morphological awareness and word reading.
Table 1. Descriptive statistics for all variables.
Min. | Max. | Mean | SD | Skewness | Kurtosis | |
|---|---|---|---|---|---|---|
Nonverbal reasoning (60) | 36 | 60 | 52.73 | 4.29 | -1.051 | 2.372 |
Vocabulary (228) | 62 | 215 | 127.88 | 32.83 | 0.052 | -0.452 |
Phonemic awareness (35) | 7 | 34 | 25.40 | 5.56 | -0.771 | 0.532 |
Morphological awareness (28) | 7 | 26 | 17.51 | 4.32 | -0.142 | -0.714 |
Word stress perception (21) | 7 | 21 | 14.38 | 3.74 | -0.105 | -0.952 |
Stress production in nonneutral derivatives (24) | 1 | 23 | 14.54 | 5.58 | -0.766 | -0.208 |
Word identification (106) | 46 | 87 | 69.31 | 10.26 | -0.493 | -0.485 |
Nonword decoding (64) | 13 | 58 | 41.76 | 8.74 | -0.494 | 0.300 |
N = 90, Standard error of skewness = 0.254, Standard error of kurtosis = 0.503, and SD is the standard deviation.
Table 2. Correlations between all variables.
1 | 2 | 3 | 4 | 5 | 6 | 7 | |
|---|---|---|---|---|---|---|---|
1. Nonverbal reasoning | – | ||||||
2. Vocabulary | 0.211* | – | |||||
3. Phonemic awareness | 0.138 | 0.207* | – | ||||
4. Morphological awareness | 0.267* | 0.526*** | 0.427*** | – | |||
5. Word stress perception | 0.177 | 0.208* | 0.359** | 0.253* | – | ||
6. Stress production in nonneutral derivatives | 0.024 | 0.308** | 0.260* | 0.245* | 0.436*** | – | |
7. Word identification | 0.353** | 0.448*** | 0.488*** | 0.629*** | 0.454*** | 0.400*** | – |
8. Nonword decoding | 0.131 | 0.298** | 0.507*** | 0.426*** | 0.377*** | 0.357*** | 0.587*** |
Correlation is significant at the 0.05 level (2-tailed). *p < 0.05; **p < 0.01; ***p < 0.001.
To answer the first research question, adult EFL learners’ scores on both lexical prosodic tasks were compared against chance level. Scores on word stress perception and stress production in nonneutral derivatives were significantly greater than chance, t(89) = 23.665 and 14.111, p < 0.001, respectively. There was no significant difference in proportion of correct responses between stress production in nonneutral derivatives (Median = 66.667) and word stress perception (Median = 71.429), Mann-Whitney U = 3405, z = −1.848, p = 0.065. In the task of stress production in nonneutral derivatives, significant difference in score was only found between high-frequency stimuli (Median = 6) and low-frequency stimuli (Median = 5), Mann-Whitney U = 3341, z = −2.047, p = 0.041. No significant difference was found between high-frequency stimuli and nonsense stimuli (Median = 5), Mann-Whitney U = 3710.5, z = −0.981, p = 0.326, or between low-frequency stimuli and nonsense stimuli, Mann-Whitney U = 3698.5, z = −1.015, p = 0.310.
To examine the contribution of word stress perception and stress production in nonneutral derivatives to word identification and nonword decoding, hierarchical multiple regression analysis was used, with other cognitive predictors as control variables. Prior to conducting the regression analyses, the assumptions of independence, normality, and homoscedasticity were examined and found to be met. Additionally, no multicollinearity problems were found (Kline 2023).
Two hierarchical regression analyses were conducted, with word identification and nonword decoding as the outcome variables. Nonverbal reasoning and vocabulary were entered into the first step as controls, followed by phonemic awareness and morphological awareness in the second and third steps, respectively. Word stress perception and stress production in nonneutral derivatives were entered in the last two steps and then the order was reversed to examine their independent contribution to the outcome variables (see Table 3).
Table 3. Hierarchical regression analyses on word identification and nonword decoding.
Step | Variable | Word identification | Nonword decoding | ||||
|---|---|---|---|---|---|---|---|
Entry β | Final β | ∆R2 | Entry β | Final β | ∆R2 | ||
1 | Nonverbal reasoning | 0.270** (p = 0.005) | 0.173* (p = 0.026) | 0.271*** (p < 0.001) | 0.071 (p = 0.497) | 0.004 (p = 0.968) | 0.093* (p = 0.014) |
Vocabulary | 0.391*** (p < 0.001) | 0.094 (p = 0.288) | 0.283** (p = 0.008) | 0.059 (p = 0.577) | |||
2 | Phonemic awareness | 0.390*** (p < 0.001) | 0.181* (p = 0.033) | 0.144*** (p < 0.001) | 0.463*** (p < 0.001) | 0.329** (p = 0.002) | 0.203*** (p < 0.001) |
3 | Morphological awareness | 0.389*** (p < 0.001) | 0.375*** (p < 0.001) | 0.091*** (p < 0.001) | 0.196 (p = 0.093) | 0.184 (p = 0.106) | 0.023 (p = 0.093) |
4 | Lexical stress perception | 0.234** (p = 0.006) | 0.178* (p = 0.038) | 0.046** (p = 0.004) | 0.192* (p = 0.049) | 0.136 (p = 0.187) | 0.031* (p = 0.049) |
5 | Stress production in nonneutral derivatives | 0.150 (p = 0.076) | 0.150 (p = 0.076) | 0.017 (p = 0.076) | 0.149 (p = 0.145) | 0.149 (p = 0.145) | 0.016 (p = 0.145) |
5 | Stress production in nonneutral derivatives | 0.215** (p = 0.008) | 0.150 (p = 0.076) | 0.040** (p = 0.008) | 0.198* (p = 0.039) | 0.149 (p = 0.145) | 0.034* (p = 0.039) |
4 | Lexical stress perception | 0.178* (p = 0.038) | 0.178* (p = 0.038) | 0.023* (p = 0.038) | 0.136 (p = 0.187) | 0.136 (p = 0.187) | 0.013 (p = 0.187) |
Total R2 | 0.568*** | 0.367*** | |||||
*p < 0.05, **p < 0.01, ***p < 0.001.
With respect to word identification, the total variance explained by all predictors was 56.8% (F (6, 83) = 18.203, p < 0.001). The two control variables accounted for 27% of variance, followed by phonemic awareness explaining 14.4% in the second step and morphological awareness 9.1% in the third step. Word stress perception added another 4.6% to the total variance in the fourth step, while stress production in nonneutral derivatives in the last step made no significant unique contribution to word identification. When the last two steps were reversed, stress production in nonneutral derivatives predicted significant additional variance in word identification after controlling for nonverbal reasoning, vocabulary, phonemic awareness and morphological awareness, and word stress perception remained a significant independent predictor of word identification beyond all the other variables.
In relation to nonword decoding, the total variance explained by all predictors was 36.7% (F (6, 83) = 8.011, p < 0.001). Nonverbal reasoning and vocabulary explained 9.3% of variance in the first step. In the second step, phonemic awareness contributed a further 20.3% of variance while morphological awareness displayed no significant unique contribution to nonword decoding. After accounting for these control variables, word stress perception and stress production in nonneutral derivatives each contributed significant additional variance in nonword decoding (3.1% and 3.4%, respectively). However, the unique predictive power of word stress perception disappeared when stress production in nonneutral derivatives was entered in the next step, and vice versa. While word stress perception demonstrated a statistically significant relationship, the p-value (0.049) was very close to the 0.05 significance threshold, warranting a degree of caution in interpreting the results.
The hierarchical multiple regression analyses provided evidence regarding the predictive role of lexical prosodic competence in word reading abilities beyond other control variables. To further explore whether and how the association between prosodic skills and reading outcomes was mediated by reading-related skills and knowledge—including phonemic awareness, morphological awareness and oral vocabulary—as proposed by Wood et al. (2009) in their theoretical model of prosodic competence, four mediation analyses were conducted. These analyses examined the direct and indirect effects of word stress perception and stress production in nonneutral derivatives on word identification and nonword decoding, respectively. The first mediation model included word stress perception as the predictor and word identification as the outcome variable. In the second mediation model, the same predictor variable (word stress perception) was used, with nonword decoding as the outcome variable. Then word identification and nonword decoding were included as the outcome variable in the third and fourth mediation models respectively, with stress production in nonneutral derivatives as the predictor variable.
Before conducting the mediation analyses, four conditions of mediation were tested (Baron and Kenny 1986; Field 2013). Firstly, the predictor variable significantly predicts the outcome variable. Secondly, the predictor variable significantly predicts the mediator. Thirdly, the mediators significantly predict the outcome variable. Finally, the effect of the predictor variable predicted on the outcome variable becomes less strong, when the mediators are controlled. All the four conditions of meditation were satisfied for the four mediation analyses, which were conducted using model 4 of the PROCESS macro for SPSS (version 4.2) (Hayes 2022). All the variables were transformed to z-scores for the following analyses. There was no significant difference between individual effects and therefore their pairwise contrasts were not presented in the following tables.
Table 4 presents the total effect, the direct effect, the total indirect effect, and individual indirect effects of lexical stress perception on word identification and nonword decoding. Indirect effects with 95% bias corrected and accelerated confidence intervals that do not contain zero are significant at the 0.05 level. The CI was computed with 5000 bootstrap samples.
Table 4. Total, direct, and indirect effect of lexical stress perception on word identification and nonword decoding.
Effects | Word identification | Nonword decoding | |||
|---|---|---|---|---|---|
Point estimate | 95% BCa CI | Point estimate | 95% BCa CI | ||
Total effect | 0.454 | 0.265–0.642 | 0.378 | 0.181–0.574 | |
Direct effect | 0.251 | 0.091–0.412 | 0.189 | 0.001–0.378 | |
Total indirect | 0.202 | 0.079–0.325 | 0.188 | 0.088–0.308 | |
Indirect effects | |||||
Phonemic awareness | 0.069 | 0.009–0.147 | 0.123 | 0.042–0.222 | |
Morphological awareness | 0.104 | 0.019–0.201 | 0.047 | −0.008 to 0.137 | |
Vocabulary | 0.029 | −0.010 to 0.088 | 0.019 | −0.031 to 0.079 | |
BCa CI bias corrected and accelerated confidence intervals; Indirect effects with 95% BCa CI containing zero are not statistically significant (p > 0.05).
Figure 2 presents the mediation model with regression coefficients for the relation between lexical stress perception and word identification. The total effect of lexical stress perception on word identification was 0.454 (t(88) = 4.773, p < 0.001). There was a significant direct effect of stress perception on word identification (β = 0.251, t(88) = 3.11, p = 0.003). The total indirect effect accounted for 44.5% of the total effect (β = 0.202, 95% CI = 0.079–0.325). Only phonemic awareness (β = 0.069, 95% CI = 0.009 to 0.147) and morphological awareness (β = 0.104, 95% CI = 0.019 to 0.201) significantly mediated the relation between word stress perception and word identification.
[See PDF for image]
Fig. 2
Effects of lexical stress perception on mediators and direct effects on word identification. *p < 0.05; **p < 0.01; ***p < 0.001.
Figure 3 displays the mediation model linking lexical stress perception and nonword decoding. The total effect of lexical stress perception on nonword decoding was 0.378 (t(88) = 3.824, p < 0.001). There was a significant direct effect of lexical stress perception on nonword decoding (β = 0.189, t(88) = 1.996, p = 0.049). The total indirect effect was 0.188, 95% CI = 0.088 to 0.308. Only phonemic awareness (β = 0.123, 95% CI = 0.042 to 0.222) significantly mediated the relationship between word stress perception and nonword decoding.
[See PDF for image]
Fig. 3
Effects of lexical stress perception on mediators and direct effects on nonword decoding. *p < 0.05; **p < 0.01; ***p < 0.001.
Table 5 presents the total, direct and indirect effects of stress production in nonneutral derivatives on the two word-level reading measures. Figure 4 presents the mediation model for the effects of stress production in nonneutral derivatives on word identification, which is similar to the model for the relationship between lexical stress perception and word identification. The total effect of stress production in nonneutral derivatives on word identification was 0.400 (t(88) = 4.095, p < 0.001). Its direct effect on word identification was significant (β = 0.201, t(88) = 2.456, p = 0.016), though relatively smaller than the direct effect of lexical stress perception on word identification. The total indirect effects of stress production in nonneutral derivatives on word identification accounted for 50% of the total effect (β = 0.200, 95% CI = 0.073 to 0.327). Phonemic awareness (β = 0.060, 95% CI = 0.009 to 0.132) and morphological awareness (β = 0.103, 95% CI = 0.019 to 0.204) significantly mediated the relation between stress production in nonneutral derivatives and word identification.
Table 5. Total, direct, and indirect effect of stress production in nonneutral derivatives on word identification and nonword decoding.
Effects | Word identification | Nonword decoding | ||
|---|---|---|---|---|
Point estimate | 95% BCa CI | Point estimate | 95% BCa CI | |
Total effect | 0.400 | 0.206–0.594 | 0.357 | 0.159–0.555 |
Direct effect | 0.201 | 0.038–0.363 | 0.197 | 0.011–0.384 |
Total indirect | 0.200 | 0.073–0.327 | 0.160 | 0.061–0.276 |
Indirect effects | ||||
Phonemic awareness | 0.060 | 0.009–0.132 | 0.094 | 0.021–0.187 |
Morphological awareness | 0.103 | 0.019–0.204 | 0.047 | −0.013 to 0.131 |
Vocabulary | 0.036 | −0.023 to 0.108 | 0.019 | −0.050 to 0.104 |
BCa CI bias corrected and accelerated confidence intervals; Indirect effects with 95% BCa CI containing zero are not statistically significant (p > 0.05).
[See PDF for image]
Fig. 4
Effects of stress production in nonneutral derivatives on mediators and direct effects on word identification. *p < 0.05; **p < 0.01; ***p < 0.001.
Figure 5 shows the mediation model for the effects of stress production in nonneutral derivatives on nonword decoding. The total effect of stress production in nonneutral derivatives on nonword decoding was 0.357 (t(88) = 3.586, p < 0.001). It made a significant direct contribution to nonword decoding (β = 0.197, t(88) = 2.10, p = 0.04). Its total indirect effect on nonword decoding was 0.160, 95% CI = 0.061 to 0.276. Phonemic awareness was the only significant mediator (β = 0.094, 95% CI = 0.021 to 0.187).
[See PDF for image]
Fig. 5
Effects of stress production in nonneutral derivatives on mediators and direct effects on nonword decoding. *p < 0.05; **p < 0.01; ***p < 0.001.
Discussion
The current study intended to investigate Chinese adult EFL learners’ lexical prosodic competence and its role in the models of word-level reading. Adult learners performed above chance levels on both the lexical stress perception task and the stress production task in nonneutral derivatives. Both types of lexical prosodic competence were unique predictors of word identification above and beyond verbal and nonverbal abilities, phonemic awareness, and morphological awareness. Lexical stress perception continued to display a unique contribution to word identification after stress production in nonneutral derivatives was additionally accounted for. Both lexical stress perception and stress production in nonneutral derivatives individually explained unique variance in nonword decoding after controlling for nonverbal reasoning, vocabulary, phonemic awareness, and morphological awareness. However, neither accounted for significant additional unique variance in nonword decoding beyond each other, suggesting a substantial overlap in their predictive contributions to decoding. Further, the relationships of both lexical prosodic competence with word identification were partially mediated by phonemic awareness and morphological awareness, whereas their relationships with nonword decoding were partially mediated only by phonemic awareness.
Adult EFL learners’ lexical prosodic competence
Consistent with previous studies (Chung and Jarmulowicz 2017; Wade-Woolley and Heggie 2015), the results of the current study indicate that Chinese adult EFL learners, with accumulating English reading experience, did exhibit better than chance levels of awareness of English lexical stress, despite the sharp differences between the prosodic systems of English and Mandarin.
In the task of assigning primary stress to nonneutral derivatives, the lack of a significant difference in performance between high-frequency and nonsense stimuli, or between low-frequency and nonsense stimuli, suggests that adult EFL learners may draw on their implicit awareness of stress patterns cued by nonneutral suffixes, when assigning primary stress to nonsense derivatives. In this case, learners may fail to find the corresponding lexical representation of either the base word or its derivative in memory, and hence may make an analogy to the stress shifting pattern in familiar and similar nonneutral derivatives, relying on their implicit awareness of stress-shifting suffixes. However, there was a statistically significant difference in performance between high-frequency and low-frequency stimuli, which may be attributable to the word familiarity effect. In assigning stress to high-frequency nonneutral derivatives, learners may not deploy the prosodic rules conditioned by suffixes, but mainly resort to the prosodic features stored in the mental lexicon (Jarmulowicz 2002). Likewise, when producing primary stress in unfamiliar low-frequency nonneutral derivatives with familiar base words, learners may rely on the fuzzy memory of the lexicon or simply copy the stress location of the familiar high-frequency base words. Taken together, adult EFL learners may access the stored prosody for high-frequency nonneutral derivatives, rely on fuzzy lexical memories for low-frequency ones or simply copy the prosodic features of their familiar base words, and may resort to the underlying awareness of suffixes that trigger stress shifts. This pattern shows that the adult EFL learners in the current study seemed to demonstrate implicit awareness of the stress shift cued by nonneutral suffixes and their explicit knowledge of it might still be developing.
Congruent with previous studies (Chan et al. 2020; Clin et al. 2009; Deacon et al. 2018; Holliman et al. 2014; Jarmulowicz et al. 2008; Jarmulowicz et al. 2007), both types of lexical prosodic competence were significantly correlated with phonemic awareness and morphological awareness in adult EFL learners. Perception of the accentuated syllable in a multisyllabic word may direct learners’ attention to the phonemes especially the full vowel within the syllable. Understanding lexical prosodic structure may enable the division between weak and strong syllables, which may stand for morphemes themselves such as un-, dis-, -ceive and -duct. Moreover, assigning stress to suffixational derivatives may bring the change of vowel quality and morphological suffixes to the focus of attention. The prosody-phoneme and prosody-morphology relationships may be reciprocal. In English, the schwa sound is often associated with weak syllables and full vowels often with accentuated syllables. Recognition of reduced and full vowels may lead to the understanding of lexical prosodic structures. Likewise, decoding English morphemes may involve stress information. For example, morphological suffixes such as -ese and -ette often carry stress themselves. A good grasp of the prosodic features of words can enhance the quality of lexical representations in the mental lexicon, and further facilitate the word reading process. The current study found a close relationship of word stress perception and stress production in nonneutral derivatives to word reading, echoing previous studies in L1 adults (Chan and Wade-Woolley 2018; Mundy and Carroll 2012; Wade-Woolley and Heggie 2015), and EFL adults (Chung and Jarmulowicz 2017).
Lexical prosodic competence and word identification
In line with what has been found in L1 third graders (Holliman et al. 2017) and fourth and fifth graders (Chan et al. 2020), adult EFL learners’ lexical stress perception has also been found to make a unique contribution to word reading beyond other metalinguistic skills. Similar findings have also been reported in L1 adult skilled readers (Chan and Wade-Woolley 2018), but it should be noted that the measurement on lexical prosodic competence in their study used the composite score of receptive and expressive prosodic skills. Hence it is not known whether the contribution of lexical prosodic competence was attributed mainly to lexical stress perception or stress manipulation. Although the predictive power of lexical stress perception in word identification was not found among Mandarin-speaking 4th graders (Chung et al. 2017), it is probable that young EFL learners’ prosodic receptive skills would develop over time with increased experience in reading multisyllabic words.
Adult EFL learners’ awareness of stress shift conditioned by nonneutral suffixes also made a unique contribution to word identification after controlling for nonverbal intelligence, vocabulary, phonemic awareness, and morphological awareness. This conflicted with the findings by Chung and Jarmulowicz (2017), who reported adult learners’ performance on stress production in nonneutral derivatives was not an important predictor of word reading. The different results may be due to the task design. Their study used only real words in the stress production task, whereas the current study included low-frequency and nonword derivatives. Compared with the use of real words in the task of assigning stress to nonneutral derivatives, nonwords may be more likely to tap into the awareness of stress shift cued by nonneutral suffixes and hence into the aspect of lexical prosodic competence that contributes to word identification. This is because nonwords prevent reliance on stored lexical knowledge and instead trigger the implicit awareness of prosody conditioned by suffixes. In contrast, performance on real words may be influenced by prior familiarity or memorization, potentially masking the true role of stress shift awareness. Overall, the inclusion of nonword derivatives in the stress production task of the current study likely provided a more sensitive measure of learners’ awareness of stress shift conditioned by nonneutral suffixes, revealing contributions to word identification that might not be apparent when only real words were used.
Additionally, the current study shows that stress production in nonneutral derivatives was a less important predictor of word identification than lexical stress perception. This suggests that adult EFL learners may tend to store the prosodic information in the mental repository of vocabulary by memorizing the stress location rather than by explicitly understanding and applying the rules of stress location conditioned by nonneutral suffixes. Their explicit knowledge of prosodic rules cued by nonneutral suffixes may still be developing.
The mediation analyses in the present study showed the same pattern underlying the effects of both types of lexical prosodic competence on word identification. Prosodic competence contributed to phonemic awareness and morphological awareness, which partially mediated the prosody-reading relationship. More importantly, echoing Chan and colleagues’ (2020) study among older children, the receptive and expressive prosodic skills had direct effects on word identification, playing additional roles in lexical retrieval beyond segmental phonological awareness and morphological awareness. Such direct effects of lexical prosodic competence on word identification were not found in younger children; instead, only indirect effects were observed (Holliman et al. 2014; Kim and Petscher 2016). It is possible that, with the accumulation of reading experience and more exposure to multisyllabic words, the influence of prosodic competence on word identification may gradually become more direct and stronger. The findings from adult EFL learners in the current study support the pathway model of prosody-reading relationship proposed by Wood et al. (2009), with phonemic awareness and morphological awareness as mediators. Moreover, the results corroborate the position that suprasegmental phonological information is essential for lexical representation quality (Wade-Woolley et al. 2022).
Lexical prosodic competence and nonword decoding
The unique variance in nonword decoding explained by each type of lexical prosodic competence, after controlling for vocabulary and phonemic awareness, suggests that prosodic competence may not be redundant with other decoding-related skills but makes a distinct contribution to nonword decoding, even for adult EFL learners who may have achieved a certain level of segmental phonological proficiency.
Consistent with the findings of Jarmulowicz et al. (2008) study with L1 third graders, stress production in nonneutral derivatives showed a direct effect on nonword decoding among adult EFL learners in the current study. This may be due to the common processing mechanisms shared by both tasks at the levels of phonemes and syllables. Adding stress-shifting suffixes to base words entails a set of expressive skills, such as manipulating the primary stress, reassembling syllables and changing vowels from reduced to full or vice versa, which may also be involved in decoding nonwords. Besides, unlike the task of stress production in nonneutral derivatives adopted by Chung and Jarmulowicz (2017), which consisted only of real words, the task in this study included not only real words but also nonsense stimuli, which might necessitate the same phonological processing as the task of nonword decoding did.
The statistically significant contribution of word stress perception to nonword decoding beyond IQ, vocabulary, phonemic awareness and morphological awareness is consistent with the findings by Mundy and Carroll (2012) among L1 adults with and without dyslexia. Word stress perception involves detecting relative pitch changes, duration changes, and intensity variations that signal which syllable in a word carries emphasis. Such sensitivity may help segmenting a phonological sequence into more manageable syllabic units and for mapping the plausible phonological representations onto written forms when decoding unfamiliar words, especially multisyllabic ones. In the early stage of learning to read in English, segmental phonological awareness is of crucial importance for children to decode simple or monosyllabic words successfully. Yet in the later stage of reading development, as learners increasingly encounter unfamiliar multisyllabic words, the common prosodic patterns captured through stress perception may be also essential in facilitating the clustering of phonemes into syllables. This, in turn, may support the accurate and efficient decoding process (Wade-Woolley and Heggie 2016).
There seems to be an overlap in the contributions of both types of lexical prosodic competence to nonword decoding. Accurate nonword decoding relies on applying phonological knowledge and skills, among which perception of word stress and production of prosodic patterns in nonneutral derivatives are important suprasegmental component skills (Wade-Woolley et al. 2022). The process of decoding multisyllabic nonwords necessitates the abilities to correctly produce the stress in the nonword and to match it to the word in their mental lexicon. In the current study, both types of lexical prosodic competence may draw upon a largely shared cognitive and linguistic resource. This includes foundational skills such as syllable segmentation, and the quality and accessibility of detailed phonological representations that encode stress patterns, which are essential in the nonword decoding task (Cutler, Dahan and van Donselaar 1997).
Both types of lexical prosodic competence showed direct effects on nonword decoding, with phonemic awareness as the sole mediator in the current study. The result is not surprising in view of the contribution of lexical prosodic competence to phonemic awareness and the crucial role of phonemic awareness in phonological decoding, which has been well acknowledged in previous studies (Ehri 2020). The findings support one pathway that Wood et al. (2009) proposed about the relationship between prosodic competence and reading ability, with phonemic awareness as the mediator.
Limitations, implications and future directions
There are several limitations that should be acknowledged and addressed in the future study. Firstly, the current study investigated two types of lexical prosodic competence and their relationships to word reading. Future studies can explore the influence of different prosodic task designs (Wade-Woolley et al. 2022) on the word reading process. It is also worth exploring the relative contributions of performance on different prosodic tasks to reading comprehension.
Secondly, although low-frequency and nonsense stimuli were incorporated in the task on stress production in nonneutral derivatives, the familiarity effect observed with real word stimuli warrants caution when generalizing about the extent to which learners consistently apply implicit knowledge of stress shift cued by nonneutral suffixes in all contexts, particularly with familiar nonneutral derivatives. Future research could employ nonsense derivatives with nonneutral suffixes as the only type of stimuli (Wade-Woolley and Heggie 2015), to disentangle the relative contributions of lexical retrieval from the mental lexicon and awareness of stress shift cued by nonneutral suffixes.
Further, while the homogeneity of the sample (e.g., similar educational background, English proficiency level) strengthened internal validity, it limited the generalizability of the findings to other populations (e.g., younger learners, learners from different L1 backgrounds, or more advanced EFL learners). Future studies could benefit from a more diverse sample, including learners at different proficiency levels and from different L1 backgrounds, to explore how prosodic competence interacts with word reading across varying contexts. Meanwhile, the sample size of the current study also limited the scope of the investigation into the full pathway model of prosody-reading relationship (Wood et al. 2009). A larger sample size is needed in future research to investigate how the full pathway model applies to readers as a function of different age groups, proficiency levels, and reading skills. Lastly, interventional studies on lexical prosodic competence training are necessary to investigate the effectiveness of prosody instruction in facilitating multisyllabic word reading.
The results of the present study have implications for EFL reading instruction. Given what this research has found about the important role of lexical prosodic competence in word reading, explicit instruction on English prosody needs to be incorporated within broader vocabulary and reading instruction in EFL settings. For EFL learners, especially those whose L1 contains a very different prosodic system than English, sufficient training on identifying word stress can help learners understand English lexical rhythm structure, which would in turn enhance their reading development. Further, as learners encounter morphologically complex words more frequently in later stages of reading development, explicit and systematic instruction on the rules governing stress placement cued by suffixes, particularly nonneutral suffixes (e.g., -ic, -ity, -graphy, -eous), may facilitate decoding accuracy and fluency, and also help learners develop more robust phonological representations of these complex words.
Summary
The current study focused on lexical prosodic competence and word reading ability among Chinese university EFL learners. The results indicated that the adult EFL learners possessed the ability to identify the primary stress in multisyllabic words and may have implicit knowledge of stress cued by nonneutral suffixes. Both types of lexical prosodic competence, i.e., word stress perception and stress production in nonneutral derivatives, explained unique variance in word identification beyond other metalinguistic skills, but word stress perception appears to be a better predictor of word identification. Both types of lexical prosodic competence predicted nonword decoding after controlling for nonverbal ability, vocabulary, phonemic awareness, and morphological awareness. Yet, neither predicted nonword decoding uniquely beyond each other. Additionally, the relationships between both types of lexical prosodic competence and word identification were partially mediated by phonemic awareness and morphological awareness. However, only phonemic awareness mediated the effect of each lexical prosodic competence on nonword decoding. Taken together, the results provide supporting evidence for incorporating lexical prosodic competence within the reading systems framework revised by Wade-Woolley et al. (2022), and for the theoretical model of prosody-reading relationship proposed by Wood et al. (2009).
Author contributions
KH conceived and designed the study, collected and analyzed the data, and wrote the manuscript.
Data availability
The data are not publicly accessible due to privacy restrictions related to participant confidentiality. Aggregated data supporting the findings of this study are presented within the manuscript. Further inquiries about accessing additional information can be directed to the corresponding author upon reasonable request.
Competing interests
The author declares no competing interests.
Ethical approval
This study was approved by the Ethics Committee of Sanda University (September 10, 2023; No. 2023013). This approval covers all research activities involving adult participants, including the study design, methodology, data collection, and the informed consent process. All procedures were conducted in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki Declaration and its later amendments.
Informed consent
Written informed consent was obtained from all individual participants during the data collection period from September 2023 to December 2023. Consent forms were signed by the participants, ensuring voluntary participation. All participants were informed of the study’s purpose, the procedures involved, the scope of data use, the measures for data protection, the potential publication of anonymized findings, and their right to withdraw at any time without penalty.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
Ashby, J; Clifton, JC. The prosodic property of lexical stress affects eye movements during silent reading. Cognition; 2005; 96,
Baron, RM; Kenny, DA. The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. J Pers Soc Psychol; 1986; 51,
Council of Europe. The common European framework of reference for languages: Learning, teaching, assessment; 2001; Cambridge University Press:
Brysbaert, M; New, B. Moving beyond Kučera and Francis: A critical evaluation of current word frequency norms and the introduction of a new and improved word frequency measure for American English. Behav Res Methods; 2009; 41,
Carlisle, JF. Awareness of the structure and meaning of morphologically complex words: Impact on reading. Read Writ; 2000; 12,
Castles, A; Rastle, K; Nation, K. Ending the reading wars: Reading acquisition from novice to expert. Psychol Sci Public Interest; 2018; 19,
Chan, JS; Wade-Woolley, L. Explaining phonology and reading in adult learners: Introducing prosodic awareness and executive functions to reading ability. J Res Read; 2018; 41,
Chan, JS; Wade-Woolley, L; Heggie, L; Kirby, JR. Understanding prosody and morphology in school-age children’s reading. Read Writ; 2020; 33,
Chung, WL; Jarmulowicz, L. Stress judgment and production in English derivation, and word reading in adult Mandarin-speaking English learners. J Psycholinguist Res; 2017; 46,
Ministry of Education, National Language Commission. China’s standards of English language ability; 2018; Beijing, The General Office of Ministry of Education: Retrieved from https://cse.neea.edu.cn/res/ceedu/1811/6bdc26c323d188948fca8048833f151a.pdf
Chung, WL; Jarmulowicz, L; Bidelman, GM. Auditory processing, linguistic prosody awareness, and word reading in Mandarin-speaking children learning English. Read Writ; 2017; 30,
Chung, WL; Jarmulowicz, L; Bidelman, GM. Cross-linguistic contributions of acoustic cues and prosodic awareness to first and second language vocabulary knowledge. J Res Read; 2021; 44,
Clin, E; Wade-Woolley, L; Heggie, L. Prosodic sensitivity and morphological awareness in children’s reading. J Exp Child Psychol; 2009; 104,
Cutler, A; Dahan, D; van Donselaar, W. Prosody in the comprehension of spoken language: A literature review. Lang Speech; 1997; 40,
Deacon, SH; Holliman, AJ; Dobson, GJ; Harrison, ECJ. Assessing direct contributions of morphological awareness and prosodic sensitivity to children’s word reading and reading comprehension. Sci Stud Read; 2018; 22,
Duanmu, S. The phonology of standard Chinese; 2007; 2nd ed. Oxford University Press: [DOI: https://dx.doi.org/10.1093/oso/9780199215782.001.0001]
Dunlea J, Spiby R, Wu S, Zhang J, Cheng M (2019) China’s standards of English language ability (CSE): Linking UK exams to the CSE (Research report No. VS/2019/0003). Retrieved from https://www.britishcouncil.org/sites/default/files/linking_cse_to_uk_exams_5_0.pdf
Dunn LM, Dunn DM (2007) Peabody picture vocabulary test-IV. Pearson Assessments, New York
Ehri, LC. The science of learning to read words: A case for systematic phonics instruction. Read Res Q; 2020; 55,
Ehri LC, Nunes S (2002) The role of phonemic awareness in learning to read. In: Farstrup A. E., Samuels S. J. (eds), What research has to say about reading instruction, vol 3. International Reading Association, p 110–139
Enderby, JL; Carroll, JM; Tarczynski-Bowles, ML; Breadmore, HL. The roles of morphology, phonology, and prosody in reading and spelling multisyllabic words. Appl Psycholinguist; 2021; 42,
Field, A. Discovering statistics using IBM SPSS statistics; 2013; London, Sage:
Goswami, U; Thomson, J; Richardson, U; Stainthorp, R; Hughes, D; Rosen, S; Scott, SK. Amplitude envelope onsets and developmental dyslexia: A new hypothesis. Proceedings of the National Academy of the Sciences of the United States of America; 2002; 99,
Hayes, AF. Introduction to mediation, moderation, and conditional process analysis: A regression-based approach; 2022; New York, The Guilford Press:
Heggie, L; Wade-Woolley, L. Prosodic awareness and punctuation ability in adult readers. Read Psychol; 2018; 39,
Holliman, AJ; Wood, C; Sheehy, K. The contribution of sensitivity to speech rhythm and non‐speech rhythm to early reading development. Educ Psychol; 2010; 30,
Holliman, AJ; Mundy, IR; Wade-Woolley, L; Wood, C; Bird, C. Prosodic awareness and children’s multisyllabic word reading. Educ Psychol; 2017; 37,
Holliman, AJ; Critten, S; Lawrence, T; Harrison, E; Wood, C; Hughes, D. Modeling the relationship between prosodic sensitivity and early literacy. Read Res Q; 2014; 49,
Jarmulowicz, L. English derivational suffix frequency and children’s stress judgments. Brain Lang; 2002; 81,
Jarmulowicz, L; Hay, SE. Derivational morphophonology: Exploring errors in third graders’ productions. Lang Speech Hear Serv Sch; 2009; 40,
Jarmulowicz, L; Taran, VL; Hay, SE. Third graders’ metalinguistic skills, reading skills, and stress production in derived English words. J Speech Lang Hear Res; 2007; 50,
Jarmulowicz, L; Hay, SE; Taran, VL; Ethington, CA. Fitting derivational morphophonology into a developmental model of reading. Read Writ; 2008; 21,
Juffs, A. Tone, syllable structure and interlanguage phonology: Chinese learners’ stress errors. Int Rev Appl Linguist Lang Teach; 1990; 28,
Kim, YSG; Petscher, Y. Prosodic sensitivity and reading: An investigation of pathways of relations using a latent variable approach. J Educ Psychol; 2016; 108,
Kline, RB. Principles and practice of structural equation modeling; 2023; 5th ed. New York, The Guildford Press:
Kotzer, M; Kirby, JR; Heggie, L. Morphological awareness predicts reading comprehension in adults. Read Psychol; 2021; 42,
Levesque, KC; Breadmore, HL; Deacon, SH. How morphology impacts reading and spelling: Advancing the role of morphology in models of literacy development. J Res Read; 2021; 44,
Lin, CY; Wang, M; Newman, RS; Li, C. The development of stress sensitivity and its contribution to word reading in school-aged children. J Res Read; 2018; 41,
Mattys, SL. The perception of primary and secondary stress in English. Percept Psychophys; 2000; 62,
Mundy, IR; Carroll, JM. Speech prosody and developmental dyslexia: Reduced phonological awareness in the context of intact phonological representations. J Cogn Psychol; 2012; 24,
Olson R, Forsberg H, Wise B, Rack J (1994) Measurement of word recognition, orthographic, and phonological skills. In: Lyon G. R. (ed) Frames of reference for the assessment of learning disabilities: New views on measurement issues. Paul H. Brookes Publishing Co, Baltimore, p 243–277
Park, JH; Kuo, LJ; Dixon, Q; Kim, H. Korean-English bilingual children’s stress cue sensitivity and its relationship with reading in English. J Psycholinguist Res; 2022; 51,
Raven, J; Raven, JC; Court, JH. Manual for Raven’s progressive matrices and vocabulary scales. Section 3: The standard progressive matrices; 2000; San Antonio, TX, Harcourt Assessment:
Wade-Woolley, L. Prosodic and phonemic awareness in children’s reading of long and short words. Read Writ; 2016; 29,
Wade-Woolley, L; Heggie, L. Implicit knowledge of word stress and derivational morphology guides skilled readers’ decoding of multisyllabic words. Sci Stud Read; 2015; 19,
Wade-Woolley, L; Wood, C; Chan, J; Weidman, S. Prosodic competence as the missing component of reading processes across languages: Theory, evidence and future research. Sci Stud Read; 2022; 26,
Wade-Woolley L, Heggie L (2016) The contributions of prosodic and phonological awareness to reading: A review. In: Thomson J., Jarmulowicz L. (eds.), Linguistic rhythm and literacy. John Benjamins Publishing Company, Amsterdam, Philadelphia, p 3–24
Whalley, K; Hansen, J. The role of prosodic sensitivity in children’s reading development. J Res Read; 2006; 29,
Wilson-Fowler, EB; Apel, K. Influence of morphological awareness on college students' literacy skills: A path analytic approach. J Lit Res; 2015; 47,
Wood, C. Metrical stress sensitivity in young children and its relationship to phonological awareness and reading. J Res Read; 2006; 29,
Wood, C; Terrell, C. Poor readers’ ability to detect speech rhythm and perceive rapid speech. Br J Dev Psychol; 1998; 16,
Wood C, Wade-Woolley L, Holliman AJ (2009) Phonological awareness: Beyond phonemes. In: Wood C., Connelly V. (eds), Contemporary perspectives on reading and spelling. Routledge, London, p 7–23
Woodcock RW (1998) Woodcock reading mastery tests—Revised normative update: Examiner's manual. American Guidance Service, Minnesota
Yeong, SHM; Fletcher, J; Bayliss, DM. Impact of early home language exposure on phonological and orthographic skills and their contributions to English literacy abilities in English monolingual and Chinese–English bilingual adults. Appl Psycholinguist; 2017; 38,
Yu, VY; Andruski, JE. A cross-language study of perception of lexical stress in English. J Psycholinguist Res; 2010; 39,
© The Author(s) 2025. This work is published under http://creativecommons.org/licenses/by-nc-nd/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.