4.4 Linear Mixed Modelling
4.5.6 Passage Reading
4.5.7.2 Group Differences in Trajectories for Writing
Over the course of the study, there was a tendency for productive measures of writing to remain fairly stable, while a relatively greater degree of change was observed in incidence of writing errors. Between t1 and t3, the EAL group maintained its higher number of T-units, increasing slightly from g = 0.41 to 0.52. The two groups diverged somewhat in terms of MLTw due to the differences in the steepness of group trajectories (Mono = 0.91; EAL = 0.49). Thus, while both groups continued to produce longer T-units at each subsequent time point, the monolingual group appeared to do this at almost twice the rate of the EAL group. Interestingly, this pattern was not found in oral narrative measures, wherein the two groups made very similar progress over time in MLUw.
On the other hand, longitudinal modelling of writing error rates revealed different patterns depending on the variable used. In terms of total error rate, the monolingual group began on a higher intercept but showed a steeper decline in error rate over time and as a result was making fewer errors than the EAL group by t3 (see Figure 4.24). Further analysis by error type revealed this pattern to be due to a decrease in the spelling error rate of the monolingual group. In contrast, the morphosyntactic error rate of the monolingual group remained very low and stable across the study, while the EAL group made fewer errors over time (Slopes = Mono = -0.01; EAL = -0.03). Despite some convergence between the groups in morphosyntactic error rate, there remained a very clear monolingual advantage by t3 (g = 0.41).
Writing development of monolingual and bilingual samples has been less often considered from a longitudinal perspective; however, some work has examined progression in spelling ability. For instance, studies of children in Canada have found evidence of an early bilingual advantage in single-word spelling (Jonejan et al., 2007; Lesaux & Siegel, 2003). In these studies, it was the bilingual group that maintained its advantage on single-word spelling measures, in contrast to findings of the present study which showed an initial bilingual advantage at t1, followed by the tendency for an interaction between time and group, such that the monolingual children were
outperforming their EAL peers by the end of the study. This pattern also contrasts with that reported in Jonejan et al. (2007) in which monolingual and bilingual children made very similar rates of progress over time. However, it is important to point out crucial differences in the analysis of spelling skills in single-word spelling tests of increasing difficulty as opposed to free writing tasks in which children choose which words to attempt. Similar to criticisms levied at the YARC (Colenbrander et al., 2017), different results might have arisen had participants been able to attempt spelling a wider range of words.
Finally, although the study into EAL writing in Cameron and Besser (2004) was not longitudinal in nature, the authors did compare writing samples of children at two educational stages; namely KS2 (age 7-11) and KS4 (age 14-16). The study found a number of similarities between the writing of children in each key stage, including similar use of adverbials and number of subordinate clauses, but also fewer agreement errors in KS2. Thus, the study provides evidence of important differences between the written language abilities of monolingual and EAL pupils up until the end of compulsory education.
Summary. At t1, children learning EAL tended to produce slightly more T-units, although these also tended to be shorter than those of their monolingual peers. Although the two groups ap- peared to differ little in terms of the average number of errors per T-unit, this picture changed once errors were disaggregated by type: specifically, EAL learners exhibited a lower rate of spelling er- rors, but a higher rate of morphosyntactic errors. The finding of a bilingual advantage in spelling is supported by the literature, and similarly, there is evidence that EAL learners make a higher proportion of morphosyntactic errors in their writing than their monolingual peers even after five years of instruction.
In terms of progress over time, the two groups made a relatively more stable rate of progress in productive measures (total T-units, MLTw) than in error rate. Effect sizes reversed direction over time for total error rate due to the monolingual group improving more quickly than the EAL group in spelling; on the other hand, while EAL learners did show a tendency to make fewer morphosyntactic errors over time, their average rate of progress was insufficient to close the gap by t3. While there was some degree of similarity between oral and written narrative measures (e.g. in total number of T/C-units, MLTw, and MLUw), some differences were observed in group slopes: specifically, the monolingual group made roughly double the rate of progress of the EAL group in MLTw, but the same rate in MLUw, suggesting some disparity between oral and written language development in the two groups. The close connection between oral and written narrative measures in general, but divergence in development of MLTw, represents a novel finding among this population of learners in the U.K.
4.5.8
Summary
In keeping with results of studies of EAL learners in the U.K. (e.g. Hutchinson et al., 2003; Bur- goyne et al., 2009, 2011a; Babayi ˘git, 2014a), at t1 the monolingual group in the present study exhibited significant advantages in receptive and expressive measures of vocabulary, expressive grammar (CELF FS common items), and spoonerism performance, and produced significantly longer and less errorful utterances in speech than their EAL learning peers. Where EAL learn- ers showed relative weaknesses in other areas, these effects were not statistically significant
(namely listening comprehension, lexical diversity in spoken language, or length or error rate of sentences in written language. In contrast, it was EAL learners who showed significant advan- tages in measures of oral and written language productivity (total C-units in speech and T-units in writing); although EAL learners showed trends for faster RAN of letters and digits as well as higher single-word reading efficiency, these effects were not statistically significant. These results also generally accord with the international and U.K.-based literature, in which bilingual learners are found to exhibit specific profiles of strengths (lexical access, single-word decoding, spelling) and weaknesses (oral language, vocabulary, and grammar; cf. ‘profile effects’ Oller et al., 2007; Cline & Shamsi, 2000; Geva & Farnia, 2012; Hutchinson et al. 2003; Burgoyne et al. 2009, 2011a).
However, the results of the present study depart from the literature in terms of passage read- ing performance, in which EAL learners performed significantly below their monolingual peers in reading accuracy, and in which the two groups did not differ significantly overall in their passage reading comprehension (although this was the case for YARC passage 3 alone). Despite these findings, however, both groups were found to be performing within the average range. Another unexpected finding was the pattern for all children to become slower over time in RAN of letters, while they became faster in RAN of digits.
A number of the group difference effect sizes in the present study are moderately to sub- stantially smaller than those reported in other studies with younger bilingual learners who are not necessarily matched with their monolingual peers on instructional experience. However, the fact that significant group effects were found after three to four years of ordinary classroom instruction suggests that day-to-day school experience is not sufficient to fully address the ‘disadvantages’ or ‘lags’ of EAL learners relative to their monolingual peers in certain skills. The need for explicit, tar- geted, oral language instruction has been recommended for all learners generally (Bercow, 2008) and specifically for children with EAL (e.g. Babayi ˘git, 2014a; Burgoyne et al., 2011; Hutchinson et al., 2003); indeed, such a recommendation would similarly be warranted from the analyses pre- sented here. Brief comparison with the results of Babayi ˘git (2014a) does appear to suggest that the EAL learners in this study were performing closer to their monolingual peers than has previ- ously been found in U.K.-based studies. To some extent, the similarity between the two groups may have been due to higher than expected performance of the EAL group, or alternatively lower than expected performance of the monolingual group (as indicated by reference to standardised scores). While further work will be required in order to determine age- or year-level expectations for EAL learners as regarding language and literacy development, it should be emphasised that such similar performance between the two groups should not result in complacency on the part of educational practitioners. Rather, it remains paramount that all learners continue to receive access to rich and high-quality language and literacy teaching (Bercow, 2008).
Following from this conclusion, questions may be asked as to the role of explicit instruction of EAL learners in altering developmental trajectories and promoting a closing of the gap in per- formance between these children and their monolingual peers. The following three chapters deal explicitly with the design and implementation of a vocabulary teaching intervention for a subgroup of EAL learners from the longitudinal cohort study. Chapter 5 will review literature concerning factors affecting word learning, effective facets of vocabulary instruction, and intervention studies with bilingual learners. Chapters 6 and 7 will then present the methods, results, and discussion of the intervention.
Literature Review II: Vocabulary Instruction for Mono- and Bilin-
gual Learners
Bilingual children, including EAL learners in England, are often found to have lower levels of vo- cabulary knowledge than their monolingual peers (Section 2.1.2.3) and indeed, the results of the longitudinal study described in Chapter 4 confirmed this pattern. The input-dependent, uncon- strained, and multidimensional nature of word knowledge, as well as its role in reading compre- hension (Section 2.3), make it a strong candidate for explicit instruction, especially within samples of bilingual learners who possess adequate decoding skill but lower levels of reading comprehen- sion than their monolingual peers (Murphy & Unthiah, 2015). Due to split exposure, the vocabulary knowledge of bilingual children tends to be distributed across their languages, often leading to a smaller stock of word knowledge in one language than that possessed by a monolingual speaker of that language (Pearson et al., 1993). While this may be an expected developmental pattern in the case of bilingual vocabulary development, it remains incumbent upon educators, especially within a ‘monolingual’ or ‘submersion’ form of education such as that employed in England (Baker, 2006; Section 1.1), to promote the vocabulary knowledge of EAL learners who are tasked with accessing the same English language curriculum and assessments as their monolingual peers.
The purpose of the following literature review is to provide some background regarding im- portant factors in children’s word learning, with the ultimate aim of informing the design and implementation of a bespoke vocabulary intervention programme described in Chapter 6. The review will begin by considering children’s implicit vocabulary acquisition as a result of exposure to oral and written language (but focusing primarily on written language), and will then move on to discuss explicit teaching of vocabulary, namely definitional and contextual approaches. Fol- lowing this, some key considerations in the design of vocabulary interventions will be discussed, including the selection of words to teach, contextual affordances, active engagement with target words, dosage and multiple exposures, and provision of child-friendly definitions. As much of this literature is based on word learning processes and instruction in monolingual children, the final section of the chapter will focus specifically on vocabulary intervention studies in populations of bilingual learners, including children learning EAL in England.
5.1
Incidental Learning of Word Meanings
School-age children are typically found to increase their word knowledge by a rate of 2,500 to 3,000 words a year (Beck & McKeown, 1991; Nagy & Scott, 2000), however this rate is somewhat smaller if one considers only root words (Biemiller & Slonim, 2001). Such a rate of learning can-
not be due solely to explicit classroom instruction, pointing to the conclusion that children must acquire a large amount of vocabulary through exposure to oral and written language (Cunning- ham, 2005; Krashen, 1989). Indeed, studies show that children and adults can successfully glean the meaning of novel vocabulary through only brief encounters in texts (Nagy, Herman & Ander- son, 1985; NRP, 2000; Martin-Chang & Levesque, 2013; Webb, 2007). Such learning may be described as incidental, as opposed to intentional, in that word meanings are acquired through activities (such as free reading) which do not have the explicit aim of teaching vocabulary, and in which the meanings of unfamiliar words are merely expected to be inferred (NRP, 2000; Swanborn & de Glopper, 1999).
The rate at which incidental learning of vocabulary occurs from reading has been found to be a function of prior vocabulary knowledge and age. In a meta-analysis of 15 studies of incidental word learning of learners in Grades 4 to 11 (ages 9 to 17), Swanborn and de Glopper (1999) found a negative relationship between density of unknown words and probability of novel word learning. For instance, at a rate of 1 unknown word for every 150 words, there is a 30% probability of learning; when the density increases to 1 unknown word for every 75 words, this probability decreases to 14%. Additionally, probability of incidental word learning increased from 8% in Grade 4 to 33% in Grade 11, suggesting that children are able to acquire higher levels of vocabulary incidentally over time, potentially as a result of access to an increasingly large stock of word knowledge, serving to decrease the ratio of known to unknown words. In McKeown’s (1985) study in the U.S., 30 monolingual fifth-graders were allocated to groups according to their performance on the vocabulary subtest of the Stanford Achievement Test (15 ‘low’ and 15 ‘high’ ability children). Nonwords were presented within a number of different sentential contexts, for example “Standing in front of it we all agreed that it seemed like a narp house” (p.485). When asked to define the target non-words, participants were awarded points for a range of behaviours, including correctly deciphering the word’s meaning, giving justifications for their answers based on the context, and correctly discriminating good and bad exemplar sentences containing the target word. The high ability group significantly outperformed the low ability group, suggesting that verbal aptitude, as indicated by prior vocabulary knowledge and verbal reasoning ability, may contribute to the ability to correctly derive meaning from context.
In summary, the results of studies of incidental word learning suggest that novel word mean- ings may be acquired through exposure alone (Nagy et al., 1985; Martin-Chang & Levesque, 2013; Webb, 2007). However, the probability of this learning is low, especially in contexts in which there is a high density of unknown words (Swanborn & de Glopper, 1999) and for learners with lower levels of prior vocabulary knowledge (McKeown, 1985). As a result, a young EAL learner with a relatively lower level of English vocabulary knowledge may face difficulties in acquiring word knowledge in an incidental fashion in comparison to a monolingual peer who is likely to have a larger stock of word knowledge to draw from. Another issue, not addressed explicitly by studies of incidental word learning discussed above, is the depth of vocabulary knowledge that is typically acquired through incidental learning (e.g. exactly what knowledge is acquired in terms of form, meaning, and function from only a small number of exposures; Nation, 2001; see Section 5.3.4 be- low). Given that EAL learners in England tend to have lower levels of vocabulary knowledge than their monolingual peers and that the two groups of children show similar developmental trajecto- ries in vocabulary acquisition as a result of engagement in the mainstream curriculum (Section
2.1.2.3), a case can be made for explicit and targeted vocabulary instruction for EAL learners in order to close the gap in knowledge with their monolingual peers. The review now turns to the role of explicit instruction in vocabulary learning, and in particular will introduce a distinction between definitional and contextual approaches to vocabulary teaching.