Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Introduction

In this chapter, we report some initial findings of a quasi-experimental study that involves teaching English derivational morphology to fourth graders in Singapore, a multilingual country where English is the medium of school instruction but not necessarily children’s dominant home language. We chose to focus on derivation because derived words are prevalent in English print, yet English derivation is a challenging morphological process to school children (Kuo and Anderson 2006; Tyler and Nagy 1989). We were particularly interested in the extent to which semester-long instruction on English derivation would accelerate the development of derivational awareness and lexical inference or word learning ability and if the relationship of derivational awareness with lexical inference would change as children progress through schooling at the primary level.

English Morphological Awareness and Its Development

Morphological awareness pertains to “the ability to reflect upon and manipulate morphemes and employ word formation rules in one’s language” (Kuo and Anderson 2006, p. 161). Carlisle argued that morphological awareness “must have as its basis the ability to parse words and analyze constituent morphemes for the purpose of constructing meaning” (2000, p. 170). Morphological awareness emerges from early spoken language acquisition and continues to develop after schooling commences. It is multidimensional and multifaceted, entailing different types and levels of insights that develop in disparate trajectories (Berninger et al. 2010; Tyler and Nagy 1989). Some insights are acquired very rapidly at the early stage of schooling; others, however, develop fairly slowly, necessitating considerable exposure and experience with printed words (Berninger et al. 2010; Mahony 1994; Tyler and Nagy 1989).

In accordance with the three major word formation processes (Plag 2003), i.e., inflection, derivation, and compounding, English morphological awareness can be of three types, including inflectional, derivational, and compound awareness. English inflectional affixes are very small in number, and inflected words are largely regular structurally, with no or very limited phonological or orthographic shift (e.g., jumps, jumping, and jumped). Awareness of inflectional morphology is thus a comparatively early acquired competence and imposes little difficulty on learning to read (Kuo and Anderson 2006).

Clark (1993) noted that compounding can be used by children as young as two for lexical coinage even before they acquire certain affixation rules. Children may refer to a person who reads as read-man rather than using the agentive rule to form the word reader. Clark et al. (1985) reported that from age 2 onward, children could identify the head in a novel compound apple-knife and pick out a picture with a knife on it. This knowledge of the modifier-head relation developed rapidly from age 2 to 4, and children above 4 generally had no problem in choosing the correct referent of a novel compound.

In comparison to inflection and compounding, derivation is more challenging to children in reading acquisition . This is because of the large number of derivational affixes in English and the fact that adding a derivational affix to a base leads to change of the meaning, and often the grammatical category, of the base; and derived forms often involve phonological or/and orthographic changes (e.g., decide and decision).

Derivational awareness can entail different levels of insights that develop with diverse timetables. Tyler and Nagy (1989) and McCutchen et al. (2008) differentiated between three types of derivational knowledge, namely, relational, syntactic, and distributional. Relational knowledge is typically measured with a morphological relatedness task in which children are to judge whether the first word in a pair comes from the second (e.g., thinker/think; corner/corn) (e.g., Nagy et al. 2006). Relational knowledge develops rapidly in early primary years, and the development levels off as students move onto upper primary and beyond (Berninger et al. 2010). Eighth graders’ performance on the morphological relatedness task was near ceiling (Tyler and Nagy 1989).

Syntactic knowledge pertains to the understanding that suffixation usually changes the grammatical category of a base, for example, brightness is a noun, brightly is an adverb, whereas brighten is a verb. It is more challenging than relational knowledge (Nagy et al. 2006; Tyler and Nagy 1989). Berninger et al. (2010) reported that similar to that of relational knowledge, syntactic knowledge development shows a nonlinear pattern – the growth was steeper in grades 1 to 3 than in grades 3 to 5. However, there is no conclusive evidence that syntactic knowledge could be fully developed among high school students (Nagy et al. 2006; Tyler and Nagy 1989). Mahony’s (1994) findings suggested that individual difference in syntactic knowledge could exist even among college students.

Distributional knowledge refers to the insight into the principle that affixation is constrained by the syntactic category of the base to which an affix is attached. For example, homeless is a viable English word, whereas jumpless is not. Understanding the distributional constraints of derivational affixation seems to be the most challenging of the three aspects of derivational knowledge. Because it involves fine-grained representation of prefixes and suffixes and knowledge about their functions, distributional knowledge requires plenty of processing experiences with derived words to develop. In Tyler and Nagy (1989), even eighth graders, who had many years of formal literacy education and print exposure, had difficulty judging whether a derivational suffix was appropriately juxtaposed with a base morpheme of a particular syntactic category.

Morphological Awareness and Word Learning

Previous analyses of the words English -speaking students were exposed to or their vocabulary repertoire suggested that morphological analysis should play an important role in learning new words. Nagy and Anderson (1984) estimated that roughly 60% of the words in the American Heritage Word Frequency Book (Carroll et al. 1971), which was considered as representing printed school English for grades 3 through 9, were semantically transparent and children could use morphological principles to derive their meanings. According to White et al. (1989), the top 20 prefixes and 20 suffixes (both inflectional and derivational), respectively, accounted for about 97% of the 2959 prefixed and 93% of the 2167 suffixed words randomly sampled from the same database (Carroll et al. 1971). Anglin (1993) reported that from grade 1 to 5, the number of derived words learned by children (about 14,000) was more than three times as much as the number of root words (about 4000) known by the same group of children. He used “morphological problem -solving” (p. 5) to refer to the process in which children use their morphological knowledge to decompose complex words (e.g., unquestionable) into their morphemic constituents (un- + question + −able) and infer meanings of novel vocabulary items based on the meanings of these constituents.

However, using morphological clues to derive meanings of unknown words is not necessarily easy, as it requires deep insights into English derivational structure as well as knowledge of the meaning and grammatical function of English derivational affixes, that is, a deep level of derivational awareness. Nagy et al. (1993) reported that even high school students tended to use the base of a derived word to establish the overall meaning of the word, without considering the syntactic and semantic features associated with the suffix(es). The issue now comes to the potential benefit of direct and purposeful teaching of morphology for learning new words by inferring their meanings. As Baumann et al. (2002) remarked, “if students are equipped with the ability to infer word meanings by … analyzing the meaningful parts of words (morphology), they have the power to expand their reading vocabulary significantly” (p. 150).

Empirically, a small but growing number of studies have addressed how morphological instruction would benefit learning of new words, and their findings, overall, support the usefulness of such instruction. For example, in comparing the effects of different vocabulary instruction strategies, Baumann et al. (2002) found that fifth graders who received instruction on some prefix families outperformed those who did not on defining isolated, unfamiliar words by writing their meanings or choosing an appropriate meanings from given choices. A similar finding was also reported in Bowers and Kirby (2010) in which fourth and fifth graders who received focused training in the morphological structure of English derived words through graphic representations (“structured word inquiry”) were more successful than the controls not only in identifying the base as the main meaning-carrying unit in a derived word (i.e., base identification), but also in using the base word knowledge to infer meanings of unknown derived words (i.e., morphological vocabulary).

Possibly because of the importance of morphological analysis to inferring new words’ meanings, morphological awareness has been found to be a significant independent predictor of vocabulary growth (Carlisle 2000). Carlisle (2000) found that third and fifth graders’ derivational awareness accounted for a large amount of variance in their vocabulary breadth, and the amount was much larger for fifth graders than for third graders, which seems to suggest that the relationship of morphological awareness to vocabulary knowledge should strengthen as children move beyond the learning to read stage. Such strengthening of this relationship may be attributed to the possibly more involvement of morphological analysis in word learning or lexical inference, which has not been empirically and longitudinally tested in a single study, even though significant correlations of derivational awareness with the ability to infer meanings of unfamiliar derivatives surfaced in a few studies that focused on diverse and separate groups of English learners (e.g., Bowers and Kirby 2010; Zhang 2013; Zhang and Koda 2012). In Bowers and Kirby (2010), children’s base identification ability significantly correlated with their morphological vocabulary or meaning definition, and after controlling for pretest vocabulary knowledge, base identification explained a significant amount of variance in morphological vocabulary for the control as well as the intervention groups. However, the study did not pretest children’s base identification and morphological vocabulary, correlate them, and compare pretest correlations against posttest ones. Consequently, the question remains to be empirically tested as to whether the relationship between morphological knowledge and word learning or lexical inference ability would change longitudinally.

Derivational Morphology in Primary English Language Curriculum in Singapore

Singapore is a multilingual society comprised of three major ethnic groups: Chinese (about 74% of the population), Malays (about 13%), and Indians (about 9%) (Singapore Department of Statistics 2010). English and the mother tongues (MT) of these ethnic groups (i.e., Chinese, Malay , and Tamil ) are stipulated as the four official languages of the country. Singaporean children, who may not speak their MT at home due to gradual familial language shift toward English (Pakir 2008), are mostly bilingual and biliterate, albeit at varying levels, because they are required to develop proficiency in English as well as their respective MT under the quadrilingual education system in the country. An international lingua franca, English, in addition to being a school subject itself, is the medium of instruction for subject matter curriculum (e.g., math and science). Consequently, English, not necessarily students’ home language, is designated as the “first school language” in school curriculum, while an MT language is taught as a “second school language,” based on the importance ascribed to the respective language in Singapore (Pakir 2008, p. 191; see also Silver and Bokhorst-Heng, this volume).

Primary English language education in Singapore adopts a curriculum entitled Strategies for English Language Learning and Reading (STELLAR), which is designed reflecting the requirements stipulated in the English Language Syllabus 2010 (MOE 2008). The STELLAR curriculum is characterized by a pedagogical model comprised of a set of strategies, such as teacher-guided reading, explicit instruction on language structures, and focused and contextualized practice of linguistic knowledge and skills (e.g., small group writing and independent writing). These pedagogical strategies aim not only for literacy development as typical of any language arts curriculum common to a monolingual English setting, but for English language learning as well (e.g., spoken communication, grammatical knowledge). The objective is to achieve “a principled blend of first language (L1) and second language (L2) teaching methods” (MOE 2008, p. 8).

Such a curricular architecture is arguably designed with good reason. Two important reasons, in addition to that of the language foundation of literacy acquisition , are worth noting: firstly, while there has been gradual home language shift toward the use of English in Singapore, a significant proportion of primary school children still use their MT or ethnic language as the dominant language at home (MOE 2011a). To this group of children, English, though designated as the first language (Silver 2005) or “First school language” (Pakir 2008, p. 191), is actually their second language. Secondly, further complicating the learning of English in Singapore schools is the use of Singlish or Singapore Colloquial English , a local variety of English that is widely spoken in the Singapore society and characterized by, for example, optional marking for plurality and tense and prevalent use of discourse particles, such as lor and meh, due to the influence from the Chinese dialects spoken by early immigrants in the country (Deterding 2007; Silver 2005). Consequently, an objective of English language education in Singapore, including primary school education, is made to facilitate the acquisition of a “standard English,” that is, “internationally acceptable” (MOE 2008, p. 7).

To researchers who are interested in morphological awareness, particularly derivational awareness, an immediate concern is if the English curriculum highlights the teaching of English derivational affixation and the close connection of morphological analysis with word learning and literacy acquisition in this context. Direct instruction on derivational morphology is included in the English Language Syllabus 2010 (MOE 2008), but briefly, and mostly in relation to vocabulary acquisition. For example, “teachers will draw pupils’ attention to how words can be formed…Teaching pupils commonly-used affixes (i.e., prefixes and suffixes) can enhance their vocabulary” (p. 106). In the STELLAR Teachers’ Guidelines (MOE 2011b), derivational morphology is listed as a component of word identification or vocabulary study and is introduced around a small number of selected prefixes and suffixes (e.g., −less and -er).

The above delineations seem to suggest that derivational morphology is an integral component of primary English language curriculum in Singapore. However, there are at least two issues. Firstly, the coverage of useful affixes is limited. A review of the affixes designated for explicit instruction in the STELLAR curriculum for Grades 1 through 3 showed that some frequent, productive prefixes and suffixes are missing, such as -ive, -ity, -ize/-ise, mis-, dis-, etc. At Grade 4, the coverage of new derivational affixes is minimal; only -ive and mini- are introduced. Researchers alert us to the “fourth-grade slump” in reading comprehension (Chall and Jacobs 2003, p. 14). One important reason for this slump is children’s enlarged exposure to informational texts from this period and onward in contrast to the inadequacy of their academic vocabulary, which is essential to comprehending lexically-rich, informational texts (Kieffer and Lesaux 2007). To prevent the slump, therefore, it is critical that academic vocabulary and strategies for acquiring it, such as morphological analysis, be taught to students.

Secondly, and probably more importantly, at the pedagogical level, our conversations with some local primary school teachers, their heads of department, and teacher educators suggested that teaching derivational morphology, even though stipulated in the STELLAR Curriculum, had not received due emphasis in pedagogical practices in Singapore. Our observations of some classes revealed a similar finding. For example, teachers often skipped the sections of the STELLAR curriculum where derivation is covered, and in addressing students’ questions about the meaning of an unknown derivative, they tended to provide a direct meaning definition instead of drawing students’ attention to morphological principles.

Research Questions

For the reasons delineated above, we developed a morphological intervention program that aimed to facilitate Grade 4 children’s acquisition of English derivational morphology, word learning, and literacy acquisition. The present chapter focuses only on morphological instruction and word learning. We had three research questions regarding this selected focus:

  1. 1.

    Does instruction on derivational morphology enhance children’s development of derivational awareness?

  2. 2.

    Does morphological instruction enhance children’s word learning ability?

  3. 3.

    To what extent is derivational awareness related to word learning ability, and does the strength of the relationship change longitudinally?

Methodology

Participants

Grade 4 children in two neighborhood schools participated in the present study, one receiving the intervention and the other serving as control. Neighborhood schools in Singapore are government-funded schools where student population is ethnically diversified. The two schools were compatible with respect to the level of the socioeconomic status (SES ) of the student population. Like all other government-funded schools in Singapore, they followed the same set of national curricula, including the STELLAR curriculum for English language education, mandated by the Singapore Ministry of Education .

In Singapore, a typical mid or upper primary English language class in a neighborhood school is comprised of 35–40 students of diverse ethnic backgrounds, with the Chinese as the dominant group. Such was also true of the two participating schools. Before the intervention was implemented in the first semester of Grade 4 and when the pretesting was conducted at the end of Grade 3, there were seven classes totaling 252 students in the intervention school (M = 36) and five classes totaling 196 students in the control school (M = 39). For the study reported in this chapter, only the Chinese children were involved, for two reasons. Firstly, the Chinese are the largest ethnic group of the student population. Focusing on the Chinese group, instead of other ethnic groups, allowed us to maintain a significant number of participants for reliable data analysis after accommodating various missing data or data attrition. Secondly, the Chinese children’s MT, Chinese, is a language that is impoverished in derivational morphology, which has been reported as particularly challenging to Chinese learners of English as compared to learners whose L1 has a productive derivational process (Ramirez et al. 2011). Consequently, with presumably limited support from their MT, Chinese children’s acquisition of English derivational morphology warrants particular attention (See also Sun and Curdt-Christiansen, this volume).

Among students in these classes, only those who were born in Singapore or had lived and studied in Singapore from before the age of 5 (i.e., before kindergarten) were included. In addition, due to attrition, lack of parental consent, or missing data, the final number of students included in the analyses was 86 in the intervention school and 79 in the control. The intervention students included 49 boys and 37 girls, with an average age of about 9.30 (SD = .33) when they were pretested at the end of Grade 3; the 79 control students included 41 boys and 38 girls, with an average age of 9.46 (SD = .56) at pretesting.

Intervention Program and Instructional Strategies

Designing and Pedagogical Principles

Drawing upon recent research and discussions on morphological intervention (e.g., Carlisle 2010; Kieffer and Lesaux 2007), we followed a few principles in conceptualizing, designing, and translating into pedagogy the intervention program. Note that while the present chapter’s focus is on morphological awareness and word learning, our intervention actually had a broader range of objectives that also included reading development . The first principle was that the program should integrate the development of knowledge as well as strategy with respect of English derivation. It should emphasize not only explicit teaching of derivational affixes and derivational structure but also the strategies to apply acquired knowledge in literacy-related activities. Our intervention program made explicit to students (1) that English derived words are constructed by juxtaposing two or more meaningful units (i.e., morphemes), and a derived word is constructed with an affix or affixes attached to a base word, and as such, it can be segmented into these components; (2) that derivation may alter the pronunciation and/or spelling of a base word in various ways (e.g., sign → signature; decide → decision); (3) the meaning and function of English derivational affixes, particularly that of suffixes in signaling the grammatical category of derived words; and (4) that the base carries the core meaning of a derived word, which is then modified by the affix(es) attached to the base. The meaning of a derived word can, therefore, be inferred by first segmenting the word into its base and affix(es) and then integrating the meanings of these components.

Developing the above knowledge and skills necessitates a set of pedagogical strategies. Therefore, the second principle that we followed, at the pedagogical level, was that the program should integrate explicit instruction, teacher modeling, and teacher-guided activities, as well as collaborative and independent student activities. In addition, there should be opportunities for teachers to re-teach and students to relearn and practice what had been previously covered as the program moved forward with new affixes being introduced. We believed that it was essential that explicit, structured instruction be provided on the form and meaning of target affixes and how they are used to form derivatives and the analytic process be modeled. In addition, students should be given ample opportunities, receptively and productively, to review and practice the various morphological skills, first through teacher-guided, scaffolded practice and then through independent work or collaborative work with their peers.

The Intervention Program

The intervention was designed around the teaching of new prefixes and suffixes. It was divided into eight sessions, including two cumulative review sessions and six regular sessions that corresponded to the six units of the STELLAR curriculum for the first semester of Grade 4. Embedding the regular sessions into the STELLAR curriculum came primarily out of flexibility considerations, that is, it would be easier for teachers to plan for their teaching of the regular curriculum while implementing the intervention program (see further details below). Altogether, 14 new affixes were introduced in the six regular sessions, including five prefixes and nine suffixes. Two affixes, including one prefix (mini-) and one suffix (-ive), were listed as target of instruction in the STELLAR curriculum. Specifically pertaining to the intervention were four prefixes (i.e., in-/im-/il-/ir-, mis-, dis-, re-) and eight suffixes (i.e., -able, -ive/-sive/-tive/-ative/-itive, -ity, -ous/-ious, -ent/-ence, -ian, -ise/-ize, -en, -th) we selected based on previous studies of the utility of English derivational affixes (e.g., Bauer and Nation 1993; Baumann et al. 2012; White et al. 1989). As to the review sessions, Session 1 provided a cumulative review of the affixes that had been introduced in the STELLAR curriculum prior to Grade 4. Session 8 was a cumulative review of the 14 new affixes taught in the intervention. Each review session took about 80 min, each regular session about 40 min. The intervention time totaled approximately 400 min.

Teachers as Intervention Implementers

The present study was a school-based and quasi-experimental one designed with an aim to align with participating schools’ current curricular arrangement. Consequently, we decided that the intervention be implemented by English teachers of the intervention school. They embedded the intervention program into the STELLAR curriculum with their adjustment, whereas the teachers in the control school taught with the regular STELLAR curriculum.

To facilitate the implementation of the intervention, an instructional resource manual was developed that covered the relationship between morphological awareness, word learning, and literacy acquisition as well as details of the eight intervention sessions with implementation guidelines, instructional sheets, and student activities. Two workshops were also conducted prior to the intervention to familiarize the teachers with the contents of the resource manual. The teachers were also given flexibilities in implementing the intervention. They had full freedom as to how to adjust classroom organization and lesson plans to accommodate the intervention. Finally, as the intervention was progressing, classroom observations were conducted in the intervention, and the control schools and a few focused group discussions (FGDs) were arranged for the intervention teachers. Our classroom observations suggested that during the intervention period, morphological instruction was never a focus of the English language classes in the control school, whereas it had been conducted by the intervention teachers, albeit at various levels of depth and duration. The FGDs suggested that overall the teachers had a positive view of the intervention, although due to concerns about inadequate time for teaching the STELLAR curriculum, the intervention was not necessarily fully implemented by all teachers.

Tasks

The following tasks were administered two times, one at the end of Grade 3 (pretest) and the other at the end of the first semester of Grade 4 (posttest). All tasks were printed on paper and group administered in classes by English language teachers. To reduce possible influence of decoding on children’s performance on the tasks, we encouraged teachers to read aloud the instructions and the items of each task as students were working on its printed version.

Morphological Relatedness

Children were to judge whether the second word in a word pair “came from,” or was morphologically related to, the first word. The task included 15 related (e.g., think and thinker) and 15 unrelated (e.g., too and tooth) word pairs. A practice item was also provided. The reliability (Cronbach’s α) of the task was .832 and .847 for the pre- and the posttest, respectively.

Affix Choice (Real)

This task touched on children’s knowledge of the grammatical function of derivational affixes. Children were to select an appropriate derived form to fill in a sentence (e.g., It is not easy to measure the ___of light.) followed by three real derived words that shared a same stem (intensely, intensify, intensity). To choose intensity as the answer, children would need to know that a noun was required of the blank and the suffix -ity nominalizes an adjective. The task had 15 test items and a practice item. The Cronbach’s α was .783 and .816 for the pre- and the posttest, respectively.

Affix Choice (Pseudo)

This task was the same as the affix choice (real) task except that it had three pseudo derivatives formed with a same decodable base. For example, I could feel the ___. (froody, froodful, froodment). If a child knew that -ment is a nominalizer suffix, he/she should be able to correctly choose froodment as the answer. This task also had 15 test items and one practice item. The pre- and posttest reliability was .730 and .879, respectively.

Morphological Production

For this task, children were to produce an appropriate derived form of a given base word to fill in the blank in a sentence that was simple both lexically and grammatically. For example, drink: The water in this area is not ____. There were 40 test items and a practice item in this task. The Cronbach’s α was .943 and .952 for the pre- and the posttest, respectively.

Meaning Inference

This task tapped children’s word learning ability or the ability to apply their knowledge of affixes and derivational structure to infer meanings of unfamiliar English derivatives. Four meaning choices with simple grammatical structures were constructed for each target word. Children were to select the best choice for each word. This task included 20 test derived words and a practice word. To make sure that the test tapped lexical inference ability rather than vocabulary knowledge, the base morphemes of the target words, as opposed to the whole words, were frequent and had been learned by the children. For example, the word familiarize was followed by four answers: to make known, to become famous, a large family, and with a knowledge of. The Cronbach’s α for the pre- and the posttest was .613 and .640, respectively.

Results

Intervention Effects

Table 12.1 shows the control and the intervention groups’ pretest and posttest scores. For the posttest, both the raw score means and the means adjusted for the pretest scores are provided. To address Research Questions 1 and 2, two sets of group comparisons with significance testing were conducted for each of the five tasks: one of the pretest scores with independent sample t-tests to address pre-intervention group equivalence, and the other of the posttest scores via ANCOVA with the pretest scores as a covariate.

Table 12.1 Means and standard deviations of control and intervention groups’ pre- and posttest scores and group comparisons
Full size table

As Table 12.1 shows, for all tasks, there was no significant pretest difference between the control and the intervention groups, indicating group equivalence prior to the intervention. The covariate, or the pretest scores, explained a significant proportion of the variance in the corresponding posttest scores, for all tasks. A comparison of the pretest adjusted means of the posttest revealed that the intervention group failed to outperform the control group significantly on the morphological relatedness task (F[1, 162] = 3.464, p = .065), and the effect size was small (Cohen’s d = .24). Similar analyses revealed that for the affix choice (real) task, the intervention group significantly outperformed the control group, F(1, 162) = 8.208, p = .005, with a small effect size (Cohen’s d = .32); for the affix choice (pseudo) task, the intervention group’s posttest performance was better than that of the control group with marginal significance, F(1, 162) = 3.795, p = .053, Cohen’s d = .22, indicating a small effect size. The intervention group also significantly outperformed the control group on the posttest of the morphological production task with a small effect size, F(1, 162) = 11.638, p < .001, Cohen’s d = .23. Finally, with regard to word learning ability, ANCOVA also revealed significantly better performance of the intervention group on the posttest, F(1, 162) = 6.873, p < .01. Like all other measures, the effect size was small (Cohen’s d = .34).

Contribution of Morphological Awareness to Word Learning

To address the third research question, that is, the relationship of morphological awareness to lexical inference ability longitudinally, correlational analyses were conducted. As shown in Table 12.2 (the two groups combined), all four morphological awareness measures, which were themselves significantly correlated, had significant and positive correlations with meaning inference. Morphological awareness measures that tapped a deeper level of morphological analysis were more strongly correlated with meaning inference (i.e., rs = .492, .517, .534 for affix choice [real], affix choice [pseudo], and morphological production, respectively; all ps < .001) than was a measure that only touched upon segmental or surface structural analysis of English derivatives (i.e., morphological relatedness; r = .346, p < .001). Multiple regression analysis further indicated that the four morphological awareness measures significantly predicted children’s word learning ability, and together they explained about 34.9% of the variance in meaning inference, F(4, 160) = 21.401, p < .001. Table 12.3 provides the correlations between the five measures separately for the control and the intervention groups. Overall, the patterns were consistent with that of the two groups combined, although the correlations appeared to be a little stronger among the intervention group than the control group. Multiple regression analyses showed that for both groups, morphological awareness significantly predicted meaning inference, F(4, 74) = 8.481, p < .001, R 2 = .314 and F(4, 81) = 13.166, p < .001, R 2 = .394, for the control and the intervention group, respectively.

Table 12.2 Correlations between pretest measures with control and intervention groups combined
Full size table
Table 12.3 Correlations between pretest measures
Full size table

The correlations between children’s posttest performance on the five tasks are shown in Table 12.4. While the correlational patterns were similar to those of the pretest, a clear difference was that the posttest correlations between the morphological awareness measures and meaning inference and between the morphological awareness measures themselves were higher than those of the pretest, for the control as well as the intervention groups. For the relationship between morphological awareness and meaning inference among the control group, rs = .581, .656, .656, and .692 (all ps < .001) for morphological relatedness, affix choice (real), affix choice (pseudo), and morphological production, respectively. The correlations of the intervention groups were, respectively, .420, .685, .744, and .705 (all ps < .001). Multiple regression analyses further confirmed that morphological awareness was more predictive of meaning inference in the posttest than in the pretest. The four measures together explained about 56.5% and 59% of the variance in meaning inference in the control (F[4, 74] = 24.034, p < .001) and the intervention group (F[4, 81] = 29.156, p < .001), respectively.

Table 12.4 Correlations between posttest measures
Full size table

Discussion

Intervention Effects

The semester-long teaching of English derivational morphology led to enhanced development of derivational awareness as well as word learning ability defined as the ability to infer meanings of unknown derivational words. Because of the wide coverage of derivational knowledge and skills in the intervention, there is no wonder that the children in the intervention classes outperformed their peers in the control classes on the two affix choice tasks and the morphological production task. And because of the intervention group’s broader and more in-depth knowledge of derivational affixes and derivational structure after the intervention, it is also no surprise that their performance was better on the meaning inference task, which necessitated segmenting an unfamiliar derivative into its morphemic constituents (i.e., base and affix[es]) and synthesizing the meanings of the constituents with consideration of the part-of-speech information signaled by the suffix in the target word.

These findings, on one hand, corroborate previous research on primary school children that focused on a similar topic. For example, Baumann et al. (2002) documented the advantage of teaching English prefix families and “morphemic analysis” (p. 150) over other vocabulary strategies for inferring meanings of unknown words. On the other hand, they showed some differences from previous studies. One difference is that a significant intervention effect failed to surface for the morphological relatedness task. We conjecture that this result may be attributed to the nature of the task itself and the development stage of the participants. The morphological relatedness task, by nature, measures structural analysis ability at the surface level, which develops rather rapidly (e.g., Berninger et al. 2010; Tyler and Nagy 1989). It seems to be a rather easily acquired competence as opposed to the other skills measured in the present study (i.e., affix choice and morphological production) that tapped a deeper level and a broader range of morphological knowledge that usually requires a longer time to develop (Mahony 1994; McCutchen et al. 2008; Tyler and Nagy 1989). Some previous studies found that normally developing children’s performance on tasks similar to the morphological relatedness task was near ceiling by approximately Grade 4 or a little earlier, without having received any morphology-focused instruction (e.g., Berninger et al. 2010; Tyler and Nagy 1989). This similarly happened to the children in the control group in the present study, particularly at the end of the first semester of Grade 4 (see Table 12.1). It seems to suggest that with increased literacy practices and enlarged exposure to printed words, upper primary students, without any need of morphological instruction, could well achieve a similar level as those who have received some instruction. It also seems reasonable, therefore, that the finding of the present study, which focused on normally developing fourth graders, differed from the small number of previous studies that had reported a significant intervention effect with respect to the ability that pertains to morphological segmentation or relatedness, because those studies were either focused on younger children, such as preschoolers or older children with reading difficulties (e.g., Berninger et al. 2003; Casalis and Cole 2009).

Another major difference is that the effect sizes of the intervention were small in the present study and smaller than those of previous studies. Bowers et al.’s (2010) recent meta-analysis of morphological instruction studies yielded an average size of Cohen’s d = .65 of 37 effects that touched upon skills at the morphological level (i.e., a sub-lexical level) (see Bowers, et al.’s Table 12.3, p. 160) and Cohen’s d = .35 of 34 effects that touched upon lexical inference or vocabulary knowledge (or lexical level; see Bowers, et al.’s Table 12.4, p. 162). On the other hand, in correspondence to these categories, the effect sizes were obviously smaller in the present study: Cohen’s ds = .24, .32, .22, and 23 at the sub-lexical level (i.e., the four morphological awareness measures) and .34 at the lexical level (i.e., meaning inference) (see Table 12.1).

The discrepancies in the effect size could be due to the many variations between the present study and the previous ones. In their meta-analyses, Goodwin and Ahn (2010) and Bowers et al. (2010) considered a variety of factors that had led to fluctuating effect sizes in previous morphological instruction research. In addition to linguistic categories of outcome measures (i.e., sub-lexical, lexical, and supra-lexical; Bowers et al. 2010), the factors could be related to participants (e.g., age/grade and type of students), instructional design (e.g., goals, length, intensities of intervention), and research design (e.g., group assignment, that is, whether control and intervention groups are natural classes of students or they are randomly assigned or matched).

While any difference in any one of the factors above could have resulted in the small(er) effect sizes of the present study, we would like to highlight the length of intervention. Goodwin and Ahn’s (2010) meta-analysis indicated that the number of hours could well moderate the size of intervention effect, disregarding the other factors that might interact with this variable of intervention length. For example, they reported that the average effect size was 0.15 and 0.31 for interventions that were 5–10 h and 10–20 h, respectively. Given that our intervention was less than 7 h and that the intervention was not necessarily implemented to its fullest length, the small(er) effect sizes obtained in the present study appear to be reasonable. Having said this, we admit that such a comparison of the effect size of our study with the one reported in Goodwin and Ahn (2010) is exploratory only, as the possible influence of length of intervention may well be moderated by many other factors like student group and research design, as we described above.

Morphological Awareness and Word Learning

We found that all four derivational awareness measures were significantly correlated with meaning inference, for both pre- and posttests, and they explained a large amount of variance in meaning inference, particularly for the posttest. Such a close relationship of morphological awareness to lexical inference is reasonable as successful resolution of the meanings of unknown derivatives necessitates not only the ability to segment the words into their constituent morphemes, i.e., base and affix(es), but also a knowledge of what these components mean, how an affix modifies the meanings of the base, and how it may or may not change the part of speech of the base. This finding is also consistent with previous research that has addressed a similar issue, for example, Bowers and Kirby (2010) on native English -speaking children and Zhang (2013) on young learners of English as a foreign language.

In addition, a few interesting findings also emerged. To begin with, we found that children’s performance on the morphological relatedness task, which addressed surface structural analysis, was less, albeit significantly, correlated with meaning inference as compared to the performance on other morphological awareness tasks. This seems to suggest that at upper primary and probably higher levels as well, deriving the meaning of an unknown complex word may be less determined by structural sensitivities; instead, knowledge of the meanings of constituent morphemes and the functions of affixes may matter more. Such an observation seems also derivable from Nagy et al. (1993) where some students at the middle grades failed to pay attention to how a suffix modified the meaning of an unknown word – they relied heavily on the base to decide the meaning of the whole word, despite their presumably high level of morphological segmentation ability.

Another finding worth noting is the strengthening of relationship between morphological awareness and lexical inference. Logic suggests that with more refined morphological knowledge (e.g., knowledge of more affixes, more in-depth understanding of the functions of suffixes, stronger representation of morphemes and their structural connections in the mental lexicon), learners are more likely to use morphological problem solving for new word learning, instead of relying largely on such vocabulary strategies as rote memory or mnemonics that seem to be more typical of early primary years. Reciprocally, with their increasing use of morphological analysis for inferring meanings of unknown derivatives as they progress to upper primary, children are processing derived words more frequently and deeply. As a result, their morphological knowledge and representation of morphological information in the mental lexicon is likely to become stronger, which further contributes to their lexical inference success. Therefore, the strengthening of the relationship between morphological awareness and lexical inference should be expected. However, the small number of previous studies failed to address this issue directly as they were largely cross-sectional and focused only on one group of students or a particular period of development (e.g., Zhang 2013; Zhang and Koda 2012). Bowers and Kirby (2010), though a longitudinal study, only pretested children’s oral vocabulary knowledge and posttested their base identification and morphological vocabulary, and consequently, there was no way to compare the relationship between the measured competencies between pre- and posttests. In this regard, the present study has provided empirical support for how the strength of the relationship between morphological awareness and lexical inference could be changing longitudinally.

Conclusions

This chapter reported some preliminary findings of a semester-long morphological intervention study on Grade 4 children in Singapore, focusing on the development of derivational awareness and word learning ability and the relationship between these competencies. Overall, the intervention accelerated the development of morphological competencies, although the effect size of the intervention was small. In addition, correlation-based analyses showed that the relationship of morphological awareness to lexical inference became stronger longitudinally, which was true of the intervention as well as the control groups.

In conclusion, we would like to point out that morphological analysis is by no means the only way for inferring meanings of unknown words. The study reported in this chapter focused only on morphological analysis of isolated derivatives, because our interest lay in how morphological awareness is related to the ability to use intra-word clues to infer word meanings. This chosen focus was not to downplay other means for lexical inferencing. Incidental learning of words draws upon structural as well as contextual analysis (Nagy and Scott 2000). Good vocabulary programs should not be restricted to morphological strategies; other strategies, such as contextual analysis and direct teaching of word meanings, should also be integral components.

The above argument seems particularly reasonable if the needs of lexically poor learners are considered with reference to the intricate relationship between morphological awareness, lexical inference, and vocabulary knowledge. While morphological awareness contributes to lexical inference, which eventually leads to a larger vocabulary, morphological awareness itself relies on large amounts of lexical processing for refinement. Consequently, the larger a learner’s vocabulary is, the more refined his/her representation of morphological units. In addition, successful lexical inference also relies on adequate vocabulary knowledge, such as root word knowledge. As Nagy (2009) pointed out, “the students with the smallest vocabularies are least likely to make the generalization from the root word to its prefixed and suffixed relatives” (p. 485). Thus, to help lexically poor students get out of the vicious cycle – smaller vocabulary, poorer representation of morphemic units, less able to use morphological analysis to learn new words, and subsequently, less able to increase their vocabulary breadth – vocabulary instruction should not stop at introducing new affixes. Instead, it should go hand in hand with other strategies, such as strategic teaching of carefully selected words and making the classroom a word-rich environment (Beck et al. 2002; Nagy 2009).