Exploring the effects of teaching word families on the vocabulary learning of L1 Chinese EFL adolescents: A mixed-methods study

1 Lexical units in pedagogy

The concept of lexical units is critically important in L2 learning, particularly for vocabulary acquisition. The three often discussed types of lexical units are word types, lemmas/flemmas, and word families. Each word type encompasses a distinct word form (see Laufer, 2021). For instance, although they share the same base word, “act,” the words “act,” “acting,” “acted,” “actor,” “action,” and “activity” each belong to a different word type. A lemma is a more extensive lexical unit comprising a base word and all of its inflections (which are all the same part of speech). For instance, “act” (as a verb), “acting,” and “acted”, when used as verbs, are part of the same lemma. Furthermore, the flemma, another related unit, disregards distinctions among the parts of speech; for instance, “act” (as a noun), “acts” (as a noun), “act” (as a verb), “acted,” and “acting,” among other forms, constitute one flemma. Finally, the most comprehensive lexical unit is a word family, which encompasses a base word and all of its inflected and derived forms. Thus, the words “actor,” “action,” “activity,” “inaction,” and “react” could be considered members of the “act” word family

Previous research has largely focused on using word families and lemmas as units for counting vocabulary. For example, studies often calculate the frequency of exposure to target words either by treating inflected forms as separate word types or by counting repeated occurrences of the same type (Reynolds & Wible, 2014). However, debate continues regarding which counting unit is most appropriate for lexical instruction. Concerning word families, earlier studies debated whether learners possess sufficient morphological knowledge to effectively employ word families as lexical units for learning and assessment (Brown, 2013; Brown et al., 2020; Laufer et al., 2021; McLean, 2018; Snoder & Laufer, 2022; Ward & Chuenjundaeng, 2009). The concept of the word family has become significant in vocabulary testing because it is assumed to reflect learners’ receptive word knowledge. If learners possess base words and relevant morphological patterns, they should be able to extrapolate their understanding to encompass derived and inflected words (Brown, 2013; Brown et al., 2020; McLean, 2018). However, skepticism persists regarding learners’ ability to connect derived and inflected words with their corresponding base words (e.g., Laufer et al., 2021; McLean, 2018; Snoder & Laufer, 2022; Ward & Chuenjundaeng, 2009). If this association is elusive, utilizing the word family as the foundational unit for assessing vocabulary size might result in an overestimation of learners’ vocabulary knowledge (Webb, 2021).

The question arises as to whether L2 learners possess a sufficient reservoir of vocabulary knowledge to enable them to deduce derived and inflected words. Research examining learners’ grasp of inflectional and derivational word knowledge has yielded divergent outcomes. Some studies have found that intermediate and advanced L2 learners exhibit the morphological knowledge necessary to comprehend word families.

Snoder and Laufer (2022) demonstrated that both intermediate and advanced adolescent Swedish EFL learners grasp derived words. These learners achieved an accuracy rate of over 90% in identifying derived words related to 80 base words, which represented the most commonly occurring 8,000 word families of English. Similarly, Laufer et al. (2021) found that the advanced L1 Hebrew EFL learners’ knowledge of derivational words was comparable to that of base words. In research conducted by Laufer and Cobb (2020), the occurrence of affixes was found to be relatively infrequent in texts, ranging from approximately 3% to 7% of derived word coverage in a variety of texts, implying that mastery of a select number of affixes could sufficiently facilitate comprehension.

Consequently, although learners may possess a limited understanding of affixes, the need for in-depth morphological knowledge may be less pronounced. However, a substantial body of evidence underscores learners’ shortcomings in terms of morphological knowledge and their capacity to deduce inflectional and derivational word forms. For example, Reynolds (2015) revealed that frequency of occurrence has a beneficial impact on the incidental acquisition of words among L1 Chinese EFL learners through reading. This effect was observed for words that maintained a consistent form or varied through inflectional changes in the reading contexts. However, this positive effect was not apparent for derivationally varied words, which suggests that despite possessing a wide range of L2 vocabulary sizes—from 2,700 to 13,500, with an average of 8,700, which is nearly comparable to the vocabulary size of doctoral students (Nation & Begler, 2007)—the L1 Chinese EFL learners in that study encountered difficulties in linking words with derivational variations, especially when exposure to L2 input was restricted, suggesting limitations in the effect of increased word exposure.

Similarly, Ward and Chuenjundaeng (2009) discovered that Thai university EFL learners encountered challenges in understanding derived words despite their familiarity with the corresponding base words, as revealed by tests that measured their knowledge of base and derived words. The participants correctly identified less than 50% of the corresponding derived words compared with their recognition of the base words. This imbalance in learners’ morphological knowledge was supported by McLean (2018), who found that Japanese EFL learners, ranging from beginner to advanced levels, experienced difficulties recalling the written receptive meanings of both base and affixed words, even when provided with sentential contexts. Similarly, Brown (2013) identified a comparable interaction between frequency and word-form variation in the word acquisition of low-intermediate Japanese EFL learners. Furthermore, Brown’s (2013) research revealed that L2 learners lacked the ability to recognize inflected and derived words with higher frequencies (e.g., the 1,000, 2,000, and 3,000 most frequent words in a corpus). Moreover, when considering words of lower frequency (e.g., the 5,000 most frequent words in a corpus), learners struggled even with the comprehension of base words.

These divergent findings might be attributed to factors such as L2 proficiency, L2 vocabulary size, morphological knowledge, and linguistic distance between learners’ L1 and L2. For example, nuanced consideration involves selecting an appropriate counting unit based on the proficiency level of L2 learners and the linguistic features of their L1s. When assessing receptive word knowledge, it is advisable to employ smaller counting units, such as word types designated for beginning learners; using larger units such as word families is more suitable for evaluating the receptive word knowledge of intermediate to advanced learners. This viewpoint is supported by Ward and Chuenjundaeng (2009), who found that the limited understanding of correspondingly derived words among Thai EFL university learners was partly attributed to their lower levels of L2 proficiency and linguistic distance between Thai and English. Similarly, in Reynolds’s (2015) study, intermediate to advanced EFL university learners faced the problem of inferring unknown derived words from successfully guessed base words. This implies the learners have difficulty benefitting from the frequency of exposure effect, which may be attributable to the differences between their Chinese L1 and English L2. The influence of L1 distance seems consistent among Japanese EFL learners irrespective of their L2 proficiency (Brown, 2013; McLean, 2018). In contrast, Snoder and Laufer (2022) demonstrated that Swedish L1 EFL learners were proficient in identifying and comprehending both inflectional and derivational variations in English, suggesting that the lower linguistic distance between Swedish and English plays a role.

Considering these factors and the suitability of lexical units, the present study aligns with the approach of selecting word families as teaching units. As the participants consist of lower L2-proficiency-level EFL learners whose L1 Chinese differs considerably from their L2 English, this approach enables us to understand whether using word families as the pedagogical unit for vocabulary teaching would be suitable for lower proficiency learners of English. Furthermore, as few studies have investigated the validity of this teaching approach, this study responds to Brown et al.’s (2021) and Webb’s (2021) calls for employing larger lexical units in instruction because learners require scaffolding in this facet of word knowledge development. Drawing on the aforementioned evidence, we adopted an explicit teaching approach that targets words learners encounter within intensive reading texts along with all of their associated forms. This approach echoes Kremmel’s (2021) suggestion to explore the legitimacy of “presenting headwords together with their inflections and derivations” (p. 963) in fostering word knowledge development and its practical application in real-world classrooms.

2 Morphological knowledge and vocabulary acquisition

An understanding of how learners internalize relationships within word families requires attention to the role of morphological knowledge. Morphological knowledge refers to learners’ understanding of how words are constructed from meaningful components such as bases, prefixes, and suffixes, and how these components contribute to the semantic and grammatical properties of complex words (Carlisle, 2000; Kuo & Anderson, 2006). Nation (2022) argues that morphological relationships play a crucial role in vocabulary learning because they facilitate both within-family and between-family word knowledge. Within-family learning occurs when understanding a base word strengthens recognition of inflected or derived forms, whereas between-family learning enables learners to generalize knowledge of frequent affixes or derivational processes to new words. According to Nation (2022), such generalization is essential for expanding learners’ vocabulary beyond what is explicitly taught or encountered frequently.

A substantial body of research has demonstrated that morphological knowledge makes a significant contribution to vocabulary growth across developmental stages and proficiency levels. In L1 contexts, children have been shown to rely heavily on morphological decomposition to acquire new lexical items from print (Anglin, 1993; Nagy et al., 1989). Experimental studies also indicate that readers who are more sensitive to the internal structure of words are better able to infer the meanings of novel morphologically complex items, particularly when the morphological relationships are transparent (Nagy et al., 1993; Tyler & Nagy, 1989). Longitudinal evidence further reveals that morphological awareness predicts vocabulary growth independently of decoding, suggesting that knowledge of word parts provides learners with analytic tools that support lexical expansion and reading comprehension (Deacon & Kirby, 2004; Nagy et al., 2006). Among multilingual learners in L1 English contexts, morphological awareness has also been shown to predict vocabulary development more strongly than lower-level reading skills such as word decoding or reading fluency (Kieffer & Lesaux, 2007, 2012).

In L2 contexts, morphological knowledge remains influential but often develops more slowly. Research shows that L2 learners frequently have difficulty recognizing the relationships between base words and their inflected or derived forms, especially when derivational morphology is less transparent or when the learners’ L1 differs structurally from English (Schmitt & Meara, 1997; Zhang & Koda, 2013). Studies have also found that L2 readers draw on morphological information when interpreting unfamiliar morphologically complex words, but that their use of such information is highly dependent on L2 proficiency and the regularity or transparency of the morphological pattern (Hayashi & Murphy, 2011; Jean & Geva, 2009). These findings complement Nation’s (2022) observation that although morphological cues can substantially support vocabulary development, learners may need repeated exposure and explicit attention to productive affixes and frequently occurring patterns in order to internalize them effectively. When morphological awareness is underdeveloped, learners may struggle to take advantage of the learning opportunities available in morphologically related forms, particularly those that are semantically opaque or irregular.

Taken together, the broad body of research demonstrate that morphological knowledge forms a crucial cognitive basis for vocabulary development. It enables learners to analyze word structure, infer the meanings of unfamiliar forms, and construct relationships among members of a word family. This foundation aligns naturally with instructional approaches that target the formal properties of words, such as base words and their associated inflectional and derivational forms.

3 Theoretical framework

The theoretical basis for teaching word families is rooted in processing resource allocation (TOPRA; Barcroft, 2000). This theory is grounded in a comprehensive understanding of human memory, particularly the concept of transfer-appropriate processing (TAP), which posits that memorization of an item is influenced by the nature of the task during both the study and testing phases (Morris et al., 1977). TOPRA pertains to the cognitive processes through which learners allocate resources to acquire and comprehend words and phrases. Learners can only manage a limited range of inputs, and the resulting learning outcomes are expected to be more pronounced in the specific aspect of the input that receives their attention.

Inspired by this theory, Barcroft (2002) undertook two practices: structural and semantic elaboration. The former involves enhancing the comprehension and retention of new words by organizing them within a coherent framework or structure, such as phonemic or graphemic structures. The latter process solidifies the understanding and retention of new words along with their meanings. Activities related to structural elaboration include mnemonic exercises such as copying a word, counting letters, and employing the keyword technique. In contrast, tasks targeting semantic elaboration focus on understanding, recalling, or using word meanings, including mental imagery, synonym generation, definition-matching, usage-example matching, and the utilization of new words to craft sentences.

As a result, semantic elaboration aids in learning the semantic properties of new words, whereas structural elaboration contributes to understanding their structural properties. For example, Barcroft (2000) compared the effects of semantic elaboration and word picture repetition on word learning among L2 learners. Barcroft revealed the inhibitory impact of semantic elaboration on productive word-form knowledge, indicating that semantic elaboration occupies processing resources that should be directed toward acquiring new word forms. This finding was corroborated by Barcroft (2002), who compared the effects of semantic (pleasantness ratings) and structural elaboration (counting letters in words).

His results indicate that English-speaking Spanish L2 learners achieved higher scores in both free and cued recall of targets in L2 Spanish when exposed to structural elaboration as opposed to semantic elaboration, which hindered L2 word-form learning. This learning effect has lent evidence to the TOPRA model, as the recall of the L2 Spanish form necessitates memory of formal properties, whereas L1 English requires memory of meaning.

Based on this theory, the present study employs the teaching of word families as a starting point for our exploration. Our research aims were twofold:

1. We aimed to determine whether teaching word families can facilitate the word knowledge of adolescent L1 Chinese EFL learners.

2. We sought to explore adolescent L1 Chinese EFL learners’ perceptions of structural elaboration in word families.

To achieve these aims, we recruited two learner groups: one received structural elaboration of word families; the other received the semantic elaboration of individual word forms, serving for baseline comparison. Accordingly, the following research questions were posed:

1 Research participants

We recruited two intact classes, which were selected via convenience sampling, comprising a total of 94 Taiwanese 10th-grade EFL learners as study participants. Participants and their parents/guardians received detailed information regarding the research and provided consent prior to participation. Although optional for some, participation was mandatory in certain cases due to curriculum requirements, with alternative activities offered for those who opted out. The word-families group included 45 students (33 females; age: M = 15.56_years, SD = .54); the individual-word-forms group included 49 students (37 females; age: M = 15.84_years, SD = .42). Prior to the experiment, participants were required to complete the reading section (maximum = 120) of the General English Proficiency Test (GEPT), which is used to measure adolescent Taiwanese EFL learners’ L2 reading abilities. L2 reading ability is found to be correlated with L2 vocabulary learning (Qian, 1999; Zhang & Zhang, 2022) and thus was selected as an indicator of their L2 ability to acquire vocabulary.

Analysis of the students’ GEPT scores indicated that the word-families group (Mdn = 102) and the individual-word-forms group (Mdn = 95.2) had comparable L2 reading abilities (U = 913, p > .05, r = -13.96) prior to participating in the study lessons. Non-parametric tests (Mann–Whitney U test) were used because of the non-normal distribution of the two datasets of L2 reading abilities.

Both participant groups were taught by the same teacher, who holds a doctorate in education. She is an experienced teacher with over 27 years of experience teaching EFL reading to 10th graders and over 30 years of teaching at the school where the intervention took place.

2 Intervention

The teaching intervention was implemented during a semester-long intensive English reading course that lasted 18 weeks. This intensive reading course followed the textbook curriculum (Anderson, 2012). The course instructions and textbook were designed to activate students’ prior knowledge, cultivate their English vocabulary, encourage them to reflect on the meaning of words and apply strategies, increase their reading fluency, and evaluate their progress. Each unit comprises 2 chapters that featured intensive text reading, and 5 of the textbook’s 12 units (10 chapters) were used in this study. The teacher applied the exact curriculum from the textbook with no modification of the material, following it precisely with the only modification being the instructional approach to the targeted vocabulary, which is described later in this section.

Students engaged in pre- and post-reading exercises. Pre-reading activities were typically performed in pairs, while many post-reading exercises served as homework assignments. During the reading stage, the teacher played a more active role in helping students understand the sound–spelling relationships, unfamiliar vocabulary, key grammatical constructions, grammatical signals for text cohesion, organization of ideas, strategy development, and detailed comprehension of the text (Nation, 2013).

Course instructions were mainly delivered in Chinese and supplemented with English (approximately 15% of the class time was conducted in English). The teacher met the students 4 times a week for 50-minute lessons over 18 weeks. The course and instructions were identical for the two groups, except for the teacher’s instructional treatment for the target words. For each target word appearing in the text, instruction began with the same initial steps for both groups. The teacher first read the sentence containing the target word and then defined the words in English and Chinese as used in the text, noting the part of speech. This information had also been written on the board, and the students were required to record it in their notebooks.

After these identical initial steps, the instructional approaches diverged. For the word-families group, after defining the target word in context, the teacher introduced all related inflected and derived forms of the target words found in the dictionary.com entry (See Table 1 in Appendix 1 of the Supplemental material). These forms were written on the board and visually grouped to highlight shared components. The teacher explicitly explained the morphological relationships, shared roots and affixes, and similarities and differences in meaning and grammatical function. Students wrote down these forms in their notebooks to reinforce the idea of word families. Because this expanded instruction sometimes required more time, the teacher tracked the vocabulary-instruction time for the individual-word-forms group carefully to ensure parity across groups. For the individual-word-forms group, after the shared step, instruction focused only on the exact word form that appeared in the text. The teacher provided additional examples using the same word form, illustrating how meaning varied with contexts or connotations and clarifying nuances of usage. All examples were written on the board, and students recorded them in their notebooks. Unlike the word-families group, this group did not receive any instruction on related morphological forms (e.g., inflections, derivations). Instruction remained narrowly focused on the specific form encountered in the text.

OPEN IN VIEWER

Table 1.

Independent samples t-tests for the word meaning recall test scores differences between the word-families group and the individual-word-forms group in the pretest and posttest.

	Word-families group(n = 45)		Individual-word-form group(n = 49)		t	df	p
	M	SD	M	SD
Pretest scores	33.73	16.926	34.33	17.997	.164	92	.870
Posttest scores	69.24	17.966	53.59	21.479	-3.814	92	.000

Note. Maximum score = 100.

Take the teaching of the target word “created” in the sentence “She created a piece of art” as an example (which is also found in the textbook). Both groups first received the definition, translation, and part of speech. The teacher read the sentence and provided both the Chinese and English definitions of “create” as “創造” and “to evolve from one’s own thought or imagination, as a work of art or an invention.” The teacher also noted the part of speech as a verb. For the word-families group, the teacher subsequently taught the meaning of all inflected and derived forms that appear at http://www.dictionary.com/, which includes create (verb), and related inflectional (created, creating) and derivational forms including creatable (adjective), intercreate (verb, used with an object), self-creating (adjective), and uncreatable (adjective). Dictionary.com was selected after comparing its entries with those of other well-known online dictionaries (Pikilnyak et al., 2021) and determining that it provided the most comprehensive coverage of inflected and derived forms. The teacher also emphasized the relationships among the distinct yet interconnected forms of the word by highlighting the similarities among the various forms and underlining the common components. All related forms were also presented visually on the board, and students connected them in their notebooks. For the individual-word-forms group, the teacher instead provided sentential examples that involved the exact form of “created” such as “according to legend, they were created in 1959 as a panicked reaction to the popularity of rock ’n’ roll” and highlighted subtle meaning differences based on context. Similarly, the teacher instructed the students to write down the word, the Chinese translation, and any other notes they wanted to take.

3 Target word selection and assessment

Fifty words taken from the textbook used in the intensive English reading course were selected as target words. The course teacher was responsible for selecting the target words to ensure the pedagogical and ecological validity of the study conducted in the intensive reading classes. The teacher employed specific criteria to select the target words for this study, focusing on the general vocabulary that would be applicable at the students’ level. The chosen words encompass various domains, including cooking, sports, personal life, and education. The teacher aimed to provide practical language skills by selecting words commonly encountered in everyday communication. Consideration of the students’ curriculum and teaching materials played a crucial role in ensuring the relevance and appropriateness of the selected words. By adhering to these criteria, the teacher sought to facilitate the students’ overall language proficiency and enable them to apply the acquired vocabulary in diverse contexts. These 50 words were presented in their base, inflected, or derived forms for the as-usual group (individual-word-forms group) and all forms introduced to the experimental group (word-families group) were documented (see Appendix 1). Across the 50 target words, the students receiving the structural elaboration instruction (i.e., word-families group) were exposed to a median of 7 forms per target word, with the number of forms ranging from 0 to 16.

Fifty words were assessed according to their base word forms. The base forms were used in the assessment to examine whether learners, particularly the as-usual group (individual-word-forms group) who encountered certain items only in inflected or derived forms in the textbook, could recognize the underlying base word. This approach also enabled us to investigate learners’ ability to make morphological inferences, an issue previously identified as challenging in L2 vocabulary research (Ward & Chuenjundaeng, 2009). The target words selected for this study encompassed a range of parts-of-speech categories, including nouns, verbs, adjectives, and adverbs. In cases in which a word could be considered both a verb and a noun, it is important to note that only the specific part of speech targeted in the textbook was focused on for assessment purposes. The average target word length was 6.8 letters; however, individual word length varied, with some words comprising 4 letters and others up to 15. The average syllable count for the target words was approximately 2.4, although some words comprised one syllable and others up to five. These characteristics reflect the diverse linguistic structures represented by the selected target words that were used in the course textbook.

Meaning recall of the 50 target words was assessed in an L2-to-L1 translation test in which students were provided with the L2 form in isolation and asked to write the L1 Chinese form. This receptive translation test, in which students provided the L1 form in response to the L2 cue, was designed to assess receptive knowledge of both form and meaning. In this task, learners were assumed to access the meaning of the L2 form and then retrieve the corresponding L1 equivalent (Webb, 2009). The pretest and posttest used the same L2-to-L1 meaning-recall format, and both assessments contained the same 50 target items. Each correct response was awarded 2 points, resulting a total score of 100 points. The predefined answer key was created collaboratively by the course teacher and a research assistant prior to scoring. The textbook meaning was adopted as the primary correct answer. In cases where the textbook meaning did not fully align with dictionary definitions, acceptable answers were broadened to include semantically relevant meanings judged to be conceptually equivalent. Correct part of speech was not required, as the test assessed learners’ receptive form–meaning connection rather than grammatical classification. Two test-graders (the course teacher and a recruited research assistant) independently marked each of the answer sheets using a predefined answer key; the inter-rater agreement between them was greater than 99.9 %.

4 Questionnaire

The 14-item questionnaire given to the participants in the word-families group inquired about their perceptions of the word-family instruction. Participants responded on a 6-point Likert scale from 6 (strongly agree) to 1 (strongly disagree). Questionnaire items were categorized into three sections.

Questions 1 through 10 relate to the usefulness of the instruction on word families for the acquisition of various aspects of word knowledge. Questions 11 and 12 concern whether the selection of target words should be initiated by the learner or the teacher. These items were included at the request of the classroom teacher to gather pedagogical feedback, but because they do not directly relate to the research questions of the present study, their results are not discussed further. Questions 13 and 14 concern the participants’ attitudes toward the instruction of word families. In addition to the 14 closed-ended questions, 2 open-ended questions were included to encourage the participants to express their thoughts, emotions, and suggestions regarding the instruction of word families. The complete questionnaire is presented in Appendix 2.

5 Procedures

Prior to the experiment, both the word-families and individual-word-forms groups were asked to complete the reading section of the GEPT to assess their reading proficiency levels, as well as the L2-to-L1 translation pretest. When the intensive English reading course started, the word-families group received instructions regarding the word families of target words selected from the textbook (Anderson, 2012); in contrast, the individual-word-forms group received instructions on the forms of the target words as appearing in the textbook. At the end of the intensive English reading course, both groups completed the L2-to-L1 translation posttest, and the participants in the word-families group were asked to complete the questionnaire described in Section III.4.

6 Data analysis

To answer the first research question, regarding whether structural elaboration for teaching word families would lead to different learning outcomes of the acquisition of the target words when compared to teaching target words only through semantic elaboration, we first analyzed the pretest and posttest scores for all 50 target words for both the word-families group and individual-word-forms group. The normality of the pretest and posttest scores was assessed using both statistical and visual methods. Quantile-quantile (Q-Q) plots indicated approximately normal distributions, and the skewness values were all below 1 (Larson-Hall, 2009). Therefore, parametric analyses were conducted. Independent sample t-tests were run to compare the statistically significant differences in target word meaning recall assessment scores between the word-families group and the individual-word-forms group on the pretests and posttests.

A further examination of the target words revealed that 12 of the 50 items appeared in the textbook only in either inflected or derived word forms and were assessed in their base word forms (see Appendix 1). This provided an opportunity for a second, more fine-grained analysis focusing specifically on these 12 items. Given Ward and Chuenjundaeng’s (2009) finding that L1 Thai EFL learners were more likely to learn derived words from base words than to infer base forms from derived forms, we used this subset to investigate whether adolescent L1 Chinese EFL learners exhibited similar patterns. Specifically, we examined whether learners in the two instructional conditions were able to associate the inflected or derived forms they encountered during instruction with the corresponding base forms used in the assessment.

For the subset of 12 items, visual inspection of Q-Q plots suggested violations of normality (Larson-Hall, 2009); hence, non-parametric analyses were conducted. We used Mann–Whitney U tests to compare the mean differences between the word-families-group’s and individual-word-forms group’s posttest scores for the 12 target words (n = 12). The Wilcoxon signed rank test was used to compare the mean differences between the pretest and posttest scores of the individual-word-forms group.

To ensure that any advantage observed for the word-families group was not simply due to the explicit instruction of the base forms of the 12 target items during instruction, forms that the individual-word-forms group did not encounter, the present study also conducted a separate analysis focusing on the remaining 38 items that were taught and tested in the same base forms. Visual inspection of Q-Q plots suggested violations of normality (Larson-Hall, 2009); therefore, Mann–Whitney U tests were used for between-group comparisons and Wilcoxon signed rank tests for within-group pre-post changes for this subset of items.

To address the second research question, regarding the word-families group, participants’ perceptions of the structural elaboration of word families’ instruction and their responses to the questionnaire were tallied and coded. The data collected from the open-ended questions were analyzed through two cycles of coding developed after following guidelines suggested by Miles et al. (2014).

The first cycle involved a combination of inductive and deductive coding. Deductive coding allowed us to generate a list of codes based on the research question, whereas inductive coding allowed for the identification of emergent codes. For example, chunks of data related to participants’ positive or negative attitudes toward note-taking of word families or the structural elaboration of word families were generated by deductive coding, whereas suggestions from the participants on improving the instruction were assigned using inductive coding. The code “positive attitudes toward note-taking” was applied to a word-families group participant’s response: “I like the way of taking notes (on word families).” The code “negative attitudes toward note-taking” was applied to another participant’s response: “The workload is heavy and challenging.”

In the second coding cycle, the codes generated from the first coding cycle were grouped into two major themes: the usefulness of the word-family instruction and attitudes toward the instruction. For example, “positive attitudes toward note-taking” and “negative attitudes toward note-taking” were categorized under the theme of “general attitudes toward note-taking.”

1 Independent samples t-tests

Descriptive statistics for the meaning recall assessment scores of both the word-families and individual-word-forms groups were calculated. As reflected in Table 1, the independent samples t-tests showed no statistically significant difference in the pretest scores between the word-families group and the individual-word-forms group (t(92) = .164, p > .05). However, a statistically significant difference in the posttest scores was found between the two groups (t(92) = -3.814, p < .001), with Cohen’s d = .79, indicating a medium effect size (Plonsky & Oswald, 2014). This suggests a meaningful difference between the two groups’ posttest scores.

These findings suggest that the word-families group, which received the structural instruction of word families, demonstrated better acquisition of the target words in base word forms compared to the individual-word-forms group, which received the semantic instruction of the target words as they appeared in the text regardless of their form (e.g., base word, inflected, or derived form).

2 Non-parametric test results

To further explore the acquisition of 12 out of the 50 target words that appeared in different forms in the textbook (i.e., inflectional and derivational forms) and on the test (base word forms), comparisons using Mann–Whitney U tests revealed no significant difference in the meaning recall pretest scores between the word-families group (Mdn = 6) and the individual-word-forms group (Mdn = 7.92), U = 937, p > .05, r = -.129. However, the word-families group (Mdn = 13.92) had significantly higher meaning recall posttest scores than the individual-word-forms group (Mdn = 7.92), U = 1704.50, p < .0001, r = .470. A Wilcoxon signed-rank test indicated that the meaning recall posttest scores of the individual-word-forms group (Mdn = 7.92) were significantly higher than their meaning recall pretest scores (Mdn = 7.92), Z = 187.00, p < .05, r = -.301. Although the pretest and posttest medians were identical, the Wilcoxon signed-rank test detected a significant improvement because many learners scored higher on the posttest, resulting in a systematically positive distribution of paired differences. Thus, the results suggest that adolescent L1 Chinese EFL learners who received semantic instruction on target word forms (i.e., inflectional or derivational) may have developed a limited ability to associate them with the base words.

To further examine whether the advantage of the word-families group was attributable to the 12 items introduced in inflected or derived forms, an additional analysis was conducted on the 38 items that were taught and tested in identical base forms. Mann–Whitney U tests revealed no significant difference in the 38-word pretest scores between the word-families group (Mdn = 26) and the individual-word-forms group (Mdn = 24), U = 1102.00, p = 1.00, r = .00, indicating baseline equivalence. However, the word-families group (Mdn = 58) had significantly higher 38-word posttest scores than the individual-word-forms group (Mdn = 42), U = 707.00, p = .0028, r = .31. A Wilcoxon signed-rank test showed that the posttest scores of the individual-word-forms group (Mdn = 42) were significantly higher than their pretest scores (Mdn = 24), Z = 1.00, p < .001, r = .61, indicating learning gains within the group. The word-families group also demonstrated a significant pre–post improvement (Mdn_pretest = 26; Mdn_posttest = 58), Z = 0.00, p < .001, r = .67. Taken together, these results show that the instructional advantage observed for the word-families group extends beyond the 12 items taught in derived or inflected forms, and is also evident for the 38 items consistently presented as base forms in both instruction and assessment.

3 Questionnaire results

To complement the quasi-experimental findings reported for the first research question, the questionnaire data were analyzed to address the second research question by providing insight into learners’ perceptions of the structural word-family instruction. The word-families group responded positively to the structural instruction. Mean ratings across items ranged from 4.07 to 5.30 on a 6-point Likert scale, indicating moderately positive to highly positive perceptions. Items relating to the usefulness of word-family instruction for learning word meaning, part of speech, spelling, word association, and Chinese translation all received high ratings (M = 4.90–5.30, SD = 0.75–1.02). Participants also reported positive attitudes toward notetaking practices (Item 13: M = 4.57, SD = 1.12) and toward the instruction itself (Item 14: M = 4.39, SD = 1.20).

Qualitative data collected from the open-ended questions enhanced the descriptive statistics obtained using the Likert-scale items and offered deeper insight into how students perceived the structural instruction. The participants reached a consensus on the positive impact of learning vocabulary through word families. Some participants explicitly acknowledged taking notes on word families in word learning because of either their usefulness or their likeness (n = 16; 36%). As one participant noted: “Although taking notes is tiring, I still find it very useful” (S1). Similarly, some participants directly conveyed their general attitudes toward the instruction of the word families, describing it as good, interesting, and appealing (n = 16; 36%), with one remarking: “This is the first time a teacher has taught word-form variations, and I think it’s a very good method” (S18).

Forty-eight percent of the participants associated the effectiveness of the word-family instruction with rapid learning and enhanced memorization of a larger vocabulary size (n = 21; 48%). One participant explained: “It helps me memorize a large number of words at once” (S42). The expanded vocabulary size, as perceived by the participants, also reflected a deeper understanding of the multiple dimensions of word knowledge, such as parts of speech and word formation. The students’ responses to the open-ended questions further revealed that they attributed these positive effects on word learning through word families to their ability to associate words with related forms and constituent components (n = 9; 20%); for example, one student explained: “Seeing how words are connected helps me understand new words more easily” (S26).

Despite the benefits, some participants expressed mixed feelings, noting that taking notes on the word families had become burdensome despite recognizing its utility and their preference for it (n = 11; 25%). For example, one student stated: “Sometimes there are too many related words, and it becomes confusing when I review them” (S33). In addition, a smaller number of participants (n = 2; 5%) voiced doubts about the usefulness of certain low-frequency forms, such as those rarely encountered in real-life contexts: “I rarely have the chance to use some of these word variations” (S23).

Regarding suggestions for instructional improvement, the participants offered the following recommendations to refine the instruction: incorporate more extensive and diverse explanations to comprehend each word family, reduce the number of word forms presented, and organize their presentation in a more coherent manner (n = 11; 25%). This is illustrated by the comment “I hope related words can be presented in a more logical order—like nouns together, verbs together, adjectives together” (S13). Some students suggested refinements to the notetaking requirements (n = 4; 9%), such as reducing the number of word forms to be recorded or excluding forms that were infrequent or not pedagogically useful. As one student suggested: “We should exclude forms that are rarely used or not common in dictionaries” (S38). A set of coded qualitative responses and additional excerpts is provided in Appendix 3.

1 RQ1: Effects of teaching word families

Regarding our first research question, the study results revealed that the word-families group, which received structural instruction on words occurring in the text and all of their associated word forms, achieved significantly higher word meaning recall posttest scores than the individual-word-forms group, which received semantic instruction only on specific word forms that occurred in the textbook. The better learning outcomes of the word-families group align with Webb’s (2021) claim that larger lexical units are suitable for pedagogy. Additionally, our results provide support for our earlier discussion, which highlighted the relationship between L2 learners’ ability to comprehend derived words and their morphological knowledge and factors such as L2 proficiency, L2 vocabulary size, and L1–L2 distance. Lower L2 proficiency presents a more significant challenge for L2 learners, and a greater distance between the L1 and L2 leads to more difficulty in acquiring morphological knowledge and inferring derived word meanings (e.g., Brown, 2013; Laufer et al., 2021; McLean, 2018; Reynolds, 2015; Ward & Chuenjundaeng, 2009). Findings from the subset analysis of the 12 words offer empirical support for this position. Learners in the individual-word-forms group demonstrated only weak associations between the inflected or derived forms encountered during instruction and their corresponding base forms tested in the assessment. This aligns with prior research showing that L2 learners, particularly those who have an L1 with greater morphological distance from English, often struggle to map derived or inflected forms back to their base forms (e.g., McLean, 2018; Reynolds, 2015; Ward & Chuenjundaeng, 2009). The present findings therefore reinforce the argument that lower-proficiency L2 learners may not yet possess sufficient morphological awareness to benefit fully from derivational or inflectional cues without direct instruction of word families.

These studies suggest that word families might be debatable as lexical units for measuring receptive word knowledge because their adoption could overestimate vocabulary size. However, in terms of teaching, this study found that explicitly teaching word families through structural elaboration significantly increases word knowledge among adolescent L1 Chinese EFL learners. In contrast, teaching individual word forms alone yielded relatively poor learning outcomes, particularly for items that required learners to connect different morphological forms. This result can be explained by Schmitt et al. (2001), who suggested that learners are more likely to establish stable form–meaning links for base words when they encounter those words together with their inflected and derived forms. This further underscores the importance of providing direct instruction on word families to expand L2 learners’ word knowledge and enhance their ability to infer base-word meanings and affixes. Consequently, the necessity of explicitly teaching word families rather than individual word forms becomes more pronounced for learners with lower L2 proficiency whose L1 is typologically distant from their L2.

Similarly, the small but statistically significant improvement of the 12 target words from pretest to posttest in the individual-word-forms group (Mdn_pretest = 7.92, Mdn_posttest = 7.92), as indicated by a Wilcoxon signed-rank test, Z = 187.00, p < .05, r = -.301, suggests that adolescent L1 Chinese EFL learners were able to make some connections between the inflected or derived forms encountered during instruction and their corresponding base forms tested, although this ability appeared to be limited. This underscores the necessity of strengthening L2 learners’ word knowledge development through the direct instruction of word families when their L2 proficiency is low. Even with the semantic elaboration of target words alone, adolescent L1 Chinese EFL learners face challenges relating inflected/derived words to their base words, achieving correct responses for only approximately 50% of the target words. The teaching of word families in the current study has proven beneficial for word learning, especially for L2 learners whose ability to associate base words with their derived forms may be weak. It is important to note that classroom teachers, whose target learners exhibit low-to-low-intermediate L2 proficiency and come from L1 backgrounds distant from English, should not expect spontaneous utilization of this knowledge when students encounter unknown derived words. The current finding reinforces Reynolds’s (2015) observation that acquiring derivational words incidentally is challenging, highlighting the significance of explicit instruction on word variation to support more robust form–meaning connections. This finding also contradicts Ward and Chuenjundaeng’s (2009) observation about Thai EFL learners, as they occasionally had some derived word knowledge inferred from the base words but rarely learned base words from the derived words.

Regarding the effect of structural and semantic elaboration on word acquisition, the findings of this study indicate that structural elaboration has a more favorable effect than semantic elaboration on word meaning recall when L2 forms are introduced to elicit responses of L1 meanings from adolescent L1 Chinese EFL learners. The positive effect of structural elaboration on word learning suggests that—compared to simply consolidating the meanings of individual words—presenting variations of words while highlighting the stems may have a more pronounced influence on word acquisition for adolescent L1 Chinese EFL learners. Consequently, structural elaboration appears to play a constructive role in enhancing meaning recall. This implies that although the L1 Chinese EFL learners in the structural elaboration group received less input on word meanings, their focused attention on the target words could facilitate their ability to infer the meanings of the base words. This observation extends Barcroft’s (2002) study that demonstrated that the structural elaboration group tended to produce more positive learning results when eliciting L2 forms, indicating that structural elaboration can facilitate the acquisition of semantic meaning. This finding seems to contradict the TOPRA model (Barcroft, 2000), which posits that the structural elaboration group has limited resources, and once their attention has been allocated to some aspects of learning, learning of the unattended aspects would be weak.

One possible explanation for the advantage of structural elaboration lies in the development of learners’ morphological awareness and the strengthening of connections among related word forms. Previous research has consistently shown L2 learners to possess only partial productive knowledge of derivative forms within a word family. For example, Schmitt and Zimmerman (2002) found that learners typically knew some, but not all, members of a word family and often struggled particularly with adjective and adverb derivatives. Similarly, Iwaizumi and Webb (2021) reported that L2 learners produced significantly fewer derived forms than L1 speakers, with lower-proficiency learners demonstrating especially limited productive derivational knowledge. These findings suggest that morphologically related forms are not automatically acquired through exposure to a single base word and may therefore require explicit instructional support. From a broader literacy perspective, morphological awareness has been shown to make a unique contribution to L1 vocabulary development and reading comprehension beyond phonological skills (Nagy et al., 2006). Explicitly presenting base words together with their inflected and derived forms may promote deeper lexical organization by drawing learners’ attention to recurring stems and affixes. Rather than assuming that knowledge of one lemma entails mastery of its family members, a position questioned in debates concerning word families as lexical counting units (Laufer, 2021), the findings of the present study suggest that instruction can strengthen the connections among related forms and enhance learners’ ability to recall meanings. In this sense, the advantage of structural elaboration shown in the present study may reflect not only improved knowledge of individual word meanings but also greater morphological awareness that supports more efficient lexical access. It should be noted, however, that morphological awareness was not directly measured in the present study. Therefore, this interpretation should be examined more explicitly in future research that includes independent measures of learners’ morphological awareness.

The present study indicates that structural elaboration seems to lead to better learning outcomes for meaning recall (i.e., semantic knowledge) than does semantic elaboration. Therefore, this finding adds significant pedagogical value to teaching practice. The different results between the present study and previous research might be attributable to differences in research design. In Barcroft (2002), word knowledge measurement was either a free or cued recall of target words in L2 Spanish, with no linguistic cues provided to measure its effect on L2 word form. In contrast, the present study used a meaning recall test that involved multiple dimensions of word knowledge: recognition of L2 word form and production of L1 meaning.

2 RQ2: Perceptions about teaching word families

In general, most participants displayed positive attitudes toward the teaching of word families. This approach facilitates rapid learning and enhanced memorization by breaking down components and exposing interconnected words. This echoes Laufer et al. (2021), who claimed that teaching morphologically related words in either a direct or indirect manner would raise learners’ awareness of the word structure and enhance their ability to detect the meaning of familiar base words and affixes. It is noteworthy that the participants mentioned that the word-family approach leads to a better understanding of the parts of speech of target words, which, in turn, aids in word learning. This perspective is consistent with previous studies that highlighted learners’ difficulty in comprehending words used in different parts of speech (Stoeckel et al., 2020). Participants also emphasized the importance of parts-of-speech instruction and called for its inclusion, a request supported by previous findings that suggest derivation, which involves changes in the parts of speech, presents additional challenges in word acquisition (Reynolds, 2015). Consequently, teaching word formation rules is beneficial for word acquisition, as an improved understanding of word affixation contributes to expanding vocabulary size (Mochizuki & Aizawa, 2000). Thus, addressing this issue is considered pivotal for improving word learning.

Still, a substantial portion of the participants expressed some preferences that were not provided by the teacher’s instruction. These students requested a more logically structured presentation of word families and additional explanation of each form. Specifically, they asked for groupings by part of speech and a systematic method of presenting the related forms. These requests suggest that although structural elaboration can facilitate learning, its effectiveness may depend on the transparency of the organization of the forms. Presenting too many derivational forms simultaneously may overwhelm certain learners, particularly when the differences between the forms are subtle or would need to be used infrequently by those learners at their current stage of learning.

This suggests that students may benefit from word-family instruction that balances breadth with depth. Rather than introducing all possible derivations of a base word, teachers may need to prioritize high-frequency and pedagogically useful forms (Bauer & Nation, 1993) and then spend more time providing explanation for each. This type of scaffolding provided by teachers may help learners consolidate morphological awareness without experiencing the confusion reported by some of the participants.

3 Pedagogical implications

These findings have important pedagogical implications. First, the results suggest that introducing word families can support more effective vocabulary learning for learners with low to low-intermediate L2 proficiency, particularly when their L1 is typologically distant from English. Teachers might consider maximizing vocabulary learning by introducing word families instead of individual word forms, especially when learners have low-to-low-intermediate L2 proficiency levels and their L1 is typologically distant from English. Teachers could make a habit of mentioning other members of a word’s family when introducing a new word. This approach encourages L2 learners to consider inflections and derivations of words. To further reinforce this habit, teachers may ask students to guess a new word’s derivation at the time of its introduction. As suggested by TOPRA (Barcroft, 2000), the task and test affect word learning outcomes; incorporating a derivation section as part of the assessment also emphasizes the importance of learning a word family (Nation, 2013, 2022).

Second, considering the challenge of learning a multitude of word variations, one possible solution is to provide a relatively small and manageable set of core word families (Cobb & Laufer, 2021). These core families include a high percentage of word families and enjoy a high level of coverage in text (Brown et al., 2020). In addition, the frequency of word families has been shown to significantly affect L2 learners’ word knowledge (Laufer et al., 2021; Snoder & Laufer, 2022). Therefore, educators may opt for high-frequency word families over high-frequency affixations when selecting word families for classroom instruction.

Third, because the instruction of the word-families group followed the structural elaboration model (Barcroft, 2002), the participants expressed a desire for semantic elaboration. They suggested that future instructions should incorporate more extensive and diverse explanations to help the word-families group comprehend each word form. This indicates that a combination of semantic and structural elaboration may be necessary for positive learning experiences.

Fourth, considering that dictionaries rarely reveal the relationship between related words or provide adequate mnemonic techniques to facilitate memorization (Nation & Webb, 2011), teachers can demonstrate stem words and their inflections and derivations and provide mnemonic techniques, such as giving signs to indicate the core meaning of the stem. This will also encourage learners to take notes on word families, as most participants recognized the value of documenting word-form variations through notetaking.

4 Limitations and future studies

This study can be improved in the following ways. First, to enhance vocabulary learning through word families, this study aimed to create an environment as close as possible to ecological validity by allowing the classroom teacher to select target words directly from the course textbook, and related forms were identified through dictionary entries rather than corpus-based word-family lists. Although this approach mirrors authentic instructional practices, this random selection of words, on the one hand, may not adequately consider the connections between word families and their affixations to achieve optimal learning outcomes. For instance, some derived forms included in the instruction may have limited occurrences in corpora (e.g., creatable, intercreate), making them less useful for learners and potentially increasing their cognitive load. Future studies may therefore consider consulting multiple sources, such as Bauer and Nation’s (1993) hierarchy of seven levels of affixes for teaching purposes and corpus analyses of affix frequency (Sánchez-Gutiérrez et al., 2018), to serve as valuable references for including pedagogically meaningful and typical word-family groupings used in L2 vocabulary research. On the other hand, the classroom teacher’s selection of target words may have resulted in learners having partial prior knowledge of several items, as reflected in the pretest scores (approximately 33–34% out of 100%). This outcome is difficult to avoid in classroom-based research where ecological validity is prioritized, and target items are taken directly from a textbook rather than selected through researcher-led vocabulary profiling or needs analysis. Although the relatively high pretest scores suggest partial familiarity, the clear posttest gains indicate that many items were not fully mastered prior to instruction. Future studies may therefore incorporate a systematic vocabulary profiling procedure to select target words with more controlled levels of prior familiarity.

Moreover, previous research has shown mixed results regarding which frequency measure best accounts for learners’ receptive knowledge of derived words. Although some studies have found that word-family frequency, rather than the frequency of individual derived forms, is a stronger predictor of learners’ recognition of derivational variants (Laufer et al., 2021; Snoder & Laufer, 2022), more recent work by Pinchbeck et al. (2022) suggests that flemma frequency may in fact provide a better representation of learner knowledge than word-family frequency. These contrasting findings highlight the need for future studies to examine more closely how different frequency measures relate to the acquisition of derived words within classroom practice. Using a single online dictionary to identify related word forms may have biased the selection of derivational items. Future research should verify the pedagogical relevance and corpus frequency of derived and inflected forms across multiple sources before including them in instruction.

Second, incorporating multiple types of word tests would be beneficial. As indicated in the TOPRA (Barcroft, 2000), the type of test can significantly influence word learning outcomes. For example, structural elaboration should facilitate form learning, whereas semantic elaboration should contribute to meaning learning. In the current study, only meaning recall was assessed, which may not have provided a comprehensive view of learners’ vocabulary learning outcomes. Furthermore, as the participants mentioned that this approach benefited all aspects of word learning and enhanced memorization, it is advisable to consider multiple tests covering more aspects of word knowledge and conduct posttests at larger intervals. Moreover, future studies could incorporate an L2 vocabulary size test to assess participants’ existing lexical knowledge and facilitate comparisons with prior research. In addition, standardized measures of overall L2 proficiency, rather than single-skill measures such as reading, would provide a more comprehensive account of learners’ proficiency (Laufer et al., 2021).

Third, the experimental (word-families group) and as-usual group (individual-word-forms group) pretest and posttest design in this current study may have some limitations. For instance, the allocation of more vocabulary teaching time in the as-usual group was intended to address potential concerns regarding unequal learning time distribution/repetition of vocabulary between the two groups. However, this adjustment inevitably deviated from the as-usual teaching approach that the comparison as-usual group was meant to represent in real classroom contexts. Thus, future studies may adopt a counter-balanced design by exposing different target words across conditions (structural, semantic, and as-usual baseline comparison) in a more controlled setting for more robust ecological validity. Moreover, the novelty effect of the structural elaboration instructional approach itself warrants further consideration. Task novelty is widely identified as a strong predictor of flow, motivation, and positive emotions in L2 learning (Zuniga, 2023), which may temporarily enhance learners’ engagement and performance. Such designs, combined with multiple measurement points, would allow researchers to disentangle novelty-driven engagement from genuine instructional effects and provide a more robust assessment of structural elaboration as a pedagogical technique. Finally, the post-experiment questionnaire was administered only to the word-families group, which prevents direct comparison of learners’ perceptions across instructional conditions. Future research should include parallel or appropriately tailored questionnaire items for the as-usual group, enabling meaningful qualitative cross-group comparisons and providing more comprehensive insights into learners’ perceptions of different instructional approaches.

Fourth, with the growing use of psycholinguistic approaches to investigate L2 learning, it would be valuable to collect online processing data from participants when presented with various word forms in both isolated and contextual settings. This would help to uncover how attention is directed toward the different components of different yet related word forms, as concentrating on one linguistic aspect might impede attention to others (Son et al., 2022).