Effects of Practice Types on the Acquisition of English Phrasal Verbs

Abstract

English phrasal verbs are ubiquitous and challenging for second language (L2) learners, particularly for those whose first language does not have an equivalent structure. This study investigates the facilitative role of three distinct L2 practice types in promoting English phrasal verb acquisition. Eighty first-year college students from China were randomly assigned to three groups: the continuation group, which was first presented with an input text and then required to complete it; the retrieval group, which was first presented with the input text and then required to engage in retrieval practice; and the trial-and-error group, which was first required to engage in trial-and-error practice before reading the input text. The effectiveness of these practice types was compared via both an immediate post-test and a 1-week-delayed post-test. The results showed that in the immediate post-test, the continuation group performed comparably with the retrieval group but outperformed the trial-and-error group. However, in the 1-week-delayed post-test, the continuation group significantly outperformed the other two groups. The findings revealed that the continuation writing task not only initially equips learners with declarative knowledge and subsequently closely integrates static L2 learning with dynamic idea expression but also enhances learners’ task self-efficacy, thereby optimally promoting phrasal verb learning and retention.

1. Introduction

Practice plays a pivotal role in enhancing L2 learning, as evidenced by extensive research (DeKeyser, 2007; Ellis & Shintani, 2013; Suzuki, 2023). This is particularly relevant in contexts such as China, where limited exposure to authentic English and time-constrained instruction restrict learning opportunities. While various practice types exist, their comparative effectiveness for mastering linguistically challenging features, such as English phrasal verbs, remains underexplored. Phrasal verbs are notoriously difficult for L2 learners due to their idiomaticity and syntactic complexity (Garnier & Schmitt, 2015; Liu, 2011), yet they are essential for fluency.

Among the potential practice types, the continuation task (C. Wang, 2016) has shown promise for L2 acquisition. The xu-argument-based continuation task, which requires learners to read incomplete input texts and extend them creatively, coherently, and logically, has been proposed as a means to facilitate language learning and achieve high-efficiency language acquisition (C. Wang, 2016; C. Wang & Wang, 2015; Yang & Lu, 2021; Chen & Zhang, 2022; Tu et al., 2025). However, empirical evidence is lacking on whether the continuation task outperforms other practice types (e.g., retrieval practice or trial and error) for phrasal verb acquisition, and additionally, how learners perceive its effectiveness.

Traditional L2 practice research has predominantly focused on performance outcomes, often neglecting the subjective experiences of learners, such as their task self-efficacy, i.e., their beliefs in their capability to accomplish tasks (Bandura, 1986; Graham, 2007). This neglect is problematic because task self-efficacy not only predicts motivation and effort (Pajares, 2003) but may also mediate the effectiveness of different practice types. For example, learners who perceive a task as achievable (high task self-efficacy) may engage more deeply, thereby enhancing the benefits of the task (Mills et al., 2007). However, few studies have investigated how different practice types influence both phrasal verb learning and the task self-efficacy of learners. In the current study, task self-efficacy refers to participants’ beliefs regarding the effectiveness of a specific type of practice for mastering target English phrasal verbs.

To address these gaps, this study investigates the effectiveness of three practice types (the continuation writing task, retrieval practice, and trial-and-error practice) on L2 phrasal verb acquisition and the perceptions of learners regarding the effectiveness of these practices. The study’s findings will inform pedagogical strategies by identifying not only which practice type optimizes phrasal verb learning but also how the perceptions of learners mediate this process. This dual focus bridges the gap between the cognitive and affective dimensions of L2 practice, offering a more holistic understanding of effective L2 practice.

2. Literature Review

2.1. The Xu-Argument

Building on the premise that language acquisition flourishes through deliberate and meaningful practice, C. Wang (2016) proposed the xu-argument, which is a tripartite framework rooted in the Chinese concept of 续 (xu), which encompasses “completion,” “extension,” and “creation.” This framework posits that language development develops through iterative cycles of textual engagement and creative reproduction (C. Wang, 2017, 2020). The continuation task serves as a practical instantiation of this theory by presenting learners with incomplete input texts that require coherent, logical, and innovative completion while adhering to the linguistic patterns and thematic content of the source text (C. Wang, 2016; C. Wang & Wang, 2015).

Two complementary theoretical perspectives shed light on the cognitive mechanisms underlying continuation tasks. First, the Interactive Alignment Model (Pickering & Garrod, 2004) suggests that effective everyday conversation arises from the interaction and alignment between language users. C. Wang (2016) extended this model to writing contexts, proposing that learners engage in “creative imitation” (C. Wang, 2021) by integrating elements from source texts while producing original content, enabling the operationalization of receptive knowledge through productive tasks. Second, the situation model framework (Zwaan & Radvansky, 1998) explicates the cognitive demands of continuation tasks through its five-dimensional structure (spatial, temporal, causal, intentional, and communicative). Learners are required to construct and maintain mental representations that align with the source text during continuation writing. This process not only enhances lexical alignment and phraseological competence (Chen & Zhang, 2022) but also reduces L1 interference, which is a critical factor in second language development.

Empirical validation of continuation tasks shows substantial impacts across various linguistic domains. Controlled experiments have demonstrated enhanced acquisition of complex lexical items (Tu et al., 2025; Rassaei, 2017; X. Zhang, 2016), verb–noun collocations (X. Zhang, 2017), and grammatical categories (Jiang & Tu, 2016), as well as improved production of relative clauses (Q. Wang & Wang, 2018) and reduced morphological errors (C. Wang & Wang, 2015). Despite the demonstrated success across various linguistic subsystems, a significant gap remains in terms of phrasal verb acquisition, a key component of English proficiency due to the high frequency and semantic opacity (Gardner & Davies, 2007; Liu, 2011). This gap drives our current investigation into phrasal verb development through the continuation task.

2.2. Acquisition Challenges and Pedagogical Approaches for Phrasal Verbs

2.2.1. Linguistic Complexity and Challenges in the Acquisition of Phrasal Verbs

Phrasal verbs, defined as formulaic sequences comprising a verb and an invariant particle functioning as a unified syntactic and semantic unit (Liao & Fukuya, 2004), represent a critical yet problematic component of English proficiency. Their pedagogical significance stems from four key dimensions: (1) high frequency in authentic discourse, with one phrasal verb occurring per 150 words in spoken registers (Gardner & Davies, 2007); (2) semantic precision enabling nuanced communication (Cornell, 1985); (3) centrality to fluency benchmarks (Liu & Myers, 2020); and (4) lexical expansion potential through particle combinatorics (Garnier & Schmitt, 2015). Despite these functional imperatives, cross-linguistic studies reveal systematic underproduction among L2 learners, particularly in Asian contexts (Chen, 2013), with avoidance strategies favoring single-verb synonyms (e.g., “invent” vs. “make up”) that are well documented in interlanguage research (Dagut & Laufer, 1985; Siyanova & Schmitt, 2007).

The complexity of acquiring phrasal verbs stems from three interrelated factors. First, semantic opacity is a defining characteristic of many non-compositional phrasal verbs (e.g., “break up”, which means “to terminate a relationship”), requiring metaphorical interpretation similar to idiomatic processing (Bolinger, 1971; Gairns & Redman, 2011). Second, the combinatorial productivity of phrasal verbs imposes an exponential learning burden, as individual verbs (e.g., “take”) can combine with more than 20 particles, while common particles (e.g., “up”) collocate with hundreds of verbs. Third, the constituent salience asymmetry leads learners to prioritize verb semantics over particle functions, as evidenced by error pattern analyses that show higher rates of particle substitution errors (Strong & Boers, 2019b).

Drawing on the notorious difficulty of learning phrasal verbs, an increasing number of L2 coursebooks, as well as online resources for both learners and teachers, emphasize practice types designed to improve the acquisition of phrasal verbs in classroom settings (Strong & Boers, 2019a, 2019b).

2.2.2. Effectiveness of Retrieval and Trial-and-Error Practices for Lexical Acquisition

Research on test-enhanced learning has firmly established retrieval practice as a highly effective pedagogical intervention for promoting long-term retention (Roediger & Karpicke, 2006; Agarwal et al., 2021). This cognitive strategy, which involves actively recalling target information, consistently demonstrates greater efficacy than passive review methods in both first (Karpicke et al., 2014) and second language acquisition contexts (Barcroft, 2007). In a study by Roediger and Butler (2011), Dutch-speaking L2 learners were tasked with studying fifteen English nouns presented on a screen under one of three conditions: word writing, retrieval practice, or a control condition in which participants simply viewed English–Dutch word pairs. Form and meaning recall tests were administered immediately after the learning session and one week later to assess participants’ retention of the target vocabulary. The findings revealed that retrieval practice led to significantly better immediate and delayed form recall compared with both word writing and the control condition.

Evidence in the realm of cognitive psychology shows that contrary to retrieval practice, trial-and-error practice, where participants are tested on items before they are presented with the to-be-learned material, influences how items are subsequently learned (e.g., Grimaldi & Karpicke, 2012; Huelser & Metcalfe, 2012). In trial-and-error practice, learners infer the correct answer on their own because the test questions provide sufficient contextual cues. Richland et al. (2009) demonstrated that even test questions unlikely to be answered correctly based on prior knowledge enhance retention of the correct responses when subsequently presented to the learner. In one experiment by Potts and Shanks (2014), participants learned the definitions for unfamiliar English words or translations for foreign vocabulary through three methods: generating a response followed by corrective feedback, reading the word and its definition or translation, or selecting from multiple-choice options followed by feedback. In a final test of all the words, generating errors followed by feedback led to significantly better memory for the correct definition or translation compared with either reading or making incorrect choices. This suggests that the benefits of generation extend beyond correctly generated items. Even when the information to be learned is novel, error-prone generation may play a powerful role in enhancing the encoding of corrective feedback.

The aforementioned review shows that not only retrieval practice but also trial-and-error practice facilitates L2 word learning. To date, Strong and Boers’ (2019a, 2019b) experiments are among the few studies to have directly compared the effectiveness of retrieval and trial-and-error practice for learning phrasal verbs. Strong and Boers (2019a) revealed superior immediate post-test performance for retrieval practice (study-then-recall sequence) over trial-and-error practice (exercise-then-study sequence). However, the participants were only asked to supply the particles for each phrasal verb, which is not sufficient for learners to acquire a phrasal verb. Strong and Boers (2019b) introduced modified experimental conditions, including two retrieval practice groups (item-by-item vs. set-based practice) and two trial-and-error learning groups (immediate vs. delayed feedback). The results demonstrated that retrieval practice significantly outperformed trial-and-error practice in the immediate post-test. However, this advantage diminished in the delayed post-test, suggesting that retrieval practice alone may not sufficiently support long-term retention and learning of phrasal verbs.

Regarding Strong and Boers’ (2019b) limited success in promoting long-term retention of phrasal verbs, one major limitation lies in their research design: the decontextualized, static, and mechanical nature of the practice condition, which fails to meet the requirements for effective second language acquisition (Ellis & Shintani, 2013). Furthermore, the participants were only required to practice particles during the treatment phase but were later expected to recall entire phrasal verbs in the post-test. This discrepancy indicates a need for practice types that can more effectively facilitate the comprehensive learning of phrasal verbs.

In the present study, the effectiveness of modified retrieval and trial-and-error practices on phrasal verb learning is further compared and analyzed. To explore how to maximize the effectiveness of practice types for phrasal verb acquisition, the continuation task is incorporated into this investigation, enabling us to determine whether it can optimally enhance phrasal verb learning. To address these gaps, the current study compares the effectiveness of three types of practice—the continuation writing task, retrieval practice, and trial-and-error practice—on the learning of English phrasal verbs. This study also evaluates the perceptions of learners in terms of how these different practices influence their learning of English phrasal verbs. The specific research questions include (1) Are there any differences in the effectiveness of different practice types for learning English phrasal verbs? If yes, why? (2) What are the similarities and differences in the phrasal verb errors made by L2 learners from different practice type groups?(3) How do learners from different groups perceive the effectiveness of different practice types in terms of assisting with learning phrasal verbs?

3. Methodology

3.1. Participants

This study recruited 80 first-year university students aged 17 to 19 who were learning English as a second language at a university in China. The participants were drawn from three parallel classes and randomly allocated to one of three treatment groups: continuation, retrieval, or trial and error. A comparison of the participants’ recent exam scores (F(2,77) = 0.564, p = 0.571) indicated that the three groups were comparable in terms of their English proficiency. Given that the quasi-experiment was conducted over a three-week period during which none of the target phrasal verbs were explicitly taught, it can be assumed that the participants’ proficiency levels remained stable throughout this study. Furthermore, all the participants had studied English as a mandatory subject for a similar duration, thus ensuring comparability across the three groups. Since Chinese lacks structural equivalents of phrasal verbs, these participants were expected to find the acquisition of phrasal verbs particularly challenging. Notably, there were no participant dropouts during the quasi-experiment, despite them being informed of their right to withdraw at any time.

3.2. Target Phrasal Verbs

Initially, we selected twenty-eight phrasal verbs from the corpus developed by Strong and Boers (2019b) as possible target items for our study. These phrasal verbs were idiomatic in nature (i.e., non-compositional) and likely unfamiliar to the students participating in the research, as determined by a college English instructor with over a decade of teaching experience at the tertiary level. To ensure that these phrasal verbs were indeed highly unlikely to be known by the 80 participants across the three treatment groups, we conducted a pilot test involving 30 students who had similar English proficiency levels but were not part of the quasi-experiment. The format of this pilot test was identical to the one-week post-test used in the actual quasi-experiment. Out of the original 28 phrasal verbs, the pilot test revealed that 10 were unknown to all 30 students. Consequently, these 10 phrasal verbs were identified as the most likely to be unfamiliar to the students in the treatment groups and were thus selected as the target phrasal verbs for this study. The 10 target phrasal verbs and their corresponding paraphrases are presented in

Table 1. The target phrasal verbs.

Target Phrasal Verbs	Paraphrase
Hang out	To spend time with friends
Turn off	To lose interest
Get out	A secret becomes known
Dive in	To start to eat food
Open up	To talk about your personal feelings
Nod off	To fall asleep for a short time
Rip off	To charge someone too much money
Pop in	To visit for a short visit
Head off	To go somewhere
Back down	To decide not to do something

.

3.3. Instrument

3.3.1. The Reading Materials and Three Practice Types

As previously mentioned, the purpose of the current study was to investigate which practice types, namely, continuation, retrieval, and trial and error, would be most effective for learning phrasal verbs. To this end, reading material and three practice types were created.

A narrative passage was meticulously selected and adapted by the researcher to serve as the input text. This narrative was derived from My Father’s Dragon by Ruth Stiles Gannett, published by Random House Books for Young Readers in 1987 (see Appendix A). The story primarily revolves around a young boy who resolves to rescue a baby dragon that has been coerced by a group of lazy wild animals into ferrying them across the river on Wild Island. The input text consists of 1139 words, with its ending omitted. The researcher modified the story by substituting ten verbs or phrases that influence the plot with ten target phrasal verbs. To draw the attention of the participants and ensure comprehension, these phrasal verbs were highlighted in bold and underlined, with their L1 translations enclosed in parentheses. Furthermore, six reading comprehension questions, designed to enhance students’ understanding, were provided at the end of the input text. For ecological validity, the input text underwent proofreading by a native English speaker. Based on the students’ English proficiency levels and their course teacher’s assessment, the known word density of the input text was determined to be 97%, satisfying the lexical threshold for meaning-focused vocabulary learning (Nation & Nation, 2001). Based on the input text, three types of practice were developed and named as follows: (1) the continuation task, (2) the retrieval practice, and (3) the trial-and-error practice. For the continuation task (see Appendix B), the instructions clearly specified that after reading the story, the participants were required to write a continuation of the story, ensuring that each of the ten target phrasal verbs was incorporated at least once. The participants were also encouraged to refer to the input text throughout the writing process to ensure coherence, logical flow, and smoothness between the original and continuation texts. The continuation writing portion had a minimum word count of 300, and the participants were allotted 55 min for this task based on the findings of the pilot study.

For the retrieval practice, the students were instructed to read the input text and subsequently complete a target phrasal verb central practice (see Appendix C). This practice focused on the form, meaning, and usage of the target phrasal verbs. During the retrieval practice, the students were encouraged to consult the input text for feedback. The time allocated for this practice was 50 min, as determined by the pilot study.

In contrast, in the trial-and-error practice, the sequence of the stages was reversed. The students began with the target phrasal verb central practice (see Appendix D) and were asked to guess the answers. Following this, they received feedback in the form of the same input text provided during the first phase of the retrieval practice. The students were also encouraged to refer to the input text for additional feedback. The time allocated for the trial-and-error practice was 50 min, which is consistent with the pilot study’s findings.

3.3.2. The Letter-Cued Target Phrasal Verb Recall Test

The pre-test, immediate post-test, and 1-week-delayed post-test all employed the same letter-cued target phrasal verb recall test to assess the participants’ productive knowledge of the target phrasal verbs (see Appendix E). Each test consisted of ten target items and five distractor items (e.g., “burst into,” “die out,” “go over,” and “break into”). Each item featured a two-sentence dialogue incorporating one phrasal verb. The initial letters of the verb and particle were provided to eliminate irrelevant or misleading alternatives. For instance, the letters “h” and “o” excluded the possibility of “going out,” leaving “hang out” as the only viable alternative. Additionally, Chinese translations were provided as cues to help the participants produce the target phrasal verbs. This approach was justified because using a phrasal verb correctly requires both spelling accuracy and contextual understanding, making it essential for the test to assess whether these criteria are met. The sentences were adapted from Strong and Boers (2019a) and the Oxford online dictionary, and they were piloted to ensure no words in the sentences would hinder comprehension. The order of the test items was counterbalanced across the three tests to minimize potential test effects. An example of this test is presented in Example 1.

Example 1:

A: --Hey, Yuki, if you’re not busy after work, do you want to h_____ o ____ (外出) with me?

B: --I’m sorry, Mike, but I’m not feeling well today. How about tomorrow?

3.3.3. Distractor Tasks

In the current study, the researcher designed a 10 min distractor task to serve as a filler activity. During this task, the participants answered ten trivia questions in their native language and solved ten simple arithmetic problems (e.g., 28 + 34 = ?). The purpose of incorporating this distractor task was to reduce the likelihood of order effects that might otherwise influence the results.

3.3.4. The Open-Ended Questionnaire

To gather the participants’ self-assessments of the effectiveness of the three treatments for learning target phrasal verbs, the researcher designed a questionnaire (see Appendix F for details). The questionnaire included a 5-point Likert scale and one open-ended question. The Likert scale was designed to evaluate learners’ perceptions of the effectiveness of the practice types, with 1 indicating “not effective at all” and 5 representing “very effective.” The open-ended question aimed to elicit the participants’ comments and attitudes toward the different practice types: “Do you believe the practice in which you just engaged is effective for learning phrasal verbs? Please provide a detailed explanation of your reasoning.” The open-ended question was designed to collect data regarding the participants’ perceptions of the effectiveness of the different practice types and the underlying learning mechanisms.

3.3.5. Procedure

Figure 1 illustrates the progression of the procedure and the timeline of this study.

A quasi-experimental design featuring a pre-test, an immediate post-test, a 1-week-delayed post-test, and additional measures was employed in the current study. The participants were randomly assigned to three conditions. One week before the treatment, the pre-test was administered to evaluate the participants’ prior knowledge of the target phrasal verbs.

During week 2, the three quasi-experimental groups took part in the treatment session before moving on to the immediate post-test. For the continuation group, the instructions clearly stipulated that participants were to first read the input text and then employ all ten target phrasal verbs to coherently and logically complete the narrative. Additionally, participants were advised to consistently refer back to the input text during the writing process. The compositions were required to be a minimum of 350 words, and the time allocated to the continuation writing task was 55 min, as determined by the pilot study. For the retrieval group, the participants were instructed to read the input text and engage in a retrieval practice exercise that incorporated all ten target phrasal verbs. They were also advised to consult the input text for feedback during the retrieval practice. The time allocated for this activity was 50 min, based on the pilot study. In the trial-and-error group, the participants initially completed the same retrieval practice as the retrieval group and then used the input text to correct their errors. The time allocated for the trial-and-error practice was also 50 min, as established by the pilot study.

After the treatment, the participants were instructed to complete a filler task consisting of 10 simple math questions (e.g., 19 + 17 = ?) and 10 trivia questions in their first language. The participants were allotted 10 min to complete this task. The purpose of this distractor task was to minimize potential order effects.

Following the filler task, the participants completed a letter-cued phrasal verb recall test as the immediate post-test, with the items randomized in their presentation order. The participants were given 15 min to complete the test, consistent with the time allocated for the pre-test.

After the immediate post-test, a questionnaire survey was administered. First, the participants were asked to rate the effectiveness of the treatment for learning target phrasal verbs on a 5-point Likert scale, with 1 indicating not effective at all and 5 representing very effective. Subsequently, the participants were required to respond to the open-ended question. The participants were allotted 10 min to complete the questionnaire.

The participants were not informed that they would complete a 1-week-delayed post-test in the following week. This delayed post-test was identical to both the pre-test and the immediate post-test, with the exception that the order of the items was re-randomized.

3.3.6. Analysis

The pre-test and post-test responses were scored dichotomously (correct or incorrect). For verbs, minor spelling errors (e.g., “get” vs. “got”) were accepted as correct if they did not impede the recognition of the target verb. For particles, no credit was granted for incorrect responses (e.g., “in” vs. “on”), as it would be impossible to distinguish between a spelling error and an incorrect word choice. Each correctly spelled word was awarded 1 point. Responses to the distractors were not included in the scoring. The total score for the spelling of the target phrasal verbs constituted a participant’s final spelling score. A one-way ANOVA was conducted on the learners’ pre-test scores to examine the potential statistical differences between groups prior to treatment. A mixed between/within-group ANOVA was used to assess the differential gains across the groups over time, and a post hoc analysis was performed when significant results were obtained from the ANOVA.

4. Results

4.1. Effectiveness of Practice Types for Learning Phrasal Verbs

The present study investigated the impact of three types of practice on the acquisition of phrasal verbs. The participants’ responses to both the pre-tests and post-tests were collected, and the extent to which they were familiar with the target phrasal verbs was assessed before and after the intervention.

Table 2. Descriptive statistics for the learners’ pre-test and post-test scores.

		Pre-Test		Immediate Post-Test		Delayed Post-Test
Treatment	N	M	SD	M	SD	M	SD
Continuation	25	0.36	0.49	8.48	1.78	6.92	1.47
Retrieval	29	0.38	0.49	8.41	1.38	5.59	2.08
Trial and error	26	0.38	0.50	6.88	1.75	3.84	1.22

presents descriptive statistics for the learners’ scores in the pre-test, immediate post-test, and 1-week-delayed post-test across all the treatment groups. A one-way ANOVA conducted on the pre-test scores revealed no statistically significant differences between the three groups (F(2,77) = 0.018, p = 0.983 > 0.05), suggesting that the three groups were comparable at the beginning of this study.

As shown in Table 2, there were significant differences between the groups in both the immediate post-test and the one-week-delayed post-test regarding their target phrasal verb knowledge recall. A one-way ANOVA conducted on the letter-cued target phrasal verb recall test scores revealed statistically significant differences between the groups for the immediate post-test (F(2,77) = 7.994, p = 0.001 < 0.01, η² = 0.172), indicating a large effect size according to Cohen’s (1988) criteria (η² ≥ 0.14 represents a large effect). Similarly, significant differences were observed in the delayed post-test (F(2,77) = 22.234, p < 0.001, η² = 0.366), which reflects a very large effect size. To identify the specific differences between the three groups, Scheffé’s post hoc pairwise comparisons (with an alpha level of 0.05) were conducted. As presented in

Table 3. Between-group Scheffé’s post hoc pairwise comparison for the post-tests.

Immediate Post-Test	Delayed Post-Test
Continuation = Retrieval	Continuation > Retrieval *
Continuation > Trial and error ***	Retrieval > Trial and error **
Retrieval > Trial and error ***	Continuation > Trial and error ***

Note: * ρ < 0.05; ** ρ < 0.01; and *** ρ < 0.001. The equals symbol (=) indicates no statistically significant difference between the two groups.

, in the immediate post-test, the continuation group performed comparably to the retrieval group, and both groups significantly outperformed the trial-and-error group. In contrast, in the delayed post-test, the continuation group outperformed both the retrieval group and the trial-and-error group, with the retrieval group performing better than the trial-and-error group.

Figure 2 illustrates the mean scores of each group across the three testing sessions. All three groups exhibited a significant improvement from the pre-test to the immediate post-test. The graph pattern reveals that following the treatment, the continuation group performed comparably to the retrieval group, whereas the trial-and-error group demonstrated relatively lower effectiveness. However, in the delayed post-test, performance declined compared with the immediate post-test, with the trial-and-error group showing the poorest results (average score of 3.84). On average, the test scores decreased by approximately 2.5 points between the immediate and delayed post-tests. Nevertheless, the continuation group still recalled a higher average number of target items (6.92) compared with the retrieval group (5.59). These findings indicate that all three treatment conditions facilitated the development of verb phrasal knowledge, with continuation practice being the most effective. Additionally, the results suggest that retrieval practice had a more pronounced long-term impact on the learners’ phrasal verb acquisition than trial-and-error practice.

4.2. Similarities and Differences in Errors Made by Different Groups

Regarding the more specific question of the phrasal verb errors made by the participants in the three treatment groups, we analyzed the constituent errors in the phrasal verbs, examining both verb errors and particle errors in detail.

Table 4. Verb errors in the immediate and delayed post-tests.

		Immediate Post-Test		Delayed Post-Test
Group	N	M	SD	M	SD
Continuation	25	0.28	0.46	0.36	0.49
Retrieval	29	0.34	0.55	0.45	0.57
Trial and Error	26	0.54	0.65	0.58	0.64

and

Table 5. Particle errors in the immediate and delayed post-tests.

		Immediate Post-Test		Delayed Post-Test
Group	N	M	SD	M	SD
Continuation	25	1	1.53	2.12	1.48
Retrieval	29	0.59	0.57	2.66	1.49
Trial and Error	26	1.23	0.9	3.35	1.44

present descriptive statistics for the learners’ verb and particle error scores in both the immediate and delayed post-tests. In the immediate post-test, the students incorrectly supplied an average of one verb compared with two particles. In the delayed post-test, these figures increased to two verbs and four particles, respectively.

To investigate whether differences in the verb error rates existed between the groups, one-way ANOVAs were conducted on the verb error scores (Table 4) of the learners for each testing occasion. The one-way ANOVA on immediate post-test scores revealed no statistically significant difference between the three groups (F(2,77) = 1.49, p = 0.231 > 0.05), suggesting that the groups were relatively comparable regarding the verb errors made during the immediate post-test. Similarly, the one-way ANOVA on the delayed post-test scores also showed no statistically significant difference between the three groups (F(2,77) = 0.927, p = 0.40 > 0.05), indicating that the groups remained relatively comparable regarding the verb errors made during the delayed post-test.

Figure 3 displays the mean verb error scores for each group for both the immediate and delayed post-tests. Although no significant differences were observed between the three treatment groups, the graph reveals a consistent slight increase in performance from the immediate post-test to the delayed post-test. Based on these comparisons, the following pattern emerges regarding the correct verb responses across the different practice groups: continuation task > retrieval practice > trial-and-error practice. However, the comparisons also indicate that the differences in learning effectiveness between the three practice types are better explained by difficulties in recalling particles rather than verbs.

To investigate whether differences in the particle errors (Table 5) existed between the groups in the immediate post-test, one-way ANOVAs and multiple comparisons were conducted on the learners’ particle error scores for each testing occasion. The results indicated that there was no statistically significant difference between the continuation group and the retrieval group (p = 0.417 > 0.05), nor between the continuation group and the trial-and-error group (p = 0.792 > 0.05). However, a significant difference was observed between the retrieval group and the trial-and-error group (p = 0.01 < 0.05).

Regarding the particle errors in the delayed post-test, a one-way ANOVA revealed that F(2,77) = 4.450, p = 0.015 < 0.05, η² = 0.104. This result indicates a statistically significant difference between the three groups. Post hoc multiple comparisons further showed no significant difference between the continuation group and the retrieval group (p = 0.416 > 0.05), nor between the retrieval group and the trial-and-error group (p = 0.228 > 0.05). However, there was a significant difference between the continuation group and the trial-and-error group (p = 0.015 < 0.05).

Based on the aforementioned analysis, both the immediate and delayed post-tests indicate that the participants in the trial-and-error group made the highest number of particle errors, whereas those in the continuation group made significantly fewer particle errors in the delayed post-test (see Figure 4).

The above data highlight the verb and particle errors in the letter-cued target phrasal verb recall tests. With respect to the verb errors, no statistically significant differences were observed between the treatment groups. In terms of the particle errors, the results revealed that the trial-and-error groups made a greater number of particle errors compared with both the continuation and retrieval groups. Furthermore, no statistically significant difference was found between the continuation and retrieval groups.

4.3. Participants’ Perceived Learning

Table 6. Participants’ perceptions of the effectiveness of the different practice types for learning phrasal verbs.

Group	Not Effective at All	Not Effective	No Opinion	Effective	Very Effective
Continuation	0%	0%	4%	60%	36%
Retrieval	0%	3.45%	10.35%	13.79%	72.41%
Trial and error	3.85%	15.38%	46.15%	30.77%	3.85%

displays the participants’ responses to the 5-point Likert scale, with anchors ranging from 1 (not effective at all) to 5 (very effective). The results reveal that 96% of the participants in the continuation group rated continuation practice as effective for learning phrasal verbs. Likewise, 86.2% of the participants in the retrieval group found retrieval practice effective for this purpose. Conversely, only 34.62% of the participants in the trial-and-error group perceived trial-and-error practice as effective for learning phrasal verbs. Notably, none of the participants in the continuation group rated their practice as ineffective, whereas 3.45% of those in the retrieval group considered retrieval practice ineffective. In contrast, 19.63% of the participants in the trial-and-error group expressed dissatisfaction with this method. In summary, Table 6 shows that most of the participants believe that the continuation writing task and retrieval practice are effective in promoting the learning of English phrasal verbs.

The responses to the open-ended questions were collected and analyzed. The following excerpts illustrate the participants’ perceptions of the effectiveness of the different practice types for learning phrasal verbs.

For the continuation writing task, a participant from the continuation group found the continuation task beneficial for learning phrasal verbs.

Example 2:

“The continuation practice is effective for phrasal verb learning. I was used to learning vocabulary by rote, but when I did continuation practice, I learned words in interesting story contexts in which the content and phrasal verbs impressed me. When I completed the story, I made sentences myself, which gave me a sense of achievement. Of course, when I did not know how to express my idea with a phrasal verb, I resorted to the story and imitated the usage of the phrasal verb. I felt relaxed because I was not afraid of having nothing to say.”

Regarding the retrieval practice, Example 3 presents the response of a participant who found the retrieval practice helpful for learning phrasal verbs.

Example 3:

“The retrieval practice is effective for phrasal verb learning. When reading the interesting passage, I noticed the salient phrasal verbs. With the paraphrases of the phrasal verbs, I can understand them easily. So, when I did the retrieval practice, I could quickly find the answers.”

Regarding the trial-and-error practice, Example 4 presents the perspective of a participant who considered the trial-and-error practice ineffective.

Example 4:

“The trial-and-error practice has little effect on phrasal verbs learning. When I did the practice, I just wildly guessed the answers. Even though I read the story after finishing the practice and searching for the correct answers, I still forget most of them, and I may put down the same wrong answers as I did in the practice stage.”

5. Discussion

The current study investigated whether the continuation writing task would optimize the effectiveness of learning English phrasal verbs compared with two other practice types, namely retrieval and trial and error. It also evaluated learners’ perceptions of how these different practices influence their learning of English phrasal verbs. The results demonstrated that, in general, the three practice types were effective in promoting learners’ knowledge of phrasal verbs, with the continuation writing task being the most effective. With respect to the participants’ perceptions, their preference for practice types was ranked as follows: the continuation writing task > the retrieval practice > the trial-and-error practice.

The first research question examined whether differences exist in the effectiveness of the various practice types for learning phrasal verbs. The results demonstrated that all three practice types positively influenced phrasal verb learning. Specifically, the continuation group significantly outperformed the retrieval and trial-and-error groups on the 1-week-delayed post-test. These findings align with the previous literature, confirming that continuation tasks optimize L2 learning (X. Zhang, 2016, 2017; Jiang & Tu, 2016; C. Wang & Wang, 2015; Q. Wang & Wang, 2018; Cui et al., 2019). Two main factors contribute to the effectiveness of the continuation writing task.

First, learners engaged in a continuation writing task naturally produce dynamic expressions of ideas when incorporating the target phrasal verbs into their continuation writing. In L2 classroom settings, learners frequently lack opportunities for dynamic idea expression, despite extensive exposure to static language drills. However, the challenge for L2 learners lies in mapping static language onto fluid, dynamic ideas (C. Wang, 2020). Studies conducted by Shah et al. (2013) and Rassaei (2017) indicated that the complexity of learning tasks influences brain activity. Specifically, their research demonstrated that when learners participate in tasks that demand a higher degree of dynamic expression, such as those involving problem-solving, critical thinking, or creative output, there is a notable increase in the number of activated brain regions compared with tasks requiring minimal cognitive effort. In the current study, the participants in the continuation writing group engaged in more complex English phrasal practice and, therefore, stimulated greater neural engagement, potentially leading to deeper cognitive processing and enhanced learning outcomes. Conversely, in retrieval tasks, learners generate knowledge of phrasal verbs through controlled and static exercises. Without “content dynamics,” static phrasal verbs remain at the declarative stage, reducing opportunities for transition into procedural knowledge, which may lead to fossilization and make L2 production laborious and effortful. Interestingly, in the immediate post-test, little difference was observed between the effectiveness of continuation and retrieval for learning phrasal verbs. In other words, in the short term, retrieval practice was equally as beneficial as the continuation writing task. This may be because retrieval practice provides learners with a foundation of declarative knowledge, which they can further build upon during practice (Lightbown, 2019). However, in the delayed post-test, the continuation writing task outperformed the retrieval practice, which demonstrates that performance levels during practice do not reliably predict long-term retention of knowledge and skills. Instead, the continuation writing task required the participants to express their own ideas using the target phrasal verbs. Although this approach may initially appear less effective and effortful at first glance, it ultimately promotes deeper encoding and more durable learning outcomes (Schmidt & Bjork, 1992; Soderstrom & Bjork, 2015).

Second, the participants’ engagement with “the situation model” provides valuable insight into how effective the continuation task was. Q. Wang (2019) posited that collocational constructions are contextually primed and evaluated due to their processing advantages. For learners to expand the input text effectively, they must first achieve a thorough understanding of the reading material and pay close attention to the target phrasal verbs. This enables the extensions they produce to remain consistent with the original context, allowing them to use the target items appropriately when completing the story. Furthermore, because learners’ continuation writing occurs within the same context as the input text, it facilitates the retrieval of the target phrasal verbs when they have the opportunity to apply them (C. Wang, 2020). The absence of a floor effect in the retrieval and trial-and-error practice conditions in this study, contrary to the findings of Strong and Boers (2019b), may be attributed to the influence of contextual factors. Additionally, to produce more target-oriented language, learners must comprehend the previous input text fully, and when they encounter difficulties in expression, they tend to repeat and imitate language from the input text. However, this does not mean that learners copy the expression from the input text; rather, it is a process of creativity, which is the ability to generate ideas, solutions, and plans that are new, novel, or unique when confronted with a specific problem. C. Wang (2021) described this imitation as “creative imitation” (meaning creation couples with language imitation), which plays a critical role in ironing out comprehension–production asymmetry, prompting learners to express ideas dynamically and pre-empting errors through modeling. Once a similar situation context occurs, the identical items can be restarted (Q. Wang, 2019). For example, in the reading material, “turn off” turns up in the following sentence: Particularly ugly old cats always turn her off (使厌倦). A participant from the continuation group generated the following sentence: Your life will turn you off if you always live in the same place. C. Wang (2016) pointed out that if a learner creates content with their currently learned vocabulary, their memory of the word could be stored for longer.

The second research question addressed the similarities and differences in the phrasal verb errors made by L2 English learners from different practice groups. The result showed the students’ failure to recall the particles, especially those underlying the post-test responses, which is consistent with Strong and Boers’ (2019b) observation that, compared with verbs, particles pose a greater challenge to L2 learners. The results indicated that the trial-and-error groups generated a significantly higher number of particle errors than both the continuation and retrieval groups, while the continuation group produced the smallest number of particle errors. This result can be attributed to the following factors. First, regarding the trial-and-error practice, it is important to recognize that merely correcting an error through immediate feedback may not be sufficient to mitigate the negative impact of that error on subsequent learning attempts. This is because the feedback provided might lack the depth or clarity needed for learners to fully grasp the correct response. Without a thorough understanding, the correct information may fail to replace the initially learned incorrect response in the learner’s memory. In the current study, the feedback embedded in the reading material did not clearly explain how the meaning of the particle contributes to the overall meaning of the phrasal verb; students may struggle to internalize the correct usage. Even when they reread the reading material afterward in an attempt to self-correct, the absence of detailed explanations can leave gaps in their understanding, making it difficult to recall the correct phrasal verb in future contexts. Therefore, for trial-and-error learning to be effective, it is crucial that feedback not only identifies errors but also provides clear, meaningful explanations that assist learners in building accurate and lasting knowledge. Second, regarding retrieval practice, extensive research in cognitive psychology (Karpick & Roediger, 2007) and second language acquisition (Barcroft, 2007; Nakata, 2017) has demonstrated that actively recalling information significantly enhances learning outcomes. This positive effect is attributed to the cognitive process involved in retrieving knowledge from memory, which strengthens retention and deepens understanding. Moreover, providing learners with foundational declarative knowledge, such as the target phrasal verb’s usage in the reading material, serves as an essential reference point. This declarative knowledge of the target English phrasal verbs enables learners to make meaningful connections as they engage in more complex tasks (Lightbown, 2019). When learners have access to basic facts and concepts, they are better equipped to apply this knowledge in context, leading to more effective and efficient language development. Therefore, integrating retrieval practice into instructional design can play a pivotal role in supporting long-term learning and retention in second language education.

The third research question concerned the learners’ perceptions of the effectiveness of the different practice types for learning phrasal verbs. The findings revealed that the participants generally perceived the continuation task as more effective compared with both the retrieval and trial-and-error practices. Moreover, the retrieval practice was also viewed as more effective than the trial-and-error practice. Let us look closely at the participant from the continuation writing task who responded to the open-ended question by stating that “The continuation practice is effective for phrasal verb learning…When I completed the story, I made sentences myself, which gave me a sense of achievement…I felt relaxed because I was not afraid of having nothing to say” (see Example 1). The response indicated that the continuation writing task effectively enhanced the participant’s task self-efficacy in learning phrasal verbs by providing successful experiences (completing stories, creating sentences independently), reducing anxiety (a relaxed creative state), and context-based task creation. Confidence and belief in one’s ability to accomplish a learning task are indicators of progress and serve as a source of influence on task self-efficacy beliefs (S. Zhang & Zhang, 2021; Bai et al., 2019; Zhu et al., 2024; Wong, 2020). This sense of efficacy not only promotes current learning persistence but may also transfer to other language tasks, leading to improvement in the L2 learner’s overall L2 achievement and foreign language learning attitude, forming a virtuous cycle (S. Zhang & Zhang, 2021). With regard to the retrieval practice, one participant stated that “The retrieval practice is effective for phrasal verb learning…when I did the retrieval practice, I could quickly find the answers,” indicating that learners tend to learn phrasal verbs under the condition where declarative knowledge is provided, which supports retrieval studies (Roediger & Butler, 2011; Karpicke et al., 2014; Smith et al., 2016) that claim that testing has a robust effect on retention. Compared with the continuation writing task, the retrieval practice was relatively effortless, resulting in relatively superficial retention. With respect to the trial-and-error practice, one participant stated that “The trial-and-error practice has little positive effect on phrasal verbs learning…I just wildly guessed the answers…I still forget most of them, and I may put down the same wrong answers as I did in the practice stage,” indicating that the trial-and-error practice may not be an effective strategy for mastering phrasal verbs. The participant’s perception suggests that simply attempting to guess the meanings and uses of phrasal verbs without sufficient contextual understanding and use may lead to confusion and retention issues. This highlights the importance of incorporating more structured and contextually rich learning techniques, such as using phrasal verbs in meaningful sentences, engaging in interactive conversations, or studying them within thematic groups.

6. Conclusions

The primary objective of this study was to investigate and compare the effectiveness of three practice types for learning phrasal verbs: retrieval practice, trial-and-error practice, and continuation task. The secondary aim was to examine how these practice types influence learners’ task self-efficacy beliefs. First, the immediate post-test results revealed that the continuation task was nearly as effective as retrieval practice. However, the delayed post-test demonstrated that the continuation task outperformed the other two practice types, thus supporting the notion that providing learners with declarative knowledge of the target language, which they can anchor during practice, is beneficial. Furthermore, enabling learners to dynamically and creatively apply this declarative knowledge enhances its long-term retention. Second, the qualitative analysis indicated that practice types fostering learners’ task self-efficacy also contribute to L2 long-term retention.

The findings of the present study carry several important implications for material writers, teachers, and learners who aim to maximize the effectiveness of practice in terms of phrasal verb learning and, by extension, other linguistic items that may lack salience or prominence in the L1 system. One implication is that effective language learning can be achieved through practice types that provide learners with declarative knowledge of the target language, which they can draw upon during practice. Furthermore, in the practice phase, highly efficient practice types should encourage learners to creatively and dynamically imitate native speakers’ expressions while also enabling them to articulate new ideas. In addition, an effective practice type should enhance learners’ task self-efficacy beliefs, which are beneficial for promoting learners’ overall language achievement.

Despite these promising implications, the current research has certain limitations that merit acknowledgment. First, this study was restricted to the “verb + particle” structure of phrasal verbs, excluding other common constructions, such as “verb + adverb + preposition.” This narrow focus limits the generalizability of the findings, as different phrasal verb structures may pose varying levels of difficulty for learners. Future research should incorporate a wider range of phrasal verb types to provide a more comprehensive understanding of how continuation tasks influence their acquisition. A second limitation lies in the potential confounding effect of the pre-test design. Although the trial-and-error practice differed from the pre-test, administering a pre-test may have inadvertently introduced a trial-and-error component, blurring the distinction between retrieval practice and trial-and-error learning. While this approach was necessary to assess the participants’ prior knowledge, future studies should implement stricter methodological controls to ensure clearer separation between the pre-testing and practice phases. Finally, the post-test design, which relied exclusively on a letter-cued phrasal verb completion task, was a form of controlled practice. While such tasks are useful for measuring explicit knowledge, they do not adequately capture the ability of learners to use phrasal verbs fluently in authentic contexts (Ellis & Shintani, 2013). To address this, subsequent research should incorporate more ecologically valid assessments, such as picture description tasks or orally elicited imitation tasks (Suzuki & Sunada, 2018), which better reflect learners’ procedural and spontaneous use of the target structures. By explicitly addressing these limitations, we hope to encourage more rigorous and ecologically valid investigations into the role of continuation tasks in phrasal verb acquisition.