Edited by: Yang Zhang, University of Minnesota Twin Cities, United States
Reviewed by: Yixun Li, The Education University of Hong Kong, Hong Kong SAR, China; Jingyi Zhang, University of Miyazaki, Japan
This article was submitted to Language Sciences, a section of the journal Frontiers in Psychology
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
This study investigates the effects of visual mnemonics and the methods of presenting learning materials on learning visually similar characters for Chinese-as-second-language (CSL) learners. In supporting CSL learners to build robust orthographic representations in Chinese, addressing the challenges of visual similarity of characters (e.g., 理 and 埋) is an important issue. Based on prior research on perceptual learning, we tested three strategies that differ in the extent to which they promote interrelated attention to the form and meaning of characters: (1) Stroke Sequence, a
Chinese writing contains the most complicated graphs in the world (
Prior research has defined visual similarity as the degree to which orthographic features of Chinese characters overlap (cf.
Based on learning theories investigating how learners perceive, process, and maintain knowledge in learning,
To fill this gap, this study developed visual mnemonics that differ in the extent to which they promote interrelated attention to the form and meaning of Chinese characters. These mnemonics were developed based on the dual-coding theory, which postulates that information coded in both verbal and visual codes has additive effects on memory (
In what follows, we introduce the difficulties of learning Chinese characters (1.1), empirical research on the learning of visually similar graphs across writing systems and within the Chinese writing system (1.2), learning strategies and the use of visual mnemonics in supporting character learning (1.3), and the present study (1.4).
The Chinese writing system is logographic in nature, given that its written symbols represent lexical morphemes instead of individual phonemes (
However, decoding (e.g., recognizing characters) and comprehension are both critical skills in learning to read (
To overcome the difficulties of learning Chinese characters, developmental research on CSL/CFL learners’ orthographic awareness has shown that learners’ sensitivity to components with different functions (e.g., semantic or phonetic) may help them to learn characters (e.g.,
Taken together, within the complex interconnections among forms, sound, and meanings of characters, forms link more reliably to meaning-bearing morphemes (
Visual similarity of graphs may influence the development of high-quality orthographic representations (
Scientific studies on learning visually similar Chinese characters are scarce and rarely focus on how methods of presenting characters influence learners’ reading and writing. To the best of our knowledge, only two journal articles have adopted rigorous pretests and posttests between subject research designs and reported comparisons on learners’ performance in learning visually similar and dissimilar characters (
Meanwhile, visual similarity between characters caused by stroke discrepancies represents a significant challenge of orthographic learning for learners who have acquired basic knowledge of Chinese characters, including L1 learners.
Visual mnemonics provide a learning strategy to strengthen memory traces of orthographic and semantic constituents, in acknowledging the relatively reliable connection between form and meaning in Chinese characters. A visual mnemonic is a learning technique used to aid the association of the to-be-memorized information (e.g., forms and meanings) with mediators (e.g., imagery to represent meaning), which are more accessible to provide better retention and retrieval for learners. The power of mnemonics has been widely acknowledged in language education (for reviews, see
By leveraging visual mnemonics and characteristics of Chinese orthography,
Key-images and Pithy Formulas are developed based on cognitive theories, namely the dual-coding theory (
The “碧” example is also an instantiation of elaboration, a strategy to organize learning content into meaningful context, helping learners to construct knowledge in their minds (
In accumulating character learning experiences, adult CSL/CFL learners may adopt various learning strategies (e.g.,
Prior to the present work, previous studies had been performed (
The purpose of this study was to examine effects of three learning strategies (Pithy Formulas with Key-images, Key-images, and Stroke Sequence) and two methods of presenting learning materials (visually similar pairs and dissimilar pairs) in reading and writing visually similar characters for non-beginning CSL learners. Based on the literature review, we asked the following research questions:
What are the effects of learning strategies (Pithy Formulas with Key-images, Key-images, and Stroke Sequence) on learning to read and write visually similar characters?
What are the effects of presenting materials (similar vs. dissimilar pairs) on learning to read and write visually similar characters?
Is there an interaction between the three learning strategies and two methods of presentation in reading and writing visually similar characters over time (immediately after learning and 1 week after learning)?
How do CSL learners perceive their learning experience with these strategies in terms of enjoyment, usefulness, ease of use, and willingness to use in the future?
A 3 (learning strategy)- × -2 (method of material presentation)- × -2 (testing time) mixed design was carried out with learning strategies (Pithy Formulas with Key-images, Key-images, and Stroke Sequence) and testing time (immediate posttest and delayed posttest) as two within-subject variables; the method of presentation (similar and dissimilar groups) as a between-subject variable. The dependent variables were the accuracy of character recognition and writing, with additional measures of learner experience ratings on the three learning strategies.
We determined a sample size of 66 by conducting a priori power analysis for sample size estimation using G*Power 3 (
Thirty traditional Chinese characters were selected from the Chinese Orthography Database (
Pairs of learning materials (30 characters; 15 pairs in each group) between both groups (i.e., similar vs. dissimilar groups).
Characters learned in the similar group |
Characters learned the dissimilar group |
||||
Block 1 | Block 2 | Block 3 | Block 1 | Block 2 | Block 3 |
埋理 | 塊瑰 | 忡怏 | 埋貢 | 賈間 | 鈦話 |
責貢 | 買賈 | 查杳 | 稚討 | 棵詳 | 李怏 |
書畫 | 問間 | 鈦鈸 | 畫責 | 塊評 | 詁查 |
稚椎 | 稞棵 | 話詁 | 計理 | 買棵 | 忡杳 |
計討 | 評詳 | 季李 | 書稚 | 瑰問 | 鈸季 |
The learning measures included a character writing task assessing productive form representation and two recognition tasks assessing form-meaning (Chinese to English) and meaning-form connections (English to Chinese), respectively. In addition to the learning measures, a learner experience survey based on prior research (
The writing task asked the participants to write a character from memory based on a given prompt of English words. They were encouraged to try their best in completing the task by being promised partial credit for their responses. Responses were scored by two schemes – an all-or-none scheme (character scoring) and a continuous scheme (stroke scoring). Stroke scoring is a partial-credit-given scheme; the score of a character is a proportion of correct strokes produced (i.e., the denominator is the character’s total number of strokes, and the numerator is the number of correct strokes in the written response). In contrast, character scoring is a strict scheme in which credit (Score 1) is given only for an exact reproduction of the whole character, while all other responses are scored 0. Scores from these two schemes may reflect the extent to which learners can recall and reproduce the character forms. Previous studies supported the higher sensitivity of using stroke scoring relative to character scoring (e.g.,
Given that partial-credit-given scoring might involve different judgments on each correctly placed stroke, one researcher scored the entire set of written responses and a second researcher independently scored one-third of the responses. These responses were selected by stratified sampling, i.e., from the pre-test, the immediate posttest, and the delayed posttest; one third of the written responses were randomly sampled. Pearson product moment correlation was performed on the cases scored by the two researchers. Inter-rater reliability in stroke scoring was 99.%, and inter-rater reliability in character scoring was 100%.
The recognition task included two subtasks: Chinese to English and English to Chinese, and a computerized multiple-choice format was adopted for both subtasks. For the Chinese-to-English recognition task, thirty characters were presented in a random order on a screen. For each Chinese character, four meanings (in English) were presented, including the correct meaning and three distractor meanings that had been paired with different characters from the same block. The participants were instructed to choose the correct meaning and then proceeded at their own paces. For the English-to-Chinese recognition task, it was particularly designed to assess the participants’ orthographic representation of visually similar characters, i.e., the ability to differentiate one character from its visually similar counterpart. This task reversed the direction of recognition by showing English words (in a random sequence) and asking the participants to choose the corresponding character. Each English word had four candidates, including the correct character, the character that was visually similar to the correct one, and two other characters that had been paired with different meanings from the same block. For both tasks, the accuracy rate was calculated by dividing each participant’s correctly responded items by the total number of items (i.e., 30) and multiplying the result by 100.
The learner experience ratings assessed the participants’ opinions of each learning strategy in four aspects: the level of enjoyment in using the strategy, the usefulness of the strategy, the ease of using the strategy, and their willingness to use the strategy in the future. These aspects were based on the Technology Acceptance Model (
The participants’ responses were made on a seven-point Likert-type scale (1 = absolutely negative, to 7 = absolutely positive). There were twelve questions in total, four questions for each strategy, and the following are examples of questions regarding use of the Key-images strategy.
(1) Enjoyment of use: Please rate, from 1 to 7 (least to most), how much you
(2) Usefulness: Please rate, from 1 to 7 (least to most), how
(3) Ease of use: Please rate, from 1 to 7 (least to most), how
(4) Willingness to use: Please rate, from 1 to 7 (least to most), how
In this study, Cronbach’s Alpha was used to test the internal consistency of the measures. The coefficient was 0.73, indicating reasonable reliability (
The procedure of the study consisted of a pretest, a learning session, a posttest (immediately after learning), and a delayed posttest (1 week after learning). All were administrated in a one-on-one fashion in an experimental lab with assistance from trained researchers specialized in teaching Chinese as a second language. Additionally, all learning measures were introduced with a practice example to make sure that the participants understood the instructions of the tasks.
Before the learning session, each participant was asked to complete the pretest as described in the measurement; all pretests shared the same form as used in the posttest and the delayed posttest. Next, the participants were randomly assigned to either the group that learned with similar pairs (the similar group), or the group that learned with dissimilar pairs (the dissimilar group). Both groups learned the same 30 characters (in pairs), while the similar group encountered 15 pairs composed of visually similar characters, and the dissimilar group encountered 15 pairs, consisting of visually dissimilar characters.
For the learning session, there were 3 blocks; a Latin square design was adopted to balance the order of the strategy and the order of the pairs. In each block, the participant used one strategy to learn 5 pairs; within each block, the sequence of pairs was randomized. Thus, all the participants experienced all three strategies in learning different character pairs (30 characters in total).
An experiment schedule and Latin square design used for balancing the order of learning strategies.
Pretest | Participants ( |
Learning blocks |
Post-test | Delayed posttest | |||
The similar group ( |
The dissimilar group ( |
||||||
1/4/7/10/13/16/19/22/25/28/31 | 1/4/7/10/13/16/19/22/25/28/31 | P | K | S | |||
2/5/8/11/14/17/20/23/26/29/32 | 2/5/8/11/14/17/20/23/26/29/32 | S | P | K | |||
3/6/9/12/15/18/21/24/27/30/33 | 3/6/9/12/15/18/21/24/27/30/33 | K | S | P |
A character writing task and two recognition tasks were administered to assess individual participants’ prior knowledge of the target characters. For participation eligibility, only when the participants wrote no more than three characters out of 30 characters in the character writing task could they proceed to the recognition tasks. Nine participants were excluded due to their writing accuracy rates being higher than 10%. This criterion was set by consulting prior research (
Descriptive statistics (
Group learned with similar pairs ( |
Group learned with dissimilar pairs ( |
|||||||||||
Measure | Strategy | Pretest | Immediate posttest | Delayed posttest | Adjusted |
Adjusted |
Pretest | Immediate posttest | Delayed posttest | Adjusted |
Adjusted |
Co-variates in the ANCOVA |
Character writing (stroke scoring) | 0.07 (0.09) | 0.54 (0.24) | 0.33 (0.20) | 0.55 | 0.34 | 0.10 (0.12) | 0.57 (0.28) | 0.47 (0.23) | 0.56 | 0.46 | 0.087 | |
0.08 (0.10) | 0.43 (0.28) | 0.32 (0.24) | 0.45 | 0.33 | 0.09 (0.10) | 0.56 (0.28) | 0.40 (0.25) | 0.55 | 0.40 | |||
0.09 (0.10) | 0.41 (0.26) | 0.27 (0.19) | 0.43 | 0.28 | 0.09 (0.11) | 0.48 (0.29) | 0.40 (0.26) | 0.47 | 0.39 | |||
Character writing (character scoring) | 0.04 (0.06) | 0.34 (0.23) | 0.17 (0.18) | 0.35 | 0.18 | 0.05 (0.09) | 0.39 (0.29) | 0.26 (0.22) | 0.38 | 0.25 | 0.041 | |
0.03 (0.06) | 0.26 (0.26) | 0.13 (0.15) | 0.27 | 0.14 | 0.03 (0.06) | 0.33 (0.27) | 0.25 (0.24) | 0.32 | 0.25 | |||
0.05 (0.07) | 0.24 (0.23) | 0.15 (0.15) | 0.25 | 0.16 | 0.05 (0.07) | 0.29 (0.25) | 0.23 (0.21) | 0.28 | 0.22 | |||
Recognition (Chinese-to-English) | 0.46 (0.18) | 0.83 (0.16) | 0.79 (0.19) | 0.84 | 0.80 | 0.56 (0.22) | 0.81 (0.15) | 0.82 (0.16) | 0.79 | 0.81 | 0.497 | |
0.43 (0.19) | 0.84 (0.16) | 0.78 (0.17) | 0.86 | 0.79 | 0.51 (0.19) | 0.81 (0.15) | 0.77 (0.16) | 0.78 | 0.77 | |||
0.48 (0.19) | 0.78 (0.17) | 0.76 (0.18) | 0.79 | 0.77 | 0.55 (0.25) | 0.82 (0.16) | 0.82 (0.17) | 0.80 | 0.81 | |||
Recognition (English-to-Chinese) | 0.49 (0.22) | 0.77 (0.24) | 0.70 (0.24) | 0.80 | 0.73 | 0.53 (0.28) | 0.75 (0.23) | 0.71 (0.24) | 0.71 | 0.68 | 0.494 | |
0.42 (0.23) | 0.73 (0.21) | 0.67 (0.23) | 0.75 | 0.70 | 0.50 (0.20) | 0.79 (0.22) | 0.71 (0.25) | 0.76 | 0.68 | |||
0.45 (0.26) | 0.76 (0.23) | 0.70 (0.24) | 0.78 | 0.72 | 0.59 (0.25) | 0.78 (0.22) | 0.79 (0.23) | 0.75 | 0.78 |
To enhance the internal validity of this study, the participants were randomly assigned to one of the two groups. The participants learned a pair on individual computers, displayed by PowerPoint software (Microsoft Office, 2019). The participants were instructed to refrain from any hand movement and to focus on learning materials on the screen without auditory input; they were informed that the display was completely controlled by the computer program with the assistance of the administer.
The learning trial for each pair was divided into an observation phase and a study phase.
An example of the observation and the study phases for learning character pair 評-詳.
For each trial, specifically, the observation phase lasted for 23 s. First, an eight-s observation to a character’s form, sound, and meaning was provided by the following sequence of events: a character’s form was shown for 1 s, followed by 1-s
Focusing on the three strategies that differed as to which they promote attention to the form and meaning of characters,
Illustrations showing the three different strategies for learning character pair (in the similar group) 評-詳.
After the learning session, the participants were given 5 min to perform a distraction task (i.e., Task Load Index;
One week after the learning session, to assess the maintenance effect of the interventions, the participants were asked to complete the same tests as they did in the pretest and the immediate posttest. Finally, a paper version of the learning materials and a debriefing sheet were given to the participants after the completion of all the tasks.
The analysis of covariance (ANCOVA) was adopted to mitigate the possible effect of the participants’ knowledge of learning materials, given that our participants were not novice learners of Chinese (
Statistical Package for the Social Science (SPSS) version 23.0 software was used. The significance level was set at α = 0.05, and partial eta-squared (
Both the continuous and the all-or-none scoring revealed that the Pity Formulas with Key-images yielded higher writing accuracies than the other two strategies immediately after learning. Also, the interactions among strategy, group, and testing time varied by scoring schemes.
When scored at the continuous level, the Pithy Formulas with Key-images showed generally better learning outcomes than the Stroke Sequence strategy for both groups in the immediate posttest. The test of regression homogeneity indicated that the slopes did not significantly differ,
For the group that learned with similar character pairs, in the immediate posttest, higher writing scores were found in the Pithy Formulas with Key-images than those for Key-images (
As for the group that learned with dissimilar pairs, in the immediate posttest, Pithy Formulas with Key-images also yielded higher writing scores than Stroke Sequence (
In addition to the significant three-way interaction, significant main effects were also found: The Strategy main effect,
When scored at the all-or-none level, Pithy Formulas with Key-images was the most effective strategy for both groups immediately after learning. We used the pretest results as a co-variate. The test of regression homogeneity showed that the slopes did not differ,
For accuracy in choosing the correct meaning based on a given character, no significant effect was found. Taking the pretest as a co-variate, the test of regression homogeneity showed that the slopes did not differ,
For accuracy in identifying correct Chinese characters based on a given English word, scores in the English-to-Chinese task reflected whether the participants can correctly differentiate visually similar characters. Taking the pretest as a co-variate, the test of regression homogeneity showed that the slopes did not differ,
The learners’ ratings of their experiences with each learning strategy were made on four seven-point Likert scales. The top of the scale (7) was the maximum positive response on enjoyment, usefulness, ease of use, and willingness to use.
The participants’ mean ratings on enjoyment, usefulness, ease of use, and willingness to use for each learning strategy.
This study supported non-beginning CSL learners to build robust orthographic representations in Chinese by addressing the challenges of learning visually similar characters. The learning intervention was conducted
The discussion is organized as follows: first, we discussed the impact of learning strategy for reading and writing as well as the learners’ experience ratings (see section “Learning Strategy of Pithy Formulas With Key-Images Supports Character Writing and Positive Learning Experiences”). Next, we explored the effects of material presentation and focused on its interaction with learning strategy under the theoretical design principles of strategies (see section “The Superiority of the Pithy Formulas With the Key-Images Over the Key-Images and the Stroke Sequence Depends on Material Presentation”). Finally, we discussed research limitation and future directions (see section “Research Limitations and Future Directions”) and then offered an overall conclusion (see section “Conclusion”).
For reading, regardless of Chinese to English or English to Chinese, all three strategies were effective for our participants. They learned the associations between form and meaning, and they were able to correctly distinguish one character from the others. This finding is in line with prior research, showing that adult non-beginning CFL learners quickly pick up the perceived patterns of characters (
In discussing the main effect of learning strategy immediately after learning, which shows that the Pithy Formulas with Key-images lead to the highest accuracy rates in writing, we revisited its design principles based on the Elaboration theory (
An alternative explanation for the strategy effect observed in the immediate posttest is learners’ motivation. The learner experience ratings (see section “Learner Experience Ratings”) showed that the learners consistently expressed positive opinions (i.e., enjoyment, usefulness, ease of use, and willingness to use in the future) on the
Next, we explored the effects of material presentation. In the context of learning to read across writing systems, echoing literature in learning distinctive features of English letters (e.g.,
Within the Chinese writing system, prior research on learning visually similar characters was relatively scarce. Also, these studies revealed discrepancy between L1 school-aged learners with different levels of achievement in Chinese (
Furthermore, depending on the two methods of material presentation, we discussed their interaction effects with three learning strategies for learning visually similar characters under the design principles of learning strategies. Theoretically, our findings supported the Dual-coding theory (
The superiority of the Pithy Formulas with the Key-images echoed the third stage of the three-stage character-based instructional framework (
Notwithstanding the effects of intervention in supporting compound character learning, several limitations of the present study must be acknowledged, and the following are our suggestions for future research. First, we are mindful of the fact that the maintenance of strategy effects did not last long. The robustness of each strategy effect merits further investigation. To deal with this issue, we suggest to vary testing times (simultaneous vs. successive) and to provide multiple practice or review opportunities to find an optimal schedule (
This study examined the synergetic effects of learning strategies and methods of presenting materials on reading and writing visually similar characters for non-beginning CSL learners. The takeaway points are twofold. For the cognitive aspect, the learning strategies emphasizing both the verbal and visual codes (i.e., the Pithy Formulas with Key-images) outperformed the visual imagery (i.e., the Key-images), which, in turn, surpassed the Stroke Sequence in writing characters immediately after learning. For the affective aspect, CSL learners’ experiences consistently revealed that
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.
The studies involving human participants were reviewed and approved by Research Ethics Review Committee of National Taiwan Normal University (NTNU). The participants provided their written informed consent to participate in this study.
L-YC and Y-YT: conception and design of the study and data analysis and interpretation. Y-YT, C-YL, and L-YC: data collection. L-YC and C-YL: manuscript writing. H-CC: providing supervision and instructions during the whole process. All authors contributed to the article and approved the submitted version.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
The sources of funding received for this work came from the “Chinese Language and Technology Center” of National Taiwan Normal University (NTNU) from The Featured Areas Research Center Program within the framework of the Higher Education Sprout Project by the Ministry of Education (MOE) in Taiwan as well as a grant from the Ministry of Science and Technology award MOST 110-2511-H-003-039-MY2. L-YC would like to thank Anwei Yu’s assistance in revising the manuscript. The authors thank the reviewers and the editor for their helpful comments.
The Supplementary Material for this article can be found online at: