Chinese character recognition and literacy development via a techno-pedagogical pivot

Abstract

This study investigates the effect of sequenced, time-delayed, multimodal Chinese language and communicative literacy supports on the ability of learners of Chinese as a Foreign Language [CFL] to recognize characters and acquire literacy via writing in Chinese. A multimodal web application was designed and developed as a writing platform and supplementary teaching tool—leveraging second language [L2] spoken Chinese understanding and pinyin spelling knowledge to overcome the well documented difficulties faced by CFL students when learning and communicating via Chinese characters. We observed and recorded a communicative writing task carried out on the platform by 14 CFL students and interviewed them upon task completion—also interviewing their instructors. Our findings suggest that designing from a perspective that is pedagogically unique for CFL classrooms afforded new types of interactions and literacy development patterns. Specifically, timed, sequenced, multimodal scaffolds anchored to students’ L2 spoken Chinese understanding and pinyin knowledge supported Chinese character recognition, production, and acquisition within communication and expression-centered tasks.

Introduction

With 20% of the world’s population, a dynastic history that stretches back more than 3500 years, the world’s second largest economy, and the world’s third most powerful military—interest in the People’s Republic of China, as well as Hong Kong and Taiwan, is predictably high. These cultural, economic, and political factors have made learning the Chinese language a priority in the US. The National Security Language Initiative (US Department of State, 2008), the 100,000 Strong Foundation, and the ongoing 1 Million Strong Initiative all address the need for increased numbers of US K12 and higher education students in Chinese language classrooms and study abroad programs. Recent reports show that Chinese is taught at more than 850 US universities and in over 700 K12 US classrooms reaching more than 200,000 K12 students (Klein, 2015; Tanner & Wang, 2015; Walker, 2016).

Given the above-mentioned trends, researchers have called for inquiry into the experiences of non-native learners of Chinese—with the aim of better serving the needs of learners and supporting teachers (Sung & Wu, 2011). Such inquiry is especially vital as Chinese holds the distinction of being one of the most difficult languages to acquire as a second [CSL] or foreign language (Liskin-Gasparro, 1982 in Shen, 2005). A large part of this difficulty stems from the challenges presented in becoming literate in Chinese. Learning to read and write via Chinese characters is commonly held as the most difficult element of learning Chinese (Shen, 2004; Yongbing, 2005) and therefore a particularly important area of study within the field of CSL/CFL research (Wong, 2017).

The challenging nature of reading and writing in Chinese is due to several factors. First, there are more than 80,000 Chinese words of which 3000 might be considered non-rare—and thus important to learn in order to enjoy functional literacy (Wong et al., 2010 in Sung & Wu, 2011). Second, there is an absence of generalized sound-symbol correspondence between characters and their pronunciation (Sung & Wu, 2011). As Yeung et. al. (2011) put it, while alphabetic scripts follow a sound-symbol or grapheme-phoneme correspondence, Chinese characters have a unique shape, a specific pronunciation, and a particular meaning—none of which reliably or robustly co-inform the others in an easily discernible way. Although a character’s shape can hold meaning and/or pronunciation clues, less than 50% of characters do (Shen, 2004; Shu et al., 2003 in Lee & Kalyuga, 2011a). Third, while printed English alphabetic letters are composed of one to four individual marks or strokes, Chinese characters can be composed of up to 43 strokes—rendering their identification and composition challenging.

Language learners who read and write alphabetic languages as their first language [L1] encounter higher effort requirements and unique challenges in the process of acquiring literacy in Chinese than they would learning another alphabetic language (Cheng & Chiu, 2018; Nation, 2001 in Sung & Wu, 2011). For example, a 20 year old English speaker learning Spanish can hear the word ‘digital’ and have a pretty good sense that is either spelled ‘digital’ or ‘dijital.’ That same learner could read the word ‘digital’ and connect it to its pronunciation—and probably to its English cognate ‘digital’ (Chen et al., 2013). In contrast, the word for ‘digital’ in Chinese is written 数字(化) and pronounced ‘shùzì(huà).’ While the character contains the phonetic radical 子 pronounced ‘zǐ’ [zee] the character does not give clues to its semantic meaning, nor which tone should be applied to the vowel ‘i’. Vowels and vowel groupings in Chinese words have one of five pronunciation options, or tones, signified by four possible markings above the vowel (the fifth has no mark). Additionally ‘shùzì’ is not a cognate of its English equivalent—nor are there any orthographic cognates between English and Chinese. Finally, 数字 shares pronunciation features with one other Chinese word (梳子, shūzi, comb). Put another way, for individuals exclusively literate in alphabetic languages, learning to write in Chinese requires making non-trivial sound-shape-meaning connections largely from memory with little meta-linguistic, semantic, or phonetic support from their L1 (Shen, 2004; Wang, 1998; Yeung et al., 2011).

Due to the above-mentioned factors, Chinese literacy acquisition has been characterized by rote memorization, repetitive and mechanistic drills, and stacks of flashcards (Allen, 2008 in Hsieh & Fei, 2009; Olmanson et al., 2018; Zhao & Xin, 2002 in Chen et al., 2013). Despite their promise and utility, phonetic and meaning-based character elements—such as 子 in the previous paragraph—are seldom strategically used by CSL/CFL learners as pronunciation and semantic aids (Chen et al., 2013; Liu & Olmanson, 2016; Sung & Wu, 2011). Instead, Wang (1998) reported 80% of students participating in his study saw character learning not as a process of learning to write and express themselves but as a time consuming, challenging process of learning each individual character in isolation. This situation places students and instructors in a position where, given the considerable effort required of CSL/CFL students to become literate in Chinese (Sung & Wu, 2011), they either have to divert considerable time away from speaking and listening development to ensure their literacy keeps pace with their aural development, or function with two sets of L2 proficiency levels wherein their literacy development is 1/3 to 1/10th of their aural L2 capacity—often with the dual-proficiency option being seen as inevitable.

Cheng and Chiu (2018) have noted that such drill-based practices conflict with the prevailing pedagogical consensus surrounding second and foreign language education. They have called for an inquiry into the use of scaffolded communicative practices—specifically writing practices—with novice-level Chinese language learners.

Within the field of education, the construct of scaffolding has been used for more than 40 years as a way to support student learning via activities that would otherwise be impossible, inappropriate, ineffectual, or intimidating (Graves & Graves, 2003 in Clark & Graves, 2005; Wood et al., 1976). Whether via an educator, a parent, a more knowledgeable peer, or a learning technology, scaffolding is based on Vygotsky’s view of learning as a social constructivist endeavor (Vygotsky, 1978). Specifically, scaffolded interaction is a way to promote growth by assisting students in the completion of more authentic or holistic tasks that are currently too complex or difficult for them to do on their own but within their zone of proximal development when provided the appropriate kind of support (Vygotsky, 1978). The appropriate type of support—determined by a teacher, tutor, or technology—is one or several task modifications that make an ambitious or authentic activity possible without rendering it too easy or eliminating productive student struggle (Clark & Graves, 2005).

Within Chinese literacy acquisition one of the most important scaffolds has been the use of pinyin. CFL/CSL courses commonly start students out with pinyin—alphabetically spelled Chinese words (e.g. the 子—Zǐ example above) (Chung, 2007). By using pinyin as a scaffold, students with a level of L2 aural proficiency are able to complete homework and decode Chinese texts that include both characters and pinyin.

This type of support affords learners the opportunity to complete reading comprehension tasks while gaining familiarity with frequently used Chinese characters without the challenge and delay of looking up each character (Chappelle, 1998; Clark & Graves, 2005). Yongbing’s study (2005) into the use of pinyin with students acquiring Chinese literacy found that a combination of extensive pinyin-supported reading followed by writing practices that targeted frequently encountered characters dramatically increased students’ character recognition capacity without the typical isolated character learning practices. Additionally, research into early literacy development for beginning Chinese learners supports the exploration of Chinese characters via multiple modalities (Curtain et al., 2016).

A holistic approach to Chinese literacy acquisition also involves gaining an understanding of Chinese orthography—namely how characters are built and which parts of characters offer semantic or phonetic clues—called radicals (Wong, 2017). As learners gain familiarity with Chinese characters, they eventually identify orthographic decoding strategies as one of the most useful ways of acquiring literacy (Shen, 2005). Initially however, beginning learners have a limited capacity to benefit from radical-based character learning—as they often lack an understanding of the Chinese writing system that would support their ability to learn more effectively (Chen et al., 2013; Shen, 2005). This lack of background knowledge often has students and teachers turning to learning aids that focus on individual characters in isolation—frequently in the form of physical and digital flashcards that include images, audio, pinyin, and radical cues.

Craik and Lockhart (1972) posited a levels-of-processing conceptual framework. Level one involved repetitive rehearsal—resulting in shallow and less durable learning. Level two included multimodal methods of connecting new knowledge to understood or intelligible information and structures. In Shen’s study (2004) of learning Chinese characters, simple memorization approaches fared less well than deeper processing approaches involving visual, textual, and oral scaffolds. While printed and digital character sheets and flashcards have been useful in offering a range of multimodal supports for character learning (Hsieh & Fei, 2009; Shen, 2005), they are not without limitations or controversy.

Specifically, the presence, location, and timing of supports relative to the target character has been a matter of debate and scholarship. Many character learning flashcards combine the character with an image depicting meaning, the written pinyin, and, for CSL/CFL learners, a translation (see Fig. 1). The historical order, position, and inclusion of specific elements on flashcards appears to not be rooted in theory so much as a matter of convention (Chung, 2007).

Fig. 1

Three examples of Chinese character learning cards (Smart Lingo, Chineasy, Semanda)

A study by Chung (2003) found that when L1 and L2 words are presented next to each other and at the same time, learners tended to pay more attention to the L1 translation—reducing the focus on the character. In a subsequent study, Chung (2007) connected this to the presence of written pinyin. Later, Lee and Kalyuga (2011b) suggested that audio pronunciation rather than printed pinyin may be a more effective character learning scaffold.

The history of technology supported language acquisition stretches back at least to language and literacy educator Jan Amos Comenius. In 1658 Comenius penned and published the first illustrated language learning textbook Orbis Sensualium Pictus [the sensory world in images] (Comenius, 2019). In the ensuing 370 years, language learners have increasingly benefited from technologies that support language development by providing the learner multiple scaffolds and models for language use. The media that has made up the foundation of language learning supports in the past century—due largely to its capacity to engage students when access to their instructor is limited—has grown from line drawings and text to include photos, audio, video, and chat (Otto, 2019). Since the beginning of this century, most learners access materials, interact with fellow learners, and express themselves in a second language via a range of technologies—making technology an integral, rather than optional, part of second and foreign language teaching and learning experiences (Chapelle & Sauro, 2019; Ertmer & Ottenbreit-Leftwich, 2010). These technologies enable meaningful, authentic exposure to, and communicative interactions with, other speakers and writers of the language (Mishan, 2005). In other words, in the 2020s the question is not whether a teacher or learner should integrate technology into language learning but rather which technologies should be integrated.

The affordances of multimedia-based technologies for Chinese literacy acquisition extend beyond increased modalities for comprehensible input (Hsieh & Fei, 2009; Krashen, 1982) to unique ways of leveraging attention. Chung (2003, 2007) demonstrated that issues of split attention and working memory limitations could be overcome, and character acquisition improved, by altering the timing and sequence of supports. Instead of showing pinyin and English translations at the same time as the Chinese character—a practice that may hinder character learning by decreasing learner effort—participants in the study were first shown the character and then shown the L1 translation five seconds later. In this way Chung (2003, 2007) found that learners, especially novice learners, were better able to focus on the character—using later-appearing translations, audio, and pinyin as confirming/disconfirming metacognitive prompts that encouraged active and increased processing with learning-assistive feedback. Paying attention to the salient elements to be learned is a central part of the cognitive theory of multimedia learning, a theory that also advocates for strategic grouping and aligns with Chung’s approach to strategic temporal sequencing (Mayer & Johnson, 2008).

The wide accessibility of networked technologies within learning contexts has also been recognized as a source of potential impact for improving learning and supplementing instruction. Furthermore, heightened attention on CALL research focused on online writing aligns with both the increased use of communicative language learning pedagogies and the use of visual elements in the production of writing (Godwin-Jones, 2018). This turn toward multimodal pedagogies (Li & Storch, 2017) as well as theories of multimodality (Jewitt, 2006; Kress, 2010), combined with the decades-long history of multimedia use in language learning settings (Chen et al., 2013), creates an environment wherein students are familiar with, may expect, and rely upon technology-supported instruction and a range of writing and literacies-related resources (Godwin-Jones, 2018; Li & Storch, 2017). The situation described above, of longstanding challenges to Chinese literacy acquisition despite the presence of networked, digital, multimodal scaffolds, presents an exciting design challenge in need of a techno-pedagogical pivot.

Educational technology design as a techno-pedagogical pivot

Educational technology design and development is a unique meta-disciplinary and cross-disciplinary field wherein both research and practical aims are supported by theory and theorizing (West et al., 2020). Viewing the design of educational technologies as an opportunity to make new types of learning and interacting possible via deep content-specific curricular and pedagogical understanding gets to the heart of a techno-pedagogical pivot. This construct is a productive response to the ways metanarratives of education influence learning designs (Olmanson, Abebe, et al., 2015). We call this systemic influence, the filtration of innovation effect. Through this effect we posit that adopting dominant positions on a number of education-related elements tends to produce designs resulting in education technologies that look and perform a lot like the ones we already have. Specifically, we observe that the sociocultural contexts that frame the organization of learning and school, act as a series of filters that constrain the design of technology-supported learning experiences (Costa et al., 2019; Olmanson, Kennett, et al., 2015). We use this understanding to remain sensitive to the positions society and institutions favor when it comes to elements such as curriculum, pedagogy, learning theory, assessment, economics, and power. In doing so, we are able to better understand the ways those positions individually and collectively influence and constrain current designed interventions and how we might make new things possible by strategically taking up different positions along one or several of these elements.

For example, designing a technology for use in Algebra I classrooms in public high schools in the US could lead to a filtration of innovation effect if the designers adhered to positions that have long influenced the teaching of algebra, namely: a sequenced, pre-defined objectives-centric view of curriculum; a direct instruction approach to pedagogy; an external, quantifiable, standardized, and individualized position on assessment; and a bias toward incrementalization and homogeneity of outcome and experience. While it is advantageous and reasonable to design with a knowledge of the contexts within which a learning technology will be used (Rieber, 2020), designs adhering exclusively to the predominant perspectives create two interrelated issues. First, it re-instantiates the ecology under which the problem in need of a technology-supported solution emerged. Second, it creates a constellation of constraints that influence what, how, with whom, and under what conditions students learn Algebra I. These dynamics exert an effect on design possibilities, particularly, they often exclude alternative perspectives on curriculum, pedagogy, learning theory, achievement, and equity—prematurely filtering out potentially novel designs that may make new types of learning possible and/or support learners who have been historically underserved by existing approaches.

The techno-pedagogical pivot is a productive response to the filtration of innovation effect (Olmanson et al., Olmanson, Abebe, et al., 2015). It is a sideways move in the face of education technology design pressures to: (a) design from within the dominant education paradigms, (b) incrementally improve existing approaches, (c) evaluate new designs with existing measures, and/or (d) add new technological capabilities to existing tools. While designing in response to these pressures often leads to useful education technologies, it also limits the range of possible designs by constraining the pathways to learning and the types of learners who are most likely to benefit from those pathways. A techno-pedagogical pivot is one in which designers acknowledge the dominant paradigms and strategically take alternative positions on one or several elements. These alternative perspectives afford a greater diversity of design starting points and trajectories—resulting in learning technologies that pivot away from incremental improvements in existing educational technologies and toward new possibilities.

In the field of Chinese language learning, technologies to support literacy development are typically designed by drawing on a number of dominant perspectives. Online, animated flashcards, matching and stylus-based character tracing apps and activities are outgrowths of a focus on decontextualized character memorization that has served as the primary vehicle for Chinese literacy development for over a century (Olmanson et al., 2018). The dominant linguistic, curricular, pedagogical, and sociocultural perspectives on learning Chinese act as filters in the design of new Chinese literacy acquisition applications. Flashcards go online, handwritten character practice is offered via tablet and stylus, yet despite the potential offered by networked multimedia-capable technologies, the dominant curricular and pedagogical focus on individual character memorization constrains design decisions.

A techno-pedagogical pivot involves designing from alternative positions on several educational and societal elements. For example, in designing DaZiBao we pivoted away from linguistics-centric, transmission-based pedagogies that privilege a focus on individual words and rote character memorization and pivoted toward communicative language learning pedagogies that support student growth via expression, scaffolded interaction, and noticing. We pivoted away from vocabulary lists as the catalyst for determining which words get memorized and instead positioned learner spoken vocabulary knowledge as the driver of character recognition and literacy growth. We made these design pivots with a theoretical and practical understanding of second language acquisition approaches that the literature suggests are productive but are not typically used in the design of Chinese literacy instruction (Design-Based Research Collective, 2003). We also surveyed existing Chinese language and language learning technologies used with other languages, eventually identifying design potential in a combination of multimodal supports and the digital character input technologies used by native speakers of Chinese in their everyday literacy practices.

Writing in Chinese via computer, tablet, or phone commonly involves the use of digital character input technologies. These OS-level tools take pinyin as user input (e.g., zi, ma) and return a row, column, or table of similarly pronounced, possible matching characters for insertion into a text (see Fig. 2). Within Chinese literacy instruction, the existence of such tools allows learners to digitally produce Chinese texts without having to first master writing the characters by hand—provided they can already recognize the character they want to insert into their text from a list of similarly pronounced characters (Stickler & Shi, 2013).

Fig. 2

Screenshot of online character input method tool [Chinese-Tools.com]. Ma is entered and a series of phonologically matching characters are shown. Pressing the corresponding number associated with one of them adds it to the composition space on the left

To synthesize the last two sections, though beginning CSL/CFL students rely heavily on repetition and memorization, the use of multiple strategies focused on input, output, cognition, and metacognition have been identified as central to Chinese literacy acquisition (Fan, 2003 in Chen et al., 2013; Sung & Wu, 2011). Over time, CSL/CFL learners become more adept at using a character's semantic and phonetic elements. Similarly, they become better able to leverage the characters they have learned in order to infer meaning and comprehend new characters and character based texts. Scaffolds such as pinyin, multimodal character representations, and time-delayed supports have been shown to be effective approaches to Chinese literacy acquisition. With the proper scaffolds, learners at the initial stages of acquiring literacy may be able to bypass a significant amount of time spent in drill-and-repeat approaches to character learning in favor of more authentic and communicative practices (Yongbing, 2005). What has yet to be fully explored and studied is the potential for writing-focused technologies, based on alternative approaches to pedagogy and curriculum, to play a more central role in language and literacy acquisition for CSL/CFL learners within and beyond the classroom (Cheng & Chiu, 2018). Thus, with this study, we test DaZiBao—a scaffolded, multimodal, research-informed web app for CSL/CFL communication and literacy acquisition. Via this test we: 1) explore the potential and value of the design to impact learning, 2) cultivate insights to apply to the next/second iteration of the DaZiBao application, and 3) gather evidence of the viability of the techno-pedagogical pivot as a guiding theory for innovative or divergent design (Barab & Squire, 2004).

Methodology

Research design and procedure

The inquiry presented herein is part of an ongoing instructional technology design and development effort founded on a conceptualization of language acquisition as multimodal, heterogeneous, and communicative. Our research question centers on developing Chinese literacy in new ways. Specifically we ask, in what ways can a scaffolded digital Chinese writing platform change the practice and experience of L2 writing and Chinese character recognition for novice CSL/CFL students? Or, put another way, to what extent can the process of communicative written expression on a digital platform lead to character recognition and literacy development [without repetitive memorization drills]? By addressing this question we explore the claims of Lee and Kayluga (2011) regarding the deleterious effects of simultaneous support provisioning. We also inquire into Chung's (2007) assertion regarding the utility of a sequenced approach to character-learning scaffolds. Additionally, we identify elements that can be re-designed in the application’s second iteration. Finally we seek to contribute to the design repertoire of the field of educational technology by describing and testing our techno-pedagogical pivot approach to design (Design-Based Research Collective, 2003).

Set in the Midwest United States, our study employs qualitative ethnographic methods such as interview, observation, and artifact analysis—particularly multi and bi-directional analysis (Halverson & Magnifico, 2013)—within a design-based research framework (Anderson & Shattuck, 2012). Aligned with Anderson and Shattuck’s (2012) analysis of what constitutes design-based research, our study: centers on participants in an adult Chinese language learning course; focuses on the design and testing of an intervention that can be used within typical Chinese-as-a-foreign-language classrooms; employs a range of methods [e.g. interview, observation, screen recording, and artifact analysis]; aims to iteratively improve the intervention [DaZiBao]; and is an active collaboration between researchers, designers, and chinese language teachers. We use bi-directional and multi-directional analysis to cross-analyze the semi-structured interviews we conducted with fourteen student participants and two teachers against our observational fieldnotes, participant screencasts, and collected telemetry (log) data (Halverson & Magnifico, 2013; Olmanson et al., 2016; Weiss, 1994; Wolcott, 2005).

Data collection took place in the Spring of 2018 in Eastern Nebraska. A member of our research team spent two weeks observing and capturing activities related to student use of our designed intervention/platform. These activities included participant observation of classroom sessions as well as one-on-one sessions with students using the intervention (Spradley, 2016). Drawing from artifact analysis (LeCompte & Preissle, 1993; Van Gog, 2007) we used an iterative analytical approach (Anfara et al., 2002) in which we coded each interview transcript, fieldnote, and screencast. Next we cross-coded each student’s interview against the screencast of their use of DaZiBao as well as the related observational fieldnotes. We generated a total of 321 base codes—assigning them where applicable. Codes were then arranged by topic, sub-ordered by frequency, and mapped thematically.

In terms of analytical procedure and sequence, first we assigned base codes to meaningful units of student and teacher interview data and fieldnotes. We also coded time-sequenced portions of student writing session screencasts. This enabled us to create and code descriptive logs of platform-supported student writing activities. We then organized base-codes into three categories: (1) student and teacher interview data regarding the experience of developing literacy in Chinese, (2) students using the DaZiBao platform, (3) student interview data related to platform use. Within and between these subgroupings we chunked, themed, and developed insights via a visual code mapping tool. For purposes of triangulation, we compared statements participants made in their interviews to their writing sessions, compositions, and telemetry data (Anfara et al., 2002).

Intervention protocol

We reached out via email to Chinese teachers in Southeastern Nebraska and asked if they would be willing to participate in a study focused on supporting character learning and written expression. We explained that our intervention was designed by a partnership of educators, researchers, and developers and that it offered sequential, multimodal scaffolds designed to help students acquire and communicate via Chinese characters based on their ability to use pinyin to spell the Chinese words and phrases that were part of their oral vocabulary. Available resources, teacher interest, and student participation afforded us the opportunity to work with two teachers and fourteen of their students during the eighth and ninth weeks of their ten-week semester.

Participating teachers were shown the platform and given time to try it out. An open-ended writing activity was jointly selected by the research team and teachers so students could compose texts that were both informal and connected to their lives while also allowing them to use as much of their spoken Chinese vocabulary as possible in their writing (Cheng & Chiu, 2018). A research team member introduced the platform to the students, providing them with platform-related support when required. In the weeks leading up to their participation, teachers said students spent time learning to read and write simple conversations about themselves and family members via pinyin but hadn’t spent any time in-class learning Chinese characters or writing. At the time of the interviews, students had spent between eight weeks and two years learning Chinese. Students spent an average of 11 min and up to 32 min completing the writing intervention task.

Design philosophy and description of the Chinese character writing platform

In designing the initial iteration of DaZiBao we borrowed from standard character-input-method workflows (see Fig. 2) and the multimodal elements of digital flashcards (see Fig. 1) to create a platform that allows learners to identify and use characters in their writing based solely on their Chinese oral vocabulary and understanding of pinyin.

This study explores the potential of what we outline in the introduction as a techno-pedagogical pivot approach to the design of an application for Chinese character learning and literacy acquisition (Olmanson et al., Olmanson, Abebe, et al., 2015). We apply this approach to overcome or side-step societal and systemic dynamics that have constrained the design of character learning interventions and technologies. In the case of supporting Chinese literacy development, the complex and divergent nature of the written language for most CFL students in North America, coupled with the heavy reliance on traditional learning approaches like vocabulary flashcards, worksheets, and dictionaries has constrained efforts in the last decade to design technologies in support of Chinese literacy development. Flashcards and worksheets have gone online, making them cheaper and more colorful and sometimes animated. Dictionaries became apps, offering fast look-up, spoken pronunciations, and stroke order diagrams. Meanwhile, some students simply used Google translate in a cut-and-paste 1950s-style grammar-translation approach. These practices and the resultant technologies—based on traditional pedagogical approaches to teaching and learning Chinese—might best be characterized as incremental efforts to improve learning from within the dominant educational paradigms.

We argue that the lack of more divergent technology-supported approaches to Chinese literacy acquisition stems from continued adherence to traditional character-learning practices in Chinese schools and teacher education programs—and the influence classroom practice exerts on educational technology designers. In other words, past designs and predominant pedagogical approaches constrain the ways we design new technologies. The resulting effect is one that perpetuates the reuse of existing design elements at the expense of new possibilities that draw on alternative approaches to teaching and learning. Applying a techno-pedagogical pivot is a productive design response to the limiting effect of the status quo. It is a sideways move around institutional and societal pressures that oblige new technological capabilities to fit within existing tools and existing patterns of teaching, learning, and assessment.

DaZiBao is as an instantiation of a techno-pedagogical pivot in that we worked to synthesize the current literature on technology-supported character learning while incorporating pedagogical approaches that are new to CSL and CFL classrooms and students —approaches that shifted the focus from traditional character memorization tools toward practices that enabled writing for communicative purposes.

The design and development team, led by the first and second authors, consisted of experienced language and literacy educators, designers, and researchers—two of whom have extensive CSL/CFL teaching experience. We developed this first working iteration of DaZiBao (Chinese Character Helper: http://chinesecharacterhelper.com/) to be a pedagogically supportive platform for Chinese writing.

Despite the prevalent use of multimedia applications for learning Chinese characters, and the use of character input methods for generating digital texts, we found no applications that were designed to aid students in learning characters via a process that combined character input with multimedia while maintaining a focus on written communication and self expression. Designed and developed without funding, DaZiBao is a web-based application that gives CSL/CFL students the opportunity to repurpose or leverage their spoken Chinese communication abilities in the production of Chinese texts. Students enter untoned pinyin (e.g., zi, ma) and DaZiBao displays the characters that share pronunciations based on the entered text. If the student recognizes their intended character from the row of characters they can select it just as they would if they were using a standard character input method (see Fig. 2). If the student does not make a selection within three seconds, an audio icon appears under each character (see Fig. 3). Students can tap or click on this icon to hear the toned pronunciation of each word. In this way, we draw on Chung’s work (2007) suggesting that a character-first presentation places the student’s focus fully on the character to be learned with additional supports like pronunciations serving as sequenced scaffolds in the form of phonemic feedback.

Fig. 3

Screenshot of DaZiBao application when ma is entered and the student has not made a selection after three seconds

Should another four seconds pass without the student selecting a word, a column of up to 10 images appear below each option (see Fig. 4). The displayed images are dynamically drawn from a repository of over 6 billion photographs and graphics (Parfeni, 2011). By drawing on such a large repository, the platform offers learners multiple images even for uncommon Chinese words. The images displayed alternate between matching a search for the Chinese word and matching a search for its English translation—in doing this we aim to support students in gaining exposure to L2 cultural insights embedded within images while offering aids that are L1 culturally familiar. The use of multiple images stands as both a way to reduce the pressure otherwise placed on one dynamically selected image to embody a word’s meaning and as a post-structural attempt to display a range of word meaning legibilities (Derrida, 1997; Olmanson et al., 2012). As with the audio support, we designed the delay in image display to maintain student focus on the character (Chung, 2003, 2007) with the images serving as a second level of sequenced scaffolds providing semantic clues.

Fig. 4

Screenshot of DaZiBao application when ma is entered and the student has not made a selection after seven seconds

If another five seconds pass without a character selection, a book/dictionary icon appears under each option. The English translations for each Chinese word are available via a tap or click on the icon which displays the translation color-coded in green under the audio and dictionary icons (see Fig. 5). As with the audio and image supports, we designed the delay in translation display to maintain student focus on character pronunciation and meaning for as long as possible before revealing the English translation which serves as a third and final level of sequenced support—providing definitive and useful feedback for literacy acquisition and Chinese composition (Chung, 2003, 2007).

Fig. 5

Screenshot of DaZiBao application when ma is entered and the student has not made a selection after twelve seconds

In these ways, we draw on research which suggests that character-first sequencing, placing supports underneath and in close proximity to the character (Lee & Kalyuga, 2011a), and color coding translations (Chung, 2007) can positively impact character learning. Additionally, we heed Shen’s call (2004) for pedagogies that help students go beyond the shallow processing of repetitive rehearsal and connect newer or new material (the characters) to known material (word meaning and pronunciation) via multimodal supports that function as sequenced feedback in the completion of expressive writing tasks (Clark & Graves, 2005; Craik & Lockhart, 1972).

Finally, our platform affords an exploration of if and when Yongbing’s (2005) findings on character acquisition via pinyin-supported reading can be extended to scaffolded student writing. As students use DaZiBao to compose texts, more frequently used characters should become familiar to them—meaning they might wait for fewer and fewer supports before making their intended selection.

Data and findings

In this section, we first consider participant perspectives on the experience of learning Chinese in general, in developing Chinese literacy in particular, and the role technology plays in their literacy development. We then look at the student experience of using DaZiBao in composing Chinese texts and their reflections on those experiences. Finally, we describe participant reactions to the specific design features. We draw on data across our 14 students and two teacher participants. The findings that follow build upon and extend an earlier round of general and teacher-focused analysis completed on subsets of the data (Olmanson et al., 2018; Olmanson et al., 2019).

Learning Chinese writing

If the language has [similarities] with the language you learned as a child, it’s easier because you have a context, and you can think ‘it’s the same but make this change.’ But when it’s Chinese, there is nothing. -Theresa.

Participants viewed learning Chinese in general and characters in particular as “difficult” and “totally different”. They attributed this difficulty partially to the distinctive gaps in writing systems between Chinese and their first language. Theresa (all names pseudonyms), a student who had lived in China for a year but was only beginning to, in her estimation, take learning Chinese seriously said it was difficult, “because [Chinese writing] is so different than the ‘characters’ we use in English.” George framed the challenge with more affect. In comparing writing in English to Chinese he said, “but the characters, I don’t know what’s what, it’s just some shapes I don’t understand, so yeah, it’s scary.” Aaron mainly avoided writing saying, “my main focus was to learn to read. I put little effort on writing because I was so intimidated.”

Our participants reported using rote memorization—in the form of repetitive writing and flashcards—to learn Chinese characters outside of the classroom. Chuck said his approach was, “writing it over and over again.” He described using a workbook which required him to translate English into Chinese characters—doing repetitive copying. “For three months I didn’t understand anything,” he said of a repetitive process he ultimately identified as helpful. Laura also used a workbook, saying, “I didn’t find [it] very useful until after I knew what [the characters] meant.” Becky summed up the process of learning characters on her own time as “obnoxious.” Describing herself as a visual person, Ashley used side-by-side copying followed by individual character production.

The other popular strategy among student participants were flashcards. Lily and Luke named flashcards as the best way they had found to learn characters. Their practice involved using flashcards “one at a time, but over and over again” to quiz each other. Laura, Evan, and Aaron also identified flashcards as important in learning characters. George however, struggled with the lack of context saying, “I’ve done a few flashcards, but it doesn’t stick as well for me because it’s isolated, it’s not tied to anything [so] it doesn’t have meaning.”

Participants also used technologies to support character production and learning. Students identified Rosetta stone, Google translation, Pleco, Chinese Character Writer, and WeChat by name. Aaron, George, Lily, and Luke used the Chinese dictionary app Pleco [http://pleco.com] for English-to-Chinese character translation. Similarly, Google translate was used by several students including Nick and Luke, who would type in English to get the character equivalent or take a picture of the character or phrase to get translated definitions. Steve, a student who was labeled by his teacher as advanced said he found WeChat to be helpful for learning Chinese characters.

Both teachers said they spent less than 10 min a week teaching characters and even less time using technology tools in support of literacy development. This resonates with previous research—finding that many teachers do not teach writing due to its increased complexity and an already overloaded curriculum (Read et al., 2014; Wang, 1998).

Our participants echoed the sentiment found in other studies as far back as Wang’s, 1998 study which predates smart phones, tablets, and most internet-based learning supports. While access to the latest devices and applications available in 2018 seemed to make Chinese character production via translation faster and easier, actual literacy development for novice students was predominantly undertaken via by-hand character memorization and online flashcard-like systems—both of which employed decontextualized drill-and-practice pedagogies.

By the time of our intervention, participants had learned greetings, how to introduce themselves, how to characterize members of their family, and the months of the year. Yet, teachers—not believing students were ready to express themselves in written form—had not assigned any homework or in-class work requiring students to express any of these ideas in written Chinese. The only writing teachers had assigned, and students had practiced, was copying and recopying several Chinese characters. So, when we asked participants about writing in Chinese, students interpreted their copying practices as writing.

Participant experiences with the first iteration of the intervention: using DaZiBao

Observations of participants composing on the DaZiBao platform as well as post-intervention interviews suggest that the use of time-delayed, sequenced, multimodal scaffolds supported students in developing literacy within the context of communicating via written Chinese. Learners leveraged their existing vocabulary to craft compositions. They used audio, image, and translation supports to both enable their writing and make learning gains. Data also suggest that students gained meta-awareness about their Chinese literacy development.

Across interviews, DaZiBao was regarded as helpful by participants—provided they knew how to pronounce words in Chinese and, as Evelyn said, “as long as you know how to spell it in pinyin, overall it helped.” Students found different supportive elements useful but tended to prefer one over the others. For example, Evan identified images as most helpful, and his screencast also showed evidence that he used images more often when composing (see Fig. 6). Lily and Lee found audio pronunciation most supportive. Lily listened to nearly every audio clip across the row of character options for each word she wrote.

Fig. 6

Video sections of Evan’s use of the platform, note his use of images [https://youtu.be/fqTRhDckpM0]

While some wanted immediate access to translations, Ashley, Steve, and Nick noted the pedagogical nature of having the audio and image supports precede translations. Ashley said, “I like how each comes at a different time…so, it gives you a chance to remember on your own first.” Steve said, “I like the way it put the pronunciation first, so I think it could be helpful for learning. I don’t have to look at a definition or whatever to see whether it is right. I can remember that in my head.” Comparing it to other tools, Nick said, “It’s a lot quicker than a dictionary, and I would say it definitely helps me learn Chinese better using this. Because when I use google translate, I would type in the English word, it would be giving me the [translation]. With this, I have to use pinyin to get the characters so this gets me more involved in learning Chinese.” In comparing the platform to character input methods (see Fig. 2) George said, “It was nice to have a little more help because [with character input methods] it just stops. I mean you have to know the character, but your program allowed you to go beyond that and gave you more help to choose the correct ones. That was nice.”

Theresa, Nick, Laura, and Aaron preferred translation. Laura waited for translations to pop up before making most of her selections. Additionally, while the supportive elements participants identified as most helpful influenced how they used the platform, character familiarity also played a role. Despite stating his affinity for the sequence of the supports, Steve’s use of the platform revealed that he often waited for and accessed character translations before making a selection. Often he and other students would hover their cursor over the spot where the dictionary icon would appear—waiting to click on it (see Fig. 7).

Fig. 7

Video sections of Steve’s use of the platform, note his use of translations [https://youtu.be/dHzsHnMW2hc]

Laura, Lily, and Luke regarded the intervention as a tool that allowed them to express themselves in written Chinese before they had memorized enough characters to do so on their own. Laura said, “[if I] didn’t have to know the character exactly, I [could] just recognize [them].” George said, “there is no way I would be able to write that in characters on my own.” Luke mentioned that he would have just used pinyin for the open-ended writing assignment if he didn’t have access to DaZiBao. Steve said, “If I were to write with my hand, I would probably look up all the characters and recopy.”

In interviews, participants like Steve above recognized that he used the intervention not solely as a tool to produce individual words but also as a method of Chinese character learning. Chuck, Dee, and Steve interpreted the need to discern the intended character from a row of possible characters as instructive. Steve said, “[when I had] to guess which one was right [I] started to remember the one I want.” Regardless of which scaffolds students preferred, the presence of multiple sequenced supports was identified as important. Theresa described her use of all the different supports, saying, “[First, I typed in] the pinyin, then I tried to remember the character. But when I wasn’t sure, I looked at the pictures, then listened to the sound, then if I still wasn’t sure, or to be very sure, I [used] the English translation.” Theresa’s process maintained a focus on the character while sequentially calling upon her previous knowledge to produce L2 text and develop a connection between meaning, pronunciation, and character-based words.

Based on observations and interview data, writing on the platform went beyond character lookup. It was also described as a learning experience that facilitated participant noticing and reflection. This was most readily observed with characters students used multiple times within the same writing episode. Nick said, “I used the character wǒ [我, I, or me] a lot, and I got [to] automatically know that’s the right one.” In his screencast, Nick typed the untoned pinyin wo three times. The first time he typed wo he listened to the pronunciation and viewed images before picking 我. The second time, he typed the phrase wo su but, upon seeing his options, erased it and replaced it with wo shi (我是, wǒ shì, I am). This time he listened to the pronunciation and viewed the translation for 我是 (I am) before selecting it. The third time he entered wo Nick didn’t wait for any of the supports, immediately picking 我. Evelyn listened to pronunciations before selecting 我 the first time and selected it with no supports the second time she typed wo. Theresa typed xuexi twice, the first time she looked at the images and used the translation to select 学习 (xuéxí, learn), the second time she typed xuexi she selected 学习 immediately. She also used images and translations the first time she typed hanyu—selecting 汉语 (Hànyǔ, Chinese), the second time she typed hanyu she selected 汉语 with no supports (Fig. 8).

Fig. 8

Video sections of Theresa’s use of the platform, note her initial use of pronunciations, images, and translations for two words followed by her selection of the same words later with no supports [https://youtu.be/aKGNC85FZiY]

The experience of using and reflecting on the use of a sequenced multimodal writing platform supported metacognitive awareness among students. Noticing occurred in a number of ways. Aaron alluded to the way the process reinforced his sense of progress saying, “I found when I was typing I didn’t use the pictures too much. I used the sound and then I just started to recognize characters that kind of looked familiar so I knew I was on the right track and I would chose it.” Luke attributed his realization of character-syllable correspondence to his use of DaZiBao. He said, “I don’t know why it took me so long to figure this out. But, if it’s a two syllable word, like mingzi, that’s two characters. If it’s a three syllable word, it’s three characters. You just have to look at it to see how long the word is to see how many characters make up that word.” This not only improved his ability to winnow his options based on syllables but it also added to his holistic understanding of written Chinese.

Dee, in her interview, described how her experience helped her reevaluate her knowledge of Chinese characters. She said initially she, “was mainly focused on what I know, which is limited, but it made me realize that I know more than I actually think, so I was kind of happy. Because I was able to identify the characters, it made me realize that I know more than I actually think. I didn’t look for definitions because I was able to identify the characters.” Dee’s experience suggests that for her, time to think and act between supports served as a catalyst for self-evaluation and meta-awareness.

Participant reflections on design features

After analyzing participant screencasts and teacher and student interview data we noted design opportunities for future iterations of the application. While most participants acknowledged the learning potential of a writing application dedicated to supporting L2 expression based on their existing spoken vocabulary, some expressed an interest in using their L1 to access new vocabulary. Both Laura and Ashley noted that if they didn’t know the Chinese word there was nothing they could do within the platform to access the character. Additionally there was an interest in expanding the character output field so that longer compositions could be accommodated without right to left scrolling.

While students often used images as a way to determine character meaning, image relevance was at times an issue. Steve said, “sometimes [the images were] confusing. For example, [the character for] little brother [had] random pictures.” Additionally, though no students brought up the issue of image age-appropriateness during their interviews, there were, across the 14 adult participants’ screencasts, more than 30 images shown to students that would likely have been deemed inappropriate within US K12 settings—part of our target context. Twentyseven times, people in thong bikinis were shown, twice people in barely-there underwear appeared, and once an image of a burned human corpse was shown.

Despite having just written an open-ended text and expressing interest in using the platform in the future as a literacy acquisition aid, participants didn’t have detailed ideas about how they would use DaZiBao after the study. Teachers and students asked for additional guidance in terms of how it could be used effectively in their process of teaching and learning Chinese literacy. Like participating teacher Ru, several students [Everlyn, Lily, Luke, Ashley, and Evan] agreed that DaZiBao could be used to practice and review characters and Luke said he might, “write a sentence in English and figure out then how to write them in pinyin, and then watch the characters coming up.” In other words, their own ideas consisted of ways it could help them do the things they had done previously within the bounds of traditional approaches to character learning.

Discussion and implications

Our findings advance and inform the field of CFL/CSL by demonstrating the capacity of novice CFL learners to acquire literacy via scaffolded communicative writing (Godwin-Jones, 2018; Vygotsky, 1978). Specifically, the capacity to recognize characters via time-sequenced multimodal scaffolds within the writing process is a synthesis and extension of the literature on technology-supported Chinese language and literacy development (Chung, 2003, 2007; Craik & Lockhart, 1972, Curtain et al., 2016; Lee & Kalyuga, 2011b; Shen, 2004). This holistic approach, embedded in the design of DaZiBao, improved participants’ ability to identify characters over a single compositional episode without relying on memorization-based lexical drills and afforded learners the chance to connect new-to-them characters to already-known-to-them pronunciations of Chinese words (Clark & Graves, 2005; Wong, 2017; Youngbing, 2005).

In general, our participants were as frustrated with the lived experience of learning to read and write in Chinese as their peers in other studies have reported being (Huang, 2006; McGinnis, 1999; Wang, 1998; Zhao, 2009). Given the lack of easily discernible phonetic and semantic clues, the amount of transferable linguistic knowledge between English and Chinese is modest at best. Confronted with having to memorize hundreds of complex character shapes one by one in a repetitive process of writing and rewriting the character or going through piles of flashcards, it is understandable as to why students and their teachers often avoid the task of learning characters. If Chinese and Taiwanese native speakers find learning and producing characters to be difficult (McBride-Chang & Ho, 2000) how much more difficult is it for students to acquire Chinese literacy as a second or foreign language?

Supports such as audio, images, and translation have been used in varying ways for hundreds of years in support of language acquisition (Otto, 2019; Shrum & Glisan, 2015). Moreover, multimodality factors heavily in the development of L1 and L2 literacies. Research studies since 2000 suggest that the simultaneous provisioning of robust language supports may not be in the students’ best interests because it runs the risk of affording immediate and undue focus on the L1 translation instead of on the L2 character (Lee & Kalyuga, 2011a). Given the sequenced multimodal scaffolds recommended by Chung (2003, 2007) and afforded to participants via DaZiBao, novice learners of Chinese did indeed use their L2 oral vocabulary and pinyin knowledge to express themselves via Chinese characters while simultaneously developing their L2 literacy.

Via this study we contribute both insights and potential pathways toward future inquiry into Chinese L2 literacy development. Our research also led to considerations for future intervention and application design iterations. In our study, participants employed limited pre-intervention approaches to learning Chinese characters—mostly centuries-old memorization-centric strategies they deemed tedious (Wang, 1998). This indicated a design opportunity to provide better supports for experiencing and learning Chinese writing, particularly platforms and tools that support learning via authentic communication.

We worked to meet this opportunity via a techno-pedagogical pivot (Olmanson et al., Olmanson, Kennett, et al., 2015). Our findings lead us to see promise in platforms similar to ours, tools that go beyond individual character memorization and instead place the focus on expression while enabling users to leverage and connect to learning they have already done. These types of tools and applications make it increasingly possible for Chinese language teachers to integrate Chinese literacy development into instruction earlier and in more authentic and meaningful ways (Fan, 2003 in Chen et al., 2013; Sung & Wu, 2011; Youngbing, 2005).

By leveraging previous work on character learning via sequenced multimodal supports (Chung, 2003, 2007; Shen, 2004), we were able to instantiate a design pivot that proved to be supportive without becoming a growth-stunting crutch. Participants relied on their pinyin knowledge and used pronunciations to winnow and identify their intended character/word. They used images to connect word meaning with characters, and they used translations to confirm and/or learn the semantic nature of the character. Our designed intervention with timed sequenced multimodal supports (Shrum & Glisan, 2015) was both a source of in-the-moment participant frustration and, upon participant reflection, was deemed supportive of their literacy development. DaZiBao enabled open-ended writing tasks and supported participants in communicative writing that historically has proven to be highly challenging or impossible for novice students of Chinese. For our student and teacher participants, who commonly position L2 writing as isolated character practice, the task of composing a text made up exclusively of Chinese characters as a way to express themselves and acquire literacy was new. The multimodal supports, designed in a research-based, pedagogically sequenced way, enabled participants to concentrate on the holistic task of expressing themselves. These multimodal scaffolds may assist learners in part by leveraging early 21st Century multimodal literacies developed over time via interacting with social media (Godwin-Jones, 2018).

The intervention created unexpected opportunities for metacognitive noticing (Chappelle, 1998). Student data aligned with our design rationale in that participants often used fewer supports than were available in selecting their intended word. In many cases, both waiting for extra support and selecting a character without waiting served to raise participant awareness that they were acquiring literacy. Our students reported that this gave them encouragement and demonstrated that they knew more than they thought they did. Previously, most students reported using flashcards for self testing and direct translation via Google translate or a dictionary app for composition. Those technologies do not offer opportunities for students to become meta-aware of their literacy development at multiple, incremental, grain-sizes. Moreover, in several cases wherein participants used the same character multiple times during the intervention they became aware that they were making literacy gains in the course of composing a single Chinese text. Students reported, and their texts and screencasts suggest, that the intervention both reinforced what they knew and proved to be a catalyst for meta-awareness in terms of self-assessment of their Chinese literacy development.

One of the affordances of using browser-based text-to-speech, a third-party image repository API, and a web-based translation API is that such an application can be lightweight while offering pronunciations, images, and translations for practically all possible Chinese words no matter how uncommon. One constraint however is that the images cannot yet be feasibly curated. Thus, there is no guarantee that returned images in each search will be relevant and age-appropriate. For example, concrete words like horse, mother, flower, school, and smile had a higher probability of returning images that are recognizable to users, while returned images for words like: because, with, philosophy, understand, and very were deemed less helpful. Regarding image age-appropriateness, despite being shown some images that could be interpreted as sexually-explicit, our adult participants did not say they had issues with the displayed images. We however see image age-appropriateness as a vital issue and are working to address this before we test the platform in US K12 classrooms.

Plans for the second iteration

Based on student and teacher participant feedback as well as our own observations of application use, we have identified three areas for the redesign and development of DaZiBao (Olmanson & Alshorman, 2021). First, participant feedback—from both students and teachers—stating they intended to use it in the future to simply practice words suggests that both groups would benefit from additional curricular and pedagogical support in order to better understand the application’s unique affordances. To that end, we are in the process of designing a Chinese literacy curriculum guidebook for teachers and beginning students that is supportive as they make this techno-pedagogical pivot (Fishman et al., 2013). This curriculum guide shows teachers how their students can use DaZiBao to complete written homework with characters as soon as they have covered those words and phrases orally in class. It offers students a pathway to literacy development based on self-expression that aligns with the curriculum their teacher is using. Additionally, we are creating resources for beginning students on how to use pinyin as a method of written communication and digital input.

Second, we have identified a number of application features that, when implemented, we anticipate will increase student usability and/or support learning. In terms of usability, we plan to increase the composition display space from a single left-to-right scrolling textbox that limited how much of their composition users could see at once (see Figs. 3, 4, and 5) to a multi-line textbox that scrolls vertically (see Fig. 9). Additionally, we plan to add a character-to-speech feature affording students the opportunity to have highlighted portions of their composition read back to them (see icon to the left of the multi-line composition space: Fig. 9). We are also implementing a feature that allows students to copy their composition into device memory with a click or tap (see icon to the right of the multi-line composition space: Fig. 9).

Fig. 9

Prototype of the second design iteration (Olmanson and Al Shorman, 2021)

Part of expressing one’s self in a new written language involves incorporating words that go beyond the vocabulary a student learns in a course. To that end, we are working to design a feature that affords students the opportunity to dictate or type in an English word which will be converted into pinyin for words they cannot pronounce—and do not know in Chinese—with a built-in timed delay so they are less likely to use it for words with which they are familiar.

Additionally, we are working on scalable ways to ensure the images used in the app are likely to be deemed age-appropriate for K-12 learners in US public schools. We have prototyped a way to filter out images involving nudity and gore by using the Azure Computer Vision API and setting the raciness score threshold to 0.169 (Microsoft, 2020). However, accessing this service, even with a paid account, is capped at 10 requests per second, meaning that, according to our testing, it will only support three simultaneous users—as the API will not return assessment results fast enough to properly filter each learner’s images.

Third, based on educator requests, we have started designing an administrative backend that affords researchers, teachers, and students access to: learning analytics and data visualization features to track their progress, tools to send their compositions via email, WeChat, or to google docs, and options to organize students by teacher or class.

We plan to study the second iteration of the application via a mixed-methods embedded design to test the effectiveness of DaZiBao to support literacy development over the course of an entire semester or K-12 school year (Bakla, 2020) (see Fig. 10). Using telemetry data we plan to measure: the number of supports used per word each time the word is selected by a particular student, the readability/reading level of the texts students produce over time, and the number of compositions as well as composition length over time. We also plan to collect screen recordings of participants using the app. Additionally, we will use pre-intervention data in the form of character recognition inventories and surveys of participant learning goals and attitudes about learning Chinese. Finally we plan to administer the American Council on the Teaching of Foreign Languages' Writing Proficiency Assessment to measure student proficiency after the semester (Surface & Dierdorff, 2004) as well as hold semi-structured interviews.

Fig. 10

Visual model of the proposed mixed-methods research design for the second iteration of the application

Via our findings, we seek to advance and inform the field of education technology design and development by offering and demonstrating the utility of two constructs, namely the filtration of innovation effect and the techno-pedagogical pivot. Theoretically and methodologically, the history of design-based research is one of small-scale iterative improvements over time rather than disruptive change (Anderson & Shattuck, 2012). Yet this may be more about the influence of filtration of innovation effects on the initial designs rather than a limitation of design-based research itself. The techno-pedagogical pivot approach to the initial design of DaZiBao and our ongoing awareness of the unique design decisions we made based on pedagogical and curricular factors during present and subsequent redesigns pair well with a design-based research framework. Based on our study, DaZiBao hastens how early in the learning process students are able to compose with Chinese characters, it diversifies the practices students can use to acquire literacy in Chinese. Additionally, it changes the amount, type of effort, and support teachers need to give their students during the literacy acquisition process (Fan, 2003 in Chen et al., 2013; Sung & Wu, 2011). These designed shifts represent new pathways for learning difficult, voluminous language content (Shen, 2004; Wong, 2017; Wong et al., 2010 in Sung & Wu, 2011), paths that are untethered to repetitive, disconnected lexical memorization.

Conclusion

In this study we inquired as to the extent that a scaffolded digital Chinese writing platform can change the process and experience of L2 writing and Chinese character recognition for novice CSL/CFL students. To do so, we tested a web application that we designed and developed as a techno-pedagogical pivot for Chinese literacy development. We drew upon the empirical L2 Chinese literacy acquisition literature but focused our design efforts on communicative pedagogical approaches to language development instead of traditional isolated lexical character learning. Additionally, in our design we marshalled the multimodal affordances associated with multimedia flashcards to support student expression over explicit character learning. Moreover, participant interactions with fewer and fewer supports led to meta-awareness of literacy growth without instructor feedback or test scores.

As multimodal social platforms are an essential way language learners express themselves and communicate, educators in the Chinese language classroom would do well to rethink writing instruction by implementing more activities using multimodal, informal, open-ended, expressive, educational technologies (Godwin-Jones, 2018; Li & Storch, 2017). While L2 Chinese literacy development will always pose unique challenges, an awareness of filtration of innovation effects, technological advances, and the growing literature on character learning afford the potential for educators, researchers, and designers to pivot toward more communicative, authentic, longitudinal, rhizomatic, and meaningful approaches to learning.

Education technology designs are impacted by societal expectations and institutional positions on multiple elements including: curriculum, learning theory, pedagogy, assessment, power, economics, and motivation. When we constrain our education technology development by designing within the bounds of societal expectations, it curbs the innovative curricular and pedagogical potential of the desinged technology and risks the continued homogenization of the learning experience. Strategically designing from divergent positions on one or several socio-educational meta-elements, in a way that also considers the current educational literature in related fields, offers a path toward the design and development of educational technologies that support learning in unique, divergent, heterogeneous ways.