The number of computer assisted language learning (CALL) products has been increasing enormously during the past decade, and a variety of CALL products have been produced in many languages, along with new CALL technologies and pedagogies (Harrison, 1998; Vincent & Hah, 1996). Regarding the Japanese language, various types of Japanese language tools – such as 1) image databases, 2) English-Japanese conversion dictionaries, 3) on-line tutorials, 4) kanji, kana, conjugation, and particle exercises, and other grammar-related exercises – have been introduced (e.g., Fukada & Hatasa, 1996; Hatasa, 1991; Inoue, 1998; Nagata, 1996; Nagata, 1998; Nara & Hirata, 1990; Tsuchiya, J, 1998; Yang & Akahori, 1999).
The need for evaluating the effectiveness of such products has been called for (Dunkel, 1991; Schuwartz, 1995), and it has been reported (Nagata & Swisher, 1995; Yang & Akahori, 1999) that products that give immediate metalinguistic feedback provide better results than those that lack such feedback. It has also been pointed out (Kulhavy & Wager, 1993) that CALL studies need a detailed analysis of learner errors before error-specific feedback strategies can be designed.
The present study will assess typing error tendencies in katakana, one of the three writing systems in Japanese, in order to develop software programs that provide detailed feedback to learners. This investigation of typing errors uses two katakana typing exercise programs developed by the researcher. The next chapter will outline previous studies of CALL, followed by the purpose of the present study.
CALL products have been used since the 1960s, and instances of their use have been increasing rapidly during the past decade (Harrison, 1998; Warschauer & Healey, 1998). The role of CALL has significantly changed along with the development of technologies and pedagogies. Warschauer and Healey (1998) divide the history of CALL into three stages: behavioristic CALL, communicative CALL, and integrative CALL.
Behavioristic CALL. In the 1960s and 1970s, computer technologies allowed learners to work at an individual pace. In this paradigm, CALL products were influenced by behavioristic learning approaches, and thus, the popular CALL products were drill-and-practice programs focusing on grammar and text manipulation.
Communicative CALL. The next stage, the era of communicative CALL, emerged in the late 1970s and early 1980s, at which time stand-alone machines were used (footnote: Stand-alone machines here refers to computer without multimedia hardware and software). During this period, many teachers shifted to a cognitive view of communicative teaching. Hence, communicative CALL products focused mainly on text reconstruction programs and simulations.
Integrative CALL. With the advent of the multimedia era in the late 1980s and early 1990s, the Internet spread out broadly, and courseware in foreign languages with sound, images, and animations was widely developed. Garrett (1987) suggests that "the integration of CALL and classroom activities can be mutually supportive and result in more efficient learning" (p. 194). Teachers shifted to a socio-cognitive view and focused more on task-based, project-based, and content-based approaches. The development of technologies and pedagogies allows us to integrate computer technology into the language learning process and to provide new possibilities of learning tools, such as on-line tutorials, e-mail communications, and electronic dictionaries.
While CALL products were widely dispersed in the mid-1990s, the need for evaluations of such products was also stressed (Schwartz, 1995; Warschauer, 1997). Schwartz (1995) examines the history of CALL and points out that many CALL products were developed without research and that they were used without teacher training. He claims that researchers must be careful not to rely heavily on student evaluation and that the products must be thoroughly reviewed in order to determine their effectiveness for pedagogical purposes and their ease of use for language learners. Armstrong and Yetter-Vassot (1994) gives an overview of technological resources for language learning and their pedagogical benefits. They claim that CALL products should be developed based on current learning theory. Warschauer (1997) analyzes current research on computer-mediated collaborative learning and stresses the necessity of a broad research in order to obtain a better understanding of the social, affective, and cognitive processes involved in computer-mediated collaborative learning.
Many studies comparing computer assisted language instruction with traditional classroom and language lab instruction, however, showed no significant difference between the two (Pederson, 1987; Morrison & Adams, 1968; Barrutia, 1970; Mellgren, 1983; Kleinmann, 1987). Pederson (1987) reviewed some of the previous studies from the late 1960s to mid-1980s which compare 1) computer assisted language instruction and language lab instruction (Morrison & Adams, 1968), 2) CALL products and traditional classroom learning (Barrutia, 1970), and 3) computer assignments and written assignments (Mellgren, 1983). These studies found no significant difference in learning. Kleinmann (1987) examined 20 commercial and one non-commercial CALL products by comparing two groups of classroom learners: one that received computer assisted instruction and one that received traditional classroom instruction. He found no significant difference between the two groups and claims that most of them were drill and practice in nature and little more than electronic textbooks.
Other features, such as immediate feedback, could be included in CALL products for effective language learning (Nagata, 1996). In fact, Teichert (1985) examines the effectiveness of CALL products with immediate feedback in comparison to conventional workbook exercises for learning German vocabulary and grammar. His findings support the effectiveness of immediate feedback; the experimental group which works on the CALL products with immediate feedback achieves a significantly higher mean on the post test, and he suggests that CALL products with such a feature may increase instructional effectiveness.
The history of computer feedback started with the provision of simple error messages (Garrett, 1987). Garrett (1987) examines the issue of grammar acquisition in CALL from a psychological perspective and reviews conventional error analysis and feedback. She claims that most software up to mid-80s is not notably successful for second language learning because most is "wrong-try again" type models that do not indicate how and why learners’ input is not correct. She claims that providing immediate and helpful feedback is a major advantage for behavioristic and cognitive reasons.
In the mid 1980s, researchers became interested in the effectiveness of consciousness-raising activities in second language classrooms (Rutherford & Sharwood Smith, 1985). Rutherford and Sharwood Smith (1985) define consciousness-raising as a "deliberate attempt to draw the learner’s attention specifically to the formal properties of the target language" (p.274). Sharwood Smith (1993), who later abandoned the term, consciousness-raising for input enhancement, claims that providing metalinguistic rules helps learners notice the input. Since the early 1990s, studies have widely reported that instant detailed feedback helps students learn grammatical structures effectively in the process of second language acquisition (Armstrong & Yetter-Vassot, 1994; Carroll & Swain, 1993; Carrol, Swain & Roberge, 1992; Doughty, 1994; Lyster & Ranta, 1997; Tomasello & Herron, 1988).
CALL researchers have also examined different types of CALL feedback (Robinson, 1991) and pointed out the need for detailed metalinguistic feedback in CALL products (Conrad, 1996; Robinson, 1991), and many CALL products with detailed feedback have been produced (Nagata & Swisher, 1995; Yang & Akahori, 1999; Yoshii & Milne, 1994, 1995). CALL products with immediate feedback were often reviewed in comparison to other types of CALL products to determine the effectiveness of immediate feedback (Conrad, 1996; Nagata, 1993; Nagata & Swisher, 1995; Yang & Akahori, 1999) or to traditional classroom instructions (e.g. Nagata, 1996; Teichert, 1985).
In a study which compares CALL products with detailed feedback and ones without, Nagata and Swisher (1995) investigated the effectiveness of metalinguistic instruction through computer feedback. They compare the effectiveness of two types of computer feedback for teaching basic Japanese sentence construction: traditional computer feedback that simply tells learners which word or letter is missing and intelligent computer feedback with detailed metalinguistic explanations. They find that when the grammar structure is complicated, detailed explanations are more helpful. These findings are supported by Yang and Akahori (1999). In their study on Web-based Japanese CALL systems developed by the authors, they compare two types of feedback systems: an intelligent feedback system that provides detailed grammatical explanations and a traditional one that only shows the correct answer, using students’ evaluations of the programs. They also find a significant difference between both systems, favoring the intelligent feedback system.
Another group of studies compared CALL products with immediate feedback and traditional classroom instruction. For example, Nagata (1996) examines whether a CALL program with detailed immediate metalinguistic feedback obtains a better result on performing grammatical skills than a traditional writing workbook in second language learning at the college level. She focuses on the problem of learning the Japanese particles (ga, o, wa, ni, and de). Her findings support the effectiveness of the use of the CALL program, and she concludes that detailed metalinguistic feedback can help students notice their errors consciously, learn grammatical rules, and improve their performances.
While such research provides strong evidence for the positive effects of detailed feedback, the need for error analysis to design feedback is also stressed (Kulhavy & Wager, 1993). Kulhavy and Wager review more than 50 CALL studies on feedback. They find no consistent pattern of results, thus proposing the need for a detailed analysis of various errors made by learners before designing a feedback strategy for correction of specific errors.
One way of analyzing errors would be to find out error tendencies for designing appropriate feedback. The next section will present the objectives of the study in detail.
The purpose of the present study is to assess the typing error tendencies in katakana loanwords in order to develop software programs that provide detailed feedback to learners.
Katakana is one of the three writing systems in Japanese, and is used mainly for foreign origin words (loanwords), as well as for onomatopoeia, colloquialisms, and for emphasis (Kang & Maciejewski, 1996). In order to transcribe foreign words in katakana as loanwords correctly, L1 English speaking learners of Japanese need to know phonological and other adjustments that occur (Ohso, 1991; Tsuchida, 1995). For example, they need to know how to adjust English vowel and consonant sounds into Japanese sounds. In addition, they need to know in which environments gemination or vowel lengthening occurs, and which vowel is inserted to break up consonant clusters. With respect to other adjustments, learners need to learn abbreviations and multiple spellings on the case-by-case basis as well as special letter combinations for katakana loanwords.
In order to help learners study katakana loanwords, many transcription rules have been introduced in terms of phonological and other adjustments during the past decade (Inagaki, 1991; Kobayashi, Quackenbush, & Fukada, 1991; Makino & Tsutsui, 1995; Ohso, 1991; Quackenbush & Ohso, 1990; Tsuchida, 1995). At the same time, a number of katakana exercise products have been produced, many of which can be downloaded from the Internet. For example, MIT has developed web-based katakana character exercises for its undergraduate students. Purdue University has also introduced a series of on-line katakana tutorials for listening, reading, and typing exercises. None of these products, however, provides detailed feedback when learners make an error.
In addition, no previous studies analyzed typing errors in katakana loanwords. With respect to error analysis of writing, Kobayashi, Quackenbush, and Fukada (1991) collect and analyze learners’ errors on katakana loanwords in order to investigate the acquisition process of these words and to investigate how effectively beginner-level learners acquire transcription rules. Each word is divided into segments, and each segment is categorized according to a set of fixed katakana transcription rules such as vowel insertions, geminations, and vowel selections; each of these rules is then subdivided into a number of more precise rules. Although they show a correct answer rate for each rule, they do not look at what types of error tendencies exist in each rule. As another example of writing error analysis of katakana loanwords, Yamagata (1999) analyzes learners’ katakana transcriptions in terms of geminations and the liquids [l] and [r]. She investigates what types of rules they use in their interlanguage to see if there is a relationship between the rules they apply and their proficiency levels. Nonetheless, her study focuses on learners’ interlanguage rules, not on error tendencies, and she does not classify the learners’ errors for each word into subcategories, which might shed light on how they made errors. Moreover, both Kobayashi, Quackenbush, and Fukada and Yamagata’s studies are based on writing errors, not on typing errors, and their results may not be applied directly to typing error tendencies.
Regarding typing errors, J. Tsuchiya and others have conducted error analyses on kanji readings and on the use of particles, long vowels, geminates, and voiced versus voiceless consonants. This work has been done for the development of a Japanese-English conversion dictionary for learners of Japanese (Tsuchiya, J. & Sugita, 1998; Tsuchiya, J & Tsuchiya, C., 1998; Tsuchiya, 1998). Based on the error tendencies found by these studies, these researchers have stored a series of potential typing errors which learners might make into a computerized dictionary program. When a learner makes an error, the program shows an alternative word and asks the learner if it is the word he or she wanted. Katakana loanwords, however, were excluded in these studies since error characteristics in loanwords are different from those of Japanese origin words (wago) and Chinese origin words (kango) (Tsuchiya, 1998).
To sum up, none of the existing katakana products provide learners with detailed feedback upon making an error. Moreover, no previous studies analyzed typing errors in katakana loanwords. The present study will thus examine the typing-error tendencies in katakana loanwords in order to design katakana software with detailed feedback. It will first summarize katakana transcription rules in the next chapter before presenting the methodology of this study in Chapter 4. After the methodology, it will report the findings in Chapter 5, by dividing common typing errors into categories.
Based on the results, samples of detailed feedback will be provided in the discussion (Chapter 6). Finally, Chapter 7 will summarize the present study and provide suggestions for further research.