UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Design of a computer-assisted speechreading training system for Japanese Fujiu, Masako 1989

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata


831-UBC_1989_A6_7 F84.pdf [ 7.29MB ]
JSON: 831-1.0097471.json
JSON-LD: 831-1.0097471-ld.json
RDF/XML (Pretty): 831-1.0097471-rdf.xml
RDF/JSON: 831-1.0097471-rdf.json
Turtle: 831-1.0097471-turtle.txt
N-Triples: 831-1.0097471-rdf-ntriples.txt
Original Record: 831-1.0097471-source.json
Full Text

Full Text

DESIGN OF A COMPUTER-ASSISTED SPEECHREADING TRAINING SYSTEM FOR JAPANESE by / MASAKO FUJIU BA. , International Christian University, Japan,1982 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES SCHOOL OF AUDIOLOGY AND SPEECH SCIENCES We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA June, 1989 . © Masako Fujiu, 1989 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. 1 further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of Audiology & Speech Sciences The University of British Columbia Vancouver, Canada Date May 29, 1989  DE-6 (2/88) i i ABSTRACT Speechreading is known to play an important role in speech perception, particularly for hearing-impaired individuals. Computer-Aided Speechreading Training (CAST) systems have been implemented for English and French, to train adults with mild-to-moderate hearing loss to improve their communicative ability with an effective use of visual speech information. Recent tests have shown that the CAST system is effective in improving speechreading performance. The purpose of this study is to design a similar system for Japanese, which will be called CAST-J. CAST-J follows the basic principles used for the English and French versions of CAST: (1) it uses the method of discourse tracking, in which training is focused on the perception of ongoing speech; (2) it aims at training subjects to integrate visual information, auditory information, and linguistic redundancy of the message; (3) it is designed for post-lingually hearing-impaired adults. Adults with cochlear implants are also considered as a possible target population for CAST-J, as long as they use auditory information of speech as a primary source in speech perception. The lessons of CAST-J are organized around the concept of viseme. Fifteen visemes of Japanese have been determined for this study, more specifically, five vowel visemes: { i }, { e }, { a }, { o }, and { u }, seven consonant visemes: { p, b, m }, { w, 0 }, { r }, { s, z, t, d, n, ts, dz }, { k, g, h }, { f }, and { j, q, J, tf, d3, ji }, and three haku visemes: { N }, { Q }, and { R }. The four sections of each lesson are designed (1) to review previously taught visemes; (2) to practice the recognition of a new viseme; (3) to practice new and old visemes by the tracking method; and (4) to recap the lesson. The lessons progress from easy-to-recognize visemes to difficult-to-recognize visemes. There are sixteen lessons, and two extra lessons are available for pre-and post-training comparison. iii The hardware required by CAST-J includes a microcomputer, a video disc player, a high resolution video monitor, and a printer. The system allows the instructor (an audiologist) to select the appropriate speaking rate for the video (fast or slow), whether the trainee will practice with or without audio signal, and to skip if wanted the optional lessons or part of a lesson, depending upon the level of the trainee. iv TABLE OF CONTENTS Page ABSTRACT ii LIST OF TABLES vi LIST OF FIGURES vii ACKNOWLEDGEMENT viii NOTATION ix 1. INTRODUCTION 1 2. SPEECHREADING IN AURAL REHABILITATION: STATE OF THE ART 4 2.1. Speechreading and speech perception 4 2.2. Speechreading: evaluation and training 7 2.3. Speechreading and auditory rehabilitation in Japan 9 2.4. Speechreading for the individuals with cochlear implants 10 3. TARGET POPULATION 13 4. VISEMES 17 4.1. Concept of viseme in general 17 4.2. Visemes for Japanese 22 4.2.1. Phonological and phonetic characteristics of Japanese 22 4.2.2. Significance of haku (mora) in Japanese 27 4.2.3. Determination of visemes of Japanese 33 5. EQUIPMENT AND INPUT/FEEDBACK MODE 44 5.1. Equipment 44 5.2. Input and feedback modes 45 5.2.1. Kanji, kana, and romaji 45 5.2.2. Input 47 V 5.2.3 Feedback 48 5.3. Orthographic rules for kana - romaji conversion 50 6. LESSONS 53 6.1. Course structure 53 6.2. Lesson structure 54 6.2.1. Review of previously learned visemes 54 6.2.2. Practice of a new viseme . 54 6.2.3. Practice of old and new visemes in discourse 55 6.3. Romaji - viseme conversion and target - response matching for feedback 58 6.4. Texts of the lessons 60 7. CONCLUSIONS 64 BIBLIOGRAPHY 67 APPENDIX A Orthographic Rules for Kana - Romaji Conversion 75 APPENDIX B Romaji - Visemic Code Conversion Rules 77 APPENDIX C Lesson Script and Corresponding Visemic Code 79 APPENDIX D Lesson Script and Literal Translation of Lesson 17 and Lesson 18 130 vi LIST OF TABLES Table Legend Page I Phonemes of Japanese and Their Allophones/Phones 24 II List of Hakus Written in Romaji 29 III List of Hakus Written in Phonetic Symbols 30 IV Examples of Loanwords Widely Used in Japanese 32 V Lesson-by-Lesson Inventory of Viseme Tokens 62 VI Viseme Inventory for Lesson 17 and Lesson 18 63 LIST OF FIGURES Legend Functional Diagram of the CAST-J System Writing System Used at Different Stages Conversion Rules viii ACKNOWLEDGEMENT I would like to express my appreciation to all those people who have contributed to this work: - Dr. Andre-Pierre Benguerel, for his considerate assistance and for the fine model he has provided as a teacher. - Noelle Lamb, for her valuable advice and assistance. - Kathy Pichora-Fuller, for her advice as the originator of CAST. - John Nicol, for sharing his expertise in computers. - Yumiko Fukuda and Ikuko Seki, for their advice regarding speechreading in Japanese. - My classmates, for their friendship. - My family and friends in Japan, for their support. ix NOTATION In this study, some punctuation marks and print faces are used for a specific situation, which is indicated on the right of each symbol. / / denotes a phoneme [ ] denotes a(n) (allo)phone \ \ denotes a viseme < > denotes a(n) (allo)phene I I denotes a grapheme « » denotes a(n) (allo)graph { } denotes a set of phonemes, visemes, etc. denotes a word with a special meaning " " denotes a quotation @ stands for "in the context of " bold type denotes emphasis of an important word S M A L L C A P I T A L S denotes a translation Italics denotes a romaji transliteration bold type italics denotes a kana / denotes separation between kana characters 1 CHAPTER 1 INTRODUCTION Computer-assisted devices for the communicatively impaired have been successfully used both in clinical and educational settings. Aural rehabilitation is one of the areas which can be refined by utilizing such devices. A Computer-Aided Speechreading Training system (CAST) has been developed by Pichora-Fuller and her colleagues as a clinical and research tool to train and assess the speechreading abilities of adults with a mild-to-moderate acquired hearing loss (Pichora-Fuller and Cicchelli, 1986; Benguerel and Pichora-Fuller, 1988; Pichora-Fuller and Benguerel, 1989). At present, English and French versions of CAST have been implemented. The purpose of this study is to outline the design and the lessons of a computer-aided speechreading training programme for Japanese (CAST-J). The CAST-J system provides an interactive environment for speechreading practice, by means of a microcomputer connected to a video disc player and a high resolution video monitor, as shown in Figure 1. The trainee views a video-recorded speaker talking on screen. He/she is given a target string (usually a phrase) that he/she has to speechread, either with or without auditory signal. Presentation of the target strings by the speaker is controlled by the computer. The trainee inputs his/her response to the computer via a keyboard. The response, echoed on screen, is analyzed by the computer, and feedback is provided by the computer on screen, in written form. If the trainee wishes to re-play the target string, or to move to another location in the lesson, he/she is able to do so by using various command keys. Although the speechreading trainer is the CAST-J system itself (both instruction and feedback are computer-controlled), the trainee still plays an active role. The data of the trainee's 2 performance are recorded, tabulated and analyzed, and they are printed out for further analysis. The detail of the system will be described in Chapters 5 and 6. This study includes a discussion of the current role of speechreading in aural rehabilitation, a survey of the linguistic characteristics of Japanese, particularly where relevant to speechreading, and a set of lessons. The actual video-recording of the lessons and training of the subjects are not included in the present study. Figure 1. Functional Diagram of the CAST System. VIDEO OUTPUT VIDEO DISC PLAYER COMPUTER CAST-J PROGRAM T VIDEO MONITOR AUDIO OUTPUT r \ AMPLIFIER & SPEAKER j c \ PRINTER V KEYBOARD 4 CHAPTER 2 SPEECHREADING IN AURAL REHABILITATION: STATE OF THE ART 2.1. Speechreading and speech perception Speechreading, a term broader than lipreading, refers to the "recognition of a speaker's words by watching his lips, facial expressions, gestures, etc., as well as using closure, rules of language, contextual cues, etc." (Sheeley, 1985, p. 1087). Besides the movements of the visible articulators (e.g. the range of jaw opening, the shape of the upper and lower lips, the appearance of the teeth, lingual movements (Fukuda, 1987), the receiver integrates various environmental and contextual cues. In general, speechreading itself is not sufficient for the interpretation of speech. However, it is now regarded as playing an important role even in normal speech perception, where it is integrated with auditory information (Massaro, 1987). The McGurk effect is a good example of this integration: people rely on visual cues even when auditory information is intact (McGurk and MacDonald, 1976). Nevertheless, speechreading has a greater importance in disambiguating speech when auditory information is degraded (Sumby and Pollack, 1954). Middelweerd and Plomp (1987) examined the effects of speechreading on speech-reception thresholds in noise. They found an improvement of approximately 4 dB, both in normal hearing subjects and in presbycusic subjects. Speechreading and auditory information are known to work synergistically in speech perception by the hearing impaired. Manner of articulation and voicing contrast are known to be better perceived through the auditory channel, while place of articulation fares better through the visual channel. (Erber, 1972; Binnie, Jackson and Montgomery, 1976; Fukuda, Sakamoto and Kuroki, 1976; Fukuda and Hiki, 1977). Information available through speechreading, however, is not sufficient when 5 place of articulation is in the middle or back of the oral cavity (McGurk and MacDonald, 1976). Recent studies in normal hearing subjects show that speechreading can be improved when supplemented with various types of auditory information such as presentation of the acoustic signal, low-pass or band-pass fdtered version of the signal (Breeuwer and Plomp, 1984), selected formant information (Breeuwer and Plomp, 1985), and/or voice fundamental frequency (FQ) information (Breeuwer and Plomp, 1986; Boothroyd, Hnath-Chisolm, Hanin and Kishon-Rabin, 1988). In the study of hearing impaired subjects, Grant (1987) reported that improvement of speechreading was observed with presentation of F„. Presentation of speech information to the cochlea, as an electrical rather than an acoustic signal (e.g. in cochlear implants), is another technique used to enhance speech perception. The significance of the use of cochlear implants in conjunction with speechreading will be discussed separately in Section 2.4. As pointed out by Massaro (1987), speechreading has a great potential in enhancing speech perception when it is combined not only with auditory information but also with other sensory information. One way of compensating for the limitations of speechreading is the manual presentation of speech cues. Manual systems developed in the past include the Mouth-Hand System developed by Forechhammer in Denmark (B0rrild, 1967), Cued Speech originated by Cornett (1970), and Q-Code proposed by Fant (1970) in Sweden. Another means of supplementing speech perception is visualization of voicing and manner of articulation by pulsing lights: the speechreader wears glasses which have several small lamps on the lenses (Upton, 1968), or the speaker wears glasses to which a ring-shaped frame with lamps is attached (Tratlnmuller, 1976, cited in Fukuda, 1987). Both devices provide the less visible part of speech information (i.e. voicing and manner of articulation) after it has been extracted from the speech signal by an electrical circuit. Tactile presentation of speech information has also been investigated by some 6 researchers. Speech information which has been made available through tactile cues includes F0 (Hanin, Boothroyd and Hnath-Chisolm, 1988; Hnath-Chisolm and Kishon-Rabin, 1988), resonance, envelope, and voicing contrast (De Filippo, 1984). In the studies of Hanin et al. (1988) and of Hnath-Chisolm and Kishon-Rabin (1988), the authors used two types of tactile display for F0: a single-channel (temporal) display and a multichannel (spatial) display. In the former case, F0 is coded as rate of vibratory stimulation and is presented to the pad of the finger. In the latter case, F0 is coded as a locus of stimulation and as a frequency of vibratory stimulation, and it is presented to the skin of the forearm. Hnath-Chisolm and Kishon-Rabin (1988) examined the effects of the two different types of display on the perception of initial consonant voicing, final consonant voicing, pitch change, and word stress in normal hearing subjects. The results indicate that the tactile presentation of F„ enhances the perception of speech pattern contrasts. The multichannel display is reported to be more effective than the single-channel display, at least for the perceptual contrast of pitch rise versus pitch fall. Hanin et al. (1988) examined the effects of the tactile presentation of F0 on the perception of sentences, both in normal hearing and hearing impaired subjects. Only the multichannel display was used for the hearing impaired, while both displays were used for the normal hearing subjects. Perception of sentences improved in all three conditions. De Filippo (1984) evaluated the spatial presentation of speech information (resonance, voicing, manner of articulation) to fingers and the temporal presentation of speech information (including spectral information) to the hand, the wrist, and the skin below the sternum. The subjects were adults who were congenitally deaf and adults with normal hearing who were made artificially deaf. All the subjects demonstrated an improvement of speech perception when speechreading and tactile information were combined. Although no significant difference was found between the two types of tactile presentation, De Filippo suggests that the temporal display with spectral information seems more efficient than the spatial display. 7 2.2. Speechreading: evaluation and training Currently, several tests for speechreading evaluation are available for English. Those tests evaluate the perception of vowels and consonants, words, sentences, and connected speech. The perception of vowels and consonants is evaluated in the lipreading screening test designed by Binnie et al. (1976). Tests of word perception include the use of PB words (Binnie, 1976), subtests of the Craig Lipreading Inventory developed by Craig (1964), and the Diagnostic Test of Speechreading, by Myklebust and Neyhus (1971). Sentence tests include the one developed by Utley (1946), the Craig Lipreading Inventory (1964), the Diagnostic Test of Speechreading (Myklebust and Neyhus, 1971), the CID Everyday Sentences Test (Jeffers and Barley, 1971; Johnson, 1976), the Jacobs Test (Sims and Jacobs, 1976), and the Lipreading Discrimination Test (Bement, Wallber, De Filippo, Bochner and Garrison, 1988). Comprehension of connected speech is most frequently assessed by the 'discourse tracking' described by De Filippo and Scott (1978). In addition, standardization of a test for cochlear implant recipients has been implemented by Spitzer, Leder, Milner, Flevaris-Phillips and Giolas (1987). Discourse tracking can be used for speechreading training as well. It consists of a procedure in which a 'receiver' (the trainee) repeats groups of words that are read or said by the 'speaker' (the trainer). In this procedure, the receiver is provided with immediate feedback by the speaker. The reception of ongoing speech is measured by calculating the number of words correctly repeated per minute. This method is one of the most popular ones, and applications to various populations of hearing impaired subjects are found in the literature (Lesner and Kricos, 1987; Owens and Raggio, 1987). Fenn and Smith (1987) suggested some modifications to the scoring system of the tracking method, claiming that despite the potential that the method has, the traditional tracking method has not been fully used for clinical purposes, due to some practical problems. They added a measure of fluency calculated as follows: the number of words correctly repeated per 8 minute is corrected by deducting 'penalty points', the number of unsuccessful attempts the subject made in identifying a word. When the subject fails to identify a word, the speaker is allowed to repeat the word a maximum of two times. If the subject does not succeed after these two trials, the word is written out and tracking proceeds. It has been shown by Fenn and Smith (1987) that the subjects are unlikely to identify the word correctly after more than two repetitions. A similar result has been shown by Gagne and Wyllie (1989): simple repetition of a test word did not help the subjects to identify the word. Therefore, besides being a valid parameter for assessing speechreading ability, this new scoring method can reduce the frustration of a poor speechreader who frequently has to spend so much time for one word. Van Uden (1983) proposes the use of visual monitoring of the learner's own speech via video replay for speechreading training: the subject sees himself speaking on the video playback and lipreads himself. In Van Uden's study of prelingually deaf children, this method was found to be particularly useful for those who had not shown progress in the previous training phase. The use of computers is another way to evaluate and train for speechreading. About a decade ago, the first computer-assisted video tape system for speechreading instruction was reported by a group of researchers at the National Technical Institute for the Deaf (NTID) as part of a computer-assisted instruction project. The system, named DAVID (Dynamic Audio Video Instructional Device), has been used successfully in the language laboratory situation, by NTID students who had a severe-to-profound hearing impairment. (Sims, Von Feist, Dowaliby, Hutchinson and Myers, 1979; Sims, Scott and Myers, 1982). While DAVID was designed for young adults with severe-to-profound hearing impairment, CAST (Pichora-Fuller and Cicchelli, 1986; Benguerel and Pichora-Fuller, 1988; Pichora-Fuller and Benguerel, 1989) is a system aiming at the pre-retirement adult with a mild to moderate hearing loss. The CAST lessons were constructed in accordance with the 9 principles of the tracking method: face to face interaction, with feedback between sender (in this case the video tape) and receiver (trainee). In the investigation regarding the effectiveness of the CAST program, a greater improvement of speechreading ability was found in the experimental group, i.e. eight normal-hearing adults who had been trained with CAST, than in the control group, i.e. eight normal-hearing adults who had not had any speechreading training (Gagne, Dinnon and Parsons, 1989). 2.3. Speechreading and auditory rehabilitation in Japan In Japan, auditory rehabilitation/habilitation has put its major emphasis on children with congenital or acquired hearing loss, for whom the acquisition of language is the primary concern. Generally, maximum use of residual hearing through amplification is the first goal of any approach involving children. Speechreading, like manual communication methods such as fingerspelling, sign language, or cued speech, is considered as a supplemental medium for auditory speech perception. In contrast to aural (re)habilitation for children, that for adults with an acquired hearing loss tends to put more emphasis on speechreading. Unfortunately, however, the topic of speechreading is given extremely limited space in the literature (Tanaka, 1980). As far as the development of assessment materials is concerned, very few studies are found in the literature. Training and evaluation of speechreading, in general, is totally dependent on the protocols of individual clinicians. In the 1970s, Tanaka and his associates developed a video system to assess speechreading ability with the goal of using the assessment results for an adequate hearing aid fitting (Tanaka, Shindo and Motomiya, 1973; Tanaka and Chiba, 1975). Despite the potential of the system for assessing adults, it was designed primarily for hearing-impaired children. Recently, a video system has been developed by Fukuda (1987) for assessing the speechreading ability of adults. However, it is specifically designed for adults with cochlear implants, and no application to other populations has been 10 attempted. 2.4. Speechreading for the individuals with cochlear implants A change in attitude towards speechreading in Japan has occurred in the mid-1980s with the development of cochlear implantation for post-lingually deaf adults. With cochlear implants, neurons of the auditory nerve are stimulated electrically through electrodes surgically implanted in the cochlea. These cochlear implants provide an auditory input to profoundly hearing impaired individuals who cannot benefit from conventional amplification, medication, and/or surgical treatment. The data reported throughout the world thus far appear to indicate that cochlear implants are a promising tool in aural rehabilitation. Presently, there exist several design options: unipolar versus bipolar electrode systems, single versus multichannel electrode systems, solitary versus multiple simultaneous electrode stimulation, transcutaneous versus percutaneous conduction, and various sound processing strategies. Common to all designs are the surgical implantation of the stimulating electrode(s) and the externally worn microphone and sound processor. Past research has shown that, in general, multichannel cochlear implants provide better speech recognition than single channel models, at least in post-lingually deaf adults (Eddington, 1980; Engelmann, Waterfall and Hough, 1981; Tyler, Tye-Murray, Preece, Gantz and McCabe, 1987; Youngblood and Robinson, 1988), although a successful case has also been reported with the single channel device (Tyler, 1988). Furthermore, continuing improvement in modern technology has allowed the use of implants in cases which had been considered contraindicated for implantation (Balkany, Gantz and Nadol, 1988). At the same time, however, the risk of possible medical complications should not be ignored (Cohen, Hoffman and Stroschein, 1988). Overall, improvement of hearing in the patients with cochlear implants appears to be encouraging: some patients have been reported to acquire the ability to use the telephone 11 (Parkin and Stewart, 1988; Youngblood and Robinson, 1988); hearing with a cochlear implant is described as superior to hearing with a hearing aid by patients who have undergone implantation (Youngblood and Robinson, 1988). It has also been reported that some people do well in open-set speech recognition tests, without lipreading (Gantz, McCabe, Tyler and Preece, 1987; Parkin and Stewart, 1988). Nevertheless, the majority of patients benefit most from the combination of auditory and visual inputs (Funasaka, Hayashibara, Takahashi, Yukawa and Hatsushika, 1987; Funasaka and Shiroma, 1987; Fukuda, Shiroma and Funasaka, 1988; Parkin and Stewart, 1988). In Japan, the first 22-channel cochlear implant was implanted by Funasaka, Hosoya, Hayashibara, Takahashi, Yukawa and Hatsushika (1986). In subsequent studies by Funasaka and his colleagues, the importance of speechreading in the whole rehabilitation process was emphasized (Funasaka et al., 1986, 1987; Funasaka and Shiroma, 1987; Shiroma and Funasaka, 1988). The patients who have undergone cochlear implantation in Japan are all adults with an acquired hearing loss. It is suggested that the ability to lipread be one of the criteria for selecting the candidates for implantation. The data for six subjects studied by Fukuda et al. (1988) exhibited enhanced speech perception three months post-operatively, when auditory and visual cues were combined, as compared with auditory cues only or visual cues only. The mean correct response scores for the above three conditions were respectively 55%, 20%, 16% for words, and 69%, 23%, 8% for sentences. Correlation coefficients indicated also the dominance of auditory perception in overall speech recognition (V + A and A: 0.72, V + A and V: 0.48). Thus, post-lingually deaf adults with cochlear implants use auditory input as a primary source of speech information, and they can increase their communicative ability to a great extent by using supplementary visual cues. Currently, cochlear implants are becoming an important option in the aural (re)habilitation of children with no or only limited residual hearing; however, no attempt to 12 perform implantation in children has yet been made in Japan. There are some data indicating the successful application of implantation of young children outside Japan (Eisenberg et al., 1983; Lyxford, House, Hough, Tonokawa, Berliner and Martin, 1988). Considering the importance of educational and developmental issues, however, this option needs to be assessed with special caution (Tyler, Davis and Lansing, 1987; Boothroyd, 1988; NIH, 1988; Simmons, 1988). 13 CHAPTER 3 TARGET POPULATION The original version of CAST was developed specifically for pre-retirement adults with acquired mild-to-moderate hearing loss (Pichora-Fuller and Cicchelli, 1986). The premises of their approach are that speechreading is "a communication task which entails three skills: visual speech perception, use of linguistic redundancy, and effective use of feedback between message sender and receiver" (p. 11). The aim of CAST is to improve speechreading, a supplement to auditory perception of speech which can be affected by various environmental factors such as background noise, speaking rate, familiarity with the topic, etc. The potential users of CAST, therefore, are those who use the auditory channel as the primary means of communication, and the visual channel as a supplemental one. Like the English and the French versions of CAST, CAST-J will aim at the post-hngually hearing impaired adults. The trainees will be expected to have a mild-to-moderate hearing loss and post-secondary-school level linguistic competency. Children are not included in the target population for this particular version of CAST, for the following reasons. First, the program for pre-lingually hearing impaired children must consider the course of cognitive, phonological, and other linguistic development of children. CAST-J, which is designed for those who can use linguistic redundancy effectively, is thus not ideally suited for pre-lingually deaf children. Second, Luxell and Ronnberg (1987) report that skilled guessing is critical for two speechreading conditions: (1) when messages have little contextual information, and (2) when longer messages are to be speechread. Third, the study by Berger (1972) indicates that a person with an early onset of deafness tends to be poorer at speechreading. Conrad (1979) reports a positive correlation between speechreading ability 14 and intelligence in pre-lingually deaf children: the more intelligent a child, the better his use of linguistic knowledge for speechreading. Conrad also points out the importance of establishing a phonological code in memory as well as acquiring other linguistic skills such as reading. Campbell (1987, p. 225) claims that speechreading is "a component set of functions that are language dependent and are localized in the left hemisphere". Fourth, CAST-J is inappropriate for post-lingual children in terms of word familiarity. If the linguistic level of a child is advanced enough, a separate CAST-J version using children's vocabulary could be useful, but since this study is only the first stage of a Japanese program, lessons specifically designed for children will not be considered. CAST-J will consider cochlear implant patients as primary potential trainees for three reasons: (1) ability to lipread is considered to be an important criterion for potential candidates to the surgery, (2) adults who have undergone a cochlear implantation have been found to have relatively good auditory perception of speech, thus they can use speechreading as a supplementary medium for speech perception, and (3) past research has shown that these patients greatly benefit from speechreading training (Funasaka, Hayashibara, Takahashi, Yukawa and Hatsushika, 1987; Funasaka and Shiroma, 1987; Fukuda, Shiroma and Funasaka, 1988). Agreement on the above points is not unanimous though. Lesner, Sandridge and Kricos (1987), for example, investigated the effects of an analytic approach for speechreading, i.e. the effects of emphasizing identification and discrimination of visemes at the segmental level on sentence-level speechreading in normal hearing subjects. When the recognition of consonant visemes was assessed in VCV utterances without auditory input, the subjects having had analytic training performed better than those with no training. However, when the improvement of speechreading was obtained from the pre- and the post-training scores, the results did not show any significant difference between the subjects whose training included feedback and those whose training did not include feedback, whether their responses were correct or not. When the recognition of sentences was assessed without auditory input, there 15 was no significant difference in performance between the three groups of subjects tested, namely those who had been trained with feedback, those who had been trained without feedback, and those without training. Based on these results, Lesner et al. (1987) concluded that (1) the analytic approach for training speechreading is not useful, at least for those who use visual information of speech as the primary means of recognizing connected speech; (2) the ability to recognize visemes at the segmental level was not reflected in the recognition of sentences when the subjects totally relied on the visual information of speech; (3) the analytic approach may be only effective for those who can use auditory input as the primary source of speech information and visual information as the secondary one, as reported by Walden, Prosek, Montgomery, Scherr and Jones (1977) and Walden, Erdman, Montgomery, Schwartz and Prosek (1981). Lesner et al. (1987) contend that individuals with cochlear implants perceive speech by relying primarily on speechreading and use the auditory channel for supplemental purposes; therefore, they should not be candidates for training which places the emphasis on contrast of visemes. The present author believes that individuals with cochlear implants are still potential candidates for training with CAST-J for the following reasons. Firstly, recent research reveals an improvement of the auditory perception of speech by these subjects and the possibility of their using audition as the primary channel for speech perception. Secondly, CAST-J does indeed have some analytic aspects in its approach (i.e. lessons include recognition of visemes at the segmental level), but it also includes synthetic aspects in its approach: CAST-J combines phrase- and sentence-level materials in its lessons, emphasizing the integration by the trainee of various contextual and linguistic cues available in a given discourse. Thirdly, CAST-J is designed with a consideration of the coarticulatory effects. These effects are believed to be one of the main reasons why the ability to contrast visemes at the segmental level is not reflected in the sentence level as found in the study by Lesner et al. (1987). Lastly, a subject has a large amount of freedom in combining the two aspects depending on what is the most salient cue (e.g. visemic cue, contextual cue) in speechreading under certain circumstances. With the rising popularity and success of cochlear implants, implementation of CAST-J should meet the needs of the patients as well as those of the clinicians. 17 CHAPTER 4 VISEMES 4.1. Concept of viseme in general The term viseme, according to Fisher (1968), refers to any individual and contrastive unit of visually perceived speech. It is conceptually analogous to the term phoneme, a contrastive unit of auditorily perceived speech. In the glossary of a textbook of audiology (Sheeley, 1985, p. 1090), viseme is defined as "a group of speech sounds that all have the same appearance when spoken". The phonemes / p /, / b / and / m /, whose labial articulations look the same, thus constitute a viseme. Sheeley also uses 'visual phoneme' as a synonym for viseme. A "word that looks the same on the lips as another word" (ibid., p. 1073) is termed a homophene of that other word. For example, "pine" is a homophene of "mine", and these two words are said to be homophenous. The first attempt to discover a 'unit of lipreading' comparable to the phoneme appears to have been made by Woodward and Barber (1960). They proposed four categories of visually perceived English consonants (i.e. viseme groups): bilabial (p, b, m), rounded labial (M, W, r), labiodental (f, v), and non-labiodental (t, d, n, 1, 8, 5, s, z, c, j, s, z, y, k, g, h). In these definitions, visemes are considered to be a re-categorization of phonemes, mainly with respect to place of articulation, and they are not independent of phonemes. In order to refine the definition of the viseme, possible meanings of viseme in hypothetical situations will be discussed in comparison with some other 'units' of speech and language. Henderson (1985) discusses the use of the term grapheme, which is used with different meanings by different authors. Originally, a grapheme denoted a minimal contrastive unit in a writing system (Sense 1). For example, in hand writing, the letter « a » belongs 18 to the grapheme I a I, while the letter « g » belongs to the grapheme I g I. Both « a » and « a » belong to the grapheme I a I, while both « g » and « g » belong to the grapheme I g I. In other words, both « a » and « a » are members (allographs) of the grapheme I a I, and both « g » and « g » are allographs of the grapheme I g I. In this sense, the role of graphemes in written language is considered parallel to that of phonemes in spoken language. The term grapheme, on the other hand, is sometimes used to denote a letter or a cluster of letters that refers to or corresponds to a single phoneme in speech (Sense 2). For example, « p » in "hoping" and « pp » in "hopping" belong to the same grapheme I p I; similarly, « gh » in "laugh", « ph » in "graph", and « f » in "fish" belong to the same grapheme I f I. In other words, « p » and « pp » are allographs (Sense 2) of the grapheme I p I, whereas « gh » , « ph » and « f » are allographs of the grapheme I f I. This definition of grapheme is based on spoken language, with a view that written language is "a mere notation for representing speech sounds" (Henderson, 1985, p. 135). Henderson claims that Sense 1 is independent of the phoneme: a grapheme is "the minimal functional distinctive unit of any writing system" (p. 137), and "not in the phoneme-representing sense" (p. 137). Although the writing system of a language might have been developed on the basis of a correspondence with spoken language, written language can exist without speaking out the script. Languages such as Latin, classical Greek, and Sanskrit are examples of this type. Extreme cases are found in written forms in music or in the formulae of physical sciences, where the script does not always need to be pronounced. (Stetson, 1937 cited in Henderson, 1985). Therefore, graphemes (in Sense 1) can exist independendy, without any reference to phonemes, even though they are procedurally defined in a similar way to phonemes. Sign language is another type of language which exists without any reference to spoken or written languages. The minimal contrastive unit of sign language is called a chereme (Stokoe, Casterline and Croneberg, 1965). The grapheme (Sense 1) and the chereme 19 are respectively the minimal contrastive unit of written language and sign language, while the phoneme is the minimal contrastive unit of spoken language. In view of these definitions, how does one define the viseme? For this author, a viseme, just as a phoneme, is a unit of spoken language or speech. From the perceptual point of view, spoken language or speech can be considered from two perspectives: audible speech and visible speech. A phoneme is the minimal contrastive unit of audible speech, whereas a viseme is the minimal contrastive unit of visible speech. It is possible to conceive a type of communication that would be performed solely by means of the visual information provided by articulatory movements, i.e. without any auditory information. If a sociolinguistic group of speakers had to communicate by means of visible speech alone, without audition, a number of easy-to-see visemes would no doubt develop, apart from the regular phonology of audible speech. Phonemes (of audible speech) are determined primarily by three elements: place of articulation, manner of articulation and voicing contrast. Likewise, each viseme could be determined by parameters such as those used for speech articulations, or preferably, by similar parameters more appropriate for describing the visual aspects of these same articulations, e.g. the positioning of the articulators, the direction of articulatory movements, and/or the duration of articulation. When visemes are considered as a visual aspect of a regular spoken language (audible speech), the number of visemes is unlikely to exceed that of phonemes, because acoustic and phonetic characteristics of speech do not necessarily correspond to visually discemable features and it is not very likely that there is some visible characteristic of speech, which is not audible but is important and useful for speechreading. Since the visemes proposed in the past have been determined specifically for the purpose of speechreading or lipreading audible speech, which consists of phonemes, it would be hard to postulate that these visemes are independent of phonemes. The definition of the viseme, as used today, appears to be fairly parallel to the definition of the phoneme used in phonology. In this study, as in most past research, a viseme is considered as a minimal unit 20 of speechreading which corresponds to one (or more) phoneme of audible speech (phoneme-based speech). Each viseme is a set whose members will be called allophenes, and each allophene in turn is a set of physical realizations which will be called phenes. In order to denote visemes and their corresponding (allo)phenes, backward slashes \ \ and angular brackets < > will be used, in the same way as / / and [. ] are used for phonemes and (allo)phones. For a given language, it is not easy to define a complete and consistent set of visemes for several reasons. Great methodological differences are found in past studies on visemes. Some authors use normal hearing subjects, while others use hearing-impaired subjects; some use identification tasks, while others use discrimination tasks. Criteria for categorizing visemes vary from author to author. For example, the hierarchical clustering analysis developed by Johnson (1967) is commonly used as a criterion for grouping visemes. It is a technique which combines objects (e.g. vowels, consonants) into optimally homogenous groups on the basis of measures of similarity (or confusability) between objects. By using a confusion matrix, one can obtain percentages of how often a particular item was confused with others. The higher the percentage of confusion, the more similar the items are. In a viseme recognition task, if a consonant was frequently identified as (or confused with) another consonant, these consonants both belong to the same visemic group (or a cluster). A 65% or higher within-cluster response rate was employed by Montgomery and Jackson (1987) to group vowel visemes, whereas a 75% or higher within-cluster response rate was employed by Walden et al., (1977, 1981) and by Lesner et al. (1987) to group consonant visemes. There are other reasons for difficulties, in addition to methodological differences: (1) coarticulatory effects may lead different investigators to partition differently the 'visemic' space into visemes; (2) the visibility of articulation varies from speaker to speaker (Benguerel and Pichora-Fuller, 1982; Montgomery et al. 1987); (3) speechreading training influences the ability to contrast visemes at the monosyllable level. Walden et al. (1977) report that hearing impaired adults, who could 21 discriminate five visemes prior to speechreading training, were able to discriminate nine visemes after training. The visemes they obtained pre- and post-training are shown below. Pre-training { < e >, < a > } { < f >, < V > } { < p >, < b >, < m > } { < J >, < 3 > } { < w > } Post-training { < 0 >, < 5 > } { < f >, < v > } {<p>, <b>, <m>} { < J >, < 3 > } { < w > } { < s >, < z > } { < t >, < d >, < n >, < k >, < g >, < j > } { <r> } { <1> } The number of English consonant visemes reported in the past ranges from 3 to 12 (Jeffers and Barley, 1971). For vowel visemes, Montgomery and Jackson (1983) reported 8 visemes, including diphthongs, while Nitchie (1979) reported 9 visemes. Visemes reported in these two studies are shown below. Montgomery and Jackson (1983) { < i >, < i > } { < U >, < 3 > } { < U > } { <a>, <o>, < A > } { < au> } Nitchie (1979) narrow { < i > } relaxed narrow { < i > } small, oval { < u> } puckered { < u > } wide { < a > } elliptical { < A > , < a >, <3> } 22 { o i > } spherical triangle { < o >, < o > } { < ou > } { < e, ae, ei, ai > } large oval { < ae > } medial oval { < e > } Despite the absence of a set of universally accepted visemes for English (or for any other language), there is a generally accepted hierarchy of visemic groups, from well-defined to less well-defined visemes (Pichora-Fuller, 1980). Benguerel and Pichora-Fuller (1982) discussed the notion of 'visual dominance' of phonemes. In their study, { / p /, / f /, / w /, / 0 /, / u / } were shown to be visually dominant, i.e. they were easily recognized in all phonetic environments of their experiment, while { / t /, / k /, / s /, / J /, / i /, / ae / } were less dominant or prone to variation: "visually dominant phonemes may be representative members of well-defined visemic categories" (ibid., p. 606). Labial, labiodental, and linguadental movements in natural linguistic contexts are the important features for the determination of visemes. 4.2. Visemes for Japanese 4.2.1. Phonological and phonetic characteristics of Japanese There exist great phonological variations between dialects of Japanese. This paper basically focuses on the Tokyo dialect, which is considered to be the basis of Standard Japanese (kydtsugo). Japanese phonemes and their main allophones are shown in Table I, which is based on the work of Hattori (1960, 1979, 1984). Consonants are grouped by manner of articulation. In the table, some modifications to Hattori's system were made by this author for practical purposes, based on the review of other works in the literature (Kawakami, 1977; Jouo, 1980). For phonemes which have been modified by this author and for those which are 23 different from the corresponding English phoneme, further description is given for each case individually. / g A / n /: In the Tokyo dialect, traditionally, / g / and / rj / have been traditionally considered as phonemically distinctive, although it is not the case in some other dialects. In a recent study, however, Inoue (1983) has shown that, even in the Tokyo dialect, / g / and / rj / are becoming phonemically non-distinctive, and that / g / is "taking over" / rj /. This tendency was reported to be particularly noticeable in the younger generation. Inoue has further stated that the older generation is more likely to view the taking over of / rj / by / g / as a deterioration. The younger generation, on the other hand, is unaware of the former difference between the two phonemes, thus it does not express any opinion about this sociolinguistic change. Since the place of articulation of these two phonemes is identical, the nasality of / rj / will not be an issue for speechreading. Therefore, in this study, the phoneme / rj / will no be considered: / g / and / rj / will be considered to constitute a single phoneme denoted as / g /. / s /, / z /: In a strict sense, the phonetic transcription of Japanese / s / and / z / include [s], [ p ], [tc], [z], and [ d? ]. Conventionally, however, the phonetic symbols [ J ]. [ tr ] and [ d3 ] are used, because [ p ], [ tp ] and [ d? ] are respectively very close to each other in terms of place of articulation, and these two groups of consonants are not phonemically distinctive in Japanese. Although the voiced fricatives [ z ] and [ % ] can occur in certain phonetic contexts, they are not phonemically distinctive from [ dz ] and [ d* ], which are the major allophones of the phoneme / z /. / r /: The phoneme / r / is usually realized by the alveolar flap [ r ] in intervocalic position (Sawashima and Kiritani, 1985). Its place of articulation varies, depending upon the following vowel: most forward before [ i ] and gradually moving backward in the context of [e], [ui], [a], [o] (Chiba and Kajiyama, 1958). In word initial position, on the other hand, a weak plosive very similar to [ d ] is observed. This weak plosive is considered to be Table I: Phonemes of Japanese and Their Allophones/Phones PHONEMES MAIN ALLOPHONES PHONETIC REALIZATION CONSONANTS stop i v i [p] [Pi /bl [ b ] [ b ] [ p] Itl [ t ] [ t ] Ul [ d ] [ d ] Ikl [ k ] [ k ] /g/ [g] [ g ] [ v] fricative /s/ [ s ] [ Q] [ s ] [ c] Izl [ dz ] [ d? ] [ d z ] [ d » ] [ z ] [ * ] [3] Ihl [ h ] [ §] [ <p] t h ] [ x ] [ 5 ] [ $ ] [ f ] affricate 1 cl [ ts ] [ tc ] [ ts ] [ tc ] flap Irl [ r] [r] [r] [1] nasal / m / [ m ] [ m ] Inl [ n ] [ ji] [ n ] [ ji] /"/ [rj] [rj] semivowel HI [ j ] [ j ] 1 w / [ w ] [ w ] VOWELS / i / [ i ] [ i ] [ j ] /e/ [e] [e][ e ] [ e ] /a/ [g] [ a ] [ g ] [ a ] Io I [Q] [o][g] /u/ [ ui] [ ui] [ ny] [ tp] SPECIAL PHONEMES /N/ [ m ] [ n ] [ ji] [ rj] [ N ] [ m ] [ n ] [ ji ] [ rj ] [ N ] / Q / [ ? ] [ P ] [ t ] [ k ] [ s ] [ c ] [ ? ] [ P"- ] [ V ] [ k- ] [ s ] [ c ] / ' / [ ? ] [ ? ] 25 a natural preparatory motion for the word initial position of the flap [ r ] (Kawakami, 1977). However, some individuals possess [ 1 ] and/or [ r ] as allophones of Japanese / r /, although these are not the main allophones (Hattori, 1984). Use of [ r ] can be considered to be due to dialectal variation rather than to the influence of a foreign language. Occurrence of [ 1 ], which is not phonemically distinctive from [ r ] and [ r ], can be considered as a variation influenced by certain phonetic environment. / j A / w /: The phonemes / j / and / w / are semivowels (or semiconsonants). Besides occurring in syllable initial position as in / ja /, / ju /, / jo /, the semivowel / j / can follow consonants, as in / pj /, / bj /, / sj /, / zj /, / hj /, / cj /, / mj /, / nj /, / rj /. The other semivowel / w / is always found in syllable initial position, except in the cluster / kw /. Vowels: Japanese has five vowels: / i /, Itl, I a /, / o /, and / u /. The vowel / u / in the Tokyo dialect has no or little lip rounding, although some other dialects do. Its major allophone is [ ui ]. Among the five vowels of Japanese, lip rounding is most obvious in / o /. The tongue position for a high back vowel is generally slightly fronted as compared to [ u ], [ ui ]. Devoicing of the vowels [ i ] and [ ui ] occurs in certain contexts (e.g. between voiceless consonants, or in word final position, when preceded by voiceless consonants, etc.), while devoicing for other vowels seldom occurs. Conventionally, Japanese vowels are transcribed phonetically as [ i ], [ e ], [ a ], [ o ], and [ ui ], thus this study will also follow this notation. / N /: The phoneme / N / has several phonetic realizations: [ m ], [ n ], [ n ], [ rj ], [ N ] depending upon the phonetic context. If / N / is followed by a labial consonant, it will be realized as [ m ], e.g. / siNpu / : [ limply ] (BRIDE); if it is followed by an alveolar consonant, it will be realized as [ n ], e.g. / teNto / : [ tento ] (TENT); if it is followed by a vowel or is used in utterance final position, it will be realized as [ N ], / teN / : [ teN ] (POINT). / Q /: The phoneme / Q / always occurs after a vowel, either in front of one of the voiceless consonants [ p ], [ t ], [ k ] [ ts ], [ t j ], [ s ], [ J ], or in word final position. 26 Exceptions occur in a limited number of loanwords in which the phoneme / Q / precedes the voiced stop [ d ], [ d3 ], [ g ]. The phoneme / Q / is usually considered as an "implosive", in the sense described by Saussure (1959, pp. 51-54). In phonetic transcriptions, / Q / is usually considered as a geminated consonant such as in the word nippon [ nip~>poN ] (JAPAN) . In the case of plosives, the phoneme / Q / corresponds to the first part of the geminated consonant, which is always an implosive, and to part of the closure. As a consequence, acoustically, a silent interval is created. While a geminated consonant is the most common realization of the phoneme / Q /, it can be realized as a combination of two different plosives, subject to certain constraints: [ nitnpON ] or [ n i lopON ] as long as [ t ] or [ k ] is not released; it can be realized as a combination with the glottal stop [ ? ] as in [ ni?poN ]. The glottal stop [ ? ] in Japanese, however, is so weak that it is not clearly observable. When the phoneme / Q / is followed by the phoneme / c /, as in / siQci / (S W A M P ) , it is not realized as [ Jitftri ] but as [ Jintji ]. Again, the first [ t ] is implosive and [ tf ] is explosive. Unlike in the case of plosives, no obvious silent interval is observed for geminated fricatives (e.g. [ issei ] U N A N I M O U S L Y , [ iff T U N ] A N I N S T A N T ) , due to the continuant nature of fricatives. However, in the [ ss ] and [ Jf ], a qualitative difference between the implosive and the explosive part can be perceived (Hattori, 1979). The phoneme / Q / corresponds to the first fricative ([ s ], [ J" ]), which is closely associated to the preceding vowel [ i ]. Hattori states that one of the characteristics of the geminated consonant is the strong linkage, or 'fester Anschluss', of the first consonant to the preceding vowel. The second fricative ([ s ], [ J" ]), on the other hand, is closely associated to the following vowel. Phonetically, [ iff U I N ] is extremely close to [ if: U I N ]; phonemically, however, they are different. Loanwords in Japanese, such as baggu ( B A G ) and beddo ( B E D ) , are realized phonetically as [ bakigui ] and [ bendo ], respectively. / ' /: The phoneme / ' / was first proposed by Hattori (1951, cited in Hattori, 1979). The purpose of proposing this phoneme was to make all the syllables start phonemically with 27 a consonant, e.g. / 'i /, / 'e /, / 'a /, / 'o /, / 'u /, and to make a semivowel always follow a consonant, as in palatalized consonants, e.g. / 'ja /, / 'ju /, /'jo /, / 'wo /. Although Hattori (1979) claims that the phoneme / ' / is the voiced counterpart of the glottal consonant / h /, neither its phonetic realization nor its acoustic characteristics are described clearly. The existence this phoneme, however, is important as far as vowels are concerned. In Japanese, a single vowel (e.g. [ o ], [ i ]), a sustained vowel (e.g. [ o: ], [ i: ]), and a successive occurrence of two identical vowels (e.g. [ oo ], [ ii ]) are phonemically distinctive. For example, [ oja ] (PARENT(S)) contrasts with [ o:ja ] (L A N D L O R D). Similarly, the word [ sato:ja ] ( S U G A R M A K E R ) contrasts with [ satooja ] (FOSTER P A R E N T S ) . Hattori suggests transcribing these latter two words phonemically as / satooja / and / sato'oja / respectively. Kawakami (1977) uses the glottal stop symbol [ ? ] as a marker of separation of two vowels, although he warns that [ 7 ] in Japanese is very weak and that it "does not exist as one consonant of Japanese." This study will employ the phoneme / ' / proposed by Hattori, not as a regular consonant, but as a special phoneme only for the case of the vowels described above. 4.2.2. Significance of haku (mora) in Japanese The mora has been defined as "the smallest unit for measuring the quantity of sound in prosody" (Onishi, 1981, p. 352). In some languages, such as classical Latin, the mora is seen as a syllable, while in some other languages, such as classical Greek, the mora is seen as an accent. In Japanese, the mora is often considered to have a constant length in time. In some languages, the concept of mora and that of syllable are ostensibly identical, but in others, they are not. Japanese belongs to the latter case. For example, the word nippon ( J A P A N ) is considered to consist of two syllables: nip + pon, but of four morae: ni + p + po + n (Hattori, 1960). Hattori (1960) was the first Japanese author to use the term mora in a linguistic sense. Kindaichi (1968), however, proposed to adopt the term haku to refer to the mora, 28 because the definition of the mora by Hattori does not always agree with the term used by other authors (e.g. Trubetzkoy, 1969) when applied to other languages besides Japanese. This study will use the term haku. In Japanese, haku is the basic unit of rhythm, and structurally, it can be either V, C(j)V, or one of the two special consonantal phoneme / N / or / Q /. The concept of haku is significant in Japanese, not only because it is the basic unit of the rhythm, but also because it is related to the writing system called kana (phonemic script). One kana corresponds to each haku except when C is followed by the semivowel /j /, as in / pj /, / bj /, / mj /, / sj /, / zj /, / cj /, / nj /, / kj /, / gj /, / rj /, / hj /. For these cases, two kanas are used: one full-size kana, followed by a small-size kana in subscript position, indicating the presence of the semivowel. Both full-size and small-size kana are transcribed in the phonemic-alphabetic script romaji. There are three styles of romaji: hydjun-shiki, nihon-shiki, and kunrei-shiki. Among the three styles, Hydjun-shiki romaji best reflects the actual pronunciation of kana. Nihon-shiki romaji and kunrei-shiki romaji (modification of nihon-shiki) are less close to the actual pronunciation. For example, the word meaning "foot" in English is pronounced as [aji] in Japanese. It is written ashi in hydjun-shiki romaji, whereas it is written asi in nihon-shiki romaji and in kunrei-shiki romaji. (See Hattori (1979) for further information.) This study will follow the hydjun-shiki romaji, which is most commonly used today. All possible hakus for the Tokyo dialect are listed in Tables II and III, in romaji and in phonetic symbols respectively. Japanese sentences are combinations of these hakus. A elongated vowel (e.g. [ o: ]) is counted as two hakus, and the symbol R - introduced by Kindaichi (1968) and elaborated on by other linguists - is commonly used as the second haku corresponding to the second half of an elongated vowel. Phonetically, the haku R is denoted by the symbol [:], and in romaji, R is denoted by a macron (a bar placed on the top of the preceding vowel). For example, a word keki , which is the romaji transliteration of the English word "cake", is a three-haku word ( ke + R + ki), and it is pronounced as [ ke:ki ]. Table II: List of Hakus Written in Romaji The list is arranged according to the traditional ordering of Japanese, hakus except N, Q, and R, are written in romaji. a i u e o ya yu yo ka ki ku ke ko kya kyu kyo ga gi gu ge go gyu gyo sa shi su se so sha shu sho za ji zu ze zo ja ju jo ta chi tsu te to cha chu cho da de do na ni nu ne no nya nyu nyo ha hi fu he ho hya hyu hyo ba bi bu be bo bya byu byo pa Pi pu pe po pya pyu pyo ma mi mu me mo mya myu myo ra ri ru re ro rya ryu ryo wa N Q R 30 Table III: List of Hakus Written in Phonetic Symbols The list is arranged according to the traditional ordering of Japanese. All hakus except N, Q, and R, are written in romaji. Hakus for loanwords are indicated in parentheses. a 1 UJ e 0 ja J U I J 0 ka ki kui ke ko kja kjui kjo ga gi gui ge go gja gjui gjo sa (si) SUI se so Ja Ji Jut (Je) Jo dza dze dzo d3a d3i d3ui (d3e) d30 ta (ti) •(tin) te to (tjui) tja tji tjui (tje) tjo (tsa) (tsi) tsui (tse) (tso) da (di) (dm) de do (djra) na ne no na J" J1UJ (ne) no ha he ho Sa Si era (Se) So m) (Oi) Oui (<De) ($0) ba bi bui be bo bja bjra bjo pa Pi pui pe po Pja Pjm pjo ma mi mui me mo mja mjui mjo ra ri rui re ro fja rjui fjo wa (wi) (we) (wo) N Q R 31 The symbol N and Q are respectively used for the special phonemes / N / and / Q /. Unlike the other consonant phonemes, which are always tied to a vowel, both / N / and / Q / are independent hakus. Kindaichi (1968) uses 112 hakus for the Tokyo dialect, while Hattori (1960) uses 110. Two hakus - R and wo - were not included in Hattori's list. When sounds occurring only inloanwords are included, the number of hakus is increased to approximately 130 (Kindaichi, 1981). While both Hattori and Kindaichi counted g(j)V and rj(j)V as separate hakus, this author will consider only 103 hakus for the Tokyo dialect, excluding rj(j)V, which is not phonemically distinctive from g(j)V for some people, and wo, which is commonly seen in loanwords and is replaced by o in conventional Japanese. Possible hakus including loanwords are listed in Table III in phonetic forms. Phonetic forms are not given for the hakus N, Q and R, because their phonetic realization varies depending upon the phonetic context. There is no definite number of hakus for loanwords; the number may vary, depending upon an individual's linguistic experience and needs. For example, for people who frequently use musical terminology or specific German nouns, the phoneme strings / tsi /, / tse /, / tsa / and / tso / may have become a part of their total phonemic inventory. The hakus widely used by most Japanese in their daily life are listed in Table IV. Some hakus are so common that people may be unaware that they are not conventional combinations of Japanese phonemes. People who have a fair amount of exposure to foreign languages or to modern technological terms may not have any difficulty in producing the hakus listed in parentheses in Table III, despite the phonological constraint of traditional Japanese. This tendency is presumably more true for the younger generation. People unfamiliar with loanwords, on the other hand, may have a rather reduced inventory of hakus. The labiodental fricative consonants [ f ] and [ v ] are not part of the phonetic inventory of conventional Japanese. The voiceless labiodental fricative [ f ], in the original pronunciation of loanwords, is replaced by the bilabial voiceless fricative [ d) ], and the voiced Table IV: Examples of Loanwords Widely Used in Japanese 32 HAKU ORIGINAL WORDS ARTICULATION [ Oi ] F E V E R [ <pi:ba: ] or [ fi:ba: ] [ <pe ] F E N C E [ OeNSiu ] or [ f e N S i n ] [ Oa ] F A N [ <paN ] or [ faN ] [ Oo ] F O U R [ $o: ] or [ fo: ] [ wi ] W E E K [ wi:kra ] [ we ] W A Y [ wei ] [ wo ] W A T E R [ wo:ta: ] [ si ] CT S C A N [ si:ti:sujkjaN ] [ ti ] T E A [ ti: ] [ di ] D I N N E R [ dina: ] [ tui ] T W O [ tm: ] [ tje ] C H E C K [ tjek^ kra ] [ d3e ] J E T [ d 3 e r t o ] [ Je ] S H E P H E R D [ Jepa:do ] 33 labiodental fricative [ v ] is replaced by the bilabial stop [ b ]. Hiki and Fukuda (1982, p. 192), however, point out that some people who are familiar with the pronunciation of English produce a sound by "pulling the lower lip backward and by touching it to the upper teeth, instead of by protruding and rounding the lips". They further report that this sound is close to English [ f ], but that it has shorter duration and less friction than English [ f ]. Inoue (1983), who investigated the production of [ Oi ] versus [ fi ] in the loanword firumu (F ILM), reports that the production of [ f ] appears to be related to an individual's educational background. One group of subjects, who were students in a private high school, exhibited a higher usage of [ f ] than of [ 0 ], whereas the other groups of subjects exhibited opposite results: a higher usage of [ 0 ] than of [ f ]. Inoue (1983) also shows that the older generation tends to use a combination of two conventional hakus [ <pui ] + [ i ] for [ fi ] in the loanword firumu ( F I L M), while the younger generation uses [ (pi ] or [ fi ] for it. Since the difference in place of articulation between [ cp ] and [ f ] is visually significant, further sociolinguistic research, including [ v ] versus [ b ] distinction will be valuable, particularly for the area of speechreading. When these loanwords are transcribed in kana, usually one full-size kana for CV is followed by a subscript small-size kana for V. For example, {[$],[ f ]} + {[ i ],[ e ], [ a ], [ o ] } is expressed with the kana for [ cpui ] plus a small size kana for { [ i ], [ e ], [ a ], [ o ] }; [ w ] + { [ i ], [ e ], [ o ] } is expressed with the kana for [ wa ] plus a small size kana for { [ i ], [ e ], [ o ] }; [ tfe ] is expressed with the kana for [ tfi ] plus a small size kana for [ e ]. 4.2.3. Determination of visemes of Japanese A few attempts to categorize visemes for Japanese can be found in the recent research literature, but the term viseme, or any translation of the word, is rarely used by Japanese researchers. Viseme is mentioned in the paper by Fukuda (1987) only as a mere introduction 34 to the concept. The present study aims at determining a set of visemes for Japanese which are suitable for the CAST-J lessons. Whether several vowel visemes should be considered or not is still controversial. In the CAST lessons developed in the past (Benguerel and Pichora-Fuller, 1988; Pichora-Fuller and Benguerel, unpublished manuscript), English vowels were categorized as a single viseme V, and French vowels were grouped into two visemes: one for unrounded vowels, the other for rounded vowels. There are two reasons for having only one or two visemic groups for vowels: (1) the definition of the viseme, in particular for vowels, is not clear enough, and (2) auditory perception of vowels is much better than that of consonants. Although it was reported that the visual discrimination of English vowels was poor, due to coarticulatory effects (Montgomery, 1987), the results may not be directly applicable to Japanese, which differs from English on many points. The determination of visemic groupings for vowels appears to depend greatly on the linguistic features of the language under consideration. The significance of vowels in Japanese and the necessity of establishing separate visemes for the vowels will be discussed next. Firstly, there are only five vowels in Japanese: / i /, / e /, / a /, / o /, / u /. Compared to several Indo-European languages (e.g. English, German, French), this number is considerably smaller. These five vowels play an important role in various linguistic aspects. In Japanese, each vowel or combination of vowels can be a meaningful word. Some of the most common and popular examples are given below: HI stomach, well, mind/feeling, difference /e / painting/drawing, branch /a/ painted in white, mutism lol tail /u/ cormorant 35 ii / good ie / house/home iu/ to say/speak ai / love, navy blue ao / blue au / to meet oi/ nephew, aging ou / to chase uo / fish ue / above, top aoi / , blue oie / family/home of aristocrat aoiuo / blue fish Secondly, there is no or little variation across dialects for Japanese vowels. The five vowels are remarkably stable throughout Japan. Thirdly, as discussed in Section 4.2.2, vowels are given more weight in terms of the structure of haku: V, CV, or C. The special phonemes (/ N / and / Q /) are the only consonant (C) which can be realized as a single haku on its own, and the occurrence of these two phonemes has been reported to be relatively low (Bloch, 1950). Bloch analyzed the frequency of occurrence of each phoneme, for all positions, from a text which consisted of two thousand phonemes. A calculation based on his results reveals that the ratio of occurrence was 0.1% for the final / N /, and 1.3% for the final / Q /. Therefore, Japanese can be considered to consist almost exclusively of hakus having a (C)V structure. The frequency of occurrence of vowels was also shown to be higher than for consonants (Bloch, 1950). When phonemes for word boundary and accent are ignored, 53.6% of the text consisted of vowels and 46.4% 36 of consonants. Although the sample used in the Bloch's study is fairly limited in size, the results illustrate one feature of Japanese: the importance of vowels in terms of frequency of occurrence. This is a remarkable difference from English, in which consonants are more frequently used than vowels. A study based on the predicted pronunciation of read text shows that in English, the relative frequency of vowels and diphthongs is 38%, while that of consonants is 62% (Flanagan, 1972). Fourthly, Japanese vowels may sometimes become voiceless: high vowels [ i ] and [ ui ] are usually devoiced between voiceless consonants. Generally, the auditory perception of vowels is better than that of consonants because of their higher intensity. In Japanese, however, vowels are not necessarily audible at all times. Lastly, evidence has been reported that, visually, Japanese vowels are easily distinguished. Fukuda et al. (1976) examined the visual discrimination of vowels in CV monosyllables for thirteen hearing impaired children. Total correct scores for each vowel were relatively high: 91% for / i /, 83% for / e /, 81% for / a /, 88% for / o /, and 80% for / u /. The most common confusions observed in their study were that of / a / with / e / (10%) and that of / u / with / o / (9%). Fukuda and Hiki (1977) also reported high correct discrimination scores of vowels for mono- and dissyllables (nonsense words) and for meaningful words. Nine subjects out of thirteen performed well for all vowels, with correct scores of 85 to 96%, although great individual variability was reported among the other four subjects. High correct scores for vowel discrimination were also found for cochlear implant patients (Fukuda et al., 1988; Funasaka et al., 1987). Stroboscopic observation of articulations in which coarticulatory effects were taken into consideration has shown that the Japanese vowels are relatively distinctive, both in the frontal and the lateral view (Fukuda and Hiki, 1982). Montgomery et al. (1987) examined the visual discrimination of the five English vowels / i A / i A / a /, / u /, / u / in various consonantal contexts. The tense vowels were found to 37 be more easily recognized than the lax vowels. The mean correct score for the tense vowels was 57.9%, whereas for the lax vowels, it was 35.7%. These scores are considerably lower than those obtained by Fukuda and her associates (Fukuda et al. 1976; Fukuda and Hiki, 1977). There are possible explanations for this discrepancy. One of them is that Montgomery et al. (1987) had limited their study to the discrimination between five vowels, four of which were, pairwise, adjacent in the vowel triangle, and thus had similar articulatory features. The five Japanese vowels, on the other hand, are more evenly distributed in the vowel triangle; articulatory features are thus relatively less similar. If the data presented in the study of Montgomery et al. (1987) is recomputed for three groups of vowels {/if, III), {/a/}, { / u / , / u / } instead of the original five, percentage of correct responses increases to approximately 85%. Overall, the discrimination of the five vowels of Japanese can be considered as relatively easy. All the factors discussed above support the consideration of each Japanese vowel as a different viseme. Therefore, the visemes for each vowel, with their main allophenes, are described as follows: \ i \ : < i >; \ e \ : < e >; \ a \ : < a >; \ o \ : < o >; \ u \ : < ui >. There is also good evidence that speechreading training for vowels is imperative. In most cases, hearing impaired people can perceive vowels better than consonants through their auditory channel, because, acoustically, vowels are low in frequency and high in intensity. Particularly with amplification, speechreading of vowels may not be indispensable for them. 38 Under certain circumstances, however, amplification is no longer helpful. For example, a hearing impaired person is most likely to turn off his/her hearing aid in noisy surroundings, e.g. on the train or the subway, a common mode of transportation in Japan. In this case, speechreading becomes a primary means of speech perception rather than a supplemental one. Three studies of the categorization of consonants will now be compared. By means of 'labiograms', Sudo (1973) conducted a quantitative investigation of the labial and facial movements in C/(j)a / monosyllables. Twelve parameters (e.g. width, velocity, duration of certain articulatory movements) were investigated in his study: four parameters for the center of the lips, two for the corner of the lips, and six for the cheeks. Sudo (1973) explains that the term 'labiograms' used in his study refers to an experimental apparatus, which consists of three CdSe cells, three DC amplifiers and one electromagnetic oscillograph. The CdSe cells transform the light waves reflected from the speaker's face into electric signals. Based on his results, Sudo (1979) proposes four groups of phonemes, in terms of the similarity of the labial and the facial movements. Although not all the consonants are included in his study, Sudo's grouping can be applied to the categorization of visemes. GROUP 1 { < p >, < b >, < m >, < pj >, < bj >, < mj > } GROUP 2 { < t >, < n >, < k >, < g >, < s >, < ts > } GROUP 3 { < kj >, < gj >, < rj >, < tj >, < z > ,< dz >, < r > } GROUP 4 { < w >, < $ > } These four groups are further divided into 10 subgroups. Thus, Sudo's study indicates that the visemes of Japanese consonants, at least in CV monosyllables, can range from 4 to 10. Fukuda and Hiki (1982) conducted a more extensive study of the visual characteristics of Japanese consonants in various phonetic environments: 100 single haku words which were either V or CV monosyllables, 15 two-haku words, 12 three-haku words, and 10 four-haku 39 words. Both frontal and lateral views of the movements of the mouth were examined from pictures taken by a special camera in which a long roll of film was driven continuously. The mouth of the speaker was illuminated by a light strobed at 5, 10, or 20 Hz. In addition to the measurement of the change in distance of various parts of the mouth, the authors utilized a type of display called 'labiogram', produced by filming small aluminum pellets placed on the important points on the lips. Using stroboscopic illumination, the traces of these points were photographed and displayed in a pseudo three-dimensional coordinate system (time, hight, and width). The waveforms of the sound was also included on the display. The authors came up with eight mouth shapes. These eight groups are listed below with the 'mouth-shape symbols' assigned by Fukuda and Hiki (1982). The letters Vf stand for 'mouth shape of the following vowel'. p: { < p >, < b >, < m > }; w: { < w >, < cp > }; t: { < t >, < d >, < n > }; r: { < r > }; s: { < ts >, < dz >, < s >, < z > }; sy: {< tj >,< d3 >,< ji >,</>,< 3 >.<? > }; y: { < j > }; Vf: { < k >, < g >, < rj>, < h > }. Fukuda and Hiki comment that these mouth shapes are discernible only in clearly articulated utterances, and can vary as a function of coarticulatory effects. There are some important observations in the study of Fukuda and Hiki (1982) dealing with the determination of visemes. Fukuda and Hiki contend that the difference between \ p \ and \ p\ + \j\is discernable: the speed of the opening movement of the lips is slower 40 in\p\ + \ j \ than in \ p \ . They report that the opening of the lower lip at 0.1 second after plosion for the utterance / pja / is almost half that for / pa /. They also confirm an observation made by Fujimura (1961): the speed of lip opening in bilabial plosives is higher than in bilabial nasals. It is an indication that the bilabial plosives (J p / and / b /) and the bilabial nasa (/ m /) may belong to different visemes. Fukuda and Hiki (1982) further suggest that Yt \ and \ s \ are discemable: \ s \ has longer duration than \ t \ possibly due to the [ +continuant ] characteristic of fricatives. Although these findings have to be taken into consideration in categorizing the visemes, there is no evidence how well a speechreader perceives these features. Further research, including a comparative study of connected speech spoken by different speakers or at different speaking rates is needed. Sato (1961) arrived at four groups for consonants when they are followed by the vowel \a\,\e\,or\o\: {\w\}; {\p\,\b\,\m\}; {\h\,\k\,\g\,\rj\}; { XtX.XdX.XnX.XsX.XzX.XtsX.YdzX.X JX.XtrX.XdsX.XjX.XrX }. When they are followed by X i X or X u X , he categorizes them into three groups, combining the latter two into one. One of the remarkable characteristics of Sato's approach is that he considers the haku as a unit of speechreading. Like other authors, Sato categorized vowels and consonants into several groups respectively. However, this categorization is not yet that of visemes in his sense. He further combined the two (the vowel groups and the consonant groups) to obtain visemes in his sense, i.e. a haku-level categorization. Sato (1961) categorized hakus into 14 groups and assigned a lipreading symbol, which corresponds to the haku-size viseme, to each 41 group. (These 14 symbols have been replaced by numbers in this study.) They are: \l\:<a>or{<h>, <k>, <g>}+<a> \2\:<e>or{<h>, <k>, < g > }' + < e > \3\:<o>or{<h>, <k>, <g>}+<o> \ 4 \ : < ui > or { < h >, < k >, < g >, < t >, < d >, < n >, < s >, < ts >, < dz >, < J >, < tf >, < d3 >, < j >, < r > } + < ui > \5\:<i>or{<h>, <k>, <g>, <t>, <d>,<n>, <s>, < ts >, < dz >, < J >, < tj >, < d3 >, < r > } + < i > \6\: { < t >, < d >, <n >, < s >, < ts >, < dz >, < J >, < tr >, < d3 >, < j >, < r > }+< a > \ 7 \ : { < t >, < d >, <n >, < s >, < ts >, < dz >, < J" >, < tr >, < d3 >, < j >, < r > }+< e > \ 8 \ : { < t >, < d >, <n >,< s >,< ts >,< dz >,<;>,< tj >,< d3 >,< j >, < r > }+< o > \ 9 \ : { < p >, < b >; < m > } + < a > \ 10 \ : { < p >, < b >, < m > } + < e > \ 11 \ : { < p >, < b >, < m > } + < o > \ 12 \ : { < p >, < b >, < m > } + < UI > \ 13 \ : { < p >,•< b >, < m > } + < i > \ 14 \ : < w > + < a > Sato (1961) emphasizes the similarity between lipreading and the decoding of unfamiliar characters. The interpretation of a spoken message without auditory signal requires the skill to identify particular labial and labiodental movements and to match them to one of the many possible hakus, using various contextual cues. Sato's approach first provides a course in which a subject is required solely to identify the 14 symbols without making any connection with their meanings. Following training in matching the movements to non-linguistic symbols in 42 connected speech, the subjects proceed to the next level: the selection of the appropriate lexical item for the symbol identified. Integrating these three studies, it is clear that {<p>, <b>, <m>} and { < w >, < 0 > } can be considered as well-defined visemes. For the less well-defined visemes, this author has decided to combine the grouping of the studies of Fukuda and Hiki (1982) and of Sato (1961), because these two studies came up with similar visemic groups, despite the fact that the authors have used totally different method for arriving at them: the study of Fukuda and Hiki (1982) is based on the quantitative measurements of articulatory movements, with a time precision of 0.1 second, whereas the study of Sato (1961) is based on his clinical research and experience with hearing impaired children learning speechreading. The seven visemes for Japanese consonants arrived at are: \p\: {<p>, <b>, <m>} \ w\ : { < w >, < <p > } \ r\ : { <r> } \ t \ : { < t >, < d >, < n >, < s >, < dz >, < ts > } \k\ : { < k >, < g >, < h > } \ j \ : { < j >, < j i >, < J >, < tr >, < d3>, < c> } \ f \ : { <f> } As discussed in Section 4.2.2., the phonemes / N / and / Q /, and vowel elongation (symbolized by R) are recognized as hakus in Japanese. Fukuda and Hiki (1982) report that, when \ p \ , \ t \ and \ k \ are preceded by N or Q, they are perceived either as consonants with a longer closure, in the case of \ p \ and \ t \ , or as consonants preceded by an elongated vowel, in the case of \ k \ . Considering these observations, this author also proposes to use N, Q, and R as haku visemes. Their realization is exemplified as follows: 43 \N\-->\p\ @_C where C is p, b, or m; \ t \ @ _ C where C is t, d, n, ji, s, dz, J, tj, d3, r, c, j; \ k \ @ _ C where C is k, g, h, i, e, a, o, ui \Q\-->\p\ @_C where C is p; \ t \ @ _ C where C is t, s, J, tf, (d), (d3); \ k \ @ _ C where C is k, (g) \ R \ --> \ i \ @ i _ ; \e\ @ e _ ; \ a \ @ a _ ; \o\ @ o _; \ u \ @ ui _ ; 44 CHAPTER 5 EQUIPMENT AND INPUT/FEEDBACK MODE 5.1. Equipment The hardware components required by CAST are: (1) a microcomputer including a hard disk, a colour graphics card and a keyboard; (2) a high-quality colour video monitor (14"); (3) a video disc player; (4) an amplifier and a speaker; and (5) a printer. (See Figure 1, p. 3.) For the English version of CAST, an IBM PC-AT microcomputer with 512-kb RAM, BCD Videolink RS 232 video controller, Sony PVM 1271Q TV monitor (14"), Sony SLO 325 video cassette recorder, and Epson FX 286 printer were used. For CAST-J, a hardware setup functionally equivalent to that for the English version of CAST will be used. The main difference is that CAST-J replaces the video cassette player by a video disc player, which provides a faster access every time the user wants to jump from one place to another. The software components of CAST-J system are: (1) an operating system; (2) a set of CAST-J programs written in a high-level programming language (C); (3) a set of CAST-J lessons consisting of the spoken materials recorded on video discs, and of the lesson texts stored on the computer with the programs. In CAST-J, the user is provided with easy-to-follow instructions on screen. The format of the screens and key functions are consistent throughout the lessons. The lesson texts are read by the speaker on screen, either with or without audio signal. The trainee inputs his/her response to the computer via the keyboard. The input is echoed on screen, in written form. The feedback provided by the computer, as well, appears on screen in written form. The detail of the lessons is described in the following chapter. For CAST-J, some modifications to the format of the screens have been made to the English version of CAST, on the basis of the 45 study by Aldham, Rochford and Warling (1987), in order to make instructions even more efficient. Further information regarding the operation of the system and the detail of the instructions given on screen are available in the owner's manual of the English version of CAST (Pichora-Fuller and Cicchelli, 1986) and in the study by Aldham et al. (1987). Currently, various highly-sophisticated pieces of equipment are on the market, and their usefulness for further research in speechreading appears promising. Sun Video, for example, allows the user to have the video image on a workstation, or to combine video images with computer-generated graphics. This type of apparatus may not be necessary for the regular training of subjects in the clinic, but it will be invaluable for research purposes. 5.2. Input and feedback modes There are several options regarding the keyboard mode and the feedback mode. The following sections will discuss which mode is most appropriate for input via the keyboard and for feedback on screen. 5.2.1. Kanji, kana, and romaji As mentioned in the previous chapter, Japanese has three different writing systems: kanji (ideograpic script), kana (phonemic-syllabic script), and romaji (phonemic-alphabetic script). Kana is further divided into hiragana and katakana. These different scripts are combined in accordance with the function of the parts of the message. Kanjis (i.e. Chinese characters adapted for use in Japanese) are used to distinguish homophonous words, which can be differentiated neither by kana nor by romaji. In general, hiragana is used to write anything that cannot be expressed with kanjis (e.g. verb inflections, function words), or to write words whose kanji form is difficult to read or write correctly. Katakana is used to write loanwords and proper nouns from foreign languages. Most printed matters issued in Japanese include a combination of these three systems. Romaji, which is an alphabetization of kana, is 46 rarely used by the native speakers of Japanese for a purpose of general communication. This alphabetic script, however, may be used for street signs and/or for textbooks of Japanese specifically written for foreigners. In addition to these conventional writing systems of Japanese, foreign words spelled in their original Western form are widely used in the Japanese society, usually for commercial and/or aesthetic reasons. There exist a number of magazines which are titled in English or in French (e.g. " more ", " croissant "). Although Japanese people are getting more and more familiar with the words written in the alphabet of the original language(s), they are not necessarily getting more familiar with the words written in the alphabet used for romaji. For instance, the English word " STOP " written as in English will be recognized by Japanese people as easily as the word written in kana or in kanji. However, when the word " STOP " is transliterated into romaji, i.e. sutoppu or SUTOPPU, or when the Japanese word meaning " to stop " is written in romaji, i.e. tomare, Japanese people will have difficulty determining what the word means. Generally, in Japan, both the alphabetic mode and the kana mode are provided as input mode to computers. Although kanjis are also available, they require the conversion from either kana or romaji as well as the search of the target character from a number of homophones. Kanjis are not suitable for the CAST-J lessons. In terms of input, it is important to keep to a minimum the time needed by a trainee to input his/her responses. In terms of feedback, one kanji usually corresponds to more than one haku, thus it is difficult to provide feedback to the trainee with viseme-size units. This leaves kana and romaji as the better options for input and feedback. Advantages and disadvantages of each mode will be discussed below. As stated in the 'notation' section at the beginning of this study, italics denote romaji, whereas bold type italics denote kana, with an italic exclamation mark (/) indicating separations between kana characters, (cf. Notation, p. viii). For example, the Japanese word meaning " rain " in English is pronounced [ ame ], thus 47 it will be transliterated ante in romaji. In kana, the same word will be written with two kanas: one kana corresponding to the haku " a " and another kana corresponding to the haku " me "; thus, in this study, it will be written as a.'me. 5.2.2. Input As far as the selection of a keyboard is concerned, neither kana nor romaji can be said to be superior to the other in all situations. Advantages and disadvantages vary, depending on the trainee's familiarity with either type of keyboard. For example, individuals who are used to typing letters in English may prefer romaji input to kana input. On the other hand, individuals who have used the kana mode from the very beginning will prefer kana input to romaji for the reverse reason. In terms of the programming of the computer, the following five aspects have to be considered: (1) There are two particular kana characters, each of which corresponds to two pronunciations ( [ he ] and [ e ]; [ ha ] and [ wa ] ), depending on the syntactic context. In the case of [ he ] and [ e ], the former corresponds to a two-viseme sequence (consonant and vowel), while the latter corresponds to only one viseme. In the case of [ ha ] and [ wa ], the consonant visemes of these two hakus belong to different visemes: < h > belongs to \ k \ , while < w > belongs to \ w \ (2) With kana input, the program would have to be sensitive to the size of each kana (full-size kana versus small-size kana), because the difference in size has phonemic significance in kana. (3) There is no adequate single Roman letter for the haku N. The upper case " N " cannot be used for the haku N, because in romaji, as in English, upper case characters are used in sentence initial position and for proper nouns. Lower case " n " is usually used for the haku N in romaji, but it is not possible to distinguish " na " ( na in kana and phonetically [ na ]) from " na " ( n!a in kana and phonetically [ Na ] ). This ambiguity can be resolved by placing a hyphen between " n " and the vowel " a " for [ Na ], (i.e. "n-a"), and by leaving " na " ( [ na ] ) as it is. The hyphenation, however, may not be 48 a common usage for all trainees. (4) In romaji, the haku R (vowel elongation) has to be indicated by a macron (a horizontal bar on the top of the vowel being elongated) as in to5i (FAR). This particular symbol is not available on normal keyboards. The use of a colon after the vowel (e.g. to.i) is one possible solution to this problem. In kana, the haku R is replaced by an additional vowel, or by a dash, which indicates elongation of the preceding vowel. (5) As for the phonetic realization of vowel elongation, there is not necessarily a one-to-one correspondence between kana and romaji. For example, the Japanese word meaning "senior high school" in English will be pronounced [ ko:ko: ], although in kana, it must be always written kolu.'ko.'u. Spelling rules for romaji, however, are not as strict as those for kana. Some people may write the word as it is written in kana, thus koukou. Others may write the word as it is pronounced, thus ko5ko5 (or ko:ko: in the present study). Furthermore, some phonetically naive people may claim that the word is pronounced [ koukou ] instead of [ ko:ko: ], based on the strong association of their intuition with kana. In such cases, it is not possible to decide whether the trainee perceived the target viseme correctly but simply wrote it out in romaji following the kana convention, or whether the trainee uses o: and ou interchangeably, regardless of the visual perception of the target. Therefore, a new set of orthographic rules of romaji has to be adhered to, in order to establish a consistent input-feedback relationship. Points (1) and (2) above can be looked at as disadvantages of the kana input, or advantages of the romaji input, while points (3) to (5) can be looked at as disadvantages of the romaji input, or advantages of the kana input. 5.2.3 Feedback An advantage of using kana for feedback messages is that it is easy for the trainee to comprehend their meaning. Since a kana can represent any of the strings (j)V, CV, Q or N, 49 the trainee can decode a kana message more quickly than a romaji message. A disadvantage of kana feedback, however, is that it is difficult to provide viseme-level feedback: most of the time, one kana contains information about more than one viseme. For example, when a trainee responds incorrectly to the consonant viseme but correctly to the vowel viseme of a given CV not the vowel but only the consonant. If the feedback provides the answer (target kana) even though the response is not completely right, the feedback is too 'generous'. On the other hand, if the feedback does not provide the answer until the trainee perceives all the visemes of a given target kana correctly, the feedback is too 'demanding'. This problem, however, may be solved by utilizing various techniques of screen display such as different colours and/or different appearance of lines. For example, a red underscore could indicate that the response for a given kana was incorrect for both the consonant and the vowel; a white underscore could indicate that only the consonant was incorrect for the given kana; finally, a yellow underscore could indicate that only the vowel was incorrect for the given kana. Alternatively, a doubled solid underscore could indicate that both the consonant and the vowel were identified incorrectly; a single solid line could indicate that the consonant was identified incorrectly; and a dotted line could indicate that the vowel was identified incorrectly. The main advantage of romaji feedback is that it can provide viseme-level feedback. Its main disadvantage is that people are not used to reading romaji, thus it will take longer for a trainee to decode a feedback message, and possibly interfere in other ways with his/her learning. As far as speechreading is concerned, there is no evidence for deciding which mode is more efficient or preferable for training Japanese subjects. For this first set of CAST-J lessons, romaji input and romaji feedback will be employed for the following reasons: (1) It is probable that a trainee can learn the layout of a keyboard without much difficulty along with the lessons. (2) Even for those who have experience only with the kana keyboard, it is advantageous to 50 become familiar with alphabetic keyboards, which are available for any computer. (3) New rules for romaji writing that have to be learned by a trainee are not complicated and limited in number. Therefore, proper instruction or the provision of a proper "list of rules" will help the trainee to learn and apply the rules correctly, without much difficulty. (4) Romaji feedback can provide viseme-level feedback most conveniently. (5) There is hope that a trainee may feel gradually less uncomfortable reading romaji messages as the training sessions progress. It is thus important for CAST-J, which is still at the pilot study stage, to continue to investigate how successfully trainees perform with different input and feedback modes. Ideally, in future implementations of CAST-J, a trainee should be allowed to choose which input mode (kana versus romaji) he/she wants to use. Likewise, it should be possible to provide the 'demanding' kana feedback as default option, and to have the trainee choose later if he/she wants to see viseme-level feedback in romaji. In the case of romaji feedback, the trainee can see which viseme he/she has misperceived. Some trainees may already know which viseme is most likely to have been misperceived, while others may not have a clue. In the first case, the demanding kana feedback (more than one viseme together) may be sufficient, whereas additional romaji feedback may be necessary in the second case. 5.3. Orthographic rules for kana - romaji conversion In order to have romaji input and feedback, it must be confirmed that several orthographic rules are known not only by the trainee, who inputs his/her responses to the computer in romaji, but also by the designer of the lessons (the author, in this study), who inputs target words/phrases to the computer. The stages where these orthographic rules are necessary are shown in Figure 2, indicated with arrows. Since not many Japanese people are used to writing Japanese in romaji, it is very important to establish orthographic rules for romaji, at least among the users of the CAST-J lessons, to avoid those input-feedback discrepancies which would be due to orthographic-level confusion rather than to the 51 misperception of visemes. These rules are listed in Appendix A. The 'intended texts' of the lessons, which are given to the speaker of the lessons for video-recording, are first written in a combination of kana and kanji, so that the speaker has no uncertainty as to the contents. Although the speaker is expected to read the intended texts as accurately as possible, in the way it is written (e.g. make a pause at phrase boundaries, these being indicated by spaces), it is probable that, after the video-recording, the intended texts will need to be revised in accordance with the performance of the speaker. The revised texts will be transliterated into romaji and will become the actual texts of the lessons, following the orthographic rules for the designer (Appendix A). These romaji-form-texts will, then, be transferred to the computer via the keyboard. Similarly, the trainee will input his/her response to the computer via the keyboard, following a subset of orthographic rules identical to those for the designer (Appendix A). The CAST-J program contains various other rules. These rules include the rules for romaji-to-visemic code conversion, both for target and response, the rules for target-response matching of visemic code, and the rules for providing feedback in romaji. A further description of these rules is provided in Chapter 6. Figure 2. Writing Systems Used at Different Stages and Conversion Rules. 52 DESIGNER'S SIDE INTENDED TEXT (Kanji & kana). t (RECORDING) VIDEO DISC (Video & Audio) (TRANSCRIPTION & REVISION) REVISED TEXT (Kana) ACTUAL TEXT (Romaji) t KEYBOARD (Romaji) Target COMPUTER PROGRAM (CAST-J) TRAINEE'S SIDE r PSYCHOLOGICAL REPRESENTATION OF , RESPONSE STRING RESPONSE STRING REPRESENTATION (Kana or Romaji) RESPONSE STRING (Romaji) KEYBOARD (Romaji) Response Rules for romaji-visemic code conversion Rules for target-response matching Rules for feedback CHAPTER 6 53 LESSONS 6.1. Course structure There are 16 training lessons in CAST-J: 5 lessons for vowel visemes, 8 lessons for consonant visemes, and 3 lessons for haku visemes. The lessons progress from the easiest and least controversial viseme to the more difficult and controversial ones, for both vowels and consonants. Haku visemes, which are considered to be advanced and supplemental, come last. The lessons cover the 15 visemes as follows: 1. \p\ <p>, <b>, <m> 2. \ a\ < a > 3. \ i \ < i > 4. \u\ < i > 5. \ w\ < w > 6. \o\ < 0 > 7. \e\ < e > 8. \ r\ < u > 9. \ t \ < t >, < d >, < n >, < s >, < dz >, < ts > 10. \k\ < k >, < g >, < h >, 11. \ w \ < w >, < F > in loanwords 12. \ f \ < f > in loanwords 13. \ j \ < j >, < J >, < S >, < ts >, < dz >, < C > 14. \N\ haku N 15. \Q\ haku Q 16. \R\ haku R 54 Two other lessons (Lesson 17 and Lesson 18) are available. These 'general lessons' are designed to be used for pre- and post-evaluation of speechreading ability, or for further practice for the trainees who have completed the training lessons. Lesson 1 through 10 constitute the minimum normal sequence: all trainees are expected to go through this set of lessons. These ten lessons include the five vowel visemes and most of the consonant visemes found in conventional Japanese. Lesson 11 through 16 are optional, i.e. the trainee should be allowed to skip some or all of these lessons when the lessons are too difficult for him/her, or when the instructor judges that these lessons are inadequate for the level of skill of the trainee. 6.2. Lesson structure Basically, the four sections of each lesson are designed (1) to review previously learned visemes (old visemes), (2) to practice the recognition of a new viseme, in syllables, (3) to practice old and new visemes by the discourse tracking method, and (4) to recap the lesson materials, with a listing of the trainee's performance statistics. 6.2.1. Review of previously learned visemes The reviewing of old visemes is accomplished by presenting the visemes in visual-only mode, in single-haku format (CV) for the consonant visemes, in two-haku format (VV) for the vowel visemes, and in three-haku format (VCCV) for the haku visemes. For this section, only one allophene is selected for each viseme, e.g. < p > is used as a typical representative for the viseme \ p \ . There is no review section for the first lesson and for the two general lessons. Viseme tokens are demonstrated and the trainee is simply asked to identify them. He/she may re-play the demonstration of a particular viseme any number of times. 6.2.2. Practice of a new viseme 55 Practising the recognition of a new viseme is accomplished by presenting pairs of visemes in visual-only mode, namely Cla - C2a single-haku contexts for consonant visemes, and - V2a two-haku contexts for vowel visemes (Vji - V2i for the viseme \ a \), where Q or Vi always stands for the target viseme, while C2 or V2 stands for the same or another viseme. The order of the pairs is random. For the training of the haku visemes, two- and three-haku sequences are contrasted, e.g. anpa - apa (or anpa) for \ N \ ; appa -apa (or appa) for \ Q \ ; and a:i - ai (or a:i) for \ R \ . The trainee is required to discriminate whether the two members of a particular pair of visemes are the same or different. After all pairs have been judged, pairs responded to incorrectly are repeated until they are judged correctly. The trainee may re-play any pair any number of times until he/she feels confident that he/she can discriminate the two visemes. After the discrimination task, all members of the target viseme are presented audio-visually. The trainee may re-play the demonstration any number of times. 6.2.3. Practice of old and new visemes in discourse Practice of old and new visemes is accomplished by presenting one or two paragraphs. Each paragraph (text) is characterized by a high frequency of occurrence of the newly learned viseme. Words selected for the texts are words which have a statistically high frequency of occurrence in Japanese (Nakano, 1975; Sato, 1982; The National Language Research Institute, 1984). In this section of the lesson, the trainee is expected to integrate visual information of speech with linguistic redundancy and effective sender-receiver feedback in perceiving speech. Each text is centered around a given topic, so that all sentences in the text are semantically related. All topics are considered to be within the everyday experience and linguistic level of the target population. The topic of each lesson is provided at the beginning of the text. Craig (1988) reports that word recognition performance was better for high-predictability sentences than low-predictability sentences and sentences with neutral-predictability 56 carrier phrase. He also reports that older adults can be misled by semantic cues just as well as by acoustic cues. In the study of Sparks, Ardell, Bourgeois, Wiedmer and Kuhl (1979), a significant change in speechreading performance was found between linguistically simple and linguistically complex sentences. The tracking performance of their subjects (the number of words correctly repeated per minute) dropped drastically, when the text was changed from linguistically simple material (e.g. novels written for children and young adults) to linguistically complex material (e.g. stories from magazines written for adults). CAST-J includes both high- and low-predictability sentences, as well as a variety of syntactic structures and phrase lengths. The text for each lesson is divided into phrases of various length (major strings), which are predetermined by the designer. Each major string is numbered, and used as a set of target words/phrases. A major string is further divided into words or phrases (minor strings), which are also predetermined by the designer. In the text, minor strings are separated by spacing. Although an indication of word/phrase boundaries is not mandatory in Japanese writing, the use of 'minor string' is imperative for the computer program to provide better feedback: the computer can compare the target and the response more accurately for short strings of characters than for long ones. The text for each lesson consists of approximately 30 to 50 major strings. In Lesson 1 through 10, each lesson has two paragraphs, the first consisting of syntactically simple sentences and the second consisting of syntactically complex sentences. The syntactically simple sentences contain few embedded sentences, while the syntactically complex sentences contain a number of embedded sentences. In Lesson 11 through 16, each lesson has one single paragraph with sentences of various syntactic structures. Lesson 17 and Lesson 18 are also one-paragraph lessons. The paragraphs used in these two lessons are considered to be visemically well balanced, and to be equivalent in terms of occurrence of each viseme, syntactic complexity, and word familiarity. They are suitable for pre- and post-training comparison. (Further information is provided in Table VI, p. 62, and in Appendix D.) 57 Jackson (1988) suggests that data on coarticulatory effects be used to select contexts: one should proceed from easy-to-see contexts to hard-to-see contexts. In general, consonants are harder to perceive when followed by / u / than when followed by / a / or / { /; vowels are harder to perceive when surrounded by consonants such as / f /, / v /, / S /, / tS /, / Z /, / dz / than in the vicinity of / h / and / g /. Although various phonetic environments are included in CAST-J, more easy-to-see contexts are used for the more controversial visemes than for the less controversial visemes. For example, in Lesson 13 (Viseme \ j \), C + / a / context in sentence initial position and / a/ +C + /a / context are frequently used, while in Lesson 1 (viseme \ p \), different vowel contexts are used. In the case of viseme \ w \ { < w > and < F > }, practice of < w > in the conventional Japanese haku wa, is emphasized in Lesson 2, whereas practice of < w > followed by other vowels and practice of < F > is emphasized in loanwords in Lesson 12. In CAST-J, in the third section of the lesson (practice of old and new visemes in discourse), the texts are presented in paragraph and/or phrase format, as in the English and the French versions of CAST. After practising recognition of a new viseme in the (previous) second section, the paragraph format is used for the introduction and the conclusion of the this section. The phrase format is used for the main part of the practice i.e. interactive training for speech perception. The speaking rate of the texts (both at paragraph level and phrase level) can be selected. The reading of the texts is recorded at two speaking rates: slow normal speech and fast normal speech. The trainee or the instructor selects the appropriate rate for each lesson. The modality of presentation can also be selected. Visual-only and audio-visual presentation are available. The trainee or the instructor can select the adequate modality for each lesson. The presentation of lessons can be made with or without amplification, and/or with or without background noise. Both responses and feedback are always displayed on screen (worksheet), throughout the 58 lessons. Compared with the live face-to-face procedure, this display is helpful in two ways: (1) it eliminates the possible ambiguity of the response; (2) it eliminates the misunderstanding of feedback by the trainee, who, in the face-to-face situation, may have the same difficulty understanding the spoken feedback as understanding the target phrases. For this particular section of the lesson, there is a ceiling to the number of times that the trainee is allowed to repeat the same target when his/her response is not correct. However, the trainee is given the freedom to skip a phrase and to come back to it later. 6.3. Romaji - viseme conversion and target response matching for feecback In the tracking procedure used in CAST-J, feedback is provided to the trainee regarding the accuracy of his/her responses. Target string and response string are converted into visemic code, following the rules of romaji - viseme conversion given in Appendix B. Vowels are coded by alphabetic characters: A, I, U, E, and O; consonants are coded by numbers: 1 through 7; and the haku visemes \ N \ , \ Q \, and \ R \ are coded by 8, 9, and a colon (:), respectively. Feedback is provided on the basis of the visemic match between target and response. The matching algorithm compares response and target, character by character. Visemes which have not yet been presented, in previous lessons or in the current one, are coded as C for consonants, / N / and / Q /, and as V for vowels and / R /. When the trainee fails to identify the existence of a certain viseme, i.e. when he does not input any response for the target viseme, the feedback for this viseme is an underscore, regardless of the level of the viseme. A typical interaction in Lesson 8 may be as follows: Transaction Presentation Mode Item Visemic code (not shown on screen) Target: videoimage yoso:sareru 7 0 4 0 : 4 A 3 E 3 U # 59 Response: Feedback: Response: Feedback: keyboard/screen osareru screen s a r e r u 0 4A3E3U* 0 4A3E3U# keyboard/screen yoso : sareru 7040: 4A3E3U# screen daihyo: tekina 4AI70:4EI4A# (matched!) However, when the trainee identifies the presence of a viseme but misidentifies it, i.e. puts the wrong viseme for the target, the feedback provides the answer for visemes which are not yet learned (C and V), but it only provides an underscore for previously learned and currently learned visemes. The following exchange illustrates this point for a trainee studying Lesson 5: Transaction Presentation Mode Target: Response: Feedback: Response: Feedback: Item video image screen keyboard/screen Visemic code (not shown on screen) shimainitottewa 7I1AI7I4094E2A* (CI1AICICVCCV2A#) keyboard/screen hikaritonde shi a nitotte 5I5A 3I4084E # 51 A 7I4094E # screen shidainitottewa 7I4AI7I4094E2A shi ainitottewa 51 AI7I4094E2A* Thus, feedback is more demanding in later lessons, where the trainee is expected to know a certain number of visemes, than in earlier lessons, where the trainee is expected to identify only a few visemes. 60 In addition to a simple position-by-position match of the visemic code, a rightward search takes place. If the visemic code for the response does not match the visemic code for the target in the corresponding position, the matching program continues to check the next characters, until the end of the minor string (i.e. until the next word/phrase boundary). If a match is found, (a number or C for a consonant, a letter or V for a vowel), the search re-start from the visemic code which is immediately right of the one matched last. The following exchange illustrates this point for a trainee studying Lesson 9: Transaction Item Visemic code (not shown on the screen) Target: do:yara watashio 40: 7A3A#2A4A7IO# Response: yaya toshio 7A7A# 40710* Feedback: ya_a t_shio 7A7A# 40710* Although the matching provided by the system is not always ideal, it can provide feedback which is not too far from that could be given by an actual instructor. 6.4. Texts of the lessons The material of each lesson is listed in Appendix C. The text is listed in the phrase format (major string) and its corresponding visemic transcription. The number of occurrences for the previously learned visemes, the target viseme, and the visemes which have not yet been learned, is listed in Table V, for each lesson individually. The haku visemes \ N \ and \ Q \ are included with the consonants, and the haku viseme \ R \ is included with the vowels. 61 The viseme distribution for Lesson 17 and Lesson 18 is shown in Table VI. An English translation for the entire texts of Lesson 17 and Lesson 18 is given in Appendix D. The texts were translated literally, so as to reflect better the linguistic complexity of the original texts. 62 Table V: Lesson-by-Lesson Inventory of Viseme Tokens Distribution of visemes are given in number of token (columns 3 to 6), and in percent (coelum 8). For each pair of lines, the first one is for consonants, and second one for vowels. Lesson Visemes Visemes Visemes Total Distribution studied studied studied number in percentage in earlier in current in later of visemes lessons lessons lessons in lesson 1 \p\ 93 291 384 0/24/76 — — 370 370 0/ 0/100 2 \a\ 47 326 373 13 / 0/87 — 91 232 323 0/28/72 3 \ i \ 35 316 351 10/ 0/90 75 92 186 353 21 / 26 / 53 4 \u\ 54 342 396 14 / 0/86 153 95 154 402 38 / 24 / 38 5 \ w\ 42 76 239 357 12 / 22 / 66 186 — 109 295 63 / 0/47 6 \o\ 47 248 295 16 / 0/84 160 71 70 301 53 / 24 / 23 7 \e\ 70 252 322 22 / 0/78 181 77 21 279 65 / 28 / 7 8 \ r \ 58 81 214 353 17 / 23 / 60 301 — 21 322 93 / 0 / 7 9 \ t \ 82 80 127 289 28 / 28 / 44 278 — 18 296 94 / 0 / 6 10 \k\ 173 86 84 343 51 / 25 / 24 320 — 19 339 94 / 0 / 6 11 \w\ 89 19 24 132 67 / 15 / 18 107 — 14 121 88 / 0/12 12 \ f \ 84 24 25 133 64 / 18 / 18 104 — 18 122 85 / 0/15 13 \ j \ 98 33 16 147 67 / 22 / 11 124 — 9 133 93 / 0 / 7 14 \N\ 157 40 4 201 78 / 20 / 2 149 — 16 165 90 / 0/10 15 \Q\ 155 37 192 81 / 19 / 0 147 — 9 156 94 / 0 / 6 16 \R\ 185 185 100/ 0/ 0 170 40 — 210 81 / 19 / 0 63 VI: Viseme Inventory for Lesson 17 and Lesson 18 Distribution of each viseme by count and percentage, in each lesson. Percentages are given in parentheses. Viseme Lesson 17 Lesson 18 \p\ % 22 ( 7.7) % 19 ( 6.7) \ w \ 7 ( 2.4) 7 ( 2.4) \ r \ 13 ( 4.5) 12 ( 4.2) \ t \ 33 (11.5) 33 (11.5) \k\ 25 ( 8.8) 29 (10.1) \ f \ 1 ( 0.3) 2 ( 0.7) \ j \ 25 ( 8.8) 22 ( 7.6) \ a\ 27 ( 9.5) 32 (11.1) \ i \ 29 (10.1) 29 (10.0) \u\ 23 ( 8.1) 25 ( 8.7) \o\ 30 (10.5) 29 (10.0) \ e\ 25 ( 8.8) 23 ( 8.0) \ N\ 12 ( 4.2) 11 ( 3.8) \Q\ 7 ( 2.4) 5 ( 1.7) \R\ 7 ( 2.4) 10 ( 3.5) Total 286 (100.0) 288 (100.0) Consonants 145 (50.6) 140 (48.6) Vowels 141 (49.4) 148 (51.4) 64 CHAPTER 7 CONCLUSIONS This study is the first attempt to design a computer-aided speechreading training system for Japanese (CAST-J). In order to implement it and to further refine it, several factors must continue to be investigated. The first factor is the definition of the viseme, in relation to coarticulatory effects. It is known that some visemes are difficult to recognize in certain phonetic contexts, and that the visibility of visemes varies, depending on the speaker (Benguerel and Pichora-Fuller, 1982; Montgomery et al., 1987). Considering the significance of haku, or the strict (C)(j)V structure, in Japanese, it may be desirable to group visemes in accordance with their visemic context. For example, as assumed in this study, one could have a single bilabial viseme \ p \ for all phonetic contexts, i.e. for any combination with the five vowels. Instead of the alveo-dental viseme \ t \ assumed in this study, however, one could consider two separate visemes: one viseme \ tj \ could be used in those contexts where it is followed by / a /, / e /, or / o /, and another viseme \ t^  \ could be used in those contexts where it is followed by / i / or / u /. Although this approach appears similar to that of Sato (1961), the difference between the two is that the approach described above still considers the viseme as a phoneme-size unit, while Sato considers it as a haku-size unit. Unfortunately, for Japanese, research on visemes and/or on coarticulation is extremely limited. In order to apply this approach to the definition of viseme, a close observation of coarticulatory effects are indispensable, and such a study must await the availability of data describing the coarticulatory effects of the preceding sound(s) as well as of the following one(s). At the same time, speaker variability and the effects of speaking rate should also be investigated closely. 65 The second factor is related to sociolinguistic changes in Japanese. Some sociolinguistic variations can be crucial for the categorization of visemes. For instance, the fricative consonant [ f ], which is not included in the phonetic inventory of conventional Japanese, has been shown to occur in certain sociolinguistic groups. Since [ f ] is considered to be one of the well-defined visemes in English (Benguerel and Pichora-Fuller, 1982), occurrence of this articulation in Japanese has a great influence on speechreading in Japanese. Rapid expansion of the use of loanwords in Japanese society and increasing opportunities to expose oneself to foreign languages may have a long-term impact on the phonetic inventory of Japanese. The third factor is the adequacy of input and feedback mode, i.e. kana versus romaji. Although it is often suspected that Japanese subjects are more familiar with kana than with romaji, there is no evidence on how much the kana mode is superior to the romaji mode, or how the romaji mode can interfere with one's expression and comprehension of the message. The fourth factor is the relation between syntactic complexity of the text and speechreading performance. Past research has shown that linguistic complexity affects the speechreading performance (De Filippo and Scott, 1978; Sparks et al., 1979). However, a study of the detail of syntactic structure (such as the number of embedded sentences) or of the structure of relative clauses is not found in the literature. Further research in this area will be helpful for the development of adequate lesson materials and the selection of an appropriate group of subjects. The fifth factor is the application of CAST-J to different populations. In the near future, serious consideration should be given to developing a similar system for children and young adults who are post-lingually deaf. Application of CAST-J to this group may simply require the adequate selection of vocabulary (for the text) that be suitable for the target population. Other potential target population for other versions of CAST-J include adults with a severe-to-profound hearing loss who cannot benefit from amplification, and adults and children with a congenital hearing loss. With a better understanding of speech perception and language development in the above populations, CAST-J will make a contribution to the enhancement of their communicative ability. 67 BIBLIOGRAPHY NOTE: Japanese references for which no English title has been provided by the original author(s) are listed with their Japanese title first, followed by an English translation by this author in parentheses. Aldham, L.J., Rochford, S.L., and Warling, A.R. (1987). Evaluation of CAST: a computerized method for lipreading training and assessment. Unpublished Undergraduate Thesis, Department of Industrial Engineering, University of Toronto, Toronto. Balkany, T., Gantz, B., and Nadol, J.B. (1988). Multichannel cochlear implants in partially ossified cochleas. Annals of Otology, Rhinology and Laryngology, 97 (Supplement 135), 3-7. Bement, L., Wallber, J., De Filippo, C, Bochner, J., and Garrison, W. (1988). A new protocol for assessing viseme perception in sentence context: the lipreading discrimination test. Ear and Hearing, 9, 33-40. Benguerel, A-P., and Pichora-Fuller, M.K. (1982). Coarticulation effects in lipreading. Journal of Speech and Hearing Research, 25, 600-607. Benguerel, A-P., and Pichora-Fuller, M.K. (1988). Design of a computer-aided speechreading training system. Ann. Bull. RILP, 22, 103-116. Berger, K.W. (1972). Speechreading Principles and Methods. Baltimore: National Educational Press. Binnie, C. (1976). Relevant aural rehabilitation. In J. Northern (ed.), Hearing Disorders. Boston: Little Brown. Binnie, C, Jackson, P., and Montgomery, A. (1976). Visual intelligibility of consonants. A lipreading screening test with inplications for aural rehabilitation. Journal of Speech and Hearing Disorders, 44, 530-539. Bloch, B. (1950). Studies in colloquial Japanese IV. Language, 26, 86-125. Boothroyd, A. (1988). Management of profound sensorineural hearing loss in children: possibilities and pitfalls of cochlear implants. Annals of Otology, Rhinology and Laryngology, 96 (supplement 128), 8. Boothroyd, A., Hnath-Chisolm, T., Hanin, L., and Kishon-Rabin, L. (1988). Voice fundamental frequency as an auditory supplement to the speechreading of sentences. Ear and Hearing, 9, 306-312. B0rrild, K. (1970). Cued speech and the mouth-hand system. A contribution to the discussion. In International Symposium on Speech Communication Ability and Profound Deafness. Ed. G. Fant. Washington, D.C: Alexander Graham Bell Association for the Deaf. 68 Breeuwer, M., and Plomp, R. (1984). Speechreading supplemented with frequency-selective sound-pressure information. Journal of the Acoustical Society of America, 76, 686-691. Breeuwer, M., and Plomp, R. (1985). Speechreading supplemented with formant frequency information from voiced speech. Journal of the Acoustical Society of America, 77, 314-317. Breeuwer, M., and Plomp, R. (1986). Speechreading supplemented with auditorily-presented speech parameters. Journal of the Acoustical Society of America, 79, 481-499. Brown, G., and Yule, G. (1983). Discourse Analysis. Cambridge: Cambridge University Press. Campbell, R. (1987). The cerebral lateralization of lip-reading. In Hearing by Eye: The  Psychology of Lip-reading. Eds. B. Dodd and R. Campbell. Hillsdale: Lawrence Erlbaum. Chiba, T., and Kajiyama, M. (1958). The Vowel: Its Nature and Structure. Tokyo: Phonetic Society of Japan. Cohen, N.L., Hoffman, R.A., and Stroschein, M. (1988). Medical or surgical complications related to the Nucleus multichannel cochlear implant. Annals of Otology, Rhinology and Laryngology, 97 (supplement 135), 8-13. Conrad, R.A. (1979). The Deaf Schoolchild. London: Harper and Row. Cornett, R.O. (1967). Cued speech. American Annals of the Deaf, 112, 3-13. Craig, C. H. (1988). Effect of three conditions of predictability on word-recognition performance. Journal of Speech and Hearing Research, 31, 588-592. Craig, W. (1964). Effects of preschool training on the development of reading and lipreading skills of deaf children. American Annals of the Deaf, 109, 280-296. De Filippo, C.L., and Scott, B.L. (1978). A method for training and evaluating the reception of ongoing speech. Journal of the Acoustical Society of America, 63, 1186-1192. De Filippo, CL. (1984). Laboratory projects in tactile aids to lipreading. Ear and Hearing, 5, 211-227. Eddington, D.K. (1980). Speech discrimination in deaf subjects with cochlear implants. Journal of the Acoustical Society of America, 68, 885-891. Eisenberg, L.S., Berliner, K.L, Thielemeir, M.A., Kirk, K.I., and Tiber, N. (1983). Cochlear implants in children. Ear and Hearing, 4, 41-50. Engelmann, L.R., Waterfall, M.K., and Hough, J.V.D. (1981). Results following cochlear implantation and rehabilitation. Laryngoscope, 91, 1821-1832. Erber, N.P. (1972). Auditory, visual, and auditory-visual recognition of consonants by children with normal and impaired hearing. Journal of Speech and Hearing Research, 15, 413-422. 69 Fant, G. (1970). Q-Code. In International Symposium on Speech Communication Ability and  Profound Deafness. Ed. G. Fant. Washington, D.C: Alexander Graham Bell Association for the Deaf. Fenn, G., and Smith, B.Z.D. (1987). The assessment of lipreading ability: some practical considerations in the use of the tracking procedure. British Journal of Audiology, 21, 253- 258. Fisher, CG. (1968). Confusions among visually perceived consonants. Journal of Speech and Hearing Research, 11, 796-804. Flanagan, J.L. (1972). Speech Analysis Synthesis and Perception, 2nd ed. New York: Springer-Verlag. Fujimura, O. (1961). Bilabial stop and nasal consonants: a motion picture study and its acoustical implications. Journal of Speech and Hearing Research, 3, 233-247. Fukuda, Y. (1987a). Transmission of speech information through the mouth shape: various methods for supplementing speechreading. Journal of the Acoustical Society of Japan, 43, 349-355. Fukuda, Y. (1987b). Use of video tapes of speech sound plus mouth shape for evaluating the speech perception ability through artificial inner ear. The Acoustical Society of Japan, Transactions of the Committee on Speech Research SP87-73, 15-20. Fukuda, Y., and Hiki, S. (1977). Interrelationship between speechreading and use of residual hearing - relationship between reception of syllable and word and its individuality. Journal of the Acoustical Society of Japan, 33, 198-207. Fukuda, Y., and Hiki, S. (1982). Characteristics of the mouth shape in the production of Japanese - stroboscopic observation. Journal of the Acoustical Society of Japan (E), 3, 75-91. Fukuda, Y., Sakamoto, Y., and Kuroki, S. (1976). Promotive effect of residual hearing upon speechreading. Journal of the Acoustical Society of Japan, 32, 271-276. Fukuda, Y., Shiroma, M., and Funasaka, S. (1988). Speech perception ability of the patients with artificial inner ear: combined use of auditory and visual perception. The Acoustical Society of Japan, Transactions of the Committe on Speech Research SP88-91, 17-23. Funasaka, S., and Shiroma, M. (1987). 22-channel cochlear implant no hyoka (Evaluation of a 22-channel cochlear implant). Japan Journal of Logopedics and Phoniatrics, 28, 265-269. Funasaka, S., Hayashibara, S., Takahashi, O., Yukawa, K., and Hatsushika, S. (1987). Speech comprehension in a deaf patient with 22 channel cochlear implant. Journal of Otolaryngology of Japan, 90, 995-1003. Funasaka, S., Hosoya, M., Hayashibara, S., Takahashi, O., Yukawa, K., and Hatsushika, S. (1986). The 22-channel cochlear implant: An introduction of the system and restoration of speech comprehension. Journal of Otolaryngology of Japan, 89, 1070-1076. 70 Gagne, J-R, Dinnon, D.A., and Parsons, J.M. (1989). Assessment of an interactive video system in training speechreading. Journal of Speech-Language Pathology and Audiology, 13, 58. Precedings of the 14th Annual Conference of Canadian Association of Speech-Language Pathologists and Audiologists. Gagne\ J-P., and Wyllie, K.A. (1989). The effectiveness of three repair strategies on the visual-recognition of misperceived words. Journal of Speech-Language Pathology and Audiology, 13, 58. Precedings of the 14th Annual Conference of Canadian Association of Speech-Language Pathologists and Audiologists. Gantz, B.J., McCabe, B.F., Tyler, R.S., and Preece, J.P. (1987). Evaluation of four cochlear implant designs. Annals of Otology, Rhinology and Laryngology, 96 (supplement 128), 145-147. Grant, K.W. (1987). Encoding voice pitch for profoundly hearing-impaired listeners. Journal of the Acoustical Society of America, 82, 423-432. Hanin, L, Boothroyd, and Hnath-Chisolm, T. (1988). Tactile presentation of voice fundamental frequency as an aid to the speechreading of sentences. Ear and Hearing, 9, 335-341. Hattori, S. (1960). Gengogaku no hoho (Methods in Linguistics). Tokyo: Iwanami Shoten. Hattori, S. (1979). Oninron to seishoho (Phonology and Orthography). Tokyo: Taishukan Shoten. Hattori, S. (1984). Onsegaku (Phonetics). Tokyo: Iwanami Shoten. Henderson, L. (1985). On the use of the term 'grapheme'. Language and Cognitive Processes, 1, 135-148. Hiki, S., and Fukuda, Y. (1982). Mouth shape of the [$] and [w] sounds pronounced in loanwords in Japanese. Journal of Acoustical Society of Japan (E), 3, 191-194. Hnath-Chisolm T. and Kishon-Rabin, L. (1988). Tactile presentation of voice fundamental frequency as an aid to the perception of speech pattern contrasts. Ear and Hearing, 9, 329-334. Hnath-Chisolm, T. and Medwetsky, L. (1988). Perception of frequency contours via temporal and spatial tactile transforms. Ear and Hearing, 9, 322-328. Hochberg, I., Rosen, S., and Ball, V. (1987). Effect of text complexity upon connected discourse tracking. Annals of Otology, Rhinology and Laryngology, 96 (Supplement 128), 82-83. Inoue, F., ed. (1983). A Sociolinguistic Study of "New Dialect" and "Language Deterioration"  - Tokyo, Metropolitan Area, Yamagata, Hokkaido. Tokyo: Tokyo University Press. Jackson, P.L. (1988). The theoretical minimal unit for visual speech perception: visemes and coarticulation. Volta Review, 90, 99-115. Jeffers, J., and Barley, M. (1971). Speechreading. Springfield: Charles C Thomas. 71 Johnson, S.C. (1967). Hierarchical clustering schemes. Psychometrica, 32, 241-244. Jouo, H. (1980). Phonetics. In Chokaku gengo shogai (Speech, Language, and Hearing Disorders). Ed. S. Horiguchi. Tokyo: Ishiyaku Shuppan. Kawakami, S. (1977). Nihongo onse gaisetsu (Summary of Phonetics in Japanese). Tokyo: Ohusha. Kindaichi, H. (1968). Nihongo onin no kenkvu (Study on Phonology in Japanese). Tokyo: Tokyodo Shuppan. Kindaichi, H. (1981). Nihongo no tokushitsu (Characteristics of Japanese). Tokyo: Nippon Hoso Shuppan Kyokai. Lesner, S.A., and Kricos, P.B. (1987). Tracking as a communication enhancement strategy with nursing home residents. Journal of the Academy of Rehabilitative Audiology, XX, 39-48. Lesner, S.A., Sandridge, S.A., and Kricos, P.B. (1987). Training influences on visual consonant and sentence recognition. Ear and Hearing, 8, 283-287. Luxford, W., House, W.F., Hough, J.V.D., Tonokawa, L.L., Berliner, K.I., and Martin, E. (1988). Experiences with the Nucleus multichannel cochlear implant in three young children. Annals of Otology, Rhinology and Laryngology, 97 (Supplement 135), 14-16. Lyxell, B., and Ronnberg, J. (1987). Guessing and speechreading. British Journal of Audiology, 21, 13-20. Massaro, D.W. (1987). Speech Perception by Ear and Eye: A Paradigm for Psychological  Inquiry. Hillsdale: Lawrence Erlbaum. McGurk, H., and MacDonald, J. (1976). Hearing lips and seeing voices. Nature, 264. Middelweerd, M.J., and Plomp, R. (1987). The effect of speechreading on the speech-reception threshold of sentences in noise. Journal of the Acoustical Society of America, 82, 2145-2147. Montgomery, A.A., and Jackson, PL. (1983). Physical characteristics of the lips underlying vowel lipreading performance. Journal of the Acoustical Society of America, 73, 2134-2144. Montgomery, A.A., Walden, B.E., and Prosek, R.A. (1987). Effects of consonantal context on vowel lipreading. Journal of Speech and Hearing Research, 30, 50-59. Myklebust H., and Neyhus, A. (1971). Diagnostic Test of Speechreading. New York: Grune and Stratton. Nakano, H. (1975). Tango no kazu wa donokurai aru ka (How many words are there?). In E. Iwabuchi, and T. Nishio (eds.), Shin nihongo koza 1: gendai nihongo no tango to  moji. Tokyo: Sekibunsha. 72 National Institute of Health (1988). Consensus Development Conference Statement: Cochlear Implant. Bethesda, MD. Nitchie, E.B. (1979). How to Read Lips for Fun and Profit. New York: Hawthorn. Onishi, M. (1981). A Grand Dictionary of Phonetics. Supervisor of the editorial board: M. Onishi. Tokyo: The Phonetic Society of Japan. Owens, E., and Raggio, M. (1987). The UCSF tracking procedure for evaluation and training of speech reception by hearing-impaired adults. Journal of Speech and Hearing Disorders, 52, 120-128. Parkin, J. L., and Stewart, B. E. (1988). Multichannel cochlear implantation: Utah-design. Laryngoscope, 98, 262-265. Pichora-Fuller, M.K. (1980). Coarticulation and lipreading. (Master's thesis, University ot British Columbia, 1980.) In Canadian Theses, 1976/77-1979/80, 2, 50023, 1983. Pichora-Fuller, M.K., and Benguerel, A-P. (1989) Computer-Aided Speechreading Training (CAST): I. Development of a rehabilitative tool. Submitted for publication. Pichora-Fuller, M.K. and Cicchelli, M. (1986). Computer-Aided Speechreading Training  (CAST): Owner's Manual. English Version 2.1. (Available from M.K. Pichora-Fuller, Department of Psychology, University of Toronto, Toronto, M5S 1A5.) Sato, K. (1982). Koza:nihongo no goi (Seminar: Vocabulary of Japanese). Tokyo: Meigi Shoin. Sato, S. (1961). Ro nancho ji kvoikushinri enshu tekisuto I.A.2: Dokushin kenkyu no bu IL (Lipreading: Educational Psychology Seminar Text for the Hearing Impaired.) Unpublished manuscript. Department of Education, Tohoku University. Saussure, F. de (1959). Courses in General Linguistics. Trans. W. Baskin. Eds. C. Bally and A. Sechehaye, in collaboration with A. Riedlinger. New York: McGraw-Hill. Sawashima, M., and Kiritani, S. (1985). Electro-palatographic patterns of Japanese /d/ and /r/ in intervocalic positions. Ann. Bull. RILP, 19, 1-6. Sheeley, E.C. (1985). Glosssary. In Handbook of Clinical Audiology, 3rd ed. Ed. J. Katz. Baltimore: Williams and Wilkins. Shiroma, M., and Funasaka, S. (1988). Jinko naiji soyosha no rihabiriteshon to hyoka. Japan Journal of Logopedics and Phoniatrics, 29, 59. Proceedings of the XXXII Congress of the Japan Society of Logopedics and Phoniatrics. Simmons, F.B. (1985). Cochlear implants in young children: some dilemmas. Ear and Hearing, 6, 61-63. Sims, D.G., and Jacobs, M. (1976). Speechreading evaluation and the National Technical Institute for the Deaf. Paper presented at the annual convention of the Alexander Graham Bell Association for the Deaf, Boston. 73 Sims, D.G., Scott, L., and Myers, T. (1982). Past, present and future computer assisted communication training at NTID. Journal of the Academy of Rehabilitative Audiology, 15, 103-115. Sims, D.G., Von Feldt, J., Dowaliby, F., Hutchinson, K., and Myers, T. (1979). A pilot experiment in computer assisted speechreading instruction utilizing the data analysis video interactive device (DAVID). Chapter 16 in "American Annals of the Deaf Symposium on Research and Utilization of Educational Media for Teaching the Deaf: Educational Technology for the '80's". American Annals of the Deaf, 124, 616-623. Sparks, D.W., Ardell, L.A., Bourgeois, M., Wiedmer, B., and Kuhl, P.K. (1979). Investigating the MESA (Multipoint Electrotactile Speech Aid): the transmission of connected discourse. Journal of the Acoustical Society of America, 65, 810-815. Spitzer, J.B., Leder, S.B., Milner, P., Flevaris-Phillips, C, and Giolas, T.G. (1987). Standardization of four videotaped tests of speechreading ranging in task difficulty. Ear and Hearing, 8, 227-231. Stokoe, W.C, Casterline, D.C., and Croneberg, C.G. (1965). Dictionary American Sign  Language. Washington: Gallaudet College Press. Sudo, T. (1973). Distinctive features of lipreading by labiogram. Bulletin of Tokyo Gakugeri University Ser.I (24), 29-37. Sudo, T. (1979). Transmitted informations on the lipreading of the syllable. Bulletin of Tokyo Gakugei University Ser.I (30), 11-20. Sumby, W.H., and Pollack, I. (1954). Visual contribution to speech intelligibility in noise. Journal of the Acoustical Society of America, 26, 212-215. Tanaka, Y. (1980). Seijin no kotense nancho (Acquired hearing loss in adults). In Chokaku gengo shogai (Speech, Language, and Hearing Disorders). Ed. S. Horiguchi. Tokyo: Ishiyaku Shuppan. Tanaka, Y., and Chiba, F. (1975). Speechreading test using a video tape recorder and its application to hearing aid evaluation. Audiology Japan, 18, 195-202. Tanaka Y., Shindo, M., and Motomiya, T. (1973). A speechreading test using a television, and communication ability in the hearing-impaired. Audiology Japan, 16, 109-119. The National Language Research Institute (1984). Studies on the Vocabulary of Senior High  School Textbooks, vol. II. Tokyo: Shuei Shuppan. Trubetzkoy, N. (1969). Principles of Phonology. Trans. C.A.M. Baltaxe. Barkley: University of California Press. Tyler, R.S. (1988). Open-set word recognition with the Duren/Cologne extracochlear implant. Laryngoscope, 98, 999-1002. 74 Tyler, R.S., Davis, J.M., and Lansing, CR. (1987). Cochlear implants in young children. Asha, XXIX. 41-49. Tyler, R.S., Tye-Murray, N., Preece, J.P., Gantz, B.J., and McCabe, B.F. (1987). Vowel and consonant confusion among cochlear implant patients: do different implants make a difference? Annals of Otology, Rhinology and Laryngology, 96 (Supplement 128), 141-143. Upton, H.W. (1968). Wearable eyeglass speechreading aid. American Annals of the Deaf, 113. 222-229. Utley, J.A. (1946). A test of lipreading ability. Journal of Speech and Hearing Disorders, 14, 109-116. Van Uden, A.M. (1983). Does seeing oneself speak benefit lipreading? Help especially for dyspraxic children. In Diagnostic Testing of Deaf Children: The Syndrome of  Dyspraxia. Ed. A. Van Uden. Silver Spring: Fellendorf Associates. Walden, B.E., Erdman, S.A., Montgomery, A.A., Schwartz, D.M., and Prosek, R.A. (1981). Some effects of training on speech recognition by hearing-impaired adults. Journal of Speech and Hearing Research, 24, 207-216. Walden, B.E., Prosek, R.A., Montgomery, A.A., Scherr, C. K., and Jones, C.J. (1977). Effects of training on the visual recognition of consonants. Journal of Speech and Hearing Research, 20, 130-145. Woodward, M.F., and Barber, CG. (1960). Phoneme perception in lipreading. Journal of Speech and Hearing Research, 3, 212-222. Youngblood, J., and Robinson, S. (1988). Ineraid (Utah) multichannel cochlear implants. Laryngoscope, 98, 5-10. 75 APPENDIX A Orthographic Rules for Kana - romaji Conversion NOTE: These rules are applied by the designer of the lessons when the actual texts are written in romaji for computer input, after the speaker has read the intended texts written in kanji and kana. They are also applied by the trainee when he/she inputs his/her responses to the computer in romaji. (1) Use hydjun-shiki romaji. (2) Place a word/phrase boundary (space), where the speaker made a pause. (3) Always use a colon (:) for vowel elongation in loanwords, e.g. sho: (SHOW) (4) When transliterating a vowel sequence such as a'.a, Hi, u!u, e.'e, olo, (a) re-write it as a vowel plus a colon (:), e.g. o.'HsHi — > oishi: (DELICIOUS) unless separation of the vowels is semantically important; in such a case, (b) re-write it as two vowels, e.g. Hi — > ii (GOOD) (5) When transliterating the kana string o!u, (a) re-write it as the first vowel plus a colon (o:) when in the context Cy_, e.g. kyo.'u — > kyo: (TODAY) or when the phonetic realization of o.'u is customarily [o:]; e.g. ko'.u'.kolu — > ko:ko: (SENIOR HIGH SCHOOL) (b) re-write it as ou when the separation of these two vowels is semantically important. e.g. olmolu —> omou (TO THINK) (6) When transliterating the kana string e!i, 76 (a) re-write it as the first vowel plus a colon (e:) only in Lesson 16; e.g. se'.Udo —> se.do (SYSTEM) (b) re-write with ei in all other cases, e.g. e.'Uga — > eiga (MOVIE) (7) When transliterating the kana string i!u, (a) re-write it as yu: in the context of to! (!ko!to); e.g. to!i!u!ko!to!ni — > toyuikotoni (IT IS THAT...) (b) re-write with iu in other contexts. e.g. mi.'u.'shi.'na.'u — > miushinau (LOSE SIGHT OF) (8) When transliterating the kana denoting the haku N, (a) re-write it as n followed by a hyphen when the haku N is followed by a (semi)vowel; e.g. an-i [ arvi ] (CARELESS) instead of am [ ani ] (BROTHER) kon-ya [ koNj'a ] (TONIGHT) instead of konya [ kojia ] (non-sense word) (b) re-write it as n when followed by a consonant, or when in utterance final position. e.g. sonna (SUCH) ichiban (BEST) (9) When transliterating the kana for the haku Q, duplicate the following consonant. e.g. nippon (JAPAN) ikkai (FIRST FLOOR) The rules (8) and (9) are illustrated without giving the kana string, due to problems with their transliteration. 77 APPENDIX B Romaji - Visemic Code Conversion Rules *** General viseme rules (Japanese) *** viseme_rule visja[] = { /* replace code, matched string, no. chars to skip */ /* a */ 'A', ti II /* i */ T, it'll * > /* u */ •u\ "u", /* e */ •E\ "e", /* o */ •0', "o", /* b */ • •1\ "b", 1* c */ '9*, "cch", '7', "ch", /* d */ '9\ "dd", •4', "d", /* f */ •6', "f, /* g */ '9', "gg". '5', "g", /* h */ '7', "hy", '5', "h", /* j */ '7', "j", /* k */ '9', "kk", '5', "k", /* m */ '1\ "m", /* n */ '4*, "ny", •8', "nb", '8', "nc", '8', "nd", '8', "nf, '8', "ng", '8', "nh", '8', 1, *8\ "nk", 1, *8\ "nm", 1, '8', _ tt nn , 1, '8', np , 1, '8', "nr", 1, '8', "ns", 1, '8', "nt", 1, '8\ "nw", 1, '8', "nz", 1, '8', l i _ l i n- , 'T, "ni", 1, '4', "na", 1, '4', "nu", 1, '4', "ne", 1, *4\ "no", 1, '8', "n", !-/* p */ '9', n _ _ tt PP . 1, '1', "P", 1 « /* r */ '3', "r", 1. /* s */ '9', "ssh", 1, '9', "ss", 1, '7', "sh", 2, '4', "s", 1, /* t */ '4', "ts", 2, '9\ "tt", 1, '4', i i | . i t 1, /* w */ •2', "wa", 1, '6\ "wi", 1, '6\ "we", 1, '6', "wo", 1. /* y */ '7', "y", 1, /* Z */ '4', It M 1 1, /* : */ • > II .11 • » 1, APPENDIX C Lesson Script and Corresponding Visemic Code LESSON 1 Review There is no review for Lesson 1. Training Visual Only Presentation: pa pa pa na pa ya pa sha pa ra pa wa pa pya pa ta pa kya pa ma pa nya pa ha pa kya pa ga pa ja pa bya pa fa pa mya pa ka pa pa za pa ba pa cha pa tsa pa rya pa da pa sa Audio-visual Presentation: pa ba ma Practice A SUM ERA MIYAKO (PLACE TO LIVE) 1. Mishiranutochide 1I7I3A4U407I4E* 2. seikatsuo hajimerubaai, 4EI5A4U0#5A7I1E3U1AAI# 3. mazu shinpaininarunoga 1A4U#7I81AI7I4A3U405A# 4. tabemonoto 4A1E104040* 5. kotobadato omoimasu. 50401A4A40#010I1A4U# 6. Donnabashodemo 4084A1A704E10* 7. swnebamiyakoto 4U1E1A1I7A5040* 8. iwarerumonono, I2A3E3U104040* 9. bimyo.na tokorode 11170:4A#4050304E# 10. atarashi: mizuni A4A3A7I:#1I4U7I# 11. najimezu, 4A7I1E4U* 12. komarukotoga arimasu. 501A3U50405A#A3I1A4U# 13. Mottomo nipponwa 1094010#7I91082A# 14. motomoto shimagunide 10401040#7I1A5U7I4E# 15. tan-itsuminzokukokka 4A8#I4U1I8405U5095A# 16. demoarunode, 4E10A3U404E# 17. ippanshiminno I91A87I1I840* 18. kurashiburiniwa 5U3A7I1U3I7I2A* 19. amari sccwa A1A3I#4A2A# 20. mitomeraremasen. 1I401E3A3E1A4E8* 21. Mononomikata, 1040401I5A4A* 22. uketomekatamo U5E401E5A4A10* 23. konpontekiniwa 5081084E5I7I2A* 24. kyo:tsu:no bubunga 570:4U:40#1U1U85A# 25. o.ku miraremasu. 0:5U#1I3A3E1A4U# 26. Mushiro mondainanowa 1U7I30#1084AI4A402A# 27. uminomukomi U1I401U50:7I# 28. utsurisumu baaidesho:. U4U3I4U1U#1AAI4E70:# 29. Mochiron ima 107I308#I1A# 30. sekaino musubitsukiwa 4E5AI40#1U4U1I4U5I2A# 31. koremadeninaku 503E1A4E7I4A5U* 32. missetsuni narimashita. 1I94E4U7I#4A3I1A7I4A# 33. Shinbun, terebinado 7I81U8#4E3E1I4A40# 34. masukomino okagede 1A4U501I40#05A5E4E# 35. samazamana jo:ho:ga 4A1A4A1A4A#70:50:5A# 36. mainichi 1AI7I7I* 37. meto mimikara 1E40#1I1I5A3A# 38. tobikondekimasu. 401I5084E5I1A4U* 39. Demo, 4E10# 40. mattakubetsu.no bunkani 1A94A5U1E4U40#1U85A7I# 41. tokekomunowa 405E501U402A* 42. totemo muzukashvmonodesu. 404E10#1U4U5A7I:10404E4U* Practice B NEZUMINO YOMEIRI (WEDDING OF A MOUSE) 1. Mukashi mukashi, 1U5A7I#1U5A7I# 2. okubukai yamano 05U1U5AI#7A1A40# 3. arumurani A3U1U3A7I* 4. nezuminooyakoga 4E4U1I4007A505A* 5. kurashiteimashita. 5U3A7I4EI1A7I4A* 6. Muraichibanno 1U3AI7I1A840* 7. bijintoyu: 1I7I8407U:# 8. hyo:banno 70:1A840# 9. musumenotameni 1U4U1E404A1E7I* 10. konoyode ichiban 5040704E#I7I1A8# 11. takumashiku 4A5U1A7I5U* 12. rippanahitoo 3I91A4A5I400* 13. omukosanni 01U504A87I* 14. mukaetaito 1U5AE4AI40* 15. oyanezumiwa 07A4E4U1I2A* 16. kibo :shiteimashita. 5I10:7I4EI1A7I4A# 17. Mazuhajimeni 1A4U5A7I1E7I* 18. erabaretanowa E3A1A3E4A402A* 19. ohisamadeshita. 05I4A1A4E7I4A* 20. Ohisamawa, 05I4A1A2A* 21. kumoni orwaretara 5U107I#0:2A3E4A3A* 22. jibunwa 7I1U82A* 23. tachimachi 4A7I1A7I* 24. mienakunatte 1IE4A5U4A94E* 25. shimaimasuyo 7I1AI1A4U70* 26. toiimashita. 40I:1A7I4A# 27. Demo, kumowa kumode 4E10#5U102A#5U104E# 28. bokuwa kazeniwa 105U2A#5A4E7I2A# 29. itsumo maketabakariiru. I4U10#1A5E4A1A5A3II3U# 30. Attoyu:mani A9407U:1A7I# 31. tobasarete 401A4A3E4E* 32. shimaimasukarane 7I1AI1A4U5A3A4E* 33. tonobemashita. 40401E1A7I4A* 34. Ippo, kazewa I910#5A4E2A# 35. donnamononimo 4084A10407I10* 36. bikutomoshinainowa 1I5U40107I4AI402A* 37. kabedesuyo. 5A1E4E4U70* 38. Bokutowa 105U402A* 39. kurabemononi 5U3A1E10407I* 40. narimasen-yo 4A3I1A4E8#70# 41. tokotaemashita. 40504AE1A7I4A* 42. Mottomo, 1094010* 43. nezumiga yowamino 4E4U1I5A#702A1I40# 44. kabewa i:mashita. 5A1E2A#I:1A7I4A# 45. Kimitachinezumini 5I1I4A7I4E4U1I7I* 46. masarumonowa 1A4A3U10402A* 47. imasen-yo. I1A4E8#70# 48. Bokuga 105U5A* 49. yononakade mottomo 70404A5A4E#1094010# 50. subarashv.to omounowa 4U1A3A7I:40#010U402A# 51. kimitachidesu. 5I1I4A7I4E4U* LESSON 2 Review Visual Only Presentation: pa Training Visual Only Presentation: ai ui ai ti ai oi ai ii ai ai ai ui ai ei ai oi ai ii ai ai Audio-visual Presentation: ai Practice A ARU AMERIKAJIN (AN AMERICAN) 1. Anekara kiita 2. hanashidearu. 3. Arutoshino aid, 4. anewa amerikano 5. shiriaino ieni 6. asobini ittakotogaaru. 7. Arunichiyo:bino asa, 8. meo somasuto 9. atarini 10. panno kaoriga 11. tadayotteita. 12. Asagohanni okuretewa 13. taihento, 14. awatete 15. fowo araj, 16. /leyao deta. 17. Daidokorono 18. rfoao a£ete, 19. aratto 20. foeo ageteshimatta. 21. Okusamadenaku 22. dannasamaga arekore 23. asagohanno shitakuo 24. shiteitanoda. 25. Aneomiruto, 26. sassoku akarui egaode 27. ateM/ ko.hi.o iretekureta. 28. Mada atatakai panni 29. bataito 30. amaz jamuo 31. tappuri nuttekureta. 32. Okusamawa 33. marfa arawarenai. 34. Hajimewa, 35. tsumani atamaga 36. agaranainokato 37. omottaga, 38. mainichiyo:bino 39. asagohanzukuriwa 40. kareni tottewa 41. mattaku atarimaeno 42. kotorashi:. 43. A/iewa, A4E5A3A#5II4A# 5A4A7I4EA3U* A3U407I40#A5I# A4E2A#A1E3I5A40# 7I3IAI40#IE7I# A401I7I#I94A50405AA3U# A3U7I7I70:1I40#A4A# 1E0#4A1A4U40# A4A3I7I* 1A840#5A03I5A# 4A4A7094EI4A* A4A505A87I#05U3E4E2A# 4AI5E840* A2A4E4E* 5AOO#A3AI# 5E7AO#4E4A# 4AI40503040* 40AO#A5E4E# A3A940* 50E0#A5E4E7I1A94A# 05U4A1A4E4A5U* 4A84A4A1A5A#A3E503E# A4A505A840#7I4A5UO# 7I4EI4A404A* A4E01I3U40* 4A9405U#A5A3UI#E5A04E# A4UI#50:51:0#I3E4E5U3E4A# 1A4A#A4A4A5AI#1A87I# 1A4A:40# A1AI#7A1U0# 4A91U3I#4U94E5U3E4A# 05U4A1A2A* 1A4A#A3A2A3E4AI# 5A7I1E2A* 4U1A7I#A4A1A5A# A5A3A4AI405A40* 01094A5A* 1AI7I7I70:1I40# A4A505A84U5U3I2A* 5A3E7I#4094E2A# 1A94A5U#A4A3I1AE40# 50403A7I:# A4E2A* 85 44. konotokibakariwa, 45. amerikajintono 46. kekkonni akogaretarashi:. 47. Ma:, imadewa 48. akirametayo:dearuga. 5040405I1A5A3I2A* A1E3I5A7I84040* 5E95087I#A505A3E4A3A7I:# 1A:#I1A4E2A# A5I3A1E4A70:4EA3U5A* Practice B NIHONGONO MOJI (JAPANESE CHARACTERS) 1. O.mukashi, 0:1U5A7I# 2. sagakuniniwa mojiga 4A5A5U7I7I2A#107I5A# 3. arimasendeshita. A3I1A4E84E7I4A* 4. Hajimete tsukawareta mojiwa, 5A7I1E4E#4U5A2A3E4A#107I2A# 5. imakara I 1 A 5 A 3 A * 6. nisennen chikakumo maeni, 7 I 4 E 8 4 E 8 # 7 I 5 A 5 U 1 0 # 1 A E 7 I # 7. chu:gokukara tsutawatta 7U:505U5A3A#4U4A2A94A# 8. kanjideshita. 5 A 8 7 I 4 E 7 I 4 A * 9. Furukukara 6U3U5U5A3A* 10. hanasaretekita 5 A 4 A 4 A 3 E 4 E 5 I 4 A * 11. kotobao 50401AO* 12. kakiarawasutameni 5A5IA3A2A4U4A1E7I* 13. atarashiku kanjiga A4A3A7I5U#5A87I5A# 14. atehameraremashita. A4E5A1E3A3E1A7I4A* 15. Jidaiga kawari, 7I4AI5A#5A2A3I# 16. yagate kanaga 7A5A4E#5A4A5A# 17. hatsumeisaremashita. 5A4U1EI4A3E1A7I4A* 18. Kananiwa 5A4A7I2A* 19. hiraganato katakanaga 5I3A5A4A40#5A4A5A4A5A# 20. arimasuga, A3I1A4U5A* 21. dochiraga hayaku 407I3A5A#5A7A5U# 22. tsukuraretanokawa 4U5U3A3E4A405A2A# 23. cunari yokuwa A1A3I#705U2A# 24. wakatteimasen. 2A5A94EI1A4E8* 25. Shikashinagara, 7I5A7I4A5A3A# 26. imano katachiga I1A40#5A4A7I5A# 27. dekiagaru kateiniwa, 4E5IA5A3U#5A4EI7I2A# 28. kanarinagai jikanga 5A4A3I4A5AI#7I5A85A# 29. kakattakotowa 5A5A94A50402A* 30. tashikadesu. 4A7I5A4E4U* 31. Hiraganawa, 5I3A5A4A2A* 32. dochirakatoieba, 407I3A5A40IE1A* 33. onnanohitoga, tegami 084A405I405A#4E5A1I# 34. aruiwa wakao A3UI2A#2A5A0# 35. kakunoni tsukawaremashita. 5A5U407I#4U5A2A3E1A7I4A# 36. Ippo:, katakanawa, I910:#5A4A5A4A2A# 37. otokonohitoga 04050405I405A* 38. kanbun-o yomutameni 5A81U8#0#701U4A1E7I# 39. kangaedasareta mojidesu. 5A85AE4A4A3E4A#107I4E4U# 40. Kanamojino okagede, 5A4A107I40#05A5E4E# 41. wareraga sosentachino 2A3E3A5A#404E84A7I40# 42. yomikakino chikarawa 701I5A5I40#7I5A3A2A# 43. hijo:ni takamarimashita. 5I70:7I#4A5A1A3I1A7I4A# LESSON 3 Review Visual Only Presentation: aa pa Training Visual Only Presentation: ia oa ia ea ia ia ia aa ia ua ia ia ia ua ia oa ia ea ia aa Audio-visual Presentation: ia Practice A KYO:TO (KYOTO) 1. Kyo.towa nishinihonno 570:402A#7I7I7I50840# 2. daihyoitekina kankoxhino 4AI70:4E5I4A#5A850:7I40# 3. hitotsudesu. 5I404U4E4U* 4. Nagai rekishio mochi, 4A5AI#3E5I7IO#107I# 5. machino naigai 1A7I40#4AI5AI# 6. itarutokoroni I4A3U4050307I* 7. kicho.na bunkaisanga 5I70:4A#1U85AI4A85A# 8. nokosareteimasu. 40504A3E4EI1A4U* 9. Shinkansenno hikarigoinara 7I85A84E840#5I5A3I50:4A3A# 10. Tokyo:ekikara 40570:E5I5A3A# 11. sanjikaninaini tsukimasu. 4A87I5A7I4AI7I#4U5I1A4U# 12. Furuitodo:jini 6U3UI4040:7I7I# 13. kindaitoshidemo 5I84AI407I4E10* 14. arimasukara A3I1A4U5A3A* 15. wakai hitotachinimo 2A5AI#5I404A7I7I10# 16. hijo:ni 5170:71* 17. ninkiga arimasu. 7I85I5A#A3I1A4U# 18. Ichinen-o tsu.jite I7I4E8#0#4U:7I4E# 19. iroirona gyojiga I30I304A#570:7I5A# 20. toriokonawaremasu. 403I0504A2A3E1A4U* 21. Aoimatsuri, AOI1A4U3I* 22. Jidaimatsuri, 7I4AI1A4U3I* 23. Daimonjiyakiwa 4AI1087I7A5I2A* 24. ippanno hitonimo I91A840#5I407I10# 25. hiroku shirareteimasu. 5I305U#7I3A3E4EI1A4U# 26. Ichidodei:kara I7I404EI:5A3A# 27. jissaini mitemiruto 7I94AI7I#1I4E1I3U40# 28. i:desho:. I:4E70:# 29. Kyo:toniwa 570:407I2A# 30. shikioriorino 7I5I03I03I40* 31. utsukushisaga ari, U4U5U7I4A5A#A3I# 32. donokisetsuni ittemo 40405I4E4U7I#I94E10# 33. kokoro hikarezuniwa 505030#5I5A3E4D7I2A# 34. iraremasen. I3A3E1A4E8* 35. Tori-wake 403I2A5E* 36. momijino jikiwa 101I7I40#7I5I2A# 37. inshoitekidesu. I870:4E5I4E4U# 38. Akiniwa A5I7I2A* 39. sumikitta ku:kinonaka 4U1I5I94A#5U:5I404A5A* 40. achikochi A7I507I* 41. saikuringusurunomo 4AI5U3I85U4U3U4010* 42. tanoshiku 4A407I5U* 43. kimochinoi:monodesu. 5I107I40I:10404E4U# Watashino shiriaiwa itsumo ujikintokini shitatsuzumio uchi, kiyomizuyaki, nishijin-ori, soshite yatsuhashio omiyageni kaimasu. 2A4A7I40#7I3IAI2A# I4U10#U7I5I8405I7I# 7I4A4U4U1I0#U7I# 5I701I4U7A5I* 7I7I7I8#03I# 407I4E#7A4U5A7I0# 01I7A5E7I#5AI1A4U# Practice B HIKO:KI (AN AIRPLANE) 1. Hiko.kiwa ochirukara 5150:5I2A#07I3U5A3A# 2. kiraidatono 5I3AI4A4040* 3. ikenga o:i. I5E85A#0:I# 4. Ririkushitekara 3I3I5U7I4E5A3A* 5. chakurikusurumadewa 7A5U3I5U4U3U1A4E2A* 6. ikitakokochiga I5I4A50507I5A* 7. shinaitoyu: 7I4AI407U:# 8. koemo 50E10# 9. hinpanni kikareru. 5I81A87I#5I5A3E3U# 10. Tashikani 4A7I5A7I* 11. ittan nanikaga I94A8#4A7I5A5A# 12. okiruto 05I3U40* 13. inochio ushinau I407I0#U7I4AU# 14. kakuritsuwa takai. 5A5U3I4U2A#4A5AI# 15. Ichikabachikano I7I5A1A7I5A40* 16. seishitsuga 4EI7I4U5A* 17. tsuyoinowa 4U70I402A* 18. jijitsuda. 7I7I4U4A* 19. Shikashi, 7I5A7I* 20. mainichi ikutsumo 1AI7I7I#I5U4U10# 21. okiteiru 05I4EI3U* 22. jido:shajikono 7I40:7A7I5040# 23. hindoni 5184071* 24. hikakushitara 5I5A5U7I4A3A* 25. ko:ku:ki tsuirakuno 50:5U:5I#4UI3A5U40# 26. hanashiwa 5A4A7I2A* 27. mettani kikanai. 1E94A7I#5I5A4AI# 28. Jitsuni anshinshite 7I4U7I#A87I87I4E# 29. riyo.dekiru 3I70:4E5I3U# 30. ko:tsu:kikanto ierunoda. 50: 4U: 5I5A840#IE3U404A# 31. Hiko:kiwa tsugitsugini 5I50:5I2A#4U5I4U5I7I# 32. seino:no takai 4EI40:40#4A5AI# 33. shingataga 7I85A4A5A* 34. kaihatsusareteiru. 5AI5A4U4A3E4EI3U* 35. Atarashi: enjin-o A4A3A7I:#E87I8#0# 36. riyo:shi, 3170:71* 37. iroirona I30I304A* 38. kaizen-o hete 5AI4E8*0#5E4E# 39. tsukuriagerareta 4U5U3IA5E3A3E4A* 40. yori hayai binga, 703I#5A7AI*1I85A* 41. ima sekaino I1A#4E5AI40# 42. kakutoshimezashite 5A5U407I1E4A7I4E* 43. tondeiru. 4084EI3U* LESSON 4 Review Visual Only Presentation: aa ia pa Training Visual Only Presentation: ua oa ua ea ua ia ua aa ua ua ua ia ua ua ua ea ua oa ua aa Audio-visual Presentation: ua 92 Practice A AKI (AUTUMN) 1. Atsui natsuga sugi, A4UI#4A4U5A#4U5I# 2. kugatsumo 5U5A4U10* 3. chu:junni naruto 7U:7U87I#4A3U40# 4. akino kehaiga A5I40#5E5AI5A# 5. kokunaru. 505U4A3U* 6. Tentakaku umakoyuru kisetsuno 4E84A5A5U#U1A507U3U#5I4E4U40# 7. otozuredearu. 0404U3E4EA3U* 8. Masshiroi kumoga 1A97I30I#5U105A# 9. sunda aozorani ukabu. 4U84A#A0403A7I#U5A1U# 10. Sujigumo, urokogumo, 4U7I5U10#U30505U10# 11. iwashigumo nadodearu. I2A7I5U10#4A404EA3U# 12. Sugasugashi: tenkiga 4U5A4U5A7I:#4E85I5A# 13. tsuzukutame, 4U4U5U4A1E* 14. supo.tsuni 4U10:4U7I# 15. aseonagasuhitoga A4E04A5A4U5I405A* 16. tokuni medatsu. 405U7I#1E4A4U# 17. Zenkokukakuchide, 4E8505U5A5U7I4E* 18. undo:kaiga okonawareru. U840:5AI5A#0504A2A3E3U# 19. Mata, ko.rakuchiwa 1A4A#50:3A5U7I2A* 20. kazokuzureya 5A405U4U3E7A* 21. guru:pude 5U3U:1U4E# 22. nigiwau. 7I5I2AU* 23. Ku:kimo oishiku 5U:5I10#OI7I5U# 24. shokuyokumo waku. 705U705U10#2A5U# 25. Kudamonoohajime 5U4A104005A7I1E* 26. akino mikakuga A5I40#1I5A5U5A# 27. kakushu ho.huni desorou. 5A5U7U#50:5U7I#4E4030U# 28. Shufutachiwa, 7U6U4A7I2A* 29. matsutakeya 1A4U4A5E7A* 30. kurio tsukatte 5U3I0#4U5A94E# 31. ryorino 370:3140* 32. udeofuruu. U4E06U3UU* 33. Geijutsusakuhinni 5EI7U4U4A5U5I87I* 34. fureru chansumo 6U3E3U#7A84U10# 35. o:kunaru. 0:5U4A3U# 36. Ongakukai, 085A5U5AI* 37. bijutsutennadomo 1I7U4U4E84A4010* 38. itsuninaku konzatsusuru. I4U7I4A5U#508 4A4U4U3U# 39. Kuwaete, 5U2AE4E* 40. dokushonetsumo 405U7O4E4U1O* 41. kyu:ni takamaru. 57U:7I#4A5A1A3U# 42. Yoruwa tsukiga 703U2A#4U5I5A# 43. akaruku kagayaku. A5A3U5U#5A5A7A5U# Mangetsuno utsukushisawa kakubetsudearu. Shizentono fukai musubitsukio itsumademo ushinawazuni itaimonodearu. 1A85E4U40#U4U5U7I4A2A# 5A5U1E4U4EA3U* 7I4E84040* 6U5AI#1U4U1I4U5I0# I4U1A4E10#U7I4A2A4U7I# I4AI10404EA3U* Practice B ANRI = FA:BURU (JEAN HENRI FABRE) 1. Konchu:kide 2. yokushirareteiru 3. Fa.buruwa, 4. Furansuno 5. mazushi: no:funo 6. musukotoshite umareta. 7. Itsumadetattemo 8. karenouchino 9. kurashimukiwa 10. hitotsumo rakuniwa 11. naranakatta. 12. Kurushi: seikatsuno . 13. renzokunimokakawarazu, 14. £arewa gakumonde 15. miotatetai toyu: 16. tsuyoikiboio 17. itsumo motteita. 18. Fa.buruga zutto 19. idakitsuzuketeita yumewa 20. su:gakushani 21. narukotodeatta. 22. Shikashi Faburuwa, 23. jitsuwa 24. do:butsugaku, 25. shokubutsugakunitsuitemo 26. sugureta saino:no 27. mochinushideatta. 28. Aruhi, 29. aru yu:meina 30. do:butsugakushakara 31. konchu:no kenkyu:o 32. susumerareta karewa, 33. mayoukotonaku 34. mushino kansatsuni 35. torikunda. 36. Katatsumuri, matsumushi, 37. mitsubachinadoo 38. kuruhimo kuruhimo 39. mitsuzuketa. 40. Hitotsuhitotsu.no 41. tsumikasanega 42. sanju:nengo 43. jukkanno 5087U:5I4E# 705U7I3A3E4EI3U* 6A:1U3U2A# 6U3A84U40* 1A4U7I:#40:6U40# 1U4U50407I4E#U1A3E4A# I4U1A4E4A94E10* 5A3E40U7I40* 5U3A7I1U5I2A* 5I404U10#3A5U7I2A# 4A3A4A5A94A* 5U3U7I:#4EI5A4U40# 3E8405U7I105A5A2A3A4U* 5A3E2A#5A5U1084E# 1I04A4E4AI#407U:# •4U70I5I10:0# I4U10#1094EI4A# 6A:1U3U5A#4U940# I4A5I4U4U5E4EI4A#7U1E2A# 4U:5A5U7A7I* 4A3U50404EA94A* 7I5A7I#6A1U3U2A# 7I4U2A* 40:1U4U5A5U* 705U1U4U5A5U7I4UI4E10* 4U5U3E4A#4AI40:40# 107I4U7I4EA94A* A3U5I* A3U#7U:1EI4A# 40:1U4U5A5U7A5A3A# 5087U:40#5E857U:0# 4U4U1E3A3E4A#5A3E2A# 1A70U50404A5U* 1U7I40#5A84A4U7I# 403I5U84A* 5A4A4U1U3I#1A4U1U7I# 1I4U1A7I4A400* 5U3U5I10#5U3U5I10# 1I4U4U5E4A* 5I404U5I404U40* 4U1I5A4A4E5A* 4A87U:4E850# 7U95A840* konchu:kitoshite shuppansareta. Senkyu:hyakuju:gonen, kagakunoshijinto utawareta Fa:buruwa kyu:ju:nisaino sho:gaini shu:shifuo utta. 5087U:5I407I4E# 7U91A84A3E4A* 4E857U:7A5U7U:504E8# 5A5A5U407I7I840* U4A2A3E4A#6A:1U3U2A* 57U:7U:7I4AI40#70:5AI7I# 7U:7I6U0#U94A# LESSON 5 Review Visual Only Presentation: aa ia ua pa Training Visual Only Presentation: wa pa wa sha wa ra wa wa wa pya wa ta wa kya wa ma wa nya wa ha wa hya wa ga wa ja wa bya wa fa wa mya wa ka wa gya wa za wa ba wa cha wa tsa wa rya wa da wa sa wa ya wa na Audio-visual Presentation: wa Practice A WATASHINO KAZOKU (MY FAMILY) 1. Watashino iewa 2. Kanagawaken 3. Fujisawashini aru. 4. Ryo:shinwa futaritomo 5. sho.~waumarede, 6. kodomowa watashito 1. futatsuchigaino imo:to. 8. Chichiwa 9. Wakayamaken shusshin. 10. Hahawa Ishikawakenno 11. Wajima shusshin. 12. Chichino shigotowa 13. kiwamete tenkinga o:ku, 14. watashitachiwa 15. kitawa hokkaidoino 16. Wakkanai, 17. minamiwa Okinawamade 18. mawatta. 19. Chichiwa 20. wakarazuyadaga, 21. hahawa wariai 22. monowakariga i:. 23. Wafaafforowtf 24. sorevva sorewa 25. kawairashikattato 26. hitono uwasanikiku. 27. Demo wareware 28. shimainitottewa 29. kowai sonzaideshika nakatta. 30. Ma:, warerano chichioyaniwa 31. fusawashi: hitodato omowareru. 32. P/<3ga imo:towa 33. mawarijukara 34. kawaigararetebarari. 35. Iwayuru wagamamamusumeda. 36. Monowasureno hidosawa 37. niwatoridesae 38. kanawanaidaw.to 39. omowaseruhododa. 2A4A7I40#IE2A# 5A4A5A2A5E8* 6U7I4A2A7I7I#A3U# 370:7I82A#6U4A3I4010# 70:2AU1A3E4E# 5040102A#2A4A7I40# 6U4A4U7I5AI40#I10:40# 7I7I2A* 2A5A7A1A5E8#7U97I8# 5A5A2A#I7I5A2A5E840# 2A7I1A#7U97I8# 7I7I40#7I50402A# 5I2A1E4E#4E85I85A#0:5U# 2A4A7I4A7I2A* 5I4A2A#5095AI40:40# 2A95A4AI* 1I4A1I2A#05I4A2A1A4E# 1A2A94A# 7I7I2A* 2A5A3A4U7A4A5A* 5A5A2A#2A3IAI# 10402A5A3I5A#I:# 2A5AI50302A* 403E2A#403E2A# 5A2AI3A7I5A94A40* 5I4040#U2A4A7I5I5U# 4E10#2A3E2A3E# 7I1AI7I4094E2A* 502AI#4084AI4E7I5A#4A5A94A# 1A:#2A3E3A40#7I7I07A7I2A# 6U4A2A7I:#5I404A40#0102A3E3U# 2A5A#I10:402A# 1A2A3I7U5A3A* 5A2AI5A3A3E4E1A3A3I* I2A7U3U#2A5A'1A1A1U4U1E4A# 10402A4U3E40#5I404A2A# 7I2A403I4E4AE* 5A4A2A4AI4A30:40# 0102A4E3U50404A# Practice B WAFU: (JAPANESE STYLE) 1. Chu:gokudewa mukashi 7U : 505U4E2A#1U5A7 I# 2. wagakunino kotoo 2A5A5U7 I 40#50400# 3. "wa "toyondeimashita. " 2 A " 4 0 7 0 8 4 E I 1 A 7 I 4 A # 4. "Wajiri'toieba, " 2 A 7 I 8 " 4 0 I E 1 A # 5. "wa"nokuniniswnu " 2 A " 4 0 5 U 7 I 7 I 4 U 1 U # 6. warewarenokotoo 2 A 3 E 2 A 3 E 4 0 5 0 4 0 0 * 7. sashitawakedesu. 4 A 7 I 4 A 2 A 5 E 4 E 4 U * 8. Tsukawareru mojikoso 4 U 5 A 2 A 3 E 3 U * 1 0 7 1 5 0 4 0 * 9. kawarimashitaga, 5 A 2 A 3 I 1 A 7 I 4 A 5 A * 10. "wa"notsuku kotobawa " 2 A "404U5U#50401A2A# 11. nihonfu: toyudmio 7 I 5086U :#407U : I 1 I 0 # 12. arawasukotoga A3A2A4U50405A* 13. wakarimasu. 2A5A3I1A4U* 14. Imawa I1A2A* 15. "heiwa"no "wa"nojiga " 5 E I 2 A " 4 0 # " 2 A " 4 0 7 I 5 A # 16. tsukawareteimasu. 4 U 5 A 2 A 3 E 4 E I 1 A 4 U * 17. Korewa, iroirona kotobatono 503E2A#I30I304A#50401A4040# 18. kumiawasega 5U1IA2A4E5A* 19. kano:na mojidesu. 5A40:4A#107I4E4U# 20. Tabemonoo arawasutokiwa 4A1E10400#A3A2A4U405I2A# 21. washoku. 2A705U* 22. Kimonono baaiwa 5I104040#1AAI2A# 23. wafuku. 2A6U5U* 24. Hokaniwa, 505A7I2A* 25. waka, wabun, washitsu, 2A5A#2A1U8#2A7I4U# 26. wasai, wagashi. 2A4AI#2A5A7I# 27. Kuwaete, saikindewa, 5U2AE4E#4AI5I84E2A# 28. wafu:yo:gashi toka 2A6U:70:5A7I#405A# 29. yo:furwagashi, 70:6U:2A5A7I# 30. aruiwa A3UI2A* 31. wafu:yo:shoku, 2A6U:70:705U# 32. yo :fu rwashokunado 70:6U:2A7O5U4A40# 33. wakenowakaranai monoga 2A5E402A5A3A4A I#10405A# 34. demawatteimasu. 4 E 1 A 2 A 9 4 E I 1 A 4 U * 35. Aruhitoni iwasereba, A3U5I407I#I2A4E3E1A# 36. Icoremo wayo:secchu:to 503E10#2A70:4E97U:40# 37. yu:kotorashi:. 7U:50403A7I:# 38. "Wa "toyu :kotobawa "2A"407U:50401A2A# 39. sunawachi 4U4A2A7I* 40. cho rwatoyu :kotoni 70:2A407U:504071* 41. naruwakeka. 4A3U2A5E5A* LESSON 6 Review Visual Only Presentation: aa pa ia wa ua Training Visual Only Presentation: oa oa oa oa oa oa oa oa oa oa ua ea oa ia aa ua aa ia oa ea Audio-visual Presentation: oa Practice A KYO:NO TENKI (WEATHER FORECAST) 1. Zenkoku.no minasama 2. ohayo.gozaimasu. 3. Kyo:no otenkio 4. otsutaeitashimasu. 5. Kyoitno zenkokutekini 6. yoiotenkito 7. narumikomidesu. 8. Yoso:saiko:kionwa 9. To:kyo:ga nijuryondo, 10. osakaga onajiku 11. niju.yondo, 12. fukuokaga 13. niju:gododesu. 14. Ko:suikakuritsuwa, 15. gozen gogotomoni 16. zerokara 17. gopa.sentodesu. 18. Kokonotokoro, 19. zenkokutekini 20. odayakanayo:kiga 21. tsuzuiteimashitaga, 22. koremo do:yara 23. konshu:ippainoyo:desu. 24. Nihonretto:o ootteita 25. ko:kiatsuga 26. higashini nuke 21. nishinoho:kara 28. ote/i&t o:habani 29. kuzurerumikomidesu. 30. Kionmo godohodo 31. hikukunarutono 32. yoso:desu. 33. Sorosoro 34. oshimeriga hoshi:to 35. omowareteitakorodesuga, 36. kisho:cho:no 37. ho:kokuniyorimasuto, 38. tokoroniyottewa 39. o:mameninaru osoremo 40. arutoyuikotodesunode 41. cfo:zo gochu:ikudasai. 4E8505U40#1I4A4A1A# 05A70:504AI1A4U* 570:40#04E85IO# 04U4AEI4A7I1A4U* 570:10#4E8505U4E5I7I# 70I04E85I40* 4A3U1I501I4E4U* 7040:4AI50:5I082A# 40:570:5A#7I7U:70840# 0:4A5A5A#04A7I5U# 7I7U:70840# 6U5U05A5A# 7I7U:50404E4U# 50:4UI5A5U3I4U2A# 5O4E8#505O4OlO7I# 4E305A3A* 501A:4E8404E4U# 505040405030* 4E8505U4E5I7I* 04A7A5A4A70:5I5A# 4U4UI4EI1A7I4A5A# 503E10#40:7A3A# 5087U:I91AI4070:4E4U# 7I5083E940:0#0094EI4A# 50:5IA4U5A# 5I5A7I7I#4U5E# 7I7I4050:5A3A# 04E85I#0:5A1A7I# 5U4U3E3U1I501I4E4U* 510810*50405040* 5I5U5U4A3U4040* 7040:4E4U# 40304030* 07I1E3I5A*507I:40* 0102A3E4EI4A50304E4U5A* 5170:70:40* 50:505U7I703I1A4U40# 4050307I7094E2A* 0:1A1E7I4A3U#0403E10# A3U407U:50404E4U404E* 40:40#507U:I5U4A4AI* 101 Practice B KOINOBORI (CARP STREAMERS) 1. Gogatsuitsukawa 505A4UI4U5A2A* 2. kodomonohide oyasumidesu. 504010405I4E#07A4U1I4E4U# 3. Konohiwa 50405I2A* 4. "tangonosekku"toyobareru "4A850404E95U"40701A3E3U# 5. otokonokono omatsuridemoarimasu. 04050405040#01A4U3I4E10A3I1A4U# 6. Kintaro.nadono 5I84A30:4A4040# 7. oningyo:o kazari, 07I8570:0#5A4A3I# 8. koinoborio tatete 50I40103I0#4A4E4E# 9. oiwaishimasu. OI2AI7I1A4U* 10. Iroazayakana I30A4A7A5A4A* 11. o:kinakoiga 0:5I4A50I5A* 12. aozorao ikioiyoku A0403AO#I5IOI705U# 13. oyogu yo:suwa, 0705U#70:4U2A# 14. honto:ni kimochiyoimonodesu. 50840:7I#5I107I70I10404E4U# 15. Musukoya magono 1U4U507A#1A5040# 16. sukoyakana seichoto negau 4U507A5A4A#4EI70:0#4E5AU# 17. chichioya, hahaoya, 7I7I07A#5A5A07A# 18. ojiichan, obaichanno 07I:7A8#01A:7A840# 19. kokoroga yoku wakarimasu. 5050305A#705U#2A5A3I1A4U# 20. Tokorode, 4050304E* 21. onnanokoga yorokobu 084A40505A#7030501U# 22. omatsuriwa 01A4U3I2A* 23. nikagetsuhodo 7I5A5E4U5040* 24. hayaku okonawareru 5A7A5U#0504A2A3E3U# 25. momonosekkudesu. 1010404E95U4E4U* 26. Kochirano oiwaiwa, 507I3A40#0I2AI2A# 27. go:kana ohinasamao 50:5A4A#05I4A4A1A0# 28. kazarimasu. 5A4A3I1A4U* 29. Hinaningyoin'cwa 5I4A7I8570:7I2A# 30. odairisamato ohimesamanohokani, 04AI3I4A1A40#05I1E4A1A40505A7I# 31. sanninkanjoya 4A87I85A8707A# 32. goninbayashinadoo 507I81A7A7I4A400* 33. fukumukotoga oddesu. 6U5U1U50405A#0:I4E4U# LESSON 7 Review Visual Only Presentation: aa pa ia wa ua oa Training Visual Only Presentation ea ua ea ea ea oa ea ia ea ea ea aa ea ia ea aa ea oa ea ua Audio-visual Presentation: ea Practice A KATEISA:BISU (SPENDING TIME WITH MY FAMILY) 1. Watashiwa bo.ekigaishao 2. keieishiteiru. 3. Eikokuni 4. dekakerukotoga o:ku, 5. ieni irunowa 6. kazoerukuraida. 7. Nippon-e 8. kaettatokiwa 9. kazokueno sa:bisuo 10. dekirudake kokorogakeru. 11. Minnano egaowa 12. menimienai sasaeda. 13. Kesawa rokujimaekara 14. suemusumeno 15. genkinakoega kikoeru. 16. So:ieba, 17. yoxhienno 18. ensokuda. 19. Do:butsuen-e 20. tsureteitte moraerurashi:. 21. Kaeriwa 22. watashiga ekimade 23. mukaenideru. 24. Sukeju:ruo 25. machigaenaiyo: 26. kiotsukenakereba. 21. Asuwa ueno musumeto 28. kodomomukeno 29. eigaomini 30. machie dekakeru. 31. Benkyo:zukueo 32. kaikaeru yakusokumo 33. shikkari oboeteoko:. 2A4A7I2A#10:E5I5AI7A0# 5EIEI7I4EI3U* EI505U7I* 4E5A5E3U50405A#0:5U# IE7I#I3U402A# 5A40E3U5U3AI4A* 7I9108#E# 5AE94A405I2A* 5A405UE40#4A:1I4UO* 4E5I3U4A5E#5050305A5E3U# 1I84A40#E5A02A# 1E7I1IE4AI#4A4AE4A# 5E4A2A#305U7I1AE5A3A# 4UE1U4U1E40* 5E85I4A50E5A#5I50E3U# 40:IE1A# 70:7IE840# E8405U4A* 40:1U4UE8#E# 4U3E4EI94E#103AE3U3A7I:# 5AE3I2A* 2A4A7I5A#E5I1A4E# 1U5AE7I4E3U* 4U5E7U:3UO# 1A7I5AE4AI70:* 5I04U5E4A5E3E1A* A4U2A#UE40#1U4U1E40# 5040101U5E40* EI5A01I7I* 1A7IE#4E5A5E3U# 1E8570:4U5UE0# 5AI5AE3U#7A5U405U10# 7I95A3I#010E4E050:# Practice B KEMURI (SMOKE) 1. Kemuriwa chi:sana tsubuno 5E1U3I2A#7I:4A4A#4U1U40# 2. atsumaridearu. A4U1A3I4EA3U* 3. Kemurino tsubuno 5E1U3I40#4U1U40# 4. hitotsuhitotsuwa 5I404U5I404U2A* 5. ningenno medewa 7I85E840#1E4E2A# 6. miwakerukotowa dekinai. 1I2A5E3U50402A#4E5I4AI# 7. Keredomo, 5E3E4010* 8. kenbikyo:o tsukaeba, 5E81I570:0#4U5AE1A# 9. sono keitaiga 4040#5EI4AI5A# 10. hakkiri mieru. 5A95I3I#1IE3U# 11. Entotsukara deru E8404U5A3A#4E3U# 12. kuroi kemurio 5U30I#5E1U3I0# 13. keiseishiteiru tsubuwa 5EI4EI7I4EI3U#4U1U2A# 14. kotaidearu. 504AI4EA3U* 15. Ippo:, I910:# 16. tabakono kemurio 4A1A5040#5E1U3I0# 17. keiseishiteirunowa 5EI4EI7I4EI3U402A* 18. dochirakatoieba 407I3A5A40IE1A* 19. ekitaino tsubuto E5I4AI40#4U1U40# 20. kangaerareteiru. 5A85AE3A3E4EI3U* 21. Yugemo mata 7U5E10#1A4A# 22. kemurino nakamato ieru. 5E1U3I40#4A5A1A40#IE3U# 23. Kemurito kenko:no 5E1U3I40#5E850:40# 24. kankeini tsuiteno 5A85EI7I#4UI4E40# 25. jikkenya kenkyu.mo 7I95E47A#5E857U:10# 26. fueteiru. 6UE4EI3U* 27. Kenko.ni 5E850:7I# 28. warui eikyo:o 2A3UI#EI570:0# 29. ataeru kemurinonakade, A4AE3U#5E1U3I404A5A4E# 30. mazu agerarerunoga 1A4U#A5E3A3E3U405A# 31. tabakodearu. 4A1A504EA3U* 32. Kin-enga 5I8#E85A# 33. sakebareru gendaiwa 4A5E1A3E3U#5E84AI2A# 34. ken-enken toyu: 5E8#E85E8#407U:# 35. kotobasae umareta. 50401A4AE#U1A3E4A# 36. Seiketsuna ku:kio eru 4EI5E4U4A#5U:5IO#E3U# 31. kenrinitsuitewa 5E83I7I4UI4E2A# 38. samazamana ikenga 4A1A4A1A4A#I5E85A# 39. aruyo:daga, A3U70:4A5A# 40. kono kenriwa 5040#5E83I2A# 41. kesshite 5E97I4E* 42. mushishitewa ikenai. 1U7I7I4E2A#I5E4AI# 43. Nipponmo 71910810* dekirudake hayaku kitsuenmo kin-enmo eraberu shakaini kaete ikanakerebanaranai. 4E5 I3U4A5E#5A7A5U# 5 I 4 U E 8 1 0 * 5 I8#E810# E3A1E3U#7A5AI7I# 5 A E 4 E # I 5 A 4 A 5 E 3 E 1 A 4 A 3 A 4 A I # LESSON 8 Review Visual Only Presentation: aa pa ia wa ua oa ea Training Visual Only Presentation: ra pa ra sha ra ra ra wa ra pya ra ta ra kya ra ma ra nya ra ha ra hya ra gya ra ka ra ja ra bya ra fa ra my a ra tsa ra rya ra za ra ba ra cha ra da ra ga ra sa ra ya ra na Audio-visual Presentation: ra Practice A KAZENO YOBO: (HOW NOT TO CATCH A COLD) 1. Daredemo kenko:niwa 4A3E4E10#5E850:7I2A# 2. sukunakarazu 4U5U4A5A3A4U* 3. kanshin-o motteiru. 5A87I8#0#1094EI3U# 4. Nenreiga 4E83EI5A* 5. takakunarunitsure, 4A5A5U4A3U7I4U3E* 6. karadano hatarakinimo 5A3A4A40#5A4A3A5I7I10# 7. henkaga arawareru. 5E85A5A#A3A2A3E3U# 8. Korekarawa masumasu 503E5A3A2A#1A4U1A4U# 9. ko.reikaga susumuto 50:3E:5A5A#4U4U1U40# 10. yoso:sareru. 7040:4A3E3U# 11. Soreni tomonai 403E7I#40104AI# 12. do:shitara yamaini 40:7I4A3A#7A1AI7I# 13. kakarazunisumukani 5A5A3A4U7I4U1U5A7I* 14. ju.tenga okarerudaro:. 7U:4E8 5A#05A3E3U4A30:# 15. Tokorode, 4050304E* 16. arayuru byo:kinonakade, A3A7U3U#170:5I404A5A4E* 17. kazewa dochirakatoyu:to 5A4E2A#407I3A5A407U:40# 18. karuku miraregachida. 5A3U5U#1I3A3E5A7I4A# 19. Keredomo, 5E3E4010* 20. korewa mukashikara 503E2A#1U5A7I5A3A# 21. manby.onomoto toxware, 1A817:0401040#40I2A3E# 22. mareni taibyo:ni 1A3E7I#4AI170:7I# 23. tsunagaranaitomo v.kirenai. 4U4A5A3A4AI4010#I:5I3E4AI* 24. Kazeni kakattara, 5A4E7I#5A5A94A3A# 25. korekuraito 503E5U3AI40* 26. hottarakashini shinaikotoda. 5094A3A5A7I7I#7I4AI50404A# 27. Yobo.nimo fudankara 7010:7110#6U4A8 5A3A# 28. chu dshinakerebanaranai. 7U:I7I4A5E3E1A4A3A4AI* 29. Sotokara kaettara 40405A3A#5AE94A3A# 30. ugaio reiko.shi, U5AIO#3EI50:7I# 31. kanarazu teo 5A4A3A4U#4EO# 32. kireini arau. 5I3EI7I#A3AU# 33. Soreigainimo, 403EI5AI7I10* 34. eiyo:no baransuno EI70:40#1A3A84U40# 35. yokutoreta shokujio 705U403E4A#705U7IO# 36. toranakerebanaranai. 403A4A5E3E1A4A3A4AI* 37. Izureniseyo, I4U3E7I4E70# 38. byo:genkinni furetemo 170:5E85I87I#6U3E4E10# 39. soreni makenai chikarao 403E7I#1A5E4AI#7I5A3A0# 40. higorokara tsuketeokukotoda. 5I50305A3A#4U5E4E05U50404A# 108 Practice B BAREE (BALLET) 1. Wakaikorokara, 2A5AI50305A3A* 2. bareriinani 1A3E3I:4A7I# 3. akogareteita. A505A3E4EI4A* 4. Arewa A3E2A* 5. rokusaikurainotokida. 305U4AI5U3AI40405I4A* 6. Terebidemita 4E3E1I4E1I4A* 7. Sorenno bareedanno 403E840#1A3EE4A840# 8. rainichiko:enno 3AI7I7I50:E840# 9. subarashisani 4U1A3A7I4A7I* 10. miserarete shimattanoda. 1I4E3A3E4E#7I1A94A404A# 11. Kareinamaino 5A3EI4A1AI40* 12. renzokuni 3E8405U7I* 13. mitOrerubakaridatta. 1I403E3U1A5A3I4A94A* 14. Sorekaratoyu .mono, 403E5A3A407U:1040* 15. jibunmo naraitakute naraitakute 7I1U810#4A3AI4A5U4E#4A3AI4A5U4E# 16. tamaranakattaga, 4A1A3A4A5A94A5A* 17. zannennagara, 4A84E84A5A3A* 18. jitsugenniwa itaranakatta. 7I4U5E87I2A#I4A3A4A5A94A# 19. Tsurai renshu:ni 4U3AI#3E87U:7I# 20. taerareru seikakudewa 4AE3A3E3U#4EI5A5U4E2A# 21. naikara 4AI5A3A* 22. yokattanokamoshirenai. 705A94A405A107I3E4AI* 23. Naraukotowa 4A3AU50402A* 24. akirameta watashidaga. A5I3A1E4A#2A4A7I4A5A# 25. sorezoreno sakuhinnitsuite 403E403E40#4A5U5I87I4UI4E# 26. arayurukotoo A3A7U3U50400* 27. shirabeta. 7I3A1E4A* 28. Mochiron kurashikkuni 107I308#5U3A7I95U7I# 29. kagiranai. 5A5I3A4AI* 30. Watashino sho:raino yumewa, 2A4A7I40#70:3AI40#7U1E2A# 31. bareeno rekishino 1A3EE40#3E5I7I40# 32. arukunio A3U5U7IO* 33. otozurerukotoda. 0404U3E3U50404A* 34. Achiranimo Kochiranimo A7I3A7I10#507I3A7I10# 35. ikitakute tamaranai. I5I4A5U4E#4A1A3A4AI# 36. Konoyumewa osoraku 50407U1E2A#0403A5U# 37. korekaramo 503E5A3A10* 38. kawaranaidaro: 5A2A3A4AI4A30:# 39. Kanaerarerunowa 5A4AE3A3E3U402A* 40. dorekurai 403E5U3AI* 41. sakininaruyara. 4A5I7I4A3U7A3A* LESSON 9 Review Visual Only Presentation: aa pa ia wa ua ra oa ea Training Visual Only Presentation: ta ja ta fa ta wa ta rya ta ra ta mya ta kya ta ma ta nya ta ha ta na ta ta hya ta bya ta tsa ta ya ta ka ta pya ta sha ta ba ta cha ta ta ta da ta za ta ga ta pa ta sa Audio-visual Presentation: sa ta za tsa Practice A KYU:YU: (AN OLD FRIEND) 1. Ototoi, 040401* 2. daigakujidaino 4AI5A5U7I4AI40* 3. do:kyu:seiga 40:57U:4EI5A# 4. Sendaikara 4E84AI5A3A* 5. tazunetekitekureta. 4A4U4E4E5I4E5U3E4A* 6. Tanakatoyu: namaeda. 4A4A5A407U:#4A1AE4A# 7. Nakanakano shu.saide 4A5A4A5A40#7U:4AI4E* 8. medatsu sonzaidatta. 1E4A4U#4084AI4A94A# 9. Tamatama shigotode 4A1A4A1A#7I50404E# 10. To:kyo:ni 40:570:71* 11. detekitatsuideni 4E4E5I4A4UI4E7I* 12. tachiyottanodatoyu:. 4A7I7094A404A407U:# 13. Totsuzendattanode, 404U4E84A94A404E* 14. isasaka awateteshimatta. I4A4A5A#A2A4E4E7I1A94A# 15. Semete hitokoto 4E1E4E#5I405040# 16. tegamika denwade 4E5A1I5A#4E82A4E# 17. jizenni 7I4E87I* 18. shirasetekureteitara 7I3A4E4E5U3E4EI4A3A* 19. motenashimodekitanoni 104E4A7I104E5I4A407I* 20. zannende naranakatta. 4A84E84E#4A3A4A5A94A* 21. Do.yara watashio 40:7A3A#2A4A7IO* 22. odorokasetakattarashi:. 040305A4E4A5A94A3A7I:* 23. Tanakano 4A4A5A40* 24. takuramukotowa 4A5U3A1U50402A* 25. daitai so:zo:gatsuku. 4AI4AI#40:40:5A4U5U* 26. Sorenishitemo 403E7I7I4E10* 27. hisabisano saikaida. 5I4A1I4A40#4AI5AI4A# 28. Nanto 4A840* 29. niju.sannenburininaru. 7I7U:4A84E81U3I7I4A3U# 30. Otagai toshio 04A5AI#407IO# 31. tottamonoda. 4094A10404A* 32. Ochitsuite mierunomo 07I4UI4E#1IE3U4010# 33. toizendaro:. 40:4E84A30:# 34. Tadashi, 4A4A7I# 35. futaritomo madamada 6U4A3I4010#1A4A1A4A# 36. hatarakizakari. 5A4A3A5I4A5A3I* 37. Toshiyoriatsukaidakewa 407I703IA4U5AI4A5E2A* 38. zettai saretakunai. 4E94AI#4A3E4A5U4AI# Practice B TANABATA UMARE (BIRTHDAY) 1. Hichigatsunanokani 2. umaretanowa 3. tashika anatano 4. oto:tosandeshitane. 5. Watashiga 6. tanjo:bio tadashiku 7. oboeteiruhitowa 8. hotondo inainodesuga, 9. gu:zen, 10. watashinoaneto onaji 11. seinengappidattanode 12. kiokushite irunodesu. 13. Hitogotonagara, 14. tanabatano 15. umaretowa 16. nakanaka 17. fuzeigaatte 18. omoimasenka? 19. Tadashi soremo, 20. maitoshi tsuyuno 21. massaichu :dewa 22. nantonaku 23. monotarinasao 24. kanjite shimaimasune. 25. Tanjo:bino 26. motsuimiwa 27. hitosorezore 28. kotonarudesho:shi, 29. iwaikatamo 30. samazamadesho:. 31. Demo, 32. darenitottemo 33. tokubetsunahide 34. arukotowa tashikadesu. 35. Otonaninaruto 36. hitotsu hitotsu 37. tas/z/o torutabini, 38. jinseitowa nanikato 39. kangaesaseraremasune. 40. Tokorode, 41. anatano seinengappio 42. oshieteitadakemasenka? 5I7I5A4U4A405A7I* U1A3E4A402A* 4A7I5A#A4A4A40# 040:404A84E7I4A4E# 2A4A7I5A* 4A870:1I0#4A4A7I5U# 010E4EI3U5I402A* 5040840#I4AI404E4U5A# 5U:4E8# 2A4A7I40A4E40#04A7I# 4EI4E85A91I4A94A404E* 5I05U7I4E#I3U404E4U# 5I4050404A5A3A# 4A4A1A4A40* U1A3E402A* 4A5A4A5A* 6U4EI5AA94E# I:40#010I1A4E85A# 4A4A7I#403E10# 1AI407I#4U7U40# 1A94AI7U:4E2A# 4A8404A5U* 10404A3I4A4AO* 5A87I4E#7I1AI1A4U4E# 4A870:1I40# 104UI1I2A* 5I40403E4O3E* 50404A3U4E70:7I# I2AI5A4A10* 4A1A4A1A4E70:# 4E10# 4A3E7I4094E10* 405U1E4U4A5I4E* A3U50402A#4A7I5A4E4TJ# 0404A7I4A3U40* 5I404U#5I404U# 407I0#403U4A1I7I# 7I84EI402A#4A7I5A40# 5A85AE4A4E3A3E1A4U4E* 4050304E* A4A4A40#4EI4E85A91IO# 07IE4EI4A4A5E1A4E85A* L E S S O N 10 Review Visual Only Presentation: aa pa ia wa ua ra oa ta ea Training Visual Only Presentation: ka pya ka ya ka sha ka ra ka wa ka na ka ta ka kya ka ha ka ma ka nya ka kya ka ga ka ja ka bya ka fa ka mya ka ka ka gya ka za ka ba ka cha ka pa ka rya ka da ka tsa ka sa Audio-visual Presentation: ka ga ha Practice A KAIGAIR YOKO: (TRAVEL) 1. ako hachinenkanni 5A50#5A7I4E85A87I# 2. kyu:kao gaikokude 57U:5AO#5AI505U4E# 3. sugosuhitoga 4U504U5I405A* 4. zo:kashimashita. 40:5A7I1A7I4A# 5. Seikakuna kazuwa 4EI5A5U4A#5A4U2A# 6. wakarimasenga, 2A5A3I1A4E85A* 7. kazokudedekakeru ke:suga 5A405U4E4E5A5E3U#5E:4U5A# 8. kekko: aruyoidesu. 5E950:#A3U70:4E4U# 9. Kaigairyoko.-wa, 5AI5AI37050:2A# 10. mukashiwa 1U5A7I2A* 11. kanari okanega 5A4A3I#05A4E5A# 12. kakarumonoto 5A5A3U104040* 13. kangaerareteimashitaga, 5A85AE3A3E4EI1A7I4A5A# 14. konogorowa 504050302A* 15. endakano okagede E84A5A40#05A5E4E# 16. kanarazushimo 5A4A3A4U7I10* 17. takakutsukanaku narimashita. 4A5A5U4U5A4A5U#4A3I1A7I4A# 18. Seikatsuga 4EI5A4U5A* 19. yutakaninari, 7U4A5A7I4A3I* 20. yokano tsukaikataga 705A40#4U5AI5A4A5A# 21. henkashitakotomo 5E85A7I4A504010* 22. kankeishiteimasu. 5A85EI7I4EI1A4U* 23. Gakuseiwa 5A5U4EI2A* 24. nagaiyasumiga 4A5AI7A4U1I5A* 25. arimasukara A3I1A4U5A3A* 26. arubaitode A3U1AI404E* 27. okaneotame 05A4E04A1E* 28. nakamato kigaruni 4A5A1A40#5I5A3U7I# 29. kakuchio 5A5U7IO* 30. kanko.surukotoga 5A850:4U3U50405A# 31. kano.desu. 5A40:4E4U# 32. Tokoroga, 4050305A* 33. shigotogaaruto 7I50405AA3U40* 34. nagaku yasumu 4A5A5U#7A4U1U# 35. wakenlwa ikimasen. 2A5E7I2A#I5I1A4E8# 36. Kekkyoku, 5E95705U* 37. gokatsuno renkyumadoni 505A4U40#3E857U:4A407I# 38. kagirareteshimaimasu. 5A5I3A3E4E7I1AI1A4U* 39. Amerikaga A1E3I5A5A* 40. ninkigatakaku, 7I85I5A4A5A5U* 41. nakademo nishikaiganno 4A5A4E10#7I7I5AI5A8 40# 42. kariforuniaga 5A3I603U7IA5A* 43. kanko.no mekkadesu. 5A850:40#1E95A4E4U# 44. Sonohokani, 45. chikaku.no kankokuemo 46. kanshinga 47. mukerareteimasu. 114 4040505A7I* 7I5A5U40#5A8505UE10# 5A87I85A# 1U5E3A3E4EI1A4U* Practice B GOMINO SHORI (GARBAGE DISPOSAL) 1. Higoro 515030* 2. nanigenaku suteteiru 4A7I5E4A5U#4U4E4EI3U# 3. gominitsuite 501I7I4UI4E* 4. kanshin-o mottakotoga 5A87I8#0#1094A50405A# 5. koremadeni arimasuka? 503E1A4E7I#A3I1A4U5A# 6. Dokode 40504E* 7. ikanaru keiroohete I5A4A3U#5EI3005E4E# 8. shoriga 703I5A* 9. okonawarete irunodesho:ka. 0504A2A3E4E#I3U404E70:5A# 10. Gominiwa 501I7I2A* 11. daidokorokaraderu 4AI4050305A3A4E3U* 12. namagominohokani 4A1A501I40505A7I* 13. reizo.koo hajimetosuru 3EI40:500#5A7I1E404U3U# 14. o:gata, kogatano 0:5A4A#505A4A40# 15. denkaseihin, 4E85A4EI5I8* 16. kagunadoga fukumaremasu. 5A5U4A405A#6U5U1A3E1A4U# 17. Kakuchide 5A5U7I4E* 18. kaishu.sareru gomino 5AI7U:4A3E3U#501I40# 19. hobo gojuppa:sentoga 5010#507U91A:4E8405A* 20. kanengomide, 5A4E8501I4E* 21. korerawa 503E3A2A* 22. kindaikasareta seiso:ko:jo:ni 5I84AI5A4A3E4A#4EI40:50:70:7I# 23. hakobikomaremasu. 5A501I501A3E1A4U* 24. Gominonakano tetsubunwa 501I404A5A40#4E4U1U82A# 25. jishakuo tsukatte 7I7A5UO#4U5A94E# 26. kaishusare, 5AI7U:4A3E# 27. shigenno sairiyo:kaga 7I5E840#4AI3I70:5A5A# 28. hakarareteimasu. 5A5A3A3E4EI1A4U* 29. Chikagorowa 7I5A50302A* 30. shinkaihatsusareta 7I85AI5A4U4A3E4A# 31. sozaiga ichihayaku 404AI5A#I7I5A7A5U# 32. kateiseikatsuni hairikomi, 5A4EI4EI5A4U7I#5AI3I501I# 33. atarashi: ko:gaito A4A3A7I:#50:5AI40# 34. narunodewanaikati 4A3U404E2A4AI5A40* 35. kenensareteimasu. 5E4E84A3E4EI1A4U* 36. Konoyo:ni, 504070:71* 37. gominikanshitemo, 501I7I5A87I4E10* 38. kaiketsushinakerebanaranai 5AI5E4U7I4A5E3E1A4A3A4AI* 39. kadaiga 5A4AI5A* 40. kanari arunodesu. 5A4A3I#A3U404E4U# LESSON 11 Review Visual Only Presentation: aa pa ia wa ua ra oa ta ea ka Training Visual Only Presentation: There is no visual only presentation for this lesson. Audio-visual Presentation: [ wa ][ wi ] [ we ][ wo ] [ <Da ][ Oi ] [ Oe ][ <Do ][ <£ui] Practice F URANKU-SAN (FRANK) 1. Furankusanwa 2. Furoridaumare. 3. Sofutoweano kaishani 4. tsutometeiru. 5. Itsumo po.kafeisude 6. go:ingumaiwei. 1. Ha:dowa:ka:de, 8. faitono katamari. 9. Tsuneni pafekutoo mezasu. 10. Wanmandakeredomo 11. feminisutodemoaru. 12. Kaiwawa wittonitomi, 13. fo:marunabademo 14. infoimarunabademo ninkimono. 15. Maiasa toreminguweaokite 16. wo:kuman-o kikinagara 17. joginguosuru. 18. Supo:tsude 19. we:tokontoro:ru. 20. Asagohanwa kafeoreto 21. kurowassan. 22. Wiikuendowa 23. ba:dowocchinguo tanoshimu. 24. Sukinaosakewa 25. wiski:to wokka. 26. Omizuwa kanarazu 21. mineraruwo:ta:. 6U3A85U4A82A* 6U303I4AU1A3E* 406U406EA40#5AI7A7I# 4U401E4EI3U* I4U10#10:5A:6EI4U4E* 50:I85U1AI6EI# 5A:402A:5A:4E# 6AI4040#5A4A1A3I# 4U4E7I#1A:6E5U400#1E4A4U# 2A81A84A5E3E4010* 6E1I7I4U404E10A3U# 5AI2A2A#6I9407I401I# 60:1A3U4A1A4E10# 18 60:1A3U4A1A4E10#7I85I1040# 1AIA4A#403E:7I85U6EA05I4E* 60:5U1A8#0#5I5I4A5A3A# 705I85U04U3U* 4U10:4U4E# 6E:405084030:3U# A4A505A82A#5A6E03E40# 5U302A94A8# 6I:5UE8402A# 1A:406097I85U0#4A407I1U# 4U5I4A04A5E2A* 6I45I:40#6095A# 01I4U2A#5A4A3A4U# 1I4E3A3U60:4A:# L E S S O N 12 Review Visual Only Presentation: aa ia ua oa ea pa wa ra ta ka Training Visual Only Presentation: NOTE: / this lesson is [ f ] not [ 0 ] fa cha fa wa fa ya fa ra fa na fa pya fa ta fa kya fa ma fa nya fa ba fa hya fa ga fa ja fa by a fa fa fa mya fa ka fa gya fa za fa sha fa sa fa tsa fa rya fa da fa pa fa cha Audio-visual Presentation: fa 119 Practice KANA TANAKOTOBANO HANRAN (A FLOOD OF LOANWORDS) 1. Gendaino fasshonaburuna 5E84AI40#6A9704A1U3U4A# 2. raifusutairutowa 3AI6U4U4AI3U402A* 3. katakanakotobao 5A4A5A4A50401AO* 4. fi:ringude tsukaukotorashi:. 61:3I85U4E#4U5AU50403A7I:# 5. Ofisudeno shigotowa 06I4U4E40#7I50402A# 6. wa.puroto fakkusu. 2A:1U3040#6A95U4U# 7. Afutafaibuwa A6U4A:6AI1U2A* 8. fittonesukurdbu. 6I9404E4U5U3A1U* 9. Dinarwa furansuryo:rino 4I4A:2A#6U3A84U370:3I40# 10. furuko:su. 6U3U50:4U# 11. Furu.too kikinagara 6U3U:400#5I5I4A5A3A# 12. naifuto fo.kude 4AI6U40#60:5U4E# 13. foagurao shishoku. 60A5U3AO#7I705U# 14. Shu:matsuwa gorufuya 7U:1A4U2A#503U6U7A# 15. safinde 4A:6I84E# 16. rifuresshu. 3I6U3E97U* 17. Amerikanfuttobo:ruwa terebikansen. A1E3I5A86U94010:3U2A#4E3E1I5A84E8# 18. Yakyumo fannara 7A57U:40#6A84A3A# 19. fainpure:ni 6AI81U3E-.7I* 20. fi:ba:suru. 6I:1A:4U3U# 21. Machidewa furawa.sho:, 1A7I4E2A#6U3A2A:70:# 22. fotokontesuto, 60405084E4U40* 23. furu:tsufea, 6U3U:4U6EA# 24. firumufesutibaruga moyo:sare, 6I3U1U6E4U4I1A3U5A#1070:4A3E# 25. kokusaifo:ramumo hirakareru. 505U4AI60:3A1U10#5I3A5A3E3U# 26. Hatashite korewa 5A4A7I4E#503E2A# 27. kokusaikananodaro:ka. 505U4AI5A4A404A30:5A# LESSON 13 Review Visual Only Presentation: aa pa ia wa ua ra oa ta ea ka fa Training Visual Only Presentation: ya na ya ya ya sha ya ra ya wa ya pya ya ta ya kya ya ma ya nya ya fa ya kya ya ga ya ja ya bya ya ha ya mya ya ka ya gya ya za ya ba ya cha ya sa ya rya ya da ya pa ya tsa Audio-visual Presentation: ya pya sha cha hya rya Practice TONARINO OYAKO (MY NEIGHBOR) 1. Kyabetsuno 57A1E4U40* 2. yasaisaradao tsukuri, 7A4AI4A3A4A0#4U5U3I# 3. gyo:zao 570:4AO# 4. yakinagara kangaeta. 7A5I4A5A3A#5A85AE4A# 5. Otonarino oyakowa 0404A3I40#07A502A# 6. yasuminiwa 7A4U1I7I2A* 7. go:kakyakusenno 50:5A57A5U4E840# 8. tabio surutoyu:. 4A1I0#4U3U407U:# 9. Urayamashi: hanashida. U3A7A1A7I:#5A4A7I4A# 10. Kyanpuni ikunoga 57A81U7I#I5U405A# 11. sekinoyamano 4E5I407A1A40* 12. wagayatowa 2A5A7A402A* 13. o:kina gyappuda. 0:5I4A#57A91U4A# 14. Kyanserumachio shite 57A84E3U1A7I0#7I4E# 15. yatto jitsugenshitaso:da. 7A940#7I4U5E87I4A40:4A# 16. Yahari nenpaino 7A5A3I#4E81AI40# 17. okyakubakari nanodaw.ka. 057A5U1A5A3I#4A404A30:5A# 18. Yuttarishita kyabinde 7U94A3I7I4A#57A1I84E# 19. yasumunowa 7A4U1U402A* 20. sazoya kaitekidaro:. 4A407A#5AI4E5I4A30:# 21. Oyaomoino 07A010I40* 22. yasashi: musukosanda. 7A4A7I:#1U4U504A84A# 23. Gyakuni 57A5U7I# 24. yasashisugiru tomoieru. 7A4A7I4U5I3U#4010IE3U# 25. Wagayaniwa 2A5A7A7I2A* 26. kyabiao omiyageni 57A1IA0#01I7A5E7I# 27. kattekitekureru 5A94E5I4E5U3E3U* 28. yakusokuo 7A5U405UO* 29. chanto shitekureta. 7A840#7I4E5U3E4A# 30. Shashinmo 7A7I810* 31. hyakumaikurai 7A5U1AI5U3AI* 32. tottekuruyamo shirenu. 4094E5U3U7A10#7I3E4U# 33. Ichinichimo hayaku I7I7I7I10#5A7A5U# 34. shuppatsushitaini chigainai. 7U91A4U7I4AI7I#7I5AI4AI# LESSON 14 Review Visual Only Presentation: aa pa ia wa ua ra oa ta ea ya fa ka Training Visual Only Presentation: anna anna anna a:na anna ana anpa apa anpa anpa anpa a:pa anta ata anta a:ta anta anta anka aka anka anka anka a:ka an-a a: an-a ana an-a an-a Audio-visual Presentation: anpa anwa anra anta anka anfa anya an-a 123 Practice MOTTO ARUKO: (LET'S WALK) 1. Bunmeino hattatsuni tomonai, 1U81EI40#5A94A4U7I#40104AI# 2. jinruino undo:ryorwa 7I83UI40#U840:370:2A# 3. kyokutanni gensho:shita. 5705U4A87I#5E870:7I4A# 4. Undotbusokuwa U840:1U405U2A# 5. himanya seijinbyo.ni 5I1A47A#4EI7I8170:7I# 6. tsunagaru shinpaimo aru. 4U4A5A3U#7I81AI10#A3U# 7. Ippanni I91A87I* 8. gendaijinwa 5E84AI7I82A* 9. kenko.o ishikishite. 5E850:0#I7I5I7I4E# 10. jibunkara susunde 7I1U85A3A#4U4U84E# 11. undo:ni hagendeiru. U840:7I#5A5E84EI3U# 12. Sarari:manno nakaniwa 4A3A3I:1A840#4A5A7I2A# 13. suimingukurabuya 4UI1I85U5U3A1U7A* 14. tore :ningukurabuno 403E:7I85U5U3A1U40* 15. kaiinninaru 5AII87I4A3U* 16. hitomo ireba 5I4010#I3E1A# 17. man-indenshao sake 1A8#I84E87A0#4A5E# 18. jitenshade 7I4E87A4E* 19. tsu:kinsuru hitomo iru. 4U:5I84U3U#5I4010#I3U# 20. Mochiron, kojinno 107I308#507I840# 21. seikatsukankyo.ya nenreini 4EI5A4U5A8570:7A#4E83EI7I# 22. awaseta undo:ho:ho:no A2A4E4A#U840:50:50:40# 23. sentakuga kanjinda. 4E84A5U5A#5A87I84A# 24. Saikinwa 4AI5I82A* 25. wo:kingu toyu:mononi 60:5I85U#407U:10407I# 26. ninkiga atsumatteiru. 7I85I5A#A4U1A94EI3U# 27. Arukukotowa A3U5U50402A* 28. ningenno kenko:no 7I85E840#5E850:40# 29. kihonda 5I5084A* 30. Donna hitodemo 4084A#5I404E10# 31. donna tokorodemo 4084A#4050304E10# 32. kantanni dekirushi, 5A84A87I#4E5I3U7I# 33. junbimo 7U81I10* 34. hotondo iranai. 5040840#I3A4AI# 35. Ashini A7I7I# 36. futanno kakaranai 6U4A840#5A5A3A4AI# 37. kutsuga kichinto 5U4U5A#5I7I840# 38. erandeareba E3A84EA3E1A* 39. ju:bunda. 7U:1U84A# 40. Juppundemo 7U91U84E10* 41. ju:gofundemo 7U:506U84E10# 42. jikan-o mitsukete 7I5A8#0#1I4U5E4E# 43. shu:kanni shiyo:. 7U:5A87I#7I70:# 124 LESSON 15 Review Visual Only Presentation: anpa anwa anra anta anfa anka an-a Training Visual Only Presentation: appa apa appa appa appa a:pa atta ata atta a:ta atta atta akka aka akka akka Audio-visual Presentation: appa assa assha atta adda accha akka agga 125 Practice MACHINO HENKA (CHANGE OF MY HOMETOWN) 1. Hisashiburini kaetta kokyo.-wa 5I4A7I1U3I7I#5AE94A#50570:2A# 2. sukkari kawatteshimatta. . 4U95A3I#5A2A94E7I1A94A# 3. Ekini oritatte E5I7I#03I4A94E# 4. bikkurishiteshimatta. 1I95U3I7I4E7I1A94A* 5. Imamadetowa I1A1A4E402A* 6. mattaku chigatta 1A94A5U#7I5A94A# 7. fu:keiga hirogatteiru. 6U:5EI5A#5I305A94EI3U# 8. Isshun, I97U8# 9. machigatte resshao 1A7I5A94E#3E97A0# 10. oriteshimattanodewanaikato 03I4E7I1A94A404E2A4AI5A40* 11. sakkakushitahododa. 4A95A5U7I4A50404A* 12. CTwtto Urashimataw.ni 70940#U3A7I1A4A30:7I# 13. nattayo:na kibundatta. 4A94A70: 4A#5I1U84A94A# 14. Machiwa shoppinguo suruhitode 1A7I2A#7091I85U0#4U3U5I404E# 15. gottagaeshiteiru. 5094A5AE7I4EI3U* 16. /tta/ dokokara I94AI#40505A3A# 17. konnaniippai hitoga 5084A7II91AI#5I405A# 18. atsumatte kurunodaw.ka. A4U1A94E#5U3U404A30: 5A# 19. Maeniwa tatta mittsushika 1AE7I2A#4A94A#1I94U7I5A# 20. nakatta birumo 4A5A94A#1I3U10# 21. imawa acchikocchini I1A2A#A97I5097I7I# 22. bisshirito 1I97I3I40* 23. tatteiru. 4A94EI3U* 24. Omottemominakatta henkani 01094E101I4A5A94A#5E85A7I# 25. hakkiriitte 5A95I3II94E* 26. gakkarida. 5A95A3I4A* 27. Mukashinomamao 1U5A7I401A1AO* 28. tamotteite 4A1094EI4E* 29. hoshikattanoni. 507I5A94A4O7I* 30. Mottomo, 1094010* 31. subetega zutto 4U1E4E5A#4U940# 32. onajijo:taide 04A7I70:4AI4E# 33. nokotteiruyoidewa 405094EI3U70: 4E2A# 34. issai hattenga I94AI#5A94E85A# 35. naikotoninatte, 4AI50407I4A94E* 36. kaette 5AE94E* 37. komatteshimaunodaro:. 501A94E7I1AU404A30: # 38. Ningendatte isshoda. 7I85E84A94E#I9704A# 39. Henkaga attekoso, 5E85A5A#A94E5040# 40. rippana seicho:nanoda. 3I91A4A#4EI70: 4A404A* 126 LESSON 16 Review Visual Only Presentation: anpa anwa appa ana akka anra anta anfa anka an-a Training Visual Only Presentation: a:i a:i a:i ai a:i ai a:i a:i a:i a:i a:i ai a:i a:i a:i ai Audio-visual Presentation: a: i: u: e: o: 127 Practice NIHONNO KYO.IKUSE.DO (EDUCATIONAL SYSTEM) 1. Nihondewa 7I5084E2A* 2. sho:gakko: rokunen, 70:5A950:#305U4E8# 3. chuigakko: sannen, 7U:5A950:#4A84E8# 4. go:ke: kyu:nenga 50:5E:#57U:4E85A# 5. gimukyo:ikudesu. 5I1U570:I5U4E4U* 6. Kono ho:ritsuwa 5040#50:3I4U2A# 7. sengoni se:te:saremashita. 4E8507I#4E:4E:4A3E1A7I4A* 8. Jugyo:ryorwa 7U570:370:2A# 9. to:zen muryo:desu. 40:4E8#1U370:4E4U# 10. Ko.ritsuno gakkomi 50:3I4U40#5A950:7I# 11. nyu:gakusurunara 47U:5A5U4U3U4A3A* 12. nyu:shino shinpaiosuru 47U:7I40#7I81AI04U3U# 13. hitsuyoiwa arimasen. 5I4U70:2A#A3I1A4E8# 14. Gaikokugowa e:goo 5AI505U502A#E:500# 15. benkyo:shimasu. 1E8570:7I1A4U# 16. Korewa kyo:tsu:no 503E2A#570:4U:40# 17. hissyu:kamokudesu. 5I947U:5A105U4E4U# 18. Watashiwa e.gono hatsuonga 2A4A7I2A#E:5040#5A4U085A# 19. jo:zunidekizu, 70:4U7I4E5I4U# 20. kuw.shita ke:kenga arimasu. 5U30:7I4A#5E:5E85A#A3I1A4U# 21. Demo su:gakuto 4E10#4U:5A5U40# 22. se:butsuwa 4E:1U4U2A# 23. ma:ma:no se:sekideshita. 1A:1A:40#4E:4E5I4E7I4A# 24. Ko:ko:shingakuwa 50:50:7I85A5U2A# 25. kojinno jiyu:desu. 507I840#7I7U:4E4U# 26. Daigakushingakuo 4AI5A5U7I85A5UO* 27. kibo:suru se:tomo 5I10:4U3U#4E:4010# 28. o:ze:imasu. 0:4E:I1A4U# 29. Sukoshidemo yu:me:na 4U507I4E10#7U:1E:4A# 30. daigakuni hairo.to 4AI5A5U7I#5AI30:40# 31. jukense :dakedenaku 7U5E84E:4A5E4E4A5U* 32. ryo:shin-ya sense :mo 370:7I8#7A#4E84E:10# 33. issho:kenme:desu. I970:5E81E:4E4U# 34. Kate:kyo:shi, juku, 5A4E:570:7I#7U5U# 35. yobikorwa daihanjo:. 701I50:2A#4AI5A870:# 36. Konoyo:na jo:kyo:ga 504070:4A#70:570:5A# 37. itsumade tsuzukunodesho:ka. I4U1A4E#4U4U5U404E70:5A# 38. Se:fuya monbushorwa 4E:6U7A#1081U70:2A# 39. jukensenso:no kaikakuto yu:keredo 7U5E84E840:40#5AI5A5U40#7U:5E3E40* 40. se:kaga agaruniwa 4E:5A5A#A5A3U7I2A# 41. so:to: nensu:ga 40:40:#4E84U:5A# 42. kakarukotodesho:. 5A5A3U50404E70:# 128 L E S S O N 17 General WATASHINO SONKEISURU HITO (A PERSON I RESPECT) 1. Anatawa A4A4A2A* 2. rekishijo:no 3E5I7I70:40# 3. jinbutsuno nakade, 7I81U4U40#4A5A4E# 4. dareo ichiban, 4A3E0#I7I1A8# 5. sonkeishimasukato kikaretara, 4085EI7I1A4U5A40#5I5A3E4A3A# 6. nanto kotaemasuka. 4A840#504AE1A4U5A# 7. Tokuteino 405U4EI40* 8. hitono namaega, 5I4040#4A1AE5A# 9. satto detekurudesho.ka. 4A940#4E4E5U3U4E70:5A# 10. Watashiwa, 2A4A7I2A* 11. konoshu.no shitsumonwa 50407U40#7I4U1082A# 12. hijo:ni kiraide, 5I70:7I#5I3AI4E# 13. itsumo, sonoba, sonobano I4U10#40401A#40401A40# 14. omoitsukide 010I4U5I4E# 15. kaitoishiteimasu. 5AI40:7I4EI1A4U# 16. Kekkyoku, 5E95705U* 17. mottomo sonkeisuru ningenga 1094010*4085EI4U3U#7I8 5E85A# 18. nanninmo to:jo:surukotoni 4A87I810#40:70:4U3U50407I* 19. natteshimaundesune. 4A94E7I1AU84E4U4E* 20. Tatoeba, aruhi, 4A40E1A#A3U5I# 21. tokugawaieyasu toittakatoomoeba, 405U5A2AIE7A4U#40I94A5A40010E1A# 22. yokujitsuwa 705U7I4U2A* 23. fukuzawayukichi toittafu:ni 6U5U4A2A7U5I7I#40I94A6U:7I# 24. mattaku betsuno jinmeio 1A94A5U#1E4U40#7I81EI0# 25. ageteshimauwakedesu. A5E4E7I1AU2A5E4E4U* 26. Moshimo, 107110* 27. onaji aitekara, 04A7I#AI4E5A3A# 28. do.itsuno shitsumon-o 40:I4U40#7I4U108#0# 29. nikai saredemoshitara, 7I5AI#4A3E4E107I4A3A# 30. taihennakotoni narimasuyo. 4AI5E84A50407I#4A3I1A4U70# 31. Nazenara, 4A4E4A3A* 32. jibunno 7I1U840* 33. saishono kotaeo 4AI7040#504AEO# 34. oboete 010E4E* 35. irarenainodesukara. I3A3E4AI404E4U5A3A* L E S S O N 1 8 General WATASHINO SUKINA TABEMONO (MY FAVOURITE FOOD) 1. Anatawa A4A4A2A# 2. sekaiju:no 4E5AI7U:40# 3. tabemonono nakade, 4A1E104040#4A5A4E# 4. naniga ichiban, 4A7I5A#I7I1A8# 5. sukidesukato kikaretara, 4U5I4E4U5A40#5I5A3E4A3A# 6. satto kaito.dekimasuka. 4A940#5AI40:4E5I1A4U5A* 7. Watashiwa, 2A4A7I2A* 8. konoyoina shitsumonwa 504070:4A#7I4U1082A# 9. hijo:ni nigatede, 5I70:7I#7I5A4E4E# 10. itsumo shibaraku I4U10#7I1A3A5U# 11. kangaenaito 5A85AE4AI40* 12. kotaega detekimasen. 504AE5A#4E4E5I1A4E8# 13. Tokubetsuna 405U1E4U4A* 14. monono namaega, 104040#4A1AE5A# 15. suguni omoitsukanainodesuyo. 4U5U7I#010I4U5A4AI404E4U70# 16. Kekkyoku, 5E95705U# 17. korezotoieru tabemonoga 503E4040IE3U#4A1E10405A# 18. nanihitotsu sonzaishinaikotoni 4A7I5I404U#4084AI7I4AI504071* 19. natteshimaundesune. 4A94E7I1AU84E4U4E* 20. Tadashi, 4A4A7I* 21. onaji imino shitsumonmo, 04A7I#I1I40#7I4U10810# 22. sentakuhan-io kagittara 4E84A5U5A8#IO#5A5I94A3A# 23. tasho:wa rakuni 4A70:2A#3A5U7I# 24. kotaerarerukamo shiremasen. 504AE3A3E3U5A10#7I3E1A4E8# 25. Tatoeba, kudamononara 4A4OElA#5U4A104O4A3A# 26. ichigo toyu:kotoga dekirushi, I7I50#407U:50405A#4E5I3U7I# 27. kunibetsudewa, 5U7I1E4U4E2A* 28. Mekishikoryo:ri toittaguaini 1E5I7I50370:3I#40I94A5UAI7I# 29. sukoshiwa hayai hanno:mo 4U507I2A#5A7AI#5A840:10# 30. fukano:dewa nainodesu. 6U5A40:4E2A#4AI404E4U# 31. Konoteidodeareba 50404EI404EA3E1A* 32. jibunmo 7I1U810* 33. hitotsuya futatsu 5I404U7A#6U4A4U# 34. nanika agerukotoga 4A7I5A#A5E3U50405A# 35. dekirudesho:. 4E5I3U4E70:# APPENDIX D 130 Lesson Script and Literal Translation of Lesson 17 and Lesson 18 Lesson 17 General WATASHINO SONKEISURU HITO Anata wa rekishijo: no jinbutsu no naka de, dare o ichiban sonkeishimasu ka to kikare tara, nan to kotaemasu ka. Tokutei no hito no namae ga, satto detekurudesho:ka. Watashi wa, kono shu no shitsumon wa hijo:ni kiraide, itsumo, sonoba, sonoba no omoitsuki de kaito:shiteimasu. Kekkyoku, mottomo sonkeisuru ningen ga nannin mo to:jo:suru koto ni natte shimaundesune. Tatoeba, aruhi, tokugawaieyasu to ittakatoomoeba, yokujitsu wa fukuzawayukichi to itta fu.ni mattaku betsuno jinmei o ageteshimau wakedesu. Moshimo, onaji aitekara, do:itsu no shitsumon-o nikai saredemoshitara, taihenna koto ninarimasuyo. Nazenara, jibun no saisho no kotae o oboeteirare nainodesu kara. A PERSON I RESPECT WHAT WOULD YOU SAY, WHEN YOU ARE ASKED WHO IS THE PERSON IN HISTORY YOU RESPECT MOST? Do YOU COME UP wrra A PARTICULAR NAME IMMEDIATELY? SINCE I HATE THIS SORT OF QUESTION, I ALWAYS NAME ONE PERSON RANDOMLY EVERY TIME I AM ASKED. ULTIMATELY, IT MAY MEAN THAT THERE EXISTS NO SINGLE PERSON I REALLY RESPECT. FOR EXAMPLE, ONE TIME, MY ANSWER WELL BE "ToKUGAWA IEYASU", AND THE NEXT DAY, IT CAN BE " F u K U Z A W A Y U K I C H l " , WHO ARE TOTALLY DIFFERENT CHARACTERS. IF THE SAME INTERVIEWER ASKS ME THE SAME QUESTION TWICE, I WILL BE IN A BIG TROUBLE, BECAUSE I CAN NEVER REMEMBER MY FIRST ANSWER. 131 Lesson 18 General WATASHINO SUKINA TABEMONO Anata wa sekaiju: no tabemono no naka de, nani ga ichiban sukidesu ka to kikaretara, satto kaito: dekimasu ka. Watashi wa, kono yo:na shitsumon wa hijo:ni nigatede, itsumo shibaraku kangaenaito kotae ga dete kimasen. Tokubetsuna mono no namae ga, suguni omoi tsukanai nodesu yo. Kekkyoku, korezo to ieru tabemonoga nanihitotsu sonzaishinai koto ni natte shimaundesune. Tadashi, onaji imino shitsumon mo, sentakuhan-io kagittara, tasho: wa rakuni kotaerarerukamo shiremasen. Tatoeba, kudamono nara ichigo to yu: koto ga dekirushi, kunibetsu de wa, mekishiko ryo:ri to itta guai ni sukoshiwa hayai hanno: mo fukano: de wa nainodesu. Kono teido de areba, jibunmo hitotsuyafutstsu nanika ageru koto ga dekirudesho:. MY FAVOURITE FOOD WHEN YOU ARE ASKED WHAT YOUR MOST FAVOURITE FOOD IS, CAN YOU ANSWER QUICKLY? SINCE I AM NOT GOOD AT RESPONDING TO THIS KIND OF QUESTION, I ALWAYS HAVE TO THINK A WHILE. No PARTICULAR NAME OF FOOD COMES UP IMMEDIATELY. U L T I M A T E L Y , IT M A Y MEAN THAT THERE EXISTS NOTHING T H A T I R E A L L Y L O V E T O E A T . H O W E V E R , THE SITUATION CAN BE A LITTLE BIT DIFFERENT, IF THE INQUIRY IS MORE SPECIFIC IN TERMS OF CATEGORIES. F O R E X A M P L E , IT IS NOT IMPOSSIBLE FOR ME T O SAY THAT I LIKE STRAWBERRIES BEST AMONG FRUIT, OR THAT I PREFER ITALIAN CUISINE TO ANY OTHER ETHNIC CUISINE. W l T H THESE QUESTIONS, I C A N PROBABLY NAME AT LEAST ONE OR TWO ITEMS. 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items