Language teaching material creation program, language teaching material creation device, and language teaching material creation method

Abstract

One representative embodiment of the present invention is a language teaching material creation program which is executed by a computer capable of communicating with an automatic conversation engine using a large language model, and which involves sequentially inputting first, second, and third prompts to the automatic conversation engine. The first prompt sets, within the automatic conversation engine, interviewer specifications corresponding to a student. The second prompt causes the automatic conversation engine to have the student input a simple personal profile, and to ask the student questions related to the profile and collect answers. The third prompt causes the automatic conversation engine to output a language teaching material customized for the student by embellishing the self-introduction on the basis of the related questions and the answers.

Description

Language teaching material production program, language teaching material production device, and language teaching material production method

[0001] The present invention can be suitably used in a language teaching material production program, a language teaching material production device, and a language teaching material production method for producing language teaching materials suitable for students.

[0002] The importance of teaching materials in language education is undeniable, and several inventions have been proposed, such as systems that can customize language teaching materials for each student.

[0003] Patent Document 1 discloses a language learning material creation system that easily creates learning materials for passive listening, which is said to be able to flexibly change and customize the learning content according to the learner's preferences, level, etc. Using input short sentences and a material database containing dictionary data, the system creates learning materials that play back long sentences made by combining each short sentence and audio data for each short sentence on the user's terminal based on set passive listening conditions.

[0004] Patent Document 2 discloses a language learning material customization system that allows users to freely customize language learning material content. Because users can freely customize language learning material content data by selecting sentences and devising sequence patterns, it is said that language learning material content data that is more suitable for the user and has a higher language learning effect can be provided.

[0005] Patent Literature 3 discloses a registration program, a registration method, and an information processing device that are said to be able to improve the quality of question and answer combinations generated. This program inputs a search query and a sentence included in the corresponding search results into a machine learning model, obtains a question sentence based on the search query and the sentence included in the search results from the machine learning model, and registers the combination of the question sentence and the sentence included in the search results as a question and answer combination in a storage device.

[0006] Patent Literature 4 discloses a method of romanizing geminated consonants in Japanese language learning materials, Japanese language learning materials using this method, and a Japanese language learning device that are said to enable efficient learning of geminated consonants. By using a romanization method for geminated consonants in Japanese language learning materials that uses a double vowel character notation in which a romanization character representing the vowel of the sound immediately preceding the geminated consonant is followed immediately by a romanization character that is the same as the vowel, it is said that it becomes possible to read and learn an pronunciation that is close to the actual geminated consonant from the romanization displayed by the Japanese language learning materials and the Japanese language learning device.

[0007] JP 2018-116190 A JP 2013-125056 A JP 2023-174053 A JP 2022-044354 A

[0008] The inventions described in Patent Documents 1 and 2 make it possible to create language learning materials suited to students. However, they only allow content data, such as pre-prepared materials, to be customized to suit each student. Therefore, in order to accommodate all kinds of students, the amount of content data that needs to be prepared in advance becomes enormous, making it unrealistic. By combining the invention described in Patent Document 3, it may be possible to automatically generate content data, such as prepared materials, to some extent, but creating a machine learning model for this would require extensive learning.

[0009] Furthermore, even if it were possible to prepare a huge amount of content data, the selection of materials from that content data and how to organize them would be left up to the students themselves or users such as teachers who know the students, which would require manual work each time.

[0010] An object of the present invention is to provide a computer program for creating language learning materials that are customized not only for each student's language level but also for their content.

[0011] The means for solving these problems will be described below, but other problems and novel features will become apparent from the description of this specification and the accompanying drawings.

[0012] A representative embodiment of the present invention is a language teaching material creation program that is executed by a computer capable of communicating with an automatic conversation engine that utilizes a large-scale language model, and that sequentially inputs first, second, and third prompts to the automatic conversation engine.

[0013] The first prompt causes the automated conversation engine to set the interviewer specifications for the student, the second prompt causes the automated conversation engine to input a brief profile of the student and ask the student relevant questions about the profile to collect answers, and the third prompt causes the automated conversation engine to output language learning materials customized for the student by adapting the student's self-introduction based on the relevant questions and answers.

[0014] A second representative embodiment of the present invention is a language teaching material production device in which a program for sequentially inputting first, second, and third prompts to an automatic conversation engine that utilizes a large-scale language model is installed in a computer that can communicate with the automatic conversation engine.

[0015] The first prompt causes the automated conversation engine to set the interviewer specifications for the student, the second prompt causes the automated conversation engine to input a brief profile of the student and ask the student relevant questions about the profile to collect answers, and the third prompt causes the automated conversation engine to output language learning materials customized for the student by adapting the student's self-introduction based on the relevant questions and answers.

[0016] A third exemplary embodiment of the present invention is a method for creating language teaching materials, in which a computer capable of communicating with an automated conversation engine that utilizes a large-scale language model sequentially inputs first, second, and third prompts to the automated conversation engine.

[0017] The first prompt causes the automated conversation engine to set the interviewer specifications for the student, the second prompt causes the automated conversation engine to input a brief profile of the student and ask the student relevant questions about the profile to collect answers, and the third prompt causes the automated conversation engine to output language learning materials customized for the student by adapting the student's self-introduction based on the relevant questions and answers.

[0018] Here, the division between the first, second and third prompts is not important, as long as the above contents are input in sequence by the prompts.

[0019] Furthermore, the student may enter his / her profile and answer questions himself / herself, or an operator may do so on his / her behalf.

[0020] The effects obtained by the present invention can be briefly explained as follows.

[0021] That is, it is possible to provide a language teaching material production program, a language teaching material production device, and a language teaching material production method for producing language teaching materials that are customized not only for each student's language level but also for the content.

[0022] FIG. 1 is an explanatory diagram showing a typical configuration example of a language teaching material production device and program of the present invention. FIG. 2 is an explanatory diagram showing a typical configuration example of an entire system including a computer on which a language teaching material production program of the present invention is installed and an automatic conversation engine that utilizes a large-scale language model used by the computer. FIG. 3 is an explanatory diagram showing a typical operation example of a computer on which a language teaching material production program of the present invention is installed. FIG. 4 is an example of first to third prompts in a second embodiment. FIG. 5 is an example of a related question and its answer (first half) in the second embodiment. FIG. 6 is an example of a related question and its answer (second half) in the second embodiment. FIG. 7 is an example of a generated script in the second embodiment. FIG. 8 is an example of first to third prompts in a third embodiment. FIG. 9 is an example of output in the third embodiment. FIG. 10 is an example of a fourth prompt in the fourth embodiment. FIG. 11 is an example of output in the fourth embodiment.

[0023] 1. Overview of the Embodiments First, an overview of the representative embodiments disclosed in the present application will be described. Reference numerals in parentheses in the drawings used in the overview of the representative embodiments merely illustrate components included in the concept of the components to which they are attached.

[0024] [1] Prompts for a generation AI that creates language learning materials from a student's own profile. A representative embodiment disclosed in this application is a language learning material creation program (20) that is executed by a computer (50) that can communicate with an automatic conversation engine (80) that uses a large-scale language model, and that sequentially inputs first, second, and third prompts (1, 2, 3) to the automatic conversation engine.

[0025] The first prompt (1) is a prompt that causes the automated conversation engine to set the specifications of an interviewer for the student. The second prompt (2) is a prompt that causes the automated conversation engine to input a brief profile of the student and ask the student relevant questions about the profile to collect answers. The third prompt (3) is a prompt that causes the automated conversation engine to output language learning materials customized for the student by adapting the profile based on the relevant questions and the answers.

[0026] This makes it possible to provide a computer program that produces language learning materials that are customized not only for each student's language level but also for the content.

[0027] [2] Dramatically Dramatizing Self-Introduction In the language teaching material production program of [1], the profile is a brief self-introduction of the student. The first prompt or the second prompt includes, as the interviewer's specifications, asking the student questions about the highlights and lowlights of their life and their dreams for the future. The third prompt includes instructing the automated conversation engine to dramatize the self-introduction by keeping the self-introduction on a theme consistent with the self-introduction and using answers to the questions about the highlights and lowlights of their life and their dreams for the future.

[0028] This allows language learning materials to dramatically introduce the students who use them, increasing their motivation to learn.

[0029] [3] Ensuring quality as language teaching materials In the language teaching material production program of [2], the first prompt further includes, as specifications for the interviewer, maintaining a friendly tone with the student and showing empathy for their personal history to create an environment that encourages responses. The second prompt is a prompt that causes the automatic conversation engine to limit the number of related questions to a predetermined number or less in a question-and-answer format. The third prompt is a prompt that specifies, as specifications for the language teaching material, the use of simple vocabulary and syntax, the use of all predetermined basic sentence patterns, and the use of a predetermined level of vocabulary.

[0030] This ensures the quality of the language learning material.

[0031] [4] Language learning materials suited to student's personal information such as occupation In the language learning material production program of [1], the profile is personal information including field of expertise. The first prompt is a prompt for setting specifications of a language learning material writer in the automatic conversation engine. The second prompt includes, as the writer's specifications, asking the student a question in a predetermined language inquiring about the student's language level and the related questions about personal information including the field of expertise. The third prompt specifies that the language learning materials to be produced by the writer are to be a set of learning material sentences in the foreign language to be studied and translations in the predetermined language, that the learning material sentences use vocabulary at the student's language level and all predetermined basic sentence patterns, and that the translations are to be literal translations rather than paraphrases.

[0032] This will ensure that the language learning materials are suited to the students' own fields of expertise, such as their occupations, and will increase their motivation to learn.Furthermore, by including personal information such as age, gender, and hobbies, the language learning materials created can be made to relate to more familiar topics.

[0033] [5] Speaking consulting function for Japanese people. In any one of [1] to [4], the language teaching material production program further inputs a fourth prompt (4) to the automatic conversation engine following the first, second, and third prompts.

[0034] The profile, the related questions, and the answers are in Japanese, the language learning materials are in English, and the fourth prompt is a prompt that instructs the automatic conversation engine to, assuming that the engine is a professional English instructor and speaking consultant for Japanese learners, faithfully convert the pronunciation of the English sentences generated as the language learning materials into katakana words as if they were pronounced by a native speaker and to display them side by side.

[0035] This allows for pronunciation that is easy for Japanese people to understand to be written alongside the English text when the language learning materials being produced are English learning materials for Japanese people, thereby improving the effectiveness of students' speaking learning.

[0036] [6] Conversion procedure for speaking consulting function for Japanese people In the language learning material production program of [5], the fourth prompt is a prompt that further instructs the automatic conversation engine to convert the generated English sentence into pronunciation data that reflects sound changes across words based on general colloquial speech, and to generate the katakana word from the pronunciation data.

[0037] This allows the pronunciation of native speakers to be faithfully expressed in katakana.

[0038] [7] Generative AI Chatbot that Creates Language Learning Materials from Students' Own Profiles A representative embodiment disclosed in the present application is a language learning material creation device (10) in which a program is installed in a computer (50) capable of communicating with an automated conversation engine (80) that uses a large-scale language model, and which sequentially inputs first, second, and third prompts to the automated conversation engine.

[0039] The first prompt is a prompt for the automated conversation engine to set an interviewer specification for the student, the second prompt is a prompt for the automated conversation engine to input a brief profile of the student and ask the student relevant questions about the profile to collect answers, and the third prompt is a prompt for the automated conversation engine to output language learning materials customized for the student by adapting the profile based on the relevant questions and the answers.

[0040] This makes it possible to provide a computer program that produces language learning materials that are customized not only for each student's language level but also for the content.

[0041] [8] Dramatically Dramatizing a Self-Introduction In the language teaching material production device of [7], the profile is a self-introduction. The first prompt or the second prompt includes, as the interviewer's specifications, asking the student questions about the highlights and lowlights of their life and their dreams for the future, and the third prompt includes instructing the automatic conversation engine to dramatize the self-introduction by keeping the theme consistent with the self-introduction and using answers to the questions about the highlights and lowlights of their life and their dreams for the future.

[0042] This allows language learning materials to dramatically introduce the students who use them, increasing their motivation to learn.

[0043] [9] Ensuring quality as language teaching materials In the language teaching material production device of [8], the first prompt further includes, as specifications of the interviewer, maintaining a friendly tone with the student and showing empathy for their personal history to create an environment that encourages responses. The second prompt is a prompt that causes the automatic conversation engine to limit the number of related questions to a predetermined number or less in a question-and-answer format, and the third prompt is a prompt that specifies, as specifications of the language teaching material, the use of simple vocabulary and syntax, the use of all predetermined basic sentence patterns, and the use of a predetermined level of vocabulary.

[0044] This ensures the quality of the language learning material.

[0045]

[10] Language learning materials suited to student's personal information such as occupation. In the language learning material production device of [7], the profile is personal information including field of expertise. The first prompt is a prompt for setting specifications of a language learning material creator in the automatic conversation engine. The second prompt includes, as the creator's specifications, asking the student a question in a predetermined language about the student's language level and the related question about personal information including the field of expertise. The third prompt specifies that the creator should generate, as the language learning material, a set of learning material sentences in the foreign language to be studied and translations in the predetermined language, that the learning material sentences should use vocabulary at the student's language level and all predetermined basic sentence patterns, and that the translations should be literal translations rather than paraphrases.

[0046] This will ensure that the language learning materials are suited to the students' own fields of expertise, such as their occupations, and will increase their motivation to learn.Furthermore, by including personal information such as age, gender, and hobbies, the language learning materials created can be made to relate to more familiar topics.

[0047]

[11] Speaking consulting function for Japanese people. A language teaching material production program (20) in any one of [7] to

[10] , further inputting a fourth prompt (4) following the first, second and third prompts to the automatic conversation engine, wherein the profile, the related questions and the answers are in Japanese, and the language teaching materials are in English.

[0048] The fourth prompt (4) is a prompt that instructs the automatic conversation engine to convert the pronunciation of the English sentences generated as the language learning material into katakana words and write them side by side, faithfully as if they were pronounced by a native speaker, assuming that the person in charge is a professional English teacher and speaking consultant for Japanese learners.

[0049] This allows for pronunciation that is easy for Japanese people to understand to be written alongside the English text when the language learning materials being produced are English learning materials for Japanese people, thereby improving the effectiveness of students' speaking learning.

[0050]

[12] Conversion procedure for speaking consulting function for Japanese people In the language teaching material production device of

[11] , the fourth prompt is a prompt that further instructs the automatic conversation engine to convert the generated English sentence into pronunciation data that reflects sound changes across words based on general colloquial speech, and to generate the katakana word from the pronunciation data.

[0051] This allows the pronunciation of native speakers to be faithfully expressed in katakana.

[0052]

[13] Method for Creating Language Learning Materials from a Student's Own Profile Using Generative AI A representative embodiment disclosed in the present application is a method for creating language learning materials, in which a computer (50) capable of communicating with an automated conversation engine (80) that uses a large-scale language model sequentially inputs first, second, and third prompts (1, 2, 3) to the automated conversation engine. The first prompt (1) is a prompt for setting the specifications of an interviewer for the student to the automated conversation engine, the second prompt (2) is a prompt for inputting a brief profile of the student to the automated conversation engine and asking the student related questions about the profile to collect answers, and the third prompt (3) is a prompt for outputting language learning materials customized for the student by adapting the profile based on the related questions and the answers.

[0053] This makes it possible to provide a computer program that produces language learning materials that are customized not only for each student's language level but also for the content.

[0054]

[14] Dramatizing a Self-Introduction In the language teaching material production method of

[13] , the profile is a self-introduction. The first prompt or the second prompt includes, as the interviewer's specifications, asking the student questions about the highlights and lowlights of their life and their dreams for the future, and the third prompt includes instructing the automatic conversation engine to dramatize the self-introduction by the answers to the questions about the highlights and lowlights of their life and their dreams for the future, while maintaining a theme consistent with the self-introduction.

[0055] This allows language learning materials to dramatically introduce the students who use them, increasing their motivation to learn.

[0056]

[15] Ensuring quality as language teaching materials In the language teaching material production method of

[14] , the first prompt further includes, as the interviewer's specifications, maintaining a friendly tone with the student and showing empathy for their personal history to create an environment that encourages responses. The second prompt is a prompt that causes the automatic conversation engine to limit the number of related questions to a predetermined number or less in a question-and-answer format, and the third prompt is a prompt that specifies, as the language teaching material specifications, the use of concise vocabulary and syntax, the use of all predetermined basic sentence patterns, and the use of a predetermined level of vocabulary.

[0057] This ensures the quality of the language learning material.

[0058]

[16] Language learning materials suited to student's personal information such as occupation. In the language learning material production method of

[15] , the profile is personal information including field of expertise. The first prompt is a prompt for setting specifications of a language learning material writer in the automatic conversation engine. The second prompt includes, as the writer's specifications, asking the student a question in a predetermined language about the student's language level and the related question about personal information including the field of expertise. The third prompt specifies that the language learning materials to be produced by the writer are to be a set of learning material sentences in the foreign language to be studied and translations in the predetermined language, that the learning material sentences use vocabulary at the student's language level and all predetermined basic sentence patterns, and that the translations are to be literal translations rather than paraphrases.

[0059] This will ensure that the language learning materials are suited to the students' own fields of expertise, such as their occupations, and will increase their motivation to learn.Furthermore, by including personal information such as age, gender, and hobbies, the language learning materials created can be made to relate to more familiar topics.

[0060]

[17] Speaking consulting function for Japanese people In any one of

[13] to

[16] , the method for creating language learning materials includes inputting a fourth prompt to the automatic conversation engine following the first, second, and third prompts, wherein the profile, the related questions, and the answers are in Japanese, and the language learning materials are in English. The fourth prompt instructs the automatic conversation engine, assuming that the automatic conversation engine is a professional English instructor and speaking consultant for Japanese learners, to convert the pronunciation of English sentences generated as the language learning materials into katakana words faithfully as if they were pronounced by a native speaker, and to display them side by side.

[0061] This allows for pronunciation that is easy for Japanese people to understand to be written alongside the English text when the language learning materials being produced are English learning materials for Japanese people, thereby improving the effectiveness of students' speaking learning.

[0062]

[18] Conversion procedure for speaking consulting function for Japanese people In the language learning material production method of

[17] , the fourth prompt is a prompt that further instructs the automatic conversation engine to convert the generated English sentence into pronunciation data that reflects sound changes across words based on general colloquial speech, and to generate the katakana word from the pronunciation data.

[0063] This allows the pronunciation of native speakers to be faithfully expressed in katakana.

[0064] The division of the first, second, third, and fourth prompts (1, 2, 3, 4) is not important, as long as the above content is entered sequentially by the prompts. Furthermore, the student's profile information and answers to questions can be entered by the student themselves, or an operator, for example, can do this for them.

[0065] 2. Details of the embodiment The embodiment will be described in further detail.

[0066] [Embodiment 1] FIG. 1 is an explanatory diagram showing a typical example of the configuration of a language teaching material production device and program of the present invention, and FIG. 2 is an explanatory diagram showing a typical example of the configuration of the entire system including a computer 50 on which the language teaching material production program 20 is installed and an automatic conversation engine that utilizes the large-scale language model used by the computer 50.

[0067] The computer 50 on which the language teaching material production program 20 is installed is a computer equipped with a processor 51, a storage device 52, and a user interface 53, and is connected to communicate with an automatic conversation engine 80 that utilizes a large-scale language model via a network 60 such as the Internet. The computer 50 may be, for example, a personal computer or a mobile information device such as a smartphone. Through the user interface, information such as text data can be input and output directly with students using language teaching materials or via language educators. Specific examples of the automatic conversation engine 80 that utilizes a large-scale language model include, but are not limited to, ChatGPT by OpenAI (registered trademark), Gemini by Google (registered trademark), and LLaMA by Meta (registered trademark). The automatic conversation engine 80 may be a chatbot that can automatically output natural responses using a large-scale language model in response to user input. It is known that the automatic conversation engine 80 can output more natural and appropriate responses by receiving commands, instructions, or specifications called prompts. Furthermore, while many known automatic conversation engines 80 are configured to be accessible and usable via the Internet, the language teaching material production device and program of the present invention can access the automatic conversation engine 80 in any manner.

[0068] The language teaching material production program 20 of the present invention is installed in a storage device 52 of a computer 50 that can communicate with an automatic conversation engine 80 that uses a large-scale language model, as shown in Figure 2, and is executed by a processor 51, thereby functioning as a language teaching material production device 10. The computer 50 does not need to be a computer that is directly operated by a user, but may also be a server computer on a cloud that provides cloud applications that users can use via a network such as the Internet.

[0069] The computer 50 executing the language teaching material production program 20 of the present invention inputs the first, second and third prompts (1, 2, 3) to the automatic conversation engine 80 in sequence.

[0070] The first prompt 1 is a prompt that sets the specifications of the interviewer for the student in the automated conversation engine 80. The second prompt 2 is a prompt that prompts the automated conversation engine 80 to input a brief profile of the student and to ask the student relevant questions about the profile and collect answers. The third prompt 3 is a prompt that prompts the automated conversation engine 80 to output language learning materials customized for the student by adapting the input profile based on the relevant questions and answers.

[0071] This makes it possible to provide a computer program that produces language learning materials that are customized not only for each student's language level but also for the content.

[0072] The profile may include, for example, a self-introduction by the student (described in the third embodiment below), occupation or field of expertise (described in the fourth embodiment below), age, gender, hobbies, etc. By setting the specifications of the interviewer for the student in the automatic conversation engine 80 in the first prompt 1, the automatic conversation engine 80 can ask probing questions related to the student's input profile in response to the second prompt 2, and have the student respond, thereby enabling the language learning materials to be created with content that is in line with the student's individual profile. The language learning materials created will have content that is close to the situation in which the student is likely to converse in the foreign language, and will be of interest to the student in line with their hobbies and preferences, thereby motivating the student to learn and providing content that can actually be used, thereby improving the learning effect.

[0073] FIG. 3 is an explanatory diagram showing a typical example of the operation of a computer 50 in which the language teaching material production program 20 of the present invention is installed.

[0074] The session with the automatic conversation engine 80 is displayed on a display, which is one of the user interfaces 53 of the computer 50 operating as the language teaching material production device 10 .

[0075] When the first prompt 1 and the second prompt 2 read from the storage device 52 in the computer 50 are sent to the automatic conversation engine 80, the automatic conversation engine 80 displays an input request sentence a on the display, which prompts the student to input their profile. For example, it displays "Could you please tell me briefly about your occupation and career?" to prompt the student to input their profile.

[0076] The student inputs profile b from a keyboard, which is one of the user interfaces 53. Other media such as file input or voice input may also be used. This can be accommodated by specifying input specifications and processing methods for other media to the automatic conversation engine 80. For example, the automatic conversation engine 80 may be instructed to perform voice recognition and have the student confirm the textual results of the recognition.

[0077] Next, the automatic conversation engine 80 generates and outputs a related question c instructed in the second prompt 2, and obtains an answer d to the related question c. The related question c and answer d are repeated, for example, a predetermined number of times, based on the specifications instructed in the second prompt 2.

[0078] After the related question c and answer d specified in the second prompt 2 are completed, language learning material e is generated in accordance with the third prompt 3 input to the automatic conversation engine 80. Based on the input profile b, the related question c and answer d are adapted into a dramatized sentence, which is output as language learning material e.

[0079] When the learning subject is English, the language learning material e is an English sentence. Furthermore, by inputting a fourth prompt 4 (described later) into the automatic conversation engine 80 and inputting this English language learning material e, language learning material f is output based on the fourth prompt 4, along with the katakana word representing the pronunciation.

[0080] Instead of input and output using a display and keyboard, input and output using other media such as voice may be adopted. Although the description has been given assuming that the students themselves input the data, a mentor or operator may assist or act on their behalf in the input. This is not limited to this embodiment, but is common throughout this specification.

[0081] [Embodiment 2] Entering a Self-Introduction as a Profile In the above-described embodiment 1, the profile entered by the student can be a simple self-introduction of the student. Preferably, the first prompt 1 further includes, as an interviewer specification, instructions to ask the student questions about the highlights and low points of their life and their future dreams and goals. Preferably, the third prompt 3 further includes instructions to the automated conversation engine 80 to maintain a theme consistent with the self-introduction and to dramatize the self-introduction based on the student's answers to the questions about the highlights and low points of their life and their future dreams and goals. Here, "highlights" refer to events in life that the student felt elated about, such as successes. Conversely, "lowlights" refer to events in life that the student felt daunted about, such as challenges. The terms "good events," "troublesome events," and so on may be used interchangeably without strict meaning.

[0082] This allows the language learning materials to become a dramatic introduction to the students who use them, increasing their motivation to learn. The highlights and low points of a student's life are important experiences that involve emotional ups and downs, and using these as language learning materials makes it easy for students to empathize and provides a self-introduction that they will often have the opportunity to share with others, thereby increasing motivation and learning effectiveness. By adding a question about future dreams and goals to Prompt 3, the language learning materials created can be dramatized even if students are unable to effectively answer the questions about the highlights and low points of their lives.

[0083] The first prompt 1 may further include instructions for the interviewer to maintain a friendly tone and demonstrate empathy for the student's personal history to create an environment in which the student is comfortable responding. Furthermore, the second prompt 2 may further include instructions for limiting the number of related questions to a predetermined number in a question-and-answer format, and the third prompt may further include instructions for the language learning material to use concise vocabulary and syntax, to use all of the specified basic sentence patterns, and to use vocabulary at a predetermined level. This ensures the quality of the language learning material. For example, in the case of English, in addition to specifying the use of all five basic sentence patterns, detailed instructions such as compound sentences, relative pronoun constructions, and conditional constructions may be included depending on the level of the student.

[0084] Figure 4 shows an example of prompts 1, 2, and 3. This example uses OpenAI's ChatGPT as the automatic conversation engine 80, which utilizes a large-scale language model. It is not necessary to clearly separate prompts 1, 2, and 3. First, "Interview Navigator is a GPT" instructs GPTs to act as an interviewer. The following statement, "designed for conducting in-depth interviews in Japanese, focusing on life highlights, lows, and future aspirations," instructs the interviewer to ask questions about the student's life highlights and lowlights and future dreams and goals. Following this, "It engages in a friendly tone, creating a comfortable environment for sharing personal stories." By providing instructions on how to conduct the interview, the system can help students relax and respond more easily.

[0085] The next instruction corresponds to Prompt 2: "The GPT tailors questions to the user's career introduction, with a limit of 10 questions per interview and one question per message." This prompts the student to enter a brief profile and then asks them related questions about that profile to collect answers. The format is a question-and-answer format and is limited to 10 questions. This allows the student to relax and not feel overly stressed during the interview. The number of questions can be adjusted appropriately depending on the student's demographics. For example, if the student is young, you can focus on their dreams and goals for the future rather than the highlights and low points of their lives, while asking fewer questions to maintain their concentration. Conversely, if the student is older, you can focus on the highlights and low points of their lives and increase the number of questions depending on their willingness to talk.

[0086] Once the interview is complete, the user can select "After concluding the interview, Interview Navigator will automatically create a self-introduction script in English." This corresponds to the third prompt 3. At this point, it is advisable to specify in detail the specifications for the English script to be generated. In the example in Figure 4, the following five points are specified: (1) "Use simple vocabulary and syntax to ensure clarity": Use simple vocabulary and syntax. (2) "maintain the interviewee as the subject of the script": Make the student the subject of the script. (3) "incorporate all five basic English sentence structures (declarative, interrogative, imperative, exclamatory, and conditional) in the script.": Be sure to use basic sentence structures (the five English sentence structures: SV, SVO, SVC, SVOO, SVOC). (4) "the script will be made based on the vocabulary in correspondence with CEFR B2.": The script should use vocabulary that corresponds to the CEFR B2 level of English. The English level can be generated to suit the student's level. On the other hand, in order to dramatize the script in an impressive way, it is best to create the vocabulary based on the B2 level, regardless of the user's level.(5) "The GPT avoids specific topics for broad discussions, culminating in an emotionally engaging script that reflects the interviewee's journey.": Avoids discussions of generalities, sticks to the interviewee's background, and creates a dramatic presentation. Finally, in general terms, it gives the basic specifications required for a script. This allows you to create a script that is easy for not only the student but also the listener to empathize with.

[0087] 5 and 6 are examples of related questions and their answers, with Fig. 5 being the first half and Fig. 6 being the second half. Fig. 7 is an example of a script generated after the questions and answers in Figs. 5 and 6 are completed.

[0088] In the first half of Figure 5, the Interview Navigator, formed by prompts 1 through 3, first asks about the student's occupation and receives a response. Praising the student for their "wonderful experience," the Interview Navigator then asks for more details about their occupation. The Interview Navigator then asks about the low points of their life, such as "a particularly memorable major challenge," hinting at "various challenges and difficulties," and asks additional questions about the highlights, such as "successful experiences." Furthermore, in the second half of Figure 6, the Interview Navigator asks questions about the student's future dreams and goals, such as "What goals and prospects do you have?" and "Are there any challenges you would like to take on in your future career, or any specific goals you would like to achieve?"

[0089] After completing these questions, the Interview Navigator generates a script for the student to introduce themselves, as shown in Figure 7. The generated script includes the highlights and lowlights of the student's life in order, such as "A significant turning point in my career was...," "This challenging period was...," and "I have since broadened my focus to...," and concludes with "My goal is...," highlighting the student's dreams and goals for the future. This script is easy for the student and the listener to empathize with.

[0090] Furthermore, the scripts that were created met the specified specifications, which were to use the student as the subject, to use the five basic sentence patterns of English, and to use an appropriate level of vocabulary.

[0091] [Embodiment 3] Language learning materials suited to the student's field of expertise, such as his / her occupation In the above-described embodiment 1, the profile can be personal information including the student's own field of expertise.

[0092] 8 shows examples of the first to third prompts in this embodiment. This is an example in which OpenAI's ChatGPT is used as the automatic conversation engine 80 that uses a large-scale language model.

[0093] The first prompt instructs ChatGPT to act as a sentence composer for language learning materials: "Sentence Composer is a professional, detailed GPT designed for English language learners, focusing on CEFR levels."

[0094] The second prompt states, "It begins by confirming the learner's CEFR level. Following this, it gathers additional information about their job, industry, age, gender, and also inquires about their hobbies." This indicates that the writer's specifications are to ask the student questions in the specified language about their language level, as well as questions about their personal information, including their area of ​​expertise.

[0095] The third prompt instructs the writer to create a set of English (the target foreign language) sentences and Japanese translations for the language learning materials. The sentences should use vocabulary at the student's language level and all five basic English sentence patterns. The translations should be literal rather than paraphrased. More specifically, "This comprehensive personal profile, including hobbies, helps Sentence Composer tailor up to 10 example sentences that align with the specified CEFR level, with the level serving as the upper limit." This prompt instructs the writer to create 10 sentences that align with the student's hobbies and that do not exceed the CEFR level. Furthermore, it instructs the writer to provide a precise Japanese translation for each English sentence, as in "Each English sentence is accompanied by a precise Japanese translation." Furthermore, as stated in the document, "The Sentence Composer ensures the inclusion of all five basic English sentence patterns at least once in its examples, providing a varied and practical learning experience while maintaining a professional interaction throughout," the document instructs students to use all five basic English sentence patterns at least once in the text they create, and finally concludes by instructing them to create language teaching materials that provide a varied and practical learning experience while maintaining professional interaction throughout.

[0096] FIG. 9 shows an example output in the third embodiment. When the first to third prompts are input, ChatGPT, the automatic conversation engine 80, functions as a GPTs "Sentence Composer." The Sentence Composer first asks the student for their CEFR level, occupation, industry, age, gender, hobbies, etc. When the student responds with "CEFR level: B1, occupation: software engineer, industry: web services, gender: female, age: 35, hobbies: child-rearing, camping, and watching videos," the Sentence Composer responds, "Thank you. I will create example English sentences for a 35-year-old software engineer with a CEFR level of B1 who works in the web services industry, including their hobbies of child-rearing, camping, and watching videos. Each sentence uses various sentence structures." As instructed, 10 learning sentences are output along with their Japanese translations. Due to space limitations, the fourth and subsequent sentences have been omitted, but the five basic English sentence patterns, such as the first, second, and third sentence patterns, are used at least once each, and the sentences are generated to include a variety of sentence patterns, including conditional statements (if statements), passive sentences, and interrogative sentences.

[0097] As described above, language learning materials can be tailored to the students' own occupations and other fields of expertise, which can increase their motivation to learn. Furthermore, by including personal information such as age, gender, and hobbies, the language learning materials can be made to relate to more familiar topics.

[0098] [Embodiment 4] The language learning material production program 20 of the present invention can input a fourth prompt 4 to the automatic conversation engine 80 following the first, second, and third prompts (1, 2, 3). This embodiment is for a case in which the profile, related questions, and answers input by the student are in Japanese, and the language learning materials are in English. The fourth prompt 4 instructs the automatic conversation engine 80 to specify that the student is a professional English instructor and speaking consultant for Japanese learners, and to convert the pronunciation of the English sentences generated in the above-described embodiment 1, 2, or 3 into katakana words that faithfully reproduce the pronunciation of a native speaker and include them in the language learning materials to be produced. This allows pronunciations that are easy for the student to understand to be included alongside the English sentences, thereby improving the effectiveness of the student's speaking learning.

[0099] More preferably, the fourth prompt 4 further includes instructions to the automatic conversation engine 80 to convert the input English sentence into pronunciation data that reflects inter-word phonetic changes based on general colloquial speech and generate katakana words from the pronunciation data. This allows the pronunciation of a native speaker to be faithfully represented in katakana words. While a word and its phonetic symbols representing its pronunciation have a one-to-one correspondence when using a dictionary as a reference, in actual sentences, pronunciations of adjacent words influence and change, and also change depending on the speaker's emotions and intentions. Therefore, by reflecting inter-word phonetic changes based on general colloquial speech in the phonetic symbols, more natural pronunciations can be generated.

[0100] Here, the pronunciation data may be text-based data such as phonetic symbols, speech data itself, or data expressed by extracting features from speech. Examples of speech features include consonants, vowels, stress, rhyme, intonation, and physical data such as frequency and sound pressure. The following is an example in which phonetic symbols are used as the pronunciation data.

[0101] 10 shows an example of a fourth prompt in this embodiment, in which ChatGPT from OpenAI is used as the automatic conversation engine 80 that uses a large-scale language model.

[0102] First, you give ChatGPT the command, "You are a professional English teacher and speaking consultant for Japanese learners. Your job is to 'convert the pronunciation of English sentences into Katakana words faithfully, so that they sound like a native speaker would pronounce them.' The conversion steps are as follows."

[0103] Next, the system provides the conversion procedure. It provides specific instructions: first, "Convert English text into phonetic symbols," and then, "Convert the output phonetic symbols into katakana words faithfully to native pronunciation." It also includes important points to note: "When converting to phonetic symbols, cross-word sound changes such as linking, reduction, flapping, assimilation, and weakening often occur, so these should be taken into consideration when converting." It also instructs that "Pronunciation should be based on general American English," and "When converting to katakana words, if there is no katakana that matches the pronunciation, select the one that sounds closest." It also informs the user that a reference example will be provided below as a "model," and instructs the user to use the output format as a copy of the example, and to output only the final result without including the conversion procedure. Finally, it provides a reference example as a "model."

[0104] FIG. 11 shows an example of output in the fourth embodiment.

[0105] When the sentence to be converted is "But that was not so easy. I tried the auditions, and I failed countless times. But actually I didn't care because I had confidence that I could make it," the pronunciation in katakana is written alongside each English sentence. In this case, the katakana is not written in 1:1 correspondence with the English word, but rather reflects the changes in pronunciation influenced by adjacent words. In this way, pronunciation that is easy for students to understand is written alongside the English sentence, improving the effectiveness of speaking learning.

[0106] As mentioned above, the pronunciation data may be voice data itself instead of phonetic symbols. In that case, the specific conversion procedure described above is realized by including in the fourth prompt a specific instruction such as "first, convert the English sentence into voice data of a native English speaker, and then convert the converted voice data into katakana words faithfully to the pronunciation."

[0107] Furthermore, the pronunciation data may be configured by including multiple conversion steps. For example, specific instructions may be given to convert a given English sentence into speech data by a native English speaker, extract features such as consonants, vowels, stress, rhyme, and intonation from the speech data, and then generate a katakana word by associating the features with the katakana that best suits them.

[0108] Alternatively, specific instructions may be given to generate katakana words by extracting features such as consonants, vowels, stress, rhyme, and intonation from a given English sentence, faithfully as a native speaker would pronounce it, without using speech data from a native speaker, and then associating the features with the katakana that best suits them.

[0109] This example shows the words being converted into katakana for Japanese people, but it can be changed to any other phonetic characters to suit the student's native language, such as Hangul for Koreans.

[0110] This embodiment has been described as the fourth prompt 4 being input to the automatic conversation engine 80 following the first, second, and third prompts (1, 2, 3). However, by inputting the above-mentioned prompt disclosed as the fourth prompt 4 alone into the automatic conversation engine 80, it is possible to realize a language learning material production program, a language learning material production device, or a language learning material production method that can convert any English sentence into a similar katakana word.

[0111] This embodiment is an example of a language learning material for Japanese-speaking students learning English, but the language to be learned may be any language. Furthermore, similar embodiments can be realized using any language that uses phonetic characters or common symbols to represent pronunciation instead of Japanese.

[0112] The invention made by the inventor has been specifically described above based on an embodiment, but it goes without saying that the invention is not limited thereto and can be modified in various ways without departing from the spirit of the invention.

[0113] The present invention can be suitably used in a language teaching material production program, a language teaching material production device, and a language teaching material production method for producing language teaching materials suitable for students.

[0114] 1, 2, 3, 4 Prompt 10 Language teaching material production device 20 Language teaching material production program 50 Computer 51 Processor 52 Storage device 53 User interface 60 Network 80 Automatic conversation engine

Claims

1. A language teaching material production program executed by a computer capable of communicating with an automatic conversation engine that utilizes a large-scale language model, the program sequentially inputting first, second and third prompts to the automatic conversation engine, wherein the first prompt is a prompt that sets the automatic conversation engine to specify an interviewer for a student, the second prompt is a prompt that causes the automatic conversation engine to input a brief profile of the student and to ask the student related questions about the profile and collect answers, and the third prompt is a prompt that causes the automatic conversation engine to output language teaching materials customized for the student by adapting the profile based on the related questions and the answers.

2. A language teaching material production program according to claim 1, wherein the profile is a self-introduction, the first prompt or the second prompt includes, as a specification of the interviewer, asking the student questions about the highlights and lowlights of their life and their dreams for the future, and the third prompt includes instructing the automatic conversation engine to maintain a theme consistent with the self-introduction and to dramatize the self-introduction with answers to the questions about the highlights and lowlights of their life and their dreams for the future.

3. A language teaching material production program as set forth in claim 2, wherein the first prompt further includes, as the interviewer's specifications, maintaining a friendly tone toward the student and showing empathy for their personal history to create an environment that encourages responses; the second prompt is a prompt that causes the automatic conversation engine to limit the number of related questions to a predetermined number or less in a question-and-answer format; and the third prompt is a prompt that specifies, as the language teaching material specifications, the use of simple vocabulary and syntax, the use of all predetermined basic sentence patterns, and the use of a predetermined level of vocabulary.

4. A language teaching material production program according to claim 1, wherein the profile is personal information including fields of expertise; the first prompt is a prompt for setting specifications of a writer of language teaching materials in the automatic conversation engine; the second prompt includes, as the writer's specifications, asking the student a question in a predetermined language about the student's language level and the related question about personal information including the field of expertise; and the third prompt specifies that, as language teaching materials to be produced by the writer, a set of teaching material sentences in the foreign language to be studied and translations in the predetermined language should be generated, that the teaching material sentences should use vocabulary of the language level indicated by the student and all predetermined basic sentence patterns, and that the translations should be literal translations rather than paraphrases.

5. A language teaching material production program according to any one of claims 1 to 4, which further inputs a fourth prompt to the automatic conversation engine following the first, second and third prompts, wherein the profile, the related questions and the answers are in Japanese, the language teaching materials are in English, and the fourth prompt instructs the automatic conversation engine, assuming that the person is a professional English instructor and speaking consultant for Japanese learners, to faithfully convert the pronunciation of English sentences generated as the language teaching materials into katakana words as if they were pronounced by a native speaker and to display them side by side.

6. A language teaching material production program according to claim 5, wherein the fourth prompt is a prompt that further instructs the automatic conversation engine to convert the generated English sentence into phonetic symbols that reflect sound changes across words based on general colloquial speech, and to generate the katakana word from the phonetic symbols.

7. A language teaching material production device having installed on a computer capable of communicating with an automatic conversation engine that uses a large-scale language model a program that sequentially inputs first, second, and third prompts to the automatic conversation engine, wherein the first prompt is a prompt that sets the automatic conversation engine to the specifications of an interviewer for a student, the second prompt is a prompt that causes the automatic conversation engine to input a brief profile of the student and to ask the student related questions about the profile and collect answers, and the third prompt is a prompt that causes the automatic conversation engine to output language teaching materials customized for the student by adapting the profile based on the related questions and the answers.

8. A language teaching material production device according to claim 7, wherein the profile is a self-introduction, the first prompt or the second prompt includes, as a specification of the interviewer, asking the student questions about the highlights and lowlights of their life and their dreams for the future, and the third prompt includes instructing the automatic conversation engine to maintain a theme consistent with the self-introduction and to dramatize the self-introduction with answers to the questions about the highlights and lowlights of their life and their dreams for the future.

9. A language teaching material production device according to claim 8, wherein the first prompt further includes, as the interviewer's specifications, maintaining a friendly tone toward the student and showing empathy for their personal history to create an environment that encourages responses; the second prompt is a prompt that causes the automatic conversation engine to limit the number of related questions to a predetermined number or less in a question-and-answer format; and the third prompt is a prompt that specifies, as the language teaching material specifications, the use of simple vocabulary and syntax, the use of all predetermined basic sentence patterns, and the use of a predetermined level of vocabulary.

10. A language teaching material production device according to claim 7, wherein the profile is personal information including a field of expertise, the first prompt is a prompt for setting specifications of a creator of language teaching materials in the automatic conversation engine, the second prompt includes, as the creator's specifications, asking the student a question in a predetermined language about the student's language level and the related question about personal information including the field of expertise, and the third prompt specifies that, as language teaching materials to be produced by the creator, a set of teaching material sentences in the foreign language to be studied and translations in the predetermined language should be generated, the teaching material sentences should use vocabulary of the language level indicated by the student and all predetermined basic sentence patterns, and the translations should be literal translations rather than paraphrases.

11. A language teaching material production device according to any one of claims 7 to 10, which is a language teaching material production program that inputs a fourth prompt to the automatic conversation engine following the first, second, and third prompts, wherein the profile, the related questions, and the answers are in Japanese, the language teaching materials are in English, and the fourth prompt instructs the automatic conversation engine, assuming that the person is a professional English instructor and speaking consultant for Japanese learners, to faithfully convert the pronunciation of English sentences generated as the language teaching materials into katakana words as if they were pronounced by a native speaker and to write them side by side.

12. A language teaching material production device according to claim 11, wherein the fourth prompt is a prompt that further instructs the automatic conversation engine to convert the generated English sentence into phonetic symbols that reflect sound changes across words based on general colloquial speech, and to generate the katakana word from the phonetic symbols.

13. A method for producing language teaching materials, in which a computer capable of communicating with an automatic conversation engine that uses a large-scale language model sequentially inputs first, second, and third prompts to the automatic conversation engine, wherein the first prompt is a prompt that sets the automatic conversation engine to the specifications of an interviewer for the student, the second prompt is a prompt that causes the automatic conversation engine to input a brief profile of the student and to ask the student related questions about the profile and collect answers, and the third prompt is a prompt that causes the automatic conversation engine to output language teaching materials customized for the student by adapting the profile based on the related questions and the answers.

14. A method for producing language teaching materials according to claim 13, wherein the profile is a self-introduction, the first prompt or the second prompt includes, as a specification of the interviewer, asking the student questions about the highlights and lowlights of their life and their dreams for the future, and the third prompt includes instructing the automatic conversation engine to maintain a theme consistent with the self-introduction and to dramatize the self-introduction with answers to the questions about the highlights and lowlights of their life and their dreams for the future.

15. A language teaching material production program according to claim 14, wherein the first prompt further includes, as the interviewer's specifications, maintaining a friendly tone toward the student and showing empathy for their personal history to create an environment that encourages responses; the second prompt is a prompt that causes the automatic conversation engine to limit the number of related questions to a predetermined number or less in a question-and-answer format; and the third prompt is a prompt that specifies, as the language teaching material specifications, the use of simple vocabulary and syntax, the use of all predetermined basic sentence patterns, and the use of a predetermined level of vocabulary.

16. A method for producing language teaching materials according to claim 13, wherein the profile is personal information including a field of expertise; the first prompt is a prompt for setting specifications of a writer of language teaching materials in the automatic conversation engine; the second prompt includes, as the writer's specifications, asking the student a question in a predetermined language about the student's language level and the related question about personal information including the field of expertise; and the third prompt specifies that, as language teaching materials to be produced by the writer, a set of teaching material sentences in the foreign language to be studied and translations in the predetermined language should be generated, that the teaching material sentences should use vocabulary of the language level indicated by the student and all predetermined basic sentence patterns, and that the translations should be literal translations rather than paraphrases.

17. A language teaching material production method according to any one of claims 13 to 16, further comprising inputting a fourth prompt to the automatic conversation engine following the first, second, and third prompts, wherein the profile, the related questions, and the answers are in Japanese, the language teaching materials are in English, and the fourth prompt instructs the automatic conversation engine, assuming that the person is a professional English instructor and speaking consultant for Japanese learners, to faithfully convert the pronunciation of English sentences generated as the language teaching materials into katakana words so as to be pronounced by a native speaker and to display them side by side.

18. A method for producing language teaching materials according to claim 17, wherein the fourth prompt is a prompt that further instructs the automatic conversation engine to convert the generated English sentence into phonetic symbols that reflect sound changes across words based on general colloquial speech, and to generate the katakana word from the phonetic symbols.

Images