Language learning support processing method, language learning support processing device and program

Abstract

We provide efficient language learning support. [Solution] A language learning support method performed by one or more processors includes acquiring presentation information including quiz information, outputting the presentation information in a manner recognizable by a user, acquiring input information entered by a user, acquiring answer information entered by a user, and enabling the input of the answer information or activating the acquired answer information when predetermined conditions are met, wherein the predetermined conditions include that a predetermined amount, predetermined percentage, or predetermined portion of the text included in the presentation information has been reproduced based on the input information.

Description

Technical Field

[0001] The present invention relates to information processing technology in language learning support.

Background Art

[0002] In recent years, the demand for language learning has increased, and learning tools and educational materials for supporting this have also diversified. For example, various styles of learning are provided, such as learning foreign language words and grammar, reading comprehension of texts, speech pronunciation practice, and answering translation problems, and these are devised to enable effective learning by combining them. Patent Document 1 discloses an example thereof.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

[0004] However, since many conventional learning methods are in a format where the learning pace and content are adjusted at the learner's own discretion, there is a problem that it is difficult to maintain concentration and the sense of tension in learning is hard to continue. As a result, the load on learning is not sufficiently ensured, the stimulation to the brain is weakened, and it is not uncommon for memory retention and operational ability improvement to become inefficient. In order to avoid such problems, there is a so-called immersion-type learning method that forcibly places the learner in a usage environment of the target language for learning, but this method imposes a large physical and economic burden related to travel and accommodation, and is difficult for general learners to implement. Accordingly, there is a demand for a new learning method that can bring out a higher concentration state and better learning effect by applying an appropriate load to the learner and requiring the brain to perform multiple processes simultaneously even in a daily environment.

Summary of Invention

Problems to be Solved by the Invention

[0005] The present invention aims to provide efficient language learning support. [Means for solving the problem]

[0006] According to a first aspect of the present invention, a language learning support method executed by one or more processors includes: acquiring quiz text; displaying the quiz text; acquiring user-inputted audio information; and, if it is determined that a predetermined proportion or amount of the quiz text has been reproduced based on the audio information, enabling the input of an answer to the quiz text or validating an inputted answer. According to a second aspect of the present invention, a language learning support method performed by one or more processors includes acquiring a quiz text, gradually displaying the quiz text, and outputting an audio recording of the quiz text, wherein the speed of gradually displaying the quiz text is slower than the speed of outputting the audio recording of the quiz text, or the outputting of the audio recording of the quiz text precedes, at least partially, the gradually displaying of the quiz text. According to a third aspect of the present invention, a language learning support method executed by one or more processors includes: acquiring a quiz text; displaying the quiz text on the display units of at least one terminal constituting a first terminal group and at least one terminal constituting a second terminal group; acquiring voice information input by a user in the first terminal group; acquiring character information input by a user in the second terminal group; and, if it is determined that a predetermined proportion or amount of the quiz text has been reproduced based on the voice information and the character information, enabling the input of an answer to the quiz text or validating an input answer in at least one terminal constituting the first terminal group. According to a fourth aspect of the present invention, a language learning support method executed by one or more processors includes: acquiring a quiz text in a first language and a quiz text in a second language having the same content as the quiz text in the first language but different from the first language; displaying the quiz text in the first language on the display unit of a first terminal; displaying the quiz text in the second language on the display unit of a second terminal; acquiring voice information input by a user at each of the first and second terminals; if each of the first and second terminals determines, based on the voice information, that a predetermined amount or a predetermined percentage of the quiz text has been reproduced, enabling the input of an answer to the quiz text at each of the first and second terminals, or validating an inputted answer; accepting user answers at each of the first and second terminals; and determining an incentive corresponding to the user answer at the first terminal based on the comparison result of the correct quiz answer with the user answer at each of the first and second terminals, and the time at which the user answer was input at each of the first and second terminals. According to a fifth aspect of the present invention, a language learning support method executed by one or more processors includes: acquiring quiz text; displaying the quiz text on the display units of at least one terminal constituting a first terminal group and at least one terminal constituting a second terminal group; acquiring voice information input by a user in the first terminal group; and, if it is determined that a predetermined amount or a predetermined percentage of the quiz text has been reproduced based on the voice information acquired in the first terminal group, enabling the input of an answer to the quiz text or validating an input answer in at least one terminal constituting the second terminal group. According to a sixth aspect of the present invention, the present invention includes: obtaining a first quiz text and a second quiz text having the same correct quiz answers as the first quiz text but with a different difficulty level; displaying the first quiz text on the display unit of a first terminal; displaying the second quiz text on the display unit of a second terminal; obtaining voice information input by a user in each of the first and second terminals; if, in each of the first and second terminals, it is determined based on the voice information that a predetermined amount or a predetermined percentage of the quiz text has been reproduced, enabling the input of an answer to the quiz text or validating an inputted answer in each of the first and second terminals; accepting user answers in each of the first and second terminals; and determining an incentive corresponding to the user answer in the first terminal based on the comparison result between the correct quiz answers and the user answers in each of the first and second terminals, and the time the user answers were input in each of the first and second terminals. According to a seventh aspect of the present invention, a language learning support method executed by one or more processors includes acquiring presentation information including quiz information, outputting the presentation information in a manner recognizable by a user, acquiring input information entered by a user, acquiring answer information entered by a user, and enabling the input of the answer information or activating the acquired answer information when predetermined conditions are met, wherein the predetermined conditions include the reproduction of a predetermined amount, predetermined proportion, or predetermined portion of the text included in the presentation information based on the input information. According to an eighth aspect of the present invention, the language learning support method described in the seventh aspect further includes obtaining the correct answer to quiz information and causing a generating AI to generate quiz information based on an input including the correct answer. The input may further include contextual information based on a predetermined viewpoint regarding the correct answer, and may further include generating and storing contextual information for each of the multiple viewpoints, and generating multiple different quiz information for the same correct answer using different contextual information. According to the ninth aspect of the present invention, the language learning support method described in the eighth aspect further includes, in obtaining the correct answer, referring to a seed database that stores candidate correct answers as seeds, and selecting the seed by applying at least one of the following: (a) cooldown control which temporarily excludes seeds from the candidate questions for which a predetermined cooldown period has not elapsed since the most recent question; (b) usage count retirement control which permanently excludes seeds from the candidate questions for which the cumulative number of questions has reached a predetermined threshold; and (c) time-series retirement control which excludes seeds classified as topical genres from the candidate questions for which a predetermined period has elapsed since registration in the database. According to the tenth aspect of the present invention, in the language learning support method described in the seventh aspect, the determination of whether or not the reproduction has been performed is made by comparing the word sequence obtained from the input information with the expected word sequence from the word at the current position of the reproduction judgment in the text included in the presentation information to a predetermined number of words ahead, and the matching is considered successful if the edit distance between the word sequence and the expected word sequence is less than or equal to a predetermined threshold in the matching. According to an eleventh aspect of the present invention, the language learning support method described in the seventh aspect further includes obtaining a group of correct answers corresponding to the quiz information, and includes obtaining first answer information at a first terminal and obtaining second answer information at a second terminal, wherein the first answer information is considered correct if it corresponds to one of the correct answers included in the group of correct answers, and the second answer information is considered correct if it corresponds to another correct answer different from the first correct answer included in the group of correct answers. Furthermore, the answer information may be obtained preferentially at the terminal that satisfies the predetermined conditions first among the first and second terminals. According to a twelfth aspect of the present invention, the language learning support method described in the seventh aspect further includes obtaining a correct answer corresponding to the quiz information, obtaining new quiz information corresponding to the correct answer if the obtained answer information does not correspond to the correct answer, and outputting presentation information including the new quiz information in a manner that can be recognized by the user. According to a thirteenth aspect of the present invention, a language learning support method executed by one or more processors includes acquiring presentation information including quiz information, outputting the presentation information in a manner recognizable by a user, acquiring input information entered by a user, acquiring answer information entered by a user, and enabling the input of the answer information or activating the acquired answer information when predetermined conditions are met, wherein the predetermined conditions include that a predetermined amount, predetermined proportion, or predetermined portion of the text included in the presentation information has been reproduced based on the input information. According to a fourteenth aspect of the present invention, a language learning support method performed by one or more processors includes acquiring presentation information including quiz information, outputting the presentation information in a manner recognizable by a user, acquiring answer information entered by a user, acquiring input information entered by a user, and, when predetermined conditions are met, enabling the reproduction of text contained in the presentation information based on the input information, or activating the reproduction that has been made, wherein the predetermined conditions include the completion of inputting the answer information, or the answer information satisfying predetermined conditions. According to a 15th aspect of the present invention, a language learning support method executed by one or more processors includes acquiring quiz information, a plurality of clue pieces related to the quiz information, and a clue answer information corresponding to each clue piece; outputting the plurality of clue pieces in a manner recognizable by a user; acquiring user input corresponding to each of the plurality of clue pieces; acquiring answer information entered by the user; and, when a predetermined condition is met, enabling the input of the answer information or activating the acquired answer information, wherein the predetermined condition includes that the user input corresponding to a predetermined number or a predetermined proportion of the plurality of clue pieces corresponds to the clue answer information. [Effects of the Invention]

[0007] According to the present invention, efficient language learning support can be provided. [Brief explanation of the drawing]

[0008] [Figure 1] An explanatory diagram showing an example of a system configuration according to the first embodiment. [Figure 2] A block diagram showing an example of the functional configuration of the server 10 according to the first embodiment. [Figure 3] A block diagram showing an example of the functional configuration of a mobile terminal 20 according to the first embodiment. [Figure 4] A flowchart of the language learning support process according to the first embodiment. [Figure 5] A diagram showing an example screen of a mobile terminal 20 according to the first embodiment. [Figure 6] A diagram showing an example screen of a mobile terminal 20 according to the first embodiment. [Figure 7] A diagram showing an example screen of a mobile terminal 20 according to the first embodiment. [Figure 8] A diagram showing an example of the configuration of the user information table 153 according to the first embodiment. [Figure 9] A diagram showing an example screen of a mobile terminal 20 according to the first embodiment. [Figure 10]Flowchart of language learning support processing according to the second embodiment. [Figure 11] Diagram showing an example of a screen of the mobile terminal 20 according to the second embodiment. [Figure 12] Diagram showing an example of a screen of the mobile terminal 20 according to the third embodiment. [Figure 13] Diagram showing an example of a screen of the mobile terminal 20 according to the fourth embodiment. [Figure 14] Diagram showing an example of a screen of the mobile terminal 20 according to the fifth embodiment. [Figure 15] Diagram showing an example of a screen of the mobile terminal 20 according to the sixth embodiment. [Figure 16] Diagram showing an example of a screen of the mobile terminal 20 according to the seventh embodiment. [Mode for Carrying Out the Invention]

[0009] Hereinafter, an example of the mode for carrying out the present invention will be described with reference to the drawings. In the description of the drawings, the same elements may be denoted by the same reference numerals to omit redundant descriptions. Further, the constituent elements described in these embodiments are merely examples, and are not intended to limit the scope of the present invention thereto.

[0010] <First Embodiment> Hereinafter, embodiments for implementing the information processing technology of the present invention will be described. The content described in the present embodiment is applicable to any of the other embodiments and respective modifications. Further, in each embodiment, the content is also applicable to any of the embodiments including the present embodiment and respective modifications.

[0011] Figure 1 is a diagram showing an example of the system configuration of a language learning support system 1 according to the present embodiment. The system may be, for example, configured to include a plurality of information processing apparatuses, and these information processing apparatuses may cooperate to execute processes. Furthermore, a system may be comprised of multiple devices (such as servers and terminals) each equipped with an information processing device as a control unit, and these devices may cooperate to perform processing.

[0012] In the language learning support system 1, for example, one or more mobile terminals 20 (mobile terminal 20A associated with user A, mobile terminal 20B associated with user B, mobile terminal 20C associated with user C, mobile terminal 20D associated with user D, ...) and the server 10 are connected in a way that allows communication via a network NW (for example, the internet). The server 10 may be a server system composed of multiple servers. Alternatively, the server 10 may be a device composed of multiple information processing devices that do not share a storage unit. The network NW may be a communication network that connects each device via wired connections, wireless connections, or both.

[0013] Server 10 has the function of, for example, transmitting language learning content such as quiz texts to each mobile terminal 20 based on a language learning program stored in the memory unit, and receiving data transmitted based on processing and operations on each mobile terminal 20, and performing appropriate processing for language learning support. Details of the functions of Server 10 will be described later.

[0014] Figure 2 shows an example of the functional configuration of server 10. Server 10 may be configured, for example, as a single server, or as a distributed server consisting of separate servers for each function. Server 10 may also be configured as a distributed virtual server created in a cloud environment, known as a cloud server. The server 10 is configured to include, for example, a control unit 11, a storage unit 15, a communication unit 14, an input / output unit 12, and so on.

[0015] The control unit 11 is a processing unit that comprehensively controls each part of the server 10 according to various programs such as system programs stored in the memory unit 15, and performs various processing related to information processing. It is composed of processors such as CPUs, GPUs, and DSPs, and integrated circuits such as ASICs.

[0016] The memory unit 15 is a storage device that includes volatile or non-volatile memory such as ROM, EEPROM, flash memory, RAM, or a hard disk drive. In particular, in this embodiment, the storage unit 15 stores a language learning support processing program 151 and a language learning support processing database 152.

[0017] The language learning support processing program 151 is a program that works in cooperation with the mobile terminal 20 to implement the language learning support processing, which will be described in detail later. The language learning support processing database 152 is a database for storing data necessary for language learning support processing and data generated by language learning support processing. For example, the language learning support processing database 152 includes a user information table 153 for storing user information. Furthermore, it stores a quiz text database 154 in which quiz texts for output to each mobile terminal 20 are registered. Note that each quiz text is associated with a corresponding correct quiz answer.

[0018] The input / output unit 12 includes devices for inputting various operations to the server 10, and devices for outputting processing results processed by the server 10. The input / output unit 12 may have an integrated input section and an output section, or they may be separate. The input unit is implemented, for example, by a keyboard or touch panel, and has the function of receiving input from the user and transmitting the information related to the input to the control unit 11. The output unit is implemented, for example, by a display, speaker, etc., and has the function of outputting the processing results processed by the control unit 11. Here, the input / output unit 12 includes, for example, a display unit 13. The display unit 13 is a display device configured to have an LCD (Liquid Crystal Display) or an OLED (Organic Electro-luminescence Display), etc., and performs various displays based on the display signals output from the control unit 11. A configuration without an output unit may be adopted, and a configuration without a display unit 13 may also be adopted.

[0019] The communication unit 14 is a communication device for sending and receiving information used within the server 10 to and from external devices via a network. Various communication methods can be applied to the communication unit 14, including wired connection via a cable compliant with a predetermined communication standard such as Ethernet or USB (Universal Serial Bus), wireless connection using wireless communication technology compliant with a predetermined communication standard such as Wi-Fi (registered trademark) or 5G (fifth-generation mobile communication system), and connection using short-range wireless communication such as Bluetooth (registered trademark).

[0020] Figure 3 shows an example of the functional configuration of the mobile terminal 20. The mobile terminal 20 is configured, for example, to include a control unit 21, a storage unit 28, an input / output unit 23, and a communication unit 22. The mobile terminal 20 is intended to be implemented by each embodiment described in this disclosure, for example, a smartphone or a tablet, but is not limited to these. Any device having the elements described below may be used, for example, a laptop or a non-portable desktop machine. In other words, it may be a general computer device that a user can use as their own machine. In Figure 1, four mobile terminals 20A, 20B, 20C, and 20D are shown. This is a general representation of the mobile terminals 20 that appear in each embodiment, and the number of mobile terminals 20 participating in the quiz session in each embodiment may be arbitrary.

[0021] The control unit 21 is a processing unit that comprehensively controls each part of the mobile terminal 20 according to various programs such as the language learning support processing program 281 stored in the memory unit 28, and performs various information processing operations. It is composed of a processor such as a CPU, GPU, or DSP, and an integrated circuit such as an ASIC.

[0022] The memory unit 28 is a storage device that, like the memory unit 15, is configured to include volatile or non-volatile memory such as ROM, EEPROM, flash memory, RAM, or a hard disk drive. In particular, in this embodiment, the storage unit 28 stores a language learning support processing program 281 and a language learning support processing database 282.

[0023] The language learning support processing program 281 is a program that works in cooperation with the server 10 to implement the language learning support processing, which will be described in detail later. The language learning support processing database 282 is a database for storing data necessary for language learning support processing and data generated by language learning support processing. For example, the language learning support processing database 282 transmits and receives information from the language learning support processing database 152 on the server 10 as needed, shares information, and stores information about the user of the mobile terminal 20.

[0024] The input / output unit 23 includes devices for inputting various operations to the control unit 21 of the mobile terminal 20, and devices for outputting processing results processed by the control unit 21 of the mobile terminal 20. The input / output unit 23 may have an integrated input section and an output section, or they may be separate. The input unit is implemented, for example, by a keyboard or touch panel, and has the function of receiving input from the user and transmitting the information related to the input to the control unit 21. The output unit is implemented, for example, by a display, speaker, etc., and has the function of outputting the processing results processed by the control unit 31. In this configuration, the input / output unit 23 includes, for example, a display unit 24, an audio input unit 25, an audio output unit 26, and an imaging unit 27.

[0025] The display unit 24 is a display device configured with an LCD (Liquid Crystal Display) or an OLED (Organic Electro-luminescence Display), and performs various displays based on the display signals output from the control unit 21.

[0026] The input / output unit 23 may also have a touch panel (not shown) integrated with the display unit 24, and this touch panel may function as an input interface between the user and the mobile terminal 20.

[0027] The sound input unit 25 is a sound input device that includes a microphone, an A / D converter, etc., and performs various sound inputs based on the sound input signal input to the control unit 21.

[0028] The sound output unit 26 is a sound output device that includes a D / A converter, a speaker, etc., and performs various sound outputs based on the sound output signal output from the control unit 21.

[0029] The imaging unit 27 is an imaging device consisting of a camera or the like, and acquires image data (still images and moving images) that is input to the control unit 21.

[0030] The communication unit 22 is a communication device for sending and receiving information used within the mobile terminal 20 to and from external devices via a network NW. Since the configuration of the communication unit 22 can be similar to that of the communication unit 14, a detailed explanation is omitted.

[0031] <Language learning support processing> The following describes the language learning support process performed collaboratively by server 10 and mobile terminal 20. The explanation below is based on the flowchart of the language learning support process shown in Figure 4.

[0032] In this embodiment, the server 10 and two mobile terminals 20, mobile terminal 20A and mobile terminal 20B, cooperate to perform language learning support processing. However, this is merely one example of an embodiment, and the number of mobile terminals 20 may be one or more.

[0033] Outputting the presented information in a manner recognizable by the user means outputting the information in any manner that allows the user to recognize the content of the presented information, and is not limited to the display of text by the display unit 24. For example, it may be done solely by audio output by the sound output unit 26, a combination of text display and audio output, or by other sensory-appealing output means.

[0034] Furthermore, in this embodiment, the display unit 24, which receives input to instruct the start of a quiz session and outputs information to the user, is not limited to but may consist of elements as shown in Figure 5 as an example.

[0035] As shown in Figure 5, the display unit 24 is composed of areas AR01, AR02, AR03, AR04, and AR05. AR01 is an area for indicating which language the user of the mobile terminal 20 is currently learning through the language learning support system 1. In this embodiment, it indicates that the user is learning English (indicated by the British flag). AR02 is an area for indicating the user's learning level in the target language in the language learning support system 1, and it indicates that the level is 1. AR03 is an area for indicating how much incentive the user has earned in the language learning support system 1, and it indicates that the user has earned a total score of 14,400 points. AR04 is an area for displaying the quiz content (quiz text, etc.) of the current quiz session in the language learning support system 1. AR05 is an area in the language learning support system 1 used to display instructions to the user or to allow some kind of input. It includes an area for entering answers to quiz questions ("Answer?:") and a button that accepts input to start a new quiz session (labeled "Next Quiz").

[0036] In this embodiment, the incentives provided by the language learning support system 1 are not limited to points, but may be any incentive. For example, they may include badges or titles, level-ups, the provision of customization elements (avatar clothing, effects, etc.), ranking improvements, feedback and praise messages, access to new content, etc.

[0037] First, each mobile terminal 20 receives input to instruct the start of a new quiz session (TA2001, TB2001). Here, a quiz session refers to a set of processes related to a single quiz, from presenting a quiz to a user linked to a mobile device 20, to the user answering the quiz and then processing the results according to the answer.

[0038] When the server 10 receives a predetermined number (for example, two) or more inputs from the mobile terminal 20 to start a quiz session (such as an input to instruct the start of a quiz session or an input indicating a desire to start a quiz session) (S1002), the server 10 starts the quiz session (S1003). In other words, in this embodiment, a quiz session starts with two users as participants when two users are present. The conditions for starting a quiz session are not limited to the number of such inputs and can be arbitrary. For example, a new quiz session may be started when a predetermined amount of time has elapsed since the end of the previous quiz session. Alternatively, a quiz session may be started based on absolute time; for example, a quiz session may be started every two minutes. In other words, the current time may be monitored by referring to the network time, and the process of presenting a predetermined quiz session may be started every two minutes, for example, at even minutes past the hour, such as 1:02, 1:04, and 1:06. In this case, if the conditions for starting the quiz session are not met, for example, if the number of users who can participate in the quiz session is less than the predetermined number (for example, less than 5, or 0 if there are no applicants), the quiz session may not be held at that time, or the quiz session may be held with only the users who have expressed their intention to participate (for example, if a quiz session was planned to be a competitive format with multiple users, but only one user expressed their intention to participate, the quiz session may be held in a single-user format instead of a competitive format).

[0039] Alternatively, pressing the button labeled "Next Quiz" in area AR05 of Figure 5 may indicate participation in the next quiz session, and this may correspond to the "input required to start a quiz session."

[0040] When a quiz session is started, the server 10 retrieves the quiz text and sends it to the mobile terminal 20 that has entered the information to start the quiz session (S1004).

[0041] The quiz information includes the quiz text. In addition to the quiz text, the quiz information may also include other information, such as category information for the correct answer corresponding to the quiz, difficulty information for the quiz, and answer choice information. The quiz text is information in text format presented to the user, and the user uses this quiz text as a clue to consider the answer to the quiz. The quiz text may be the entirety of the quiz information or a part of it.

[0042] Retrieving the quiz text means obtaining the quiz text from the quiz text database 154 stored in the memory unit 15 of the server 10, according to the learning level of the user of the mobile terminal 20 that has entered the input to start the quiz session.

[0043] Here, the quiz text may be obtained not from a database 154 that has been pre-registered, but from a database registered on an external server, or furthermore, quiz text generated within server 10 through or in cooperation with an external server may be obtained.

[0044] The means of generating quiz text are not particularly limited. For example, quiz text may be obtained through a web search, or text and information may be extracted through a web search and generated based on them, or it may be automatically generated by a text generation system using natural language generation technology (hereinafter referred to as "generative AI"). Here, "generative AI" refers to a system that generates natural language text using natural language processing technologies, including not only large language models (LLMs), but also rule-based systems, statistical methods, machine learning models, etc. When generating quiz text, it is preferable that the quiz text be automatically generated by inputting (prompting) predetermined conditions (e.g., keywords, conversation history, browsing history, etc.) based on the user's learning level, interests, current events, or past answer history on the mobile device 20 into the generating AI. Furthermore, the generation method also includes accessing external information sources such as news sites, blogs, SNS posts, FAQ databases, online encyclopedias, and learning platforms on the web, and analyzing and extracting such information to construct the quiz text. This allows for the dynamic generation of optimized quiz text tailored to the user's proficiency level, interests, and the social context at that time, contributing to improved learning efficiency and increased motivation.

[0045] The generation AI may be executed within the server 10, or it may be a generation AI executed on an external server and used via a network NW.

[0046] As described above, when using a generative AI to generate quiz questions, the quality of the generated questions varies greatly depending on how the input is provided to the generative AI. A common method involves inputting conditions such as genre and difficulty level (prompts) to simultaneously generate pairs of questions and correct answers. However, this method requires the generative AI to simultaneously satisfy both the natural expression of the question and the factual accuracy of the correct answer, which can result in unnatural-sounding questions or questions where the correct answer is revealed within the question itself.

[0047] In contrast, the preferred method for generating quiz questions in this embodiment involves pre-determining a word or phrase (hereinafter referred to as "seed") that will be the correct answer to the quiz, and having the generating AI generate the quiz questions using the seed as input. That is, after determining the correct answer in advance, the generating AI is instructed to "create a question so that this word is the answer," and then generates the question. By doing so, the generating AI can focus solely on constructing a natural-sounding question, resulting in a more fluent and natural-sounding question. Furthermore, since the correct answer is already known, it becomes easier to verify afterward whether the correct answer is included in the generated question.

[0048] Furthermore, when selecting the aforementioned seeds, it is preferable to register and store multiple seeds in a database in advance, assigning attributes such as genre and difficulty level. When creating questions, a seed that meets predetermined conditions is selected from the database and used to generate the question text, thereby ensuring consistent question quality. Additionally, by recording the number of times each seed has been used and the date of its last use, and managing seeds that have been used more than a predetermined number of times as targets for discarding, it is possible to prevent the same correct answer from being repeatedly presented. In particular, for seeds belonging to genres that are timely and prone to becoming outdated, such as sports match results, video works, and music, it is preferable to apply different discard criteria (for example, discarding based on the number of days elapsed since the last use) than for regular seeds.

[0049] Furthermore, this seed-first generation method allows for the separation of quiz information generation and correct answer management processes, resulting in improved overall system maintainability and scalability. Additionally, by combining this with a configuration that generates contextual information from multiple perspectives for the same seed to generate quiz information, it becomes possible to systematically accumulate quiz information from diverse contexts for a single correct answer. This promotes effective vocabulary acquisition by repeatedly exposing users to the correct answer word in various contexts.

[0050] Each mobile terminal 20 that receives the quiz text displays the received quiz text (TA1005, TB1005). For example, in area AR04 of Figure 6(a), the quiz text "In 2024, scientists found a new kind of this sea animal. It has eight arms and lives very deep in the ocean." is displayed, and at the bottom of the screen (area AR05), "Read this text aloud." is displayed instructing the user to read (reproduce) this text, resulting in the display format shown in Figure 6(b).

[0051] Here, the quiz text may be displayed as a whole at once, or it may be displayed sequentially from the first word onwards, or any other display format is acceptable. For example, the color of the quiz text may be gradually faded in to make it visible. Alternatively, the text may be displayed intermittently, starting with one word at a time. However, since the user cannot begin the reading aloud until the first word is displayed, it is preferable to start accepting input of the audio information described later when the first word (or the first predetermined number of words) is displayed.

[0052] Then, at a timing corresponding to the display of the quiz text (when the entire quiz text is displayed, or, if the quiz text is displayed sequentially from the first word, when a word in a predetermined order (for example, the first word or the third word) is displayed), each mobile terminal 20 begins accepting input of voice information to read and reproduce this quiz text (TA2006, TB2006). When it is determined that a word has been reproduced based on the user's voice information input, it is preferable that the display mode for the word that has been determined to have been reproduced is changed to a different mode than that for the words that have not yet been reproduced, as shown in Figure 6(c).

[0053] Then, it is determined whether the user has reproduced a predetermined amount or a predetermined proportion of the quiz text (TA2007, TB2007). Reproducing a predetermined amount or a predetermined proportion of the audio information of the quiz text means that each word of the quiz text is reproduced by the input audio information, and the amount of the reproduced words, or the ratio of the number of reproduced words to the total number of words in the quiz text, reaches or exceeds a predetermined threshold. If a predetermined amount or a predetermined proportion of the quiz text is reproduced on the mobile terminal 20A (TA2007; Y), the answer to that quiz text becomes available for input (TA2008, TB2008). For example, as shown in Figure 6(c), words in the quiz text that have already been voice-inputted (depending on the language, this may be based on characters, syllables, etc., or any arbitrary unit) are displayed in a different manner than words that have not yet been voice-inputted. When a predetermined amount or percentage (in the example shown in Figure 6(c), the predetermined amount is 80%) of voice information in the quiz text is input, an input field for entering the answer, "Answer?: ", is displayed in area AR05. On the other hand, if a predetermined amount or proportion of the quiz text has not yet been reproduced (TA2007;N), the acquisition of audio information will continue. The same applies to the mobile terminal 20B.

[0054] Here, it is possible that the mobile terminal 20 can already input an answer before steps TA2008 and TB2008. In this case, it is preferable that in steps TA2008 and TB2008, the entered answer is not made available for input, but rather the entered answer is validated. That is, although the user has entered an answer on the mobile terminal 20, the system has not yet validated the answer, and the process of validating the answer (accepting the answer) is performed only after a predetermined amount or a predetermined percentage of the quiz text has been reproduced.

[0055] The response is provided by the user inputting character information into area AR05 via the input / output unit 23. The means of inputting character information can be arbitrary; for example, it may be done by user operation such as a software keyboard, touch input, or handwriting input, or it may be done by voice input, with the input voice being analyzed and the resulting character information being input. Furthermore, the method of answering is not limited to users directly inputting text information; it may also involve displaying options and allowing users to select from them by tapping or other actions.

[0056] Furthermore, informing the user that they can now enter an answer is not limited to displaying an input field for entering the answer. For example, the display of at least a part of the screen may be changed, such as by inverting the display of area AR04 where the quiz text is displayed. In addition, an audio message to that effect (for example, "You can now answer") may be output.

[0057] When input is received into the input field for the answer, the input content is sent to the server 10 (TA2009, TB2009). Figure 7(d) shows how it is displayed when the answer "octopus" is entered into the input field. Meanwhile, the server 10 accepts answers for a predetermined period (for example, 20 seconds) starting from the time corresponding to the start of the quiz session (S1010). This acceptance of answers is performed independently of the processing related to enabling answers on the mobile terminal 20.

[0058] Furthermore, when accepting responses after enabling responses, it is preferable to obtain not only the content of the response but also the time the response was submitted. By doing so, the speed of response input, calculated based on the time the response was entered, can also be included as a factor in the evaluation of the responses described later. More preferably, the time when the response was made available for a similar purpose is also obtained in conjunction with the response.

[0059] If an answer is entered from the mobile device 20 within the specified period, the process described in step S1011 is performed. On the other hand, if no answer is entered from the mobile device 20 within the specified period, the system is processed as if no answer was received, or as if information that had been pre-prepared as an incorrect answer was entered.

[0060] In other words, while server 10 accepts answers from the time corresponding to the start of the quiz session, mobile terminal 20 will not be able to answer unless a predetermined amount or a predetermined percentage of the quiz text is reproduced (e.g., answers will not be accepted, entered answers will not be valid, etc.). Therefore, users need to read the quiz text aloud as quickly as possible to make it possible to answer, and then think of and enter answers corresponding to the quiz text.

[0061] The time corresponding to the start of the quiz session, which is the trigger for the start of the predetermined period for accepting answers, can be any time related to an event that starts the quiz session. For example, if the quiz session starts at a set time (e.g., every 2 minutes), it could be the start time (absolute time), the time when the processing related to the start of the quiz session is completed, such as when the entire quiz text or the first word is displayed, or the time when the user presses a button related to the start of the quiz session.

[0062] The server 10 evaluates the answers entered from each mobile terminal 20 based on a comparison with the correct answer (for example, whether it is correct, whether it is close to the correct answer, etc.), and also evaluates how quickly the answer was submitted (for example, the elapsed time from the time corresponding to the start of the quiz session to the time the answer was received, or the elapsed time from the time the answer became available to the time the answer was received, etc.), and determines points as an incentive based on these factors (S1011). Figure 7(e) shows an example where the result of the server 10's evaluation of the answer (that it is correct) is output to the mobile terminal 20. The determined points are then awarded to the user of each mobile terminal 20 (S1012). Figure 7(f) shows an example where the points awarded to the user (+400 points) and the total points acquired (increased from 14400 points to 14800 points) are displayed, and a button to start a new session (an icon labeled "Next Quiz") is displayed, and the button is about to be pressed.

[0063] The following quiz text is an example of a quiz where partial credit is awarded for answers that are not the correct answer but are close to it. "Please provide the Japanese name for the country known as 'United Kingdom' in English." The correct answer is the official name, "The United Kingdom of Great Britain and Northern Ireland," but partial credit may be awarded for answers such as "United Kingdom." Alternatively, depending on the learning level, "United Kingdom" may be marked as incorrect and no partial credit may be awarded.

[0064] The evaluation criteria for determining this incentive may be an absolute evaluation based on the content and speed of the answers, a relative evaluation among the mobile devices 20 of the users who participated in that quiz session, or a relative evaluation among the mobile devices 20 including past quiz sessions in which the quiz text was presented. In the case of a relative evaluation within a single quiz session, for example, the incentive may be given only to the user who answers the fastest (or in order of speed) correct answers in a buzzer-style format.

[0065] In determining the incentive, it is also conceivable to further consider at least one of the following factors in addition to the comparison results and response speed. Note that in S1011 shown in Figure 4, the process of determining the incentive according to each user's response content and response speed is shown, but in determining the incentive, it is also conceivable to further consider at least one of the following factors in addition to these.

[0066] Firstly, one approach is to consider the method of inputting the response information. For example, since the nature of the processing required differs between responses entered via voice input and responses entered via text input, a configuration that provides a higher incentive for responses entered via voice input may be adopted.

[0067] Secondly, one approach is to consider the language of the text included in the presented information. For example, in a quiz session involving users learning multiple languages, fairness can be ensured among users learning different languages ​​by adjusting incentives according to differences in language difficulty and the amount of information provided.

[0068] Thirdly, one approach is to consider the difficulty level of the text included in the presented information. By providing higher incentives to users who reproduce and answer with more difficult text, it is expected that users will be motivated to challenge themselves with difficulty levels beyond their own learning level.

[0069] Fourthly, one approach is to consider the learning history of the user who entered the answer information. For example, this could involve providing a higher incentive to a user who has previously answered incorrectly for the same correct answer but answers correctly this time, or providing a higher incentive for the first correct answer. In this way, appropriate incentives are provided according to the user's learning progress, promoting the maintenance of continuous learning motivation.

[0070] Then, server 10 terminates the quiz session and sends a start confirmation notification to each mobile terminal 20 to initiate a new quiz session (S1013). Then, the processes TA2001, TB2001 and below are repeated. Of course, depending on user settings or system settings, it may not be necessary to perform the process of starting a new quiz session after an existing one has ended.

[0071] Furthermore, in order to manage information for each user, the server 10 stores a user information table 153, such as the one shown in Figure 8. The user information storage table 153 stores information such as each user's ID, username, and points earned, as well as information necessary for generating quiz text in quiz sessions, such as the language to be learned, the language the user is fluent in (for example, their native language), learning level, and areas of interest. For example, the content of the quiz text in a quiz session needs to be generated in the user's language to be learned, but it may be prepared independently of the user's attributes and interests, or it may be prepared based on the user's attributes and interests. For example, when a user whose area of ​​interest is movies participates in a quiz session, movie-related quiz text may be generated.

[0072] According to this embodiment, the system presents the user with a quiz text in the target language and prompts them to reproduce the quiz text and answer within a predetermined time. This requires the user to read the quiz text aloud, understand and analyze its content, formulate an answer, and input it within a limited time. This simultaneously involves understanding the target language, producing speech output, comprehending content, and processing thoughts in a short period, thereby increasing the load on the brain. As a result, it offers the advantageous effect of improving learning effectiveness.

[0073] <First variation (1)> In the first embodiment described above, language learning is conducted in a competitive manner by multiple users of mobile devices 20. However, the system is not limited to this, and a single user of a mobile device 20 may learn the target language in a single quiz session. In other words, during a quiz session, the server 10 may interact with only a single mobile device 20 and provide the user of the mobile device 20 with incentives based on the evaluation of their answers and response speed. Doing so will achieve the same effects as in the first embodiment.

[0074] <First variation (2)> In the first embodiment described above, only the quiz text was displayed on the mobile terminal 20. However, the invention is not limited to this, and in addition to the quiz text, the sound output unit 26 may also output an audio recording of the quiz text being read aloud. By outputting this audio recording, users can understand how to pronounce the quiz text, providing support for accurately mastering the pronunciation of the target language.

[0075] Preferably, when reading the audio, the mobile terminal 20 may also display a video of the person reading the audio. Such a video may be a video of the person actually reading the quiz text, or it may be a video generated by a generation AI. Figure 9(g) shows an example screen when a video of the person reading the audio corresponding to the quiz text is displayed. In this way, users can watch the movements of the person reading the audio (especially lip movements and facial expressions) and imitate the pronunciation of the person reading the audio, receiving support in accurately acquiring the pronunciation of the target language.

[0076] Furthermore, when the sound output unit 26 of the mobile terminal 20 outputs audio that reads out the quiz text, it is preferable that the audio output speed is slower than the normally expected speed, or that the reading begins after a predetermined time has elapsed (for example, 3 seconds) after the quiz text is displayed.

[0077] Furthermore, the reproduction of words contained in the quiz text may be based solely on voice information input by the user from the mobile device 20, or it may be considered that the reproduction of words contained in the quiz text is performed based on the timing of the voice output of each word performed by the mobile device 20. In other words, if the user reproduces part of the quiz text but stops at a word whose pronunciation they do not know, the reproduction of that word may be recognized when the mobile device 20 reaches and finishes reading that word. The user can then continue to reproduce the remaining words and will be able to complete the reproduction of a predetermined amount or a predetermined percentage of the quiz text.

[0078] In this way, if the user can reproduce the quiz text themselves, they can do so without waiting for the mobile device 20 to read it aloud, allowing them to input their answer to the quiz text more quickly. On the other hand, if there are words that the user has difficulty reproducing (reading, correct pronunciation, etc.), the mobile device 20 will read those words aloud for them, and continue the reproduction process, allowing the user to input their answer to the quiz text. Therefore, this prevents the frustration of being unable to answer the quiz text because the user understands the content but has difficulty reading it aloud, and also promotes learning of words whose reading is unknown.

[0079] <First variation (3)> In the first embodiment described above, when it became possible to answer, an input field for entering the answer was displayed, allowing the user to enter the answer. However, this input may be performed in various forms. For example, this may be done by tapping the input / output unit 23. Alternatively, the answer may be given by voice input. In such a case, as shown in Figure 9(h), if a predetermined amount or percentage of the quiz text has not been reproduced and the answer is not yet possible, an icon with a diagonal line through the microphone will be displayed in area AR05 to indicate that voice input is not possible. Then, as shown in Figure 9(i), when the answer becomes possible, an icon with no diagonal line through the microphone will be displayed to indicate that voice input is possible. When voice input is performed, the mobile terminal 20 (or server 10, etc.) performs voice recognition processing to determine what kind of answer has been entered.

[0080] <First variation (4)> In the first embodiment described above, after the user's answer is entered into the mobile terminal 20, the system may wait for a predetermined amount of time to elapse before evaluating the entered answer, or it may evaluate the answer immediately after it is entered. For example, in the latter case, if the answer is not correct, the system may output a message to that effect to the mobile terminal 20 (for example, displayed on the display unit 24), and if it is within a predetermined period or within a predetermined number of re-answers, the system may give the user an opportunity to re-answer, or it may not be necessary to give an opportunity to re-answer. Furthermore, if the correct answer is not entered within a predetermined period, it is preferable to output information regarding the correct answer to the mobile terminal 20 and notify the users participating in the quiz session.

[0081] <First variation (5)> In the first embodiment described above, it was explained that a response is made possible when a predetermined amount or a predetermined percentage of voice information is reproduced. However, the threshold corresponding to this predetermined amount or predetermined percentage may be fixed (for example, always 80%), or it may be variable for each quiz session or for each group of quiz sessions, depending on user attributes such as learning level, the content of the quiz text, and the user's choice. For example, it is preferable to set a lower threshold for users with a low learning level, and it may also be possible to set a higher threshold depending on the user's choice. In particular, when the threshold is set high, it is preferable to set it so that the incentive given in response is large.

[0082] <Second Embodiment> In the first embodiment, it was described that the quiz session was conducted in a form in which each mobile terminal 20 competed against each other. However, the second embodiment is characterized in that the quiz session was conducted in a form in which multiple mobile terminals 20 cooperated with each other.

[0083] Figure 10 is a flowchart of the language learning support process in the second embodiment. In the language learning support process in the second embodiment, mobile terminal 20C and mobile terminal 20D cooperate to respond to the quiz text. Mobile terminals 20C and 20D may have the same configuration as mobile terminal 20.

[0084] The following describes the differences from the first embodiment. In the second embodiment, processing is carried out in the same manner as in the first embodiment, with the mobile terminal 20 and the server 10 communicating with each other. However, the difference from the first embodiment is that in the first embodiment, mobile terminals 20A and 20B are treated equally (corresponding users A and B compete), whereas in the second embodiment, user C of mobile terminal 20C supports user D of mobile terminal 20D. For the sake of explanation, in the second embodiment, only mobile terminal 20C (user C) supports mobile terminal 20D, but the number of supported mobile terminals 20 is not particularly limited and may be multiple. Similarly, the number of supported mobile terminals 20 is not particularly limited and may be multiple. Furthermore, in the following, user C of mobile terminal 20C will be referred to as the auxiliary input user, and user D of mobile terminal 20D will be referred to as the respondent user.

[0085] First, when a new quiz session is started, the server 10 sends the acquired or generated quiz text to the mobile terminals 20C and 20D (S1004).

[0086] In the mobile terminals 20C and 20D, the received quiz text is displayed (TC2005, TD2005), similar to the first embodiment. Furthermore, in the mobile terminal 20D, voice information input is acquired from the respondent user who reads aloud (reproduces) the displayed quiz text (TD2006).

[0087] On the other hand, in the mobile terminal 20C, the system acquires character information input from an auxiliary input user to input words contained in the displayed quiz text. Here, the input of text information by an auxiliary input user in the mobile terminal 20C may be via voice input or via non-voice user operation (software keyboard, touch input, handwriting, etc.). In this embodiment, text information is used as a comprehensive concept that includes both voice input (including voice input being recognized as voice and consequently becoming text information) and non-voice user operation input. Furthermore, in the following description, the input of words included in the quiz text in the mobile terminal 20C will be described as non-voice user operation input.

[0088] Figure 11 shows an example of a screen displayed on the mobile terminal 20C. After the quiz session begins, as shown in Figure 11(j), user C can input words from the displayed quiz text on mobile terminal 20C. Meanwhile, on mobile terminal 20D, voice input is performed to reproduce the displayed quiz text, similar to the first embodiment, and it is determined that the reproduction of the words read aloud is complete. The second embodiment is characterized in that the users of each of these mobile terminals 20 can collaborate to reproduce the words contained in the quiz text.

[0089] Since input regarding the quiz text is performed jointly by mobile terminal 20C and mobile terminal 20D, it is possible that mobile terminal 20C may reproduce one word first, while mobile terminal 20D may reproduce another word first. In this case, as shown in the example in Figure 11(k), words that have already been reproduced (words reproduced by mobile terminal 20C or mobile terminal 20D) are displayed in a different format than words that have not yet been reproduced. Furthermore, words reproduced by the auxiliary input user via mobile terminal 20C are underlined, making them even more distinguishable from words reproduced by the respondent user via mobile terminal 20D.

[0090] Furthermore, for words that are reproduced on mobile terminal 20C before those on mobile terminal 20D, an incentive is provided to the auxiliary input user of mobile terminal 20C according to the reproduction (TC2013). In the example shown in Figure 11(k), the score increased from 3900 points to 3920 points (an increase of 10 points for each word input) in response to the earlier reproduction of the two words "scientists" and "animal". Furthermore, the provision of incentives to auxiliary input users on the mobile terminal 20C may be done all at once after step TCD2007, as described later, or it may be done before step TCD2007, in an appropriate manner each time a word eligible for incentives is reproduced.

[0091] In this way, each mobile terminal 20 inputs text information corresponding to the displayed quiz text, and it is determined whether a predetermined amount or percentage of the quiz text has been reproduced by combining the inputs from mobile terminals 20C and 20D (TCD2007). When the reproduction of the predetermined amount or percentage is complete (TCD2007; Y), mobile terminal 20D becomes able to input the answer to the quiz text, similar to the first embodiment (TD2008). Meanwhile, mobile terminal 20C displays a message indicating that the input is complete, and processing for that quiz session on mobile terminal 20C ends. If the reproduction of a predetermined amount or percentage has not yet been completed (TCD2007;N), both mobile terminal 20C and mobile terminal 20D return to steps TC2006 and TD2006, respectively, and the acceptance of input for reproduction continues.

[0092] Subsequently, similar to the first embodiment, the server 10 accepts responses for a predetermined period (S1010), while the mobile terminal 20D acquires the responses and transmits them to the server 10 (TD2009). The server 10 then determines points based on the evaluation of each respondent user's response content and response speed (S1011). The determined points are then awarded to each respondent user (S1012).

[0093] Furthermore, in the mobile terminal 20C, an output indicating that input is complete is issued (TC2014) based on the determination that the predetermined amount or predetermined percentage of step TCD2007 has been reproduced. In other words, the mobile terminal 20C may not be required to input an answer corresponding to the quiz text.

[0094] According to this embodiment, the respondent user is shown a quiz text in the language to be learned, and is required to reproduce the quiz text and provide an answer within a predetermined time. Therefore, the respondent user needs to read the quiz text aloud, understand its content, analyze it, and output an answer within a limited time. In this way, a significant load can be placed on the brain regarding the processing of the language to be learned, and as a result, the learning efficiency of the respondent user can be improved, similar to the first embodiment.

[0095] Furthermore, since the assisting input user helps the responding user reproduce the presented quiz text, the assisting input user can also improve their own learning efficiency in the target language. In particular, even beginners with limited knowledge of the target language, children in countries where the target language is spoken, and individuals with intellectual disabilities, who do not have a sufficient understanding of the target language, can contribute to other users participating in the quiz session.

[0096] For example, it is possible to implement a format where a team consisting of a respondent user and an input assistant user competes against other individuals or teams. In particular, if there are words that the respondent user has difficulty reproducing, the input assistant user can input such words, enabling cooperative play. Through the efforts of the input assistant user, their team can contribute to scoring more points than other teams. In other words, the input assistant user can participate in the quiz session with high motivation and engage in language learning.

[0097] Furthermore, the respondent user may have multiple mobile devices 20 (first group of devices). Similarly, the auxiliary input user may also have multiple mobile devices 20 (second group of devices). In other words, the first group of devices and the second group of devices may be formed by at least one mobile device 20 each, and these two groups may work together as a team to perform these processes and conduct language learning.

[0098] Specifically, the quiz text is reproduced using audio information on the mobile terminals 20 of the respondent users constituting the first terminal group, and the quiz text is reproduced using text information on the mobile terminals 20 of the auxiliary input users constituting the second terminal group. A predetermined amount or percentage of the reproduction is determined based on both this audio and text information.

[0099] Furthermore, in this configuration, the reproduction by the auxiliary input user on the mobile terminal 20C is not limited to performing the reproduction sequentially from the first word of the quiz text, but may be performed on any word in the quiz text. For example, if there is a word in the quiz text that is difficult to pronounce, the auxiliary input user can support the respondent user by inputting that word before the respondent user. Alternatively, information about such words or the fact that they are difficult to pronounce may be registered in advance and linked to the quiz text, and when the quiz text is displayed on the mobile terminal 20C, such words may be treated with priority over other words (changing the color, increasing the font size, giving higher points, etc.) to encourage the auxiliary input user to input them preferentially.

[0100] Furthermore, when a respondent user sees a quiz text and discovers a word whose pronunciation they don't know, they can select that word by tapping it on the mobile device 20D. The selected word is then sent to the mobile device 20C and displayed preferentially, prompting the assistive input user to input it.

[0101] <Third Embodiment> In the first embodiment described above, a form was described in which multiple users compete to learn a language (for example, English) in a single quiz session. However, the third embodiment is a form in which multiple users compete to learn multiple languages ​​in a single quiz session.

[0102] The processing flow in the third embodiment is almost the same as the processing flow in the first embodiment (for example, the flowchart shown in Figure 4). The difference is that, in each mobile terminal 20, if the language to be learned by the corresponding user is different, the display and input are provided in a form translated into the respective language to be learned.

[0103] Figure 12 shows an example screen display when there are three mobile devices 20 participating in a single quiz session, each learning a different language.

[0104] Figure 12(l) shows an example screen when the language to be learned is English. In this embodiment, each mobile device 20 can see the status of other users sharing the same quiz session (whether they can answer, what answers have been entered, etc.). In addition, although input from other users is displayed in English on the user's mobile device 20, this does not necessarily correspond to the actual input content, but rather indicates that words entered in the language to be learned on another user's mobile device 20 are translated and displayed in English on the user's mobile device 20. Specifically, Figure 12(l) shows that the language being learned is English, the quiz text and other materials are presented in English, and the user has read aloud up to the word "animal" in the quiz text, but has not yet been able to answer (the input field corresponding to "You" is shaded). On the other hand, it is shown that other users B and C have entered "jellyfish" and "octopus" as their answers, respectively.

[0105] Figure 12(m) shows an example screen when the target language for learning is Spanish. "LEVEL" and "SCORE" have been replaced with their corresponding Spanish words, "NIVEL" and "PUNTOS," respectively, and the quiz text presented is the same as in Figure 12(l), but in Spanish. Furthermore, it is shown that user A has already entered the answer "medusa" (in the input field corresponding to "t?"), while another user A has not yet been able to answer, and yet another user C has entered the answer "pulpo."

[0106] Figure 12(n) shows an example screen when the target language for learning is Japanese. "LEVEL" and "SCORE" have been replaced with the corresponding Japanese words "level" and "score," respectively, and the quiz text presented is the same as in Figure 12(l), but in Japanese. Furthermore, it is shown that user A has already entered the answer "octopus" (in the input field corresponding to "you"), while another user A has not yet been able to answer, and yet another user B has entered the answer "jellyfish."

[0107] In this way, even users learning different languages ​​can participate in quiz sessions with similar quiz texts in each language, achieving the same effect as in the first embodiment.

[0108] Furthermore, when preparing such quiz texts, differences may arise in the amount of information and workload that users must reproduce or understand through speech, etc., due to the nature of the languages ​​themselves (for example, "speed limit" in English becomes the lengthy "Geschwindigkeitsbegrenzung" in German, and it is likely that reproducing or answering a quiz text with similar content will take more time in German than in English). In light of these differences, it is preferable to keep the content of the quiz texts the same, but to apply a handicap (such as changing the incentive depending on the language, changing a predetermined amount or percentage to make it possible to answer, etc.) according to the amount of information (number of characters, number of syllables, etc.) in the quiz texts and answers translated into the target language, or to adjust the quiz texts so that they have a similar amount of information when preparing or translating them.

[0109] <Fourth Embodiment> In the first and second embodiments described above, we explained the forms in which matches are played with individual users and a pair of responding users and auxiliary input users, respectively. However, the forms in which matches are played are not limited to these. In the fourth embodiment, multiple users form teams and cooperate to play against other teams.

[0110] Figure 13 shows an example of the screen displayed on a mobile device 20 when a quiz session is conducted in which teams composed of multiple users compete against each other. It is assumed that a team consists of four users, whose usernames are You, D, E, and F.

[0111] The basic processing flow is the same as in the first embodiment, but the differences are explained below. Server 10 prepares (acquires or generates) quiz text and sends it to each mobile terminal 20. Upon receiving the quiz text, the mobile terminal 20 displays the received quiz text. Here, You and E are only responsible for inputting the answers corresponding to the quiz text, while F is assigned the role of both providing the answers corresponding to the quiz text and reproducing the displayed quiz text.

[0112] F reproduces the displayed quiz text, but even if the amount or percentage reproduced has not reached a predetermined amount or percentage, You and E can input the answer corresponding to the quiz text on their respective mobile devices 20. In other words, in the fourth embodiment, the completion of the reproduction of a predetermined amount or percentage of the quiz text does not correspond to the enabling of answer input on the mobile device 20. However, the evaluation of whether an entered answer is correct or not is only performed after it is determined that F has completed a predetermined amount or percentage of reproducing the quiz text. For example, as shown in Figure 13(o), since F has not yet reached the predetermined amount or percentage of reproducing the quiz text, the input fields for You, D, and E's answers are shaded. However, You and D have already entered their answers. However, You's input is "no idea...", indicating that they do not know the correct answer to the quiz and have communicated this to the other users.

[0113] In this embodiment, the system is configured so that answers from other users on the same team are visible on each mobile device 20. However, it is not limited to this configuration, and the system may be configured so that answers from other users are not visible, or so that only the answers of a specific user (for example, a user who is responsible for both reproduction and answering) are visible.

[0114] Then, when F determines that a predetermined amount or percentage of the quiz text has been reproduced, the already entered answers are evaluated. In the example screen shown in Figure 13(p), You and E have entered some answers, but both are incorrect, so an icon including an "X" mark indicating incorrectness is displayed to the right of the input field.

[0115] Here, if Server 10 is still accepting answers within the specified period, each user responsible for answering can either enter a new answer or modify their answer to aim for the correct answer. Figure 13(q) shows an example screen illustrating the process when an answer is changed. F enters the new answer "scallop," and E changes their original answer "shark" to "octopus." Since the correct answer is "octopus," an icon including a circle (○) indicating that the answer is correct is displayed to the right of the input field for E's answer, which has been changed to the correct answer. In contrast, F's answer is incorrect, so an icon including an "x" (×) is displayed to the right of the input field.

[0116] Thus, in the fourth embodiment, the tasks of reading aloud and answering are divided within the team, allowing all necessary processing to be carried out collaboratively. Through this mechanism, the learning efficiency of each user regarding the target language can be improved. In particular, from user F's perspective, while they are solely responsible for reproducing the quiz text, multiple people, including F, can answer the quiz questions. This allows for learning without becoming overly burdensome, while still leaving room for individual answering.

[0117] For the sake of explanation, it has been stated that there is one user responsible for reproducing the quiz text, but the number of users responsible for reproduction is not limited to one; there may be multiple users. For example, if there are two users responsible for reproduction, the goal is to complete a predetermined amount or percentage of reproduction by either of the two users reproducing different parts of the quiz text, or by having the two users reproduce overlapping parts (if it is determined that correct audio information has been input from either of the users for a word in the quiz text, that word may be considered reproduced).

[0118] Furthermore, users responsible for reproduction may be required not to answer the quiz text. For example, they may not be allowed to enter answers after completing a predetermined amount or percentage of the reproduction.

[0119] Furthermore, when conducting multiple quiz sessions, roles may be fixed for each session, or roles may be changed (for example, by rotation) for each session. Also, when different users are teamed up in each of the multiple quiz sessions, depending on the user's choice, one person may always be responsible for both reproduction and answering, or one person may always be responsible for answering only.

[0120] Furthermore, it is preferable that the audio reproducing the quiz text provided by the user is output from the sound output unit 26 of another user's mobile device 20 via the network. In addition, the voices of users who are not responsible for reproduction may also be output from the sound output unit 26 of another user's mobile device 20. In this way, users can recognize areas where they tend to make mistakes by referring to the audio provided by their study partners (teammates), and they can participate in quiz sessions by helping each other in the same target language, thus enabling them to learn the language with high motivation.

[0121] This allows for control over the process, for example, enabling users who have difficulty reading aloud to focus on answering questions during a quiz session, or allowing them to take on the role of reading aloud to overcome their weakness and concentrate on that skill. Furthermore, each user may take on roles equally according to their learning level, or they may change roles according to their learning level (for example, a user with a higher learning level may be responsible for both reading aloud and answering) during the quiz session.

[0122] <Fifth Embodiment> In the first to fourth embodiments described above, the condition for answering or evaluating an answer was to complete the reproduction of a predetermined amount or a predetermined percentage of the presented quiz text. However, the fifth embodiment is characterized by gradually displaying the quiz text and also outputting an audio recording of the quiz text being read aloud.

[0123] Figure 14 shows an example of a screen displayed on the mobile terminal 20 in the fifth embodiment. In Figure 14(r), the quiz session has just begun, and only the first part of the quiz text is displayed. The bottom of the screen indicates that voice input for answering is not yet permitted. As time progresses, the screen transitions to Figure 14(s), where a certain amount of quiz text is displayed, and the bottom of the screen indicates that voice input for answering is permitted. After further time has passed, Figure 14(t) shows that the entire quiz text is displayed, the user has entered voice information for their answer, and an output indicating that the answer is correct is displayed.

[0124] Here, the speed at which the quiz text is gradually displayed is slow, assuming a reading speed for beginners. For example, if the target language for learning is English, it is preferable to have a speed of approximately 50-100 wpm (words per minute). Furthermore, it is preferable that the sound output unit 26 of the mobile terminal 20 reads aloud each word displayed in the quiz text in conjunction with its display.

[0125] Furthermore, it is preferable that the audio output of the quiz text from the sound output unit 26 of the mobile terminal 20 is faster than the speed at which the quiz text is gradually displayed on the display unit 24. In other words, although the same quiz text is output from both the display unit 24 and the sound output unit 26, they are not synchronized, and the audio output is output at a faster speed.

[0126] Alternatively, the speed at which the quiz text is displayed may be approximately the same as the speed at which the audio output of the quiz text is output from the sound output unit 26. However, in this case, it is preferable that the audio output of the quiz text from the sound output unit 26 precedes the display of the quiz text by a predetermined time (for example, about 3 seconds). Note that the audio output of the quiz text preceding the display may mean that the start of the audio output precedes the start of the display, or that initially the audio output and display are simultaneous or the display output precedes the audio output, but at some point the audio output takes precedence. For example, if the quiz text contains difficult words, the presence of such words may be registered in advance, and until such words are presented, the display output will precede the audio output of those words, and then the display will follow.

[0127] In this way, users can understand the content of the quiz from the audio output of the quiz text, but if they have difficulty hearing, they can understand the content of the quiz again from the display output and earn points by entering the correct answer or an answer close to the correct answer. Furthermore, it is desirable to structure the system so that users receive fewer points for delayed responses compared to those who respond early. This will motivate users to strive for quick responses to earn more points, encouraging them to concentrate on the listening comprehension and enabling more effective language learning.

[0128] Furthermore, the audio output of the quiz text by the mobile device 20 may be limited to a means that only the user can hear the output, such as earphones, rather than using a sound output means that anyone nearby can hear, such as a speaker on the mobile device 20.

[0129] Furthermore, when answering, participants may be allowed to answer once a predetermined amount or percentage of the quiz text is displayed, or they may be allowed to answer without waiting for the predetermined amount or percentage to be displayed. In other words, they may be allowed to answer from the beginning of the quiz session, or they may be allowed to answer after a predetermined time (for example, 3 seconds) has elapsed since the start of the quiz session.

[0130] Furthermore, when answering, responses may be accepted in a buzzer-like manner. Specifically, a buzzer button may be provided on the screen of the mobile device 20, and when the button is pressed, the audio output and display output of the quiz text may be temporarily stopped to accept the answer. If the answer at that time is incorrect, the audio output and display output may be resumed. Alternatively, when the buzzer button is pressed, only the display output may be stopped, and the audio output may continue to the end. In other words, the user may be provided with information about the entire quiz text through audio output, while the correct word may not be visually confirmed by the display output.

[0131] Thus, instead of the user reproducing the quiz text as in the first to fourth embodiments, the system (mobile terminal 20, etc.) outputs the quiz text aloud, while the quiz text is displayed at a slower speed or at a later timing, gradually displaying it. This provides the learner (user) with an environment where they can concentrate on listening. Furthermore, even if they miss a word during listening and are unable to answer the quiz at that point, they can later identify that missed word through the display of the quiz text and answer the quiz, albeit at a later timing. In other words, it also has the effect of allowing learners to recognize words that are difficult to hear while continuing their learning.

[0132] As a variation, the quiz text may be output with an accent (dialect, etc.) specific to the country or region where the target language is spoken. For example, by outputting audio of English with a New Zealand accent, users will need to concentrate on listening to the New Zealand accented English, while simultaneously being able to understand words that were difficult to hear by referring to the text displayed afterward. This allows users to become familiar with the dialect and accent of the language, which is effective if they are planning to emigrate to or study in a country or region with that accent.

[0133] <Sixth Embodiment> In the first to fifth embodiments described above, the content of the presented quiz text was the same despite differences in language. However, the sixth embodiment is characterized in that, based on some trigger or condition, the content of the quiz text differs between the mobile terminals 20, although the correct answer remains the same.

[0134] Figure 15 shows an example of a screen displayed on the mobile terminal 20 in the sixth embodiment. Figure 15(u) shows an example screen displayed on a mobile device 20 associated with a user whose target language of study is English and whose learning level is 4 (assuming an intermediate level). Unlike the example screen shown in Figure 5, for example, the quiz text in Figure 15(u) is "Known for clever problem-solving and camouflage, this ocean dweller can slip through narrow gaps with eight limbs." The correct answer corresponding to this quiz text is "octopus," just like in the example screen shown in Figure 5. In other words, the correct answer is maintained the same while the quiz text is prepared and displayed for the intermediate level, depending on the learning level. The same applies to Figure 15(v), where the quiz text uses more advanced vocabulary: "Highly intelligent and solitary, this cephalopod displays extraordinary camouflage and puzzle-solving skills with eight agile arms," ​​but the correct answer is the same as "octopus." This displays quiz text prepared for advanced learners, based on the user's learning level being 7 (which is considered an advanced level).

[0135] By answering the quiz text displayed in this way, users can earn corresponding points or other incentives. Furthermore, when answering, as in the first embodiment, the user may be required to reproduce a predetermined amount or percentage of the quiz text via voice input before answering, or this may not be required.

[0136] This configuration can also be used as a handicap, especially when users compete against each other. For example, when users with different learning levels compete against each other, while keeping the correct answer the same, each mobile device 20 can display and input different quiz texts according to their learning level. This makes it appear as if each user is thinking about and inputting an answer to the same quiz text, thus mitigating the advantage of users with higher learning levels and enabling competitive quizzes.

[0137] Furthermore, this approach eliminates the need to gather users with the same learning level into the same quiz session; users with different learning levels can participate in and compete in the same quiz session, making it less likely for situations to arise where there aren't enough users for competition to take place. Moreover, based on the third embodiment, users of any learning level who are learning other languages ​​can also participate in the same quiz session.

[0138] Furthermore, the configuration of presenting quiz texts with the same correct answer but different content on multiple devices has a unique effect, independent of the feature of reproducing the quiz text: by presenting multiple users with questions of different difficulty levels, languages, and content for the same correct answer, it allows users to participate in the same quiz session regardless of their attributes or learning level. This enables diverse users with different target languages ​​and learning levels to share a single quiz session, relaxing the requirements for a quiz session to be established and is expected to improve learning motivation through competition and interaction among diverse users.

[0139] Furthermore, the first text included in the first presented information and the second text included in the second presented information only need to have the same correct quiz answer but different content, and the nature of the difference is not particularly limited. In addition to the differences in language and difficulty level mentioned above, for example, the texts could be written from different perspectives or contexts regarding the same correct answer (for example, presenting a quiz text based on historical background to one terminal and a quiz text based on a scientific perspective to another terminal for the same correct answer), have different lengths, use different types of vocabulary, or have different writing styles. Any difference in expression, content, or format while sharing the same correct quiz answer is acceptable. In this way, multiple users can work on the same correct quiz answer from diverse perspectives simultaneously, thereby increasing the diversity of the quiz session and the learning effect.

[0140] Furthermore, the content of the displayed quiz text is not limited to the learning level; it may also be changed based on arbitrary conditions or triggers. For example, users could be allowed to select a more difficult level of text if they wish, or users with a high accuracy rate, such as those who have answered correctly consecutively, could be provided with a more difficult quiz text regardless of their learning level (and vice versa). Users who answer quickly after the quiz text is displayed could also be provided with a more difficult quiz text regardless of their learning level (and vice versa). Users who frequently use the non-visual hint function (for example, displaying quiz hints by tapping a button on the screen) could be provided with a less difficult quiz text regardless of their learning level (and vice versa). Alternatively, the difficulty level could simply be changed randomly for each quiz session.

[0141] The difficulty level of the quiz text can be changed for each user individually, or it can be changed for the entire quiz session. Furthermore, if the quiz text is of a high difficulty level, a greater incentive may be provided compared to when the difficulty level is low.

[0142] In this way, by changing the difficulty level of the quiz text while keeping the correct answer the same based on certain conditions or triggers, users can learn at a quiz level that suits their situation and preferences. In particular, in a competitive format between users, although it appears that they are competing based on the same quiz text content, the competition is actually more competitive because handicaps are imposed according to their learning level, etc. Furthermore, it does not necessarily require gathering users with the same or similar learning level, or even the same target language, thus making it easier to establish a quiz session.

[0143] <Seventh Embodiment> The seventh embodiment is characterized by providing a function to support the user when there are words that are difficult for the user during the process of inputting a predetermined amount or a predetermined percentage of the presented quiz text into the mobile terminal 20 using voice information provided by the user.

[0144] Figure 16 shows an example of a screen displayed on the mobile terminal 20 in the seventh embodiment. Figure 16(w) shows an example screen displaying the received quiz text on the display unit 24 of a mobile terminal 20 that has received the quiz text from the server 10. After the quiz text is displayed, the user begins to reproduce it by inputting voice information, and Figure 16(x) shows the state after the user has read aloud up to the word "cause".

[0145] The screen shown in Figure 16(x) simulates a situation where, with the quiz text displayed, the user is unable to correctly pronounce the next word, "ionization," and the reading for reproduction stops. If the quiz text is to be read sequentially, the reading of words after "ionization" cannot be performed unless the user correctly pronounces "ionization," resulting in a situation where the reproduction of a predetermined amount or percentage of the quiz text cannot be completed.

[0146] In such cases, if a user performs a predetermined operation (for example, pressing a “HINT” button (not shown), tapping any part of the screen, tapping a part of a word that cannot be pronounced correctly (in this case, “ionization”), or inputting a specific sound (for example, “I don’t know”), or if a sound with a commonly mispronounced pronunciation of the target word is input a predetermined number of times, or if a predetermined time (for example, 3 seconds) elapses without the correct pronunciation being input, then information about the word (preferably information about pronunciation such as phonetic symbols, but not limited to this; it may also be semantic information or synonym information) is displayed on the display unit 24 as shown in Figure 16(y) to support the user in understanding the word. Alternatively, the correct pronunciation of the target word may be output from the sound output unit 26. In this case, if the user fails to input the correct pronunciation of the word for an extended period, or if a predetermined operation is performed by the user, it is preferable to output the correct pronunciation of the target word again from the sound output unit 26.

[0147] Alternatively, if a predetermined time (e.g., 5 seconds) elapses without the correct pronunciation being input, or if a predetermined number of incorrect pronunciations of the target word are input after the above-mentioned information about the target word has been output, or if there is voice input indicating that the user has given up on reproducing the word (e.g., "skip"), or if the user performs a predetermined operation, the system may consider the word to have been correctly reproduced, or skip the reproduction of that word, and accept voice input for reproducing subsequent words.

[0148] In this way, users can complete the reproduction of a predetermined amount or percentage of the quiz text and be able to answer the quiz text. That is, in the short time allotted for a quiz session, users are required to read aloud and answer as quickly as possible, while simultaneously grasping the correct pronunciation of difficult words, reading the quiz text correctly, understanding the quiz content, and entering their answers. This keeps the brains of foreign language learners moderately "busy," resulting in a learning effect that deepens memory and understanding.

[0149] <Eighth Embodiment> The eighth embodiment relates to a language learning support system that presents a hint to the user, prompts the user to input the correct string corresponding to the hint, and enables answer input only after the completion of the string input.

[0150] The language learning support system 1 in this embodiment has the same system configuration as the first embodiment (Figure 1), and the server 10 and the mobile terminal 20 work together. The server 10 acquires hints and the correct answer strings corresponding to the hints as quiz information. The hints are information that serves as clues for the user to recognize and recall the correct answer strings, and their type is not particularly limited. In acquiring quiz information, as in the first embodiment, in addition to acquiring it from a database, a method can be adopted in which a generating AI generates quiz information using the correct answer strings as a seed.

[0151] The following are examples of specific forms of hints: Firstly, each character of the correct string or its outline (hollow characters where the shape of the character is represented only by the outline, with the inside of the character being blank) is displayed, and the user inputs the character by tracing the character or outline. Secondly, an example of the correct string is displayed next to the answer field, and the user writes the character in the answer field while looking at the example. Thirdly, a picture or photograph corresponding to the correct string is displayed, and the user recalls the correct string from the picture or photograph and writes the character in the answer field. Fourthly, a question text related to the correct string is displayed, and the user writes the character in the answer field by referring to the question text. Fifthly, an audio recording of the correct string being read aloud or an audio recording related to the correct string is presented to the user, and the user writes the character in the answer field using the audio as a clue. These forms may be used individually or in combination. In other words, the user should input the corresponding string in response to some information (hint) about a string that has been output.

[0152] The means of character input are not particularly limited, and any input method can be used, such as touch operation via the touch panel of the mobile terminal 20, handwriting input using a stylus, or input using a software keyboard. Here, the tracing input and the handwriting input differ in nature. In tracing input, the user traces the shape of the characters or outline of the correct answer string, which are displayed as hints, whereas in handwriting input, the user autonomously recalls and writes the correct answer string, and the latter requires a higher level of proficiency. In the embodiment in which characters are handwritten in the answer field, the input handwritten characters can be recognized by a handwriting recognition means (optical character recognition module or machine learning model, etc.), and the correctness can be determined by comparing the recognized string with the correct answer string.

[0153] The control unit 21 determines whether the completion rate of character input has reached a predetermined threshold. Here, the completion rate of character input is quantified by, for example, the percentage of characters that have been traced out of all characters, or the percentage of the number of characters entered in the answer field relative to the number of characters in the correct answer string. When the completion rate reaches or exceeds a predetermined threshold (for example, a predetermined percentage or a predetermined amount), it is determined that a predetermined condition has been met, and the system proceeds to the answer phase. This predetermined condition corresponds to "reproduction of a predetermined amount or predetermined percentage of the quiz text" in the first embodiment, and in this embodiment, character input constitutes the reproduction means. The means of answering are not particularly limited and can be any means such as voice input, tap input, keyboard input, sign language input, etc.

[0154] When predetermined conditions are met, the mobile terminal 20 presents the user with instructions prompting them to provide the correct string of characters. The method of presenting these instructions is not particularly limited and can be any manner that the user can recognize as prompting them to respond, such as displaying text such as "Try saying this word out loud!" on the display unit 24, outputting the same content as speech from the sound output unit 26 using speech synthesis, or displaying a person giving the instructions in sign language on the display unit 24. The control unit 21 (or server 10) performs recognition processing on the user's response input and determines whether it is correct or incorrect by comparing the recognition result with the correct string of characters. In the case of a response provided by voice input, it is preferable to set the criteria for voice recognition loosely according to the user's level of proficiency, similar to the modified example in the first embodiment.

[0155] If the answer is determined to be correct, the server 10 awards the user a predetermined number of points. The points awarded may be weighted based on factors such as the accuracy of the character input (including the degree of agreement between the tracing path and the outline, the accuracy of the stroke order, etc.), the speed of answering, and the accuracy rate. If the answer is incorrect, the server may provide additional hints to encourage the user to try again, and after a predetermined number of incorrect answers, the correct answer may be presented.

[0156] Furthermore, the types of hints and the format of text input can be changed in stages according to the user's proficiency level. For example, in the initial stages, a format with many guides, such as tracing hollow letters, may be adopted, and as proficiency improves, the difficulty level can be gradually increased by transitioning to a format that presents only pictures or photographs, or only audio. Also, similar to the second to fourth embodiments, quiz sessions can be conducted in a form in which multiple users cooperate or compete.

[0157] Furthermore, as a variation of the eighth embodiment, the text included in the presented information is written in a language different from the language the user is learning (for example, their native language or English, a language the user has already acquired), and the user is asked to read and reproduce the text aloud, while simultaneously being asked to input a string of characters in the language the user is learning as the answer information via handwriting input on the touch panel. For example, for a user learning Russian, the presented information is displayed in English (for example, "What is the Russian word for apple?"), the user reads and reproduces the text aloud, and then inputs "яблоко" as the answer via handwriting input on the touch panel. With this configuration, since the text to be reproduced is in a language different from the language the user is learning, even beginners who have not yet acquired the writing system can reproduce it. On the other hand, by having the user handwrite input characters in the language the user is learning as the answer, the shape of the characters, stroke order, and the correspondence between words and meanings can be trained simultaneously, resulting in a high learning effect, especially for users in the early stages of acquiring the writing system.

[0158] According to this embodiment, users can learn the shapes of characters by writing them using hints as clues, and then train their understanding of the string by subsequently inputting answers. Because writing characters and inputting answers are performed sequentially within a single quiz session, a complex load is placed on the brain, resulting in a high learning effect. Furthermore, since a variety of hint types and answer methods can be set, this embodiment is applicable to a wide range of learning needs and user characteristics, such as hiragana learning for young children, character acquisition for foreign language learners, and for people with autism or hearing impairments (including sign language users).

[0159] <Variation> In the above embodiment, the server 10 is central to processing tasks such as preparing and sending quiz text and calculating points to be awarded to users, but the system is not limited to this configuration. For example, the system may not have a server 10, and processing may be completed within the mobile terminal 20. Alternatively, if multiple mobile terminals 20 are participating in a quiz session, one or more of those mobile terminals 20 may be responsible for or share the processing tasks.

[0160] Furthermore, the quiz text generation process does not necessarily need to be centrally executed on server 10; it may be generated in a distributed manner by any of the mobile terminals 20 participating in the quiz session. Moreover, the quiz text may be generated by another device not participating in the quiz session (for example, any mobile terminal 20) and sent to the mobile terminals 20 participating in the quiz session. By adopting this configuration, where the quiz text generation process is executed by a device at the network end, so-called edge computing, advantages such as reduced network traffic and reduced response delay can be obtained.

[0161] Furthermore, the quiz text may be generated within server 10, or it may be obtained by server 10 querying an external system (for example, a system related to generation AI). For example, the quiz text may be obtained by prompting with keywords such as the learning level of the target quiz session, the topics of interest of the participating users, and current news information.

[0162] Furthermore, while it is preferable, the user reads the quiz text in the order it appears from the beginning, it is not limited to this. For example, any words in the text may be read aloud in any order, and the user may be deemed ready to answer if a predetermined amount or percentage of the words are reproduced. Furthermore, instead of reading aloud all words in the quiz text, the system may be configured to designate only specific words as the target for reading aloud. The method for selecting the words to be read aloud is not particularly limited; for example, difficult words may be selected, or they may be selected randomly. In this case, the condition for enabling the answer may be that all of the designated words to be read aloud have been read aloud, or that a predetermined percentage of the total designated words to be read aloud have been read aloud.

[0163] Furthermore, the correct answer to a quiz text does not necessarily have to be unique. For example, a quiz text may have multiple correct answers (e.g., "Which country has a flag composed of only two colors, red and white?"). Moreover, a quiz text may require multiple correct answers (e.g., "What are the three primary colors of light?"). In this case, answering one correct answer may be sufficient, or a predetermined number of correct answers (e.g., all three) may be required. Additionally, if only a portion of the required answers are answered correctly, fewer points may be awarded compared to answering all the required answers correctly (for example, if three correct answers are required, 3 points may be awarded for answering all three correctly, and 2 points for answering two correctly, with points awarded proportionally).

[0164] Furthermore, when the audio input unit 25 can accept audio information read aloud by the user in order to reproduce the quiz text, it is preferable to make the recognition criteria more lenient than in normal speech recognition when determining whether the pronunciation of the target word has been entered correctly, given that language learners are generally not proficient in the pronunciation of the target language. In other words, it is preferable to adjust the criteria appropriately according to user attributes, etc.

[0165] Examples of ways to loosen recognition criteria include the following: (1) Lower the score threshold for phoneme recognition. Conventional speech recognition engines have a mechanism that determines a mismatch if the score of each phoneme (or subphone) falls below a certain threshold. However, in the learner pronunciation mode of the present invention, this threshold is lowered or handled flexibly so that even if there is a small difference in pronunciation or accent, it will not be considered "incorrect" and will be recognized as a similar phoneme. (2) Allow common error patterns For example, if the target language is English, common mistakes made by Japanese learners, such as confusing [r] and [l] or mispronouncing [th] as [s] or [z], are anticipated, and such phonetic discrepancies are not marked as "incorrect." (3) Handling of differences in accent and intonation The acoustic model is set to a broad range to allow recognition of regional pronunciations with different accents for the same word internationally, as well as shifts in stress placement that learners often get wrong. (4) Set the cost of DTW (Dynamic Time Stretching) gradually. In speech recognition, matching is performed with a large tolerance for time axis adjustments, even if there are variations in the learner's pronunciation speed (e.g., slow or fast). (5) Utilize phone-level partial match scores The system expands the criteria for "partial match," which is defined not only by whether the entire target word perfectly matches, but also by whether a certain number of similar phoneme sequences are present. For example, if the final score exceeds a certain threshold, it will be judged as "OK." (6) Learning that takes into account native language interference For example, the system adjusts the acoustic model and pronunciation dictionary by incorporating the respective native language interference patterns, such as creating an English speech recognition model for Japanese native speakers or an English speech recognition model for Korean native speakers. In addition, by estimating pronunciation tendencies based on the user's profile in advance and registering information such as the user's native language, learning history, and common errors, the system optimizes the "tolerance range" for each native language and individual user. (7) Learning that takes liaisons, etc. For example, in French liaison and English linking, sound changes occur when words are pronounced consecutively. However, because learners find it difficult to consciously read these changes, even if a word that should be pronounced with liaison or linking is read alone, it will still be considered correct. In such cases, it is preferable to output the following on the mobile device 20 after the word is reproduced (preferably in the latter half of the quiz session). (a) A statement indicating that pronunciation without liaison was detected in the text. Example: “ils ont” is pronounced as “ils on” and the message “Liaison omitted” is displayed. (b) Comparison with the correct pronunciation Example: Display "Standard liaison form: ***" and "Your pronunciation: ****" This allows learners to clearly identify where they went wrong.

[0166] Furthermore, the leniency of speech recognition may be adjusted according to the learner's level. For example, in beginner mode, the recognition threshold could be significantly relaxed, allowing even slight deviations to be considered close to the correct answer, while at intermediate and advanced levels, the recognition threshold could be gradually tightened, requiring more accurate pronunciation. Furthermore, points may be awarded based on the discrepancy between the correct pronunciation and the input pronunciation. For example, when comparing input that is nearly correct with input that is within an acceptable range of correctness despite some discrepancies, it is preferable to award a greater incentive to the former than to the latter.

[0167] Similarly, even when users reproduce the quiz text by typing rather than using voice input, minor spelling errors may be tolerated and considered within the range of a correct answer. For example, if a single letter of spelling is mistakenly entered with a nearby letter on the keyboard, or if too many or too few letters are entered, these may still be considered within the range of correct answers. In the former case, for example, if the input is from a keyboard, even if the input is from a nearby letter on the keyboard (for example, on a QWERTY keyboard, if "computer" is misspelled as "conputer" or "comouter" because the letter n is next to the letter m and the letter i / p is next to the letter o), these may still be considered within the range of correct answers.

[0168] Furthermore, users may choose which quiz session to participate in for each session, or the server 10 may determine which quiz session a user should participate in based on user information (including themes of interest registered in the user information table 153, as well as user attributes, history, etc.), or the quiz sessions may be held within user groups set up by the user. Furthermore, the quiz text in a quiz session may be determined based on the user information of the participant in the quiz session, or themes related to the quiz text (for example, "sports," "entertainment," "geography," etc.) may be set in advance, and users may select a quiz session to participate in based on those themes.

[0169] Furthermore, in the first embodiment, etc., the system was configured to allow the user to input an answer corresponding to the quiz text when a predetermined amount or percentage of the displayed quiz text was reproduced by the user's input, but the system is not limited to this configuration. For example, the user may input an answer even before the predetermined amount or percentage of the quiz text is reproduced by the user, and only after the predetermined amount or percentage is reproduced by the user will the system accept the already entered answer (the system recognizes that the answer has been completed). Also, for example, the trigger for enabling the input of an answer corresponding to the quiz text is not limited to when the predetermined amount or percentage of the displayed quiz text is reproduced by the user's input, but the trigger may be when any condition including this case is met.

[0170] Furthermore, in the first embodiment, etc., the responses of participating users are comprehensively managed within the framework of a quiz session occurring over a certain period, and elements such as the content of the response and the speed of the response are comprehensively evaluated, but the management method is not limited to this. For example, if information related to the quiz (quiz text, correct quiz answer, etc.) as well as the time corresponding to the start of the quiz session and the time of the response can be obtained, then even if the quiz is conducted at different times on each mobile device 20, it is possible to determine whether the response is correct and how quickly it was answered. As long as the quiz text is the same, the responses on each mobile device 20 can be evaluated using the same criteria.

[0171] In other words, even if there are slight differences in the start times of different quiz sessions, it is possible to manage the answers from each mobile device 20 as if they were participating in the same quiz session. Furthermore, even if the start times of the quiz sessions are completely different from those of other mobile devices 20 (for example, on different days), the answers can be evaluated using the same criteria. That is, the evaluation of answers to the same quiz text can be performed across different quiz sessions.

[0172] Furthermore, in the above embodiments, various programs and data related to various processes are stored in the storage unit, and the processing unit reads and executes these programs to realize the processes in each of the above embodiments. In this case, the storage unit of each device may have recording media (recording media, external storage devices, storage media) such as memory cards (SD cards), CompactFlash® cards, Memory Sticks, USB memory, CD-RWs (optical discs), and MOs (magneto-optical discs), in addition to internal storage devices such as ROM, EEPROM, flash memory, hard disks, and RAM, and the above programs and data may be stored on these recording media.

[0173] Further detailed modifications of the above-described correct answer-first (seed-first) question text generation method will be explained. In these modifications, it is preferable to adopt a two-stage generation method in which "contextual information (hints)" is provided as an intermediate product between the seed and the question text in order to systematically generate diverse question texts from the same seed.

[0174] In the first stage, the generating AI is given a seed and a predetermined perspective (hereinafter referred to as "angle") related to that seed as input to generate contextual information. The angles are predefined, such as "historical background / turning point," "etymology / origin," "correction of common misconceptions," "record-breaking / extreme cases," "specific individuals involved in the event," "specific locations involved in the event," "cause-and-effect relationship," "cultural significance / impact," "application to real society," and "classification / definition perspectives." The contextual information is based on the combination of the seed and the angle, and does not directly reveal the correct answer, but consists of one or more sentences describing factual background information related to the seed.

[0175] In the second stage, the generation AI is given a seed and contextual information as input to generate quiz questions. By generating questions that are relevant to the content of the contextual information, the direction of the questions is naturally determined by the contextual information, resulting in questions that are more specific and natural in expression.

[0176] By generating contextual information for each of multiple angles and storing it in a database, it becomes possible to systematically generate multiple types of problem statements corresponding to the number of angles from a single seed. For example, if the seed is "Newton," a problem statement using contextual information based on "historical background" (e.g., "In the 17th-century scientific revolution, the law of universal gravitation was formulated, making it possible to explain the motion of celestial bodies in a unified way") and a problem statement using contextual information based on "correcting common misconceptions" (e.g., "The anecdote that he conceived of gravity after seeing an apple fall is said to be a later fabrication, but his research forms the foundation of modern physics") will generate problem statements with the same correct answer but from completely different perspectives.

[0177] Furthermore, it is preferable to manage which angles of contextual information have already been generated and stored for each seed, and to control the generation of duplicate contextual information for the same angle. This has the effect of preventing monotony in user learning, as even when the same seed is presented repeatedly, the question text is always provided from a different perspective.

[0178] Further detailed modifications regarding the operation of the seed database will be described. When a large number of seeds are registered in the seed database, if questions based on a particular seed are repeatedly presented in a short period of time, it may lead to a situation where users can anticipate the correct answers, potentially impairing the learning effect. In addition, depending on the genre, some seeds may lose their freshness over time and become unsuitable for questions. In this modification, server 10 has a lifecycle management function that manages the lifecycle of seeds from multiple independent perspectives in order to resolve these problems.

[0179] The first management mechanism is cooldown control. After the server 10 presents a question based on a seed to the user, it temporarily removes that seed from the list of potential questions until a predetermined cooldown period (e.g., 7 days) has elapsed. This prevents the repeated presentation of questions based on the same seed in a short period of time and promotes the even use of diverse seeds in the seed database.

[0180] The second management mechanism is retirement control based on usage count. The server 10 manages the cumulative number of times each seed has been used, and permanently removes seeds from the list of candidates for use when that count reaches a predetermined threshold (e.g., 50 times) (hereinafter referred to as "retirement"). By preventing excessive repeated use of specific seeds, the diversity of questions can be maintained over the long term.

[0181] The third management mechanism is the time-based retirement control for topical genres. Server 10 defines whether each genre is topical or not, and seeds classified as topical genres (e.g., news, social trends, etc.) are retired after a predetermined period (e.g., 180 days) has elapsed since the date of registration of the seed in the database. Since the freshness of information is directly related to the quality of the questions for topical seeds, a retirement condition independent of that for non-topical genres is established. The cooldown control, usage count retirement control, and time-based retirement control are each evaluated and applied independently, and seeds that meet any of the conditions are excluded from the list of potential questions.

[0182] In this way, by independently applying cool-down control, usage count retirement control, and time-based retirement control, it is possible to continuously maintain the quality of the seed database while always providing users with fresh and appropriate quiz information. Each of these controls evaluates the suitability of the seeds for question generation from a different perspective, and by using them in combination, it is possible to simultaneously satisfy three quality requirements: prevention of short-term repetition, ensuring long-term diversity, and maintaining timeliness.

[0183] In the seed-first generation method described above, the seed is primarily selected from a database managed by the system. However, it is also possible for the user themselves or another person, such as the user's teacher, to specify the seed and generate quiz information. For example, if the user wants to deeply learn a particular word or group of words, or if another person believes the user should deeply learn those words or groups of words, the user or another person can input the word or group of words into the mobile device 20, and this can be used as a seed to generate quiz information for the AI, which then presents the quiz to the user. In this way, the user can proactively select learning targets based on their own learning challenges and interests, or another person can select the user's learning targets, resulting in more purposeful and efficient vocabulary acquisition.

[0184] As an extension of this configuration, a configuration in which a user or another person inputs web articles or other external content into a mobile device 20, and the system automatically extracts seeds from words or expressions contained in the content to generate quiz information, is also conceivable. Specifically, when a user or another person inputs the URL or text of a web article written in the target language, the server 10 acquires the content and, according to predetermined rules, randomly or using a generating AI, extracts words or expressions from the content that are deemed to be of high significance to the user in light of their learning level, etc., as seeds. Quiz information is generated based on the extracted seeds and presented to the user. In this way, the user can repeatedly learn in a quiz format words and expressions directly related to the content they have actually encountered, or words and expressions directly related to the content that another person considers appropriate for the user, thereby promoting both content comprehension and vocabulary acquisition simultaneously. Furthermore, since the quiz is based on content that the user is interested in or that is considered appropriate for the user, it also contributes to maintaining motivation to learn.

[0185] In the above explanation, adjusting the threshold for the recognition accuracy of individual words was described as a way to loosen the criteria for speech recognition. However, when a user fluently reads a quiz question, multiple words may be spoken consecutively in one breath, and the speech recognition engine may obtain a sequence of words containing multiple words at once as output. In such cases, a method that matches only the single word at the current position of the recall judgment may result in a judgment of recall failure even if the word is read correctly, because the word at the current position and the word in the recognition result do not directly match.

[0186] To resolve this issue, this modified version employs a "pre-read window matching" method, which attempts to match the word sequence obtained from the speech recognition result with the expected word sequence from the current position of the word in the recall judgment up to a predetermined number of words ahead (for example, up to 3 words ahead). In other words, if a sequence of consecutive words included in the speech recognition result is determined to correspond to and match multiple words ahead of the current position, these multiple words are considered to have been recalled together, and the recall of multiple words is recognized at once. This ensures that the recall progress is accurately updated even when multiple words are read aloud fluently in succession.

[0187] Furthermore, in this modified version, instead of requiring only a perfect spelling match for the words during matching, a "approximate matching" is incorporated, which considers matching successful even if the edit distance (e.g., Levenshtein distance) between the word to be matched and the expected word is below a predetermined threshold. Edit distance is defined as the minimum number of insertion, deletion, and substitution operations required to convert one string to the other. As a result, even if pronunciation variations specific to non-native speakers (such as consonant omission, vowel changes, and differences in accent placement) are reflected in the speech recognition result, they can be tolerated as reasonable differences in pronunciation, and the result can be determined to be a successful reproduction.

[0188] By combining the aforementioned look-ahead window matching and approximate matching, the following two problems can be solved simultaneously: Firstly, accurate progress tracking when multiple words are recognized consecutively through fluent reading (solved by look-ahead window matching). Secondly, flexible reproduction determination when the recognition result does not perfectly match due to variations in pronunciation or accent (solved by approximate matching). It is preferable that the width (number of words) of the look-ahead window and the threshold for the edit distance in approximate matching be adjustable according to the user's learning level and language used. For example, a narrower window width and a smaller threshold may be applied to advanced users to more strongly demand accurate pronunciation.

[0189] A more detailed modification will be described regarding the relationship between the quiz text and the text to be reproduced. In the above-described embodiment, the quiz text displayed to the user and the text to be reproduced under the predetermined conditions were the same. That is, the requirement for satisfying the predetermined conditions was to reproduce the text displayed as the question exactly as it was in audio. In this modification, the server 10 presents the user with reading material (hereinafter referred to as "read-aloud text") that describes background information or related information of the quiz question (hereinafter referred to as "question text"), and adopts a configuration in which the read-aloud text is the target of reproduction under the predetermined conditions.

[0190] The screen display configuration in this modified example will now be described. Server 10 displays, for example, the following text in the main part of the user's display area: "Born in 17th-century England, this scientist formulated the law of universal gravitation while at Cambridge University, and later served as President of the Royal Society." In a separate display area, the following problem statement text is displayed: "Who formulated the law of universal gravitation and laid the foundation of classical mechanics?" The text to be read aloud does not directly reveal the correct answer, but it is preferable that it is structured as a text that includes factual background information related to the correct answer.

[0191] The user aims to fulfill predetermined conditions for enabling or validating an answer by reading aloud the audio text. Server 10 manages the progress of the audio text reproduction by comparing the audio information with the audio text, and determines that the predetermined conditions have been met when a predetermined percentage or amount of the audio text has been reproduced. After the predetermined conditions have been met, the user can input an answer to the question text. The question text remains displayed even while the audio text is being read aloud, allowing the user to view the question text and think about their answer while reading aloud.

[0192] In this modified example, "quiz text" is a concept that includes both the audio text and the question text. That is, the entire series of displayed content presented to the user as a quiz constitutes the "quiz text," and the audio text to be reproduced is a part of it. The predetermined condition is "reproduction of a predetermined proportion or amount of the quiz text," and in this modified example, the reproduction of the audio text portion that constitutes the quiz text is the subject of the determination of the predetermined condition.

[0193] The learning effects of this modified version are explained below. Users are required to read aloud the text accurately and quickly (improving reading speed and pronunciation accuracy), and at the same time, to understand the content of the text and derive the correct answer to the quiz (improving reading comprehension and vocabulary). Furthermore, since the question text is displayed while the text is being read aloud, a situation is created where three processes—reading aloud, comprehension, and answer thinking—occur simultaneously. As a result, a learning state with a high load on the learner's brain is achieved, similar to the previously described embodiment, leading to a high learning effect. In addition, if the text to be read aloud is longer than the question text, the cognitive load from reading aloud is further increased, resulting in the additional effect of promoting the training of more advanced language processing abilities.

[0194] Furthermore, the audio reading text can be generated by Server 10 using a generation AI based on seeds stored in the seed database. For example, if the seed is "Newton," Server 10 generates an audio reading text describing background information about Newton and presents it in a separate display area from the problem statement text. By combining this with the contextual information generation mechanism using angles described above, it is possible to generate audio reading texts with different angles for the same seed, systematically providing diverse reading comprehension materials.

[0195] Furthermore, the text to be read aloud is not limited to content related to the correct answer corresponding to the quiz text, but may also be content independent of the quiz question. For example, it may be any sentence, or even a string of meaningless words. In other words, it may be a configuration that places a dual cognitive load on the user, as they read the question and think about the answer while simultaneously reading aloud something else.

[0196] Further modifications regarding the criteria for determining reproduction under predetermined conditions will be described. In the above-described embodiment, the progress of reproduction was managed by the ratio or absolute number of reproduced words to the entire text. In this modified example, the server 10 predetermines specific words or specific parts (hereinafter referred to as "predetermined parts") in the text as the subject of reproduction determination, and considers the predetermined conditions to be met when it determines that the predetermined parts have been reproduced based on the audio information. Examples of criteria for selecting predetermined parts include a configuration in which parts are selected randomly, a configuration in which words deemed important for learning (e.g., learning target vocabulary, keywords, technical terms, etc.) are targeted, a configuration in which specific positions in each sentence (e.g., the first N words of each sentence) are targeted, or a configuration that combines these.

[0197] The designated portion may or may not be disclosed to the user. In the disclosure configuration, the designated portion of the text can be visually indicated to the user, for example, by highlighting, underlining, or using color to emphasize it. This allows the user to clearly understand the portion to be reproduced and concentrate on its pronunciation, making it suitable for intensive practice of specific vocabulary or expressions. On the other hand, in the configuration where the designated portion is not disclosed to the user, the user cannot know which portion is to be judged, which encourages them to read the entire text aloud and promotes the maintenance of continuous attention to the entire text.

[0198] Thus, the disclosure or non-disclosure of certain parts can be selected according to the learning objectives and difficulty level, and in either embodiment, by using qualitatively and spatially identified locations rather than quantitative ratios as the criteria for reproduction assessment, it is possible to give learning significance to accurately pronouncing specific words in the text.

[0199] Further variations in the structure of the quiz information will be described. In the embodiment described above, the quiz information consisted only of text, but in this variation, the server 10 may display quiz information consisting of a combination of images and text information. As a specific example, the quiz information consists of one or more images presented to the user and text describing a question related to the image.

[0200] For example, one configuration may display an image of a famous building along with a question such as, "What city is this building located in?", or an image of a person along with a question such as, "Who is this person?". The image itself is not the object of reproduction, but the satisfaction of the predetermined conditions is determined by the reproduction of the text included in the presented information (the question text, or a separately displayed audio text).

[0201] In this modified example, the determination of the predetermined conditions is made by comparing the audio information with the text contained in the presented information, and the content of the image itself is not subject to reproduction judgment. This makes it possible to apply the answer gate mechanism, which is conditional on the reproduction of text, to visual and intuitive problem formats that use images, thereby achieving both diversification of problem formats and ensuring learning effectiveness.

[0202] The following describes a modified version. This modified version describes language learning support using a set of correct answers consisting of multiple correct answers corresponding to quiz information. In the previously described embodiment, it was assumed that there was only one correct answer to the quiz information, but this modified version describes the case where there are multiple correct answers to the quiz information. For example, for quiz information such as "Name one South American country," a set of correct answers consisting of multiple correct answers such as Brazil, Chile, Argentina, Peru, and Colombia is obtained from a server or the like. The number of correct answers included in the set of correct answers is not particularly limited and is set appropriately according to the content of the quiz information.

[0203] In this modified example, users input answer information on each of the multiple terminals. An answer is considered correct if the answer information entered on each terminal corresponds to one of the correct answers included in the set of correct answers. The handling of cases where the same correct answer is submitted on multiple terminals is not particularly limited, and various methods are possible, such as validating only the answer from the terminal that submitted the answer first and invalidating the answer from terminals that submitted the same correct answer later, suppressing duplicate answers by displaying the correct answers already submitted on other terminals on the display of each terminal, or simply invalidating duplicate answers. The number of terminals is not limited to two, and the same can be implemented in a configuration where three or more terminals participate.

[0204] In this way, by using a set of correct answers consisting of multiple correct answers and having multiple devices share the task of answering different correct answers, each user aims for a different correct answer, preventing them from easily giving the same answer by referring to other users' answers, and encouraging each user to think and answer independently. Furthermore, by using quiz information with multiple correct answers, the range of quiz sessions that can function even as the number of participating devices increases broadens, making it easier to realize multi-participant quiz sessions.

[0205] The following describes a modified version. This modified version describes language learning support using correct answer conditions corresponding to quiz information. In the previously described embodiment, it was assumed that there was one correct answer to the quiz information, and that correctness or incorrectness was determined by whether the answer information matched that correct answer. However, this modified version describes a method in which the degree of correspondence between the answer information and the correct answer conditions is evaluated, and an incentive is determined according to that degree. The correct answer conditions are criteria for evaluating the answer information obtained from the user, and a method in which conditions based on semantic similarity with a predetermined answer can be considered. Here, semantic similarity refers to the degree of semantic similarity between the answer information and the predetermined answer, and refers to a value quantified by, for example, the distance in a vector representation that quantifies the semantic relationship between words, the hierarchical distance of words in a thesaurus, or the similarity evaluation result by a generative AI, but is not limited to these.

[0206] When using correct answer criteria based on semantic similarity, the higher the semantic similarity between the answer information and the given answer, the higher the degree of correspondence is evaluated, and a higher incentive is awarded. For example, for the quiz information "What are synonyms for happy?", a semantically closer answer such as elate will be evaluated as having a higher degree of correspondence than an answer such as joyful, and a higher incentive will be awarded. It is also conceivable to use a threshold-based evaluation method, where a high incentive is awarded when the semantic similarity is above a predetermined threshold, and a low incentive is awarded when it is below the threshold. In this way, by determining the incentive according to the degree of correspondence, incentives are awarded not only for answers that are not completely correct but also for answers based on partial understanding, and it is expected that this will promote gradual language acquisition while maintaining the user's motivation to learn.

[0207] The following describes another variation. This variation describes an embodiment in which a predetermined category is used as the correct answer condition. In particular, this variation realizes an embodiment in which, even if the user does not fully understand the meaning of the quiz information in the context of foreign language learning, the user can infer the content being asked from known words contained in the quiz information and answer. For example, if the user is presented with quiz information in a foreign language that they have not yet learned, the user will infer the content being asked from known words contained in the quiz information and answer. If the quiz information asks about hobbies, words or expressions that are generally recognized as hobbies will satisfy the correct answer condition, and if it asks about favorite foods, words belonging to the food category will satisfy the correct answer condition. The predetermined category is identified based on the quiz information, and may be configured to be automatically identified from the quiz information using, for example, a generative AI. In this way, by using a category-based correct answer condition, it becomes possible for the user to answer even when they do not fully understand the meaning of the quiz information, and training of advanced language processing ability, such as inferring the meaning of words in the context of a foreign language, is realized. In other words, a learning cycle is formed in which the user reproduces the text of the quiz information in a foreign language while simultaneously inferring its meaning and answering, which involves a complex cognitive load, and effective language acquisition is promoted.

[0208] Thus, the structure of the quiz, which involves inferring the meaning of quiz information and answering while performing the cognitively demanding task of reproduction, has a unique learning effect as training for practical communication skills in a foreign language environment, independent of the characteristic of text reproduction. In other words, the process of inferring meaning from a foreign language text whose meaning cannot be fully grasped, using known words as clues, is a skill that is required on a daily basis in a real foreign language environment, and training this skill is expected to improve practical language proficiency.

[0209] Furthermore, the answer information is not limited to single words or expressions, but may also be in sentence form. For sentence-form answers, it is necessary to interpret their semantic content and determine whether they belong to a category. For example, a system may be configured to use a generative AI to interpret the semantic content of the answer information and determine whether that content belongs to a predetermined category. For example, if a user answers "I like Star Wars." to the quiz question "What is your favorite movie?", the generative AI interprets the semantic content of the answer information and determines that it belongs to the movie category, thus determining it is correct. On the other hand, if the user answers "I like Star Wars." to the quiz question "What was the last movie you saw?", it is determined that the semantic content does not match the category corresponding to the quiz question, and is therefore incorrect. In addition, a system may be configured to provide a higher incentive when the answer information is in a semantically appropriate sentence form compared to when only a single word is answered. For example, a higher incentive may be given for answering "I like Star Wars." to the quiz question "What is your favorite movie?" than for answering "Star Wars". This approach increases users' motivation to respond using appropriate written expressions in a foreign language, thereby promoting the acquisition of more practical language proficiency.

[0210] In relation to the above modification using multiple correct answer sets, we will now describe a control that prioritizes receiving answer information from terminals that satisfy predetermined conditions first. When multiple terminals participate in the same quiz session and each terminal inputs answer information aiming for a different correct answer included in the correct answer set, When users on each terminal reproduce the quiz text (text included in the presented information) at their own pace, the timing at which predetermined conditions are met on each terminal may not coincide. In this modified example, the server 10 controls the system to prioritize receiving answer information from the terminal that first satisfies the predetermined conditions (for example, the terminal that first completes the reproduction of a predetermined amount of text included in the presented information).

[0211] For example, in a buzzer-style quiz session, the user whose device first meets certain conditions may be granted the right to answer first, while other devices may be prevented from receiving answer information until the answer processing on the device granted the right to answer is complete. In this way, users who reproduce the text faster and more accurately are given priority in answering, and a learning cycle with a high cognitive load of "reading quickly and answering quickly" is naturally formed. As a result, a fair competitive environment among users is maintained while the motivation to learn is enhanced.

[0212] Further modifications regarding the re-presentation of questions in the event of an incorrect answer will be described. In the above embodiment, an example was described in which, as a process for handling an incorrect answer by the user, information regarding the correct answer is output to the mobile terminal 20 and the quiz session is terminated. In this modification, if the answer information obtained from the user does not correspond to the correct answer corresponding to the quiz information (i.e., it is an incorrect answer), the server 10 generates or obtains new quiz information corresponding to the correct answer and outputs presentation information including this information to the user's terminal.

[0213] Specifically, for example, if a user incorrectly answers "jellyfish" to the correct answer "octopus," the server 10 uses a generation AI to generate new quiz information from a different perspective with "octopus" as the correct answer (for example, using "octopus" as the seed in the seed-first generation method described above), and sends the presentation information including this to the user's terminal to re-present the question. In this way, even after incorrectly answering a word, the user can immediately try another question with the same correct answer, and deepen their knowledge from multiple contexts related to the same word while waiting to answer correctly. In combination with the first modified example (4) (processing before and after steps TA2008, TB2008), new quiz information related to the correct answer is presented after the initial incorrect answer, and the user is asked to reproduce and answer it using voice input, thereby realizing an active knowledge retention process starting from an incorrect answer.

[0214] In this way, by repeatedly presenting quiz information from different contexts related to the same correct answer in response to an incorrect answer, users can deepen their understanding of the word in various contexts in which it is used, rather than simply memorizing the correct answer. As a result, it has the effect of promoting long-term memory retention of the correct answer itself while preventing dependence on a specific quiz format. In addition, by presenting different quiz information each time an answer is given incorrectly, it also has the effect of maintaining the motivation for users to continue the learning session even after making a mistake.

[0215] <Variations of reverse processing> In the first to eighth embodiments described above, a process was described in which a predetermined condition is met upon the reproduction of a predetermined amount, predetermined proportion, or predetermined portion of the text contained in the presented information, and only when this predetermined condition is met is the input of an answer to the quiz information or the activation of an input answer (hereinafter referred to as "sequential processing"). This modification relates to a form that employs processing in the reverse order (hereinafter referred to as "reverse processing").

[0216] In reverse processing, the user first inputs answer information for the quiz. Once the input of the answer information is complete, or once the answer information meets predetermined conditions (e.g., matching the correct answer, belonging to a predetermined category, etc.), the user can reproduce the text contained in the presented information. The result of the quiz session is then determined when the reproduction reaches a predetermined amount, a predetermined percentage, or a predetermined portion. In other words, in reverse processing, the input of the answer information triggers the start of the reproduction, and the completion of the reproduction triggers the determination of the session result.

[0217] The specific flow of the reverse processing is explained below. First, the server 10 acquires presentation information including quiz information and outputs it to the display unit 24 of the mobile terminal 20. The user recognizes the presentation information and inputs answer information for the quiz information. At this point, input for reproduction (voice input, text input, etc.) is not possible, or even if input is made, it will be invalid. When the input of answer information is completed, or when the input answer information satisfies predetermined conditions, the control unit 21 moves to the reproduction phase and enters a state where it accepts input from the user for the reproduction of the text contained in the presentation information. The user reproduces the text contained in the presentation information, and when it is determined that the reproduction has reached a predetermined amount, predetermined percentage, or predetermined portion, the result of the quiz session is determined, and the server 10 grants an incentive to the user.

[0218] In reverse processing, the text to be reproduced preferably includes background information or related information concerning the correct answer corresponding to the quiz information. For example, if the correct answer to the quiz information is "pandemic," the text to be reproduced may include background information related to the correct answer, such as "The 1918 Spanish Flu spread worldwide, killing tens of millions. It was named after Spain, which freely reported the outbreak as a neutral country in World War I." In this way, by reproducing text that includes background or related information related to the correct answer, users can experientially acquire the meaning, usage, and context of the correct answer through reading it aloud. The text to be reproduced can be generated by the generating AI using the correct answer as a seed, similar to the seed-first generation method described in the first embodiment.

[0219] The means of reproduction in reverse processing are not particularly limited, and various reproduction means described in the first to eighth embodiments can be applied. In particular, when reproduction is done by reading aloud, the user reads aloud a text containing background information or related information related to the correct answer after inputting the answer information. As a result, the user reads aloud while comparing their own answer with the content of the text, and can digest the meaning, background, and usage of the correct answer from multiple perspectives. Consequently, it has the effect of promoting a deeper understanding of the context in which the correct answer is used, rather than simply memorizing the correct answer. Furthermore, because the session result is not finalized until the reproduction is complete, the user is motivated to reproduce quickly in order to complete the session as soon as possible, thereby maintaining the workload as foreign language speaking training.

[0220] The incentives awarded in the reverse processing may be determined based on the accuracy of the reproduction (e.g., the proportion of reproductions, the speed of reproduction, the accuracy of pronunciation, etc.). The incentives may also be varied based on the correctness of the response information; for example, users who enter a correct answer may receive a higher incentive upon completion of the reproduction, while users who enter an incorrect answer may receive a lower or no incentive upon completion. Alternatively, a fixed incentive may be awarded upon completion of the reproduction, regardless of whether the answer is correct or incorrect.

[0221] Furthermore, forward processing and reverse processing may be switched for each quiz session, or the user may be allowed to choose arbitrarily, or the server 10 may automatically select based on the user's learning history and proficiency level. Also, forward processing and reverse processing may be combined within a single quiz session, for example, in the form of reproducing a predetermined percentage of the text contained in the presented information → inputting answer information → reproducing the remaining percentage.

[0222] <Ninth Embodiment> The ninth embodiment relates to a language learning support system that presents a user with multiple pieces of clue information, prompts the user to input words or phrases associated with each piece of clue information, and enables the user to answer a comprehensive quiz once a predetermined number or percentage of inputs have been made.

[0223] In this embodiment, the clue information refers to information in any medium that allows the user to uniquely associate a corresponding word or phrase to some extent, and is not limited to images, videos, audio, text, etc. Furthermore, these may be combined to form a single piece of clue information. For example, clue information could include a photograph of a person, an image of a specific building, an image of a national flag, an audio clip of a specific piece of music, or a video clip of a specific scene.

[0224] Server 10 acquires multiple clues and corresponding words or phrases (hereinafter referred to as "correct clue information") as quiz information and transmits them to the mobile terminal 20. The display unit 24 and sound output unit 26 of the mobile terminal 20 output the received multiple clues to the user. The user inputs words or phrases associated with each clue. The control unit 21 compares the input words or phrases with the correct clue information and, if it determines that a predetermined number or proportion of clues have been correctly entered, it determines that predetermined conditions have been met and enables the user to answer the comprehensive quiz.

[0225] As a concrete example, the system presents multiple clues related to a country (such as an image of the country's flag, an image of its capital, an image of a famous building, an image of a local product, etc.), and the user inputs a word or phrase corresponding to each clue (such as the name of the flag, the name of the capital, the name of the building, the name of the local product, etc.). For example, the user might input "Swiss flag" for an image of a flag with a white cross on a red background, "Alps" for an image of an Alpine landscape, "clock" for an image of a clock, and "chocolate" for an image of chocolate. If a predetermined number of inputs are made correctly, the system allows the user to answer a comprehensive quiz asking "What is the name of this country?", and the user answers "Switzerland."

[0226] Another example involves presenting multiple elements from a literary work as clues, and having the user input a word or phrase corresponding to each clue. For instance, if Sherlock Holmes is the subject, the user might input "pipe" for an image of a pipe, "violin" for an image of a violin, "Baker Street" for an image of Baker Street, and "magnifying glass" for an image of a magnifying glass. If a predetermined number of inputs are correct, the user is then able to answer a comprehensive quiz question, "These are associated with which fictional detective?", and the user might answer "Sherlock Holmes."

[0227] According to this embodiment, the user is required to perform complex cognitive processing, such as associating and inputting corresponding words or phrases from multiple pieces of clue information, in order to arrive at the final answer to the comprehensive quiz. This allows for the simultaneous acquisition of individual words or phrases and training of the ability to synthesize and infer them, resulting in a high learning effect. Furthermore, since the media types of the clue information can be set in a variety of ways, learning that utilizes multiple senses such as sight and hearing becomes possible, which also has the effect of being able to accommodate a wide range of user characteristics. [Explanation of Symbols]

[0228] 1. Language Learning Support System 10 servers 20 Mobile devices NW Network

Claims

1. A language learning support method that is executed by one or more processors, To obtain the information presented, including quiz information, The aforementioned information is output in a manner that the user can recognize, To obtain input information entered by the user, To obtain the response information entered by the user, When the predetermined conditions are met, the input of the response information is enabled, or the acquired response information is made valid. A language learning support method that includes, The aforementioned predetermined conditions include the fact that a predetermined amount, predetermined proportion, or predetermined portion of the text included in the presented information has been reproduced based on the input information.

2. A language learning support method according to claim 1, The aforementioned input information includes audio information.

3. A language learning support method according to claim 1 or 2, The aforementioned quiz information includes the aforementioned text.

4. A language learning support method according to claim 1, Outputting the aforementioned information in a manner recognizable by the user includes gradually displaying the text contained in the aforementioned information and outputting audio that reads the text aloud.

5. A language learning support method according to claim 4, The speed at which the text is displayed is slower than the speed at which the audio reading the text is output.

6. A language learning support method according to claim 4, Outputting audio that reads the aforementioned text precedes, at least partially, the gradual display of the aforementioned text.

7. A language learning support method according to claim 1, The output of the aforementioned presentation information in a manner recognizable by the user is performed by the display units of the first terminal and the second terminal. The input information includes voice information entered by the user at the first terminal and text information entered by the user at the second terminal.

8. A language learning support method according to claim 1, Acquiring the aforementioned presentation information includes acquiring first presentation information which includes the first text and second presentation information which includes the first text which has the same correct quiz answer as the quiz information and the second text which has different content from the first text. Outputting the aforementioned information in a manner recognizable by the user includes displaying the first information on the display unit of the first terminal and displaying the second information on the display unit of the second terminal. Acquiring the aforementioned input information includes acquiring the aforementioned input information entered by the user at each of the first terminal and the second terminal, Obtaining the aforementioned response information includes receiving user responses at each of the first and second terminals, Furthermore, the language learning support method includes determining an incentive corresponding to the user's response in the first terminal based on the comparison result between the correct quiz answer and the user's response in the first terminal and the second terminal, respectively. The reproduction included in the predetermined conditions is the reproduction of a predetermined amount, predetermined percentage, or predetermined portion of the text contained in the first presented information or the second presented information, based on the input information, for each of the first and second terminals, as individually determined for each terminal.

9. A language learning support method according to claim 8, The first text and the second text use different languages.

10. A language learning support method according to claim 8, The first presentation of information, which includes the first text, and the second presentation of information, which includes the second text, differ in difficulty level as quizzes.

11. A language learning support method according to claim 8, The determination of the aforementioned incentives is made in addition to the comparison results, (a) Response speed regarding the above response information, (b) The manner in which the response information is entered, (c) The language of the text included in the information presented above, (d) The difficulty level of the text included in the information presented above, (e) The learning history of the user who entered the above answer information, It is carried out based on at least one of the following.

12. A language learning support method according to claim 1, Outputting the aforementioned information in a manner recognizable by the user includes displaying the text contained in the aforementioned information on the display units of the first terminal and the second terminal, respectively. Acquiring the aforementioned input information includes acquiring the input information entered by the user at the first terminal, Acquiring the aforementioned response information includes receiving user responses at the second terminal, and the reproduction included in the predetermined conditions is the reproduction of a predetermined amount, predetermined percentage, or predetermined portion of the text included in the presented information based on the input information acquired at the first terminal.

13. A language learning support method according to claim 2, In the reproduction of the text based on the input information, if the word to be reproduced is not reproduced within a predetermined time, information regarding that word shall be displayed on the display unit. It also includes.

14. A language learning support method according to claim 2, The reproduction of the text based on the input information is performed by changing the speech recognition criteria according to the attributes of the user who inputs the information.

15. A language learning support method according to claim 1, To obtain the correct answer to the aforementioned quiz information, The AI ​​generates the quiz information based on the input including the correct answer. It also includes.

16. A language learning support method according to claim 15, To obtain contextual information based on predetermined perspectives regarding the correct answer, It further includes, The input includes the contextual information.

17. A language learning support method according to claim 16, The process involves using the correct answer and one of a predefined set of viewpoints as input to generate the contextual information using the generating AI, The process involves using the correct answer and the contextual information as input to cause the generating AI to generate the quiz information, Includes.

18. A language learning support method according to claim 17, To generate and store the contextual information for each of the aforementioned multiple viewpoints, Using different contextual information based on each of the aforementioned multiple perspectives, multiple different quiz pieces of information are generated for the same correct answer. It also includes.

19. A language learning support method according to any one of claims 15 to 18, In obtaining the correct answer, a seed database is referenced, which stores candidate correct answers as seeds. (a) Cooldown control that temporarily excludes seeds from the list of potential questions for which a predetermined cooldown period has not elapsed since the most recent question, (b) Usage count retirement control that permanently removes seeds from the list of candidates for questions when the cumulative number of questions asked reaches a predetermined threshold, (c) Seeds classified as genres with a current event nature should be added to the database. Time-based retirement control, which removes a candidate from the list of potential questions after a specified period has elapsed since registration, Selecting the seed by applying at least one of the following, Includes.

20. A language learning support method according to claim 2, The determination of whether or not the reproduction has been performed is made by comparing the word sequence obtained from the audio information with the expected word sequence in the text from the word at the current position of the reproduction judgment to a predetermined number of words ahead.

21. A language learning support method according to claim 20, In the matching process, the matching is considered successful if the edit distance between the word sequence and the expected word sequence is less than or equal to a predetermined threshold.

22. A language learning support method according to claim 20 or 21, In the matching process, if it is determined that multiple consecutive words included in the word sequence correspond to and match multiple words preceding the word at the current position, those multiple words are considered to have been reproduced collectively.

23. A language learning support method according to any one of claims 20 to 22, At least one of the predetermined number and the threshold for the editing distance is adjusted according to the learning level of the user inputting the voice information.

24. A language learning support method according to claim 1, The process further includes obtaining a set of correct answers consisting of multiple correct answers corresponding to the aforementioned quiz information, Obtaining the aforementioned response information includes obtaining first response information at the first terminal and obtaining second response information at the second terminal. The aforementioned language learning support method is The first answer information described above is considered correct if it corresponds to one of the correct answers included in the set of correct answers, The method further includes determining that the second answer information is correct if it corresponds to another correct answer that is different from the first correct answer included in the group of correct answers.

25. A language learning support method according to claim 24, Of the first and second terminals, the terminal that first satisfies the predetermined conditions will be given priority in acquiring the response information.

26. A language learning support method according to claim 1, To obtain the correct answer corresponding to the aforementioned quiz information, If the acquired answer information does not correspond to the correct answer, new quiz information corresponding to the correct answer is acquired. To obtain new presentation information, including the aforementioned new quiz information, It also includes.

27. A language learning support method according to claim 1, To obtain the correct answer conditions corresponding to the aforementioned quiz information, To evaluate the degree of correspondence between the acquired response information and the correct answer conditions, Incentives are determined based on the degree of the aforementioned response, It also includes.

28. A language learning support method according to claim 27, The aforementioned correct answer conditions include conditions based on semantic similarity with a predetermined answer, The degree of correspondence is evaluated based on the semantic similarity between the response information and the predetermined response.

29. A language learning support method according to claim 28, The degree of the aforementioned correspondence is evaluated using a predetermined threshold.

30. A language learning support method according to claim 1, The correct answer condition corresponding to the aforementioned quiz information is to obtain a predetermined category, The system determines that the answer is correct if the acquired response information, or the semantic content of the response information, belongs to the predetermined category. It also includes.

31. A language learning support method according to claim 1, The aforementioned input information includes character information, The aforementioned text information is entered by handwriting input on a touch panel.

32. A language learning support method according to claim 31, Outputting the aforementioned information in a manner recognizable by the user includes displaying characters or their outlines corresponding to at least a portion of the text. The aforementioned handwriting input is performed by tracing the displayed characters or outlines.

33. A language learning support method according to claim 1, The text included in the aforementioned information is written in a language different from the language the user is learning. The aforementioned answer information is entered by handwriting input on the touch panel, with the correct answer corresponding to the quiz information as a string in the language the user is learning.

34. A language learning support device, An acquisition unit that acquires presentation information including quiz information, An output unit that outputs the aforementioned presentation information in a manner that can be recognized by the user, An input information acquisition unit that acquires input information entered by the user, A response information acquisition unit that acquires response information entered by the user, When predetermined conditions are met, a control unit enables the input of the response information or validates the acquired response information, Equipped with, The aforementioned predetermined conditions include the fact that a predetermined amount, predetermined proportion, or predetermined portion of the text included in the presented information has been reproduced based on the input information.

35. On the computer, To obtain the information presented, including quiz information, The aforementioned information is output in a manner that the user can recognize, To obtain input information entered by the user, To obtain the response information entered by the user, When the predetermined conditions are met, the input of the response information is enabled, or the acquired response information is made valid. A program that executes, The aforementioned predetermined conditions include the fact that a predetermined amount, predetermined proportion, or predetermined portion of the text included in the presented information has been reproduced based on the input information.

36. A language learning support method that is executed by one or more processors, To obtain the information presented, including quiz information, The aforementioned information is output in a manner that the user can recognize, To obtain the response information entered by the user, To obtain input information entered by the user, When the predetermined conditions are met, the system enables the reproduction of the text contained in the presented information based on the input information, or activates the reproduction that has been made. A language learning support method that includes, The aforementioned predetermined conditions include the completion of inputting the response information or the fact that the response information satisfies the predetermined conditions.

37. A language learning support method according to claim 36, The aforementioned input information includes audio information, The text to be reproduced includes background information or related information concerning the correct answer corresponding to the quiz information.

38. A language learning support method according to claim 36 or 37, The system further includes enabling the determination of the results of the quiz session, or validating the determined results, on the condition that the reproduction reaches a predetermined amount, a predetermined percentage, or a predetermined portion.

39. A language learning support device, An acquisition unit that acquires presentation information including quiz information, An output unit that outputs the aforementioned presentation information in a manner that can be recognized by the user, A response information acquisition unit that acquires response information entered by the user, An input information acquisition unit that acquires input information entered by the user, When predetermined conditions are met, a control unit enables the reproduction of the text included in the presentation information based on the input information, or enables the reproduction that has been made. Equipped with, The aforementioned predetermined conditions include the completion of inputting the response information or the fact that the response information satisfies the predetermined conditions.

40. On the computer, To obtain the information presented, including quiz information, The aforementioned information is output in a manner that the user can recognize, To obtain the response information entered by the user, To obtain input information entered by the user, When the predetermined conditions are met, the system enables the reproduction of the text contained in the presented information based on the input information, or activates the reproduction that has been made. A program that executes, The aforementioned predetermined conditions include the completion of inputting the response information or the fact that the response information satisfies the predetermined conditions.

41. A language learning support method that is executed by one or more processors, To obtain quiz information, multiple clues related to the quiz information, and the correct answer information for each clue, Outputting the aforementioned multiple pieces of clue information in a manner that the user can recognize, Obtaining user input corresponding to each of the aforementioned multiple pieces of clue information, To obtain the response information entered by the user, When the predetermined conditions are met, the input of the response information is enabled, or the acquired response information is made valid. A language learning support method that includes, The aforementioned predetermined conditions include the fact that the user input corresponding to a predetermined number or proportion of the multiple pieces of clue information corresponds to the correct clue information.

42. A language learning support device, An acquisition unit that acquires quiz information, multiple pieces of clue information related to the quiz information, and correct clue answer information corresponding to each piece of clue information, An output unit that outputs the aforementioned multiple pieces of clue information in a manner that can be recognized by the user, A user input acquisition unit that acquires user input corresponding to each of the aforementioned multiple pieces of clue information, A response information acquisition unit that acquires response information entered by the user, When predetermined conditions are met, a control unit enables the input of the response information or validates the acquired response information, Equipped with, The aforementioned predetermined conditions include the fact that the user input corresponding to a predetermined number or proportion of the multiple pieces of clue information corresponds to the correct clue information.

43. On the computer, To obtain quiz information, multiple clues related to the quiz information, and the correct answer information for each clue, Outputting the aforementioned multiple pieces of clue information in a manner that the user can recognize, Obtaining user input corresponding to each of the aforementioned multiple pieces of clue information, To obtain the response information entered by the user, When the predetermined conditions are met, the input of the response information is enabled, or the acquired response information is made valid. A program that executes, The aforementioned predetermined conditions include the fact that the user input corresponding to a predetermined number or proportion of the multiple pieces of clue information corresponds to the correct clue information.

Images