JP2005107483A - Word learning method, word learning apparatus, word learning program, and recording medium with the program recorded thereon, and character string learning method, character string learning apparatus, character string learning program, and recording medium with the program recorded thereon - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a new technique for learning character strings enabling a learner to reasonably and efficiently learn character strings by presenting to the learner character strings stored in a character string dictionary and attribute information of the character strings. <P>SOLUTION: Extraction conditions for a test are set and, based on the extraction conditions, character strings are extracted out of the character string dictionary. The extracted character strings are outputted to the learner, and responses from the learner are collected. By analyzing the collected responses, character strings that the learner does not know are estimated. Then, extraction conditions for extracting important character strings out of the estimated character strings are set and, based on the extraction conditions, character strings and attribute information of the character strings and familiarity are extracted out of the character string dictionary. The extracted character strings are rearranged in the order of familiarity and outputted together with the attribute information to the learner. Thus, the learner can learn the character strings starting from the one that the learner does not know and in the order of importance. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention records a word learning method and apparatus for enabling a learner to learn words efficiently in a language learning / education scene, and a word learning program and program for use in realizing the word learning method Recording medium, character string learning method and apparatus for enabling learner to efficiently learn character strings such as kanji, and character string learning program and program for use in realizing the character string learning method The recording medium.

  In the conventional word learning device, a technique of randomly extracting a word from the word dictionary storage unit and outputting it to the learner, or a technique of outputting a predetermined word to the learner has been used.

  In addition, a technique has been used in which a word previously searched by a learner is identified as an unknown word, and the word is output to the learner in consideration of the number of searches for those words (see, for example, Patent Document 1). .

  From now on, when learning kanji reading, etc., the same technique is used to extract random kanji from the kanji dictionary storage unit and output it to the learner, or to output predetermined kanji to the learner. Will be.

In addition, a technique is used in which kanji previously searched by the learner is identified as an unknown kanji and the kanji is output to the learner in consideration of the number of times the kanji is searched.
Japanese Patent Laid-Open No. 4-97192

  When a learner learns words in a certain language, it is rational that the learner learns in order from words that the learner does not know and important words, that is, words that are well known by humans who are native speakers of the language. It is efficient. In order to enable this method, it is necessary to accurately estimate words that the learner does not know and to identify the importance of the words in the language.

  However, the conventional technique does not consider these points, so word learning is not rational and efficient.

  For example, in a technique of randomly extracting words from a word dictionary storage unit and outputting them to a learner, neither estimation of a word that the learner does not know nor identification of the importance of a word is considered. Therefore, word learning cannot be performed reasonably and efficiently.

  In addition, for example, in a technique of outputting a predetermined word to a learner, estimation of a word that the learner does not know is not considered. In addition, the technique does not consider the importance of words. Therefore, word learning cannot be performed reasonably and efficiently.

  Further, for example, in the word learning device disclosed in Patent Document 1, the word searched by the learner using the device is specified as an unknown word. However, since it is considered that there are many words that the learner does not know even if the learner does not search, the technique disclosed in Patent Document 1 estimates only a part of the unknown word. Moreover, since the words output to the learner are limited to the words searched by the learner, there is a problem that the words that are not searched cannot be learned.

  In the technique disclosed in Patent Document 1, a word is output to a learner in consideration of the number of word searches. However, even if the number of searches by the learner is large, it is not always an important word in the language. Therefore, it cannot be said that the technique considers the importance of words.

  The problems of the prior art described above are also problems that occur when various kinds of attribute information possessed by character strings such as kanji are learned.

  According to the present invention, for the above problem, based on the estimation of the word unknown to the learner and the identification of the importance of the word, the words that the learner does not know are output to the learner in descending order of importance. The purpose is to provide a new word learning technique that enables rational and efficient word learning.

  Then, the present invention solves the above problem by assigning character strings that the learner does not know to the learners in descending order of importance based on estimation of the character strings that the learner does not know and identification of the importance of the character strings. An object of the present invention is to provide a new character string learning technique that enables rational and efficient character string learning by outputting.

[I] Word Learning Device of the Present Invention To achieve this object, the word learning device of the present invention presents a word stored in a word dictionary and semantic information of the word to a learner, In order to perform a process for learning a word by a learner, (1) an extraction unit that extracts a prescribed number of words and a value indicating the importance of the word stored in association with the word dictionary; ) An estimation means for estimating the learner's vocabulary by presenting the extracted word to the learner and determining whether or not the learner knows the word according to a response input to the learner And (3) setting means for setting extraction conditions for a word to be learned to be presented to the learner based on the determination result of whether or not the learner knows the presented word, and (4) a word dictionary To satisfy the extraction conditions set by the setting means A word and a value indicating the importance of the word stored in association with the word and semantic information of the word are extracted, and according to the order of the value indicating the importance, the learner is notified of the word and the word. Presenting means for presenting meaning information and (5) instruction means for instructing the re-execution of processing to the extracting means when the processing of the presenting means ends.

  In such a configuration, a familiarity that is a value indicating the degree of subjective familiarity of the word may be used as the value indicating the importance of the word.

  The word learning method of the present invention realized by the operation of each of the above processing means can be realized by a computer program, and this computer program is provided by being recorded on an appropriate recording medium such as a semiconductor memory. Alternatively, the present invention is realized by being provided via a network, installed when the present invention is carried out, and operating on a control means such as a CPU.

  In the word learning device of the present invention configured as described above, a prescribed number of words and a value indicating the importance of the word stored in association with the word are extracted from the word dictionary.

  For example, by dividing a range of values indicating the importance of words according to a specified number at equal intervals, and extracting words stored in the word dictionary in association with the divided values as extraction targets, from the word dictionary Then, a prescribed number of words and a value indicating the importance level of the words stored in association with them are extracted.

  Subsequently, these extracted words are presented to the learner, and the learner's vocabulary is estimated by determining whether or not the learner knows the word according to the response input thereto.

  For example, whether or not the learner knows the presented word by collating the meaning information about the presented word input by the learner with the meaning information stored in the word dictionary in association with the presented word By judging the above, the vocabulary of the learner is estimated.

  Subsequently, based on the determination result of whether or not the learner knows the presented word, an extraction condition for the word to be learned to be presented to the learner is set.

  For example, the lowest value among the values indicating the importance of a word judged to be known by the learner is identified, and the importance of the word defined by the identified value and a value lower than that value The section of the value indicating is set as the extraction condition.

  Subsequently, from the word dictionary, a word satisfying the set extraction condition, a value indicating the importance of the word stored in association with the word, and semantic information of the word are extracted, and the order of the value indicating the importance Thus, the learner is made to learn the word by presenting the word and the meaning information of the word to the learner.

  For example, by setting the above-described section as an extraction condition, it becomes possible to extract an important word that is not known to the learner from the word dictionary. By presenting the word meaning information of the word to the learner according to the order of the values indicating the importance, the learner learns the important word that is not known to the learner.

  Then, by repeating this series of processing, processing is performed so that the learner learns words that are not known to the learner and that are important words in descending order of importance.

[Ii] Character string learning device of the present invention To achieve this object, the character string learning device of the present invention provides a character string stored in a character string dictionary and attribute information of the character string to a learner. (1) A value indicating the importance level of the character string stored in association with the specified number of character strings from the character string dictionary in order to perform a process of learning a character string by presenting And (2) presenting the character string extracted by the extraction means to the learner, and determining whether or not the learner knows the attribute information of the character string according to the response input thereto (3) setting means for setting the extraction condition of the character string to be learned to be presented to the learner based on the determination result of the determination means, and (4) setting of the setting means from the character string dictionary The character string that satisfies the extracted conditions and the corresponding character string Extracts the value indicating the importance of the character string and the attribute information of the character string, and presents the character string and the attribute information of the character string to the learner according to the order of the value indicating the importance And (5) instruction means for instructing the extraction means to re-execute the processing when the processing of the presentation means is completed.

  In such a configuration, a familiarity that is a value indicating the degree of subjective familiarity of the character string may be used as a value indicating the importance of the character string.

  The character string learning method of the present invention realized by the operation of each of the above processing means can be realized by a computer program, and this computer program is recorded on an appropriate recording medium such as a semiconductor memory and provided. Or provided via a network, installed when implementing the present invention, and operated on a control means such as a CPU, thereby realizing the present invention.

  In the character string learning device of the present invention configured as described above, a prescribed number of character strings and a value indicating the importance of the character string stored in association with the character string are extracted from the character string dictionary.

  For example, by dividing a range of values indicating the importance of a character string into equal intervals according to a specified number, and extracting a character string stored in a character string dictionary in association with the divided value as an extraction target, A prescribed number of character strings and a value indicating the importance of the character string stored in association with the character string are extracted from the character string dictionary.

  Subsequently, the extracted character strings are presented to the learner, and the learner's knowledge is determined by determining whether the learner knows the attribute information of the character string in accordance with the response input to the learner. Estimate force.

  For example, by comparing the attribute information about the presented character string input by the learner with the attribute information stored in the character string dictionary in association with the presented character string, the attribute information of the presented character string is obtained by the learner. The learner's knowledge is estimated by determining whether or not he / she knows.

  Subsequently, the extraction condition of the character string to be learned to be presented to the learner is set based on the determination result of whether or not the learner knows the presented character string.

  For example, the lowest value among the values indicating the importance of the character string judged to be known by the learner is specified, and the character string defined by the specified value and a value lower than that value is specified. An interval of values indicating importance is set as an extraction condition.

  Subsequently, a character string satisfying the set extraction condition, a value indicating the importance of the character string stored in association with the character string, and attribute information of the character string are extracted from the character string dictionary. The learner is made to learn the character string by presenting the character string and the attribute information of the character string to the learner according to the order of the indicated values.

  For example, by setting the above-described section as an extraction condition, it becomes possible to extract a character string that is not known to the learner and is important from the character string dictionary. The learned character string and the attribute information of the character string are presented to the learner according to the order of the value indicating the importance, thereby allowing the learner to learn the important character string that is not known to the learner. It is.

  Then, by repeating this series of processing, processing is performed so that the learner learns important character strings in the descending order of importance, which are character strings that the learner does not know.

  According to the present invention, rational and efficient word learning that learns in order from words that are unknown to the learner and that are important words, that is, words that are well known by humans whose native language is the language of the word. Can be performed.

  According to the present invention, it is possible to learn in order from a character string that is not known to the learner and an important character string, that is, a character string that is well known by a person whose native language is the language of the character string. In addition, efficient character string learning can be performed.

[I] Word Learning First, the best mode of the present invention applied to word learning will be described.

  The present invention uses the vocabulary number estimation method disclosed by the present inventors in Japanese Patent No. 3331286 to estimate words that the learner does not know and to identify the importance of the words as “ Words that learners do not know by using word intimacy as described in “Vocabulary characteristics of“ Volume 1 ”, Noriaki Amano, Kimihisa Kondo, (Sanseido) 1999 (hereinafter referred to as Reference 1) The main feature is to output in descending order of importance.

  Here, in the vocabulary number estimation method disclosed by the present inventors in Japanese Patent No. 3331286, a predetermined number of words are selected from a group of words to which closeness is given and presented to the subject. The number of subjects' vocabulary (the number of words the subject remembers in their heads) by counting words that are more familiar than those with the lowest familiarity The method of estimating is taken.

  The word familiarity is a numerical value indicating the degree of subjective familiarity, and is expressed by a real value in the range of 1 to 7, for example. The closer to 1, the less familiar, the closer to 7, the more familiar. In order to determine the familiarity of words, there is a method in which the familiarity of each word is rated by a large number of people with numerical values of 1, 2, 3, 4, 5, 6, and 7, and the rating values are averaged. The intimacy obtained by such a method for about 80,000 Japanese words is stored in a database (see Reference 1).

  Word familiarity is widely used not only for academic research related to language, but also for applications such as infant language teaching materials, word comprehension tests for the hearing impaired, and the creation of aphasia test batteries. The word familiarity is also used in the vocabulary number estimation method disclosed by the present inventors in Japanese Patent No. 3331286.

  Important words that everyone knows are always intimate. For example, the familiarity of the words “onion” and “rice ball” is 6.5 or more (see Reference 1). Conversely, unimportant words that are not well known by everyone are always less familiar. For example, the intimacy of the words “Sasame” and “Suiwai” is 2.0 or less (see Reference 1).

  Therefore, it is extremely reasonable and reasonable to use intimacy as a criterion for identifying importance, that is, an important word.

  FIG. 1 illustrates the configuration of the best mode of the word learning device of the present invention.

  As shown in this figure, the word learning device of the present invention sets a word dictionary storage unit 11 that stores a word, a word meaning sentence that describes the meaning of the word, and a closeness of the word, and a word extraction condition. An extraction condition setting unit 16, a word extraction unit 12 that extracts words from the word dictionary storage unit 11 based on the set extraction conditions, a word output unit 13 that presents the extracted words to the learner, and the learner's A reaction collection unit 14 that collects the reaction, and a reaction analysis unit 15 that estimates a word that the learner does not know by analyzing the collected reaction and obtaining a lower limit of the familiarity in the word that the learner knows Have.

  The differences from the prior art are (1) having a reaction collecting unit 14 and a reaction analyzing unit 15 for collecting learner responses, and (2) estimating words that the learner does not know in the reaction analyzing unit 15. And (3) a point having an extraction condition setting unit 16 that sets a word extraction condition based on the estimation result and the word familiarity. It becomes possible to extract and output words in the order of importance.

  The implementation of the present invention thus configured comprises a test stage and a learning stage.

"Test stage"
The test stage is a stage for estimating words that the learner does not know using closeness.

  In the extraction condition setting unit 16, the familiarity is set to a certain section [a, b], the number of extractions is set to n, and the word extraction condition is set. The extraction is performed at almost equal intervals in the interval [a, b]. Since the word dictionary storage unit 11 stores a word, a word meaning sentence describing the meaning of the word, and the closeness of the word, the word (group) whose closeness matches the extraction condition is stored in the word meaning sentence. At the same time, it can be extracted by the word extraction unit 12.

  The extracted word (group) is sent to the word output unit 13 and presented to the learner one by one. Simultaneously with this presentation, the presented word and the meaning sentence of the word are sent from the word output unit 13 to the reaction collecting unit 14. In response to this presentation, the learner answers by a method of answering whether or not the presented word is known or a method of answering the meaning of the word. The learner's answers are collected by the reaction collection unit 14.

  The collected responses are sent to the reaction analysis unit 15 together with the word (group) and the meaning of the words. The reaction analysis unit 15 determines whether or not the learner knows the word. In the case of an answer method for determining whether or not the learner knows, this determination method is based on the answer itself, and in the case of the answer method based on the meaning, the meaning of the meaning is verified and the determination is made based on whether the answer is correct.

  Thereby, a data pair of a word and information on whether or not the word is known is obtained for the word group. The vocabulary number estimation method disclosed by the present inventors in Japanese Patent No. 3331286 is applied to this data pair to calculate the vocabulary number known to the learner. By combining the data pair with the familiarity of each word, the lowest familiarity of the word known to the learner is obtained. If the lowest intimacy value is t, the learner knows words with intimacy values higher than t, but the learner knows words with intimacy values lower than t. It can be estimated that there is no.

"Learning stage"
The learning stage is a stage where the learner learns words based on the output of the word learning device of the present invention.

  By using t obtained in the test stage and a certain positive value c, the extraction condition setting unit 16 sets the range of familiarity as the extraction condition in the section [tc, t], so that the learner It is possible to extract words that are unknown and have high intimacy, that is, words with high importance.

  Based on this condition, the word extraction unit 12 extracts the word (group), the word meaning sentence describing the meaning of the word, and the closeness of the word from the word dictionary storage unit 11. The word output unit 13 sorts the extracted words (groups) in descending order of closeness, and presents the words (groups) and their meaning sentences to the learner. The learner learns the output word.

  Since the total number of words remembered by the learner has increased after learning, the test stage is performed again to estimate the number of words after learning. Based on the estimation, a learning step is performed in which a word that is not known to the learner and has high importance is extracted and learned.

  As shown in FIG. 2, by repeating the test stage and the learning stage described above alternately, it is possible to always output words that are not known to the learner and have high importance. Therefore, rational and efficient word learning is possible.

  If the extraction condition is set so that the number of words known to the learner is zero and all words are extracted without performing the test stage, a word list arranged in order of importance can be obtained. If this is output and printed, a word list that can be learned in order of importance is obtained. It is also possible for the learner to learn independently using this word list.

  As described above, in the present invention, first, the extraction condition setting unit 16 sets a test word extraction condition, and the word extraction unit 12 selects a word from the word dictionary storage unit 11 based on the word extraction condition. By extracting and outputting those words to the learner in the word output unit 13, collecting the learner's reaction to the output in the reaction collecting unit 14, and analyzing the collected reaction in the reaction analyzing unit 15, Estimate words that the learner does not know. Subsequently, a word extraction condition for extracting a word having high importance from the words estimated by the extraction condition setting unit 16 is set, and the word extraction unit 12 selects a word from the word dictionary storage unit 11 based on the word extraction condition. And the word meaning sentence of the word and the familiarity of the word, and the word output unit 13 rearranges the words obtained by the extraction in the order of familiarity and outputs them together with the word meaning sentence to the learner. To learn words.

  According to the present invention, according to the present invention, a rational and efficient learning is performed in order from words that are not known to the learner and important words, that is, words that are well known by humans who are native speakers of the language. You will be able to learn words.

[Ii] Learning Kanji Next, the best mode of the present invention applied to learning Kanji (Kanji reading learning is assumed as a specific example) will be described.

  The present invention uses the kanji memory number estimation method disclosed in Japanese Patent Application Laid-Open No. 2002-312499 by the present inventors to estimate kanji that the learner does not know and to specify the importance of kanji as “ Learners know by using kanji intimacy as described in Japanese Vocabulary Characteristics, Volume 5, Nariaki Amano, Kimihisa Kondo, (Sanseido) 1999 (hereinafter referred to as Reference 2) The main feature is to output in descending order of importance from no kanji.

  Here, in the kanji memory number estimation method disclosed by the present inventors in Japanese Patent Laid-Open No. 2002-312499, a predetermined number of kanji characters are selected from a kanji group to which familiarity (intimacy) is given to the subject. By counting the kanji that is more familiar than the lowest kanji among the kanji groups that were presented and answered “know”, The number of kanji stored in the head is estimated.

  The familiarity of kanji is a numerical value indicating the degree of subjective familiarity, and is expressed by a real value in the range of 1 to 7, for example. The closer to 1, the less familiar, the closer to 7, the more familiar. In order to obtain the familiarity of kanji, there is a method in which the familiarity of each kanji is rated by a large number of people with numerical values of 1, 2, 3, 4, 5, 6, and 7, and the rating values are averaged. The intimacy obtained by such a method for about 6300 kanji characters in Japanese is stored in a database (see Reference 2).

  The familiarity of kanji is used not only in academic research related to language but also in the kanji memory number estimation method disclosed by the present inventors in JP-A-2002-312499.

  The important kanji that everyone knows is always intimate. For example, the familiarity of the Chinese characters “kai” and “hon” is 6.5 or more (see Reference 2). On the other hand, insignificant kanji that no one knows is always intimate. For example, the familiarity of Chinese characters “魑” and “龠” is 2.0 or less (see Reference 2).

  Therefore, it is extremely reasonable and reasonable to use intimacy as a criterion for identifying importance, that is, important kanji.

  FIG. 3 shows the configuration of the best mode of the kanji learning apparatus of the present invention.

  As shown in the figure, the kanji learning device of the present invention includes a kanji dictionary storage unit 21 that stores kanji, reading information of the kanji, and familiarity of the kanji, and an extraction condition setting for setting a kanji extraction condition. A kanji extraction unit 22 for extracting kanji from the kanji dictionary storage unit 21 based on the set extraction condition, a kanji output unit 23 for presenting the extracted kanji to the learner, and collecting learner responses And a reaction analysis unit 25 that estimates the kanji that the learner does not know by analyzing the collected reactions and obtaining the lower limit value of the familiarity in kanji that the learner knows.

  The differences from the prior art are (1) having a reaction collecting unit 24 and a reaction analyzing unit 25 for collecting learner's reactions, and (2) estimating kanji characters that the learner does not know in the reaction analyzing unit 25. (3) a point having an extraction condition setting unit 26 for setting the extraction condition of the Chinese character based on the estimation result and the familiarity of the Chinese character. It becomes possible to extract and output in the order of importance.

  The implementation of the present invention thus configured comprises a test stage and a learning stage.

"Test stage"
The test stage is a stage in which kanji that the learner does not know is estimated using closeness.

  In the extraction condition setting unit 26, the familiarity is set to a certain section [a, b], the number of extractions is set to n, and the kanji extraction condition is set. The extraction is performed at almost equal intervals in the interval [a, b]. The kanji dictionary storage unit 21 stores kanji, reading information of the kanji, and the familiarity of the kanji, so that the kanji (group) whose intimacy matches the extraction condition is read information of the kanji. At the same time, it can be extracted by the kanji extraction unit 22.

  The extracted kanji (group) is sent to the kanji output unit 23 and presented to the learner one by one. Simultaneously with this presentation, the presented kanji and information on the reading of the kanji are sent from the kanji output unit 23 to the reaction collecting unit 24. In response to this presentation, the learner answers by a method of answering whether or not he / she knows the presented kanji or a method of answering the reading of the kanji. The learner's answers are collected by the reaction collection unit 24.

  The collected responses are sent to the reaction analysis unit 25 together with information on kanji (group) and readings of the kanji. The reaction analysis unit 25 determines whether the learner knows kanji. In the case of an answer method for determining whether or not the learner knows, this determination method is based on the answer itself, and in the case of an answer method by reading, the reading is collated and the determination is made based on whether the answer is correct.

  Thus, a data pair of kanji and information on whether or not the kanji is known is obtained for the kanji group. The kanji memory number estimation method disclosed by the present inventors in JP-A-2002-312499 is applied to this data pair to calculate the kanji memory number known to the learner. By combining the data pair with the familiarity of each kanji, the lowest familiarity of the kanji known to the learner is obtained. If this lowest familiarity value is t, the learner knows kanji with a familiarity value higher than t, but the learner knows kanji with a familiarity value lower than t. It can be estimated that there is no.

"Learning stage"
The learning stage is a stage where the learner learns kanji based on the output of the kanji learning device of the present invention.

  By using t obtained in the test stage and a certain positive value c, the extraction condition setting unit 26 sets the range of intimacy as the extraction condition in the section [tc, t], so that the learner It is possible to extract a kanji that is unknown and has a high degree of familiarity, that is, a kanji with high importance.

  Based on this condition, the kanji extraction unit 22 extracts the kanji (group), the reading information of the kanji (group), and the familiarity of the kanji (group) from the kanji dictionary storage unit 21. The kanji output unit 23 rearranges the extracted kanji (group) in descending order of familiarity, and presents the kanji (group) and its reading information to the learner. The learner learns the output kanji.

  Since the total number of kanji remembered by the learner has increased after learning, the test stage is performed again to estimate the number of kanji memories after learning. Based on this value, a learning step is performed in which a kanji that is unknown to the learner and has a high importance is newly extracted and learned.

  As shown in FIG. 4, by alternately repeating the test stage and the learning stage described above, it is possible to always output kanji characters that are not known to the learner and have high importance. According to this, rational and efficient learning of kanji is possible.

  If the extraction condition is set so that the number of kanji known to the learner is zero and all the kanji are extracted without performing the test stage, a kanji list arranged in order of importance can be obtained. If this is output and printed, a kanji list that can be learned in order of importance can be obtained. It is also possible for the learner to learn independently using this kanji list.

  As described above, in the present invention, first, the extraction condition setting unit 26 sets a test kanji extraction condition, and the kanji extraction unit 22 selects a kanji from the kanji dictionary storage unit 21 based on the kanji extraction condition. By extracting and outputting the kanji to the learner in the kanji output unit 23, collecting the learner's reaction to the output in the reaction collecting unit 24, and analyzing the collected reaction in the reaction analyzing unit 25, Estimate kanji that the learner does not know. Subsequently, a kanji extraction condition for extracting a kanji having a high degree of importance from kanji estimated by the extraction condition setting unit 26 is set, and the kanji extraction unit 22 sets a kanji from the kanji dictionary storage unit 21 based on the kanji extraction condition. And the kanji reading information and the familiarity of the kanji, and the kanji output unit 23 rearranges the kanji obtained by the extraction in the order of the familiarity and outputs them to the learner together with the reading information. Let learners learn kanji.

  According to the present invention, according to the present invention, a kanji that is not known to the learner and an important kanji, that is, a kanji that is well known by a person who is familiar with the kanji is learned in order, which is rational and efficient. You will be able to learn kanji.

  Next, an embodiment of the present invention applied to word learning will be described in more detail.

  FIG. 5 illustrates an embodiment of the device configuration of the word learning device 100 of the present invention.

  As shown in this figure, the word learning device 100 of the present invention includes a word dictionary 110 that stores a word, a meaning sentence that describes the meaning of the word, and a closeness of the word, and a word that executes a word learning process. A learning program 120 and a working memory 130 serving as a working memory area for the word learning program 120 are provided, and words that the learner does not know are output in descending order of importance through the dialogue apparatus 300. , Process the learner to learn the word.

  Note that the word dictionary 110 may store words according to the order of alphabets or the like, or may store words according to the order of familiarity.

  6 and 7 show an embodiment of the processing flow executed by the word learning program 120. FIG. Next, according to this processing flow, the word learning processing executed by the present invention will be described in detail.

  When a learning request for a word is issued from the learner, the word learning program 120 starts with the most familiarity that the learner knows in step 10 as shown in the processing flow of FIGS. A word extraction condition for specifying a low word is set. Specifically, by setting the familiarity interval [a, b] and the extraction number n, the word extraction condition for specifying the lowest familiar word known to the learner is set. is there.

  Subsequently, in step 11, based on the set extraction condition, information (words, word meaning sentences, and intimacy) of the number of words specified by the extraction condition is extracted from the word dictionary 110. Specifically, by dividing the set familiarity interval [a, b] into (n + 1) equal intervals, n familiarities are specified, and information on words matching those familiarities is extracted. As a result, information on the number of words specified by the extraction condition is extracted from the word dictionary 110.

  Subsequently, in step 12, when it is determined whether or not all the extracted words have been presented to the learner and it is determined that all the words have not been presented, the process proceeds to step 13 where the extracted words are extracted. Select one of the words that has not been presented. Subsequently, in step 14, the selected word is presented to the learner, and in subsequent step 15, it is determined whether or not the learner knows the presented word according to the answer from the learner. Return to 12.

  Here, regarding the determination of whether or not the learner knows the word presented, if the learner is asked to input a “know” or “don't know” answer, the decision is made based on the answer. Will do. Further, when the learner inputs the word meaning of the presented word, the determination is made by collating with the word meaning sentence obtained from the word dictionary 110.

  In this way, when it is determined in step 12 that all the extracted words have been presented to the learner by repeating the processing of step 12 to step 15, the process proceeds to step 16 and presented to the learner. Identify the least familiar word that the learner knows.

  Subsequently, in step 17, the number of learners' vocabulary is estimated by counting words having an intimacy higher than the intimacy of the identified word among the words stored in the word dictionary 110. That is, the present inventors estimate the number of vocabulary of learners according to the vocabulary number estimation method disclosed in Japanese Patent No. 3331286.

  Subsequently, in step 18, a word extraction condition for learning is set using the familiarity of the word specified in step 16. Specifically, a certain positive value c is set, and the familiarity t of the identified word is used, and a word that falls within the interval whose familiarity is [tc, t] is to be learned. In addition, word extraction conditions for learning are set.

  Subsequently, in step 19, based on the set extraction condition for learning, word information (word, meaning sentence, and intimacy) that satisfies the extraction condition is extracted from the word dictionary 110. That is, information of words whose intimacy is from tc to t is extracted from the word dictionary 110.

  Subsequently, in step 20, the extracted words are rearranged in the order of familiarity. Subsequently, in step 21, it is determined whether or not all the rearranged words have been presented to the learner, and when it is determined that not all the words have been presented, the process proceeds to step 22 where the rearrangement is performed. From the replaced words, one that has not been presented is selected according to the order of arrangement.

  Subsequently, in step 23, the selected word and its meaning text are presented to the learner so that the learner learns the selected word, and then the process returns to step 21.

  In this way, by repeating the processing of step 21 to step 23, it is determined in step 21 that all rearranged words (all words extracted in step 19) are presented to the learner. Then, the process proceeds to step 24, where it is determined whether or not a word learning end request is issued from the learner, and when it is determined that a word learning end request is not issued, the process returns to step 10 to repeat a series of processes. Return.

  On the other hand, when it is determined in step 24 that the learner has issued a word learning termination request, the process proceeds to step 25, where the most learned word possessed by the learned word is stored in the word dictionary 110. The number of words after learning of the learner is estimated by counting words having a higher familiarity than a low familiarity. That is, the present inventors estimate the number of vocabularies after learning by the learner according to the vocabulary number estimation method disclosed in Japanese Patent No. 3331286.

  Subsequently, in step 26, the number of vocabulary before learning of the learner specified by the number of vocabularies initially estimated in step 17 and the number of vocabulary after learning of the learner estimated in step 25 are output and processed. Exit.

  In this way, the word learning program 120 executes the processing flow of FIGS. 6 and 7 so that the learner learns important words that are words that the learner does not know. .

  Next, a specific example of the word learning process executed according to the present embodiment will be described.

  In this word learning process, a word dictionary 110 that stores 70000 words, meaning sentences and intimacy of the words is used.

  First, as a test stage, the word learning program 120 sets word extraction conditions in step 10 described above, with the word extraction range set to a = 1 and b = 7, and the number of extracted words set to n = 50. . Based on this extraction condition, 50 words were extracted from the word dictionary 110 in step 11 described above. Then, in steps 12 to 15 described above, the extracted words are presented to the learner one by one, the learner is made to answer whether or not the presented word is known, and the answers are collected.

  As a result, in step 16 described above, t = 5.1 is obtained as the familiarity of the word with the lowest familiarity that the learner knows, and the estimated value of the number of vocabulary of the learner before learning is obtained from this familiarity. As a result, 26000 words were obtained.

  In the subsequent learning stage, by setting c = 0.1 in step 18 described above, the word extraction condition for learning that is a word whose intimacy is in the interval of [5.0, 5.1] is set. Set. Based on this extraction condition, in step 19 described above, the word whose intimacy is in this section and the word meaning and intimacy of the word are extracted from the word dictionary 110. Then, the extracted words are rearranged in the order of familiarity in the above-described step 20, and the extracted words are presented to the learner together with the word meaning one by one in the above-described steps 21 to 23 for learning.

  When it is determined in step 21 that one learning has been completed, the test stage and the learning stage are performed again by returning to step 10 through step 24, and this is repeated 100 times. .

  The total time required for the word learning process was 20 hours, and the number of vocabularies after learning of the learner obtained in the above-described step 25 was 30000 words. That is, it was confirmed that 4000 words could be learned.

  In order to verify the effectiveness of the present invention, a total of 20 hours of word learning is performed by the conventional technique, and then the vocabulary of the learner is estimated by the vocabulary number estimation method disclosed by the present inventors in Japanese Patent No. 3331286. Number was measured.

  As a result, with the technique of randomly extracting words from the word dictionary 110 and outputting them to the learner, the number of vocabularies increased from 26000 words to 26800 words. In other words, 800 words were learned. Further, in the technique of outputting predetermined words to the learner, the number of vocabulary has increased from 26000 words to 27000 words. In other words, I was able to learn 1000 words. In the technique disclosed in Patent Document 1, the number of vocabulary increased from 26000 words to 26500 words. In other words, 500 words were learned.

  From comparison with these conventional techniques, it was confirmed that when the learning time is the same, the present invention can learn 4 to 8 times as many words as the conventional technique.

  Next, an embodiment of the present invention applied to kanji learning (learning kanji reading as a specific example) will be described in more detail.

  FIG. 8 illustrates an embodiment of the device configuration of the kanji learning device 200 of the present invention.

  As shown in this figure, a kanji learning apparatus 200 according to the present invention includes a kanji dictionary 210 that stores kanji, information about the kanji reading, and the familiarity of the kanji, and a kanji learning program 220 that executes kanji learning processing. And a working memory 230 that serves as a working memory area for the kanji learning program 220, and outputs kanji characters that the learner does not know via the dialogue device 300 in descending order of importance. Process to learn kanji.

  The kanji dictionary 210 may store kanji according to the order of kanji reading or stroke count, or may store kanji according to the order of familiarity.

  FIG. 9 and FIG. 10 show an embodiment of the processing flow executed by the kanji learning program 220. Next, according to this processing flow, the kanji learning process executed by the present invention will be described in detail.

  When a learning request for a kanji is issued from the learner, the kanji learning program 220 firstly, in step 30, as shown in the processing flow of FIG. 9 and FIG. Set the kanji extraction conditions to identify low kanji. Specifically, by setting the familiarity interval [a, b] and the extraction number n, the kanji extraction condition for identifying the kanji with the lowest familiarity known by the learner is set. is there.

  Subsequently, in step 31, based on the set extraction condition, information on the number of kanji characters (kanji, reading and intimacy) specified by the extraction condition is extracted from the kanji dictionary 210. Specifically, by dividing the set familiarity interval [a, b] into (n + 1) equal intervals, n familiarities are specified, and information on kanji characters matching those familiarities is extracted. As a result, information on the number of kanji characters specified by the extraction condition is extracted from the kanji dictionary 210.

  Subsequently, in step 32, it is determined whether or not all the extracted kanji have been presented to the learner, and when it is determined that all the kanji have not been presented, the process proceeds to step 33 to extract the extracted kanji. Select one of the kanji that has not been presented. Subsequently, at step 34, the selected kanji is presented to the learner. At step 35, it is determined whether or not the learner knows the kanji presented by the learner according to the answer from the learner. Return to 32.

  Here, whether or not the learner knows the kanji presented by the learner is determined based on the answer when the learner inputs “know” or “don't know” answers. Will do. Further, when the learner inputs the presented kanji reading information, the determination is made by collating with the reading information obtained from the kanji dictionary 210.

  In this way, when it is determined in step 32 that all extracted kanji characters have been presented to the learner by repeating the processing of step 32 to step 35, the process proceeds to step 36 and presented to the learner. The kanji with the lowest familiarity that the learner knows is identified.

  Subsequently, in step 37, the kanji memory number of the learner is estimated by counting kanji having a closeness higher than the intimacy of the specified kanji among the kanji stored in the kanji dictionary 210. That is, the present inventors estimate the learner's kanji memory number according to the kanji memory number estimation method disclosed in JP-A-2002-312499.

  Subsequently, in step 38, the kanji extraction conditions for learning are set using the familiarity of the kanji specified in step 36. Specifically, a certain positive value c is set, and a kanji that has a closeness of [tc, t] is used as a learning target using the familiarity t of the specified kanji. Then, kanji extraction conditions for learning are set.

  Subsequently, in step 39, based on the set extraction conditions for learning, kanji information (kanji, reading and intimacy) satisfying the extraction conditions is extracted from the kanji dictionary 210. That is, information on Chinese characters whose familiarity falls from tc to t is extracted from the Chinese character dictionary 210.

  Subsequently, in step 40, the extracted Chinese characters are rearranged in the order of familiarity. Subsequently, in step 41, when it is determined whether or not all the rearranged kanji have been presented to the learner, and when it is determined that not all the kanji have been presented, the process proceeds to step 42, where From the replaced kanji, one undisplayed kanji is selected according to the order of arrangement.

  Subsequently, in step 43, the selected kanji and its reading information are presented to the learner, so that the learner learns the selected kanji, and the process returns to step 41.

  In this way, by repeating the processing of step 41 to step 43, it is determined in step 41 that all rearranged kanji (all kanji extracted in step 39) are presented to the learner. Then, the process proceeds to step 44, where it is determined whether or not a kanji learning end request has been issued by the learner. Return.

  On the other hand, when it is determined in step 44 that the learner has issued a kanji learning end request, the process proceeds to step 45, and the kanji stored in the kanji dictionary 210 has the most learned kanji. The number of kanji stored after the learning of the learner is estimated by counting the kanji having a higher familiarity than the low familiarity. That is, the present inventors estimate the kanji memory number after learning by the learner according to the kanji memory number estimation method disclosed in JP-A-2002-312499.

  Subsequently, in step 46, the learner's kanji memory number before learning specified by the kanji memory number estimated first in step 37 and the learner's kanji memory number after learning estimated in step 45 are output. To finish the process.

  In this way, the kanji learning program 220 executes the processing flow of FIGS. 9 and 10 to process the kanji that the learner does not know and that is important for the learner to learn. .

  Next, a specific example of the kanji learning process executed according to the present embodiment will be described.

  In this kanji learning process, a kanji dictionary 210 that stores 6300 kanji characters, reading information of the kanji characters, and closeness is used.

  First, the kanji learning program 220 sets the kanji extraction conditions in the above-described step 30 as a test stage, with the kanji extraction range set to a = 1, b = 7, and the number of extracted kanji characters set to n = 50. . Based on this extraction condition, 50 kanji characters were extracted from the kanji dictionary 210 in step 31 described above. Then, in the above-described step 32 to step 35, the extracted kanji is presented to the learner one by one, the learner is made to answer whether or not the presented kanji is known, and the answers are collected.

  As a result, in step 46 described above, t = 5.1 is obtained as the familiarity of the kanji with the lowest familiarity known to the learner, and the kanji memory number before learning of the learner is estimated from the familiarity. A value of 600 characters was obtained.

  In the subsequent learning stage, by setting c = 0.1 in step 38 described above, the kanji extraction condition for learning kanji that falls within the interval whose familiarity is [5.0, 5.1] is set. Set. Based on this extraction condition, in the above-described step 39, the kanji whose intimacy is in this section, the reading information of the kanji, and the intimacy are extracted from the kanji dictionary 210. In step 40, the extracted kanji characters are rearranged in the order of familiarity, and in steps 41 to 43, the extracted kanji characters are presented to the learner together with the reading information one by one. .

  When it is determined in step 41 described above that one learning has been completed, the test stage and the learning stage are performed again by returning to step 30 via step 44, and this is repeated 50 times. .

  The time required for this kanji learning process was 10 hours in total, and the number of vocabulary after learning of the learner obtained in the above-mentioned step 45 was 800 characters. That is, it was confirmed that learning of 200 characters could be realized.

  In order to verify the effectiveness of the present invention, kanji learning is performed for a total of 10 hours by a conventional technique, and then the learner uses the kanji memory number estimation method disclosed in Japanese Patent Application Laid-Open No. 2002-312499. The number of stored kanji was measured.

  As a result, in the technique of randomly extracting Chinese characters from the Chinese character dictionary 210 and outputting them to the learner, the number of stored Chinese characters has increased from 600 characters to 650 characters. In other words, I was able to learn 50 characters. In the technique of outputting predetermined kanji to the learner, the number of stored kanji has increased from 600 to 700. That is, 100 characters were learned.

  From comparison with these conventional techniques, it was confirmed that when the learning time is the same, the present invention can learn twice or four times as many kanji characters as the conventional technique.

  Although the present invention has been described with reference to the illustrated embodiments, the present invention is not limited thereto. For example, although familiarity is used in the embodiment, it is possible to use appearance frequency or other reasonable variables as importance instead of familiarity.

  The present invention can be generalized as a method for learning a character string and attributes (characteristics) of the character string.

  In other words, instead of a combination in which the character string is “word” and the character string attribute is “meaning (meaning)”, or a combination in which the character string is “kanji” and the character string attribute is “reading”, for example, a character If the string is "word" and the string attribute is "reading", the string is "verb" and the string attribute is "verb honorific form", or the string is "word" When the column attribute is “corresponding foreign language word”, the character string is “Kanji” and the character string attribute is “stroke order”, or the character string is “Kanji” and the character string attribute is In case of “Meaning”, the character string is “Hiragana” or “Katakana” and the character string attribute is “Stroke”, or the character string is “Hiragana” or “Katakana” and the character string attribute is “ The present invention can also be implemented in combinations such as “reading”.

It is a figure which shows the structure of the best form of the word learning apparatus of this invention. It is a figure which shows the algorithm of the word learning of this invention. It is a figure which shows the structure of the best form of the Chinese character learning apparatus of this invention. It is a figure which shows the algorithm of the kanji learning of this invention. It is one Example of the apparatus structure of the word learning apparatus of this invention. It is one Example of the processing flow which a word learning program performs. It is one Example of the processing flow which a word learning program performs. It is one Example of the apparatus structure of the Chinese character learning apparatus of this invention. It is one Example of the processing flow which a Chinese character learning program performs. It is one Example of the processing flow which a Chinese character learning program performs.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11 Word dictionary memory | storage part 12 Word extraction part 13 Word output part 14 Reaction collection part 15 Reaction analysis part 16 Extraction condition setting part 21 Kanji dictionary storage part 22 Kanji extraction part 23 Kanji output part 24 Reaction collection part 25 Reaction analysis part 26 Extraction conditions Setting section

Claims (20)

In the character string learning method for causing the learner to learn the character string by presenting the character string stored in the character string dictionary and the attribute information of the character string to the learner,
A process of extracting a specified number of character strings and a value indicating the importance of the character string stored in association with the character string dictionary;
Presenting the extracted character string to the learner and determining whether or not the learner knows the attribute information of the character string according to a response input thereto;
Based on the determination result, a process of setting extraction conditions for a character string to be learned to be presented to the learner,
From the character string dictionary, a character string satisfying the extraction condition, a value indicating the importance of the character string stored in association with the character string, and attribute information of the character string are extracted, and the order of the values indicating the importance And providing the learner with the character string and the attribute information of the character string.
Character string learning method. The character string learning method according to claim 1,
In the above setting process, the lowest value among the values indicating the importance of the character string judged to be known by the learner is specified, and defined by the specified value and a value lower than the specified value. To set the interval of the value indicating the importance as the extraction condition,
Character string learning method. In the character string learning method according to claim 1 or 2,
In the extraction process, a range of values indicating the importance is divided at equal intervals according to the specified number, and a character string stored in the character string dictionary in association with the divided value is extracted as an extraction target. That
Character string learning method. The character string learning method according to any one of claims 1 to 3,
It is configured to use a closeness that is a value indicating the degree of subjective familiarity of the character string as the value indicating the importance,
Character string learning method. In a character string learning device that causes a learner to learn a character string by presenting the character string stored in the character string dictionary and attribute information of the character string to the learner,
Means for extracting a prescribed number of character strings and a value indicating the importance of the character string stored in association with the character string dictionary;
Means for presenting the extracted character string to the learner and determining whether or not the learner knows the attribute information of the character string according to a response input thereto;
Based on the determination result, a means for setting the extraction condition of the character string to be learned to be presented to the learner,
From the character string dictionary, a character string satisfying the extraction condition, a value indicating the importance of the character string stored in association with the character string, and attribute information of the character string are extracted, and the order of the values indicating the importance And a means for presenting the character string and attribute information of the character string to the learner,
Character string learning device characterized. The character string learning device according to claim 5,
The setting means specifies the lowest value among the values indicating the importance of the character string determined to be known by the learner, and is defined by the specified value and a value lower than the specified value. To set the interval of the value indicating the importance as the extraction condition,
Character string learning device characterized. In the character string learning device according to claim 5 or 6,
The extracting means divides the range of values indicating the importance according to the specified number at equal intervals, and performs extraction using a character string stored in the character string dictionary associated with the divided value as an extraction target. That
Character string learning device characterized.   The character string learning program for making a computer perform the process used for implement | achieving the character string learning method of any one of Claim 1 thru | or 4.   A recording medium on which is recorded a character string learning program for causing a computer to execute processing used to implement the character string learning method according to claim 1. In a word learning method for causing a learner to learn a word by presenting the word stored in the word dictionary and the meaning information of the word to the learner,
Extracting a predetermined number of words and a value indicating the importance of the words stored in association with the word dictionary;
Presenting the extracted word to the learner, and determining whether the learner knows the word according to a response input to the learner, and estimating the learner's vocabulary power;
Based on the above determination result, a process of setting extraction conditions for a word to be learned to be presented to the learner,
A word satisfying the above extraction condition and a value indicating the importance level of the word and meaning information of the word stored in association with the extraction condition are extracted from the word dictionary, and the learner is determined according to the order of the value indicating the importance level. A process of presenting the word and meaning information of the word,
Characteristic word learning method. The word learning method according to claim 10,
In the setting process, the lowest value among the values indicating the importance of the word judged to be known by the learner is specified, and the value is defined by the specified value and a value lower than the specified value. Setting the section of the value indicating the importance as the extraction condition,
Characteristic word learning method. The word learning method according to claim 10 or 11,
In the extraction process, the range of values indicating the importance is divided into equal intervals according to the specified number, and extraction is performed using words stored in the word dictionary in association with the divided values as extraction targets. ,
Characteristic word learning method. The word learning method according to any one of claims 10 to 12,
In the above estimation process, the learner knows the presented word by comparing the meaning information about the presented word input by the learner with the meaning information stored in the word dictionary in association with the presented word. To determine whether or not
Characteristic word learning method. The word learning method according to any one of claims 10 to 13,
Providing a process of instructing re-execution of the process of the extraction process when the process of the presenting process is completed;
Characteristic word learning method. The word learning method according to any one of claims 10 to 14,
It is configured to use a closeness that is a value indicating the degree of subjective familiarity of the word as the value indicating importance.
Characteristic word learning method. In a word learning device that causes a learner to learn a word by presenting the word stored in the word dictionary to the learner and the meaning information of the word,
Means for extracting from the word dictionary a prescribed number of words and a value indicating the importance of the words stored in association therewith;
Means for estimating the learner's vocabulary by presenting the extracted word to the learner and determining whether or not the learner knows the word according to a response input thereto;
Based on the determination result, means for setting a word extraction condition to be learned to be presented to the learner,
A word satisfying the above extraction condition and a value indicating the importance level of the word and meaning information of the word stored in association with the extraction condition are extracted from the word dictionary, and the learner is determined according to the order of the value indicating the importance level. A means for presenting the word and meaning information of the word,
Feature word learning device. The word learning device according to claim 16, wherein
The setting means is defined by specifying the lowest value among the values indicating the importance of the word judged to be known by the learner, and a value lower than the specified value. Setting the section of the value indicating the importance as the extraction condition,
Feature word learning device. The word learning device according to claim 16 or 17,
The extracting means divides the range of values indicating the importance according to the specified number at equal intervals, and performs extraction using words stored in the word dictionary in association with the divided values as extraction targets. ,
Feature word learning device.   The word learning program for making a computer perform the process used for implementation | achievement of the word learning method of any one of Claims 10 thru | or 15.   The recording medium which recorded the word learning program for making a computer perform the process used for implementation | achievement of the word learning method of any one of Claims 10 thru | or 15.

Images