JP5384298B2 - Vocabulary learning method executed by computer, vocabulary learning apparatus, vocabulary learning program, and computer-readable recording medium - Google Patents

Description

  The present invention relates to a learning method and a learning apparatus executed by a computer.

  Usually, it takes many years and great efforts of language learners to acquire the vocabulary amount necessary for efficient communication in a foreign language or a second language.

  Reading over a wide area is one way to learn linguistic abilities such as grammar and vocabulary by reading a large number of originals at an appropriate level of difficulty.

  It is said that language users need to learn a vocabulary of 15000 to 20000 words in order to read with little trouble from unknown vocabulary (I. Nation in Learning Vocabulary in Another Language published by Cambridge University Press) , 2001).

  There is limited time for learners to learn new vocabulary. It is therefore necessary to use this time as efficiently as possible.

  Words or phrases in a language (or subject area such as physics) differ in their utility. In general, a word with a high frequency is high in utility, and a word with a low frequency is low in utility. However, utility may vary depending on the specific subject area and curriculum.

  In this application, when the term “word” is used, it means a word, phrase, term, or other vocabulary unit. Vocabulary acquisition (or learning) means the process of acquiring vocabulary units and their related pronunciation, writing, concepts, or other relationships. Therefore, vocabulary acquisition (or learning) is a standard task related to language learning, or a more general educational task acquired from language materials such as textbooks or videos in physics or geography. Can mean.

  The learner should not spend time learning the less effective words before learning the more effective words. For example, if a learner lacks knowledge of a basic term, the learner may face problems with understanding the context in which the advanced term is used, so Acquisition is even more difficult. The existence of such a gap in knowledge makes it difficult to connect new knowledge with existing knowledge. As a result, a cognitive load (cognitive load) is applied and causes frustration.

  Cognitive load is a term used in the field of psychology, meaning the collective of what is required of a person's mind when performing tasks such as language learning, logical thinking in language, or general problem solving To do. If the cognitive load is too high, it can be difficult or even impossible to perform tasks intensively. For example, when reading a book, trying to memorize the meanings of many words that are not very familiar can be expensive and make it difficult to understand the text. Therefore, it is beneficial to reduce unnecessary cognitive load.

  It is difficult for a learner to determine whether a particular word has high or low utility. Furthermore, since a certain word is a keyword in the language teaching material in use, the word teaching material may have a relatively high utility. Therefore, in order to understand the language teaching material, it is necessary to first understand the word. Thus, utility depends in a sense on the context of the language teaching material in use. In addition, the current ability of the learner will change as a result of learning. Thus, the utility of words can vary with time and the learner and other factors.

  The ideal system for vocabulary acquisition is to inform words with high utility in the context of language teaching materials and in the learner's current knowledge, and also to allow the learner to time a possible learning act. It is highly recommended.

  Traditionally, the vocabulary acquisition process has been managed through the teacher and curriculum, or by the learner himself, and managing the task itself is clearly a burden.

  When a process is managed by a teacher or curriculum, the extent to which individual weaknesses can be determined and corrected is extremely limited. The learner may lack knowledge, and if that cannot be resolved, the learner's future progress will be hindered.

  A ranked vocabulary list is an important source of information for managing vocabulary acquisition because frequency is associated with utility. Prior art includes systems that use vocabulary ranked by frequency of occurrence (GB2352543A, US7108512, US20070269778A1, WO06121542A2, US20060063139A1, US6726486, JP2000047559A2). Some of these systems do not include any readers or language teaching materials, but are based solely on flashcard initiatives (activities) or quizzes (GB2352543A, US7108512, US20060063139A1, US6726486, JP2000047559A2). .

  None of the above systems, including a reading unit and a ranked vocabulary list, clearly recommends whether a particular word should be learned immediately or left behind.

  For example, US20070269778A1 displays the difficulty level and frequency of words selected by the learner in the reading unit. However, it does not actually recommend whether each given learner should learn a particular word now.

  WO06121542A2 generates a preferential target vocabulary learning list that is used for other learning components. The method described in WO06121542A2 does not analyze the learning material that the learner is currently reading when generating the preferential target vocabulary learning list. WO06121542A2 does not use a target word list that clearly indicates to a particular learner whether a particular word should be learned.

  Some prior art systems give clear recommendations but do not do so in a context sensitive manner.

  Follow You! The system (Chi-Chiang Shei. FollowYou! An automatic language lesson generation system. CALL, 14 (2), 2001) is a system based on a ranked vocabulary list. The learner's knowledge is indicated by a pointer to a rank in the list. The system presumes that the learner knows all the words with higher ranks. Learner Follow You! When a certain text is opened, the system selects a number of words present in the text and currently ranked below the learner's knowledge rank as words to learn. The only characteristic of text that is considered when selecting a learning word is whether the word is present or absent. Follow You! Does not consider contextual characteristics such as local frequency and semantic filed membership in imported text when selecting words to learn. Words that are not in the textbook in use and that are higher than the lowest-ranked practice word are assumed to be known thereafter. This approach creates deficiencies in the learner's vocabulary knowledge. Follow You! Lists the learning words on its user interface and takes the recommended form. However, as mentioned above, Follow You! Does not take into account the contextual properties of the text in use. Furthermore, the recommendation cannot be updated at an appropriate timing, for example, immediately after the learner has verified the knowledge of a certain word.

  REAP System (Heilman, M. & Eskenazi, M. (2006). Language Learning: Challenges for Intelligent Tutoring Systems. Proceedings of the Workshop of Intelligent Tutoring Systems for Ill-Defined Domains. 8th International Conference on Intelligent Tutoring Systems Select the appropriate reading materials for each learner based on their profile and curriculum profile, which is a vocabulary list ranked by frequency of occurrence. Track active and passive vocabulary by calculating how often learners have learned about a particular word and how often a learner has touched a particular language. Based on one profile, REAP can be used for 2 to 4 learning words (ie Read out short sentences that contain a certain percentage of words that the system thinks that the learner has learned (words that the learner has not yet learned) (J. Brown and M. Eskenazi. (2005). Student, text and Proceedings of the IASTED International Conference on Human-Computer Interaction 2005. Phoenix, AZ.) However, REAP does not use a method that is sensitive to the context of utility. Are not selected based on the read text, they are pre-selected REAP is a learner within a given curriculum profile (vocabulary list ranked by frequency of occurrence). The learning word is selected by prioritizing the word with the highest frequency and the word with the lowest frequency that has not yet fully understood. When the text is presented to the learner, to take the recommendations of the form, the words to be learned is highlighted. And because the profile is not updated until the learner has read the read text and finished the word practice, the status of the word cannot be changed immediately (J. Brown and M. Eskenazi. (2004). Retrieval of authentic documents for reader-specific lexical practice. Proceedings of InSTIL / ICALL Symposium 2004. Venice, Italy).

  Therefore, Follow You! Both the system and the REAP system consider the learner's current knowledge state (when the text is first accessed), but do not use the context of the text in use in determining word utility. Furthermore, both systems cannot recommend learning actions immediately or at a favorable timing in response to constantly changing learner knowledge.

  There is a need for a method of determining the current word utility according to the language teaching material in use and the learner's current knowledge state and changing knowledge state.

  In the first embodiment of the present invention, the following vocabulary learning method is realized. That is, in a vocabulary learning method executed by a computer after selecting a text for transmitting information to a learner, the text includes a plurality of words, and at least one of the plurality of words in the text is included. Assigning a status to the word based on the first index, the second index, and the third index for the word, and transmitting the assigned status to the learner at least upon request And wherein the assigned status is selected from a plurality of predetermined statuses, each status associated with at least one corresponding recommended action taken by the learner. The first index depends on the learner, and the second index is the text. Is an independent relation to the door, the third indicator is dependent on the other words in the above text.

  The assigning step determines a utility value related to context agility of the word based on the second index and the third index, and assigns the status based on the utility value related to the context agility. And 2 allocation steps.

  The determining step may further be configured to determine a utility value related to the context sensitivity based on the first index.

  The assignment step may be configured to assign a status of “study” indicating that a learning action is recommended to a predetermined number of words determined to have the highest utility value regarding the context sensitivity.

  The assigning step may be configured to assign a status of “study” indicating that a learning action is recommended to a word whose utility value relating to the context sensitivity is higher than a predetermined threshold.

  The assignment step may be configured to assign a status of “ignore” indicating that a learning action is not recommended to a word to which a status of “study” is not assigned.

  The second index may have a configuration independent of the learner.

  The second index may be a general utility function of the word.

  The above general utility is the word frequency, concreteness, word difficulty, word region structure, personal learning list structure, imageability (the meaning of the word can be easily imagined in the head) It can be based on any one or more of the importance to a subject or curriculum.

  The second indicator may be zero for all words that are not of a particular type, such as nouns, or otherwise for the general utility of the word.

  The third indicator may be a function of the utility of the word within the context of the text.

  The utility of the word in the context of the text may be determined based on the frequency of the word in the text.

  The utility of the word within the context of the text may be determined based on the relative frequency of the word in the text compared to the overall frequency of the predetermined range. . Here, the predetermined range is, for example, all the words of a certain language sample in the text. The sample is selected so as to include many typical words from the language, and is, for example, listed in a dictionary.

  The third index may be configured to be a function of the local importance of the word in the text.

  Each word is associated with a tag attached to indicate a part of a word group related to a common idea, and the third index indicates the frequency of the tag associated with the word in the text. The configuration may be a function.

  The assigning step may be configured to assign a status so as to limit the number and / or density of words to which a “study” status indicating that a learning action is recommended.

  The first index may be a function of knowledge about the word that the learner has.

  The first index may be configured to indicate the degree or degree of acquisition of the word by the learner.

  The first index may be related to the number of times the learner has seen or read the word.

  The first index effectively applies a distribution, such as a Gaussian distribution, depending on the learner's level or degree of acquisition of the word, and the distribution potentially selects a word to learn. You may be comprised so that it may narrow down to a useful narrow range of words.

  It may be configured to include a second determination step of determining from the first index a Boolean value indicating whether or not the learner has learned the word.

  For at least some of the words in the text for which the first index is determined to be classified into a predetermined category, the assigning step is subordinate to the first index, the second index and the second index The above-described status may be assigned independently of the three indexes.

  The predetermined category may include a word that is considered to be already learned by the learner.

  The allocation step is a configuration in which a word determined to belong to the predetermined category is assigned a “mastered” status indicating that a learning action is not required or that review is recommended. It's okay.

  The assigning step may include at least one of the words when a signal indicating that one or more of the first index to the third index has changed or that there is a possibility that the index has changed is transmitted. The status may be reassigned to the part.

  The transmission of the text to the learner may include a display of the text on a display, provision of the text in a printed state, and / or an audio display of the text.

  According to the second aspect of the present invention, the vocabulary learning device having means for processing at least a part of a word included in a text selected for transmitting information to a learner includes a first index for the word, An assigning means for assigning a status to the word based on a second index and a third index; and a communicating means for transmitting the assigned status to the learner at least in response to a request. Is selected from a plurality of predetermined statuses, each of the statuses being associated with at least one corresponding recommended action taken by the learner, and the first indicator is assigned to the learner. The second index is independent of the text, and the third index is another value in the text. It is dependent on the word.

  According to the third aspect of the present invention, when loaded into a control program for controlling an apparatus that executes the method according to the first aspect of the present invention, or when loaded into the apparatus, the apparatus becomes the second aspect of the present invention. A program for providing such a device is provided. The program may be executed on a carrier medium. The carrier medium may be a recording medium. The carrier medium may be a transmission medium.

  According to a fourth aspect of the present invention, there is provided an apparatus programmed according to a program according to the third aspect of the present invention.

  According to a fifth aspect of the present invention, there is provided a recording medium recording a program according to the third aspect of the present invention.

  Embodiments of the present invention provide context-specific wording in language materials that the learner is currently using to display recommended learning behaviors to the learner as soon as the learning system changes. Provide a learning system that can immediately determine the utility of.

  In this application, when the term word is used, it should be reconfirmed to mean a word, phrase, term, or other vocabulary unit.

  In a preferred embodiment, a language learning device such as an electronic book reading device including a learner model and a vocabulary model as disclosed in the method and apparatus according to GB 0702298A0 is modified to accommodate the system.

  The system combines information from an analysis of the learner's model, vocabulary utility model, and the language teaching material in use. The system can re-determine the utility of words as soon as changes occur in the learner model, vocabulary model, or language teaching in use. Changes can occur as a result of learner actions or internal processing of the system.

  Based on the utility of the word, the system can make recommendations to the learner about possible learning behaviors such as whether a particular word should be learned now or later.

  The system is context sensitive in order to take into account the specific characteristics of the language material currently being learned and the current (estimated) knowledge state of the learner. The characteristics of these language teaching materials include, but are not limited to, local word frequency, importance of local words, and semantic relationships between words in the language teaching materials.

  The learner model for a particular learner tracks each word in the word list for whether or not the learner has learned a particular word, and the learner performs actions such as reading the word in context in the system. It can be updated each time it is done.

  The vocabulary model associates each learner with the general utility of that word for all learners.

  Embodiments of the invention have one or more of the following advantages.

  The advantage of the system is that it reduces cognitive load and makes vocabulary acquisition more efficient and effective.

  A further advantage of the system is that recommendations can be expressed, i.e. the status of any or all words in the language material can be clearly shown to the learner.

  A further advantage of the above method is that it can immediately reflect changes in the referenced information source. In particular, changes in the learner model are immediately considered in order to adapt to the choice of learning words (and the above general status of specific words).

  A further advantage of the above method is that it can fill in the vocabulary knowledge deficiencies in certain learners.

  A further advantage of the above method is that it works for all kinds of language teaching materials (speech or document), and for all disciplines such as physics and history, or in all areas including the vocabulary that the learner needs to learn. It is in.

1 is a block diagram illustrating a preferred embodiment. It is a flowchart which shows the status determination process of a word. The example of the text displayed on a text reading interface is shown. It is an excerpt of the frequency of words in a text example. It is an example of a learner model. An excerpt of the vocabulary model. An excerpt of the ranking of unlearned words according to context sensitive utility. A text example of the highlighted learning word is shown. An alternative display method of learning words is shown. A method of displaying the word status “IGNORE” will be described. The display method of the word status of MASTERED is shown. An excerpt of the ranking of unlearned words based on context sensitive utility measurements that take into account the semantic domain structure.

  The preferred embodiment of the present invention provides a timely word status determination and notification mechanism in a device based on language learning and especially reading for vocabulary learning.

  FIG. 1 is a block diagram showing a configuration of a preferred embodiment.

  A device for language learning, in particular a device for vocabulary learning, comprises a text reading interface 100. The text reading interface 100 displays the text 130 being used. The apparatus includes a learner model 110 that stores vocabulary knowledge of a particular learner. The apparatus includes a vocabulary model 120 that associates words with general utility values. Word status determination unit 140 has access to learner model 110, vocabulary model 120, text 130 in use, and optionally information source 150. The word status determination unit 140 determines the word status and notifies the word status notification unit 160 of the word status. The word status notification unit 160 controls the display of the word status on the text reading interface 100.

  One skilled in the art will appreciate that an apparatus for language learning may include additional configurations and that the configurations can communicate with each other in a manner not explicitly shown in FIG. . Those skilled in the art may implement the configuration shown in FIG. 1 as individual configurations, or implement some of them, or a combination of all of them, as a single configuration. You will understand that.

  FIG. 3 is a diagram illustrating an example of a text read interface 100 referred to below.

  The function of the configuration shown in FIG. 1 will be described in detail.

  Text read interface 100 serves to display various electronic actions that can be envisioned, including, but not limited to, moving between a page and a selected word. When the text is read into the text reading interface 100, a signal related to the event is transmitted.

  The learner model 110 stores an evaluation of the learner's word acquisition level. Learner modeling is well known in the prior art, and any suitable learner model can be used in a preferred embodiment. Since it is not possible to directly observe whether or not the learner has actually learned a word, the conventional learner model evaluates the degree of acquisition of a specific word based on the evidence obtained from the interaction between the system and the learner. In a preferred embodiment, the learner model stores the probability (probability value) that a particular learner has mastered a particular word. Numeric values such as probabilities are converted to Boolean values (ie true / false) by applying a threshold. For example, probabilities greater than 0.8 are converted to true (ie, the learner has already mastered this word) and other probabilities are converted to false. FIG. 5 shows an example of a learner model at a certain time in a table.

  The learner model 110 receives an inquiry from the word status determination unit 140. That is, when the learner model 110 receives an inquiry about a certain word, the learner model 110 returns an evaluation degree of the learner's word acquisition degree regarding the word. As soon as any change occurs from the initial state, the event signal is transmitted. For example, the state of the learner model can change when the learner reads a word within the context of the text in use, or when the learner demonstrates knowledge by answering correctly in a vocabulary test.

  The evaluation level of the learner's word acquisition level returned by the learner model 110 is considered to be a first index for the word. The first index is used to determine a status (eg, “STUDY”, “IGNORE”, “MASTERED”, etc.) of at least a part of the word. Since the first index is influenced by the learner's word acquisition level, it can be said that the first index depends on the learner.

  The vocabulary model 120 associates a word with a numerical value as a general utility. The model can include any word list deemed useful for learning. In a preferred embodiment, the model includes only open-class words such as nouns, verbs, and adjectives. However, the model can be designed to support a specific language curriculum, for example, to learn such words by including only words related to animals and plants. Alternatively, for example, when learning a physics subject, the model may include terminology relating to motion, gravity, and the like. The vocabulary model 120 is a vocabulary list ranked by frequency of occurrence, or word frequency, specificity, word difficulty, word region composition, personal learning list composition, imageability (word Ranked by a combination of either the importance to a subject (for example, the importance to physics) or the importance to a curriculum. The general utility value (utility value) of each word is determined by referring to the vocabulary list or other factors not limited to this list. The vocabulary model 120 may be selected from a plurality of vocabulary models depending on the interest of the learner, the course that the learner takes, and the like.

  The general utility of a word provided by the vocabulary model 120 is considered to be the second index for that word. This second indicator is used, at least in part, to determine the status of the word (eg, STUDY, IGNORE, MASTERED, etc.). Since the second index relates to the general utility of words in a broader sense than the selected specific text, it can be said that the second index is independent from the text.

  The word status determination unit 140 can query the vocabulary model 120 for the general utility of any given word. As soon as any change in the internal state of the vocabulary model 120 occurs, an event signal can be sent. For example, when the curriculum that is the basis of the model is changed by the teacher, the state may be changed. FIG. 6 is a table showing an example of the vocabulary model.

  The word status determination unit 140 may expand the general utility by referring to additional information sources 150 and taking into account additional factors such as semantic domains. This will be described in detail below.

  The word status determination unit 140 determines a word status by performing a word status determination process using the learner model 110. The word status determination step determines the status of words in the text 130 using analysis of the learner model 110, the vocabulary model 120, and the text 130 in use. This process is described in detail below.

  The first two of the information sources (learner model 110 and vocabulary model 120) used to determine the status of words in text 130 are associated with a first index and a second index, respectively. The third source (analysis of text 130 in use) is considered to be the third indicator. The third indicator is used, at least in part, to determine the status of the word (eg, STUDY, IGNORE, MASTERED). It can be said that the third index depends on other words of the text because it is based on part of the remaining text or on the entire analysis.

  In the preferred embodiment, the word status values include, but are not limited to, MASTERED, STUDY, and IGNORE, respectively, that the learner has already mastered and needs to learn positively. Or should be ignored. A further possible status value is ASSUMED_KNOWN, which corresponds to a word that is assumed to be known by the learner without actively learning through a reading-based language learning device. After determining the word status, the word status determination unit 140 notifies the word status notification unit 160 of the status.

  The word status notification unit 160 includes means for displaying the status of each word displayed on the text reading interface 100 to the learner, thereby making clear recommendations. In the preferred embodiment, words with a status of STUDY are recommended that the learner should currently learn the word, eg some visuals such as reverse video, bold typeface, different text color, etc. Highlight the word by means. Words with other statuses (IGNORE, MASTERED) are not highlighted to make recommendations that the learner should not learn the word. FIG. 8 shows an example of a text readout interface that is visually highlighted.

  Other types of learning practices recommended include reviewing words, listing words for later learning or review, associating personal notes or drawings with words, information about words (eg, definitions or personal Memo) can be included.

  Other types of displays include displays such as pop-up information boxes (see below), sounds (see below), changing the shape of the mouse cursor, and displaying word status in designated areas on the screen.

  FIG. 2 is a flowchart showing a word status determination process performed by the word status determination unit 140.

  When a signal related to the event is received in step 200, the word status determination process starts. Here, the event is that the text 130 is loaded into the text reading interface, the learner model 110 is changed, the vocabulary model 120 is changed, and the learner selects a word in the text reading interface 100. Or some other state change. One skilled in the art will understand that event signaling is just one possible way to initiate the process. Other methods include directly requesting the above functions performed by other components or message passing mechanisms.

  The second step 210 obtains words that are present in the text 130 being used.

  Step 220 assigns MASTERED to all words already learned by the learner in the text in use according to learner model 110.

  Step 225 obtains word information from the text 130 in use. The frequency of words in the text 130 in use can indicate, for example, how many times an opportunity is given to learn a particular word. The above steps can perform a linguistic analysis of the text and divide the text into vocabulary units such as words. (Note again that the term word represents a word, phrase, term, or other lexical unit in the text.) The above steps can also convert the words in the text into a standard expression, such as the original or quoted form. This is consistent with the format used for vocabulary model 120 and learner model 110. Linguistic analysis may include stemming, POS tagging (part-of-speech tagging), named entity recognition, phrase recognition, or other steps not limited to this list. Finally, in the preferred embodiment, the frequency analysis is performed by calculating the number of times each unique word appears in the text. FIG. 4 shows an example of the frequency of words obtained from text. Other types of word information can also be collected. For example, the importance of a word in the text in use can be determined by using known statistical means such as mutual information or t-score and the relative frequency of words in the text and its frequency in common languages. Can be computed by comparing.

  Step 230 ranks the remaining words by context sensitive utility values. This value is calculated according to at least two other values: general utility and frequency of occurrence in text 130. In the preferred embodiment, the function is defined as follows:

CSU (w) = k × GU (w) × Freq (w)
GU (w) is the general utility of the word w based on the vocabulary model 120, Freq (w) is the frequency of words in the text 130 calculated in step 210, and k is a constant. FIG. 7 shows an example of the ranking generated by step 230.

  More generally, the function is defined as follows:

CSU (w) = k * f (w) ^a * g (w) ^b * h (w) ^c
f is a function of the learner model (above “first indicator”), g is a function of general utility (above “second indicator”), and h is a text analysis in use (above “third indicator”). Index "). a, b, and c are constant weighting factors, and k is a normalization constant. For example, in one variation, g assigns a value of zero or a general utility value to all words that are not nouns. In one variation, f returns a direct estimate that the learner has learned the word w. In this case, the context sensitive utility value is not a simple Boolean value represented by the distinction between mastered and unskilled, but takes into account the degree of mastery. In another modification, the f function may use a learned value distribution such as a Gaussian distribution. This has the effect of narrowing the CSU value to a narrower range of potentially useful words. In another variation, the f function includes the number of times the learner has viewed (or read) the word w. In another variation, the h function returns the local importance of the word w in the text in use. The constants a, b, c and k are adjusted by experiment.

  Step 240 assigns a status of STUDY to a predetermined number of words ranked higher in the ranking list, or to all words having a context sensitive utility value that exceeds a predetermined threshold.

  Step 250 assigns an IGNORE status to the rest of the ranked list.

  Finally, step 260 notifies the word status notification unit 160 of the assignment of word status.

  The word status determination step will be described using an example. FIG. 3 shows a relatively long one-page example sentence (130) displayed on the text reading interface 100. FIG. 7 is an example of the ranking generated in step 230 using the text example of FIG. 3, the learner model example of FIG. 5, and the vocabulary model example of FIG. The post-rankings ranked in the top 20 are shown.

  In FIG. 7, the word with the highest context-sensitive utility value in the example text is “say”. “Man” has a lower general utility than, for example, “year”, “time”, and “make”, but the text 130 has a relatively high frequency. Higher context sensitive utility than them.

  In a preferred embodiment, for example, the learner selects a word from the text reading interface 100, or the learner model 110 is updated as a result of recalculation and the learner has learned the word, etc. When any system event occurs, the word status determination process is performed again, and the display of the word status in the text read interface changes immediately and significantly.

  Those skilled in the art will recognize that the optimization steps of the preferred embodiment may recalculate context sensitive utility values only, for example, for words for which the assessment of proficiency has changed in the learner model, or learn in the text reading interface. It will be understood that context sensitive utility values can only be calculated for words selected by the person.

  Various modifications according to the preferred embodiment are possible.

  In a variation according to a preferred embodiment, the context sensitive utility function of step 230 takes into account the semantic domain composition in the text 130. A further knowledge source 150 assigns a semantic tag to each word in the text 130. Examples of semantic tags include emotion, food, home, money, sports, movement, number, material, communication, social life, time, technology, body parts, and the like. The frequency of each type of semantic tag is taken into account according to the context sensitive utility used in step 230 and ranked into candidate learning words. In this embodiment, the h function (above) takes into account the frequency of word semantic tags in the text 130 in use. FIG. 12 is an excerpt of the ranking of unlearned words based on this context-sensitive utility index.

  In another variation of the present invention according to a preferred embodiment, when a learner selects a word in the text readout interface 100, a recommendation is displayed using a pop-up information box. FIG. 9 is an example of a pop-up indicating that the word “fater” is the status of STUDY and should be learned now (in this example, the pop-up not only suggests STUDY but also allows the word to be learned immediately However, in other examples, the user is linked to a learning task or directed to some other learning resource). FIG. 10 is an example of a pop-up indicating that the word superior is an IGNORE status and should be learned later (this is presented with the suggestion of SKIP in the pop-up). FIG. 11 is an example of a pop-up that shows what the term be has a MASTERED status and may currently learn to remember a word (this is presented using a pop-up called REMEMBER?). In general, any information or recommended learning behavior can be connected to such pop-ups or other user interface means.

  In another variation of the preferred embodiment, step 220 assigns a MASTERED status or an ASSUMED_KNOWN status to all words in the in-use text that the learner has mastered relative to the learner model 110. The MASTERED status is assigned to a word when a learner has previously performed a learning act on a word using a language learning device. ASSUMED_KNOWN is assigned when the learner is not performing a learning act on the word using the language learning device. By continuing to record the words that the learner has performed, the learner model 110 or other elements in the system can function.

  In another variation of the preferred embodiment, the learner can control the maximum density of appearing words assigned the status STUDY in step 240, rather than using a predetermined number of words. The learner can control the density from 0% to 100%, for example, using slide control or the like.

  In another variation of the preferred embodiment, the language teaching material is in audio or video format. The learner can navigate through teaching materials and interrupted playback. This is similar to navigating through text material and word selection in the text reading interface 100. When a word is selected from audio or video teaching materials, a learning action is recommended based on the word status determined by the word status determination unit 140. The word status notification unit 160 may display the learning word by inserting an audio signal or a visual signal into the audio teaching material or the video teaching material.

  In another variation of the preferred embodiment, the subject to be learned is not the language itself. Text subjects are, for example, information contained in textbooks or encyclopedias about subjects to be studied such as physics, geography or history. The text material may be the learner's primary language. The vocabulary model 120 may assign general utility to the words related to the subject and may include words related only to the subject. Context sensitive utility indicators may also use additional information sources 150 that define essential relationships between words included in the vocabulary model 120. As a result, the word status determination unit 140 can assign the status of STUDY only to words for which related essential words have already been acquired based on the learner model 110. The general utility of the word (above “second index”) is independent of the learner, or is configured and realized so that different general utility values are returned for the same word for each user. May be.

  As described above, the first index, the second index, and the third index are determined for a certain word, and these three indexes are used to determine the status of a word such as STUDY, IGNORE, or MASTERED, for example. Explained. However, if the first index places these words in a predetermined word category, it may only be necessary to refer to the first index only. For example, for a word, if it is indicated that the learner has already learned the word sufficiently, the status of the word can be set to MASTERED without referring to other criteria related to the word. Since the second index and the third index are not calculated for a word that has already been learned, some steps can be omitted.

  One or more of the various components can be controlled by a program that operates in the device or the device. The operating program can be stored in a computer readable recording medium or embodied in a signal such as a downloadable data signal obtained from an Internet website. The appended claims should be construed to include the operating program itself, or the recording, signal, or other form on a carrier medium.

Claims (29)

A vocabulary learning method executed by a computer after selecting a text to be communicated to a learner,
The text includes a plurality of words, and for at least some of the plurality of words in the text,
An assigning step for assigning a status to the word based on the first index, the second index, and the third index for the word;
Communicating the assigned status to the learner at least upon request, and
The status to be assigned is selected from a plurality of predetermined statuses,
Each of the statuses is associated with at least one corresponding recommended action taken by the learner,
The first index is dependent on the learner, the second index is independent of the text, the third index is dependent on other words in the text ,
The above allocation step is
A determination step for determining a utility value related to context dependency indicating that the word depends on a context of a language teaching material currently learned for the word based on the second index and the third index;
A computer-executed vocabulary learning method comprising: a second assignment step for assigning the status based on a utility value relating to the context dependency . 2. The computer-implemented vocabulary learning method according to claim 1 , wherein the determining step further determines a utility value relating to the context dependency based on the first index. The above allocation step is
Words of a predetermined number of utility value is highest and decisions about the context-dependent, according to claim 1 or 2, characterized in that to assign the status of representing "study" that learning behavior is recommended Vocabulary learning method executed by a computer. The above allocation step is
Utility values for the contextual dependency is higher word than a predetermined threshold value, from claim 1, wherein the assigning status indicating "study" that learning behavior is recommended in any one of 3 A vocabulary learning method executed by the computer described. The above allocation step is
5. The computer-executed vocabulary learning according to claim 3 or 4 , wherein a status of "ignore" indicating that a learning action is not recommended is assigned to a word to which a status of "study" is not assigned. Method. 6. The vocabulary learning method executed by a computer according to any one of claims 1 to 5 , wherein the second index is independent of the learner. The vocabulary learning method executed by a computer according to any one of claims 1 to 6 , wherein the second index is a general utility function of the word. The above general utility function is based on one or more of word frequency, concreteness, word difficulty, word region composition, personal learning list composition, imagery, importance for a subject or curriculum The vocabulary learning method executed by a computer according to claim 7 . Above the second index, a value of zero for all the words are not a particular type, or by a computer according to claim 7 or 8, characterized in that assigning a value of the general utility function of the words Vocabulary learning method to be performed. 10. The computer-implemented vocabulary learning method according to claim 9 , wherein the specific type is a noun. 11. The computer-implemented vocabulary learning method according to any one of claims 1 to 10 , wherein the third index is a utility function of the word within the context of the text. 12. The computer-implemented vocabulary learning method according to claim 11 , wherein the utility of the word within the context of the text is determined based on the frequency of the word in the text. The utility of the word within the context of the text is determined based on the relative frequency of the word in the text compared to the overall frequency of the predetermined range. vocabulary the computer-implemented method of claim 11 or 12. 14. The computer-implemented vocabulary learning method according to claim 13 , wherein the predetermined range includes all words belonging to a certain language in the text, or at least a predetermined word sample of the language. The third indicator, vocabulary learning method executed by a computer according to any one of claims 11 14, characterized in that a function of the local importance of the word within the text. Each word is associated with a tag attached to indicate a part of a group of words that share a common idea,
16. The computer-implemented vocabulary learning method according to any one of claims 1 to 15 , wherein the third index is a function of the frequency of the tag associated with a word in the text. . The vocabulary learning method executed by a computer according to any one of claims 1 to 16 , wherein the first index is a function of knowledge of the word possessed by the learner. 18. The computer-implemented vocabulary learning method according to claim 17 , wherein the first index indicates a degree or degree of acquisition of the word by the learner. 19. The computer-implemented vocabulary learning method according to claim 17 or 18 , wherein the first index is related to the number of times the learner has seen or read the word. 20. The computer-implemented vocabulary learning method according to any one of claims 17 to 19 , wherein a Gaussian distribution of an acquired value of the word of the learner is applied as the first index. The computer according to any one of claims 17 to 20 , further comprising a second determination step of determining a value indicating whether or not the learner has learned the word from the first index. Vocabulary learning method. For at least some of the words in the text for which the first indicator is determined to be classified into a predetermined category,
The computer according to any one of claims 1 to 21 , wherein the assigning step assigns the status depending on the first indicator and independent of the second indicator and the third indicator. Vocabulary learning method to be performed. The first index is a function of knowledge about the word that the learner has,
For at least some of the words in the text for which the first indicator is determined to be classified into a predetermined category,
The assigning step is dependent on the first indicator, assigns the status independent of the second indicator and the third indicator,
17. The computer-implemented vocabulary learning method according to any one of claims 1 to 16, wherein the predetermined category includes words that are considered to be already learned by the learner. The assigning step assigns a “mastered” status indicating that a learning action is not necessary or that review is recommended to a word determined to belong to the predetermined category. A vocabulary learning method executed by a computer according to claim 23 . The above allocation step is
When a signal indicating that one or more of the first index to the third index has changed, or that there is a possibility that the index has changed, is transmitted to at least a part of the words. The vocabulary learning method executed by a computer according to any one of claims 1 to 24 , wherein the vocabulary learning method is assigned again. Transmission of the text to the learner includes display of the text on a display, provision of the text in a printed state, and / or audio display of the text. 26. A vocabulary learning method executed by the computer according to any one of 25 . A vocabulary learning device having means for performing processing on at least part of words included in a text selected to convey information to a learner,
Assigning means for assigning a status to the word based on the first index, the second index, and the third index for the word;
A transmission means for transmitting the assigned status to the learner at least in response to a request;
The status to be assigned is selected from a plurality of predetermined statuses,
Each of the statuses is associated with at least one corresponding recommended action taken by the learner,
The first index is dependent on the learner, the second index is independent of the text, the third index is dependent on other words in the text,
The above allocation means is
Determining means for determining a utility value relating to context dependency indicating that the word depends on a context of a language learning material currently being learned for the word based on the second index and the third index;
A vocabulary learning device comprising: a second assigning unit that assigns the status based on the utility value relating to the context dependency . A control program for controlling the apparatus for performing the vocabulary learning method according to any one of claims 1 to 26. A computer-readable recording medium on which the control program according to claim 28 is recorded.

Images