WO2003079312A2

WO2003079312A2 - Tutorial based learning

Info

Publication number: WO2003079312A2
Application number: PCT/GB2003/001056
Authority: WO
Inventors: Terry Miles; Timothy Shirra
Original assignee: Anglo European Training Centres Ltd.
Priority date: 2002-03-12
Filing date: 2003-03-12
Publication date: 2003-09-25
Also published as: GB0205767D0; AU2003216803A1; WO2003079312A8

Abstract

This invention relates to a method of analysing language for tutorial based learning, comprising providing a first portion of speech or text for tutorial analysis, providing a second portion of speech or text, wherein the first portion and the second portion are related by tutorial content, dividing the first portion of speech or text into a first plurality of segments, and dividing the second portion of speech or text into a second plurality of segments, wherein each of the first plurality of segments corresponds to one of each of the second plurality of segments, the segments being created for the purpose of tutorial analysis.

Description

TUTORIAL BASED LEARNING

Field of the Invention

The present invention relates to the field of any learning or tuition based on the use of language, and more specifically to remote tuition using remote computers, in which a student and a tutor are preferably interconnected by the Internet. The invention particularly but not exclusively relates to such tuition utilising both text-based and audio-based learning and assessment. The invention further particularly but not exclusively relates to foreign language learning, or subject learning based on the use of language. Background to the Invention

In the current art, the practice amongst language teachers of providing individual language learners with corrections, and sometimes comments to their language is well-known.

This is because it is widely recognised that an important requirement in 'learning a foreign language', as that concept is most widely understood, involves the learner attempting to produce the target language (i.e. the learning language) themselves and discovering what parts of their production are regarded by the target language community as acceptable, successful communication, and what parts of their production lead to a breakdown in communication.

Most experienced language teachers would agree that the most effective learning happens when the language to which a student is being exposed or which a student is being expected to produce is language that is personally relevant to that student, and linguistically within their current scope of competence.

This is why experienced teachers do not restrict themselves to one particular authored source of content (such as a coursebook), because no materials can ever be relevant and 'at the right level' all the time to all people.

Commonly, the most structured way of enabling a student to produce language which will then receive full feedback is for the teacher to ask the student to write an essay or composition. The teacher will then mark and correct mistakes in the text, sometimes using codes to describe the linguistic category of the mistake, in order to encourage the student to think more deeply about the mistake. Sometimes the teacher will point out good uses of language as well as mistakes. If the teacher is being careful and taking their time, they will not place the corrections on the same page as the one where the original mistakes are marked, in order to encourage the recognised 'good practice' on the part of the student of thinking about the mistakes and possible corrections before looking at the correction that the teacher has made. Sometimes, but much less commonly, the input material that a teacher will use is the student's spoken language. In order to do this in anything more than a tokenistic, randomly-sampled way, the student must be recorded and a transcription made from the recording. However, there are very few examples in the industry literature of the use of audio-recording and transcription as a means of providing whole-language processing on spoken language.

In the past few years, it has become more common for more adventurous teachers to encourage students to use computer word processing tools to write compositions and to use word processing tools themselves to undertake marking and correcting, and occasionally commenting. However, marking, correcting, and commenting are laborious processes even when the tutor is a fast typist, and it is widely recognised that to give a 'full' processing of a student's language, complete with linguistic categorisation of mistakes and corrections, codes and comments, that the tutor will be being paid a 'one-to- one' rate explicitly to do so, or will be subsidising the process through the use of their own or their institution's time unpaid. It is more common for a teacher under the pressure of time to simply mark and correct work.

The open-ended nature of word processing software, and the need to automate repetitive tasks, such as entering linguistic codes, has led some in the industry to develop automated subroutines or 'macros' within the word processing software. This is how one marking and correction tool, 'Markin' by Martin Holmes, developed. Initially it was a set of macros, which he then developed into a stand-alone product, which teachers can now buy as a tool to help them mark and annotate text.

Even with products such as 'Markin', the process remains highly laborious. One of the key reasons for this is that the comments and explanations that teachers make when they use this kind of software are written comments. In particular this does not enable the tutor to use the stress and intonation in his voice to convey meaning as effectively

There is also little scope for indicating problems and solutions with respect to a student's pronunciation. However, the weakest point in the art currently is the assumption that, once the student receives the corrections and comments from their tutor, she or he will know how best to work with the corrections to 'phase out' the mistakes and incorporate the new 'model' language into their active competence. However, it is the very common experience of teachers in the language teaching profession that students keep making the same mistakes, even after the teacher has indicated the mistake on several occasions and made sure the student is aware of the correction.

It is rare to find a student who has the pedagogical confidence and awareness to develop their own full systems for prioritising their practice in such a way that they will have ready an awareness of what they have just learnt, what is only half-learnt but still needs work, and what is top priority and needs full attention still.

The inventors have thus identified a need for a system which provides students with a personalised, properly structured and prioritised, and enjoyable framework for 'eradicating' inter-language mistakes and for learning new, model language.

It is therefore an object of the present invention to provide an improved tutorial based learning system, which addresses one or more of the above problems. Summary of the Invention

According to the present invention there is provided a method of analysing language for tutorial based learning, comprising: providing a first portion of speech or text for tutorial analysis; providing a second portion of speech or text, wherein the first portion and the second portion are related by tutorial content; dividing the first portion of speech or text into a first plurality of segments; and dividing the second portion of speech or text into a second plurality of segments, wherein each of the first plurality of segments corresponds to one of each of the second plurality of segments, the segments being created for the purpose of tutorial analysis.

In a further aspect of the present invention there is provided a computer program product containing program code for performing a method of analysing language for tutorial based learning, the method comprising: providing a first portion of speech or text for tutorial analysis; providing a second portion of speech or text, wherein the first portion and the second portion are related by tutorial content; dividing the first portion of speech or text into a first plurality of segments; and dividing the second portion of speech or text into a second plurality of segments, wherein each of the first plurality of segments corresponds to one of each of the second plurality of segments, the segments being created for the purpose of tutorial analysis.

In a still further aspect the present invention provides a method of analysing language for tutorial based learning, comprising: providing a first electronic file including a first portion of speech or text for tutorial analysis; providing a second electronic file including a second portion of speech or text, wherein the first portion and the second portion are related by tutorial content; electronically dividing the first portion of speech or text into a first plurality of segments; and electronically dividing the second portion of speech or text into a second plurality of segments, wherein each of the first plurality of segments corresponds to one of each of the second plurality of segments, the segments being created for the purpose of tutorial analysis. In another aspect the present invention provides a computer system for analysing language for tutorial based learning, comprising: means for receiving a first electronic file including a first portion of speech or text for tutorial analysis; means for receiving a second electronic file including a second portion of speech or text, wherein the content of the first electronic file and the second electronic file are related by tutorial content; means for electronically dividing the first portion of speech or text in the first electronic file into a first plurality of segments; and means for electronically dividing the second portion of speech or text in the second electronic file into a second plurality of segments, wherein each of the first plurality of segments corresponds to one of each of the second plurality of segments, the segments being created for the purpose of tutorial analysis.

In a still further aspect the present invention provides a computer system for analysing language for tutorial based learning, including: at a first computer: means for receiving a first electronic file including a first portion of speech or text for tutorial analysis; means for receiving a second electronic file including a second portion of speech or text, wherein the content of the first electronic file and the second electronic file are related by tutorial content; and means for transmitting first and second electronic files; and at a second computer: means for receiving the first and second electronic files; means for electronically dividing the first portion of speech or text in the first electronic file into a first plurality of segments; and means for electronically dividing the second portion of speech or text in the second electronic file into a second plurality of segments, wherein each of the first plurality of segments corresponds to one of each of the second plurality of segments, the segments being created for the purpose of tutorial analysis.

The present invention also provides an Internet website providing computer software for facilitating a method of analysing language for tutorial based learning, the method comprising: providing a first portion of speech or text for tutorial analysis; providing a second portion of speech or text, wherein the first portion and the second portion are related by tutorial content; dividing the first portion of speech or text into a first plurality of segments; and dividing the second portion of speech or text into a second plurality of segments, wherein each of the first plurality of segments corresponds to one of each of the second plurality of segments, the segments being created for the purpose of tutorial analysis.

The present invention produces tutorial (tutorial practice materials) according to a Productive Access to Language (PAL) methodology, and uses specially- designed software to facilitate the delivery of a product which is personalised to an unprecedented degree.

The invention enables the authoring by student and tutor of completely personalised, multimedia materials, that can be produced and accessed in a wide variety of forms. The invention facilitates the engagement of the learner regularly in transcribing their own language, and enables a situation in which the materials that the learner uses for their learning are made largely by themselves.

To do this profitably and cost-effectively whilst being able to maintain consistent standards on a large scale, the invention provides software tools for use by a tutor and a student that enable a relatively small amount of time spent by the tutor to generate a proportionately large amount of practice material and opportunity for the student.

A key aspect of the present invention is the way a learner's spoken or written language is captured and transcribed (by the learner) in the form of digital audio (preferably windows '.wav' files) and digital text, and the way in which larger chunks of language are broken down into smaller chunks, then corrected, commented on, and stored in a database. The organisation of this series of small chunks or segments of student and tutor language (text and audio) in a database that allows the material to be later accessed and practised in a multitude of ways is extremely advantageous. The software is designed in such a way that the 'componentalisation' process is an integral part of the process of reflection and evaluation by the tutor, and therefore does not stimulate extra cost. Brief Description of the Figures

The present invention will now be described by way of non-limiting example with reference to the accompanying drawings in which: Figure 1 illustrates a general example of access to a website hosted on a server by remote terminals using the internet;

Figure 2 illustrates the preferred main components of a computer terminal for performing the present invention; Figure 3 illustrates the steps performed by a student in carrying out a tutorial according to a preferred embodiment of the present invention; and

Figure 4 illustrates the steps performed by a tutor in carrying out an assessment of a tutorial in accordance with a preferred embodiment of the present invention; Description of Preferred Embodiments

The present invention is described herein by way of reference to a particular exemplary embodiment of a language tutorial-based learning system. It will be apparent to one skilled in the art that the invention is not limited in its applicability to such a specific learning system, and may have more general - applicability. Alternative applications of the technique of the present invention will be discussed herein below. In particular, the present invention is not limited to the teaching of language skills.

For the purposes of the description of the exemplary embodiment, it is assumed that the language tutorial-based learning system is centred on a service that is offered for use through an Internet portal hosted by a server. As illustrated in Figure 1 , server 100 hosts the Internet site providing a language tutorial-based learning system. A user of the service accesses the website using a computer terminal such as designated by reference numeral 102 or 104, connected though the Internet, generally designated by reference numeral 106. As can be seen in Figure 1 , the server is shown connected to the Internet by connections 108, and the computer terminals 102 and 104 are shown connected to the Internet by connections 110 and 112 respectively. The interconnection of remote computer terminals to a website providing a service is well-known to one skilled in the art, and Figure 1 is not intended to provide a detailed implementation of such interconnections.

For the purposes of the described embodiment of the invention there are, broadly speaking, two types of user of the language tutorial-based learning system website: students and tutors. Students preferably register for tutorials on the website, and as will be described further herein below are thereafter able to access learning information. Tutors access the website, or are provided with information via the website, in order to tutor the students. Referring to Figure 2, there is illustrated in more detail the computer terminal 102 of Figure 1. In Figure 2, the computer terminal 102 is shown to include a main processing unit 204, a display unit 202, a keyboard 206, a speaker unit 208, and a microphone 210. Each of the display unit, keyboard, speaker and microphone are interconnected to the main processing unit via communication lines 212a to 212d. The main processing unit is also provided with a connection to the Internet communication link 110. As will become apparent in the following description, the computer terminal 102 provides all the necessary means for a student or tutor to perform tasks associated with the present invention. In a preliminary phase of the language learning, not shown in detail here and well within the understanding of one skilled in the art, a student registers for tutorials on the website, and thereafter begins their tutorial programme. The registration process is typically carried out by a student accessing the website via a remote computer terminal, and carrying out a registration process which includes providing details as to the type of language learning desired, personal information, and payment information.

The tutorial programme itself is preferably based around a modular type structure, where a student progresses from module to module, and completes a particular module and then submits the completed module for assessment by the tutor. The tutor then assesses the students attempt at that module and provides feedback, also on a modular basis. As will be discussed further hereinafter, interaction between the student and the tutor is preferably via the website providing the language learning service.

In registering for a language course, one of the basic pieces of information that is obtained is the language that the student wishes to learn or improve - the "target language", and the main language (for example the mother tongue) of the student - the "anchor language". The framework of a typical tutorial for a student is described hereinafter with reference to Figure 3. Within a given tutorial the student performs tasks and prepares materials for assessment by the tutor. In this embodiment this is preferably facilitated by online software at the website. In an alternative it may be facilitated by local, or offline, software. The student therefore preferably performs the tutorial at a remote compute terminal such as terminal 102, after downloading the tutorial module from the website of the tutorial provider.

The performance of the tutorial by the student is facilitated by student software, provided via software downloaded from the website and/or running on the local computer.

In a step 302 the student prepares a task, which may be set by or in conjunction with the tutor. A task set in conjunction with a tutor may be based on feedback provided on a previous tutorial module, or information provided at the time of registration. The task may be, for example, a presentation, a role- play, a report, or a composition. This step uses functionality for enabling the student to prepare the performance of the next task in the online or offline software.

In a step 304, the student performs the task using the software, working from notes made in 302. Performing the task may involve recording a spoken piece, using the microphone 210, or composing a written piece, using the keyboard 206. Tasks are preferably performed bilingually, i.e. in the target language and in the anchor language. Tasks may also involve use of tutor- provided audio and text as part of preparation scenario.

In a step 306 any spoken task is recorded by software in the computer terminal 102 as one of a series of audio files, either in the target language, the anchor language, or both.

Alongside step 306, there may also be a step of performing a written task by typing in text on the keyboard 206 in the target language, anchor language or both, which text is stored in the computer terminal as text files. In a step 308 the software provides a playback facility to enable the student to listen, reflect on, and transcribe their language. Control of the playback facility is from the keyboard 206 or mouse (not shown in Figure 2, but assumed to form part of the keyboard 206) of the computer, enabling maximum ease of use.

In a step 310 the software provides for a student-commentary facility, allowing the student to mark their own language learning, attempt self-corrections, and add audio or text comments, to specific answers or as general comments. This information may subsequently be used by a tutor.

If the tutorial has taken place offline, the software compresses audio files (if necessary) and transfers the audio files, text files and any picture files (if used), via the internet to the hub-server 100. This data is preferably transferred in the . form of the files themselves plus database-ready XML (Extended Markup Language) files.

If the tutorial is performed online, then as represented by step 212 in Figure 2 the audio files and text files are stored in a database associated with the server 100, to await transferral to a tutor. The student can at any time access the website to upload picture files or other files that may be linked to the tutorial materials they are submitting to a tutor.

Once the student completes the tutorial, and the completed module transferred to the server hub, a series of 'back-office' functions are performed. These back-office functions may, for example, check the tutor's schedule and provide an estimate back to the student of when the assessment of the tutorial will be completed by the tutor. The back-office function may also send an automated e-mail to the tutor allocated to the student to advise the tutor that the tutorial preparation has been completed and is ready for assessment.

The marking, correcting, commenting, and explaining of the tutorial by the tutor, together with the setting of new challenges by the tutor, is facilitated by tutor software.

Student work is preferably presented to the tutor for tutorial processing as a unit (or module), which for administrative purposes has a globally unique reference string as an identifier. Such a unit may consist of one or more sets, a set consisting of one or more clips. If there is more than one clip in a set, they are included in the same set if they are related thematically. A clip contains the actual text data of the work, and if the work was a spoken performance, the names of audio files that were made when the spoken performance was recorded. The typical length for a clip is a paragraph. A clip can contain anchor language text, reference to an anchor language sound file, target language text, and reference to a target language sound file. For example, if the student performs and transcribes a bilingual spoken presentation consisting of five main points, delivered in both anchor language and target language, and this is the only piece of work that they are submitting to the tutor as the current unit, the data structure for the unit will be as follows:

The unit consists of one set; The set consists of five clips;

Each clip consists of: the anchor language text transcribed from the current anchor language spoken point ('paragraph'); the name of the sound file that contains the recording of that spoken anchor language point; the target language text transcribed from the current target language spoken point ('paragraph'); and the name of the sound file that contains the recording of that spoken target language point. The operation of the tutor's tasks in assessing the tutorial will now be described with general reference to the flow chart of Figure 4. It should be noted that the tutor may obtain the tutorial to mark in a variety of ways. For example the tutorial file may be sent to the tutor's remote computer terminal by e-mail, or the tutor may log onto the website to access the tutorial. In this example, it is assumed that the tutor logs onto the website. Upon logging on, in a step 402 the tutor is presented with the option to open a unit which is already in the database, or to import a unit into the database from a set of XML files and audio files that have been downloaded to local disk from the web server. Upon choosing a unit to work on, the tutor is presented with a confirmation screen via which can be seen all the text produced by the student, and any audio files produced by the student can be listened to. The tutor can choose to proceed with work on the unit, or reject it and choose another. If the tutor's login identity does not match the database's record of the allocated tutor, he is prompted to confirm that he wishes to assume responsibility for the processing of the unit, or to close the unit.

If working in a networked situation, and if the tutor proceeds with the unit, the software preferably 'locks' the database records for the current unit, to prevent accidental simultaneous processing by other tutors on the system. In a network situation, the software may also copy all source audio files from the network server's drives to a temporary folder on a local drive for faster read- write access. If any audio files are in a compressed format, the software may uncompress them in order that parts of them can be deleted if necessary. In cases where the unit has been previously worked on and it is being opened again, the software reads from the database the stage reached, and takes the tutor to the screen at which the processing or assessment of that unit was last exited.

The tutor software guides the tutor through stages in order to complete the tutorial. The tutor can leave the process at any point and resume later at the point at which they left it.

In a step 404, the tutor familiarises themselves with the students work. The principal purpose of this stage in the process is to allow the tutor to situate himself mentally in relation to the task ahead and to allow him to record and store any initial impressions he has of the student's work. In cases where the student's work is verbally performed, the tutor may do this by listening to the performance. In cases where the student's work is written, this is done by reading the whole text.

At any point during this stage, the tutor is able to create a note consisting of text and/or a spoken recording. The number of notes it is possible to create is limited only by the disk-memory on the computer being used. In tutorials where the student's work is a spoken performance, the tutor is presented with a playback panel for listening to the audio (first pass), with note-taking functionality for initial impressions. Audio playback and control is preferably simply provided by the keyboard or mouse. The tutor software provides functionality to play, pause, rewind and forward the audio file. There is also functionality to select a portion of the audio file for looped playback, allowing the tutor to listen repeatedly to a selected portion.

When a selection is being made or played back, a yellow bar preferably appears behind the standard slider to indicate that the audio player is now in 'selection mode'. At the beginning and end of the yellow bar there are small anchors that can be dragged to move the start or end points of the selection.

For written only tutorials, the student's written text is displayed on screen, with a scrolling facility if the text does not fit onto the screen.

Note taking is preferably facilitated by on-screen 'sticky-note'-style devices, which enable typed or spoken notes to be recorded. These notes may be free form and may be accessed at any later point in the process. During this stage, although the emphasis is on listening, the tutor can access any of the texts or audio files made by the student, including student commentary files, in both the student's anchor language and the student's target language. In a step 406, the tutor identifies 'chunks' or segments of the text files and the audio files included in the tutorial. In this way chunks or segments of each text and audio file are created. More particularly, as will be further described herein below, the segments of text and audio are matched to each other, such that each text segment has a corresponding audio segment. This step enables the tutor to reflect once more on the whole piece of work, before focusing on individual parts of it for detailed correction and commentary

More importantly, the text (and the audio) is broken down into a series of chunks or segments of optimal length. The tutor applies skill and judgment in deciding the optimal length for any chunk. The decision may be based on what length chunk the student will find optimal for retention in memory when performing listen-and-repeat exercises. Importantly, the audio files are broken down into chunks or segments so that they match the chunked or segmented text word for word.

The segmentation of the text and audio into corresponding chunks offers many advantages in the context of tutorial based language or content learning, as will become apparent in the following description.

The segmenting of the text and audio facilitates, for example, amendment of the transcription of the audio files in tutorials where this is not accurate. In performing this process a screen is displayed consisting of a text box, a grid, a main audio player and a 'child' audio player, which is invisible when the screen is first displayed.

Before this screen is displayed, in order to aid the tutor, a subroutine preferably prepares the clip level text or texts by 'pre-chunking' them: for each clip, it finds every full stop in the text and inserts two carriage returns. In alternative arrangements the text may be parsed to look for different delimiters, such as commas.

Once the screen is displayed, the text is then presented in a manner corresponding to the pre-chunked text. The tutor can choose whether to accept the chunk as it is presented following automatic parsing, or to create a new one; either by deleting the carriage return at the end of the chunk, thus joining it to the next one, or by inserting a new carriage return before the end of the chunk.

It should be noted that the use of carriage return inputs at the keyboard to identify the divisions of the chunks is only one possible way of creating the chunks. The divisions of the chunks could be marked with any input from the keypad or mouse.

When the automatic parsing subroutine has executed, the screen is displayed with the main audio player controls visible, and the first automatic pre- chunked clip in the text box. The tutor can navigate backwards and forwards in the audio file in normal, 'non-selecting' mode using the audio player controls. However, when the tutor wishes to select a chunk of audio, they can click on, or trigger from the keyboard a 'selection down' button on the audio player controls, which plays the audio file from that point whilst displaying a yellow selection bar, which indicates that an audio chunk or segment is being selected. The screen focus then shifts to a 'selection up' button on the audio player controls, which is used to allow the tutor to indicate the end of the desired audio chunk or segment selection simply by pressing, for example, the 'enter' key on the board.

It should be noted that the terms 'selection down' and 'selection up' are a matter of implementation choice. What is important is the functionality they afford the tutor using the tutor software. Specifically they allow the beginning and end of a particular chunk of an audio stream to be identified. The tutor marks the beginning and end of a chunk of audio to match the beginning and end of a chunk of text displayed on the screen.

It should be noted that preferably the audio stream is marked such that, following a copy being made, the audio stream preceding the start of the chunk and the audio stream following the end of the chunk, are deleted, leaving just the chunk.

Once, in this embodiment, the 'selection up' key has been pressed, the focus of the screen moves to a 'copy across' button on the screen displayed audio control. Clicking on this button takes the portion of the audio represented by the yellow selection bar, i.e. the selected audio segment, and copies it to the 'child player', which becomes visible and automatically starts playing the selected chunk. The purpose is for the tutor to confirm that this is indeed the audio chunk or segment that matches the displayed text segment.

If the tutor makes a mistake, a 're-select' button on the audio control display on the screen can be selected, which hides the child player and places the focus back onto the main player, which is readied again for selecting an audio chunk or segment.

Once the tutor invokes the 'copy across' button, the focus of the screen moves to a 'save chunk' button displayed on the screen.

If the chunk is now as the tutor desires, and it matches the first text chunk or segment as desired by the tutor, the tutor can trigger the 'save chunk' button by pressing 'enter' or an alternative input means. When the 'save chunk' button is clicked, a message box is displayed with the chosen chunk of text, asking the tutor to confirm that this is the chunk to be saved. The tutor presses the 'enter' key or clicks on a 'yes' button to confirm. Then the first chunk or segment in the textbox is removed from the text box and is saved to the 'chunks grid' which is below the text box on the screen.

Each text chunk is given a text chunk number when it is saved. The text chunk number is automatically allocated, depending on the text chunk's number in a series of text chunks created. Thus when a text chunk is saved to the chunks grid, it is given the next available number in the series. The chunked audio file is also saved, with the same chunk number, to hard disk.

Screen focus then moves back to the main player, which automatically starts playing the audio file, starting from the end of the last selected chunk, with the next text chunk displayed in the text box. Before the audio file is played, the tutor may choose to edit the text chunk provided by the automatic parsing.

Preferably, the text chunk is automatically displayed in the correct size/format, any preparing or editing of the text chunks having been carried out by the tutor prior to beginning the entire audio chunking exercise. Then once the text chunk is displayed, the audio player commenced from the end of the last chunk, and the tutor selects the end of the chunk associated with the current text chunk with a single keystroke or mouse depression.

If the tutor is certain that the transcript for this text chunk is accurate, the process may be accelerated by pressing a shortcut key. This performs the marking of the end of the audio chunk, and the copying across and the invoking of the 'Save Chunk' button, in just one action. Such a shortcut action is particularly envisaged where the tutor has fully utilised the student work familiarisation in step 404.

The design of the tutor software is preferably such that the whole process of chunking and matching can be achieved in real time with minimal effort. If the transcript is accurate, a chunk of text and its corresponding audio-chunk is made using as little as 2 keystrokes per chunk, without stopping the soundtrack at all. The chunks, after they are made, are stored visibly in the 'chunk grid' below the text box. If the tutor makes a mistake, or changes his mind, he can at any point choose a chunk from the chunk grid and either amend the text or re- select the audio portion that accompanies it. At any point during this stage, if the tutor has difficulties determining the optimal length for a particular chunk, the tutor can, by clicking to display popup screens, refer to the whole text of the student's work and to any comments (audio or text) that the student has made as part of their student commentary.

If there is more than one clip, then at the end of each clip the software reads the next clip in the sequence from the database, closes the current main audio file, reads the name of the next main audio file in the sequence, and opens it.

The tutor proceeds with the chunking and matching until all clips have been stored away as text chunks with associated audio chunks. After the creation of the text and audio chunks or segments, in a step 408 the tutor assesses the tutorial by marking and correcting it, and where appropriate adding a text or audio commentary. The purpose of this step is to enable the tutor to mark and correct any mistakes made by the student, preferably to categorise the mistakes by applying one or more codes to them and, if appropriate, to provide clues and/or explanation comments in text and/or audio, to help the student understand the reason why it is a mistake and to understand the correction.

In this step, the tutor software enables the tutor to correct and/or comment on the text (grammar, vocabulary, expression) and, if the submitted work was from a spoken performance, the pronunciation of the student.

The software is designed for optimal speed and ease of use by minimising the need to use a mouse. It preferably presents two text boxes on screen. On the left contains the Original text version' of the text produced by the student, and is for reference only. On the right also starts off by containing the original version, but can be edited by the tutor if it contains any mistakes, in order for it to become the 'model text version'. The texts displayed in the original text version and model text version text boxes are the ones that were chunked in the previous stage. Thus, an original text version chunk is displayed alongside a model text version chunk. As such, a third association is created: a model text chunk associated with the original text chunk, and also associated with the original audio chunk.

The tutor works directly in the model text version text box by selecting items of text and overtyping them with corrections, or inserting new text, or deleting existing text. Any words or phrases thus selected preferably appear in the original textbox underlined in red. Once the tutor has finished typing the corrected item or deleting superfluous text, the tutor presses the 'enter' key to indicate that the correction is completed.

At this point a pop-up panel is displayed, containing several rows of buttons, each of which displays a three-letter code representing a category of mistake. The buttons are read from a database to allow for variation in the sets of codes to cater for different target languages. The tutor selects from one to three codes to apply to a current item or mistake, by clicking on the relevant code buttons, or navigating to the desired button and pressing 'enter'.

Following this, a grade panel is displayed containing three buttons. The tutor then chooses what grade (level of seriousness) applies to a particular mistake.

All information about the item is then displayed in a row towards the bottom of the screen: the item number, the original text, the code(s) applying to the item, the solution/correction to the item, and the grade. At this or any later point, the tutor can click on a button next to each item row to add up to two alternative solutions for the item in question.

The tutor can also choose to delete or edit an item at any point. Each chunk can contain as many items as there are words in the original chunk. Thus there is created a further set of items associated with the displayed chunk.

In addition to correcting, coding and grading text items, for each item in each corrected chunk, the tutor is able to make comments in text and/or verbal

(audio) form. The tutor can make two kinds of comment: a clue comment and an explanation comment. The form of a clue comment is such that it provides the student with a clue to the nature of the mistake they have made and/or to the correction of the mistake. The tutor provides clues in this way to encourage the student to think harder about their language and enhance the learning. Explanation comments are provided by the tutor to help the student to understand and acquire a correction by providing an explanation of grammatical or lexical patterns. The tutor can choose to make an explanation there and then, or choose one from an existing stock of pre-existing explanations associated with mistakes of that kind.

Clue comments and explanation comments are made from a separate commentary screen, consisting of three audio devices (one for playing the original student chunk, one for recording a clue comment, and one for recording an explanation comment), and two text boxes (one for storing a textual clue comment and one for storing a textual explanation comment).

If the student's work included a spoken performance, the tutor is also expected to listen to the corresponding audio-chunk, which is available from a player on the same screen, with the aim of assessing the student's pronunciation.

The tutor can choose also to comment on and correct errors in pronunciation that the student has made. To do this, the tutor invokes a 'correct pronunciation' button. Invoking this displays a new screen that contains three audio devices: a 'what-the-student-said' player, an audio-explanation player- recorder, and a model-version player-recorder. The screen also preferably contains three text boxes. One to display the original text context, one to display a pseudo-phonemic representation of what the student said, and the third to display a pseudo-phonemic representation of the model way of saying that item. The pseudo-phonemic representations are created through known techniques in text editing.

From this screen the tutor can select the portion from the original audio-chunk that the student should listen to. The tutor does this using the yellow-bar selection device described in the chunking process. The tutor can embed this 'quoted' portion into a verbal explanation of the problem, which is being recorded in the second audio player. Finally, using the third audio player, the tutor can record a discreet model version, which can be used by the student for targeted listen-and-repeat practice.

Optionally, to help convey the nature of the problem and the solution, the tutor can use the pseudo-phonemic textboxes. These have alongside them buttons which change the font-style of any text inside them which is selected. Fonts can be made larger or smaller, changed from lower to upper case, or from upper to lower case. If these text boxes are used, the contents of them are stored as rich text format (rtf) in the database, so that they can be displayed back to the student exactly as the tutor intended. When the tutor has finished with the current pronunciation item, the tutor invokes a 'save item' button, which saves all audio files and rich text files, if any, and re-displays the main correction screen.

When the tutor has finished with the current chunk or segment, a 'next chunk' button is invoked. This saves all items in the current chunk to the database. At any point the tutor can move back through previous chunks to add, delete or amend items.

During this stage a tutor can also add a third type of comment: a 'pure comment'. The purpose of this pure comment is not to correct but to allow the tutor to make a comment about anything the student has said. This pure comment may be to respond to a question that a student has posed in their own student commentary; it may be to praise the student for having made good use of language, or it may be to respond to the content of what the student has said.

To add a pure comment, the tutor invokes a 'pure comment' button, which displays a separate screen. In this screen, there are two textboxes, one in which the whole student text is displayed for reference, the other blank to receive the tutor's (optional) text comment. There are two audio devices displayed: one enabling playback of the student's original audio, and selection of a portion of it for reference; the other enabling the tutor to record his comment. If necessary, the tutor can select from the textbox the part of the text to which he is referring, which will then be saved as the referred-to text of that pure comment item. When the tutor has finished making a pure comment, the main correction screen is once again displayed.

At any point, if the tutor is in doubt as to a meaning the student intended to convey, and if the tutor is familiar with the student's anchor language, the tutor can refer to the anchor language versions of the student clips (texts and audios) if they exist (i.e. if the student worked bilingually), in the hope that the meaning will have been conveyed more clearly.

The tutor moves through the chunks in this way correcting them, and where appropriate, providing: i) clue comments; ii) explanation comments; and iii) pure comments, and correcting pronunciation errors. In order to avoid the tutor falling into the trap of 'not seeing the wood for the trees', it is possible to display on a separate screen the whole model text as it gets built, chunk by chunk. When the last chunk has been reached, and the tutor clicks on the 'next chunk' button, the software moves to a further stage as discussed below.

In a step 410, the tutor analyses the tutorial and comments thereon. The purpose of this step is to enable the tutor to see, graphically, if any patterns have emerged in the mistakes made by the student, and to use that understanding to decide what comments to make. This is an overall analysis of the tutorial.

All items from all chunks are displayed on screen in rows on a grid. The grid is laid out thus: column one displays the item number, column two, the item itself in context, with the item highlighted. Column three, the code(s) applying to the item, column four, the item's solution in context, with the solution highlighted. Column five displays the item's grade.

By clicking on the column headers, the tutor can sort the grid according to different criteria. In one embodiment the default is to sort on the item code column, since as long as the tutor has been consistent in applying codes, this reveals what kind of mistake has been made most frequently by the student. The tutor can then choose to select one or more item-rows about which to make a clue and/or explanation comment. If more than one item is selected, the comment(s) will be stored in the database as applying to all of the selected items, so that when the student comes across each item, they will be routed each time to the same comment that applies to the whole group.

This provides an efficient way for the tutor to deal with repeated mistakes.

In the previous step (step 408), the tutor was able to make clue and explanation comments by jumping from the correction screen to the Comments screen. But if the tutor prefers, he can wait until this step (step 410) to make any or all of his clue or correction comments, as a result of having seen and been able to analyse any patterns of mistakes made by the student. In a following step 412, the tutor creates an audio model of the tutorial. The purpose of this step is to enable the tutor to make appropriate audio-model material that may be used in different exercise types to help the student acquire the model language. By the end of the step the tutor has preferably recorded at least one model spoken performance of the whole model text, and in the process marked the points in the sound file where each chunk begins. These marker flags may then be used by the software to automatically break down the 'parent' model sound files into chunk-length 'child' sound files. These audio model chunks are newly created but have a one-to-one correlation with the chunks in step 406 in analysing the student's work. The software then presents the tutor with all the model-chunks made, in the form of a 'scrollable autocue'. At any given point, the font of a current chunk being read by the tutor is changed so that it stands out from the chunks before and after it.

After dictating each model chunk aloud into the microphone, the tutor presses the spacebar to advance to the next chunk. When the spacebar is pressed, the software stores a marker flag so that later, it can make separate, chunk- length sound files. Pressing escape reverts to the beginning of the chunk and starts the recording of the current chunk again.

The tutor continues like this until the recording has finished at the last chunk. When this happens, the main whole-text-length sound file is saved and then a copy of it is broken up into separate, chunk-length files that are saved individually. The tutor is then asked to make another model version, in case it is desired to make one that is faster, or slower, or in a different style, from the one that has just been made.

In a step 414 the tutor finalises the tutorial assessment. In finalising the tutorial assessment, the tutor may make an introductory commentary and a 'rounding-off and pointing-forward' commentary to frame the detailed corrections and comments.

The introductory commentary will be the first thing that the student will hear and see when they receive their processed unit from the tutor. This may refer back to the last unit done by the student, and to any questions that have been raised by the student since then, and will give the student the tutor's overall impressions of the quality of the preparation of the work, the effort made and the language used. It may include comments about the content, and anything else that the tutor deems appropriate to include in an introductory commentary.

In the rounding-off commentary, the tutor reviews the most important points of the tutorial and points the student forwards, providing advice on what areas to focus on, and pointing to new challenges. It includes a recommended topic and task for the next unit to be completed by the student. On screen the tutor has access to a number of audio-recording devices, each of which has a text box next to it. There is also access to the previous stage's screen, where the tutor can refresh his memory of all the corrections that have been made. There is also access to the student's initial commentary, which the tutor may wish to respond to as part of his introductory comments. The tutor records his introductory comments in each of a number of default categories, and optionally can enter accompanying text comments. Default categories include: evaluation of exercises done by the student during the last unit; responses to questions raised by the student since the last unit; evaluation of the quality of the student's preparation for this unit, evaluation of the extent to which the student stretched himself in this unit; assessment of the performance in this unit; fluency achieved; grammatical accuracy; etc. The tutor can also click to add a new category of comment, or change the category of a comment already made.

If the tutor so wishes, and is recording comments in the student's target language, the tutor can choose to change to the student's anchor language and flag this (or vice versa - change from the student's target language to their anchor language). The tutor may also click to add an extra recording device so that comments in both languages can be provided.

As a part of making the rounding off comments, the tutor will be required to recommend the task for the next unit that the student is going to do (unless the student has indicated that he does not wish to proceed any further with this target language). The tutor will enter the topic of the recommended next task, and a number of 'prompt questions' or instructions that will aid the student in preparing their next tutorial. The tutor can choose to enter fresh new questions or instructions, or select template ones from the database. Questions and instructions are preferably always in text form, but audio can be added.

Finally, in a step 416, the tutor uploads the assessed tutorial. The software compresses all audio files into MPEG-Layer-3 Format Windows Wav files. It then converts all data from the unit to a series of XML files and uploads these and the sound files to the web server, in order to be accessed up by the student.

Although the present invention has been described herein with reference to the particular example of foreign language tuition, it is not so limited in its applicability. The present invention may apply to any tutorial technique using language to drive the tutorial, but for which language learning is not the purpose of the tutorial. For example, the technique may apply to teaching mathematics.

Claims

CLAIMS:

1. A method of analysing language for tutorial based learning, comprising: 5 providing a first portion of speech or text for tutorial analysis; providing a second portion of speech or text, wherein the first portion and the second portion are related by tutorial content; dividing the first portion of speech or text into a first plurality of segments; and 10 dividing the second portion of speech or text into a second plurality of segments, wherein each of the first plurality of segments corresponds to one of each of the second plurality of segments, the segments being created for the purpose of tutorial analysis.

15 2. A method according to claim 1 , wherein the steps of dividing include applying a critical analysis to the speech or text to determine the optimum size of a portion of speech or text for tutorial analysis.

3. A method according to claim 1 , wherein the steps are performed in 20 software.

4. A method according to claim 1 , wherein the first portion is speech provided by a learner.

25 5. A method according to claim 4, wherein the first portion of speech is in a learning language.

6. A method according to claim 1 , wherein the second portion is text provided by a student.

7. A method according to claim 6, wherein the second portion of text is in a learning language.

8. A method according to claim 1 , wherein the steps of dividing are performed using tutorial criteria

9. A method according to claim 8, wherein the steps of dividing are performed by a tutor.

10. A method according to claim 1 , further comprising providing a third portion of speech or text, wherein the third portion is related to one of the first or second portions by tutorial content; and dividing the third portion of speech or text into a third plurality of segments, wherein each of the third plurality of segments correspond to one of each of the first or second plurality of segments, the third segments being created for the purpose of tutorial analysis.

11. A method according to claim 10, wherein the third portion of speech is provided by a tutor.

12. A method according to claim 11 , wherein the third portion of speech is model speech corresponding to the first portion of speech.

13. A method according to claim 10, further comprising providing a fourth portion of speech or text, wherein the fourth portion is related to one of the first or second portions by tutorial content; and dividing the fourth portion of speech or text into a fourth plurality of segments, wherein each of the fourth plurality of segments correspond to one of each of the first or second plurality of segments, the fourth segments being created for the purpose of tutorial analysis.

14. A method according to claim 13, wherein the fourth portion of text is provided by a tutorial analysis.

15. A method according to claim 14, wherein the fourth portion of text is model text corresponding to the second portion of text.

16. A computer program product containing program code for performing a method of analysing language for tutorial based learning, the method comprising: providing a first portion of speech or text for tutorial analysis; providing a second portion of speech or text, wherein the first portion and the second portion are related by tutorial content; dividing the first portion of speech or text into a first plurality of segments; and dividing the second portion of speech or text into a second plurality of segments, wherein each of the first plurality of segments corresponds to one of each of the second plurality of segments, the segments being created for the purpose of tutorial analysis.

17. A method of analysing language for tutorial based learning, comprising: providing a first electronic file including a first portion of speech or text for tutorial analysis; providing a second electronic file including a second portion of speech or text, wherein the first portion and the second portion are related by tutorial content; electronically dividing the first portion of speech or text into a first plurality of segments; and electronically dividing the second portion of speech or text into a second plurality of segments, wherein each of the first plurality of segments corresponds to one of each of the second plurality of segments, the segments being created for the purpose of tutorial analysis.

18. A method according to claim 17, wherein the steps of electronically dividing are performed in software.

19. A method according to claim 18. wherein the steps of electronically dividing are performed automatically in software by parsing the content of the electronic file to identify delimiters.

20. A method according to 19, wherein the steps of electronically dividing further include editing the automatically divided portions based on a critical analysis.

21. A method according to claim 20 wherein the step of editing includes combining automatically divided segments and further dividing divided segments.

22. A method according to claim 20 wherein the editing is performed by an operator of a computer system applying tutorial criteria.

23. A method according to claim 17 wherein the first portion is digitised speech.

24. A method according to claim 23 wherein the first portion of digitised speech is provided by a learner.

25. A method according to claim 24 wherein the digitised speech is electronically recorded using a microphone connected to a computer system.

26. A method according to claim 17 wherein the second portion is electronic text.

27. A method according to claim 26 wherein the electronic text is provided by a learner.

28. A method according to claim 27 wherein the electronic text is electronically provided using a keyboard connected to a computer system.

29. A method according to claim 17, further comprising providing a third electronic file including a third portion of speech or text, wherein the third portion is related to one of the first or second portions by tutorial content; and electronically dividing the third portion of speech or text into a third plurality of segments, wherein each of the third plurality of segments correspond to one of each of the first or second plurality of segments, the third segments being created for the purpose of tutorial analysis.

30. A method according to claim 29, wherein the third portion is

3 digitised speech provided by a tutor.

31. A method according to claim 30, wherein the third portion of digitised speech is model speech corresponding to the first portion of digitised speech.

10

32. A method according to claim 17, further comprising providing a fourth electronic file including a fourth portion of speech or text, wherein the fourth portion is related to one of the first or second portions by tutorial content; and 15 electronically dividing the fourth portion of speech or text into a fourth plurality of segments, wherein each of the fourth plurality of segments correspond to one of each of the first or second plurality of segments, the fourth segments being created for the purpose of tutorial analysis.

20 33. A method according to claim 32, wherein the fourth portion is electronic text provided for tutorial analysis.

34. A method according to claim 33, wherein the fourth portion of electronic text is model text corresponding to the second portion of text.

T

35. A computer system for analysing language for tutorial based learning, comprising: means for receiving a first electronic file including a first portion of speech or text for tutorial analysis; 30 means for receiving a second electronic file including a second portion of speech or text, wherein the content of the first electronic file and the second electronic file are related by tutorial content; means for electronically dividing the first portion of speech or text in the first electronic file into a first plurality of segments; and means for electronically dividing the second portion of speech or text in the second electronic file into a second plurality of segments, wherein each of the first plurality of segments corresponds to one of each of the second plurality of segments, the segments being created for the purpose of tutorial analysis.

36. A computer system according to claim 35, wherein the first electronic file comprises a portion of digitised speech for tutorial analysis.

37. A computer system according to claim 36, wherein the second electronic file comprises a portion of text for tutorial analysis.

38. A computer system according to claim 37, wherein the means for electronically dividing the second electronic file includes means for parsing the text to identify predetermined delimiters, the predetermined delimiters defining each text segment.

39. A computer system according to claim 38 wherein the means for electronically dividing the first electronic file includes means for playing the audio file, and input means for making points in the audio file corresponding to the position of the delimiters in the text file.

40. A computer system according to claim 35 wherein the segments of the first electronic file are allocated speech numbers and the segments of the second electronic file are allocated corresponding text numbers.

41. A computer system according to claim 35 wherein the corresponding first and second plurality of segments are stored in a memory means.

42. A computer system according to claim 35 further comprising means for outputting the text and audio segments.

43. A computer system according to claim 42 wherein the means for outputting the text and audio segments includes means for transmitting the segments to a remote computer device.

44. A computer system according to claim 35 wherein the means for receiving the text and audio file comprises means for receiving the files from a remote computer device.

45. A computer system according to claim 44 wherein the remote computer device is a remote server.

46. A computer system according to claim 35 wherein the means for receiving an audio file includes a microphone.

47. A computer system according to claim 35 wherein the means for receiving a text file includes a keyboard.

48. A computer system according to claim 35 wherein the means for presenting the text and audio file include a VDU.

49. A computer system according to claim 35 wherein the speech and audio segments are created by an operator simultaneously reading the text and listening to the speech, and determining the segments.

50. A computer system according to claim 35, further comprising means for receiving a third electronic file including a third portion of speech or text, wherein the third portion is related to one of the first or second portions by tutorial content; and means for electronically dividing the third portion of speech or text into a third plurality of segments, wherein each of the third plurality of segments correspond to one of each of the first or second plurality of segments, the third segments being created for the purpose of tutorial analysis.

51. A computer system according to claim 50, wherein the third electronic file comprises a portion of digitised speech provided by a tutor.

52. A computer system according to claim 51 , wherein the third portion of digitised speech is model speech corresponding to a first portion of digitised speech.

53. A computer system according to claim 50, further comprising means for receiving a fourth electronic file including a fourth portion of speech or text, wherein the fourth portion is related to one of the first or second portions by tutorial content; and means for electronically dividing the fourth portion of speech or text into a fourth plurality of segments, wherein each of the fourth plurality of segments correspond to one of each of the first or second plurality of segments, the fourth segments being created for the purpose of tutorial analysis.

54. A computer system according to claim 53, wherein the fourth portion is electronic text provided for tutorial analysis.

55. A computer system according to claim 54, wherein the fourth portion of electronic text is model text corresponding to the second portion of text.

56. A computer system for analysing language for tutorial based learning, including: at a first computer: means for receiving a first electronic file including a first portion of speech or text for tutorial analysis; means for receiving a second electronic file including a second portion of speech or text, wherein the content of the first electronic file and the second electronic file are related by tutorial content; and means for transmitting first and second electronic files; and at a second computer: means for receiving the first and second electronic files; means for electronically dividing the first portion of speech or text in the first electronic file into a first plurality of segments; and means for electronically dividing the second portion of speech or text in the second electronic file into a second plurality of segments, wherein each of the first plurality of segments corresponds to one of each of the second plurality of segments, the segments being created for the purpose of tutorial analysis.

57. A computer system according to claim 56, further including a remote server, wherein the remote server includes means for receiving the first and second electronic files from the first computer, and means for ^' transmitting the first and second electronic files to the second computer.

58. A computer system according to claim 57, wherein the first and second computers are connected to the remote server via the Internet.

59. A computer system according to claim 57, wherein the remote server hosts an Internet website providing a tutorial based learning service.

60. A computer system according to claim 59 wherein the first computer is associated with a learner registered with the Internet website providing a tutorial based learning service and the second computer is associated with a tutor registered with the Internet website providing a tutorial based learning service

61 . An Internet website providing computer software for facilitating a method of analysing language for tutorial based learning, the method comprising: providing a first portion of speech or text for tutorial analysis; providing a second portion of speech or text, wherein the first portion and the second portion are related by tutorial content; dividing the first portion of speech or text into a first plurality of segments; and dividing the second portion of speech or text into a second plurality of segments, wherein each of the first plurality of segments corresponds to one of each of the second plurality of segments, the segments being created for the purpose of tutorial analysis.