CALICO - The Computer Assisted Language Instruction Consortium
Vol 3, No. 3 (March 1986)
CALL with Audio Output
John CollettAbstract:
The need to include an audio component in CAI work in French is discussed, and an attempted solution is described in terms of a suite of programs designed to facilitate the creation and use of dictation practice materials. Pre-existing recordings can be used in the preparation phase, or
new recordings can be made as the program runs. An index of recordings is automatically updated. For each text a setting can be made to require exact punctuation or to tolerate punctuation differences when it is used. In student use of the program, useful features include: the option of instructions if needed; repetition of any phrase by pressing a single key; the progressive build-up on the screen of the complete and correct text, with those segments in which a student error occurred being highlighted; the option of repeating incorrectly answered segments at the end of the exercise; and of finally hearing and seeing the exercise; and of finally hearing and seeing the complete text simultaneously. Supplementary programs enable the inclusion of
accented letters and the production on screen of the correctly formatted text.
KEYWORDS: French, RAAD, audio cueing, DICTEE, dictation, software
The first flush of computer aided instruction in French has not fitted easily into the general modes of language teaching of recent years, in which oral and communicative aspects of language use have been predominant. The nature of the hardware has been a limiting factor, and has until now restricted the work done in the French Department at Waikato to programs involving the student in reading and writing only. The problem was discussed by Ariew (1984, pp. 43-47), where he commented:
It is ironic that most foreign language educational software is mute...the microcomputer is therefore an advanced technology with an antiquated pedagogy. Audio cueing is an essential component of up-to-date foreign language software (p.45).
Looking for the Audio
The limitations imposed on our programs by the absence of the audio component were acknowledged, but it promised to be difficult to find machinery which would be affordable, compatible with existing hardware, and also effective. Several methods of voice storage and production exist, and three contrasting examples are mentioned here.
1. Voice synthesizers: A voice synthesizer produces sound by means of an artificial larynx. Systems using this method of sound production have not so far produced sounds which are adequate for language teaching. "The most expensive equipment will produce something resembling a Dalek with laryngitis" (Davies, 1982, p. 55). Even if the technology improves considerably, to the extent that a machine will produce an utterance which does not sound like a machine speaking, it is doubtful if it will ever sound like a real person. It must inevitably lack that essential timbre of humanity by which a speaker reveals mood, personality, origins, attitudes or emotions. The use of voice synthesizers does not appear to be the road for language teachers to follow.
2. Digitized storage of human voices: Another approach to voice production has been by means of voice digitization—the storage as digital signals of thousands of minute samples of real human speech. Ariew (1985, p. 45) stated that the quality of production thus obtained can be very good, but this method requires so much memory space that presently it is not a viable solution to language teachers' needs.
3. Computer controlled tape recorders: A tape or cassette recorder which can be controlled by a computer seemed to be the best approach to the problem. Such machines are not cheap, however, and Ariew (p. 45) described how a cheaper solution was obtained by "simplifying the connections from the Apple to the tape recorder and controlling only the forward movement of the tape (play and pause)." But the facilities for replay and random location which are lost by this simplification are, in the author's opinion, absolutely essential if some of the specific strengths of computer-aided learning are to be retained.
While on study leave in 1984, the author sought advice on these matters in Paris and London. At the Centre International d'Etudes Pedagogiques at Sevres and at the Universite de Paris VII (Jussieu) an impressive linking of machines had been achieved by M. Garrigues, under the direction of A. Monnerie (author of Intercodes). One screen showed slides of market stalls, shops, etc., while a tape recording of a related conversation was heard. The slides appeared again to accompany a second screen on which a computer program presented a grammatical exercise.
37
The material used in this program was selected portions of the original conversation. Monnerie and Garrigues acknowledged that they had as yet produced material on minor grammatical points only, but they had great faith in the possibilities of the system. The hardware was impressive, but the brand names were unknown to the author, and appeared to be unavailable outside France. (The reverse also applied. Many of the French researchers in CAI were full of praise for BBC microcomputers, but could not obtain them because of import restrictions designed to protect their own electronics industries.)
In London, inquiries through CILT, CET, and similar organizations were disappointing. At first, only the work done by Professor R. Last in Dundee appeared to involve computer software for language teaching with audio output, but a visit was out of the question, since Scotland was snow-bound at that time. An Information Guide published by CILT revealed, however, that Last and others were using Tandberg tap recorders in their work (Davies 1982, p. 5). Tandberg had developed a system called AECAL (Audio Enhanced Computer Assisted Learning) in which a tape recorder can be controlled by a microcomputer. Linking two such machines is not in itself particularly noteworthy. Cassettes are regularly used for storing computer programs by users who have not yet graduated to a disc drive, but this use of a tape recorder is strictly linear and involves much pressing of keys to play, stop, rewind, etc. These limitations had been overcome in the Tandberg machine, and inquiries were pursued in that direction. The British branch of the company, when asked about Tandbergs being used for language teaching at sites which were within visiting distance, suggested the Luton College of Higher Education. A letter to Mr. Alan Page, Head of the Languages Department at that college, elicited an invitation to spend a day there. A brief demonstration of what he had achieved in his first exploratory steps with a BBC microcomputer linked to a Tandberg TCCR 530 was enough to show that this was one way in which the elusive audio element could be added to at least some programs used for teaching French at Waikato.
His program contained the numbers 1 to 20 recorded in five sets, each in a different language. The first step was to choose a language, and then a mode of study. The two modes were: (1) hear a number, type it in, and then see confirmation of your typed response, or (2) see a number, say it, and then hear the tape version for confirmation of your attempt. The numbers had been recorded in sequence, but appeared at random. So mode 1 might go thus: you hear "sept"; you type "7"; you see "7". And mode 2: you see "7"; you say "sept"; you hear "sept." The program and tape would be ready with the next item within a second or so. The whole thing may appear trivial, but its enormous potential for language teaching was obvious. Of particular importance was that the tape could move to any point, with no significant delay, and without any keys on the recorder having to be pressed. These impressions reinforced the high opinion the writer has always had of the engineering in Tandberg machines. He recalled being impressed, twenty years ago, even by their pause button, with which a single-syllable word could be split in two, and both parts remain clearly audible, and similarly by their fast forward and fast rewind facilities, in which "fast" was the operative word.
The TCCR 530 contains its own microprocessor to control direction, speed, time measurement, and so on, when the machine is used as an ordinary tape recorder. This processor can also communicate with a microcomputer via the latter's RS423 serial printer port—that is, in the same way as the microcomputer links to a printer. Messages contained in computer programs are sent from the computer to the tape recorder, telling it, for example, to play as far as a certain point on the tape, to stop, to rewind until an empty patch of tape is found, to send back to the computer information about the tape counter at that point, and so on. The "echo" from the tape recorder back to the computer is usually a message identical to the one sent, to confirm that the command was correctly received and carried out, but the returned message can also contain other information for the program to use, such as a counter reading or whether or not a signal is present at a particular point.
At any point in the program where action by the tape recorder is needed, a command containing an item from the Tandberg Command list must be sent to the recorder, and its response must be received. A programming feature which is particularly well implemented in BBC Basic is its use of Procedures—a simple method of invoking a particular section of a program at any time and as often as needed. The Tandberg Operating Instructions (p. 39) contain an example of a procedure designed to pass messages to and from the Tandberg, and, as recommended, this example has been used as the basis for the procedure used in our programs. The following are the operations carried out by the procedure:
—Selection of the input and output devices; the recorder and the computer respectively.
—Clearing buffers of previous contents.
—Printing the command in question to the selected output device.
—Checking legality of command and sending an error message if needed.
—Masking the 8th (parity) bit of each byte so that the data formats of the two machines match.
—Termination of command with a carriage return signal.
—Resetting of input and output devices, to return control from the tape recorder to the computer.
The piece of the program that invokes this procedure is "PROCt(CMD$)." In plain English this says "Pass to the PROCedure which handles the tape recorder the set of letters which have been defined as CMD$, and have it act upon them in the appropriate way." For example, if CMD$ is "ST" the tape will stop.
The Tandberg is very accurate in its response to commands involving the tape counter. If you have a recording starting at, say, 0030 (30 seconds into the tape), and give the commands GT0030, PL, then that is precisely what the machine will do—go to 0030, and play. It is so accurate that it may well clip the beginning of the recording, and it pays to follow the advice given in the instructions (p. 29): "With GT you must always build in a safety margin at the start of an audio block where PL or RE is the next command. This means deducting a small amount (1 or 2 seconds) from the Counter to avoid clipping the start." Then there is the question of spaces between recorded
patches. The Tandberg instructions (p. 31) advise that an audio block be preceded by a gap of silence of at least three seconds.
This kind of consideration, of which only two examples have been given, suggested the desirability of being able to easily and quickly make minor adjustments to the tape and counter setting, for example, to change 0030 to 0028, or to advance the tape by
38
three seconds to build in the required gap of silence automatically. A user defined function called "FNadjust" was written for this purpose and is used in the recording programs. An entry such as "FNadjust(t$3)" will, if t$ is, say, "0030," return the string "0033." Or "FNadjust(t$,-2)" will return the string "0028." A function program is needed because the results of ordinary addition and subtraction sometimes need conversion. If t$ is "0058," for example, and the second argument is 3, then the returned string must be 0101 (58 seconds plus 3 seconds = 1 minute and 1 second). As an initial exercise in putting the linked machines to work, it was decided to write a dictation program. It would require two major parts, one for the teacher recording the text to be used (RECORD), and one for students (DICTEE). There is also TANDEM (similar to RECORD except that the actual tape recording already exists), and several small servicing programs which the main programs chain to if and when needed. The functions of these small programs are listed below. They are held separately to ease problems of program size, and to enable one or two of them to be used on their own.
ACCENT: defines the ten red function keys so that they will produce accented letters.
HELP: provides instructions to teacher or student if and when needed.
EDIT: allows for the insertion of accented letters into scripts prepared in the BBC Word Processor.
INDEX: automatically builds up an index of recordings for consultation by students who wish to know what texts are available. The actual list is held in a file called TITLES.
FORMAT: offers a student the option, at the end of a dictation exercise, of seeing and hearing the text at the same time. This program can also be used separately by teachers for a quick check on the contents of data files and the associated recordings.
When these programs were completed and all parts were running smoothly, the user documentation was written.
PREPARING A DICTATION TEXT
A French text which is selected for use in the DICTEE program must be stored in a TEXT file on disc. The BBC Word Processor VIEW can be used to do this.
The layout of a dictation file should be thus: the title of the extract should be entered in the first line. The title will not be recorded or heard unless specifically repeated for that purpose in the second line of the file, but it will appear centered at the top of the screen when recording and dictation are in progress. The second and subsequent lines, up to a maximum of 30, should contain the dictation text, arranged so that one dictation segment occupies each line. Here is a very brief sample:
LES VACANCES
Certains de mes amis
vont passer leurs vacances
au bord de la mer
avec leurs parents.
Each text is therefore limited to 29 segments of a maximum of 40 characters each (29 + title = 30 lines). Setting the editor to a longer line and then running the program would work, but the screen presentation would become untidy. This version is written in BBC Mode 1; an alternative version could be written in Mode 0, at up to 80 characters per segment.
EDITING TEXT
Our sample text, LES VACANCES, has been entered as a file named TEXT. In the version above, words containing accents were deliberately avoided. But if, for example, the text said not "vont passer" but "ont passe" then one would need to use the program EDIT, written for the express purpose of putting accents into the file named TEXT.
The program title first is shown followed by a screening of the file called TEXT with each of its lines numbered and with the function key definitions for various accent marks appearing at the foot of the screen. The lower half of the screen becomes the workspace.
To leave working space on the screen if the text contains more than 15 segments, the first 15 segments only are offered for editing. When they are done, the remaining segments are screened for editing in a similar way, and when the whole text has been edited it is rescreened for a final check. A title plus 29 lines of text will just fit on the screen.
The text is now ready to record.
RECORDING TEXT
To record the text, the Tandberg TCCR 530 must be linked to the BBC Microcomputer, it must have its input level suitably set (5 on a scale of 0 to 10 seems satisfactory), and a microphone attached. The text which the RECORD program will present for recording must be in the file called TEXT. First, the program asks if you wish to see instructions for each step as the program runs. The instructions, if needed, are supplied by means of the separate program called HELP. The first recording task is the plain text, from beginning to end, without pauses. Before we start, the program asks at what point on the tape we wish this recording to begin and requires the answer to be given in minutes and seconds. The minimum setting possible is 0012 (12 seconds) since the program always advances to that point before starting, so as to be clear of the lead-in patch of tape. As soon as the starting point has been entered, the recorder will go into "fast forward" mode and the cassette will automatically wind to the point requested.
The text to be recorded is presented on the screen. When you are ready to begin recording, the space bar is the only control which needs to be touched. Press it once to begin recording, and again to stop recording. A message on the screen will tell you the starting and finishing points on the tape of this part of the recording, the former being confirmation of the one you requested, the latter just for information. Then you are ready to begin recording the text again, in separate segments. Each phrase appears on the screen as it is to be recorded. Type a space to begin and another to end the recording of each phrase. While recording is on, the word RECORDING (in red!) is plainly visible next to the printed phrase. A temporary file called TAPE is constructed to contain not only the text but also all details of timing which will be needed
39
by the program DICTEE when students use it. When recording is finished, the old file called TEXT is deleted, and the contents of the temporary file called TAPE are placed in a new TEXT file before TAPE is deleted. This may sound complicated, but the outcome is simple. The file called TEXT which used to contain these lines: LES VACANCES
Certains de mes amis
vont passer leurs vacances
au bord de la mer
avec leurs parents.
now contains these lines (sample times shown here):
0028LES VACANCES0040
0044Certains de mes amis0048
0051vont passer leurs vacances0056
0059au bord de la mer0103
0106avec leurs parents.0110
The file TEXT must now be renamed TEXTn, where n indicates its position in the set of files, and it is then ready for use by students in the program DICTEE. (Each side of a disc can hold up to 31 files. eleven are used to operate the DICTEE system, leaving space for up to 20 study texts per drive). The first line which contains the unrecorded title is used as a storage space for the starting and finishing points of the recording of the plain text. Here, 0030 had been requested as a starting point. The speaker probably went into action at about 0031, and finished at about 0038—seven seconds for sixteen words—a somewhat slow and deliberate rendering in this case, it would appear. But the point to be made is that the program has extended very slightly the length of the recording period, adding two seconds at each end by means of the "FNadjust" function. These new parameters are the ones which DICTEE will work to. And so it is with the figures which precede and follow each segment of text in the file. They are all automatically extended by two seconds in each direction to leave a safety margin.
The TITLES file is automatically updated by the program INDEX whenever RECORD is used.
USING PRE-RECORDED TAPE
The program TANDEM is for use instead of RECORD in cases where there already exists a recording of the material to be used. (The advantages of a variety of authentic voices in listening materials need not be reiterated here.) The program simply brings together a script and a recording in a format which the program DICTEE can use. Both must be prepared in such a way what they will fit together as required.
WHEN HELP IS NEEDED
The program "HELP" screens instructions if requested. All users, both teachers and students, will need to be told how to handle the various steps of the programs as they proceed. But when they become familiar with the program such instructions can be an annoyance. There is therefore an option built into RECORD, TANDEM, and DECTEE of seeing instructions or not. To save space in the memory of the main programs, the instructions have been placed in a separate program which is only chained to when needed. The HELP program should not be run on its own.
AN INDEX OF FILES
After the "instructions or not" option, students are asked which text they wish to use. If they wish to see a list, they can press RETURN. This list is held in the file TITLES, which is automatically updated each time the RECORD or TANDEM programs are run. At the first use of RECORD or TANDEM, there should be no file called TITLES on the disc. It will be created automatically by the program, and updated at each subsequent use. After each run of RECORD or TANDEM, the updated TITLES list is screened to show the user that the new title has been added. The program INDEX can also be run on its own for a quick listing of files, in which case the list displays the starting point of each file on the tape as well as its title, to help teachers keep track of tape contents. Students will see, for example, "1 LES VACANCES"; teachers will see "1 0030 LES VACANCES."
THE CENTRAL PROGRAM "DICTEE"
All the other files and programs mentioned exist to facilitate the smooth running of the program DICTEE. To use DICTEE the cassette recorder should of course be switched on, the input level turned right down, and the output level turned up to whatever point is comfortable. After dealing with the options of seeing instructions and/or a list of texts, a student types in the number of the text chosen. The screen is then cleared, brief instructions are given, the title of the text is displayed, and the recording of the complete text is heard. As soon as the complete text has been played a message advises, "Type a space when you're ready." and the program moves into the dictation phase proper. The following reminder of procedures appears in a box at the top of the screen.
Type in each phrase you hear.
Press COPY for a repeat.
Use DELETE to make changes.
Press RETURN when each phrase is finished.
Just below these instructions another area is defined as student work space, and a prompt = indicates where the student's typed answer will appear. Below this, the entire text will be built up, phrase by phrase, as the exercise proceeds. Each segment will be added to the text whether or not the student's attempt is correct, but with the following differences in the latter case. The correct version of the segment which the student entered incorrectly is highlighted by a colored background in its position in the text. The reminder of procedures at the top of the screen is replaced by a statement that the recording of the incorrectly answered phrase is now going to be played again and advice to look and listen carefully is given.
And so it proceeds to the end of the text, with a student being able to hear each phrase as many times as desired, just by pressing the COPY key. In the normal running of the program it will accept as correct those phrases of which the punctuation does not match that of the stored segment. Case differences will also be tolerated. The only exceptions are the apostrophe and the hyphen which, as in "s'appelle-t-il," are a more integral part of the phrase than, say, a comma or a full stop. But we are all familiar with formal dictation procedures in which punctuation is spelled out along with the segments of text. If one wishes to use DICTEE for practice in this mode, then a version of the program can be invoked which tolerates no differences whatsoever between the stored segment and a student's version. To achieve this, place as asterisk just before the title in the data
40
file. Thus, for example, "LES VACANCES" will allow punctuation difference, "*LES VACANCES" will not. At the end of the exercise the student is shown a score—the number of phrases correct out of the number tried, and the percentage correct. Then "Do you wish to try again the parts you got wrong? (Y/N)." If the student types Y, the machine goes to the point in the tape which holds the first incorrectly answered phrase, and the phrase is played again, with all options operating as before. And so on through the text, this time playing only those phrases in which an error occurred on the first run. The recorder can go to any point on the tape, as determined by the student's first performance. Such flexibility is an important feature of CAI work, and it is essential for it to be possible on the tape as well as in a computer program.
The final stage is to offer the student the option of seeing the complete text while listening again to the first recording (no pauses). This dual presentation, with the audio and visual versions appearing together, should leave with the student a strong impression of the relationship between the written and spoken forms—which is what dictation is all about.
The dictation segments are originally entered one per line, but when the text is screened for the student, it is correctly formatted. A one-space inset from the left at the beginning of a segment invokes a new paragraph in the formatted version. The code which deals with the correct formatting of the text is held in a separate program called FORMAT, to which DICTEE chains when required. The formatting program is held separately for two reasons: to ease the memory loading problem in the main program; and to enable this useful feature to be available to teachers, so that a nicely formatted version of study texts can easily be seen. The only command needed is CHAIN "FORMAT," and the program asks which text is to be screened and played. When FORMAT is accessed during student usage, this information is already known, and the formatting is done at once.
Summary
This then is the set of programs developed in the first few weeks or so in which the Tandberg has been available. An obvious next step is to develop listening comprehension exercises, in which students will hear, for example, a dialogue, while reading and answering questions about the speakers, their location, their topic of conversation, and so on. And other ideas are already crowding in.
A conversion of this set of programs to run on a VAX computer is needed, since this is the computer normally used by French students at Waikato for their CAI work. The one Tandberg recorder is unlikely to be sufficient if CAI with audio output develops at anything like the rate of "traditional" CAI activities. We are already thinking in terms of a language laboratory equipped with video terminals and computer keyboards as well as cassette recorders and headphones. The language laboratory may then more fully realize some of its early promises.
Tape recorders have been helping us to teach languages for many years. Computers are a later but no less effective arrival on the scene. The set of programs described above illustrates one way in which these two machines can be used together, providing us with a useful teaching tool and adding yet another thread to the rich fabric of language teaching.
ENDNOTES
A catalogue of French teaching programs prepared by CREDIF (1985) listed about thirty sets, including work done in France, Denmark, Sweden, the Netherlands, Eire, Israel, Canada, the United States, and New Zealand. The list was clearly not exhaustive, since no reference to any work in the U.K., for example, was included, but it is worth noting that the reference to Garrigues' work contained the only mention of any form of audio output.
"Logabax 529" computers; "Erca" video terminals.
CILT: Centre for Information on Language Teaching and Research, 20 Carlton House Terrace, London SW1.
CET: Council for Educational Technology, 3 Devonshire Street, London W1.
Tandberg, Ltd., Revie Road Industrial Estate, Elland Road, Leeds.
BIBLIOGRAPHY
Ariew, R., "Computer-assisted foreign language materials: advantages and limitations." CALICO Journal, 2, 1, Sept. 1984, pp. 43-47.
Coll, J., The BBC Microcomputer User Guide. BBC, London, 1982.
Davies, G. and Higgins, J., Computers, Language and Language Learning. Information Guide No. 22, CILT, London, 1982.
CREDIF, Catalogue F.L.E.O. Francais langue etrangere et ordinateir. ENS de Saint Cloud, Paris, 1985.
Davies, G., "Micros in modern languages." Educational Computing, 3, 8, 1982, p.30.
England, E., "Teaching English as a foreign language and computing: Problems and principles." Ph.D. thesis, UWIST, Cardiff, 1984, pp. 360-367.
Last, R.W., "Computer assisted learning: single element in total teaching process." Educational Computing, 1, 8, 1980, pp. 25-27.
Mounier, E., "Z Alons Bibliographiques." Etudes de Linguistique Appliquee 50, Didier Erudition, 1983, pp. 114-120.
Tandberg TCCR 530 Operating Instructions. Tandberg A/S, Norway, 1983.
Editor's note: Due to space considerations the source code for DICTEE was not reproduced here with the article. Please write to the author for this item.
Author's Address
John Collett
French Department
University of Waikato
Private Bag, Hamilton, New Zealand
41
Tidak ada komentar:
Posting Komentar