TTS IN EFL CALL – SOME PEDAGOGICAL CONSIDERATIONS

by Wlodzimierz Sobkowiak

School of English , Adam Mickiewicz University

Poznan, Poland

sobkow@amu.edu.pl

Abstract

Rule-based Text-to-Speech synthesis (TTS) is discussed from the point of view of English as a Foreign Language (EFL) Computer-Assisted Language Learning (CALL). The perspective is pedagogical rather than technological. Some didactically salient characteristics of TTS are considered, such as (a) its feasibility as a pronunciation model, (b) control afforded over accentual and phonostylistic variation of speech, (c) the prospects of multimodal synthesis ('talking heads'). Some internet website addresses featuring TTS information, products and demos are provided.

1. Introduction

CALL is different things to different people.  To some it is the use of text editors in the process of writing a homework assignment.  To others it is surfing the internet for language tasks and exercises.  To others yet it is when they stick that EFL CD in the drive and start on lesson number five.  To learners it brings the welcome air of novelty to break the boredom of the language classroom.  To teachers it means that extra effort and stress of preparing and running the technology-assisted lesson, with Murphy spitefully poised to spoil the carefully constructed scenario.  To administrators CALL is that expensive foible of teacher X, forever in need of new equipment, software and servicing.  To developers it is yet another opportunity to build in one of the newest hard- or software gimmicks to get the edge over competition.  To parents it is sometimes the only motivation to finally give in and buy that idolatrous icon of modernity – the computer.  To enthusiasts CALL is the promised land of language teaching and learning, with learners acquiring language knowledge and skills effortlessly from the machine, and the teacher walking benignly among workstations and laptops, offering help and guidance when needed.  To technophobes CALL is a monster which can expose their ignorance and inadequacy, which threatens their (human/ist) ego, and which will ultimately destroy language learning as we know it today, transforming language teachers into beggars, and learners into cyborgs.

In the middle of all this commotion and controversy, CALL has now secured itself a safe position in all areas and on all levels of education, and it is certainly one of the most vibrant themes of didactic reflection and research at the beginning of the third millennium.  Some of its thrust comes from the tempestuous development of computer technology; some from the concurrent, but seemingly independent, changes in FLT methodology over the last decade or so – changes away from the pure communicative paradigm, in the direction of more form-oriented teaching and learning (e.g. Doughty & Williams 1998, or Ellis 2001).  It is linguistic form, after all, that computers can (so far) manipulate much better than meaning or pragmatics.

In what follows I will consider the actual and potential impact of one of the cutting-edge computer technologies on one area of EFL which has traditionally been the most form-oriented of all – pronunciation.  In particular, I will look at Text-to-Speech (TTS) synthesis as part of Human Language Technologies (HLT) or Natural Language Processing (NLP).  The discussion will, however, be rather low-tech, with no reference to Hidden Markov Models (HMMs), Fast Fourier Transforms (FFTs), Dynamic Time Warping and the like.  For these, the reader is referred to some introductory sources, such as Cole et al. 1996, Hovy at al. 1999 or Granström, House & Karlsson 2002.  Instead, the perspective will be that of a (Polish) phonetics teacher and materials developer.

2. TTS in EFL CALL

There are, fundamentally, two types of speech synthesis: by concatenation, whereby previously recorded human speech is segmented and recombined, and by rule, whereby no previous human recording is necessary in any form, but rather a complex set of rules derives phonemic representations from spelling, and then generates the segmental and suprasegmental acoustics.  While the process is rather more involved than what the preceding sentence might suggest, the TTS synthesis by rule has now overcome most teething problems and reached human-like quality (cf. e.g. Dutoit 1997 and 1999).  The high-end TTS engines are rather expensive, and research to improve especially the prosodic properties of synthesized speech is still under way, but the technology is now reaching the stage where it can be applied to CALL.

2.1. Synthesized speech as a FLT pronunciation model

TTS synthesis has been in existence for a few decades now, mostly in applications for use by professionals and visually impaired people.  The general public has first come into contact with 'robotic' speech in a variety of telephone information systems, such as train or air timetables.  The two quality criteria proposed for such systems have been intelligibility and naturalness (d'Alessandro & Liénard 1996), rather than phonetic correctness with reference to some FLT norm.  Thanks to both technological developments (processor speed, cheap memory, better DA converters) and linguistic research, all three criteria have now been reached.  Not only is (top-quality) synthesized speech intelligible and natural (click here for a few short demos: http://elvis.naturalvoices.com/demos/), but it can also actually function as a model of pronunciation.  For example, Filoglossia, a CALL package with (modern) Greek as a foreign language, already employs TTS synthesis: http://www.ilsp.gr/filoglossia_plus_eng.html, and WordPilot (http://www.compulang.com), also has this feature.

This creates a completely new situation in pronunciation-oriented CALL.  So far, most of the CALL CD-ROMs (and most of the internet bandwidth for on-line speech-enabled courses and applications) was taken by audio (and video) recordings of speech in a variety of formats, including the space-saving mp3.  It was believed that only good quality pre-recorded native speech can be profitably used in an educational CALL setting.  But if the trick can be done with a TTS algorithm of a few megabytes, the freed space (bandwidth) can be used for other software, including e.g. artificial intelligence routines, which are crucial in sustaining a continued dialogue between the machine and the human.  And, naturally, speech synthesis is by far less expensive than recording a team of highly trained human speakers.

The space and bandwidth savings are rather trivial technicalities, of course, from the point of view of FLT.  Even if some CALL objectives and functionalities had to be compromised in the past due to space restrictions, the advent of DVD and wideband internet is guaranteed to change this facet of CALL completely.  This has already happened in the US , and will eventually come to Poland as well.

2.2. Accent control

But there are more pedagogically interesting ramifications of TTS.  One of them is the degree of control which the developer has over synthesized speech.  Practically all speech variables can be manipulated at will.  Take accent, for example.  EFL CALL packages and electronic dictionaries for learners have so far catered for at most a few selected accents of English, usually the British RP on the one hand, and the American GA on the other.  Some CALL programs would also include texts spoken in other accents, e.g. Australian, because learners' acquaintance with a variety of 'Englishes' has been at a premium in the communicative approach to EFL teaching and learning.  But that was the limit of what could be achieved with pre-recorded speech.  The learner could not listen to the same text spoken in different varieties of English as this would require wholesale duplication of recordings, and this was (and is) not feasible for a number of reasons.  A well-tuned TTS synthesis system will allow the learner to pick the accent at will because – to the extent that accentual variation is rule-governed – the developer will be able to program the salient variables, such as, e.g. vowel quality, rhotacism, flapping, and the like for American English.  The simple TTS synthesis plug-ins available for free off the web include the British and American accents as a matter of course (e.g. ReadPlease: www.readplease.com).

2.3. Phonostylistics

Nor is this the end of programmable phonetic variation, of course.  Consider phonostylistics, i.e. variation related to the tempo and style (roughly casualness) of speaking.  While accents are usually rather sharply categorized into their respective pigeon-holes, at least for pedagogical purposes in the EFL setting, the phonostylistic variation is evidently a continuum with no clear-cut boundaries, from overcorrect citation-form enunciation at one end, through reading and scripted speaking, to highly reduced allegro speech at the other end.  In real life all this variation is heavily context-dependent, of course, with subject matter, situation, speakers/listeners, and a host of other factors all playing their roles.  All this colourful phonostylistic kaleidoscope was of necessity mapped onto a grayscale of rather formal speech, with each text frozen forever in one stylistic rendition given to it by the recorded speaker.  And yet, as is well known, understanding casual fast speech in naturalistic native settings is among the hardest tasks which the learner of EFL pronunciation must face.

Style-aware TTS synthesis could go a long way to help learners in their endeavours.  The phonetic exponents of phonic styles in English are (at least partly) rule-governed and reasonably well understood: vowel reduction to schwa, schwa deletion, sonorant syllabicity, palatal coalescence, alveolar assimilations and elisions, for example, are all phonostylistically sensitive, and have gathered a sizeable bibliography.  It is, I believe, mostly a matter of time for the TTS engines to be equipped with appropriate phonostylistic routines and algorithms.  These will not only contribute to the overall impression of naturalness of synthesized speech, but will actually support a variety of phonostylistics-oriented tasks and exercises in CALL packages.  Even (learners') electronic dictionaries would benefit, for – although today they only speak headwords rather than definitions or example sentences – even single lexical items spoken in isolation can be phonostylistically more or less appropriate: those marked as informal, slang or taboo are sometimes paradoxically pronounced in an incompatibly high style by the bored list-reading speaker.

To the extent that phonetic variation can be coded orthographically, tweaking orthography can crudely simulate a number of phonemic and allophonic processes, such as elisions or assimilations.  For example, entering 'tem players' will TTS-synthesize a bilabially assimilated nasal alveolar, while 'coat' will obviously generate a Polglish-devoiced pronunciation of 'code'.  There is ample space for experimentation here, but care should be taken as various TTS engines will react differently to nonce strings (mostly depending on the availability and structure of the built-in lexicon).  For the sake of my T6 conference (http://www.ictconference.gliwice.pl/) presentation in Gliwice I attempted a rather crude approximation to the in-spe subtle control over phonostylistic aspects of English speech.  I uploaded a short text for synthesis with the Festival TTS engine on the Bell Labs web page (http://www.bell-labs.com/projects/tts/; this service is now discontinued, but see ATT webpage at the URL address specified below).  One version of the text had orthographic alterations meant to simulate some phonostylistic phenomena: final alveolar stop deletion, affricate weakening, nasal 'bleaching'.  The text which I uploaded was (without brackets): "My name is Radek.  I welcome all presen[] in hall C-1 at professor Sobkoviak's lec[s]ure.  My task is convincin[] you abo[w ] the high level of Bell speech synthesis".  The effect (http://elex.amu.edu.pl/~sobkow/tts/RadekFastFestival.wav) is best audially compared to the unaltered version (http://elex.amu.edu.pl/~sobkow/tts/RadekFestival.wav): "My name is Radek.  I welcome all present in hall C-1 at professor Sobkoviak's lecture.  My task is convincing you about the high level of Bell speech synthesis".  Some 'casual speech' effects are of course more convincing than others, but notice that all were achieved by a rather primitive method of orthographic manipulation.  With full control at the level of the TTS engine, real phonostylistics can easily be attempted.

2.4. Foreign accent simulation

The accentual, dialectal and phonostylistic variation within the native norm does not exhaust the pedagogically useful control possibilities in TTS synthesis.  It would be technically rather easy to simulate a foreign accent, for example to better demonstrate to the learner the areas which need improvement (e.g. final devoicing in Polish English).  This is something which few native speakers would be capable of, and while it is not unimaginable to employ non-natives for the job, it would clearly verge on impossible to be able to control just the wanted phonetic variables to the exclusion of others.  Expert non-native phoneticians could be resorted to for a better command of the phonetic interlinguistic intricacies, but this would again be troublesome for other reasons.  The precise phonetic control afforded by a TTS system can hardly be improved.

Simulating a foreign accent of English by computer for didactic purposes is not a new idea.  In 1997 Hyouk-Keun Kim created his Korean Accented English Pronunciation Simulator (http://odin.prohosting.com/hkkim/cgi-bin/kaeps/kaeps_home.htm), rightly noticing that "Most adult ESL/EFL learners [...] do not recognize the problems of their English pronunciation", and that it might be a good idea to demonstrate these under computer control.  Eventually a rule-based KAEPS system was set up, simulating "three types of English pronunciations in the IPA symbols: 1) a phoneme-based English pronunciation, 2) a desirable allophone-based American English pronunciation, and 3) a possible Korean accented English pronunciation".  While Kim's system has never advanced beyond accented graphemic (i.e. IPA) representation, it would be easy enough to attach the IPA-to-speech engine to it.  After all, most TTS systems use phonetic transcription at some stage of the synthesis process.

Notice that, like in the case of phonostylistic variation, L1-accented English speech is really a continuum of interlanguage, roughly correlated with the traditional dimension of proficiency.  To take a Polglish example: initially Polish learners of English will tend to force the English vowel system into the Procrustean bed of their native 6-point system.  At this stage, all of English //:A, /Q/, /Ã/ tend to fuse into Polish /a/.  Later /Q/ tends to emancipate itself and the back and central vowels take their respective positions, with /Ã/ the last to be properly interpolated between Polish /a/ and /e/.  An L1-sensitive TTS system would be able to dynamically adjust its parameters to realistically simulate spoken Polglish at these various stages of proficiency.  Needless to say, the relevant phonetic bibliography on the dynamics of English-targeting interlanguage speech development is by far more modest than that devoted to intra-English variation.  This is particularly true of such minority L1's as Polish, unfortunately.  The bottom line is, then, that much more fundamental research would be needed to feed into the creation of a Polglish speech simulator.

2.5. Non-human speech

If the argument is accepted that listening to variably L1-accented speech, produced and manipulated under computer control, might help learners first notice and then get rid of their accent, the logical extension is to take it one step further, like Keller & Zellner-Keller (2000a) did, noting that "speech synthesis allows [...] the creation of sound examples that could not be produced by a human being (e.g., speech with intonation, but no rhythm)".  Demonstrations like this are located somewhere between teaching foreign pronunciation as one practical language skill (on a par with reading and writing, say) and teaching foreign phonetics as declarative knowledge.  The latter is done as part of the so-called 'descriptive grammar' in the Polish academic philological setting.  The recent rapprochement of these two kinds of courses, partly stimulated by the 'focus-on-form' movement in EFL mentioned earlier, makes the Kellers' idea quite attractive.  Notice, incidentally, the peculiar paradox: TTS synthesis was initially developed (and is still perfected) to make artificial speech sound as human-like as possible, but for some didactic applications it is actually beneficial to create more 'robotic' speech.

2.6. Visual synthesis

The most advanced TTS systems now available go beyond the simple (?) unimodal audio synthesis, into the exciting world of face animation, featuring the so-called 'talking heads' or animated agents.  One of the most successful applications of this cutting-edge computer technology to CALL has been the University of Colorado Center for Spoken Language Understanding (CSLU) "Baldi" project (http://cslr.colorado.edu/toolkit/main.html).  In brief, it is an NLP environment focused on the use of TTS synthesis and Automatic Speech Recognition (ASR), enhanced with the animated face ("Baldi", Figure 1; Baldi has recently been superseded by other faces, but the essentials of the system remain) simulating phonetically realistic articulatory movements in real time.

Figure 1. "Baldi". http://elex.amu.edu.pl/~sobkow/tts/Baldi1.jpg

Visual object programming, speech spectrography and many other components are integrated in the Rapid Application Developer which makes it possible to create a simple dialogue schema in minutes, which can then be built into another application, such as CALL for example (see http://www.haskins.yale.edu/haskins/heads.html for a comprehensive interactive overview of many other 'talking head' projects).

What is most exciting in the package (which is free for educational purposes) is the novelty of using the animated face to enhance speech synthesis and make the spoken exchange more realistic.  Baldi not only moves his lips and eyes to provide the much needed – especially in the context of learning a foreign language – visual information to aid intelligibility.  It can also 'go transparent', exposing the realistically rendered inner articulators in full motion, down to the root of the tongue (see Figure 2).

Figure 2. Baldi gone transparent. http://elex.amu.edu.pl/~sobkow/tts/Baldi2.jpg

This is an incredible resource for pronunciation learners, of course: they can listen to natural (if synthesized) speech and see how it originates in the mouth.  The head is quasi-3D; it can be rotated in all three dimensions with the mouse, and the amount of transparency can also be adjusted at will, the extreme leaving just the articulators on screen.

The CSLU toolkit, where Baldi lives, has so far been used mostly to assist speech and language therapy of native American children, but its application to EFL CALL (and other L1's – Baldi can be programmed for any language whatsoever) is just a matter of time.  Also, it is enough to go to the movies nowadays to see the level of realism which animation of human-like synthetic actors has achieved (e.g. "Shrek" or "Lord of the Rings"; see also Thalmann & Thalmann 1990).  In a few short years animated anthropomorphic agents will be used in CALL, which will be hard to tell apart from video-recorded real human speakers.  One technical consequence of this will be – like with the TTS synthesis – that more CD space will be freed from the enormously memory-hungry current video files.  It is much harder to predict learner reactions to (semi-intelligent) speaking and animated human-like agents acting as conversation partners in settings which are now only available in video conferencing.  Learners may relate to these artificial personas to the extent which may be pedagogically relevant, with both its pros and cons.

3. TTS on-line demos

There are many players in the field of TTS synthesis, both academic and commercial.  Most would offer some information about their research and products on the Internet, including passive and interactive demos.  The latter are of course by far more exciting from the point of view accepted here, i.e. that of a EFL pronunciation teacher.  While most demos would only allow to enter a rather short piece of orthographic text for synthesis, some measure of deliberate spelling manipulation is afforded (like I did above with Radek). 

Without attempting a comparative description and analysis of the many TTS systems available on the Web, I will close this section with a few best-known URL addresses, and let the interested reader experiment on his/her own.

Some of the best-known TTS systems and their manufacturers:

·         Centre for Speech Technology Research: http://www.cstr.ed.ac.uk/projects/festival/

·         ScanSoft's RealSpeak: http://www.scansoft.com/realspeak/demo/

·         Prosynth Project: http://www-users.york.ac.uk/~lang19/

·         ATT Labs: http://www.research.att.com/projects/tts/

·         Elan Speech: http://www.elantts.com/

4. Conclusions

Text-to-Speech Synthesis is of course only one branch of the NLP tree.  In this text no attention was paid to Automatic Speech Recognition (ASR), Machine Translation (MT), multimodal man-machine communication, automatic information extraction (data mining) and summarization, language generation, multilingual resources, speaker/language identification, speech evaluation techniques, Artificial Intelligence (AI), and a number of other areas in one way or another concerned with natural language processing.  All of them are potentially of interest to foreign language educators.  Some have been recognized as such, and work is going on exploiting their potential, e.g. ASR.  Some are still in statu nascendi, struggling with technological problems and lack or inadequacy of relevant linguistic research, e.g. AI.  But it is a safe guess that sooner or later (most probably – sooner) all of these human language technologies (HLTs) will arrive at the door of Foreign Language Teaching, EFL in particular (and soonest), demanding to be accommodated.  FL teachers should better be prepared, lest pedagogy be compromised for sheer technology.

References

d'Alessandro, C., Liénard, J.-S. (1996) "Synthetic speech generation". In R.Cole et al. (eds). Chapter 5.2.

Cole, R. et al. (eds) (1996) "Survey of the state of the art in human language technology". http://cslu.cse.ogi.edu/HLTsurvey; accessed 13.4.2003

Delcloque, P. (ed.) (2000) Proceedings of InSTIL: Integrating Speech Technology in Learning. University of Abertay Dundee , Scotland .

Doughty, C., Williams, J. (1998) Focus on form in classroom second language acquisition. Cambridge : Cambridge University Press.

Dutoit, T. (1997) An introduction to text-to-speech synthesis. Dordrecht: Kluwer Academic.

Dutoit, T. (1999) "A Short Introduction to Text-to-Speech Synthesis". http://tcts.fpms.ac.be/synthesis/introtts.html

Ellis, R. (ed.) (2001) Form-focused instruction and second language learning. Oxford : Blackwell.

Granström, B., House, D., Karlsson, I. (eds) (2002) Multimodality in language and speech systems. Dordrecht: Kluwer Academic.

Hovy, E. et al. (eds) (1999) Multilingual information management: current levels and future abilities. http://www.cs.cmu.edu/~ref/mlim/index.html

Keller, E., Zellner-Keller, B. (2000a) "Speech synthesis in language learning: challenges and opportunities". In P. Delcloque (ed.)

Keller, E., Zellner-Keller, B. (2000b) "New uses for speech synthesis". The Phonetician, 81, 35-40.

Thalmann, N., Thalmann, D. (1990) Synthetic actors in computer-generated 3D films. Tokyo: Springer Verlag.

MULTIMEDIA, THE WEB AND FORMAL EFL EXAMS

by Pawel Topol
Adam Mickiewicz University ,
Poznan, Poland
topol@amu.edu.pl
http://main.amu.edu.pl/~topol

Abstract

Most CALL software available are material-presenting packages or sets of exercises and tests (or a combination of both). The article discusses two (author and co-author) examples of how computer technology can support real EFL exams:

·       A multimedia CD-ROM package, “FCE”, that prepares for the First Certificate in English exam. The package contains more than 1,000 linguistic tasks, as well as two full-time simulations of the real exam.

·       The “Multimedia Distant English Courses…” for Polish students of technical universities (within the Leonardo da Vinci European Union project). The student learns, practices and takes on-line tests and exams.

Background - Glottodidactic Software

Computer software supporting learning and teaching has a long history. Educational programs were popularized shortly after the introduction of home computers, for example Sinclair Spectrum, Commodore and Atari. The programs were usually single tests, small sets of exercises and/or short practical tasks for the learner. Their modest forms and content could be explained by limited memory and generally low capabilities of contemporary hardware. The scope and shape of educational programs changed and expanded with time, together with the rapid development of computer technology. During the last twenty years or more simple computer-based didactic tasks transformed into complex, multipurpose, multimedia packages.

Languages, especially English, were those “privileged” teaching subjects in humanities that were fairly well supported with educational software. It seems now that English as a foreign language (EFL) has the richest library of computer educational tools among all languages, if not all teaching subjects. This situation is rather natural: English has been an international language for years, first computers came from English-speaking countries. The status of English has been strengthened lately by the popularization of the Internet and WWW.

Are there many EFL computer packages available on the Polish market too? I think there are very many interesting programs. Unfortunately, they are badly advertised. Most ads appear in popular-science magazines, e.g. CHIP, Enter, ComputerWorld. I am afraid they are not periodicals that are extremely popular among English teachers. A few years ago I started my own database which contained basic information about CALL (computer-assisted language learning) packages supporting EFL released in Poland . The database contains more than a hundred records now. This means that an English teacher or a learner of the English language has a choice of over 100 CD-ROM packages in Poland now. It is worth mentioning that many of them are sets of CD’s. Some contain 6, 8 or even 10 discs (“Euro Plus+”, “Tell Me More”, Langmaster series etc.). However,  the information about the programs came mostly from those popular-science magazines mentioned above. I am sure there are a lot more packages available on the Polish market.

What kinds of CALL packages do we find then? I would divide them into four groups: (1) multimedia encyclopedias, thematical guides, lexicons and other encyclopedia-type publications, (2) language courses, electronic books and different material-presenting software, (3) language games, learning-through-entertainment software, and (4) programs checking the learner’s knowledge or skills: collections of exercises and tests. The first group comprises mostly encyclopedias (e.g. Britannica, “Microsoft Encarta”, "Microsoft Bookshelf for Windows”, Hutchison, Grolier) and dictionaries (e.g. Collins Cobuild, Webster, Oxford ). The second group is represented mostly by recognized EFL courses published previously in the traditional, paper form, now transferred onto an electronic carrier, CD-ROM. The packages are, of course, supported by exercises and tasks that make use of the computer’s capabilities, and would be hard or impossible to provide traditionally, on paper (e.g. “Euro Plus+” based on the “Flying Colours” series by Heinemann). The third group are either games which contain some educational content (e.g. “All-in-One Language Fun”) or educational packages supplied with some entertaining elements (e.g. “Lingualand”). The fourth group comprises programs which describe and test selected grammatical, lexical or phonetic issues (e.g. “Professor Henry” series, “Fonetyka w pigułce”). Large collections of exercises and tests also belong to this group (e.g. “PopEnglish”, “ETeacher”).

In the discussion on different CALL applications, we cannot forget about the Internet. I would distinguish two types of language teaching/learning aids here: educational webpages and web-based CALL software. There are hundreds or thousands of EFL WWW pages. Let us systematize them now:

·        homepages of educational publishers known from their traditional, paper production: coursebooks, periodicals, books (Longman, Cambridge Press, Collins, etc.);

·        pages by editors of dictionaries and encyclopedias (Webster, Oxford , Collins, etc.);

·        homepages of traditional and virtual educational institutions and organizations: universities, colleges, schools, courses and other forms of formal training;

·        educational (e.g. EFL) portals;

·        homepages of commercial and non-commercial educational software publishers;

·        websites – collections of traditional teaching and learning materials: different tests and exercises, English literature in the original or adapted to instructional needs, sets of language tasks for language improvement and many other teaching/learning aids;

·        so-called private pages: websites designed by individual, informal, not institutionalized WWW users: educators, teachers, students and hobbyists. Many of those represent a very high professional standard.

Web-based computer software are programs located in the Internet. They are operated remotely from the individual user’s computer screen. They should not be confused with downloadable software that is located on the web. These programs, when downloaded and installed, work locally from the user’s hard disc. The best examples of remote (web-based) CALL software are on-line dictionaries, encyclopedias and glossaries. On-line language courses belong to this group too. Connecting with such a page activates a special remote program which guides the user through the course, presents material, runs exercises, displays results, and performs many other organizational tasks.

Summing up, almost all CALL programs support the process of learning and teaching. The teacher receives tools that can be incorporated in the classroom along the language course. The learner can use different programs individually in order to improve the general knowledge and skill of the language, or work on selected linguistic issues. Thus, CALL packages are teaching/learning aids, additional to all sorts of traditional materials.

Very few computer applications are designed for formal examination use. On the other hand, very few existing formal, official language exams make use of computer technology, in Poland too. One of the reasons is that software developers and producers are not willing to get involved in special-purpose computer applications for marketing reasons. However, I believe the market response to such software production could be quite positive, at least in the case of common, official or state exams. For example, a program preparing for “matura” (secondary education final exam), or English Departments entrance exams at Polish universities, would sell quite well.

EFL exam on the computer screen – a local example

It is generally known that using teaching aids and materials in addition to the basic textbook along the course of a foreign language is a highly preferable activity. Similar dependence takes place on the side of the learner. The student can improve the process of learning by using different additional tools and materials, including computer software. In the case of a general English course the situation is rather comfortable, because the variety of different programs available is very wide. It is much worse in the case of real formal exams. Such packages are hardly ever produced.

An example of such a package I would like to describe briefly refers to the worldwide known examination, Cambridge First Certificate in English. The exam is very popular in Poland , and its popularity is growing year by year. FCE exams are performed by local agendas of the British Council in several cities and towns in Poland (more information is  available directly at http://www.britishcouncil.pl/education/polfce.htm, the website of the BC Polish branch). The package I am introducing holds the same name, “FCE”, and has been produced by SuperMemo World, a computer company in Poznań , Poland . More information about the package can be found at http://www.supermemo.pl/index.php?page=21160 .

This is a multimedia package on CD-ROM which prepares for the FCE exam mentioned above. It is neither a textbook nor an English course – it is a huge collection of exercises and linguistic tasks which can help the user estimate his/her linguistic competence and skill in the context of the FCE exam standards. The package contains over 1,000 tasks similar to those the user can expect at the real examination. The word similar is used here intentionally – one should not expect exactly the same questions. However, I have tried to make all the tasks as close to the original ones as possible, both in content and form. Hence the package “brings” the user much closer to the real exam tests than most or even any other preparatory, computer-based materials.

The linguistic content is based on both formal and colloquial English on the intermediate level. The exercises are divided into five groups according to the division of the original exam into the following parts parts: Reading , Writing, Use of English, Listening and Speaking. The groups are comparable in size, which means that the user can expect about (or a little more than) 200 questions and tasks in each group.

The huge collection of exercises constitutes the main part of the program, however, there is another powerful feature of the “FCE” package: exam simulations. This is a separate module which contains two simulations of the real FCE exam. Here, the user is given questions to answer, problems to solve and tasks to do in the same sequence and similar form. The time is limited and the user is given the score after completing the whole exam. This part of the package is quite exceptional when compared with other EFL multimedia production. Presumably no other program available in Poland , or elsewhere, does the two things at the same time: prepares and tests exactly for a specific exam, and simulates it.

A crucial question in this discussion is the reliability of the simulation, to what extend it is possible to imitate real class situations, whether all the examination tasks can be performed via the computer screen effectively,  whether the final grade is representative, etc. First, let it be stated clearly that “full” computer simulations of face-to-face situations are hardly possible. Even the best, the fastest and the most capacious modern computers cannot handle a natural discussion. The situation is similar with open questions. The longer the student’s utterance is, the more difficult it is for the machine to assess it. Thus, there are parts of the exam that cannot be simulated by the computer by any means, e.g. essays. The computer can check spelling quite easily, though it is much harder with grammar, or almost impossible with style. It is far beyond the computer’s capabilities to take a position on the student’s metaphors, wit, irony or meta-language. One can hardly imagine a situation in which the computer could replace the examiner in the Speaking part of the exam.

The Reading part is rather easy to simulate. Even if the text is too long to fit the computer screen, the user can scroll it in a separate window and answer the given questions. The Use of English is probably the easiest part to be adapted to computer-based exercises – it checks grammar, syntax, vocabulary, word collocations and other elements which can be atomized, i.e. presented on the screen in the form of short questions and answers. The Listening part can also undergo certain atomization. First, the student listens to the whole of the text. When specific questions appear on the screen, the user is given opportunity to play back the part of the text the question refers to, usually two or three paragraphs. This portion is short enough to draw the user’s attention to the issue, and long enough to mask the issue in question. Speaking had to be atomized because performing a real conversation with the computer is still technologically impossible. The student is given a question and then asked to answer it aloud. Then a number of alternative answers are displayed, and the student can compare his/her response with the suggested ones.

The same method of comparison is used in Writing, the most difficult part to simulate. The student’s task is to write an essay. First, the title is given and a more detailed explanation of the task. Then, specific instructions of how the task should be performed are displayed, together with the very detailed criteria of how the essay will be evaluated and assessed. The student is asked to write the essay on paper, read it again and compare the essay with a model answer. A model essay is displayed on the next screen. The student is then asked to grade him/herself according to the grading scale given before. It is enough to click on the appropriate button on the screen: “excellent”, “very good”, “good”...

One would say the reliability of the test suffers when students grade themselves. This is true, however, the dilemma is whether to include such an element (being aware of its limited reliability), or to exclude the whole part from the simulation. The former solution seems to be good compromise. Nevertheless, we are considering hiring a tutor. The user could send in the texts electronically, the tutor would correct them, and the learner would receive them back with some additional comments.

The learner can use both the simulation and the main parts of the package repeatedly. After the simulation is completed, a report is displayed which then can be saved on disk for future reference. As for the main parts, each answer to single questions is recorded by the program. Then the computer calculates the so-called forgetting curve, and plans in time  repetitions of unanswered or wrongly answered items.

The main types of activities in both simulations and main parts of the package are:

·        multiple choice test (one correct choice out of three or four);

·        filling blanks (missing word or phrase in a sentence);

·        matching exercise (drag-and-drop technique);

·        pointing the appropriate element on the screen (e.g. “click on the person’s forehead”);

·        indicating the incorrect element in a sentence;

·        descriptive tasks which require longer utterances, exemplified by model answers and possible alternatives.

Almost all texts that appear on the screen have been recorded by native speakers. The user can play them back freely at any time.

Both the simulations are time limited, they are 3-hour versions of the test. In my opinion they can help the student feel the climate of the genuine FCE exam. I also believe that people who prepare to the exam individually can benefit from the package as well as those who attend official preparatory courses.

EFL course exam – a remote example

The computer screen, or rather its working area, is called “Desktop”. The term is indeed accurate: the user arranges this area with most frequently used tools, just like on a regular desk. The computer desktop presents them in the form of icons. The icons are actually shortcuts to programs installed on local drives, to documents or folders, as well as remote applications located “somewhere” on the web.

Formal web-based instruction takes usually two forms: (1) there are virtual educational institutions which function totally on line, (2) some “traditional” schools and/or universities offer selected courses that can be available electronically, in the Internet. They are often specialization courses or courses for special purposes, integrated projects, or single subjects performed remotely – foreign languages including. I would like to present an example of such a virtual, remote EFL examination application.

The EFL platform I will discuss briefly belongs to the Leonardo da Vinci Project performed under the aegis of the EU. Generally speaking, it is a multi-module course dedicated to students of Polish technical universities. The course covers some specialized subjects, technological issues, and is provided in English only. The full name of the program is „Multimedia Distant English Courses for Polish Users in Legal, Banking and Finance, Science and Technology, and Safety Training Sectors with Elements of European Union Regulations and Standards”. The venture is a result of co-operation between two Polish technical universities: Poznań TU and Gdańsk TU, and the University of Wolverhampton, Great Britain.

“The Leonardo da Vinci (LdV) is one of the three education programmes of the European Union implemented in Poland which contribute to the implementation of its vocational education and training policy. The overriding aim of the programme is to promote activities focused on upgrading the quality of VET systems and adapting the system of education to the labour market needs. The programme provides a framework for transnational cooperation in the sphere of raising quality, fostering innovation and promoting a European dimension in vocational education and training systems, approaches and practices.”

The above definition comes from LdV materials. More details are available at  http://www.bkkk-cofund.org.pl/, the site of the Polish Agenda of the program – “The Task Force for Training and Human Resources”. The largest source of information about LdV can be found at http://europa.eu.int/comm/education/leonardo/leonardo2_en.html.

As I mentioned above, the course is dedicated to Polish students, though provided in English. A general idea of the project design was to join the content-related issues with the communicative aspects. The objective of the course is to train the student in specialization subjects on the one hand, and to check the student’s linguistic competence on the other. I am a co-author of one of the modules.

The Internet Technologies Module (P. Topol, A. Adamczyk) takes a special position in the whole course because it covers the issues which should constitute basic knowledge of the Internet among students of any technical school. Therefore, this module has been made obligatory for all the students undergoing any of the specialization modules. Here is the list of most of the issues discussed in the ITM:

·        standard Internet services (electronic mail, telnet, ftp, discussion groups, irc, gopher, www);

·        advanced services (video on demand, audio- and videoconferences, distant education, e-commerce, e-banking, ip telephony, push channels, portals and vortals, etc.);

·        internet services and browsers (Microsoft Internet Explorer vs. Netscape Communicator, off-line browsers and Teleport Pro);

·        html and additions (server-side additions, client-side additions);

·        authoring tools for the www (programs for graphics authoring, programs for music authoring);

·        intranet – practical issues (creating and managing resources);

·        technologies and tools of the intranet;

·        security issues (network security, information encryption).

The student reads portions of information on a few screens and then does exercises. Each fragment (chapter) is provided with tests, and there are additional exercises and tasks at the end of the whole module. There are different types of manual tasks, e.g.:

·         different forms of multiple choice test: pointing the appropriate answer with the mouse, clicking on the correct element (checkboxes, radio-buttons), selecting the answer from a pull-down menus (classification of correct answers);

·         filling gaps: writing words or phrases;

·         matching (drag & drop technique): two columns of items, hierarchy, ordering.

There are also problem-solving activities. The first screen of such an exercise introduces a short text to be displayed on the next screen, and presents some keywords, key phrases or key issues the student should pay special attention to. Then the full text is displayed which never exceeds one computer screen. The student reads the text without time limits. The third screen shows the correct answers and gives some supplemental information and/or comments. Sometimes screen 3 contains additional tasks for the student and the answers are given on screen 4.

Many exercises are supported by audio recordings performed by native speakers. In some exercises the student listens to a text first and then writes the answer. There are many exercises where the student can check the pronunciation having given the answer in the written form. The main texts in the module have not been audio-recorded because of technical and organizational reasons. This issue is being discussed now, they may be recorded in the next edition of the project.

The student’s answers in exercises and tests are checked automatically by the computer in most cases. One of the main goals of the project design was to make the process of student assessment possibly highly automatic. There are situations, however, where some corrections have to be made by a tutor or a person monitoring the student’s progress.

The Leonardo webpage can be entered in two ways: as a student and a tutor. The student is allowed to look through the course and perform tasks. Each time an answer is given, the student is informed about its correctness immediately. The student can go through an exercise more than once, then the computer stores the results for each take respectively. The tutor can monitor the learners’ achievements: which student completed which tasks with what result. The tutor can browse all the student’s answers, comment on them and also make corrections to the grades given by the computer. This last feature has been added in case an unpredictable factor occurs, and a “manual” correction or assessment will be necessary.

Each module is supplied with some additional tools: chat, calendar and discussion group. Students can chat both with other students and tutors. The calendar functions as a mutual reminder of events, terms, due times, etc. Both tutors and students can add to the calendar. This refers to the discussion group too.  All the tools are internal parts of the system. They are available only for those who have logged in to the server. Logging is authorized and password-checked.

The student completes the course and a final grade is calculated. All the student’s answers along the module are taken into consideration: both from intra-tests attached to each chapter and from the final tasks. The final result is actually a compound of the student’s professional preparation as well as linguistic competence. Thus, LdV is neither a strictly technical course nor a typical EFL project – it is a combination of both.

Conclusion

Both examples described in the article were put in practice in 2001. “FCE” met with positive reviews in popular magazines. It was among a few packages that received an award of the “Product of the Year 2001” in the category of educational computer software, granted by the “CHIP” magazine. The market response to the package is also optimistic. However, no research has been done on its actual usefulness, i.e. if it really proved a value for those who used the package for their own practice and then took the FCE exam. I am considering a survey that would gather such opinions. There are obstacles, e.g. how to reach those people, though the Internet might be the right tool to perform such an investigation.

The latter example, the LdV system, received positive reviews too. It was tested by students of Technical University in Poznań . The project is by all means successful, however, there are elements both in the system and the courses that need improvement. It is worth mentioning, however, this was the first phase of the LdV program. Now we are looking forward to its second edition.

To conclude with a general remark, EFL examinations seem to be a forgotten or neglected area on the computer market. Software developers and producers concentrate mostly on general language courses or multi-purpose sets of exercises. The projects discussed in the article show that supporting foreign language examinations by computer technology can be beneficial for both parties: those who take exams and who organize them.

SOURCES AND SUPPORT FOR PRESERVICE ENGLISH TEACHERS AS TECHNOLOGY USERS
by Roberta Devlin-Scherer
Seton Hall University , South Orange,
New Jersey, United States
devlinrb@shu.edu

Abstract

This article describes technological applications (Inspiration Software, Puzzles, Games, and Quizzes, Virtual Field Trips, WebQuests) that English Language Arts preservice teachers were able to develop and conduct with middle school students in two urban school sites.  Preservice teachers’ reactions to their efforts are included with recent web sources that guided their work. Successes and problems are noted. Cooperating teacher encouragement, access to computers, and support from the university class appear to be factors that contributed to their accomplishments.

Introduction

Internationally, there is interest in infusing technology in English programs and concern for adequate training for teachers (Hughes, 2003). Reports, articles, and conferences in the United States have advocated adoption of technology to foster student learning (DOE, 2000; McLester, 2002; NASBE, 2001; November, 2001; NECC, 2001; ICTE, 1999; Wisconsin University System, 2001; Wodarz, 2001). The timing is appropriate. An anticipated teacher shortage in the United States provides teacher education an opportunity and responsibility to revise their programs to prepare preservice teachers to use technology effectively in classrooms. Since 1999, the United States Department of Education has supported an initiative for promoting technology use by educators, the Preparing Tomorrow’s Teachers to Use Technology (PT3) grant program. Participants have been charged with forming partnerships to implement innovative ways to address the need to expand the use of technology in K-12 schools and in teacher education programs. Preservice teachers are beginning to learn computer skills in their teacher education programs, but often in isolated courses. They usually have not used technology with students in their field experiences (Gibson, 2002; Whetstone, L. & Carr-Chellman, A., 2001).

During the three years of the PT3 grant at Seton Hall University in South Orange , New Jersey , the university worked closely with two nearby urban middle schools to increase the use of technology in teaching. The purpose of this article is to discuss technology applications English Language Arts (ELA) preservice teachers were able to use with grade 5-8 students as well as the sources that enabled them to develop the varied applications. ELA preservice teacher perspectives on their lessons are shared as well as factors that contributed to their success.

Background

For nearly a decade, Seton Hall University has advocated and supported faculty use of technology. All undergraduate students and faculty have been provided laptops, and training for faculty on ways to infuse technology in their teaching has been continuous and extensive. In 1999, EDUCAUSE recognized the university for superior campus networking. In 2001, Yahoo Internet Life’s “America’s Most Wired Colleges” ranked the university 14^th among United States' campuses and first among Catholic universities, for its hardware, academics and services in technology.

The secondary education program at Seton Hall has a strong commitment to develop teachers who become reflective practitioners with the depth and breadth of knowledge needed to make informed decisions about student learning. Increasingly, these decisions involve uses of technology. As part of the PT3 grant, the program has continued to examine and employ technology in its courses in order to help preservice teachers use technology effectively. In Computer Fundamentals, a freshman requirement, students developed projects using Microsoft applications, Excel, PowerPoint and created web sites, a helpful background to lead into teacher education students’ creation of technology applications in a field experience associated with the Instructional Theory into Practice course. In this class, preservice teachers reviewed lesson plan sites for technology lessons, planned and taught lesson plans using different models of teaching, and demonstrated and discussed various ways to infuse technology into teaching. They were taught how to use the Inspiration program for webbing ideas for writing or project development. PowerPoint informational presentations were discussed as ways to stimulate student conversation and introduce open-ended questions about the material as well as to serve as outlines of important ideas. It was noted that virtual field trips and web quests could be developed with PowerPoint. A 60 hour field experience in urban schools was connected to this course. In the class, website resources which provided descriptions and examples of the technology were shared, and some class time was devoted to development of the technology application to help preservice teachers get started. Preservice teachers were allowed to work in teams to complete and present technology lessons. Technology project descriptions in the course are in Appendix A following the references.  

While this class is composed of all subject majors, this article will focus on the work of English majors only, who are typically the largest group in the class. Selected readings about technology practices in schools and discussions with colleagues in the PT3 Grant and in local schools were the basis for technology projects implemented in the class. In the next sections of the paper, the technology application will be described, ELA preservice teacher reactions to conducting the technology application with students will be shared, and current sources for its development will be listed. The sources for the technological applications are listed with brief descriptions, so readers will be able to find those that interest them. In addition, samples of ELA preservice teacher work are provided on the Portfolio Handbook in the Techprojects section found at http://pirate.shu.edu/~devlinrb/portfolio/techprojectsdefined.html (when users go to the eportfolio to see Techprojects preservice students created there will occasionally be a dialogue box asking them to put in a password, they should click cancel and will be able to proceed).

Technology Lesson Plans

For this first assignment, ELA preservice teachers were asked to survey lesson plan sites and find five technology sample lessons that focused on different teaching areas in English: literature, writing, grammar, media. They were to select lessons that fostered communication, collaboration, and critical thinking (Flores, A., Knaupp, J.E., Middleton, J.A., & Staley, F.A., 2002). Below are selected representative popular sites; others are available under the title, Helpful Lesson Plan Sites, at http://pirate.shu.edu/~devlinrb/ on the right side of my homepage.

Lesson Plan Sites: http://www.awesomelibrary.com; http://school.discovery.com/schrockguide/index.html; http://www.atozteacherstuff.com

Cooperative Learning Lesson Sites: http://sps.k12.mo.us/coop/cybercoop.html; http://www.educationplanet.com; http://www.cotf.edu/ete/modules/modules.html

For ELA preservice teachers an extensive, specialized list of sites was developed, in these categories: media literacy, writing, grammar, and literature. English Language Arts Sites, grades 4-12, is at http://pirate.shu.edu/~devlinrb/ on the right side of my homepage.

Teacher Web Site

Preservice candidates found developing this web site at http://www.teacherweb.com for their cooperating teacher easy to use and update (TeacherWeb.com, 2000).  Writing up the “T