Packed PowerPoint slide show here.

The paper describes the ideas and experience of designing and developing interactive multimedia courses for university students, an alternative method of teaching and learning foreign languages. The project is sponsored by the EU Leonardo da Vinci programme.

Distance Learning (DL) is attracting more attention and has started playing an increasingly important role - not only at universities but also all other places. DL offers independent studies by enabling immediate access to customised learning. The best results are achieved when coupled with interaction with a tutor. DL has some disadvantages, and to compensate for lack of the traditional face-to-face classroom environment, heavier use of multimedia, also voice recognition is desired. DL courses equipped with the access to digital libraries will form a base for knowledge centres.

The MDEC project serves the purpose of preparing the university students for better use of English language in the following areas:

• Legal issues,
• Science & Technology issues,
• Internet Technology issues,
• Occupational Safety issues,
• Finance & Banking issues.

The paper gives a description of organisation of the courses and description of language tasks, which are of synchronous and synchronous type. Asynchronous material is meant for individual study with special vocabulary lists, grammar resource links, audio and video files as a student support.Synchronous sessions are in the form of chat modules.

Testing results of pilot groups are presented. They concern the evaluation of the following elements:

• the level of difficulty and appropriateness of the language material,
• the selection of exercises,
• the effectiveness of the use of multimedia,
• the usability of interfaces,
• the modes of assessment.

Our aim is to examine two different accounts of the structure of Polish numeral phrases (NumP's) from the NLP point of view. We regard numerals as both (a) practically important (frequent in a corpus of a language) and (b) theoretically interesting (because of their unique syntactic behaviour). We want to present theories implemented, i.e. working automatic syntactic analyser (ASA), not just elegant and logically sound account. Thus theoretical issues will be supported and illustrated by examples of analyses produced by syntactic analyser ASA.

By a NumP we understand here an NP that contains a cardinal numeral (simple or complex) and a nominal constituent. The approaches referred to below are the following: (a) a "classical" one, language-specific, given in Saloni-Świdzinski (1981), and (b) a "generative" one, UG-oriented, within the HPSG and P&P framework (Przepiórkowski (1999), Rutkowski (2000)).

NumP's seem to be one of the most challenging problems of a rigorous description of Polish. Various practical consequences of the choice of the theoretical background for our automatic syntactic analyser ASA is discussed in the paper. We limit ourselves to problems connected with NumP's in Subject position:

(1) Osiem nowych wiolonczel wymaga jeszcze nastrojenia.

eight new cellos need_3pers,sing yet tuning

"Eight new cellos still need tuning."

seven-hundred twenty three percussionists played_3pers;pl on oboe

"Seven hundred and twenty three female percussionists played the oboe."

We also examine adjectival phrases that may agree with constituents of the "numeral subject" in a number of ways:

these_gen=nom eight sopranos_gen were tipsy

"These eight sopranos were tipsy."

These_gen=acc,pl seven basses_gen seemed on tight-fisted_acc,pl

The main issue discussed in the paper is that of Case for "numeral subjects". We implement two accounts: a nominative interpretation, typical of the "classical" theory, and an accusative one suggested within the generative paradigm. We compare the results of both types of the analysis, presenting parsing trees for each of them.

Although we do not incline towards either of the two solutions, the former turns out to be more accurate, as the one which is Polish-specific. Actually, automatic analysis of Polish sentences is a purely technical business. When designing a parser for Polish, we do not see any need to take other languages into account; UG is of no use for this purpose.

In our opinion, implementation provides the best technique of evaluation of linguistic theories nowadays. We believe that only those formal models that are practically applicable as, e.g., disambiguation devices fit in with the demands of the XXI st century linguistics.

Przepiórkowski, A.: Case Assignment and the Complement/Adjunct Dichotomy. A Non-Configurational Constraint-Based Approach. Ph.D. Dissertation, Tuebingen 1999.

Rutkowski, P.: Składnia polskich liczebników głównych w ujęciu generatywnym. MA Thesis, Warszawa 2000.

Saloni, Z., and Świdzinski, M.: Składnia współczesnego języka polskiego. Warszawa 1981 (2nd ed. 1983, 3rd ed. 1986, 4th ed. 1998).

General overview

Teachers of English face new challenges due to the development of Information Technology and the Internet. The achievements of technology may change most of their teaching practice. Computers as tools for teaching and learning have become widely widespread. The Internet as a source of information and a means of communication, is cheap and also easily accessible. If teachers of English don't want to remain relics of an ancient civilisation, they must learn how to apply the new technology into their teaching practice, and reconsider the basic principles of teaching a foreign language. The first questions are: What is the real-life use of English in the era of the Internet? What does "teaching for communication" mean with English being a global language? Although these basic questions may never be satisfactorily answered, they are still worth considering.

Everyday perspective

There are challenges which teachers face in their everyday practice. The government programme Interkl@sa equipped almost 6000 schools with a small lab of 10 computers. These are mainly used for teaching basic computer skills, but they can also be used for teaching a foreign language. Thus, teachers of English are expected to know how make use of computers in teaching and learning a language via the Internet. Language teachers have never learned with the help of computers, and have never been taught how to teach with computers. This is challenging teacher training institutions. Access to the Internet is transforming education on everyday basis. Changes caused by IT have to be implemented very quickly because the coming generation will be computer literate and they will require up-to-date standards of technology in education, particularly in language teaching.

Description of the research

According to the Language Learning and Technology journal the new situation of a foreign language teacher has never been examined before anywhere in the world.

To study this, I have prepared a questionnaire survey intending to find internal and external factors which prevent language teachers from using the computers and the Internet in class.

Four hundred schools randomly chosen out of the six thousand schools involved in the Interkl@sa programme all over the country have been sent the questionnaire. The number of schools chosen from each of the 16 administrative regions of Poland is in proportion with the number of schools equipped with computers in the region.

Teachers were asked about:

• their professional background
• their computer literacy
• access to computers
• their willingness to learn computer skills and computer-based methodology of teaching a foreign language
• their opinion on computer assisted language teaching
• main obstacles in using computers for language teaching
• factors which may help them to introduce technology-enhanced teaching.

The research is at the stage of collecting the data. It will be completed by the end of February 2001. The results are expected to help understand the situation of Polish teachers of foreign languages, and prepare a training programme which should encourage them to apply computers in their teaching practice.

The wide spread of the Windows sound file format has dramatically extended the possibility of developing digital speech processing (DSP) software based on the commonly used PC soundcard, which has practically removed into oblivion the expensive autonomous devices. A range of PC-based DSP software systems have been developed in Europe and America and the research worker is often at a loss as to which product to buy.

To meet the need for reliable evaluation of DSP software, we have reviewed a number of currently developed systems, evaluating the possibilities they offer. In particular, we have paid attention to the quality of spectrographic image, formant and F₀ tracking accuracy, possibility of spectral filtering and ease of editing in the time dimension. We have also evaluated the accuracy of a number of miscellaneous, sometimes unparalleled, functions DSP programmes may be employed to perform.

In addition to these professional applications, with the phonetician as the user and the focus on his needs and expectations, general user-friendliness of the programmes was examined to learn about the chances of possible employment in non-professional environments beyond the research lab, especially in the field of teaching foreign language pronunciation and the correction of speech pathology in hearing deficient patients.

The results of our analysis have revealed that spectrographic visualization of sounds obtained in most DSP programmes in the default mode suffers from inadequate images which can be repaired only by tedious operation of the variables in the setup, which can indicate a breach between software designers and laboratory practitioners. The only notable exception in by far the best default setup producing well balanced dynamic spectrograms with easily operated controls was found in SpeechStation 2 by Sensimetrics.

Ever since the computer-mediated modes of communication have emerged and, subsequently, started to enter the foreign language school curriculum, numerous attempts have been undertaken by the researchers to assess and fully understand all the corollaries that electronic network communication really brings about, especially when placed within the pedagogically grounded, educational perspective. As computer networks have led to the previously unthinkable and, often, truly exciting new possibilities in transforming the way people communicate with each other, as well as allowed possibilities for substantially more equal and active participation in the process of knowledge creation, the great expectations of how computer networks are bound to enhance foreign language learning in the foreseeable future have been overly aroused. Historically, however, educators' expectations regarding the degree to which new technologies indeed do revolutionise language learning have not necessarily been borne out in practice.

Therefore, this paper will take a close look into both the advantages and disadvantages of the electronic mode of communication and, by drawing on the "critical perspective approach", will report on the already-known empirical research in this particular field of computer-assisted applied linguistics. Consequently, an attempt will be made to thoroughly analyse the variety of factors involved in the process of bringing the learners together with texts and with other speakers of a target language in networked environments.

Power Point slide show here

I wish to report in this paper on the problems encountered in my PhD work (http://main.amu.edu.pl/~przemka/rsearch.html), where, using semi-manual word combination retrieval methods, I attempted to compare quantitatively the correlation between the proficiency levels of writers in English and their application of idiomatic combinations involving high-frequency primary verbs. The necessarily small, argumentative/expository collections (partly pooled from the International Corpus of Learner English resource) feature advanced and intermediate EFL learner varieties, native English learner varieties, native English expert writing, as well as contrastive native-tongue material (Polish).

Some of the reported difficulties, as always in corpus work, will apply to compilation and sampling standards (the ubiquitous issue of representativeness); others will concern the inadequacies of computer tools that complicate, bias or block findings in learner language phraseology. Using as illustration sample analyses of the phraseology of GIVE and TAKE (as well as other primary verbs, if needed), I would like to raise, inter alia, the issues of:

1. how text quality can affect quantitative comparisons of idiomaticity (e.g. topic skewness; spelling variation, incl. hyphenation; proficiency level);
2. how the error margin in POS-tagging accuracy can affect phraseological statistics;
3. how the bottleneck in semantic and other word disambiguation, also in learner language, constrains corpus-derived results in phraseology;
4. how collocation statistics (MI, t-score, z-score, multi-word chunks) tend to yield pedagogically uninteresting data;
5. how corpus analysis software tools are unsuited to fast and smooth annotation and processing of phraseology, especially in learner language.

At the end of the presentation, I would like to point out the features of a browser/ concordancer/ editor dedicated for faster learner-corpus work (hot links with a dictionary, enhanced annotation facilities, powerful corpus saving options, etc.)

Power Point slide show here.

1. Introduction: online learning of English (online lessons) -- what it is, what are its benefits over traditional coursebook instruction (recency, authenticity, novelty, variety, choice, interactivity, etc.)
2. The elements of an online classroom: equipment (lab, hardware, software), students, teacher -- what should each part be like, the consideration of how the three are changing, the conclusion that while the two others are sure to improve in the very near future, the teacher is bound to become the weakest point in the whole process
3. Teachers of English nowadays: different level of education, different educational backgrounds, different level of computer skills, different approach to in-service training -- but Information Technology as one of the requirements of awans zawodowy (professional promotion), using the Internet in the teaching process, WWW, email, class website
4. Training online teachers: on the one hand need to teach them technical skills (including operating the mouse), working with the software, equipment, on the other hand it is more crucial to show them the possibilities the Internet offers to language classroom, how it can assist and replace the coursebook, help the teacher, etc; what are teachers' expectations; what are school administration requirements
5. The reality of in-service teacher training: the discussion of two courses (a 20-hour and a 40-hour one), presentation of syllabuses, time allocation for each point, discussion of how they went, reactions of participants
6. Conclusion: because teacher is an essential figure of an online classroom, and because universities and colleges do not provide teachers training in online teaching, it is necessary to conduct in-service teacher training courses in IT for English teachers, incorporate some elements of methodology of online teaching into university curricula, with training concentrating on so much on computer skills, but rather on noticing and exploiting the infinite possibilities given by the Internet and computers

In this paper, we present a computational implementation of a grammar of (a fragment of) Polish. This implementation is based on a formal grammar of Polish developed at the Institute of Computer Science, Polish Academy of Sciences, and formalized within the Head-driven Phrase Structure Grammar (Pollard and Sag, 1994; PrzepiĂłrkowski, 2000).

The implementation takes into consideration the following aspects of Polish:

• kinds of phrase structures of Polish including phrases introduced by complementizers and so-called "clause-union" ("verb clusters") infinitival environments;
• agreement (subject-verb, adjective-noun, etc.);
• syntactic case assignment (including case within numeral phrases and the genitive of negation);
• negation (including Negative Concord);
• binding (i.e., the resolution of pronominal anaphors);
• coordination (to a limited degree);
• and unbounded dependencies such as:
• wh-questions;
• relative constructions.

This implementation is at the moment unidirectional, i.e., it analyses ("parses") sentences of Polish giving as output HPSG structures corresponding to those sentences, but it does not generate sentences on the basis of corresponding HPSG structures. Outputs of the parser, i.e., HPSG structures, are more general than the usual parse trees returned by the garden-variety parsers of contex-free grammars (such as the first computational grammar of Polish, Szpakowicz 1978, 1986): apart from the information about constituency, they represent other kinds of morphosyntactic, syntactic and -- to a limited degree -- semantic information. The output of the parser for a simple sentence, Janek idzie, is presented at the end of this abstract; for lack of space, it cannot be discussed here, and it is given here merely to illustrate the richness of the output structures.

Such an HPSG parser has a number of potential applications. First of all, it can be used as a parsing module in a machine translation system; for example, one of the largest and most ambitious machine translation systems, Verbmobil, is based on HPSG grammars of German, English and Japanese (Wahlster, 2000). This is possible because HPSG parsers, unlike typical context-free grammar parsers, produce syntactic and, importantly, semantic analyses of sentences; these semantic representations can subsequently be used as the basis for the actual translation ("transfer"). Another possible application of such parsers is for the teaching of syntax, in this case, the syntax of Polish.

The parser discussed here has been implemented in the unification-based ALE system (Carpenter and Penn, 1995) based on SICStus Prolog, in the Unix environment. Depending on the interest of the organizers of the CAAL workshop, and on the technical conditions at the workshop venue (availability of a computer running SICStus Prolog and, more importantly, of an LCD projector) a prototype version of the parser can be presented during the workshop.

Carpenter, B. and Penn, G. (1995). The Attribute Logic Engine (Version 2.0.1). User's Guide. Carnegie Mellon University, Pittsburgh.

Pollard, C. and Sag, I. A. (1994). Head-driven Phrase Structure Grammar. Chicago University Press / CSLI Publications, Chicago, IL.

Przepiórkowski, A. (2000). Slavic linguistics formally -- HPSG. To appear in George Fowler, ed., The future of Slavic linguistics. Bloomington, IN: Slavica Publishers.

Szpakowicz, S. (1978). Automatyczna analiza składniowa zdań pisanych. Ph. D. dissertation, Uniwersytet Warszawski, Warszawa.

Szpakowicz, S. (1986). Formalny opis składniowy zdań polskich. Wydawnictwa Uniwersytetu Warszawskiego, Warszawa.

Wahlster, W., editor (2000). Verbmobil: Foundations of Speech-to-Speech Translation. Springer-Verlag, Berlin.

The paper discusses conceptual and technical issues related to a computational linguistic project (part of a larger European Union project) whose aim was to build a database of syntactically annotated sentences of written Polish.

Below, we describe the aims of the project, linguistic and implementational issues dealt with in the project, as well as potential applications of the final product.

Aims The aim of the project was to construct an extensive database of sentences of written Polish. This database ("test suite") is extensive in two senses: First, it covers syntactic phenomena of Polish near-exhaustively. Second, each sentence is exhaustively annotated.

These aims enforced certain constraints on the design of the database. First, in order to represent as many syntactic phenomena as possible, also those rarely occurring in corpora, it was decided that the test suite will contain elicited (rather than attested) sentences. Second, since each sentence in the test suite is annotated not only for its constituent structure, but also for various kinds of linguistic phenomena (utterance type, agreement, negation, etc.), a formalism chosen for annotation had to be specific enough to be computationally usable, and linguistically sophisticated to provide a basis for representation of all types of linguistic information. As a result, sentences are annotated with Attribute-Value Matrices (AVMs) as used in Head-driven Phrase Structure Grammar (HPSG), cf. Pollard and Sag (1994), one of the most popular linguistic frameworks used in computational linguistic applications. This choice also allows for homogeneity of annotations: AVMs can be used for encoding syntactic trees as well as other kinds of linguistic (morphological, syntactic and semantic) information.

Linguistic Issues The database described here was designed as a test suite; in fact the design was based on existing test suites for various European languages created within the TSNLP project (Lehmann et al., 1996, Oepen et al., 1998). The database contains not only grammatical, but also ungrammatical sentences, violating various linguistic rules.

The main linguistic problem that had to be faced when creating the test suite was ensuring that it really is syntactically exhaustive.

To this end, thanks to a close cooperation between linguists and computational linguists, a comprehensive hierarchy of linguistic phenomena of Polish was constructed. The main groups of phenomena in this hierarchy are: Types of utterances (declarative, imperative, interrogative); Tense, Aspect and Modality; Complementation (possible valence frames); Diathesis (possible changes in predicate-argument structure of a predicate, e.g., passivization); Modification; Agreement (within various kinds of phrases); Coordination; Negation (including negative concord and genitive of negation); Word Order (including positioning clitics word order within NPs); and, additionally, Punctuation. Linguistic phenomena within each group are organized into a hierarchy. Now, for each leaf, i.e., a maximally specific phenomenon, in this hierarchy, the test suite contains sentences illustrating this phenomenon, sentences violating it, and sentences showing its interaction with other phenomena.

The granularity of the phenomena represented can be illustrated by the following examples of rather idiosyncratic phenomena fully covered by the test suite: agreement and case assignment within and to numeral phrases, coordinated elliptical constructions, use of resumptive pronouns in co-marked relative clauses.

Implementation Issues The HPSG test suite for Polish is a database of Polish sentences (the HPSG test suite proper) implemented in Delphi (Borland) in the Microsoft Windows NT environment, see Bartkowski (2001). Correct sentences are augmented with their (one or more) HPSG representations (AVM structures) constructed according to the HPSG signature. An additional, auxiliary, database contains Polish wordforms (a dictionary).

The content of the database is restricted by two files: an HPSG signature and the hierarchy of linguistic phenomena represented in the database. The most important groups of operations on the test suite are entering, modifying, searching and viewing data. Annotated sentences can be added interactively or non-interactively from the text file whereas parses (AVMs) are added only interactively via a specialized graphical user interface. There are several simple search options but it is also possible to enter a query in the form of a regular expression. Introduced parses can be viewed both as AVM structures and as trees.

Applications The most common use of test suites is evaluating computational grammars (i.e., parsers) of a given language: if a parser is able to parse most or all of the correct sentences and none of the ungrammatical sentences in the test suite, then its coverage is very extensive. Thus, the empirical adequacy of parsers can be quantitatively evaluated by examining how they deal with respect to the data in the test suite. They can also be qualitatively evaluated by comparing the parses they produce to the exhaustive annotations contained in the test suite. So far, preliminary tests have been made on the prototype version of the AS parser, see Bień et al. (2001), which implements the largest computational grammar of Polish.

Similarly, reference grammars, textbooks and even particular syntactic analyses can be evaluated against the data in the test suite. Due to various viewing formats (AVMs or trees) and operations (folding or unfolding attributes, hiding selected attributes, expanding tags, etc.), the database can be used as a tool to teach HPSG.

A longer term application of this AVM-annotated test suite concerns automatic grammar induction using stochastic methods, similar to the applications described, e.g., in Osborne (2000), Johnson et al. (1999), Johnson and Riezler (2000) and Riezler et al. (2000).

Bartkowski, W. (2001). Komputerowa baza analiz gramatycznych w formalizmie HPSG. Master's thesis, Uniwersytet Warszawski, Wydział Matematyki, Informatyki i Mechaniki, Warszawa. In preparation.

Bień, J. S., Szafran, K., and Woliński, M. (2001). Experimental parsers of Polish. In Proceedings of FDSL 3, Leipzig. In preparation.

Johnson, M. and Riezler, S. (2000). Exploiting auxiliary distributions in stochastic unification­based grammars. In Proceedings of the 1st Meeting of the North American Chapter of the Association for Computational Linguistics (ANLP­NAACL 2000), Seattle, WA.

Johnson, M., Geman, S., Canon, S., Chi, Z., and Riezler, S. (1999). Estimators for stochastic "unification-based" grammars. In Proceedings of the 37th Annual Meeting of the Association for Computational Linguistics (ACL'99), College Park, MD.

Lehmann, S., Oepen, S., Regnier­Prost, S., Netter, K., Lux, V., Klein, J., Falkedal, K., Fouvry, F., Estival, D., Dauphin, E., Compagnion, H., Baur, J., Balkan, L., and Arnold, D. (1996). TSNLP -- test suites for natural language processing. In Proceedings of COLING 1996, Kopenhagen.

Oepen, S., Netter, K., and Klein, J. (1998). TSNLP -- test suites for natural language processing. In J. Nerbonne, editor, Linguistic Databases, CSLI Lecture Notes. CSLI Publications, Stanford.

Osborne, M. (2000). Estimation of stochastic attribute­value grammars using an informative sample. In Proceedings of COLING 2000, Saarbrueken, Germany.

Pollard, C. and Sag, I. A. (1994). Head­driven Phrase Structure Grammar. Chicago Univ. Press, Chicago.

Riezler, S., Prescher, D., Kuhn, J., and Johnson, M. (2000). Lexicalized stochastic modeling of constraint­based grammars using log­linear measures and EM training. In Proceedings of the 38th Annual Meeting of the Association for Computational Linguistics (ACL'00), Hong Kong.

Even though there are many misgivings about reusing existing lexical resources, i.e. dictionaries, for natural language processing (for a discussion cf. Wilks and Slator and Guthrie 1996), these dictionaries do continue do be so used. Some of them are re-reused quite often, indeed. An example of a monolingual dictionary can be Longman Dictionary of Contemporary English. As for bilingual dictionaries, at present Oxford dictionaries are quite often popular. There is a certain discrepancy, however, between the use of those dictionaries in highly advanced applications, which, however, are not usually commercially available for copyright reasons, and in strictly commercial ones, which, however, are not as interesting.

The commercial electronic publications have sometimes striking weaknesses. The Oxford bilingual dictionaries, available as Superlex. Oxford Bilingual Dictionaries on CD-ROM, have no morphological analyzer for any of the four languages present (English, German, Spanish, French). All they can do is to display a cross-reference entry.

This paper will discuss one of bilingual dictionaries which are used both in advanced research and in commercial applications. It is a medium-size English and Polish dictionary, written by Zygmunt Saloni and the author. It has been used in three computer applications so far. First, it was made available in one of the early bilingual dictionaries in Poland written for Macintosh computers (Piotrowski and Saloni 1995). The dictionary was quite simple. Next, it was also used in research carried out in a joint European project (STEEL) by Xerox (France), universities of Prague, Tuebingen, Warsaw and Tadeusz Piotrowski (cf. Piotrowski 1999), in a prototypical translating dictionary, which uses two-level approach to inflection in English and Polish. Finally, it was used in a prototype of a bilingual dictionary with a morphological analyzer of both English and Polish, which will be presented at the workshop. The dictionary was done by Morphologic (Hungary).

Its popularity, so to say, in those computer applications probably results from two facts: one is that the authors own exclusive copyright to the text, the other is that the dictionary was structurally tagged by the authors and therefore can be fairly easily converted into other formats, into a variety of fonts in case of a simple dictionary like Angela, or into SGML, as is the case with the Xerox or Morphologic. This experience, however, will allow me to draw some comparisons between the approaches in those projects.

Longman Dictionary of Contemporary English. 3d ed. Harlow 1995, Longman.

Piotrowski, Tadeusz, Zygmunt Saloni. 1995. Angela. for Macintosh computers. Warszawa: Programac

Piotrowski, Tadeusz, Zygmunt Saloni. 1999. Nowy słownik angielsko-polski i polsko-angielski. 3d edition. Warszawa: Wydawnictwo Wilga

Piotrowski, Tadeusz. 1994. Z zagadnień leksykografii. Warszawa: Wydawnictwo Naukowe PWN

Piotrowski, Tadeusz. 1999. "Tagging and conversion of a bilingual dictionary for XeLDA, a Xerox computer-assisted translation system". In: Papers in Computational Lexicography COMPLEX'99 Proceedings. Budapest: Hungarian Academy of Sciences 1999, ss. 113-120

Superlex. Oxford Bilingual Dictionaries on CD-ROM. Oxford 1996, Oxford University Press

Wilks, Yorick, B. M. Slator, Louise M. Guthrie. 1996. Electric Words. Cambridge: MIT Press

This presentation summarizes a five-year practice in using multimedia computer facilities as a perfect technical means in the English language training at different levels of language acquisition.

Being strongly communicatively oriented I could not be satisfied with the opportunities provided by typical educational computer programmes basically relying on the student's individual practice, which is mainly limited to performing a certain set of typical exercises . At this stage the computer becomes a fast and nicely visualized book of exercises.

Later, an attempt was made to activate the student's speech performance. Programmes with voice recording and sampling came into existence. However nice it may appear, it is not communication.

So far I am strongly convinced that a computer can not replace a humanised lesson in shaping student's communicative abilities correctly. Only through a dialogue supervized by the teacher students can become aware of the language appropriecy and master language functions in natural conversation. Trying to come to a compromize, I discovered that software which can be defined as "guided cinema games" is a well adaptable material for the purpose.

As the basic product I have been using the Humongous Entertainment series which presents linguistically levelled and topically variable programmes. The advantage of such programme is the possibility to make a pause or to swap pages, to replay a scene or to set up a task to be solved. Every game (or page of game) acquires its clearly defined Study stage which usually covers particular grammar points. The teacher gets a booklet with a set of typical classroom activities that should provoke students to use the desired structures on the basis of the scene on the screen.

The classroom arrangement in such case is limited to 10-12 students sitting in a horseshoe with only one, preferably large, monitor. This seems to be a solid advantage, since not many schools can afford a full-sized computer classroom.

The scheme of the interactive presentation is the following:

• A brief review, classification and evaluation of the English Language multimedia programmes available on the Polish market. Those programmes have been tried by the language school Miteinander in Przemysl where the author is running his courses. Basic advantages and drawbacks of the programmes under consideration will be presented.
• Guided cinema games - classification, structure and applications in particular language teaching situations. Typical examples of programmes will be demonstrated.
• Varieties of exercises applied by the author.
• (optionally) A workshop on task-based designing of exercises for classroom use.

Power Point slide show here.

Introduction. The teacher, with his/her knowledge, skills, experience, attitudes and emotions, is -- despite the often aired fears of the profession -- an indispensable element of the Computer-Assisted Language Learning (CALL) classroom. But in order to function properly in this challenging milieu s/he must accept new roles and obligations, as well as adjust some of his/her attitudes concerning the technological enhancements of the didactic process. The teachers' knowledge and experience of, as well as attitudes toward, electronic learners' dictionaries are among the most vital elements of the new situation in which teachers and learners find themselves more and more often in Polish schools. I decided to probe some of them, with particular attention paid to the changing attitudes relative to the teachers' experience of machine-readable dictionaries (MRDs) of English as a foreign language.

Subjects and data. Twenty-five in-service teacher-trainees doing their one-semester post-graduate study of TEFL in the School of English, Adam Mickiewicz University, were asked about their experience and attitudes toward electronic EFL dictionaries. Twelve yes-no-? questions were asked, eight of them duplicated from a much more comprehensive questionnaire circulated among over six hundred Poznań students of EFL in May 1998 (see Sobkowiak 1999). The questions ranged from rather technical issues of phonetic transcription to ones probing general attitudes to (MRDs), as shown below. The four 'new' questions, extending the coverage of the questionnaire to include the teacher perspective, are asterisked. The ordering of all twelve questions is reverse-alphabetic.

1. Words can be searched by their sound in a computer dictionary
2. *Using computers with pupils in class must disrupt the flow of the lesson
3. *There is access to computer(s) where I teach
4. Speed is the main advantage of a computer dictionary
5. Pictures are the main advantage of a multimedia computer dictionary
6. One can search the whole text (including definitions and example sentences) in a computer dictionary
7. It's easier to use a computer dictionary than a traditional one
8. In a multimedia computer dictionary, phonetic transcription is useless
9. In a computer dictionary words having similar meaning can be listed
10. *I have used a computer dictionary of English recently in class
11. *I have at least one computer dictionary on my PC at home
12. Computer dictionaries cannot show phonetic transcription on screen

Results. The results have no pretense -- from a small sample like this one -- to reaching the full rigour of an empirical questionnaire study, but they are certaily suggestive to the point of proving it worth while to conduct a more thorough investigation. The most significant observations derived from answers to the preceding questionnaire appear to be the following:

• Not a single teacher "has used a computer dictionary of English recently in class", even though most (15/25) have one at home, as many as 19 have "access to computer(s) where [they] teach" and only three teachers agreed that "using computers with pupils in class must disrupt the flow of the lesson".
• Correct answers to the more technical questions having to do with phonetic transcription, like 1 or 12, appear in less than half of the returns: thirteen respondents ticked <?> in these questions. The less technical questions, like #6 and 9, were (usurprisingly) easier: 15 and 22 <yes> answers, respectively.
• About half of the teachers value speed (question 4: 12 <yes>'s) and ease of use (question 7: 12 <yes>'s), but only five believe that "pictures are the main advantage of a multimedia computer dictionary".

Some interesting tendencies can be observed when data are grouped according to the professed respondents' experience of MRDs: I divided them, just like in my 1998 study, into 15 'experts', i.e. those who declared having "at least one computer dictionary" at home, and 10 'ignorants', who do not. Experts are more knowledgeable about some of the technical aspects of MRDs, such as their ability to show phonetic transcription on screen (47% <yes>, as opposed to 20% ignorants), but -- surprisingly -- not about some others: 53.3% experts, as opposed to 70% ignorants believe (correctly) that "one can search the whole text (including definitions and example sentences) in a computer dictionary" (experts may have older MRDs, and ignorants may have inflated expectations). There are also differences in attitude: one in three experts believes that "pictures are the main advantage of a multimedia computer dictionary", and not a single ignorant concurs.

Conclusion. If it is indeed true that "student concerns about being able to cope with the demands of technology appear to cause some to choose less advanced language learning tasks", as Debski and Gruba (1998:54) discovered, it is the obligation of the teachers to change these attitudes. If teachers themselves continue to have bad stereotypes and fears of educational technology, as appears from this pilot study, no amount of computing machinery lavished upon schools will help. Teacher trainers should think about it very seriously: the main challenge of electronic learners' dictionaries is in the head of the teacher.

Cameron,K. (ed.). 1998. Multimedia CALL: theory and practice. Exeter: Elm Bank Publications.

Debski,R. & P.Gruba. 1998. "Attitudes towards language learning through social and creative computing". In K.Cameron (ed.).1998.51-56.

Sobkowiak,W. 1999. Pronunciation in EFL machine-readable dictionaries. Poznań: Motivex.

Creating WWW pages is no longer an exclusive domain of professional programmers. A web-page can be created relatively easily with the use of high-level (or object-) programming tools, or even word processors. This is a great challenge for both educational institutions and individual teachers and educators. There are many examples of ToEFL web-pages. Most of them are created by educational institutions, but many - the so-called "private home pages" - are designed by teachers or learners of English. Very many teaching and learning resources can be found there.

The paper discusses web-page designing as a tool in the hands of individual teachers. The following issues are analyzed:

• the potential of educational web pages,
• teaching and learning resources that can be found on existing web pages,
• the promises and challenges of private home-page creation,
• the methods and tools for web-page design.

In other words, from the teacher's perspective, the three crucial questions are answered:

• What? -- What is (and what can be) the content of educational web pages?
• Why? -- Why is it advantageous to create private web-pages?
• How? -- How to design a page, possessing very little or no programming knowledge?

The first two generations of CAI programs appeared not quite satisfactory from the pedagogical point of view. Tasks that they contained demanded mostly operational and analytical thinking and, as a result, did not contribute to the growth of the students' intellectual creativity and general humanistic development. A remedy could be the application of Artificial Intelligence (AI) technology, as it could supply computers with the ability to converse with the user, administer intellectually demanding tasks, and even adapt to the learner's personal (cognitive and affective) features [2]. This last ability could be achieved only with an advanced student model, the component of a program collecting information related to each student's learning processes.

From the analysis of pedagogy-related literature devoted to CALL it appears that there are no established pedagogical standards for student modelling [3, 4]. Such standards would help software designers to apply modelling techniques to CALL in full accordance with modern trends in education, associated with humanistic pedagogy [7]. The lack of pedagogical standards has resulted in didactically unsatisfactory projects of CALL systems and their student models. Most frequently, they are products of the designers' arbitrary decisions based on intuition and previous experience with CALL. It seems necessary to work out an alternative methodology which would employ pedagogical theory and the results of linguistic research. One possible suggestion for such a theory-and-research-based methodology of student modelling for CALL is as follows:

1. work out a framework for construing pedagogical standards for CAI systems in general,
1. choose a particular pedagogical theory to underlie the system,
2. based on the general framework (step 1), formulate a set of CAI standards rooted in the chosen theory (step 2),
3. from the set of CAI standards (step 3) derive a set of standards for a student model,
4. relying on CAI standards (step 3) and on a self-chosen approach to foreign language teaching, outline a CALL system as an environment for a student model,
5. with the pedagogical standards for a student model (step 4) and the outline of a CALL system (step 5) as the foundation, work out the design of a student model for this CALL system.

Ad. 1 - The framework consists of eight questions, which specify two main problems universal in education: what is the purpose of learning and upbringing, and by what means can this purpose be achieved?

Ad. 2 - As for the choice of a general pedagogical theory, one possible choice is Wincenty Okoń's Theory of Versatile Education (TVE), as it is consistent with the "humanistic paradigm" [5, 6, 8]. The main adjustment made to the general framework with regard to TVE is the introduction of four types of didactic resources, methods, and activities: receptive, explorative, emotional and practical. Besides, it is necessary to consider which traditional didactic principles are consistent with TVE (presumably: the principle of systematisation, of effectiveness, of accessibility, of individualisation and socialisation), and what didactic goals are achievable with this theory.

Ad. 3 - Considering principles of TVE and traditional didactic principles compatible with TVE, the final set of pedagogical parameters for a CAI program will include the following items:

1. The program should make use of four types of resources recommended by TVE.*
2. These resources should enable learners to recept, explore, experience and apply knowledge with the help of four types of methods.*
3. These methods should be favourable to four routes of learning.*
4. The resources, methods and routes of learning should be consistent with:

Ad. 4 - From the parameters prepared for a CAI system in step 3, it is possible to draw the following set of standards related to a student model: (Near each parameter the source has been mentioned, i.e. the number of a corresponding parameter form the set given above.)

• A student model should record the learner's routes of learning based on the four categories of resources and methods.* (from parameter 3)
• A student model should model facts and processes that are important from the point of view of the methodology of the teaching of a given school subject. (from parameter 4a.)
• A student model should record the learning process in all its stages (from parameter 4b. - "the principle of systematisation")
• A student model should collect personal factors influencing the learning effects. (from parameter 4b. - "the principle of effectiveness")
• A student model should collect personal factors influencing the reception of the program's content. (from parameter 4b. - "the principle of accessibility")
• A student model should monitor the processes of the learner's individualised and socialised development. (from parameter 4b. - "the principle of individualisation and socialisation")
• A student model should monitor the learner's approaching the didactic purposes of the program. (from parameter 4c.)

Ad. 5. - It appears that the foreign language methodology best suited to TVE is Communicative Language Teaching (CLT) [1, 9]. The two approaches are derived from the humanistic tradition and, as such, promote a holistic, versatile development of the learner. The result of applying TVE and CLT to intelligent CALL is a Communicative TVE-based ICALL System (CoTIS).

Ad. 6 - With the pedagogical parameters (step 4) and the outline of CoTIS, it is possible to put forward a theoretical design of a student model in this system. As was argued above, the choice of information to be gathered and utilised by the student model is not dictated by some accidental factors, but results from theoretical assumptions made in advance. This feature is deemed to be the main advantage of this design.

The student model collects data about the student's linguistic and educational background and learning needs. It also monitors: developing motivation, learning facts, making generalisations, formulating rules, consolidating knowledge, and applying knowledge to practice. It analyses communicative competence improvement, monitors personal development, searches for sources of persistent learning problems, and scrutinises the development of the learner's autonomy. Suggested modelling techniques include cognitive task-tracing and collaborative student-questioning.

This is a theoretical student model, unlikely to be applied in near future. The goal was to point towards possibilities of enhancing student modelling in future, so that student models in CALL systems could perform their important pedagogical function successfully.

______________________

1. Brown, Douglas, H., 1994. Principles of Language Learning and Teaching. San Francisco: Prentice Hall Regents
1. Chwiałkowska, Ewa, 1991. Sztuczna inteligencja w systemach eksperckich. Warszawa: Mikom
2. Ephratt, Michal, 1992. "Developing and Evaluating Language Courseware." Computers and the Humanities 26: 249-259
3. Manning, Patricia, 1990. "Methodological Considerations in the Design of CALL Programs." CITE Report No. 131. Milton Keynes: Open University
4. Okoń, Wincenty, 1967. Podstawy wykształcenia ogólnego. Warszawa: WSiP
5. Okoń, Wincenty, 1995. Wprowadzenie do dydaktyki ogólnej. Warszawa: Wydawnictwo Żak
6. Rogers, Carl, 1983. Freedom to Learn for the 80's. Ohio: Bell and Howell
7. Tanaś, Maciej, 1997. Edukacyje zastosowania komputerów. Warszawa: Wydawnictwo Żak
8. Tarone, Elaine, and George Yule, 1991. Focus on the Language Learner. Oxford: Oxford University Press

The CSLU Toolkit is a set of tools that let us investigate and create spoken language systems. The package was initially created as a platform for teaching language to profoundly deaf children. It includes speech recording, playback, synthesis and recognition, alignment of facial animation and speech as well as gives an opportunity to incorporate external multimedia content. The powerful features of the Toolkit can be controlled and programmed within a robust, intuitive, graphical Rapid Application Developer. Multimedia and interactive dialogues are now just a mouse-click away. The whole set of applications available in the package let an imaginative and open-minded teacher create a wealth of applications that can be used in EFL teaching.

There are numerous CALL packages on the market employing high quality multimedia content and state-of-the-art technologies, e.g., speech recognition. However, CSLU Toolkit has some features that make it different and unique as a tool for EFL teaching. First of all, it is its enormous flexibility; only the teacher's imagination sets limits for exercises, lessons and activities that can be developed with CSLU RAD. Instead of some fixed, predetermined exercises and activities that can be found in 'standard' CALL packages, teachers can use the Toolkit to build their own, highly customized exercises, with speech recognition and multimedia content that better suit their and their students' needs.

Another of the Toolkit's components -- "Baldi" -- a conversational agent with facial animation that can be aligned with speech sounds, can aid teaching pronunciation. A multitude of configuration options make Baldi a very interesting as well as instructive interlocutor. Changing Baldi's emotions, voice characteristics, race and look may make him more human-like and attention-grasping. Advance features, such as making his face transparent, showing a cross section of his articulators moving, can enhance teaching phonetics.

Given new, powerful tools, it is still the human teacher and student who decide on how they can and should use them.

Last update: 4.6.2001 by Włodzimierz Sobkowiak