Toegepaste Taalwetenschap in Artikelen - Volume 33, Issue 1, 1989

At the Humanities Faculty of the Free University of Amsterdam intensive efforts are being made to integrate computer-assisted learning in the curriculum of modern language teaching. The courseware is written in VULCO, an authoring language developed in-house and specifically designed for non-scientific applications.The paper describes a design concept for vocabulary acquisition courseware. The concept is characterized as 'modular', 'integrated' and 'learner-controlled'. The course-ware consists of a number of independent modules such as texts, a lexical knowledge base and various types of exercises. The design is 'integrated' in the sense that vocabulary learning proceeds from an authentic text in which new items are marked; at any point in the learning process this text together with the associated lexical database can be accessed. Exercises are related to this text. The didactic rationale for this approach is to provide constant confrontation with original reading material. The courseware is 'learner-controlled' in the sense that the learner is not compelled to adhere to a prescribed learning path - although this may be imposed by the teacher if so desired.The paper also briefly reviews an experimental study with the system for students of Dutch as foreign language.

Educational software for language learning is an emerging field with scientific experience since the early sixties. Its full potential will have to be reached dynamically by continuing research and by a better production. It is difficult to say how the most efficient language software will look ten or twenty years from now. A number of general criteria can already be indicated.Richness of content: high programming techniques and the optimal use of condensing procedures for stocking make it possible to include maximal contents within one software package. One disk should contain hundreds if not thousands of working hours for the user. That means that the production time shifts more to the making of content than to the programming.External versatility: by placing the content into databases it becomes possible to build a situation-independent program which allows to generate, within one package, a high number of specific lessons. This way the independence of the user is guaranteed.Didactic ergonomy concerns itself with many aspects. As to content, educational software should match any initial situation, be ready for use, contain a build-up planning for years, allow speedy overviewing, respect freedom of use, and limit error analysis. The programming ergonomy deals with screen presentation, the use of colours, the reaction speed and the user options given.

In second language teaching a great deal of attention is paid to the learning of communicative skills. The computer program PRESTIGE (Productive and Receptive English, a Smart Tutor and Intelligent Generator of Exercises) subscribes to the views currently held in communicative language teaching. Among these are learning the meaning of a word through deduction of the word's meaning from contexts, and emphasizing the use of productive language. The computer program PRESTIGE, written in PROLOG, consists of several parts:A parser, which checks if syntax, spelling and inflection of pupils' input sentences, words, and verbs are correct. The parser is the heart of the program, and is capable of analysing most English sentence structures which are used in the first years of secondary education.A database of descriptions of words and context sentences in everyday English (taken from Collins COBUBLD English Language Dictionary). Both databases can be consulted by the pupil at any moment and thus constitute an extensive help facility.A database of texts. These texts are taken from a method called "Notting Hill Gate" (Malmberg).A generator of exercises. The program is able to generate exercises for both receptive and productive language use.Receptive: a story fragment is selected from the database. For every word, pupils can ask for a description (in English) or for a context sentence in which the word is used.Productive: several different exercises can be generated. They vary from cloze texts: the parser leaves out words that belong to a certain grammatical class (adjectives, prepostions, etc.), to free language productions: writing a composition. Furthermore, exercises can be generated in which interrogative or negative sentences have to be made. When the pupil is writing a composition, the parser checks whether the sentences are grammatically correct. Here as well, all help options from the databases are available.Pupil registration. For every pupil the program keeps track of the words that (s)he has studied, what kind of help (s)he has asked for, and what the results of the exercises are.The advantage of PRESTIGE over existing courseware for second language teaching is the open structure of the program, which makes free input of language possible. In the article the possibilities of the parser are shown and illustrated by examples. The article concludes with a discussion of the uses of PRESTIGE as an intelligent generator of exercises, and some plans for further research.

In the present article the author explains what concordancers are and what their contribution to language learning can be.A concordancer is a computer program that allows one to look for one or more keywords, prefixes or suffixes in contexts. It gives a listing of each occurrence of every word in a text where words are listed in a given order. The length of the contexts may be one or more lines. A concordancer is distinct from a dictionary in that it doesn't give information about word class, etymology, pronunciation and an example of every meaning of the word. A concordancer displays a word in the context that particular word appears in. So the contexts depend on the corpus the concordancer searches in.Concordancers can be used by teachers, students and writers of text books. Teachers can illustrate their grammatical explanation with contexts (examples). They can generateexercises for their students on the basis of the contexts e.g. for guessing the meaning of particular words or for deducing grammatical rules.Statistical information about the frequency of words can help textbook developers to see how often and where (new) words appear in the texts.Concordancers also have some limitations. The value of the concordance 'tool' is determined by the corpora it is based upon. The variety of texts in a corpus e.g. news-papers, transcriptions of the use of oral language and literature determine the variety of contexts. The requirements of a corpus depend on the educational objectives and the target group. The author describes a few possibilities to obtain corpora. Finally she warns against the relative authority of texts.

An adventure is a new type of computer game which has become immensely popular in the course of the 1980s. This article is about the possible role of adventures in foreign language learning and teaching (FLL).First there is a brief explanation of what adventures are, the different types and the way communication within the game can take place in natural language. Examples are given for French, German and English.Adventures can play a role in FLL in several respects: -they stimulate discovery learning procedures -they encourage the use of certain reading strategies -they are suitable contexts for vocabulary learning -they can present contexts for communication.Moreover adventures can be looked upon as a new type of literary text, which learners can read as an alternative to a book (some adventures are known as interactive fiction).The article then presents a set of criteria for FLL: For advanced levels text adventures are more suitable than graphic ones from the point of view of language learning, because they present a rich language environment. Graphic ones may be more suitable for beginners. Adventures should accept a variety of syntactic patterns and provide adequate semantic analyses, so that the student gets appropriate feedback. A certain tolerance to spelling is needed, or easy correction options should be available. The program must show the student the type of language it accepts. Hint-files to help students when they get stuck are important and possibly an on-line glossary might be useful. The vocabulary used must not be too exotic and the plot not too complex.It is finally demonstrated that the Infocom adventure SHERLOCK meets these requirements to a large extent.

What aids do teachers who consider making their own courseware have at their disposal today?In this article some of these are discussed, and special attention is paid to such notions as authoring programmes and authoring systems.At the same time some criteria are presented which enable the teacher-author to decide which programming tool he or she considers the most suitable in his or her position.Next, attention is drawn to the importance of the didactic plan i.e. writing out the didactic script, before 'programming' (the production stage) is started. Especially with a view to the pre-production stage a so-called grade project is introduced, which offers the possibility of developing courseware systematically, the didactic plan being the central point.In the second part of this article Mrs. W. van Straalen (University of Leyden) describes how she together with her team (of students) has come to a (well-)considered choice of an authoring system (Calis) and has been able to write student-friendly programmes.To give a concrete example: she analyses the programming process of training the use of prepositions after the adjectives listed in L'essentiel de la grammaire française, appendix Vili.

One of the most important questions that has to be answered before we can start developing courseware for computer assisted language learning concerns the role feedback could or should play in this type of courseware.This question is the more important because in the last few years two tendencies seem to be developing in CALL:-one that stimulates and elaborates 'traditional' CALL-programmes, adapted as much as possible to the characteristics of the individual learners, taking into account the level of knowledge of the learner and providing feedback that is specifically adapted to the learners' responses; -other CALL-developers tend to propagate 'open-ended' programmes like adventures, simulations, that they consider more interesting and more motivating because they are closer to real-life situations.Making the first type of programme requires a huge amount of work: foreseeing and analyzing learners' responses and developing adequate feedback. The second type of programme is available and ready for use, no programming is necessary - but individual control and correction are not possible. Moreover, it is not very clear what learners do learn when working with these programmes.An essential difference between the two types of programmes is in the feedback they provide. If we want to be able to make a choice between these two approaches to CALL, we should know more about the role of feedback. That is what the project at the Univer-sity of Amsterdam described here wants to achieve.

At the Nijmegen Institute for Cognition Research and Information Technology (NICI), linguistic modules have been developed to make computer simulations of human language processing possible. Those modules are also used in several practical applications in the field of computer-assisted instruction. In order to improve the results of te teaching of grammar and spelling in Dutch and Flemish schools, we developed two programs: BOUWSTENEN, an intelligent program for Dutch grammar teaching, and SPELRAAM, an intelligent program for the instruction of the spelling of Dutch verbsThe innovative aspects of BOUWSTENEN are:- the explicit grammar which makes a clear distinction between grammatical functions and categories at different levels- the constructive method: the program offers constructive as well as analytical exercices- the two-dimensional display of the syntactic structure of sentences- mmediate feedback adapted to the knowledge level of the student and explaining what is wrong as well as why it is wrong.An elaborate description of BOUWSTENEN is given, in which the architecture of the program and the elements (examples, exercises, explanation) that are part of every lesson are discussed. Two lessons are described in detail: the finite verb lesson and the direct object lesson.In the future, BOUWSTENEN could be elaborated into a complete Intelligent Tutoring System by developing a curriculum network as a base for a didactic component and a student model.Knowledge of all the rules for the spelling of regular Dutch verbs makes SPELRAAM also an intelligent program. By practising with SPELRAAM, a student learns to spell Dutch verb forms correctly by applying a small set of rules. The program is able to diagnose errors made by the student and to help the student corect them.

At the Institute of Applied Social Sciences in Nijmegen a computer based language training programme is being developed and experimentally evaluated in secondary schools. The programme called Edittraining, is based on the principle of the editing-test. This editing-test is a test of general language competence and consists of a reading text in which randomly chosen words have been randomly inserted; the number of the 'intruders1 is about 12 per cent of the original number of words. The main task is to delete these intruders. To execute the task not only lexical but also grammatical competence is needed. The four main components of the programme are: a set of 12 reading texts processed as editing tasks, grammatical feedback supplied with various text elements, a students' grammar, and a lexicon.In experiments executed in secondary schools empirical data have been obtained on the effectiveness of the programme. The main hypothesis is that learning by Edittraining leads to improvement of language proficiency. The specific hypotheses pertain to the improvement of general language competence, reading competence, grammatical compe-tence and knowledge of grammatical concepts.The design chosen was a pre- and post-test design, with experimental and control groups. The pre-tests used were a cloze-test and a grammatical knowledge test; the post-tests were the same cloze-test, a parallel version of the grammatical knowledge test, a multiple choice test for reading comprehension, and a grammatical skill test. The results of a covariance-analysis showed a positive significant effect for the experimental group on the cloze-test, on a subtest of the grammatical knowlegde test called 'concepts' and on a subtest of the grammatical skill test called 'composition of noun and preposition groups'. Most of the grammar subtests turned out to be rather easy for these pupils. Analyses for subgroups, with less than or equal to 60 or 80% correct on the pre-test, did not greatly alter the results except for the cloze-test. For the group with less than or equal to 50% correct on the pre-test of the cloze-test, the effect of the Edittraining was a little stronger than for the group as a whole. We can conclude that learning by Edittraining leads to a significant improvement of general language proficiency and of some grammar subskills but not of reading comprehension and the other grammar subskills that were tested.Other specific hypotheses concerning special versions of the programme such as working in groups or individually, positive versus neutral feedback, and feedback following each separate part or the whole of the text, showed no clear results.

Text-to-speech systems generally consist of two components. The first one converts the input text to an abstract, linguistically relevant, representation. Usually, this is a phoneme representation of the input text, with markers for (word, morpheme, syllable) boundaries, word stress, and sentence accent. The second component converts this transcription into a physical speech sound. Two aspects of natural speech are most important to be imitated in this latter step: (a) natural prosody (speech rate, segment duration, pitch, etc.), and (b) representation of phonetic adjustement between phonemes. The resulting synthetic speech is mainly used in special-purpose applications, although a wider use is foreseen for the future.

In Model Schools Project West-Netherlands the Free University of Amsterdam (Department of Cognitive Psychology) and the State University of Utrecht (Researchgroup on Mathematics Education and Educational Computer Centre) study the way the computer can be used as an aid in secondary education. In the model school (Cals College Nieuwegein) five departments (Dutch Language, Mathematics, Foreign Languages, Geography, Home Economics) are developing and trying series of experimental lessons in which the computer is used as an aid to students and teachers. Existing (educational) software is elaborated with worksheets and teacher guidelines, aimed at an optimal integration of the software into the curriculum. During the schoolyear 1987/1988 the English Language teacher, supported by researcher and subject matter experts, has given a series of lessons on "reading and prediction", viz. the use of function words in a text. In four lessons the students worked on (a) choosing an appropriate consecutive phrase given a main phrase with function word (supported by the program "Sequitur"), (b) identifying the meaning of function words in a text (without computer), (c) learning the meaning of the major function words (supported by a Dutch program "Word Meanings"), and (d) completing sentence with a function word (supported by the Dutch program "Doka"). The experimental lessons indicated the contributions to be expected of the programs used and also the shortcomings to be remedied in future releases.