Toegepaste Taalwetenschap in Artikelen - Volume 11, Issue 1, 1981

In computational linguistics three classes of models have been developed for the automated treatment of texts in natural language. One class is best characterized as a set model. Language is defined as a set of words. On these words normal arithmetic computations are performed. The set model has led to frequency counts. Frequency counts of natural language material have proved to be of little importance to language analysis and the study of language learning.The second class of models is best characterized as formal linguistic models. Here language is defined as not merely a set of words but more as a set of sentences. On these sets of sentences more than purely arithmetic operations can be performed. Important notions in these models are transformations, recursivity or even grammars. This class of models has led to the adaptation of context free grammars to natural language. The weak point in this class of models is the inappropriateness of formal grammars to human language.The third class of models can be defined as artificial intelligence models. Here the computer is used to simulate human verbal behavior. Language processes are defined as processes of understanding language. Linguistic knowledge is not defined outside the vocabulary or outside these processes. This class of models has led to the application of Minsky's frame theory to natural language processing. The lexicon here is defined as a procedural fact device in the language processor itself. This class of models is most promising for the study of language learning and the role of vocabulary in this language learning process.

Students of economics at the University of Antwerp, UFSIA (Belgium) have been asked about the usefulness of grammatical information in the Longman Dictionary of Contemporary English (DCE) and the Oxford Advanced Learner's Dictionary of Current English (ALD).The dictionaries have been used in English lessons as an alternative and active way of brushing up grammatical knowledge, in order to develop the habit of consulting these reference works.The grammatical coding system in DCE is more comprehensive than in ALD. With a few exceptions most of the symbols are learnable.DCE presupposes some grammatical knowledge on the part of the user of the dictionary.Some inconsistencies with A Grammar of Contemporary English by R. Quirk, the linguistic adviser of DCE, are pointed out: the treatment of- transitivity- the copula- infinitive constructions as object of a verbIn the future the coding of grammatical information will have to be improved. With a number of translating dictionaries from foreign languages into Dutch to appear, containing grammatical data similar to ALD and DCE, the need for greater uniformity of presentation and especially of the principles of grammatical coding is urgent.

Especially from the viewpoint of one who translates texts from one language into another, it is obvious that interlinguistic one-to-one correspondences of lexical words are rather hard to find. The first and foremost circumstance that accounts for this phenomenon is that speakers of different languages conceptualize reality in different ways. Even if one subscribes to the hypothesis that there is a universal stock of semantic features by means of which designata can be analysed, it remains true that languages differ considerably with regard to which features their speakers consider relevant to situations, and which not. Thus seemingly corresponding words may have different designata, to the degree that translation requires a clear contextual setting according to which equivalence relationships can be established.Polysemy, hyponymy and synonymy have largely differing distributions even in affiliated languages. A polysemous SL lexeme may require the coice of a TL lexeme which is not its immediate correspondent according to the bilingual dictionary. Conceptual synonyms (or, rather, quasi-synonyms) should be dealt with suspiciously, since differences in connotation, mode or register may necessitate the use of one term to the exclusion of the other. Even semantically equivalent terms may show different grammatical behaviours in SL and TL in that, for example, the TL imposes selection restrictions on a given term which do not apply to its correspondent term in the SL. Further complications may derive from collocational differences between languages. Finally, quite a lot of interlinguistic lexical anisomorphy, is due to differences in culture and social conventions - even to the extent that semantically and pragmatically corresponding expressions take on astonishingly deviant forms.In order that they may account for these differences translation didactics as well as foreign language teaching can rely on models which have been elaborated by certain present-day disciplines whose concern is the study of linguistic phenomena. Thus componential analysis still provides a sound basis for comparative lexical analysis, while the theory of speech acts and Permyako's 'paremiological model' explain how semantic and pragmatic invariance goes hand in hand with socio-cultural variability.

Les auteurs rendent compte des multiples difficultés rencontrées en composant deux vocabulaires de français destinés aux élèves hollandais. Elles en énumèrent 14, de nature très variée, les illustrant chaque fois d'exemples concrets.I) le temps consacré à l'étude est primordial et dépend des autorités qui en fixent le nombre d'heures. 2) la politique joue un rôle: un enseignement intégré conduira à une harmonisation des programmes. 3) le dynamisme du temps: des changements s'opèrent dans la langue elle-même, dans les méthodes d'enseigne-ment, dans le climat pédagogique. 4) l'élève demande de l'efficacité dans l'ap-prentissage et ses intérêts ne correspondent pas forcément à ce qu'on lui pré-sente. 5) le professeur, débordé, n'a pas le temps de se tenir au courant de l'actualité linguistique et méthodologique. 6) le type d'établissement pourrait influencer la sélection et la présentation des mots. 7) 1'examen pose un grand nombre de problèmes. 8) le programme n'est que très vaguement déterminé. 9) Les rapports entre compréhension et expression ne sont pas définis, pas plus que leur place aux différents niveaux de l'enseignement. 10) Les niveaux de langue: faut-il enseigner le français écrit ou le français parlé?II) la sélection des mots peut être influencée par leur fréquence, la situation, le hollandais ou des considérations didactiques.12) la méthodologie en vigueur.13) la présentation des mots peut se faire de 8 manières différentes.14) l'apprentissage des mots peut varier d'un mot à l'autre et d'un élève à 1'autre.L'acquisition d'un vocabulaire est donc fonction d'une série de variables souvent encore mal déterminées. Il est grand temps de s'occuper sérieusement d'un programme cohérent pour l'enseignement des langues modernes dans le secondaire et jusqu'à présent personne n'a envisagé de le faire.

On of the major concerns in the tradition of foreign language teaching has been the question of how FL learners learn new vocabulary. This issue comprises two questions, i.e. how are FL words grasped in their complex meanings and how are they remembered? In this paper this dual issue is tackled by proposing the notion of a didactic lexicon, which is seen as an optimal presentation of word meanings.This contribution discusses Fillmore's 'Scenes-and-frame semantics' and the contextual approach to vocabulary learning and describes an experiment in contextual vocabulary learning and testing.A didactic lexicon is defined as a variabele set of lexical items explained in ordinary language, possibly with the help of visual materials or trans-lations, enabling the learner to build up the strongest possible associations between FL words and their prototypical scenes, and revealing the underlying structure of the lexicon of the FL.Finally some strategies for developing a 'didactic' lexicon are presented.

As there is a general complaint about the poor lexical competence of students in non-English speaking universities, and as there is certainly no hope of remedying this by means of an increase in the academic staff, we have decided to measure our students' English vocabulary mastery (recalling of words) at the very start of their anglistic studies, so as to provide them with adjustable mastery-learning strategies, performable at computer terminals.In this paper we present the test's components and its measuring capacity. We use the 2000 most frequent English words, obtained by means of a merger of three objective frequency lists of about 1.000.000 words (the Brown Corpus, the Leuven Drama Corpus and the Lancaster-Oslo-Bergen Corpus), presented in a descending rank list; to each word the inflected forms and derivatives have been added. We also have combined this objective frequency with two subjective selections; the defining vocabulary of 2000 LDOCE-words (Longman Dictionary of Contemporary English, 1978) and 1500 words from J.C. Richards' Familiarity Count (1971).The words of these two lists that do not overlap with the frequency-words are considered as components of the test-corpus.The students who score below the means obtained for this test (in the Flemish part of Belgium + 111/150 in October 1980) get the opportunity to increase their vocabulary mastery by means of having to study a number of texts; they have to learn the new words (outsiders) in these texts and practise again the words they supposedly know.The analysis of the texts, the ordering according to the degrees of difficulty, the definition of outsiders, the construction of exercises on the outsiders and on the supposedly-known words between positions 501 - 2000 in the descending ranklist, the exercises on cohesion, cloze, polysemy - all these operations have been subjected to automation by computer, and the software developed thus far is operational.In the present paper the results obtained during the session of Mastery-Learning from October 1980 to March 1981 is presented, explained and discussed.

Hunter Diack's Standard Literacy Tests (SLT's) were constructed as English vocabulary tests intended to cover all the vocabulary of the English language. Originally, these tests were devised to assess the vocabulary command of native speakers. They were labeled tests of 'literacy' on the assumption that the extent of a person's vocabulary command in a language reveals the width and quality of his reading.In each SLT, English words are presented out of context. Subjects are requested to give one meaning for each word encountered, in any manner they choose, e.g. by giving a paraphrase, by using the word in an appropriate context, etc.If, as the author assumes, the SLT's are valid, a subject's score is a clear indication of the proportion of words in the domain he knows, i.e. of the absolute number of English words he knows.Clearly, tests like the SLT's are useful also in EFL teaching contexts, especially at advanced levels, where the amount of vocabulary (whether the result of systematic training or unsystematic exposure) is large, calling for tests which cover an unrestricted, non-specific domain of vocabulary. Estimates of the size of foreign learners' English vocabulary as represented by the SLT's may serve a variety of purposes, e.g. in individual diagnostics, or in EFL research.The SLT's were tried out on a particular population of EFL students, viz. Dutch students of English at university level. Four SLT's were administered to these subjects in four experiments between 1976 and 1980. The tests were slightly adapted, both in item content and in test format (e.g. translation into Dutch was accepted as a response mode).The results of the extensive experimentation concerning these tests (reliability and validity data, psychometric test indices etc.) are presented.

In order to specify false friends for French speakers using a general dictionary of English (LDOCE : Longman Dictionary of Contemporary English. 1978), we propose a system of tags and cross-references. Special attention is paid to an outline of the system of syntactic specifications that will warn the users of the dictionary. Detailed syntactic specifications are possible in LDOCE because it makes use of a sophisticated system of grammati-cal fields and codes. Moreover, the dictionary is available as a data base, developed in the LONGMAN-LIEGE project, so that the false friends dimension can gradually be incorporated in our monolingual dictionary.

One of the aims of every linguistic theory is to provide an exact and systematic description of the meaning of any sentence. The notion of "valence" is widely used in West- and East-European linguistics (Tesnière, Helbig/Schenkel, Apresjan, Mel'cuk, etc.) in order to account for the connection between syntactic structures and word meanings. So, in John saw the children it is due to the inherent valences of SEE that John is under-stood as 'seeing' and the children as 'seen': the referents of these two sentence elements carry specific meaning aspects of the verb. They only function in this way if they meet the formal characteristics set by the verb.In this paper I furnish a definition of "valence" and a corresponding discovery procedure. This procedure yields an explicit set of valences of any particular word. Valences are considered aspects of the meaning of a (set of) word(s) X, carried by the referent of a (set of) word(s) Y; as a corrolary, the meaning of this Y cannot be completely understood without taking into account the functioning of its referent as the carrier of the valence meaning. The discovery procedure does not assign a specific valence meaning to a verb if the functioning of the corresponding sentence element can be (completely) accounted for on the grounds of its own form. To illustrate this point, a valence labelled "price" which is claimed by Apresjan for the Russian verb ARENDOVAT' (='to lease'), would be rejected by the procedure because the corresponding sentence element has a form and a meaning which are not in function of its being dependent to ARENDOVAT', but would have the same form and meaning in other contexts.An analysis of the syntax and semantics of the Dutch verb WACHTEN (='to wait') illustrates the procedure. A short contrastive analysis of Russian and Dutch 'wait' shows some interesting differences in their comparable valences and in the syntactic peculiarities of adjuncts to these verbs.

The approaches discussed in this paper refer to the problems involved in accessing words in the Mental Lexicon. Once such access has taken place all the other information associated with a word becomes available.The first section concentrates upon describing three models for lexical access which are defined in terms of the notion of an access code, the format in which information is presented to effect an attachment to the internal representation. The models involving direct sound-based access and the combined sound and visual access systems are extended by a morpheme based model, in which all access codes are more abstract than usually proposed.The second major section is devoted to a more detailed description of two specific models for lexical access, the Logogen model of Morton and the Active Search model of Forster. Following these the point is made that a model of the Mental Lexicon and its systems needs to be supported by a model for the specific tasks used to test such models and a number of stimulus attributes are described. The possible minimal inferences which can be drawn from effects on such attributes are listed. A third, combined model is proposed which has elements of both active and passive accessing systems, called the Short-List model, which offers a better analysis of the experimental data which is defined by specific tasks as well as the structure of the accessing systems.

In the literature on the organisation of the bilingual lexicon, two closely related issues have played a major role. In the so-called independence-interdependence issue, the central question is: Do bilinguals have two separate storage and retrieval systems, one for each language, or is there just one common memory store? A second group of studies centers around the 'compound-coordinate' distinction: 'compound' bilinguals are supposed to have only one semantic system that serves for both languages, whereas on the other hand 'coordinate' bilinguals have a separate semantic system for each language.In order to resolve the above controversies researchers have employed different experimental techniques. Many of these studies have resulted in distinct and apparently contradictory conclusions. The main reason for these unclear results seems to be the defining of the level of language processing involved in these studies: the variety of experimental techniques used has caused an obvious incompatibility of the various results obtained.One other complication that has repeatedly been neglected within the exper-iments reported in the literature concerns the existence of individual differences in bilingual subjects. By studying the distinction between 'compound' and 'coordinate' bilingualism many investigators have demonstra-ted the relevence of its consequences for the selection of bilingual subjects to be used.Two experiments in which an attempt has been made to minimise the effect of some of the fundamental problems mentioned above, are globally discussed in the context of two models for the organisation of the monolingual lexicon (Forster, 1976; Morton, 1979). Within these experiments we investigated the effects of word-frequency and word-repetition within and between languages in respectively Dutch-English staff-members of the English-American Institute at the KUN and Dutch-English students, in order to explore the sharedness of lexical 'access-files' and for 'input-logogens'. In these experiments, that support an input-mechanism common to both English and Dutch, we have used cognates, words similar in form in both languages and differing in meaning (boot) or similar in meaning (hand).

For some language (e.g. Latin, Russian) learners need to acquire only one skill: Reading. This article proposes a learning method for beginning reading in a foreign language wich consist of a dictionary in which lexical and grammatical information are combined. The working of this Dictionary Search Method is demonstrated. With the help of a search-algorithm, learners may, quite independently, grasp the meaning of authentic texts. As lexicon and grammar are fully combined in the reading (or learning to read) process, there is no risk of a gap between formal language knowledge and the application of it.From psycholinguistic literature we know that the reader needs less grammatical information than the writer or speaker. Traditional pedagogic grammar reflects the production oriented language knowledge. This is demonstrated by comparing fragments of a reception- and production-oriented language knowledge. In designing reading courses in a foreign language we have to consider this fact which is further clarified by some examples from different languages. Another very important fact to be considered is the level of conceptual knowledge the reader has of the subject. Thus a course for highly specialised learners may well be very different from a course for beginning students, concerning the amount of formal language knowledge to be learned. The integrated dictionary offers all the necessary grammatical information. The learners themselves decide on the parts they use and acquire.

Word recognition is an important part of the reading process. During word recognition a Dutch reader of English will regularly process English words wich in different ways resemble Dutch words. Many of them are cognate words. Some examples are: block, dance, beer, bath. If it could be shown that the meaning of such English words need not be learned but becomes available to a reader in the same way as it becomes available when their Dutch counterparts are processed in a Dutch context language teaching could benefit from that insight. Using a lexical decision paradigm with pseudo-cognates as stimuli (3 to 5 letter words) it was shown that the following types of cognates were recognised as such during visual word recognition in English by Dutch subjects (first year university students not specializing in English):1. Dutch cognate words that had kept their spelling in an English context.2. Dutch cognate words that in an English context were still homophonous with their Dutch version and differed in the spelling of only one phoneme from their Dutch counterparts (cf. the Dutch-English cognate pair TEKST-TEXT).Of all other types of cognate words, also those that were homophonous with their counterparts in the other language but differed in the spelling of two or more phonemes, the cognate status was not recognised. Applications of the results are discussed.

In the recent literature on the methodology of foreign language teaching there is a growing consensus that new words should be presented in texts and not in vocabularies. The theoretical background of this assumption is explained through Van Parreren's theory of trace systems in memory.In particular it is argued that a good retention of new words requires that the traces of meaningfully related words be connected in many diffe-rent ways. Words in texts are already so connected (whereas words in vo-cabularies etc. are not). Moreover it is also possible to treat words in texts in a number of different ways.Although it is thus generally acknowledged that presenting words in texts is preferable, little is known about the nature of the psychological processes of comprehending and retaining new words when presented in texts. In order to gain insight into these processes a qualitative investigation was carried out. Texts in different foreign languages were read by adult subjects. These texts contained several words unknown to them. These words were tested twice: after a short and after a longer interval. While recalling the meaning of the words the subjects had to think aloud.These recall protocols were qualitatively analysed with respect to a number of questions. It was concluded that presenting words in texts offer many possibilities to ''embed" words in meaningful memory systems.However some linguistic and psychological conditions (concerning the text and the treatment of the text by the subjects) have to be taken into ac-count. In particular attention has to be paid only to a part of the unknown words, i.e. to the words in linguistically favourable positions. As a fa-vourable psychological condition for comprehension and retention diversity of treatment of the target words by the subjects seems to be required.

Applied Linguistics is generally regarded as a multidisciplinary field in which didactics, psychology and linguistics participate. It is remarkable that within the context of foreign language teaching the focus is mainly on the didactic experiment and on the construction of psycholinguistic hypotheses. Yet for a linguistic-didactic experiment to be relevant, insight in what is to be taught, viz. language,is necessary. Many variants of language teaching could have been avoided with a better linguistic insight. Moreover, a better linguistic understanding in applied linguis-tics leads to a better distinction between the views of linguists on language didactics and psycholinguistics and the descriptions of language they give.In this paper the relation between grammar and vocabulary is discussed. It is argued that this distinction is based more on definition than on reality. Stressing the importance of the role of vocabulary does not imply denying or minimising the importance of grammar. On the contrary, the traditional task division in linguistics between grammar and lexicology has led to a sterile grammatical description. Recent tendencies in linguistics now show a more integrated description of grammar and vocabu-lary.Finally, with regard to the didactically important problem of vocabu-lary selection, some remarks are made concerning the difference between selection on the basis of linguistic properties and selection on the basis of usually arbitrary non-linguistic idiosyncrasies of words and the influence of this on teaching material. This is illustrated with examples from language courses.

The acquisition of WF-rules by children learning their mother-tongue has hardly been studied up to now for English or Dutch. Yet, it seemed to me that the acquisition of WF proceeds much more slowly than the acquisition of phonology, syntax and flexional morphology.To test this hypothesis a WF-test of 211 items was submitted to a sample of 109 thirteen-year-old children of different social categories and to a control group of 20 adults. An adapted version has been submitted later on to a sample of 80 sixteen-year-old adolescents. Both versions try to measure the knowledge of WF-rules, and not the knowledge of certain morpho-logically complex words.It appears from the test that the children master only half of the WF of Dutch : only 47 % of the items have been correctly formed, understood or judged. Even using a less rigid criterion - not mastery, but WF-ability in general - they don't do much better (55 %) . The lexical-morphological ability of the adolescents reaches 69 %; i.e. one fourth higher than that of the children. The proportion of normative answers rises from 39 % of all answers given by children over 54 % for the adolescents to 73 % for the adults.The author's conclusion that children - and to a lesser extent, adolescents -do not master Dutch WF-rules, is illustrated with the discussion of four test items. The author stresses the desirability of teaching WF-rules at the right age and of correctly distinguishing between knowledge of words-the norm- and knowledge of the rules to form and understand complex words, i.e. the language-system.

The basic point behind the present paper is the general premise that the student, irrespective of the method of instruction followed, is suddenly confronted with a large number of new words at the point when he moves from pedagogically determined text material to "authentic" materials, such as newspaper articles, advertisement, literary texts, etc.An attempt is made to have a clear insight, both quantitatively and qualitatively, into the vocabulary gap that requires bridging, via an analysis of the type of newspaper extract used in class (for reading comprehension) and of Dutch essays produced by French-speaking pupils (for written production). These two elements are further compared with a basic vocabulary list and with essays produced by Dutch-speaking pupils in their mother tongue.The following conclusions become apparent :a) among the "outsiders", i.e. words not in the basic vocabulary, a restricted number of loanwords (+ 15 %) and non-derived or non-compound words (± 25 %) were noted; b) the majority of "outsiders" were derived or compounds and chiefly verb forms ; c) compared with their Dutch-language peers the majority of French-speakers in using the items referred to in (b) are less ready to "play" with lexical variants; d) the characteristic noted in (c) is not only attributable to a more limited competence in word formation but also due to the fact that bilingual dictionaries provide insufficient information on the nuances expressed through prefixes or particles.The above observations lead to the conclusion that it is desirable to direct more attention to non-inflectional morphology at all levels -university, teacher training in particular, textbook and grammar instruc-tion. Research and conclusions relate to the teaching of languages made up of a rich non-inflectional morphology (in this case Dutch) to native speakers of a language whose morphology is relatively poor in non-inflectional items (e.g. French), and are not necessarily applicable to other languages