Paul Bogaards Tuning the dictionary to the skills of intermediate leamers Abstract. Entries in the bilingual dictionary may have quite different structures. In compiling a dictionary for a particular user group it is important to know which type of entry is most effective. After a brief overview of some aspects of the structure of entries in receptive dictionaries, this paper describes four experiments which were aimed at gaining some insight in the ways intermediate learners of French take advantage of different entry structures. The results show that highly structured entries allow users to find more translations in a g1ven period of time. Acquaintance with the principles structuring a particular dictionary enhances translation retrieval even more. 0. Several years ago I was asked to compile a new bilingual dictionary for Dutch secondary school students having leamed French for two or ·three years. The two volum~s of this medium size dictionary had to be compiled as a sort of excerpt of the "big" Van Dale French-Dutch and Dutch-French dictionaries, which had been elaborated for professional users and very advanced leamers. About one third of the text contained in the big dictionaries was to be entered in the new desk dictionanes. This was a rather new situation in the history of Dutch lexicography, because up to then medium size bilingual dictionaries had also been the biggest ones available. They were used by translators and other professionals as well as by leamers of almost any level. As a result, the first thing to do was to. think about the specific characteristics of this product that was meant tobe used primarily by intermediate leamers of French. One of the first texts I came across during my search for literature on the theory and practice of the adaptation of dictionaries to specific user groups, was the Summary Report of a conference on lexicography held in Bloomington (Indiana) in 1960. In this report, the first point, "on which the conference expressed unanimity or very general agreement", has been very often cited ever since, and always with approval. lt reads as follows: Dictionaries should be designed with a special set of users in mind and for their specific needs (Householder 1962: 279). I am sure that many lexicographers and many publishers have taken to heart this statement, but I am afraid they have taken too literally the words in mind. That is to say that I failed to find very many studies about what skills or other qualities characterize special sets of users or about the specific needs of such sets of users FLuL 23 (1994) Tuning the dictionary to the skills of intermediate learners 193 (see Bogaards 1988). Unfortunately, this situation has not changed fundamentally since then. In this paper I want to discuss four experiments in which one particular point of the skills the dictionary user is supposed to have was investigated. lt concems the problem of finding the relevant information in entries with different structures. Before reporting on the experiments, I will give a brief discussion of entry structure in bilingual dictionaries. Because the experiments have only used the receptive part of the dictionary, i.e. the French-Dutch part for Dutch leamers of French, I will limit this discussion to that type of dictionary. Fora more complete treatment of entry structure in bilingual dictionaries the reader is referred to Baunebjerg Hansen (1990). 1. The structure of entries in receptive bilingual dictionaries Entries in receptive bilingual dictionaries begin with the headword, which is most of the time accompanied by grammatical information. Sometimes the different grammatical uses of the headword are numbered or distinguished by letters, but they may also be separated by commas or semicolons. This holds also for the different meanings, which may be formally characterized or which may have to be inferred from the translations or equivalents. In some cases a distinction is made, for each meaning, between the most frequent or else the primary or preferred translation and other, less frequent, more contextually restricted translations or stylistic synonyms. One further possible difference concems the presentation of examples: they may imniediately follow the translation of a given meaning, or there is first an overview of the more or less general, decontextualized translation equivalents of all the meanings, followed by a block of examples. To make the discussion somewhat more concrete, let us have a look at the entry assurer in four different dictionaries with French as the source language (see page 194). The entry in figure 1 seems tobe the least structured one. All translations are given as separate Dutch equivalents of the French verb, and examples and grammatical models of different uses are given in the second part of the entry. In figure 2 the verb has been split up in two grammatical categories, a transitive verb (1 vt) and a reflexive verb (II vpr). Furthermore, the translations of assurer as a transitive verb are presented in nine sections (indicated by semicolons), some of which have several synonyms (separated by commas). FLuL 23 (1994) 194 assurer verzekeren; veilig stellen; vastzetten; waarborgen; verzekeren (tegen schade); voorzien (in); qch. a qn. iem. iets verzekeren; qn de qch. iem. de verzekering van iets geven; s'contre zieh verzekeren (tegen); s'de qn iem. in verzekerde bewaring stellen; s'sur ses pieds stevig gaan staan; s' de qch. zieh overtuigen van iets Fig. 1: Wolters assurer I overg bevestigen, vastzetten, steunen, stutten, vastmaken; vastberaden maken, vertrouwen geven; verzekeren; verschaffen; waarborgen; beveiligen; zorgen voor; voorzien in; beweren (dat...); II wederk: s'zieh verzekeren; zieh vergewissen; verwerven, verkrijgen; s'~ de zieh vergewissen van; zieh verzekeren van, zieh verschaffen; s'de (la personne de) qn iem. in verzekerde bewaring nemen; s'que... ook: zieh ervan overtuigen, dat... Fig. 2: Kramers assurer [asyRe] 0 1 vt a (certifier) a qn que to assure sb that; que to affirm ou contend ou assert that; cela vaut la peine, je vous assure it's worth it, I assure you. b (confirmer) qn de amitie, bonne foi to assure sb of; sa participation nous est assuree we have been assured of his participation, we're guaranteed that he'II take part. c (Fin: . par contrat) maison, bijoux, vehicule to insure (contre against) personne to assure. qn sur la vie to give sb (a) life assurance ou insurance, assure sb's life; faire qch to insure sth, have ou get sth insured; etre assure to be insured. d (Journir) fonctionnement, permanence etc to maintain; surveillance to ensure, provide, maintain; service to operate, provide. pendant la greve, les mineurs n'assureront que les travaux d'entretien during the strike the miners will carry out ou undertake maintenance work only; on utilise des appareils electroniques pour la surveillance des locaux electronic apparatus is used to guard the prernises ou to ensure that the premises are guarded; l'avion qui assure la liaison entre Geneve et Aberdeen the plane that operates between Geneva and Aberdeen; l'armee a du le ravitaillement des sinistres the army has had (to be moved in) to ensure ou provide supplies for the victims; (Jur) sa propre defense to conduct one's own defense; la direction d'un service to head up a department, be in charge of a department; - Je remplacement de pieces defectueuses to guarantee the replacement of faulty parts ou to replace faulty parts; - Je suivi d'une commande to follow up an order. e (procurer, garqntir) une situation a son fils to secure a position för one's son; cela devrait leur une vie aisee that should ensure that they lead a comfortable life ou ensure a comfortable life for them; ,; a lui a assure Ja victoire that ensured his victory ou made his victory certain. f (rendre sur) bonheur, succes, paix to ensure; fortune to secure; avenir to make certain. (Mil) les frontieres contre to make the frontiers secure from; (fig) ses arrieres to ensure one has something to fall back on; cela m'assure un toit pour les vacances that makes sure l'11 have a roof over my head ou that ensures me a roof over my head for the holidays. g (affermir) pas, prise, echelle to steady; (fixer) echelle, volet to secure; (Alpinisme) to belay. il assura Paul Bogaards ses lunettes sur son nez he fixed his glasses firmly on his nose. 2 vi (*: etre a la hauteur) to be very good. il assure avec les femmes he's very good with women; ne pas ~ to be useless* ou no good*; je n'assure pas du tout en allemand I'm useless* at German ou no good* at German at all. 3 s'assurer vpr a (verifier) s'que/ de qch to make sure that/ of sth, check that/ sth, ascertain that/ sth; assuretoi qu'on n'a vole check if ou make sure ou ascertain that nothing has been stolen; assure-toi si le robinet est ferme check if ou make sure the tap (Brit) ou faucet (US) is off; je vais m'en ~l'11 make sure ou check. b (contracter une assurance) to insure o.s. (contre against). (se premunir) s'~ contre attaque, eventualite to insure (o.s.) against; s'~ sur la vie to insure one's life, take out (a) life assurance ou insurance. c (se procurer) ~ l'aide de qn/ la victoire to secure ou ensure sb's help/ victory; il s'est ainsi assure un revenu in this way he made sure of an inccime for himself, he thus ensured ou secured himself an income; s'~ l'acces de to secure access to. d (s'affermir) to steady o.s. (sur on); (Alpinisme) to belay o.s. s'~ sur sa selle/ ~es jambes to steady o.s. in one's saddle/ on one's legs. e (fitter: arreter) s'~ d'un voleur to apprehend a thief. Fig. 3: Robert & Collins assurer [asyre] <f> < ➔ to3> I <onov.ww.><inf.> 0.1 dat varkentje wel kunnen wassen; II <ov.ww.> 0.1 verzekeren ⇒ betuigen, (be)zweren, de verzekering geven van 0.2 verzekeren ⇒ assureren 0.3 waarborgen ⇒ veilig stellen, zeker stellen 0.4 verzorgen ⇒ belast zijn met, de verantwoordelijkheid hebben voor, waarnemen 0.5 verzekeren ⇒ bezorgen, verschaffen, voorzien van 0.6 beveiligen ⇒ beschutten, vrijwaren 0.7 verzekeren ⇒ bevestigen, vastmaken, vastzetten 0.8 <bergsport> zekeren · ♦ 1.3 l'avenir de ses enfants voor de toekomst van zijn kinderen zorgen 1.4 ~ une permanence*; ~ son service dienst doen, hebben, zijn werk verrichten; un service, une liaison een verbinding onderhouden 1.5 ~ des munitions a l'armee het leger met munitie bevoorraden; ~ une reute a qn. iem. een jaargeld toekennen 1.6 ~ ses frontieres zijn grenzen beveiligen 1.7 <fig.> ~ sa contenance een vastberaden houding aannemen; ~ ! es volets de luiken vastzetten 1. 'I[ <scheep.> - Je pavillon de vlag verzekeren 4.2 <wederk.> s'~ contre qc. zieh tegen iets verzekeren 4.5 <wederk.> s'~ (de) qc. zieh van iets verzekeren; <wederk.> s'~ Ja faveur de qn. iemands gunst winnen 4.6 <wederk.> s'~ contre qc. zieh tegen iets indekken, zieh tegen iets wapenen 6.1 ~ qn. de sa sincerite iem. van zijn oprechtheid verzekeren; 6.2 qc., qn. contre l'incendie iets, iem. tegen brand verzekeren; 6.3 sa conduite passee nous assure de l'avenir zijn staat van dienst in het verleden vormt een waarborg voor de toekomst; m s'~ <wk.ww.> 0.1 zieh vergewissen ⇒ zieh overtuigen 0.2 een stevige houding aannemen <ook fi_g.> 6.1 s' ~ de qc. zieh van iets vergewissen, overtuigen 6.2 s'sur ses jambes stevig op zijn benen gaan staan; s'~ sur sa seile stevig in het zadel gaan zitten 8.1 - (a) qn. que <+ aant.w.> iem. verzekeren dat; s'~ que <+ aant. w.> zieh ervan vergewissen, overtuigen dat Fig. 4: Van Dale FLuL 23 (1994) Tuning the dictionary to the skills of intermediate learners 195 In figure 3 there are three grammatical categories: transitive verb (1 vt), intransitive verb (2 vi) and reflexive verb (3 vpr). Two of these uses have several meanings, which are indicated by letters 1• Some of these meanings (e.g. la, lb) are only translated in the context of a given structure or within a sentence, others (e.g. ld) have broader, less contextualized translations and are illustrated afterwards in sentences. All examples follow immediately the meaning which they are bound to illustrate or which they otherwise belong to. In figure 4 we find a similar subdivision in three uses: intransitive verb (1 onov. ww.), transitive verb (II ov. ww.), and reflexive verb (III wk. ww.). In this case, for each meaning, indicated as 0.1, 0.2, etc, one or more decontextualized translations are given; an arrow ( ⇒) separates primary translations from synonyms. For each grammatical use, all examples are grouped together. In order to indicate which example belongs to which meaning, the number of the meaning is repeated before the example; this is always the second digit of the two digit code. In case there is no link with a particular meaning, the sign 'I[ is used. The first digit in the code indicates the grammatical category of the word the headword in the given meaning is combined with. Tue meaning of these first digits is as follows: 1. noun 6. preposition 2. adjective 7. numeral or article 3. verb 8. conjunction 4. pronoun 9. interjection 5. adverb So, in category II (transitive verb), example '1.6 assurer ses frontieres zijn grenzen beveiligen' (to make one's frontiers secure) shows a noun (category 1) in combination with meaning number 6 ofassurer, whereas example '6.1 assurer q[uelqu'u]n. de sa sincerite iem[and] van zijn oprechtheid verzekeren' (to assure somebody of one's sincerity) shows the use of a preposition (category 6) with meaning number 1 of the same verb. lt is not my intention to give a detailed account of all the structural differences between these four entries or to defend a given structure against the sometimes fierce attacks which have been made (see Hausmann 1988). Nor do these four structures exhaust the possibilities; they just give some idea of the great variation that can be found in different dictionaries. My point here is that, to the best of my knowledge, these or other solutions for the presentation of the rieb and very complex information which is to be found in dictionaries have never been tested or even compared for their effectiveness or their suitability for different user groups. The senses are also explained in French through synonyms (e.g. la certifier) or by means of possible objects (e.g. 1c maison, bijoux, vehicule). These indications being intended for the productive use by the French user, I will not consider this feature here. FLuL 23 (1994) 196 Paul Bogaards Having been invited to adapt a dictionary including the structure of fig. 4, the Van Dale structure, one of the first questions I had to answer was whether or not the two digit code could be maintained in a dictionary that was to be used by a public of non professional dictionary users. I have addressed that question in the following series of experiments. 2. The experiments 2.1 First experiment Subjects. Tue subjects were 204 students in different types of classes and in different years in one school. Students at level 1 were in their fourth year of French in a lower section of the educational system; those at levels 2, 3 and 4 were in their third, fourth or fifth year of French respectively in a higher section. Students of level 5 were in their fourth year of French in the highest, pre-university section. All students were native speakers of Dutch. There were two classes with about the same number of students at each level. Method and materials. All students were given a text and a micro-dictionary. The text (see page 197) was taken from a popular French newspaper. lt had been translated into Dutch, except for a number of words I was pretty sure most students would not know. These were underlined and numbered in the French text and replaced with <lots in the Dutch translation, which was at the right of the French text. The subjects were asked to look up the untranslated words in their microdictionary and to write down the translations on an answersheet. The French-Dutch micro-dictionaries contained photocopies of all the pages they needed to find the relevant information. The pages were in alphabetical order. Half of these microdictionaries contained photocopies of the "big" Van Dale dictionary, the other half containing photocopies of the medium size Kramers dictionary. As the entries of the Van Dale dictionary are much longer than those in the Kramers dictionary and each page contains much more information, extra pages of the Kramers dictionary had been added in order to give each student about the same amount of text. All students in one class worked with the same micro-dictionary. In two classes of the same level different micro-dictionaries were used. In order to make sure that all students knew what they were expected to do and to make them feel familiar with the micro-dictionary they had to work with, there was first a training session using the first part of the text. All questions were answered, but no explicit instruction was given in the use of any type of dictionary. At the end of the training session the first part of the text was read out and all solutions were discussed. After that, the subjects were given 15 minutes to do the test. The test contained the remainder of the newspaper item in which 27 items had not been translated. FLuL 23 (1994) Tuning the dictionary to the skills of intermediate leamers 197 UN ENFANT ATTAQUE PAR QUATRE CHIENS Een kind aangevallen door vier honden Plus jamais, Rerny, onze ans, ne croira que le chien est l'arni fidele de l'hornrne, le cornpagnon ideal de jeu pour les enfants. L'ecolier a failli etre devore vif au cours des fetes pascales a Saint-Cyr-La- Riviere par quatre gros bergers< 1 J. "Heureusernent que rnon fils portait un blouson< 2 J en cuir double d'une fourrure<3l, explique Alain Colinet, le pere de Rerny. Sinon, i1 aurait ete tue. Heureusernent aussi que les chiens ont fait tornber Rerny la face contre terre et non a la renverse< 4 l. L'enfant a pu se proteger le cou, sinon iJ aurait ete egorge< 5 l." Dans son lit d'höpital, les jambes couvertes d'epais bandages < 6l, Rerny raconte: "J'ai toujours eu peur des gros chiens. Aussi, quand j'ai aper9u ces deux rnolossesm qui rne barraient< 8 l la route, j'ai fait derni-tour< 9 l. Alors deux autres gros chiens ont bond{ 10 l d'un fosse< 11 l." Nooit rneer zal Rerny (11 jaar) denken dat honden trouwe rnensenvrienden zijn, ideale speelrnakkers voor kinderen. In Saint-Cyr-La- Riviere is deze scholier tijdens het paasweekend bijna levend verslonden door vier grote ..... (1). "Gelukkig droeg rnijn zoon een leren < 2 J gevoerd rnet .....<3 l, legt Alain Colinet, Rerny's vader, uit. Anders zou hij doodgebeten zijn. Gelukkig ook hebben de honden Rerny rnet zijn gezicht tegen de grond gegooid en niet .....< 4 l. Het kind heeft zijn nek kunnen bescherrnen, anders zou hij .....< 5 l zijn." In zijn bed in het ziekenhuis, rnet zijn benen dik in .....< 6l, vertelt Rerny: "Ik ben altijd bang geweest voor grote honden. Dus, toen ik die twee .....< 7l zag die rne de weg .....< 8l, heb/ ben ik .....< 9l. Toen ..... <10 J er twee andere grote honden uit een .... .0 1 l." Text 1: First part of the text used in the first, second, and third experirnent Results. Tue results are presented in Table 1 (see page 198). As can be seen, there are only small differences between the classes that worked with the Van Dale dictionary and those that used the Kramers dictionary. Tue mean number of words looked up is 19.0 and 19.3 respectively for the total numbers of students working with one dictionary. Tue total mean correct scores were 89.7 (s.d. 9.3) for Van Dale and 88.2 (s.d. 9.5) for Kramers; this difference was statistically not significant. This means that intermediate students are able to look up some 19 words in 15 minutes and that about two of these look-ups lead to inacceptable translations. This result was taken to mean that secondary school .students were able to find about the same number of translations with about the same degree of correctness in a traditional medium size dictionary andin the "big" Van Dale dictionary with its highly structured entries. Some additional findings were that intermediate students sometimes choose a wrang entry (lacer instead of lacerer), that the nesting of words, i.e. the presentation of several entries in one article, may lead to relatively high numbers of mistakes, that a clear structure of the entries can help to find more correct translations F.LuL 23 (1994) 198 Paul Bogaards and that it is important to avoid lang entries, because many students are not willing to scan langer stretches of text (see Bogaards 1985). All in all the results indicated that there seemed to be no <langer in including the two digit code in the new medium size dictionaries: this feature clearly did not hamper the successful use of the dictionary. Level 1 2 3 4 5 Total "big" Van Dale 1 Kramers N subj N words N errors Level N subj N words 14 18.1 1.9 1 15 17.3 23 15.3 2.4 2 23 14.9 18 19.6 1.8 3 22 22.4 20 20.5 1.5 4 25 22.7 25 21.3 1.9 5 19 19.0 100 19.0 1.9 Total 104 19.3 Table 1: Mean numbers of words found and of errors made with Kramers and with the "big" Van Dale 1 N errors 1.7 1.7 3.5 2.1 2.8 2.4 2.2 Second experiment 2 Subjects, method, and materials. As soon as the desk dictionaries bad been published, the first experiment was replicated. Photocopies of the pages containing the relevant information were made from the new Van Dale dictionary. These were used in the same procedure and with the same text as in the first experiment. The subjects were 120 students at the same levels and in a similar school as in the first experiment. The expectation was that these students would have better results because the entries in the Van Dale desk dictionary were shorter than those in the "big" Van Dale and better structured than those in the Kramers. Results. The results of this experiment are presented in Table 2 (see page 199). A comparison with the results of the first experiment shows that the differences are still very small. The number of words looked up is 18.4 and the mean correct score for the total number of students is 87.3 (s.d. 10.5). These numbers do not differ significantly from those found for the Kramers dictionary in the first experiment, but the correct score is significantly lower (p < .05) than the one found for the "big" Van Dale dictionary. 2 This experiment has been carried out by Esther Putter. FLuL 23 (1994) Tuning the dictionary to the skills of intermediate learners Van Dale desk dictionary Level N subj N words N errors 1 15 15.1 1.2 2 23 17.2 2.4 3 39 20.1 3.0 4 23 20.4 2.0 5 20 19.2 2.3 Total 120 18.4 2.2 Table 2: Mean numbers of words found and of errors made with the Van Dale desk dictionary 199 A more qualitative analysis of the results reveals that at least 20% of the errors are due to the choice of a wrong entry, that many students ignore typographical indications such as parentheses, and that they make a relatively higher number of errors when the translation is not given immediately after the headword, i.e. when there are other types of information, such as grammatical codes, phonetic transcription, etc.· between the foreign word and its translation. As this experiment was carried out in another school than the first one, the slightly lower results could be explained by a difference in the cömposition of the sample. However, the fact that the results in this experiment did not differ significantly from those obtained with the Kramers dictionary suggested that another explanation might be appropriate as well. lt was hypothesized that differences that might exist between the use of the two types of dictionary, one with a traditional entry structure (Kramers) and one with the Van Dale structure, were nullified by the fact that the students first had to find the appropriate page and then the appropriate entry on that page before they could choose the most suitable translation. This means that a possible gain in time due to the clear structure of the entries may be wiped out by differences in skill in the manipulation of the alphabet and in the overall handling of a dictionary. This possibility led to the next experiment. 2.3 Third experiment 3 Subjects, method, and materials. Subjects in this experii: nent were 181 students at levels 2 to 5 in still another school of the same type as those in the first and the second experiment. Tue same text was used in the same procedure, but in this experiment the photocopies of the relevant pages in the micro-dictionaries were not given in alphabetical order, but in the order in which the students needed them. In 3 This experiment and the fourth one have been carried out by Sandra Neelis. FLuL 23 (1994) 200 Paul Bogaards this way, differences in the skill of manipulating the alphabet was for a great part excluded. Because less time seemed to be needed in this procedure, only 10 minutes were given for the translation of the 27 items of the test. Results. The results of this experiment are shown in Table 3. This time the number of correctly found translations is significantly different (p < .05): 17 .9 (s.d. 3.4) for Kramers and 20.1 (s.d. 3.7) for Van Dale. Level 2 3 4 5 Total Kramers Van Dale desk dictionary N subj N words N errors Level N subj N words 23 18.0 2.2 2 31 20.4 18 19.0 1.8 3 24 21.1 17 21.1 2.1 4 21 24.2 24 22.5 2.5 5 23 22.9 82 20.1 2.2 Total 99 22.2 Table 3: Mean numbers of words found and of errors made with Kramers and with the Van Dale desk dictionary N errors 2.7 2.0 1.5 2.1 2.1 The difference between the total mean correct scores was not statistically significant. As can be seen in Diagram 1, there is a 'ceiling effect': considerably more students working with Van Dale had finished the whole test of 27 items within 10 minutes than was the case for students working with Kramers. This means that the Van Dale entries allow a higher working pace than the Kramers' without leading to more errors. 24 22 20 18 16 8 6 4 2 0 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 items Legenda IZ) van [? ale miJ Kramers Diagram 1: Numbers of subjects having finished different numbers of items FLuL 23 (1994) Tuning the dictionary to the skills of intermediate leamers 201 Tue errors made in this experiment have been classified in two categories. More than half of the errors (52% in the case of Kramers, 58% in the case of Van Dale) are due to a lack of sensitivity to the context: where more than one equivalent are given, students choose a translation that clearly does not fit the context. About one third of the errors (37% with Kramers, 35% with Van Dale) are the result of the choice of a wrang entry (and denote even less attention to the context). Tue residual 10% of the errors are not interpretable in terms of these two classes. The negative influence of nesting that was found in the first experiment was not present here; that is to say, it did not cause more errors, but it might have slowed down the working pace. A conservative interpretation of these results confirms the former conclusions: the high degree of structuring in the Van Dale entries does not hinder intermediate leamers. A more daring conclusion would be that this structuring helps to find a larger number of correct translations in a given period of time. In order to find some confirmation for this last idea, the fourth experiment was set up. 2.4 Fourth experiment Subjects. 108 students in four classes at two different schools have participated in this experiment. They were all at level 5. Method, and materials. 59 students, one whole class at each school, were presented with an explanation of the essentials of the Van Dale two digit code. Only the grammatical codes for the noun, the adjective, the verb, and the preposition (categories 1, 2, 3, and 6 respectively) were used. This choice was made because it was felt that the whole series of nine categories would have been too much, and because the chosen categories seemed to be the only ones that would be obvious to most of the subjects. The other students have been made familiar with other aspects ofthe Van Dale dictionaries (symbols and abbreviations). In both cases the explanation was followed by a training session of 10 minutes; in this period of time, the subjects had to look up 15 words in sentences (see Text 2). They could ask all sorts of questions, but only questions about the aspects they were supposed to familiarize themselves with were elaborated upon. 1. Les clochards sont toujours dans le besoin. Zwervers ..... altijd ..... 2. Ce sont des meubles en bois dur. Dit zijn ..... meubelen. 3. Attention! Tute fourres dans un guepier. Pas op! Je .............. 4. 11 rentre le ventre pour ne pas paraitre trop gros. Hij ........... om niet al te dik te lijken. 5. Quanta ces deux camarades, Ja partie est egale. Wat die twee vrienden betreft, ........ . Text 2: First items of the materials used in the fourth experiment FLuL 23 (1994) 202 Paul Bogaards All subjects bad tbe same micro-dictionary, wbicb consisted of the relevant pages of the Van Dale desk dictionary. Tbe pages were in the order of tbe words to be translated. In order to minimize all differences in tbe time needed to locate tbe word on eacb page, tbe relevant entry bad been marked with a red asterisk. Tbe subjects were given 10 minutes to translate 27 words in sentences. with explanation without explanation Class N subj N words N errors Class N subj N words N errors Al 30 22.7 0.3 A2 21 18.9 0.3 B 1 29 21.1 0.5 B2 28 17.8 0.3 Total 59 21.9 0.4 Total 49 18.2 0.3 Table 4: Mean numbers of words found and of errors made by subject; ; who have had or have not had an explanation of the Van Dale two digit code Results. Table 4 gives the results of this experiment. Tbose subjects wbo bad been introduced to the use of the two digit code correctly looked up 21.5 words, wbereas those wbo bad studied tbe symbols and abbreviations looked up only 17.9 words. This difference is statistically significant (p < .05). The difference between the correct scores as a percentage of the total number of words looked up is not significant, bowever; tbe proportion of errors made is about the same. So, it seems again, but now in a clearer way, that the Van Dale structure may indeed belp to find words more quickly, ev: en in the case of intermediate leamers. lt sbould be kept in mind, bowever, that the subjects were at level 5 (the bigbest level in this series of experiments), tbat all entries used were rather sbort, and tbat only four rather "obvious" grammatical categories bave been tested. A qualitative interpretation of tbe errors made, as well as comments made by some subjects baving worked on tbe two digit code, indicate tbat these subjects bad bad two types of difficulties. The first one was tbat they did not know with wbicb item in tbe sentence the word was supposed to be combined. In a sentence like tu te fourres dans un guepier it was unclear wbether the verbal form fourres was combined with a noun (guepier: 1) or witb a preposition (dans: 6). Tbe second difficulty concerned tbe fact that subjects were unable to determine tbe category of the item the word was combined witb. Tbis was for instance tbe case in (i/ a maintenant) la gorge prise wbere prise was looked for under 3 (verb) but was presented under 2 (adjective). FLuL 23 (1994) Tuning the dictionary to the skills of intermediate learners 203 3. Discussion In the first and the second experiment reported on in this paper, the subjects themselves had to find their way through the micro-dictionary they were given. They had to determine the canonical form of the unknown word, to find the page containing that fortn, to locate the entry on that page, and to extract the relevant information from the entry. This means that various types of knowledge and skills were required to get to the place where the information was to be found: e.g. knowledge of French morphology, skill in coping with the alphabetical presentation of words in dictionaries, and more specifically in the truncated micro-dictionary they were given. In these conditions, no significant differences were found between the Van Dale dictionaries and traditional dictionaries. In the third and the fourth experiment, however, parts of the preparatory steps were left out by presenting the pages in the order they were needed or even by marking the entry on the page. In these more artificial circumstances, the Van Dale structure led to a gain in speed (not in correctness); this gain in speed was even more important when subjects had been taught the principles of the two digit code. Is this to say that this type of highly structured entries is profitable to the nonprofessional users of bilingual dictionaries? There are several reasons not to draw such a hasty conclusion. As was stated by Janssens (1991) the good use of the two digit code is based on four presuppositions: (1) the users read the int: roduction or the instructions for use in order to get acquainted with the system; (2) the users are able in the same way as the Van Dale lexicographers to determine the combination word, i.e. the contextual element which is most important; (3) the users know the grammatical categories and are able in the same way as the Van Dale lexicographers to recognize the grarnmatical categories of words in the source language as well as in the target language; (4) the users do not think the system is too complicated and are ready to memorize the nine grarnmatical categories together with their codes. As to the first point, it is generally known that dictionary users do not read introductions or instructions for use (see, for instance, Bogaards 1988). Fortunately, as may be clear from our results, and as is claimed by the publisher, those who are not acquainted with the two digit code are not hampered by its presence. Moreover, the results of the third experiment seem to indicate that even uninformed users may profit from the Van Dale structuring. The gain in speed that this dictionary allows seems to be lost, however, when users work on their own and are not presented in due time with the page where they can find the relevant information. As to the second point, it has been seen that students have difficulty in finding the right combination word. So do the lexicographers. lt is far from obvious, as has been stated by Hausmann (1988), why in the entry poche the following examples have different codes: FLuL 23 (1994) 204 1.1 [es mains dans la poche (combination with a noun) 3.1 n'avoir pas sa langue dans sa poche (combination with a verb) 6.1 l'affaire est dans la poche (combination with a preposition). Paul Bogaards Furthermore, a test done by Janssens (1991) indicates that about one hundred French-speaking university students of Dutch and Dutch-speaking language students make the same choice as the Van Dale lexicographers in only 4 out of 10 cases. Finally, Tops (in an intemal report) has made a distinction between free examples and fixed expressions. In fixed expressions the context of a given word is clear, because it is fixed, and the combination word can easily be determined. This does not hold for free examples, however. In this case the lexicographer may choose any convenient context. This means, for instance, that a word like part may be illustrated by have an important part at the election, which may be categorized as 2.2 (combination with an adjective), or by play no part in ..., which should be put under 3.2 (combination with a verb). So, the user is unable to know in advance where he has to look in order to find a given use or meaning .. Tue third point provokes similar criticism. Our fourth experiment, which was run with students at the highest level only, has shown that even for the "obvious" categories used, there may be difficulties. This point is confirmed by the results of another test done by Janssens with the same subjects as mentioned above. In this test 20 cases were proposed with the combination word underlined; most of the items contained one of the other, less "obvious" categories like pronouns, conjunctions, etc (categories 4, 5, 7, 8, and 9). Even highly qualified university language students recognize the right category in only half of the cases. So, the third presupposition does not hold, at least not for leamers at all levels, not for the total of nine categories, and not for all words of the foreign language. · Janssens (1991) states that only 17% of bis subjects said they used the two digit code, but that, if they did, they benefited most from the second digit (the one that corresponds to the semantic distinctions). Most subjects found the code too complicated and bad not memorized it. No research has been done about the use of the code in real life situations. In addition, one may wonder whether users who are acquainted with the system are willing to activate their grammatical knowledge and their knowledge of the two digit code at any time they need the dictionary, for instance when they are mentally occupied by understanding a difficult text or while writing a text. 4. Conclusion Much research needs to be done before lexicographers can do what they were urged to do more than 'thirty years ago: write dictionaries which are tuned to the needs and skills of specific user groups. As far as the aspect that has been treated in this paper is concemed, it seems that in receptive bilingual dictionaries entry structures like the one used in the Van Dale dictionaries are potentially beneficial FLuL 23 (1994) Tuning the dictionary to the skills of intermediate leamers 205 for the user. But many questions remain. Let me finish by formulating some of them: 1. At which word do users expect to find the translation (or the explanation) of different types of examples (see Bogaards 1992) for different types of tasks (see Hausmann 1988)? 2. Which elements are considered to constitute a context for a given word? 3. To what extent do gramrnatical categories or syntactic relationships play a role in the search strategies of users? 4. lf gramrnatical categories play an important role, how many can be used in a search code? And: are digits the best way to mark them or could other rnnemotechnical means (e.g. abbreviations) be of any help? 5. Should examples be given irnmediately after the meaning they are related to or in a group? If they are to be presented in a group, should their link with a given meaning be maintained (see Hausmann 1988)? References BAUNEBJERG HANSEN, Gitte (1990): Artikelstruktur im zweisprachigen Wörterbuch. Überlegungen zur Darbietung von Übersetzungsäquivalenten im Wörterbuchartikel. Tübingen: Niemeyer. BOGAARDS, Paul (1985): "Ben vergelijking van het gebruik van twee woordenboeken door leerlingen van -het a.v.o.". In: Boou, G.E./ GoEMAN, A.C.M./ REENEN, P. Th. van (eds.): Corpusgebaseerde woordanalyse. Jaarboek 1985. Amsterdam: VU, 17-19. BOGAARDS, Paul (1988): «Apropos de l'usage du dictionnaire de langue etrangere». In: Cahiers de lexicologie 52, 131-152. BOGAARDS, Paul (1992): "French dictionary users and word frequency". In: TOMMOLA, Hannu [et al.] (eds} Euralex '92 Proceedings. Tampere (Studia Translatologica, Ser. A, vol. 2), 51-61. HAUSMANN, Franz Josef (1988): "Grun~probleme des zweisprachigen Wörterbuchs". In: HYLD- GAARD-JENSEN, Karl/ ZETIBRSTEN, Arne (eds.): Symposium on Lexicography III. Proceedings of the Third International Symposium on Lexicography May 14-16, 1986 at the University of Copenhagen. Tübingen: Niemeyer, 137-154. HOUSEHOLDER, F. w. (1962): "Summary Report". In: HOUSEHOLDER, F. W./ SAPORTA, s. (eds.): Problems in lexicography. Bloornington: Indiana University Press 1962, 279-282. 'JANSSENS, G. (1991): "Tweetalige woordenboeken van en naar het Nederlands. De (verschillen in de) gramrnaticale achtergrond van de (onderscheiden) gebruikers". In: Forum der Letteren 32, 108-113. Kramers = Prick van Wely, F.P.H. (s.d.): Kramers' Woordenboek Frans (36th ed.). Den Haag: Van Goor. Robert & Collins = Duval, Alain/ Sinclair Knight, Lorna (1993): Le Robert & Collins. Dictionnaire franrais-anglais, anglais-franrais. Senior. (3eme ed.). Paris: Dictionnaires Le Robert. Van Dale = Al, Bernard [et al.] (1990): Van Dale Groot Woordenboek Frans-Nederlands. (2nd ed.). Utrecht/ Antwerpen: Van Dale Lexicografie. Wolters = Verrijn Stuart-Alma, H. (1970): Frans Woordenboek Frans-Nederlands. (15th ed.). Groningen: Wolters-Noordhoff. FLuL 23 (1994)