A RESEARCH-BASED STUDY OF FOREIGN STUDENTS' USE OF GRAMMATICAL CODES IN FIVE LEADING BRITISH LEARNERS' DICTIONARIES

Grammatical codes are one of severa! ways of including grammar in leamers' dictionaries. In our research we focussed on the usability and user-friendliness of leamers' dictionaries as regards grammatical information. The results presented and discussed in this article are based on answers obtained by a questionnaire that tested the understanding of codes found in five leading British monolingual learners' dictionaries and the success of the explanations of the same codes provided in the front matter of each dictionary. The results are presented by dictionaries and by codes. The most important finding of this research is that the understanding of the code and thus its usefulness depends on the code itself rather than on the dictionary


l. Introduction
Dictionaries , especially monolingual dictionaries designed to meet the needs of a foreign language leamer, contain a large amount of information of various types .Early studies conducted into dictionary use (Bamhart 1962, Quirk 1973, Bejoint 1981, Tomaszczyk 1979) show that dictionaries are most often used for checking the meaning of unknown words.Apart from meaning, other types of information are also included and used by dictionary users but to a lesser extent than meaning.Undoubtedly, one of the important elements of monolingual learners' dictionaries is the inclusion of grammar (McCorduck 1993) .Grammar has been included in leamers' dictionaries since the very beginning (Cowie 1999), but the amount of grarnmar in dictionaries has increased noticeably in the course of tirne (Bejoint 2000: 29).Typically, all monolingual leamers' dictionaries include the part of speech label, which represents grammatical information about words , where they can operate in the syntax of sentences , and what their combinatorial possibilities are.Other grammatical information refers to various grammatical properties of each individual word class, such as countability in nouns, verb patterns, attributive , postpositive, and predicative use of adjectives, as well as gradability in adjectives and adverbs (Herbst 1989, Jackson 2002, Landau 2001, Cowie 1999).Grammatical information is provided in the form of codes or abbreviated phrases as well as more indirectly, i.e. in definitions and examples of use.The question of what should be taken into consideration when including grammatical information has been widely discussed (Jackson 1985, Lemmens and Wekker 1986, Sinclair 1987), but it is of the greatest importance to know what dictionary users understand and make use of when consulting their dictionaries in terms of grammar.The only way to find this out is to carry out research into different aspects of dictionary use , including grammar.Surveys should enquire into students' awareness of the existence of grammatical information in learners ' dictionaries.Severa!aspects have been tested so far, such as syntax, complementation codes and labels (Nuccorini 1992), the obligatory use of the article, verb patterns, and countability (Tono 2001), and verb patterns, especially verb complementation (Bogaards, van der Kloot 2001).The results show that grammar is rarely looked up by the test subjects; they do not use all the available information, and they tend to choose the first definition in the entry rather than read the whole entry (this applies especially to long entries) , even though useful grammatical information may be included later in the entry.Several comparisons of codes used in different learners' dictionaries have also been made, investigating the system of grammatical codes (Aarts 1991, Dalgish 1995, Strevens 1987), the thoroughness and consistency in the use of grammatical codes (Heath 1982), and the transparency of codes (Aarts 1991, Herbst 1989, Strevens 1987).The most commonly expressed criticism refers to the fact that codes are not sufficiently transparent, therefore not sufficiently user-friendly.The main proposal is that a coding system should be self-explanatory and reasonably easy to use .
In order to test the transparency and usability of grammatical codes included in the latest editions of five leading British monolingual dictionaries, we carried out a survey, the results of which are presented and discussed in this article.Research was undertaken to prove our hypothesis that dictionary users do not regularly consult the front matter in order to check a code oran abbreviated phrase they do not understand.We, therefore, wanted to find out to what extent the codes or abbreviated phrases are understood without being explained (are they so clear that the explanation is redundant, or are they so difficult that they cannot be understood without an explanation ?) .

Test Subjects
Research was carried out among second-and third-year students of the Faculty of Economics (hereafter referred to as FE) and fourth-year students of the Faculty of Arts (hereafter referred to as FA) , Department of English, University of Ljubljana, Slovenia.Together the groups comprised 162 respondents.One hundred and fifty-seven respondents (i.e .96.9 %) speak Slovene as their mother tongue, and 5 (i.e. 3 .1 % ) speak some other language as their mother tongue.The test subjects were asked to give details about the number of years they had been studying English.Most of the respondents (i.e.69.l %) had been studying English for 10-12 years; 15.4 % had been studying it for less than 10 years, and 15.4 % for more than 12 years.Our test subjects were considered as belonging to a quite homogenous group: most of them attended schools in Slovenia; they had the same number of English lessons per year and covered the same syllabus.For these reasons we considered it unnecessary for them to undergo a placement test.However, we did enquire about the students' average grades in English in grammar school.The majority of students indicated that their average school grade in English was either A or B (37 % stated Grade A and 34.6 % stated Grade B); 24.l % had a Grade of C and only 4.3 % a Grade of D.

Test Design
The questionnaire consisted of two parts: a Dictionary User Profile Form and a Dictionary Research Test.The Dictionary User Profile Form was completed by all test subjects and was aimed at obtaining information about the dictionary users, i.e. their mother tongue, how long they have been studying English, and their grades in grammar school (cf.2.1 Test Subjects).
The Dictionary Research Test concentrated on the respondents' abilities to decipher the grammatical codes included in five leading British monolingual learners' dictionaries: COBUILD4, OALD6, CALD2, LDOCE4, and MED I.It consisted oftwo tasks.In both tasks , the respondents were subdivided into five groups depending on the dictionary tested.Thirty-two students were given ten grammatical codes or abbreviated phrases from MEDI, 32 from OALD6, 31 from LDOCE4, 36 from COBUILD4, and 31 from CALD2.In Task 1, the students were supposed to explain the code or the abbreviated phrase taken from the above-mentioned dictionaries.It was aimed at testing students' understanding of the codes.The students did not have to use the appropriate terminology; they could also explain their understanding in their mother tongue or provide an example .Task 2, on the other hand, included the same codes with the explanations taken from the front matter of each dictionary.It was aimed at testing the understanding of the explanations of these codes or abbreviated phrases.

Procedure oj data collection
For the purpose of this study, data were collected by means of a questionnaire.Test subjects were given the questionnaire, and the researchers explained what they had to do.The respondents were supposed to complete the questionnaire in 45 rninutes.The respondents' answers were appropriately coded and prepared for a statistical analysis using the Microsoft Excel program.Standard statistical methods were used for the data processing, which was carried out by the SPSS for Windows, version 11.

Results of the Dictionary Research Test
In Task 1, the range of correct answers in all the dictionaries tested is quite broad (from 12.5 % to 87 .5 % in OALD6, from 3.2 % to 90 .3% in LDOCE4, from 11.1 % to 88.9 % in COBUILD4, and from O.O % to 90.3 % in CALD2).The only dictionary where the percentage of correct answers is somehow more balanced is MEDl, where the difference between the best and the worst result is also quite large but still smaller than in other dictionaries (between 46.9 % and 96.9 %).
In Task 2, where the understanding of the explanations of the codes or abbreviated phrases was tested, COBUILD4 bas to be mentioned as the dictionary where the difference between the explanation that was understood by the highest percentage of the respondents (97 .2% ) and that which was understood by the lowest percentage of the students (22.2 % ) is most noticeable.In MED 1, the difference between the explanation that was understood by the highest percentage of the students ( 100 .O % ) and that which was understood by the lowest percentage of the respondents (62 .5 %) is the smallest, whereas the results range from 58 .1 % to 100.0 % in CALD2, from 43 .8% to 100.0 % in OALD6 , and from 38 .7 % to 100.0 % in LDOCE4 .
Let us consider the individual codes and abbreviated phrases included in all the above-mentioned dictionaries.
The codes or abbreviated phrases in MED 1 do not seem to pose great problems to dictionary users.Among the codes that were the most problematic we have to mention code 8 ' in imperative', code 1 'linking verb', and code 7 'usually progressive'.Code 6 ('usually passive') with 96 .9%, code 4 ('never before noun') with 87.5 %, and code 10 ('in negatives or questions ' ) with 87.5 % are the most successful ones.
A comparison of results concerning the explanation of the codes found in the front matter shows that only the explanation for code 6 ('usually used in the passive') was understood by ali the test subjects.The explanations for code 3 ('adjectives that can only be used before a noun ' ), code 4 ('adjectives that can never be used before a noun'), code 10 ('usually used in negative sentences or questions'), and code 9 ('usually used in the infinitive') also proved successful, since they were all understood by more than 90 % of the respondents .The respondents found the explanations for code 7 ('usually used in the progressive') and code 8 ('usually used in the imperative') the most difficult ones, since they were understood by 62.5 % and 65.6 % of the test subjects respectively.Here a parallel can be drawn with the results in Task 1, where the same codes were understood by a relatively small number of respondents.For more detailed results concerning the success in deciphering the codes and understanding their explanations in MED 1, see Table 1.Among the grammatical codes taken from OALD6, code 10 (' VN speech') proved to be the most difficult one, since it was correctly explained by only 12.5 % of the respondents .The most difficult part of this code is most certainly the second element ('speech'), since 71.9 % of the respondents correctly explained code 3 ('VN' ), which corresponded to the first element in code 10.Less than 50 % of the test subjects were able to explain the meaning of code 4 (' VN + adv./prep.')and code 9 ('VN -ing' ).
Code 2 (' pt') and code 5 ('V-ADJ') were correctly explained by 81.3 o/o and 87.5 % of the respondents respectively, a result which means that they are among the most successful codes in OALD6.
As regards the explanations of the codes, code 2 (' past tense') was the most successfully explained, which is understandable, given that all language leamers are familiar with the term 'past tense'.Apart from code 2, code 3 ('verb + noun phrase'), code 1 ('past participle') , code 5 ('verb + adjective') , and code 7 ('verb + that clause' ) are also explained in such a way that the majority of respondents are able to understand them.The most problematic code seems to be code 10 ('verb + noun phrase + direct speech') , whose explanation was understood by only 43.8 % ofrespondents.The same code also yielded the worst result in Task 1, and even the explanation provided in the front matter does not help as much as it should.Detailed results of student success in diciphering the codes and understanding of the explanations of the codes are presented in Table 2.In comparison to MEDI and OALD6, the codes included in LDOCE4 are less transparent.Only code 5 ('only before noun') was understood by more than 90 % ofthe respondents.As many as five codes were understood by less than 50 % of the students, which is far from being satisfactory.Code 9 ('+ adj/adv') was deciphered by 3.2 % ofthe respondents only.The explanation of this code ('shows that an adverb of degree is used before adjectives and adverbs') seems to be more successful, since 51.6 % of the respondents were able to understand it.Apart from code 9, code 10 (' also + plural verb BrE') was also a problematic one, since less than 20 % of the test subjects managed to explain it correctly.Again , the explanation ('shows that a group noun can take a plural verb in British English') was understood by 77 .4% of the respondents, which is quite a good result in comparison to the ability of the test subjects to explain the code.Another code that gave students problems is code 8 ('sentence adverb') , which was appropriately explained by only 23.3 % of the respondents .The explanation of this code ('shows that an adverb modifies a whole sentence') provided in the front matter is not of great help to dictionary users, either, since it was understood by only 38.7 % of the respondents.Apart from code 5 (already mentioned) the most successful codes include code 7 ('only after noun'), and code 6 ('not before noun'), as they both were successfully deciphered by more than 80 % of the respondents.Similarly, the explanations of these three codes were clear enough to be understood by the majority of respondents (the explanations for codes 5 and 6 were understood by all the respondents, and the explanation for code 7 by 96 .8%).The greatest difference between correct answers that show students' ability to decipher the code and those that show that respondents understood the explanation of the code can be found with code 2. Only 41.9 o/o of students understood the code, as opposed to 100.0 % of students who understood the explanation of the code.Further details concerning students' ability to decipher the code and understand its explanation are presented in Table 3. Generally speaking, codes in COBUILD4 give cause for concern.Only one code (code 7 -'V-PASSIVE') was understood by 88.9 o/o of the students.The results show that code 4 ('N-VAR') with 11.1 %, code 1 ('PREP-PHRASE') with 13.9 %, and code 3 ('N-COUNT-COLL') with 27.8 o/o were the most difficult ones.The results showing how well the students understood the explanation of the codes indicate that the explanations for code 7 ('passive verb'), code 2 ('comparative form'), and code 9 ('imperative ') were understood by the majority of respondents, whereas the explanation for code 4 ('variable noun') was understood by only 22 .2o/o of respondents.For more detailed results, see Table 4.The questionnaires tested the success of 50 codes -i.e. 10 codes taken from each dictionary -but the only code that was not understood by a single respondent was code 5 ('M') from CALD2.Even the explanation itself (i.e.'phrasal verb with a particle that can come before or after the object') is not really revealing, since only 58.1 o/o of the students managed to understand it.This result is not a complete failure, but if we compare it with other results for explanations of codes in this dictionary, it is the worst one.Besides code 5, we have to mention code 4 ('L') and code 8 ('R'), which were appropriately explained by less than 10 o/o of the respondents.Again, a single letter is not self-explanatory; therefore students did not manage to decipher 'L' as referring to linking verbs and 'R' as referring to reflexive verbs.Also worth mentioning is code 9 ('S'), which was correctly explained by 35.5 o/o of the students.All the users understood this code after having read the explanation.A comparison of the results showing how well the students understood the explanations of the codes reveals a completely different picture.As many as three explanations (code 3 -' adjective that is placed only before a noun'; code 9 -' singular' ; code 10 -'verb usually used in the passive') were understood by all respondents.Effective explanations also include those for code 6 (' verb with an object followed by a noun oran adjective' ) and code 7 ('verb with an object followed by the -ing form of another verb '), which were understood by more than 90 % of the test subjects.More detailed results are presented in Table 5.

Discussion
These results clearly show that the test subjects find some codes or abbreviated phrases fairly transparent, whereas others prave to be quite difficult.It is surprising, for example, that respondents had problems with the abbreviated phrase 'in imperative', since they come across the term 'imperative' in the course of leaming English, but it is understandable that they did not know the meaning of the abbreviated phrase 'linking verb', since this is not used when teaching English grammar at school.The term 'linking verb' is known only to the students of English who get acquainted with it in their grammar courses at the university, when discussing the classification of the verbs.The general dictionary user, on the other band, cannot possibly be familiar with the meaning of this term.
The term 'progressive' also poses problems because it is not used in textbooks, the result being that students are not familiar with it.At school , the term ' continuous ' is used, and it can be assumed that this term would be much better understood .It has to be stressed that all the textbooks that are officially approved for use in primary and grammar schools in Slovenia are published by the same publishing houses as the monolingual leamers' dictionaries (e.g .Cambridge Englishfor Schools -Cambridge University Press; Project -Oxford University Press; New Headway English Course -Oxford University Press; Matrix -Oxford University Press; Natura/ English -Oxford University Press) .The same terminology can justifiably be expected in textbooks as well as dictionaries intended for the same target audience.It is true that in primary school and at least at the beginning of grammar school the level of English is not sufficient to enable the use of the dictionaries tested, which are intended for advanced learners.Advanced leamers, however, have experienced many years of leaming English, and the basic leaming aid they have used bas been the school textbooks.One would, however, expect a better result with 'v-cont' (in COBUILD4) , where the term 'continuous' is used, as opposed to the term 'progressive', which is used in other leamers' dictionaries.It is possible that our respondents did not manage to understand the abbreviated form of the term (cont for continuous), but even the results showing how many students understood the explanation provided for this code are not as high as one would expect.The students seem to have been puzzled by the entire expression 'continuous verb', which is used in the front matter of this dictionary, as they are more familiar with the term 'continuous tense' rather than with the term 'continuous verb'.
Another complicated abbreviated phrase is 'sentence adverb', since a general dictionary user does not understand the concept underlying this term.This is a term language learners do not come across when learning English; they become acquainted with it as late as their grammar classes at the university (if they study English).The terms 'variable noun', 'prepositional phrase', 'collective countable noun', and 'reciprocal verb' are also much too complicated for an average dictionary user.The results for these abbreviated phrases are unsatisfactory and the reason for such poor results is most certainly the terminology used.
What about one-letter codes?The code 'S' used in CALD2 proved to be quite difficult, which is surprising if we know that this code refers to 'singular ' .In this particular example, the difficulty lies in the code itself not in the terminology used.In other leamers ' dictionaries the code 'sing.' is used (OALD6 and COBUILD4), or the word 'singular' is written in full (MED 1 and LDOCE4).We also cannot ignore the code 'M' (used in CALD2), which is the only code (out of all 50 codes tested) that was not understood by a single respondent.What is it that makes this code so complicated?As we leam from the explanation of this code, it refers to a 'phrasal verb with a particle that can come before or after the object'.The code 'M' suggests that the object can occupy the medial position between the verb and the particle, but the code itself is not self-explanatory.How can a dictionary user know that 'M' stands for medial position?
In codes consisting of severa!elements, such as 'VN speech', 'VN + adv./prep.','VN -ing', we can presuppose that it was the second element that the respondents were unable to understand, taking into account that the code ' VN' was not problematic.Probably, the reason cannot be attributed to the fact that the students do not know what 'speech', 'adv./prep.' or '-ing' mean; it is rather that they do not know what the entire code refers to.The same holds true ofthe codes '+ adj/adv', 'always + adv/prep' found in LDOCE4, since they consist of elements that average dictionary users would know, but probably users are unable to link the elements and understand what the entire code means.Consequently, the verbalization of these two codes ('shows that an adverb of degree is used before adjectives and adverbs' and 'shows that a verb must be followed by an adverb or a preposition') is perfectly clear, and is thus understood by all respondents.However, the explanation does not always prave successful, which is confirmed by the explanation of the code 'VN speech' , meaning 'verb + noun phrase + direct speech' .The test subjects understood all parts of the explanation even the last part, i.e. 'direct speech', since they get acquainted with this term in the course of leaming English, but in all likelihood, they were not able to make a connection between all three elements of the explanation.
With a high degree of certainty it can be claimed that understanding of the code or abbreviated phrase and thus its usefulness depends on the code or abbreviated phrase itself rather than on the dictionary used.The compilers of the dictionaries provide as much grammatical information as possible, much of which is included in the form of grammatical codes or abbreviated phrases.As is shown in our research, the transparency and simplicity of the codes or abbreviated phrases are of the utmost importance if we want users to understand them and to make use of them.These results also confirm the findings of Aarts (1991), Herbst (1989), andStrevens (1987) .

Conclusion
There is no doubt that dictionary reference skills involve highly complicated cognitive skills.Dictionary skills are problem-solving skills: users are trying to solve a particular linguistic conflict by consulting a dictionary.However, it should be emphasized that dictionary use is not as simple as one would imagine.A dictionary abounds with information of various types, and very often a dictionary user is not even aware of the wealth of information he/she can find when consulting a dictionary.We strongly believe that students should be made aware of the whole range of information included in monolingual learners' dictionaries, and the inclusion of codes and abbreviated phrases is no exception.The front matter of each dictionary where the types of information included in this particular dictionary are explained can be of great help.Unfortunately, many dictionary users never read it, the result being that they cannot be regarded as proficient dictionary users.The educational system can play an important role, since the teaching of dictionary skills could be integrated into the foreign language syllabus.An important tool in the training of dictionary users is the use of separate companion booklets that are meant to train users and offer them practice in using the special features, including grammatical codes.Systematic dictionary training is absolutely necessary to enable learners to become efficient dictionary users.

Table 1 :
Success in deciphering and understanding the grammatical codes in MEDI (in%)

Table 2 :
Success in deciphering and understanding the grammatical codes in OALD6 (in % )

Table 3 :
Success in deciphering and understanding the gramrnatical codes in LDOCE4 (in%)

Table 5 :
Success in deciphering and understanding the grammatical codes in CALD2 (in%)

Codes from MEDl with explanations
1. linking verb verbs that are followed by a noun or adjective complement describing the subject