Coding systems in monolingual English learners' dictionaries: form and utility. (Linguistics).

The present paper attempts to investigate and evaluate the structure of verb codes in selected pedagogical dictionaries as well as assess the usefulness of encoded syntactic information to the learner. The collection of primary sources chosen for the analysis comprises the editions of the Oxford advanced learner's dictionary published since 1974, i.e. OALDCE3 (1974), OALDCE4 (1989), OALDCE5 (1995) and OALDCE6 (2000), all the editions of the Longman dictionary of contemporary English, i.e. LDOCE1 (1978), LDOCE2 (1987) and LDOCE3 (1995), the Collins COBUILD English language dictionary (1987, henceforward COBUILD 1), the Collins COBUILD English dictionary (1995, henceforth COBUILD2) and the Cambridge international dictionary of English (1995, hereafter CIDE). The first two editions of the Oxford advanced learner's dictionary, i.e. OALDCE1 (1942) and OALDCE2 (1963), have not been taken into consideration since the coding systems employed there hardly differ from that in the 1974 edition.

The paper consists of nine parts. The rationale for encoding syntactic information on verbs in the pedagogical dictionary is discussed in the first one. Proposals for a user-friendly form of verb codes are dealt with in part two. The third section presents a brief diachronic account of the introduction of verb codes into pedagogical lexicography. Section four is concerned with the coding system employed in OALDCE3. Then, constituent parts of verb codes in the remaining dictionaries are subjected to scrutiny. Specifically, section five is concerned with symbols for the verb, and section six -- with symbols for the components of the complementation structure. The main types of coding system are identified and assessed from the vantage point of the learner in the seventh part of the paper. Results of selected pieces of research on dictionary use and the tenets of relevant theories of language teaching are subsequently referred to with a view to throwing light on students' attitudes to verb codes. Concluding rema rks, offered in the final part of the article, complete the study.

1. Introduction

Coding systems in pedagogical dictionaries have been invented in response to the need to provide the learner with adequate syntactic information on verbs in an appropriate form. As Cowie (1983a: 100) points out, devising a system of grammatical labeling which properly reflects syntactic complexities, and is thus reliable, but remains lucid and usable calls for great ingenuity. Elsewhere he emphasizes that the considerable weight of syntactic information on verbs needs to be presented in a form which is both economical and maximally explicit (Cowie 1978: 260-261).

Accuracy of description and ease of accessibility are the two main aims any dictionary is likely to pursue (Herbst 1999: 229). Unfortunately, it is difficult to satisfy both criteria concurrently. On the one hand, as Jackson (1985: 58) observes, the more detailed the syntactic information, the more elaborate and often impenetrable the system of presentation becomes. Conversely, information about how to incorporate a newly learnt word into phrases and sentences is essential to the learner, hence the need for easily understood mechanisms for providing such information. Accuracy and user-friendliness should thus, in Herbst's (1999: 229) view, be seen as the poles between which the design of a dictionary has to find its place.

Striking a balance between accuracy on the one hand and intelligibility and accessibility of syntactic information on the other is of paramount importance especially in dictionaries designed for foreign learners, whose reference skills are necessarily limited (Bejoint 1981: 211). Cowie (1983b: 136) defines reference skills as "the skills which the user is assumed to possess, or can be expected to acquire, in handling a dictionary and making effective use of the information it contains". He also remarks that not only reference skills but also the reference habits associated with the content of the dictionary and its organization are slow to change. Thus, should the pedagogical dictionary become a sophisticated reference tool, it will run the risk of outstripping the often rudimentary reference skills of those it is designed to serve. What Cowie (1987: 184) refers to as "the gap between descriptive sophistication and reference skills of many ordinary users" should therefore be bridged. Bejoint (1981: 211) concl udes ruefully that "[l]exicographers are in fact faced with an alternative: either to tune dictionaries to the existing skills of the user, thus producing a simple-easy-to-use book, or to market a beautifully contrived reference work, requiring skills that are beyond what can be expected of the average user." (1)

Despite the problems that the incorporation of syntactic information into the verb entry entails, such information can on no account be absent from the pedagogical dictionary. For one thing, English verbs are notoriously difficult to learn. To explain difficulty in acquiring certain vocabulary items it is necessary to refer to Sweet's (1899 [1964]: 141-143) distinction between encyclopedic and lexical words. The first category of words consists of low-frequency technical or specialized terms. Sweet (1964: 142) takes it for granted that foreign students of average sophistication either know their meanings already or do not find it difficult to learn them, since such words are "barren from the linguistic point of view, for they offer neither varied shades of meaning nor irregularities of form, nor do they enter into idiomatic combinations or special grammatical constructions". Conversely, lexical words, which constitute the second category, are high-frequency, necessary everyday words. It is those words that po se the most serious learning problems. According to Cowie (1983b: 136), the learning difficulties inherent in lexical words arise from the variety of senses and combinations in which they are used. He lists many verbs as illustrative examples of this category. The nature of the pattern or patterns in which a verb occurs is believed to be one of the fundamental difficulties encountered by the foreign student of English, especially in view of the fact that many verbs allow several complementation types (Palmer 1938 [1949]: vii). Besides, they are "choosy; not all verbs can appear in all sentences, even when the combinations make perfect sense" (Pinker quoted in Hamdan and Fareh 1997: 197) and analogies with the mother tongue may, more often than not, lead the learner up the garden path.

The reference needs of foreign learners, which do not coincide with those of native speakers, also justify extensive syntactic coverage in the pedagogical dictionary. Whereas for native speakers grammar means parsing and analysis, for foreign students it means synthesis because, as Hornby (1965: 108) puts it, "they need to compose, not pull to pieces". Furthermore, Anglophones, by and large, take it for granted that their encoding competence is adequate and use dictionaries almost exclusively for decoding. Foreign students, by contrast, use their dictionaries much more for encoding than native speakers do (Whitcut 1986: 117). For one thing, unlike the latter, they cannot be assumed to have internalized much of the grammatical system of the language they speak (Cowie 1983a: 105). In addition, native speakers have a better command of the semantic rules and restrictions governing the use of individual words, so they know intuitively, e.g., whether a verb combines with an animate or an inanimate object (McCorduck 1993: 39). That is why learners' dictionaries offer more detailed and explicit treatment of syntactic properties of words than those intended only for native speakers. (2) McCorduck (1993: 83) points out that syntactic information about a word is for learners often the most essential information in their production of the second language. Patterns of usage are a case in point. They include verb patterns which show the learner how to use verbs to form correct sentences. For Palmer (1949: 276) they amount to a panacea for learners' errors. He holds that "[e]xcept by guess-work and chance the student of a foreign language cannot use a verb correctly in a sentence without knowing to what pattern or patterns it belongs."

The significance of the information on verb syntax in learners' dictionaries may be ascribed also to the fact that the verb is the sine qua non of a sentence. This statement is the cornerstone of valency theory. (3) In the light of this theory valency is the capacity a verb has for combining with particular patterns of other sentence constituents and the centrality of the verb consists in the fact that the kind of verb selected for a sentence determines its basic structure (Allerton 1982: 1, 2). The verb is therefore the item which governs the rest of the sentence because its complementation features shape the syntactic structure of any sentence in which it occurs.

Hornby (1965: 110) asserts that "[o]ne of the most useful helps that may be given in a dictionary for foreign learners of English is guidance on verb patterns." Unfortunately, the word class label, which provides basic information on the syntactic operation of an item by indicating the place or places it may occupy in the syntactic structure (Jackson 1985: 54), does not provide sufficient syntactic information. Although in the case of the verb the word class label implies, in keeping with valency theory, that the verb is the focal point of the sentence, the question of complementation is not touched upon at all. Labeling verbs as transitive or intransitive, or, in other words, the traditional specification of the subclasses, does not supply enough information to enable the learner to build acceptable clauses, either. The information on the number of complements, which transitivity in fact boils down to, is far too little. What the learner needs is, in Jackson's (1985: 56) words, "the information ... how to use that item Indeed within this information on use must be a specification of the syntactic operation of the lexical item, how it fits as an individual item into the general syntactic patterns of the language." Idiosyncratic information about the verb is thus indispensable. In this regard, the learners' dictionary should specify which clause patterns a verb may enter, what complementation it may take, which items are obligatory, optional or deletable (Jackson 1985: 56). Subjects and objects with which the verb is routinely used should also be identified (Orszagh 1969: 219). The foregoing idiosyncratic facts about each verb may be subsumed under Krishnamurthy's (1993: 68) blanket term "the features of the syntactic environment of a verb".

It should be noted, however, that scrutiny of the syntactic behavior of verbs may often lead to a fineness of detail which may turn out to be impractical and perhaps unnecessary. Some complementation patterns may in fact not be very common or may be restricted in their occurrence to certain levels of style. Not only would their inclusion in the learners' dictionary require an inordinate amount of space, but, what is more, it might be more misleading than useful to the learner (Herbst 1984: 6). Typicality should thus be made the guiding principle underlying the choice of patterns to be included in the dictionary. As Hanks (1993: 121) points out, the foreign learner struggling to encode English naturally needs guidance not so much on what is possible as on what is typical. However, even when the choice of syntactic patterns is restricted to the most typical ones, lexicographers still have to economize on space to avoid making their dictionaries unwieldy.

Cowie (1984: 156) notes that "extreme economy" is the distinct merit of encoded syntactic information. Moreover, even though codes take up as little space as possible, they convey a lot of information on the complementation properties of verbs. They usually consist of one or a few letters, digits or abbreviations, whose significance is spelled out and exemplified in explanatory tables or keys outside the alphabetical index of the dictionary. McCorduck (1993: 17) remarks, however, that the economy and the exhaustiveness obtained through the use of codes are unfortunately vitiated by their usually abstract nature and the complexity of the explanations accompanying them. It is only natural, then, that a desired feature of verb codes is that they be transparent. The user should thus be able to tell at a glance what information a code represents and thereby dispense with time-consuming and irksome toing and froing between entries and illustrative tables or charts (Herbst 1996: 329).

2. Proposals for coding systems

Cowie (1984: 155) takes into consideration ways of making codes accessible to the learner. Having admitted that their often algebraic appearance might be off-putting to many students, he suggests substituting self-explanatory labels, whose mnemonic value encourages quick memorization of patterns, for arbitrary alpha-numeric symbols. Attractive though it is, such a policy is of limited applicability, since transparency of verb codes is, in Cowie's (1984: 156) opinion, an unacceptable trade-off for valuable space.

The shape of the coding system in the pedagogical dictionary is of primary concern to Lemmens and Wekker (1986: 13, 99-100), who set out minimal conditions for a new coding system. They argue for grammatical codes which, although unambiguous and easy to use, would provide the dictionary user with useful and straightforward information about the grammatical function and the grammatical category of the elements which may follow the verb as its complements. Every verb entry, or, if appropriate, every sub-entry, should, in their view, include one or more codes, all of which would be self-explanatory. They emphasize that codes should be such that the dictionary user should have all the relevant information at hand and thus find it unnecessary to consult the explanatory section. Still, they admit that it would remain necessary to provide a guide in the introductory pages to explain elementary function and category labels to the absolute beginner. However, to facilitate comprehension without recourse to the guide, c odes must be as explicit and complete as possible. In other words, they should not be collapsed and no part of the code should be put in brackets (Lemmens and Wekker 1986: 13-14).

The question of coded syntactic information is also addressed by Aarts (1991). With a view to making verb codes simple and accessible to the learner, he defines conditions that the system of verb codes should meet, namely: the number of codes and the number of symbols reduced to a minimum, transparency of symbols and the use of category symbols rather than those denoting sentence functions (Aarts 1991: 577). Besides, he maintains that when the symbol V is followed by n it should be assumed that a passive construction is possible, unless the label stating otherwise is provided. Importantly, only one symbol for the verb, i.e. V, is deemed necessary (Aarts 1999: 31).

It is noteworthy that the above proposal does not chime in with that of Lemmens and Wekker. Unlike Aarts, Lemmens and Wekker do not impose any constraints on the number of codes. What is more, their minimal conditions imply that accuracy and exhaustiveness of syntactic information take priority over the number of codes. Moreover, whereas Lemmens and Wekker argue for transparent codes, even at the expense of valuable space, Aarts does not agree to sacrifice space for clarity. He implies that codes should be simple and short, but they must not become opaque. Besides, Aarts (1991: 580) explains that the suggested use of category symbols in codes has important consequences, since it means that the same code can be given to verbs that are in fact different from a syntactic point of view and, for instance, take similar, though not exactly the same, complementation patterns. He is of the opinion that codes should be simple and accessible to linguistically naive learners rather than complicated enough to convey infin itesimal syntactic differences (Aarts 1991: 581). After all, the place to discuss syntactic issues in detail is, in his view, in the grammar, not in the dictionary. Conversely, Lemmens and Wekker (1991: 231) assert that the use of both formal and functional categories in verb codes makes description more accurate without impinging on the accessibility of encoded syntactic information. They claim that learners may be assumed to be familiar with these categories since, as a rule, they are also users of well-known grammars of English (Lemmens and Wekker 1986: 9). In view of the fact that the classification of words into parts of speech and the division of sentences and clauses into elements are common to mainstream comprehensive English grammars, they see no reason to dispense with either formal or functional categories in the dictionary. (4)

3. Verb codes in learners' dictionaries -- a historical perspective

The use of grammatical codes, concomitant with the prominent trend in pedagogical lexicography towards the provision of detailed and varied information on language production, has become an integral feature of the presentation of verb syntax in the vast majority of pedagogical dictionaries. Still, there have been dictionaries which purported to meet the needs of foreign learners but contained almost no grammatical information or even less than is typically found in dictionaries for native speakers.

West and Endicott's The new method English dictionary (1935 [henceforth NMED]), claimed to be the very first learners' dictionary (Cowie 1999: 3; Summers quoted in Stark 1999: 29), is a case in point. The preface to NMED makes it clear that this dictionary is written especially for the foreigner and, as an entirely original feature, defines words by means of a controlled vocabulary (West and Endicott 1935 [1965]: iii, iv). The question of verb syntax is, however, nonexistent there. Not only does NMED provide no information on complementation of verbs, but it also does not even give part of speech labels for headwords. Neither are they given in its revised edition, published in 1965 under the title An international reader's dictionary (henceforward IRD). It was only in the second, 1977, edition of IRD that labels were "added to tell the user which part of speech is being defined" (West and Endicott 1977: v). What is more, a distinction between transitive and intransitive verbs was introduced and marked accordi ngly. Nonetheless, "if a verb is neither intransitive nor transitive, or if ... it can be both transitive and intransitive, it has been marked simply v" (West and Endicott 1977: vii).

With the exception of NMED, the learners' dictionaries of the 1930s and 1940s are especially noteworthy as aids to encoding. Two more learners' dictionaries appeared in that period: Palmer's Grammar of English words in 1938 (henceforth GEW), followed in 1942 by OALDCE1 by Hornby, Gatenby and Wakefield. Information on verb syntax featured prominently in both of them. This should not come as a surprise, given that Palmer and Homby were the driving force behind the research into verb patterns, which proceeded in parallel with work on vocabulary control (Cowie 1999: 6). A detailed subclassification in the sets of verb patterns in the aforementioned dictionaries was arrived at by taking account not of transitivity in general, but of the kinds and order of grammatical structures following the verb. It is this classification scheme that Hornby adopted as the principal means of incorporating grammar in subsequent editions of OALDCE. Moreover, both Palmer and Hornby indicated in their dictionaries the syntactic functi ons of verbs in each of their different senses. Codes showing in which pattern a verb may occur were first employed in Palmer's GEW. Still, they are essentially the same as in the first two editions of OALDCE in that they take the form of alphanumeric references to verb patterns explained at length in other parts of the dictionary. Hornby (1965: 110) himself admits that the references to verb patterns that he employed in OALDCEl and OALDCE2, elaborated so as to provide as much help as practicable in limited space, are "not an entirely new departure in lexicography". It is necessary, therefore, to do justice to Palmer and his groundbreaking method of presenting verb syntax in the pedagogical dictionary.

The learners' dictionaries published afterward have also made a point of their systems of coding syntactic information on the verb. Of all the general-purpose learners' dictionaries, only the Chambers universal learners' dictionary from 1980 does not use any system of syntactic coding. Verbs are just classified there as transitive and intansitive and labeled vt and vi respectively (Huang 1985: 60; Stark 1999: 29). Even though the other learners' dictionaries published in the last three decades embody the same principles of encoding and explanatory key as GEW, OALDCE1 and OALDCE2, they have developed fresh systems of symbols to encode syntactic features. (5)

4. Coding system in OALDCE3

Before embarking on an analysis of symbols for the verb and its complements in the selected pedagogical dictionaries it is necessary to explain why codes used in one of them, that is OLADCE3, are not amenable to being split into their constituent parts and scrutinized likewise. Verb codes in OALDCE3 tell the user nothing about individual patterns themselves and in that regard they fail to fulfill the chief function assigned to them, i.e. to indicate complementation structures typifying the verb (D'Elia 1992: 233). In fact, they only reflect the ordering of patterns in the total scheme. A code in OALDCE3, which consists of the capitals VP followed by a number, links an individual dictionary entry to the appropriate section of explanatory tables. Even though the main codes range from [VP1] to [VP25], as is the case in OALDCE2, some of them are further subdivided. Capital letters other than V and P are used to indicate such subdivisions, e.g., [VP18A], [VP18B], [VP18C]. As a result, the total number of distinctl y labeled verb patterns amounts to fifty-one, which means that it is double what it was in the previous edition. The division of [VP6] into [VP6A] and [VP6B], for instance, is made on the grounds that some transitive verbs with nominal objects allow passivization while others do not (OALDCE3: xxxii). Such creation of small subpatterns within a given verb pattern on the basis of transformational differences between subclasses of verbs adds to the complexity of the system. The solution where one code is given to both passivizable verbs and nonpassivizable ones, the latter being accompanied by the label no passive, appears to be more straightforward.

Verb patterns in OALDCE3 are not defined solely on the basis of grammatical criteria. Two criteria are in fact used, that of form and that of meaning, which makes the system of codes unnecessarily complicated. There seems to be no need to make semantic subclassifications in the system of verb patterns since semantics is dealt with in the explanatory sections of verb entries. Principles governing the choice and ordering of verb patterns should thus be strictly syntactic. An example of such redundant semantic subclassification can be found in the pattern [VPl8], which is subdivided as follows: "[VP18A]. In this pattern the verb is used with a noun or pronoun and a bare infinitive. The verb indicates physical perceptions... [VP 18B]. A small number of verbs which do not indicate physical perceptions are used in this pattern. Make and let are examples ... [VP18C]. Have is used in this pattern when it means wish, experience or cause." (OALDCE3: xxxvi). There is no need to distinguish syntactically between these th ree patterns and code them separately. They might just as well be described as consisting of the verb followed by a noun phrase or a pronoun and a bare infinitive. Similarly, the only (semantic) difference between [VP19A] and [VP19B] is that the verb in the construction subject-verb-direct object-adjective/noun indicates physical perception in [VP19B] but not in [VP19A] (OALDCE3: xxxvii). However, this property of the verb does not have a bearing on the structure of the pattern in which the verb occurs. It is clear, then, that the use of semantic criteria paves the way for an unnecessarily large number of codes.

The form of codes in OALDCE3 leaves a lot to be desired. As they are in no way self-explanatory, it is unfeasible to work out what they mean just by looking at them. The meaning of each code must always be looked up in the introduction or inside the dust jacket of the dictionary. The principle behind the verb patterns in OALDCE3 is that "the learner is not expected to memorize these verb patterns. They are a simple reference system, a practical tool to guide the learner who wants to form correct sentences" (OALDCE3: xxxiii). Nonetheless, becoming familiar with this system of codes calls for exceptional dedication on the part of the learner. The system is neither transparent nor mnemonically organized, so that no resemblance between patterns such as want to do sth and want sb to do sth can be deduced from the codes (Herbst 1996: 329).

Hornby (quoted in Cowie 1990: 341-342) deemed it necessary to indicate both form and syntactic function of each postverbal constituent in his system of verb patterns. It is therefore two kinds of information about the same item in the complementation structure rather than different types of label attached to different adjacent elements that should be given as long as the treatment of verb complementation is to take account of function as well as form. However, the actual description of codes in OALDCE3 falls short of this target since verb complementation is described in terms of either parts of speech or grammatical functions, or a combination of these, as shown in the respective examples corresponding to the description headings above verb pattern tables in OALDCE3:

(1) [VP2D] subject + vi + adjective/noun/pronoun,

(2) [VP12B] subject + vt+ IO + DO,

(3) [VP23] subject + vt + DO + noun,

where IO and DO stand for indirect and direct objects respectively (OALDCE3: xxx, xxxiv, xxxviii). Labels for sentence elements are absent from descriptions of as may as fourteen verb codes. (6) It is clear then that OALDCE3 mixes categorial and functional information instead of providing, in tune with Hornby's original intention, both types of information separately and consistently. The descriptive and pedagogical shortcomings of codes employed in OALDCE3 thus give substance to Herbst's (1996: 354) remark that these codes "were not perhaps designed primarily with a lexicographic purpose in mind".

5. Symbols for the verb in the dictionaries published after 1974

In the subsequent pedagogical dictionaries codes are not reduced to opaque cross-references to explanatory sections, but they are designed to convey information on complementation requirements of the verb. To serve this purpose, symbols representing the verb itself and its complements have replaced alphanumeric references to the places verb patterns occupy in explanatory tables. The symbols for the verb in the learners' dictionaries presented in Table 1 are discussed below.

Table 1

Symbols for the verb

Verb class\Dictionary                    OALDCE                LDOCE

                             4         5         6         1
Intransitive                 I         V         V         I
Copular/Linking              L         V         V         L

Monotransitive               T         V         V         T
Ditransitive                 D         V         V         D
Complex transitive           C         V         V         X
Other                                                      V
Ergative
Ergative linking
Ergative reciptrocal
Reciprocal

Number of different symbols  5         1         6         3

Verb class\Dictionary               LDOCE                  COBUILD

                             2         3         1         2
Intransitive                 I         I         V         V
Copular/Linking              L         linking   V         V-link
                                       verb
Monotransitive               T         T         V         V
Ditransitive                 T         T         V         V
Complex transitive           T         T         V         V
Other
Ergative                                         V-ERG     V-ERG
Ergative linking                                           V-LINK-ERG
Ergative reciptrocal                                       V-RECIP-ERG
Reciprocal                                       RECIP     V-RECIP

Number of different symbols  3         3         3         6

Verb class\Dictionary        CIDE


Intransitive                  I
Copular/Linking               L

Monotransitive                T
Ditransitive                  T
Complex transitive            T
Other
Ergative
Ergative linking
Ergative reciptrocal
Reciprocal

Number of different symbols   3

As Table 1 shows, LDOCE1 and COBUILD2 employ the largest number of symbols in order to refer to the verb. Whereas OALDCE4 uses one symbol fewer than the aforementioned dictionaries, there is only one symbol for the verb in the next two editions of OALDCE. COBUILD1 and COBUILD2 are the only dictionaries that have symbols for ergative and reciprocal verbs. Ergative linking verbs and ergative reciprocal verbs are marked accordingly in COBUILD2 alone.

Jackson (1988: 184) defines the term ergative as the alteration between transitive and intransitive patterns in which the object of the transitive pattern appears as the subject of the intransitive one, as in

(4) Mary opened the door,

(5) The door opened.

There seems to be no need to design a code for ergative verbs provided that examples showing how they can be used are supplied. The word ergative is most probably not even part of the passive vocabulary of many Anglophones, let alone that of foreign learners. It is worth noting that in the remaining dictionaries the contrast between (4) and (5) is regarded as involving separate intransitive and transitive uses of the same verb, the difference between which is reflected in distinct codes and in examples which help to clarify the context.

Reciprocal verbs allow the interchangeable use of the subject and the object without altering the meaning of the verb, as in

(6) Mary met John,

(7) John met Mary. (COBUILD1: 1201).

On top of that, the two noun groups can be linked by the conjunction and and function as the subject of the verb. In that case the verb does not take a new object, except for the optional phrase each other used to emphasize the reciprocal meaning, for instance:

(8) John and Mary met (each other). (COBUILD1: 1201).

Although the code for reciprocal verbs is explained in the dictionaries in question, students may still have difficulty interpreting it. The practice followed by all the other dictionaries with regard to reciprocal verbs, which in fact corresponds to that adopted in the case of ergatives, no doubt spares the learner such problems. The labels used in COBUILD2 for ergative linking verbs and for ergative reciprocal verbs are even more likely to remain abstruse to learners. Unfortunately, explanations of these symbols in the grammar section in this dictionary are far from illuminating. The student learns that an ergative reciprocal verb behaves both like an ergative and a reciprocal verb. The term ergative linking verb remains even more arcane, since the learner finds out that the verb it denotes sometimes behaves like a linking verb, but it also occurs in non-linking patterns, which makes it similar to an ergative verb (COBUILD2: xxix). It is clear, therefore, that the assertion of the editors of COBUILD2 (vii) that the abbreviations used in this dictionary are nearly all familiar to any learner is nothing short of an overstatement. The concern about the proliferation and opacity of codes in other dictionaries designed for foreigners, expressed in COBUILD1 (vii) in the words "[we did not want] to have hundreds of grammar codes, as some dictionaries have, which can only be understood by looking up another part of the book", does not thus herald transparency and a small number of verb symbols either in COBUILD1 or in its successor.

Judging by the matching number of verb symbols, COBUILD2 appears to find itself where LDOCE1 stood two decades earlier. In LDOCE1, however, unlike in COBUILD2, each of the major categories of verbs is given a different capital letter symbol. The basic categorization of verbs was modeled on A communicative grammar of English by Leech and Svartvik (Procter 1976: 315).

The use of letters for recognizable descriptive labels is to make the coding system easy to remember. Moreover, such verb symbols are to tell the user whether a verb can have an object (LDOCE1: xxviii). Hence I for intransitive verbs, L for verbs which link subject and its complement, D for ditransitive verbs, i.e. the verbs which take direct and indirect objects. Unfortunately, not all verb symbols in LDOCE1 have mnemonic value. X, which labels the complex transitive verb, i.e. the verb of which the object has the complement realized by a noun phrase, an adjective phrase or an adjunct (LDOCE1: xxxiii), is a case in point. Similarly, the symbol V, associated with verbs in general, is in LDOCE1 used to denote a specific category thereof, i.e. verbs which take "a 2-part DIRECT OBJECT. The first part is a nounlike expression, and the second is an infinitive with or without to, an -ing form or a past participle" (LDOCE1: xxxi). (7) In view of the fact that the symbol V does not stand for a corresponding recogniza ble descriptive label that could be attached to a category of verbs, it has been specified as other in Table 1. The analysis of the symbols for the verb in LDOCE1 shows then that even though some of them are mnemonic, the system as a whole is still a far cry from one which is "easily remembered and requires no knowledge of grammatical theory to be fully understood", as is asserted in the introduction (LDOCE1: vii-viii).

A similar set of symbols characterizes OALDCE4. In fact, it is only the symbol C, labeling complex transitive verbs used in place of X, and the absence of V that differentiate the sets of verb symbols in these two dictionaries.

In contrast to LDOCE1 and OALDCE4, the next two editions of LDOCE as well as CIDE do not have special symbols for the classes of transitive verbs. Instead, the symbol T is used for all the verbs that cannot occur on their own. Monotransitive, ditransitive and complex transitive verbs are thus all given the same symbol even though they have widely different complementation patterns. The symbol for intransitive verbs has been retained in all the dictionaries under discussion, whereas the symbol L has been replaced by a descriptive label only in LDOCE3. It is therefore only the distinction between the three major verb classes that verb symbols in these dictionaries indicate.

The categories transitive and intransitive are not distinguished in OALDCE5, OALDCE6 or in either edition of COBUILD, where only the symbol V is used for both these classes of verbs. With the exception of COBUILD2, the same symbol is also used for copular verbs. The use of one symbol for the verb should be seen as advantageous since there is evidence, referred to by McCorduck (1993: 89), that while learners, for the most part, have difficulty remembering the significance of the term transitive, they grasp the concept of verb plus object with ease. Besides, as Aarts (1991: 572) emphasizes, what students need to know about a verb is whether it can be used on its own, and, if not, by how many and what type of elements it must or can be followed. It turns out that one symbol for the verb is sufficient to answer these questions. In the four dictionaries under discussion the symbol V by itself means that the verb does not take an object. However, if the verb does require complementation, V is followed by additional symbols or labels that automatically assign a category value to the verb (OALDCE5: B4; OALDCE6: B6). (8) In view of the fact that, as Aarts (1999: 23) points out, students are not only ignorant of what symbols such as I, L, T, D and C mean but, worse yet, they do not bother to look them up, restricting the number of verb symbols to just one appears to be a better solution. In this way dictionaries dispense with reference to the five major verb classes which presupposes a rather advanced knowledge of verb categories.

The last two editions of OALDCE have the simplest system of labeling verbs. Since the number of symbols for the verb is reduced to just one, these dictionaries eliminate the codes which are not transparent to grammatically uninitiated users.

6. Symbols used with the verb symbol

Symbols for the verb do not provide adequate information on the complementation structure of the verb. In the case of T or V it is only additional symbols accompanying them that reveal the specific complementation requirements of a given verb. Even the less general verb codes are not specific enough. The symbol C, for instance, while indicating that object complementation must follow the object, tells the learner nothing about the grammatical categories which may perform this function. In Table 2 are shown the symbols which the selected learners' dictionaries employ in their verb codes with a view to furnishing more details on syntactic properties of the verb than the verb symbol alone may supply.

Table 2

Symbols accompanying verb symbols in codes

       OALDCE                     LDOCE                 COBUILD

4            5       6           1         2       3            1
a           adj     ADJ          *        adj     adj           A
f           adv     ADV       1a, 1b      adv     adv          ADJ
g         inf (no   inf          2         n      pre          ADV
            to)                                    p
i           ing     ing          3        obj                   C

n            n       N        4a, 4b    obj (i)                INF
p            p      prep      5a, 5b,     obj                  ING
                                5c        (d)
pr          pr      that      6a, 6b     prep                  NG
s          that      to          7       that                   O
                    inf
t         to inf     wh          8       to-v                 PAST
                                                              PART
w           wh                   9       to-v                 PREP
                                         v-ed                REPORT-
                                                               CL
                                         v-ing               to-INF
                                          wh
10          10       9          15        13       3           12

            COBUILD         CIDE

4             2
a            adj           adj
f            adv           adv
g             ed         adv [M]

i            inf       infinitive
                       without to
n            ing            n
p             n            obj

pr            p           prep
S            prep      that-clause

t            that     to-infinitive

w           to inf    [two objects]
              wh          v-ed

          whether/i       v-ing
                         wh-word
10            12           13

As can be seen in Table 2, LDOCE1 employs the largest number of symbols to accompany the verb symbols in codes. It contrasts sharply with LDOCE3, where the number of such supplemental symbols has been reduced to just three, so that it constitutes but one fifth of what it was in 1978. The remaining dictionaries do not differ so significantly in this respect since they have from thirteen to nine symbols. It also transpires that the use of only one symbol for the verb does not entail the largest number of additional symbols in codes. In OALDCE4, for instance, where verb complementation is indicated through a classification scheme built around the five main verb types, there are as many symbols which signify further discriminations as in OALDCE5, where only one verb symbol is used.

6.1. Symbols in the Oxford dictionaries

In OALDCE4 a set of one-letter abbreviations represents various phrase and subordinate clause types, e.g., a -- adjective phrase, n -- noun phrase, t -- to infinitive clause (Cowie 1990: 344-345). It should be noted that OALDCE4 employs symbols representing formal categories only, rather than ones denoting sentence functions. (9) Some of them, e.g., f for that-clause, are infelicitous. By and large, however, the coding system in OALDCE4 should be viewed as a major improvement on that in OALDCE3. The symbols in this dictionary, in contrast to those used by its predecessor, have, in the main, mnemonic value. Once learners have understood what they mean, they are not likely to forget them quickly and hence there is little need for them to consult explanatory sections of the dictionary.

The symbols used in OALDCE5 are more transparent than those in OALDCE4. The use of abbreviations consisting of the first two letters instead of single lower case letters, e.g., adj instead of a, undoubtedly enhances transparency. The symbol f has likewise been replaced by the more straightforward that. Like in OALDCE4, no symbols for functional categories are used.

With the exception of the symbol p standing for pronoun in OALDCE5 but absent from OALDCE6, and for some minor typographical changes, there is hardly any difference between these two dictionaries with respect to the symbols used in verb codes.

6.2. Symbols in the Longman dictionaries

In LDOCE1, unlike in OALDCE4-6, numbers and letters accompany verb symbols in codes. The function of the former is to "give information about the way the rest of a phrase or clause is made up in relation to the word being described" (LDOCE1: xxxii). In crude terms, numbers indicate what may follow the verb and have the same meaning irrespective of the verb symbol which they follow, namely: [empty set] -- no complement or object, 1 -- one or two noun or pronoun objects or complements, 2 -- a bare infinitive, 3 -- a to-infinitive, 4 -- an -ing form, 5 -- a that-clause, 6 -- a clause or a phrase introduced by a wh-word, 7 -- an adjectival complement or a noun object followed by an adjectival complement, 8 -- an -ed form, 9 -- an obligatory adjunct, usually a phrase used adverbially (LDOCE1: xxxiii-xxxiv). It is the consistent use of numerals which always have the same meaning wherever they appear that Procter (1976: 316) considers the greatest advantage of this method of encoding syntactic information. Unfortuna tely, interpretation of lower case letters is more problematic. The symbol [a], for instance, can be interpreted in three different ways depending on the combination in which it occurs. In [1a] and [4a] it means that the particle in a compound verb is not movable and that it invariably follows the verb. In [5a] it shows that the word that is optional in the following that-clause. In [6a] it indicates that a wh-word introduces a finite clause, rather than one including the full infinitive, in which case the code [6b] is required. Lower case letters thus make the syntactic description of the verb much more elaborate, even though only the combinations included in Table 2 are possible. All in all, the symbols in LDOCE1 are not immediately comprehensible to the user. Besides, in the light of the above analysis it is by no means obvious that they can be assimilated with ease. Reference to both formal and functional levels in the description of verb complementation adds to the complexity of the system. No wonder, th en, that Hausmann and Gorbahn (1989: 50) find the difficult-to-read grammatical symbols in LDOCE1 "the innovation which started out on the wrong foot", an obvious deficiency of a newly conceived work. Still, Ellegard (1978: 235) points out that the system of symbols in LDOCE1 is superior to that of OALDCE3 since it has more structure, and therefore less redundancy, than the latter. Unfortunately, however, both of them are opaque.

Greater transparency of grammar codes was one of the aims of the revision of LDOCE1. While the sophisticated grammar codes in this dictionary were allegedly well-received by those particularly interested in grammar, many learners found them difficult to remember (LDOCE2: F9). As Table 2 shows, symbols used in verb codes in LDOCE2 are much more transparent than those in the previous edition as a result of the substitution of immediately interpretable symbols for abstract ones. Since abbreviations are employed, and not digits, it is only natural that the new system is more space-consuming. Besides, the use of obj(i) and obj(d) for indirect and direct objects respectively, and n/ad] rather than comp for complement, bears testimony to the confusion of the levels of form and function.

LDOCE3, where only three symbols are used with the symbols for the verb, employs the most space-consuming method of specifying complementation patterns of the verb. This dictionary has gone a step further than its predecessors and replaced symbols with pattern illustrations. (10) The use of pattern illustrations, or collocations, instead of symbols has the most obvious advantage of not requiring the user to know any grammatical terminology at all. On the whole, it also meshes nicely with the general phrase-orientation of LDOCE3 (Herbst 1996: 329). Still, Aarts (1999: 27), for instance, remains skeptical about the merits of this innovation and wonders whether the simplification of the code system in LDOCE3 has not been too radical. He rightly concludes that "[t]he editors probably believe that phrases and collocations can do the same job as codes. This is a question which deserves further research" (Aarts 1999: 28). Nonetheless, pattern illustrations in LDOCE3 confirm Nesi's (2000: 74) claim that an economical alternative to grammar coding has not yet been found.

6.3. Symbols in the COBUILD dictionaries

The system of symbols used in COBUILD 1 bears a resemblance to that in the concurrently published LDOCE2 inasmuch as the symbols in both dictionaries consist, for the most part, of a few letters and represent both formal and functional categories. The symbol A, which stands for adjunct, is not immediately comprehensible. It seems that the term adjunct itself may presuppose too much familiarity with linguistic theory to be useful to many learners. The other symbols for functional categories used in this dictionary, C for complement and O for object, are not explained in separate entries. The user wishing to find out what they denote is referred to the boxed entries dealing with compound codes such as V+A, V+C, V+O+A, V+O+C. The symbol REPORT-CL, unique to this dictionary, is also worthy of comment. It indicates that the verb may be followed by reported clauses of any type (COBUILD1: 1227). It appears that symbols showing exactly which category of reported clauses may be used with a given verb are more informat ive.

Functional categories are absent form COBUILD2. The label REPORT-CL has also been dropped and the symbol PAST PART has been replaced by the more straightforward -ed. All in all, symbols in COBUILD2 hardly differ at all from those used in the last two editions of OALDCE. The consistent description of possible syntactic constructions in terms of formal categories only can be seen as an improvement on COBUILD 1.

6.4. Symbols in CIDE

CIDE uses the function label obj. In the case of transitive verbs, it follows the headword no matter whether the verb in question is monotransitive, ditransitive or complex transitive (CIDE: 1616). What is more, it does not supply any information that is not also provided by the symbol [T]. There seems to be no reason, therefore, why [T] should be used in view of the fact that the label obj is attached to all transitive verbs anyway and, on top of that, makes it possible to show the distinction between obligatory and optional objects, the latter being indicated as (obj). Oddly enough, the symbol in question also has a different meaning. When used in complex codes following examples, it denotes a sentence constituent (CIDE: 1616). Moreover, the symbol [two objects], unique to CIDE, shows that both a direct and an indirect object follow the verb (CIDE: 1575). Apart from these symbols indicating sentence functions, CIDE uses category symbols. The labels [infinitive without to] and [to infinitive] appear to be un necessarily long in comparison with other symbols employed there, especially in view of the fact that respective abbreviations in other dictionaries are no less transparent. (11)

The above analysis of symbols used in verb codes in the selected dictionaries reflects attempts to make verb codes less abbreviated and impenetrable, hence more accessible to the user. It transpires also that the sets of symbols used in LDOCE2 and COBUILD1 are quite similar to each other. The symbols in the last two editions of OALDCE and COBUILD2 are virtually the same, with CIDE still differing from the three, and LDOCE3 pursuing the most distinctive path by using pattern illustrations.

6.5. Verb codes and underlying syntactic differences between similar patterns

The analysis of verb codes makes it possible to answer the question whether verb codes in the dictionaries under discussion account for underlying syntactic differences. (12) With this end in view, attention will be paid to three constructions, namely: want sb to do sth, advise sb to do stir and know sb to be a liar; illustrated by the following examples:

(9) I want her to answer the question,

(10) I advise you to answer that question,

(11) I know him to be a liar,

and labeled by Quirk et al. (1985: 1176, 1203, 1215) as monotransitive, ditransitive and complex transitive respectively. Though they are similar on the surface, the complementation patterns following want, advise, and know are underlyingly different. The following syntactic tests bring out the differences:

(12) *She is wanted to answer the question.

(13) You are advised to answer the question.

(14) He is known to be a liar.

(15) *I want her that.

(16) I advise you that.

(17) *I know him that.

(18) I want that question to be answered by her.

(19) *I advise that question to be answered by you.

(20) *I know that liar to be him. (13)

Without going into detail, it is clear that each of the structures allows a different set of transformations, which brings to light their underlyingly different nature. However, it is only OALDCE4 that takes account of these distinctions and employs three different codes for the patterns in question. In LDOCE1 and CIDE two codes are used, whereas in the remaining dictionaries only one code is given. (14) The tendency to provide only one code in the cases where, on the basis of transformational constraints, it is possible to draw a distinction between superficially similar constructions is commendable since, according to Aarts (1991: 576), the learner, as a rule, wants information about surface structure possibilities only. The coding system accounting for the deep structure may therefore be judged unnecessarily complicated for the average, not grammar-hungry dictionary user.

7. Types of coding system

In the light of the above analysis it is advisable to divide the coding systems in the selected dictionaries into two groups, according to the categories of linguistic description represented by the symbols accompanying verb symbols in codes. The results are presented in tabular form below. The labels NFor and NFun stand for the number of formal symbols and the number of functional symbols respectively.

Table 3

Coding systems according to the categories of symbols

                       Symbols in verb codes
Formal                   Formal and Functional
Dictionary                    Dictionary        NFor      NFun

OALDCE4                         LDOCE1           4         11
OALDCE5                         LDOCE2           3         10
OALDCE6                        COBUILD1          3         9
COBUILD2                         CIDE            2         11

In none of the dictionaries do codes denote sentence functions only. In fact, the structure of verb complementation is described in terms of either word classes, i.e. formal categories alone, or both formal and functional ones, the latter always being in the minority. (15) This conclusion substantiates Herbst's (1999: 233) claim that general learners' dictionaries tend to focus on the formal aspect of verb complementation. The tendency should be regarded as positive, since, as Heath and Herbst (1988: 316) observe, functional categories presuppose, as a rule, more familiarity with grammar than can be expected of many learners. The use of word classes in the description of verb complementation, in turn, results, according to Bogaards (1996: 305), in such great transparency of encoded syntactic information that learners can be expected to use the verb correctly without even being forced to read examples.

Interestingly, Ilson (1989: 1975-1976) equates coding systems with semiotic ones and argues that they may be analyzed in the same way as the languages they describe. In fact, he distinguishes three classes of coding systems and compares them to words, phrases and definitions respectively. The codes in OALDCE3 are, in his view, units that behave like single words and, as such, are in stark contrast to those in COBUILD1, which are similar to phrases whose elements are meaningful and can occur in other combinations. The codes [VP17] and [V+O+to-INF] are given as examples of the two polar opposites. In between are codes which, like the Aristotelian definition, consist of just two major elements. Ilson (1989: 1976) compares an OALDCE1 code to an analytical dictionary definition and calls the capital letter representing the verb the genus and the number following it the differentia, both of which are meaningful and independently recombinable, as is the case in phrase-like codes.

By way of recapitulation of the results of the analyses of symbols used in codes, the types of coding system in the consulted pedagogical dictionaries are presented in Table 4.

Table 4

Types of coding system

Neither transparent  Mnemonic  Transparent  Pattern illustration
   nor mnemonic

      OALDCE3         LDOCE1     LDOCE2            LDOCE3
                     OALDCE4    COBUILD1
                                COBUILD2
                                 OALDCE5
                                 OALDCE6
                                  CIDE

In OALDCE3, where patterns are just numbered from [VP] to [VP25], neither the structure of a pattern nor any resemblance between similar syntactic constructions can be seen from the codes. In LDOCE1 and OALDCE4, by contrast, the coding systems are mnemonic in that similarities between patterns can be seen from a code, even though the meaning of the code is not immediately obvious to the user. This is not the case in totally transparent systems, which occur in the majority of pedagogical dictionaries. It is these systems that are by far the best since it is apparent at a glance what sort of information is being conveyed. It is not necessary, therefore, to look up a code in the explanatory sections of the dictionary. Finally, LDOCE3 uses pattern illustrations rather than codes.

8. Utility of verb codes

Pedagogical lexicography has been strengthening its reputation for the user-centered innovation since the mid 1970s (Cowie 1984: 163). This is the period which has also seen a growth in the use of coding systems in learners' dictionaries. Unfortunately, Quirk's wry comment (quoted in Crystal 1986: 78) to the effect that some of the dictionary features which appear to be central to the lexicographer are decidedly peripheral to the ordinary dictionary user holds true for verb codes in the pedagogical dictionaries published in the 1970s.

Several research projects which laid the groundwork for the compilation of LDOCE2 show that students find the mnemonic codes in LDOCE1 off-putting and impenetrable (LDOCE2: F8). Not only are learners unable to recall the meaning of verb codes, but, worse still, they are often ignorant of the fact that they convey syntactic information on verbs. In a survey conducted among 160 students at the universities of Augsburg and Erlangen-Nurnberg only 70 were aware of the fact that OALDCE3, which they used at school, supplied syntactic information in the form of verb codes (Herbst and Stein 1987: 120). Herbst (quoted in McCorduck 1993: 22) reports that many of the German students of English that he surveyed did not realize that the codes in LDOCE1 and OALDCE3 concerned verb syntax. These findings should not come as a surprise in view of the fact that many teachers, not to mention learners, brush aside grammar codes. West's study (quoted in Nesi 2000: 73) revealed that "[t]eachers felt that both systems [i.e. those in OALDCE3 and LDOCE1] presupposed a linguistic sophistication which most students did not possess and no one claimed that they had successfully trained their classes to use either system, most stating that they did not even try." Herbst (quoted in Whitcut 1986: 119) reports that a group of Bavarian teachers using OALDCE3 in the classroom had never understood Hornby's pattern [VP6A], the common transitive pattern. Aarts (1999: 16), while finding Hornby's idea of verb patterns brilliant, deems the form of his codes useless. Considering their opacity and the fact that they are but cross-references to explanatory charts in the dictionary, they are indeed inaccessible to teachers and students alike, except perhaps to those who are prepared to memorize them.

Bejoint's (1981) study is said to have broken new ground by surveying in detail the dictionary preferences of foreign learners (Hartmann 1987: 21). (16) In his survey of students' needs and reference skills he relied on the answers of over a hundred informants who were students of English at the University of Lyon in their second, third and fourth years of study. A vast majority of the students used OALDCE3 and LDOCE1; the percentage of students using monolingual dictionaries not designed for foreign users was negligible (Bejoint 1981: 214). Thus the information yielded by the study pertains, for the most part, to the two learners' dictionaries. The results highlight a considerable demand among foreign learners for grammatical information. No less than 53 percent of the informants admit they look for such information in their dictionaries (Bejoint 1981: 215). The need for grammatical information is in fact the second most important motive for consulting dictionaries, the meaning of words being the first one. It transpires, however, that learners do not appreciate and utilize the wealth of syntactic information offered. 55 percent of the subjects admit that they never use syntactic codes. The study also shows that learners who read the introductory sections in their dictionaries where the coding systems are explained are few and far between. Bejoint (1981: 216) reports that about 90 percent of the students neglect the introductory matter: one third of them do not consult it at all and more than half read it only cursorily. Still, only 10 percent of the students acknowledge some difficulty comprehending codes. This number is somewhat low when compared with the other results of the research under discussion. Bejoint (1981: 211) is of the opinion that it is students' reluctance to confess to a failure to understand that accounts for this number. Paradoxically, students appreciate guidance on syntax but fail to make the most of it. Pedagogical dictionaries are not used as fully as they should be, and many students are not even aware of the wealth of grammatical information they contain. In fact, they turn out to be no more useful to learners than dictionaries designed for native speakers since learners apparently fail to take advantage of those features of pedagogical dictionaries which are meant to facilitate language learning and use. Bejoint (1981: 220) concludes ruefully that "[i]t is for encoding that students need the most information, it is encoding information which is the most difficult to supply, and yet it is the information which students use the least." Lexicographers' disappointment is best verbalized by Whitcut (1986: 121): "[w]e are bursting to impart a whole mass of information to a public that does not seem to want to listen. It is sobering to reflect that the user is free to reject the whole thing."

To account for students' reluctance to use codes it is necessary to take note of the fact that the most impressive achievements in the treatment of syntax in pedagogical lexicography, manifested in elaborate coding systems, have coincided with a period of waning interest in the teaching of grammar (Cowie 1984: 164). The structuralist linguistic theory as well as behaviorist learning theories relying on listening, drilling, remembering and adding to the already familiar structures, which dominated language teaching in the 1950s and 1960s, were superseded by the communicative approach in the 1970s (Kaliski 1992: 98-99). The corollary of this approach, resting on the assumption that communication in the limited sense of getting the message across is in itself an adequate linguistic achievement, is not only the overemphasis on speaking, but also the neglect of grammatical correctness (Herbst and Stein 1987: 121). Learning grammar and vocabulary is believed not to be enough to communicate effectively since, as Wid dowson (quoted in Swan 1985: 4) observes, "[i]t is possible for someone to have heard a large number of sentence patterns and a large number of words which fit into them without knowing how they are put to communicative use." It is thus not so much grammatical competence as communicative competence that people share. That is why grammar explanations are considered to have been overvalued in the past and grammatical explanations are readily set aside. It is for these reasons that students who have been exposed to the more communicative approach to language instruction are, as a rule, ignorant of grammatical terminology. Students' disinclination to master the system used to codify grammatical patterning may thus be put down to their ignorance of grammar, sanctioned by the communicative approach to language teaching. Besides, the premise that a vague message is all that is necessary for the interlocutor to guess the meaning from the context obviates the need to consult a dictionary to word the message more caref ully. In that regard, the communicative emphasis in foreign language teaching not only discourages dictionary use but actually runs counter to it (Herbst and Stein 1987: 121). This false sense of security and the superficiality of their knowledge of language may thus be seen as other reasons behind students' disregard of verb codes.

The general neglect of verb codes gives rise to the lexicographer's dilemma. On the one hand, lexicographers may attempt to improve coding systems. Alternatively, they may abandon the grammatical coding scheme altogether (Lemmens and Wekker 1986: 11). As has been shown in the previous sections, the former is the course of action that lexicographers have opted for. Even though economical and almost mathematically symmetrical systems of codes prove to be quite easy to describe, the fact that learners cannot interpret such minimalized and obscure codes defeats the purpose of providing them (McCorduck 1993: 145). That is why the coding systems in consecutive pedagogical dictionaries have been made more transparent so that the meaning of the individual code is more readily intelligible and thereby requires less dependence on a separate key where codes are listed and explained.

At this point it is necessary to refer to the results of a more recent piece of research. A study conducted by Harvey and Yuill (1997) and concerned with the use of COBUILD 1 while writing shows that informants overwhelmingly locate grammatical information in examples and, to a lesser extent, in definitions (Harvey and Yuill 1997: 267). The extra column provides assistance in only about 10 percent of the successful grammatical searches. It transpires, therefore, that even though the codes in COBUILD1 are not as opaque as the alphanumeric ones in LDOCE1 and OALDCE3, students are still reluctant to use them.

By way of explanation, one might refer to Willis's (1993: 8) claim that it is the inadequacy of language description that makes students devise and implement learning strategies which do no depend on grammatical description of the language. This implies that it is deficiencies in verb codes in COBUILD1 that make students infer the relevant information from other parts of the microstructure. As has been shown in the preceding sections, verb codes have been further modified in the pedagogical dictionaries published since 1995. The codes used in COBUILD2 are also markedly different from those in the previous edition. The time seems ripe, therefore, for research which would make it possible to answer the question whether more user-friendly verb codes are more frequently and readily used by learners. (17)

9. Conclusions

The above study of coding systems shows that while today's learners' dictionaries can be seen as continuing a consistent encoding tradition begun in pedagogical lexicography before World War II, verb codes have been considerably altered in the course of time. Changes in the systems of codes in the learners' dictionaries should be regarded as beneficial to the learner. For one thing, codes no longer constitute cross-references to explanatory charts, as was the case in the early pedagogical dictionaries, but being composed of symbols for the verb and its complements, they themselves convey information on the complementation structure of the verb. Especially praiseworthy is the shift from mnemonic coding systems toward transparent ones, which appreciably enhance the accessibility of encoded syntactic information to the learner. The trend toward simplification and lesser abstraction of coding systems is welcome and should be continued, since the more obvious the meaning of a code, the more usable the code is like ly to be for learners who, as a rule, are loath to turn to a separate table or key for consultation. Its concomitants, that is, reduced accuracy of codes and their more space-consuming form, should be seen as an acceptable trade-off for clarity.

In view of the fact that greater transparency of codes entails their simplification, distinctions between them and, consequently, between the types of information they represent are blurred. This, coupled with the fact that codes are grossly underused, points to the need for a system for presenting grammatical information whereby the learner is not entirely dependent on verb codes. Moreover, the fact that, as Harvey and Yuill's (1997) study shows, students are far more willing to access grammatical information by analogy rather than through coding, is all the more reason why verb syntax should also be shown indirectly in the entry by means of such vehicles for uncoded syntactic information as definitions, examples and usage notes. (18)

DICTIONARIES

Cowie, Anthony P. (ed.)

1989 Oxford advanced learner's dictionary of current English. (4 edition.). Oxford: Oxford University Press. [OALDCE4].

Crowther, Jonathan (ed.)

1995 Oxford advanced learner's dictionary of current English. (5 edition.). Oxford: Oxford University Press. [OALDCE5].

Homby, Albert S. (ed.)

1974 Oxford advanced learner's dictionary of current English. (3. edition.). Oxford: Oxford University Press. [OALDCE3].

Procter, Paul (ed.)

1978 Longman dictionary of contemporary English. (1 edition.). Harlow: Longman. [LDOCE1].

1995 Cambridge international dictionary of English. Cambridge: Cambridge University Press. [CIDE].

Sinclair, John M. (ed.)

1987 Collins COBUILD English language dictionary. (1 edition.). London -- Glasgow: Collins. [COBUILD1].

1995 Collins COBUILD English dictionary. (2. edition.). London: HarperCollins Publishers. [COBUILD2].

Summers, Della (ed.)

1987 Longman dictionary of contemporary English. (2 edition.). Harlow: Longman. [LDOCE2].

1995 Longman dictionary of contemporary English. (3 edition.). Harlow: Longman. [LDOCE3].

Wehemeier, Sally (ed.)

2000 Oxford advanced learner's dictionary of current English. (6. edition.). Oxford: Oxford University Press. [OALDCE6].

* This paper is based on research carried Out for the author's MA thesis (Dziemianko 2001).

(1.) At this point, Bejoint points up the need for adjustments to the dictionary. It is noteworthy, however, that honing students' reference skills by deliberate teaching is also a viable alternative.

(2.) In this regard, it is instructive to compare the Oxford advanced learner s dictionary (1974) with the Shorter Oxford English learner's dictionary (Onions 1969, 1). The former, for the verb bring, provides grammatical codes for five out of the six major subsenses. The latter simply makes use of the label v, which appears only once at the beginning of the relevant entry.

(3.) Valency theory was advanced by Tesniere in his 1959 book Elements de syntaxe structurale, excerpts from which are quoted in Allerton (1982: 1-10).

(4.) As will be shown in section eight, students' familiarity with even basic grammatical terminology can by no means be taken for granted.

(5.) Following Aarts's (1991: 571-577; 1999: 22-29) example, the term symbol used with regard to verb codes denotes any letters, digits or abbreviations that comprise codes.

(6.) [VP2D], [VP4B], [VP6E], [VP255], [VP13A], [VP18B], [VP18C], [VP19A], [VP19B], [VP19C], [VP20], [VP21], [VP22] and [VP25] are the codes in question.

(7.) LDOCE1 remains very vague about what exactly is meant by nounlike expression, the term used also in the description of L, T, and D. examples given in LDOCE1 are what to do, president, fool. Lemmens and Wekker (1986: 65) doubt, with reason, whether the ordinary dictionary use will understand what the nounlike characteristics are that these examples share.

(8.) Regrettably, even though the symbol in question is used as described in COBUILD2, it is not included in the list of notations inside the front cover of this dictionary, nor is it explained in the grammar section. The learner is thus left to infer from examples which it accompanies in the verb entry that it denotes the intransitive verb. By contrast, in the previous edition, where V labels also linking verbs, its significance is spelled out in the appropriate special boxed entry (COBUILD1: 1613).

(9.) Both formal and functional categories are nonetheless incorporated in the description of verb codes in the section of the dictionary where codes are explained.

(10.) Herbst (1996: 329) points out that the use of pattern illustrations is typical of bilingual dictionaries. Interestingly, LDOCE3 was not the first monolingual learners' dictionary to adopt the system. Longman Language Activator (1994) develops precisely the same method of indicating grammatical patterns.

(11.) Although space-consuming, these labels are the only ones that fulfill the last minimal condition stipulated by Lemmens and Wekker and discussed in section 2.

(12.) This issue was addressed by Aarts (1991: 575), who analyzed codes for seven underlyingly different structures in OALDCE4, LDOCE2 and COBUILD1.

(13.) For more syntactic tests showing differences between monotransitive, ditransitive and complex transitive verbs see Quirk et al. (1985: 1216-1220).

(14.) It should be noted that LDOCE2 and COBUILD1 do not encode know sb to be a liar. Each of these dictionaries, however, gives only one code for the other two structures under scrutiny. Interestingly, the study conducted by Aarts (1991: 576) led to similar conclusions inasmuch as it showed that the deep syntactic structure was reflected in codes only in OALDCE4. In the other two dictionaries consulted by Aarts codes concerned only the surface syntactic structure.

(15.) See Table2 for the total number of additional symbols in verb codes. The following codes represent functional categories in the dictionaries: [empty set], 1, 7, 9 in LDOCE1; obj, obj(d), obj(i) in LDOCE2; A, C, O in COBUILD1; obj and [two objects] in CIDE.

(16.) Bejoint (1981: 208) points out, however, that his research must be seen against the background of Tomazczk's study (1979), the first generally available one where the most important vantage point is that of the needs of the learner. Tomaszczyk, unlike Bejoint, went beyond the English monolingual context to investigate requirements of the foreign language learner and translator. His study concerned also bilingual dictionaries and in his heterogenous sample of informants there were Polish students. For more details see Tomaszczyk (1979: 104-117).

(17.) This question will be explored in the author's doctoral dissertation.

(18.) Correspondence between codes and examples is discussed in detail in a separate paper (Dziemianko forthcoming).

REFERENCES

1991 "Lexicography and syntax: The state of the art in learner's dictionaries of English", in: James E. Alatis (ed.), 567-582.

1999 "Syntactic information in OALD5, LDOCE3, COBUILD2 and CIDE", in: Thomas Herbst - Kerstin Popp (eds.), 15-32.

Alatis, James E. (ed.)

1991 Linguistics and language pedagogy: The state of the art. Washington D. C.: Georgetown University Press.

Allerton, D. J.

1982 Valency of the English verb. London: Academic Press.

Bejoint, Henri

1981 "The foreign student's use of monolingual English dictionaries: A study of language needs and reference skills", Applied linguistics 2: 207-222.

Bogaards, Paul

1996 "Dictionaries for learners of English", in: International journal of lexicography 9: 277-320.

Cowie, Anthony P.

1978 "Problems of syntax and design of a pedagogical dictionary", Rassegna Italiana di linguistica applicanta 10: 225-264.

1983a "On specifying grammar: On specifying grammatical form and function", in: Reinhard R. K. Hartmann (ed.), 99-107.

1983b "The pedagogical/learner's dictionary: English dictionaries for the foreign learner", in: Reinhard R. K. Hartmann (ed.), 135-144.

1987 "Syntax, the dictionary and the learner's communicative needs", in: Anthony P. Cowie (ed.), 183-192.

1990 "Verb syntax in the revised Oxford advanced learner's dictionary: Descriptive and pedagogical considerations", in: Anthony P. Cowie (ed.), 341-347.

1999 "Learners' dictionaries in a historical and a theoretical perspective", in: Thomas Herbst - Kerstin Popp (eds.), 3-13.

Cowie, Anthony P. (ed.)

1987 The dictionary and the language learner: Papers from the EURALEX seminar at the University of Leeds, 1-3 April 1985. Tubingen: Niemeyer.

1990 Euralex '90 proceedings: Actes del IV Congreso International. Barcelona: Bibliograf.

Crystal, David

1986 "The ideal dictionary, lexicographer and user", in: Robert Ilson (ed.), 72-81.

D'Elia, Catherina

1992 "Syntactic information in monolingual dictionaries: The Oxford Learner's dictionary", Lingvisticae investigationes 16: 233-265.

Dziemianko, Anna

2001 Verb syntax in monolingual English learners' dictionaries. [Unpublished MA thesis, Poznan: Adam Mickiewicz University.]

forthcoming "Encoded syntactic information on verbs in monolingual English learners' dictionaries: distribution in the microstructure and congruity with illustrative material".

Ellegard, Alvar

1978 "On dictionaries for language learners", Moderna Sprak 72: 225-242.

Fisiak, Jacek (ed.)

1997 Studia Anglica Posnaniensia 32. Poznan: Wydawnictwo Nakom.

Hamdan, Jihad -- Shehdeh Fareh

1997 "Dictionaries as a potential source of error for Arab EFL learners: Evidence from verb argument structures", in: Jacek Fisiak (ed.), 197-216.

Hanks, Patrick

1993 "Definitions and explanations", in: John M. Sinclair (ed.), 116-136.

Hartmann, Reinhard R. K.

1983 "Four perspectives on dictionary use: A critical review of research methods", in: Anthony P. Cowie (ed.), 11-28.

Hartmann, Reinhard R. K. (ed.)

1983 Lexicography: Principles and practice. London: Academic Press.

Harvey, Keith -- Deborah Yuill

1997 "A study of the use of a monolingual pedagogical dictionary by learners of English engaged in writing", Applied linguistics 18: 253-278.

Hausmann, Franz Josef - Adeline Gorbahn

1988 "COBUILD and LDOCE II: A comparative review", International journal of lexicography 2: 44-56.

Hausmann, Franz Josef -- Oskar Reichmann -- Herbert Ernst Viegand -- Ladislav Zgusta (eds.)

Heath, David -- Thomas Herbst

1989 Worterbucher, dictionaries, dictionnaires: An international encyclopedia of lexicography. 3 vols. Berlin -- New York: Walter de Gruyter.

1988 "Review of Longman dictionary of contemporary English (new edition)", English language teaching journal 42: 315-317.

Herbst, Thomas

1984 "Adjective complementation: A valency approach to making EFL dictionaries", Applied linguistics 5: 1-11.

1996 "On the way to the perfect learners' dictionary: a first comparison of OALD5, LDOCE3, COBUILD2 and CIDE". International journal of lexicography 9: 321-357.

1999 "Designing English valency dictionary: Combining linguistic theory and user-friendliness", in: Thomas Herbst -- Kerstin Popp (eds.), 229-253.

Herbst Thomas - Gabriele Stein

1987 "Dictionary-using skills: A plea for a new orientation in language teaching", in: Anthony P. Cowie (ed.), 115-127.

Herbst, Thomas -- Kerstin Popp (eds.)

1999 The perfect learners' dictionary (?). Tubingen: Niemeyer.

Hornby, Albert S.

1965 "Some problems in lexicography", English language teaching 19:104-110.

Hornby, Albert S. -- E. V. Gatenby -- H. Wakefield (eds.)

1948 The advanced learner's dictionary of current English. (1st edition.) Oxford: Oxford University Press.

[1958] [Reprinted by Polish Scientific Publishers]

1963 The advanced learners' dictionary of current English. (2nd edition.) Oxford: Oxford University Press.

Huang, G. F.

1985 "The productive use of EFL dictionaries", Regional English language centre journal 16: 54-71.

Ilson, Robert (ed.)

1985 Dictionary, lexicography and language learning. Oxford: Pergamon Press.

1986 Lexicography: An emerging international profession. Manchester: Manchester University Press.

1989 "Present-day British lexicography", in: Frans Josef Hausmann et al. (eds.), 1967-1983.

Jackson, Howard

1985 "Grammar in the dictionary", in: Robert Ilson (ed.), 53-59.

1988 Words and their meaning. London: Longman.

Kaliski, Terry

1992 "Computer-assisted language learning (CALL)", in: P. Roach (ed.), 97-109.

Krishnamurthy, Ramesh

1993 "The process of compilation", in: John M. Sinclair (ed.), 62-80.

Lemmens, Marcel -- Hermann Wekker

1986 Grammar in English learners' dictionaries. Tubingen: Niemeyer.

1991 "On the relationship between lexis and grammar in English learners' dictionaries", Cahiers de l'Institut de Linguistique de Louvain 17: 227-242.

McCorduck, Edward Scott

1993 Grammatical information in ESL dictionaries. Tubingen: Niemeyer.

Nesi, Hilary

2000 The use and abuse of EFL dictionaries: How learners of English as a foreign language read and interpret dictionary entries. Tubingen: Niemeyer.

Nickel, Gerhard (ed.)

1976 Proceedings of the 4th International Congress of Applied Linguistics. Stuttgart: HochschulVerlag.

Onions, C.T. (ed.)

1969 The shorter Oxford English Dictionary. 2 vols. (3rd edition.) Oxford: Clarendon Press.

Orszagh, Ladislas

1969 "Wanted: Better English dictionaries", English language teaching 23: 216-222.

Palmer, Harold E.

1938 A grammar of English words. London: Longman.

[1949] [Reprinted, Edinburgh: Bishop and Sons, Ltd.]

Procter, Paul

1976 "The design of a dictionary for language teaching and learning", in: Gerhard Nickel (ed.), 309-317.

Quirk, Randolph -- Sidney Greenbaum -- Geoffrey Leech -- Jan Svartvik (eds.)

1985 A comprehensive grammar of the English language. London -- New York: Longman.

Roach, P. (ed.)

1992 Computing in linguistics and phonetics. London: Academic Press.

Sinclair, John M. (ed.)

1993 Looking up: An account of the COBUILD project in lexical computing. London: Harper Collins Publishers.

Stark, Martin

1999 Encyclopedic English dictionaries: A study of their design features from user perspective. Tubingen: Niemeyer.

Summers, Della

1994 Longman language activator: Harlow: Longman.

Swan, Michael

1985 "A critical look at the communicative approach", English language teaching journal 39: 2-12, 76-87.

Sweet, Henry

1899 The practical study of languages: A guide for teachers and learners. London: Oxford University Press.

[1964] [Reprinted, in: Language and language learning series.]

Tomaszczyk, Jerzy

1979 "Dictionary users and uses", Glottodidactica 12: 104-119.

West, Michael - James Endicott (eds.)

1935 The new method English dictionary. London: Longman.

[1965] [Reprinted, London: Green and Co. Ltd.]

1977 An international reader dictionary. (2nd edition.) London: Longman.

Whitcut, Janet

1986 "The training of dictionary users", in: Robert Ilson (ed.), 111-122.

Willis, Dave

1993 The lexical syllabus: A new approach to language teaching. London: HarperCollins Publishers.