- Cross-Referencing as a Lexicographic Device

Ab t act: The mediostructure, that is the system of cross-referencing, is a lexicographic device S r b used to establish relations among different components of a dictionary. This paper th.t can e I· d th . . I li .. I em' ses on different mediostructura strategtes an ell prachca app cation m genera syn ONC foc~ . s ...... e structure of dictionaries is discussed from a metalexicographic perspective in dlet/onane. j II order to explain the system of cross-referencing. It is shown how textual cohesion, achieved by the Interaction of the various structural components, is promoted by the use of a system of cross-referencing and improved by an innovative approach towards a mediostructure-orientated lexicog-


Introduction
In spite of the fact that lexicography has been practised for centuries, metalexico.gr~phy, that is the theory of lexicography, is a relative new subdiscipline metalexicography and the lexicographic practice. The metalexicographical influence has transformed lexicography into a scientific practice with a very definite purpose, viz. the production of dictionaries. The production of diction_ aries as a result of the scientific practice of lexicography should lead to the establishment of the cultural practice of dictionary use. The user-perspective, So prevalent in modem-day metalexicography, compels lexicographers to compile their dictionaries according to the needs and research skills of well-defined target user groups. The dominant role of the user has had a definite effect on the compilation of dictionaries as well as on the evaluation of their quality. Good dictionaries do not only display a linguistically sound treatment of a specific selection of lexical items. They are also products that can be used as linguistic instruments by their respective target user groups. The better they can be used, the better dictionaries they are.
The quality of dictionary use, that is the degree of success a user experiences when consulting a dictionary and employing the retrieved information, is determined by a variety of features, but one of the most important characteristics of a good dictionary is its accessibility. Accessibility leads to an unambiguous retrieval of the information presented on both the macro-and microstructural levels. Any theory of lexicography should present strategies to improve the linguis!ic quality of dictionaries. However, this should be preceded by strategies to enhance the w'ay in which the target user can identify the data he/she is looking for in order to retrieve the necessary information and to utilise it for decoding or encoding purposes.
Dictionary research has lead to the establishment of different structures of printed dictionaries, e.g. in addition to the macrostructure, microstructure and access structure also the mediostructure. The mediostructure, that is the system of . cross-referencing, is a lexicographic device that can be used to establish relations between different components of a dictionary. According to Wiegand (1996: 11) it interconnects the knowledge elements represented in different sectors of the dictionary on several levels of lexicographic description to form a network. Working with a dictionary as a carrier of texts, the mediostructural entries can guide the user between different texts, e.g. between the central text and any text in the front or back matter or between various articles functioning as subtexts in the central word-list. An article-internal mediostructural relation assists the user to relate various microstructural entries employed in the same article.
This paper focuses on different mediostructural strategies and their practical application in general synchronic dictionaries. The structure of dictionaries is discussed from a metalexicographic perspective in order to explain the domain of application of a systp.m of cross-referencing. It is shown how textual cohesion, achieved by the interaction of the various structural components, is promoted by the use of a system of cros~-referencing and improved by an innovative approCich towards a mediostructure-orientated lexicography. Although the mediostrticture of dictionaries is a central topic of this paper, references to 19 of mediostructures will only cover a small segment of this structural the theoryt A detailed discussion can be found in Wiegand (1996).
. cornponen. ba sic terms relating to a theory of mediostructures some W' nd (1996) gives an exposition of the fundamental terms employed in a th leg a of mediostructures. According to his theory, a leXicographer refers the d.e~~ary user from a reference position to a reference address. This is usually dl~e by means of a reference entry and gives the user access to additional ~ vant lexicographic data. A reference relation is established between the r:f:rence entry and the reference address. In Webster's Ninth New Collegiate ~ictionary (W9) the article of the lemma sign frog contains the following entries: any of various smooth-skinned web-footed largely aquatic tailless agile leaping amphibians ... -compare TOAD In this example the specific slot in the article of the lemma sign frog is the reference pOSition, and the lemma sign toad, the separate macrostructural entry to which the user is referred, is the reference address. Here the reference entry consists of two separate text segments, .i.e. the entry marking the reference relation (compare), henceforth referred to as the reference marker, and the entry indicating the reference address (toad). A variety of reference markers is used in different dictionaries and often also in one dictionary, e.g. text segments like see, compare, ~, =:>, etc. In the English-Dutch translation dictionary Van Dale Groot Woordenboek Engels-Nederlands a single arrow is used as one of the reference markers. In the article of the lemma sign track system the reference entry "~ tracking" consists of the reference marker "~" and the entry tracking, indicating the reference address.
A rather interesting example of cross-reference is found in Dictionary of Lexicography (OL) in its treatment of the entries circular reference and reference circularity. The first is referred to the second, and the second again to the first: The niched lemmata, included as run-on entries, consist of the unexplained compounds fratsvlieer. (aerobat), frats brander, fratsgolf (freak wave) and fratsongeluk (freak accident), with the lexical item fratsas word initial stem. However, it is not the same sense of the lexical item frats that functions in these self-explanatory compounds. Two sense discriminations occur in the treatment of the lemma frats. To assist the user in achieving the correct semantic interpretation a relation has to be established between the different niched lemmata and the relevant sense discriminations. One of the ways in which this can be done, is by means of a mediostructural procedure that is aimed at an article-internal address. The lemma sign fratsvlieenier is followed by the entry (by 1) and the other compounds by the entry (by 2). These are cross-references with the text segment by functioning as a reference marker, and the text segments 1 and 2 functioning as reference address indicators. These reference entries guide the user from the reference position to an address in the same article, i.e. the meaning paraphrases presented for the respective sense discriminations. A second type of reference address is the external reference address. The cross-reference exceeds the boundaries of the article. Two search domains can be identified for external reference addresses. Dictionary articles are texts but they also function as sub texts of the central word-list which is the dominating lexicographic text. Th~ external address can be located either elsewhere in the central word-list, e.g. another lemma sign or a specified microstructural element in another article, or in a separate text outside the central word-list. Compare the articles of gyro and stow in Collins Dictionary of the English Language (CED) and Longman Dictionary of Contemporary Englis~ (LDOCE) respectively.
IS exp arne rn s ta bl: as a verb with one object as well as an additional descriptive word or ta hrase , e.g. put + it + in the box. The text ~lement away in the q~oted text segp nt is the additional word to be used With the verb stow. In this example the ::t segment X9 is a reference entry indicating an external address located in another text of the dictionary.
Quite often a combination of external and internal reference addresses are given in one refe~ence en~y. In The Co~cise Oxford Dictionary the article of the lemma sign ghostmg con tarns the followrng entries: the appearance of a 'ghost' (see GHOST n.4) or secondary image in a television picture.
In this example the meaning paraphrase of the lemma sign ghosting is the reference position containing a triple address which consists, as the main address, of an external reference address located in the central word-list, i.e. the lemma sign ghost, as well as two additional internal addresses, i.e. a secondary address, the nominal function of this lexical item, and a tertiary address, the fourth polysemous sense of this item. The last two reference addresses identify text segments in the article of the lemma sign ghost.
The use of a mediostructural strategy of external reference addresses endeavours to enhance the functionality of a dictionary as a source reflecting aspects of the linguistic reality. One of the real problems experienced by the users of alphabetically ordered dictionaries is the decontextualisation of lexical items. Bolinger (1985: 69) maintains that lexicography is an unnatural occupation: "It consists in tearing words from their mother context and setting them in rowscarrots and onions and beetroot and salsify next to one another -with roots shorn like those of celery to make them fit side by side, in, an order determined not by nature but by some obscure Phoenician sailors who traded with Greeks in the long ago." He continues this argument by saying that "half of the lexicographer's labor is spent repairing this damage to an infinit:ude of natural connections that every word in any language contracts with every other word, in a complex neural web knit densely at the center but ever more diffusely as it spreads outward". According to him "a bit of context, a synonym, a grammatical category, ... and a cross-reference or two" are "the additives that accomplish the repair" .
From both a semantic and a pragmatic perspective the lexicon has to be regarded as an ordered set of lexical entries. However, the alphabetical ordering of a dictionary defies the network of semantic relations existing between this set of lexical entries. The mediostructure of a dictionary is a powerful mechanism to re-establish some of the lexical relations. Dictionaries employ the mediostructure to refer the user to external addresses which are linked with the lemma sign of the reference position article in relations such as synonymy, oppositeness of meaning, hyponomy, dialectat stylistic, chronolectic and other forms of variation, etc. For the language learner as well as the seasoned native speaker of any given language these cross-references represent an added value which assists them in improving their communicative potential. South African dictionaries should employ external reference addresses in a more general and consistent way. However, it is of extreme importance that these strategies be explained comprehensively in the front matter of the dictionary.
The third category of reference address is the dictionary external reference address. This mediostructuralprocedure links a text segment in a dictionary to a source outside the dictionary. In A Dictionary of LAnguage Planning Terms Cluver (1993) puts the strategy of dictionary external reference to good use. The back matter of the dictionary contains a bibliography of sources in which more information regarding the terminology treated in the dictionary can be found. Many articles contain condensed bibliographical references which leads the user to the bibliography in the back matter. This is the reference position. from where the user is guided by means of a complete reference to the specific source. The condensed bibliographical references in the articles are clearly indicated by the reference marker "Bibl.". In the article of the lemma sign primary language the following text segment is fcund: "Bibl. Miihlhausler 1986: 9". The bibliography gives the full reference, i.e. "Mi.ihlhausler, P. 1986. Pidgin and creole linguistics. Oxford: Basil Blackwell." By means of the dictionary external reference address the lemma sign is linked to this external source. A variety of other reference addresses can also be identified but they are not relevant for the present discussion.
For the African languages, apart from the disruption of semantic relations, alphabetical ordering has serious detrimental consequences for grammatical relations. Many traditional compilers, although following an alphabetical ordering in principle, regard the importance of combined semantic and grammatical cohesion as too important to break. This view implies that in the case of African languages the mediostructure is incapable of re-establishing the most relevant lexical relations. In most dictionaries this results in a hybrid approach where different derivations, sometimes a hundred or more, of a single word are treated within the article of a nominal or especially verbal stem in a complex article with numerous sublemmas and sublemmatic addresses, in addition to being entered as separate lemmas in their appropriate alphabetical positions.
In this way mediostructure is exhausted/overused for the sole purpose of maintaining structural links. Little or no realization of mediostructure as a powerful access structure is achieved. Once referred back to the main cluster (4), it is unlikely that the user will be able to work out the meaning, especially for those cases which lie relatively deep in the modular structure as in the case of dithekollibno. The user has to look up this word under the singular thekol-liAano in (S) and is then referred to reka in (4) and eventually, after having struggled through this lengthy article, he finds thekollisano at the end of (4) with no translation equivalents given. (Compare Prinsloo (1994) for similar examples and a detailed discussion on problematic aspects of the lemmatization of verbs.) This obsession with keeping together what in their view semantically and grammatically "belong together" thus results in extremely user-unfriendly entries in which successful retrieval of information virtually becomes impossible. It could be argued that the utilization of cross-references simply for the sake of grammatical binding is nonfunctional.
It was stated in the introduction that one factor in the evaluation of a dictionary is the extent to which it is useful to the user. Dictionaries such as these fail in this main criterion. Students consequently opt for less sophisticated dictionaries with less information categories and less exhaustively treated lemmas, i.e. a lower density of information.
t reference address to which the user is referred from (6)(a) is In (6) tJ:te ~ac econd entry for mollo, (6)(c), has no relation to molelo whatso-uncertall1. k e ~gs worse, the reference in (6)(a) is to mo11o instead of'mo11o.
ever. To mati' e .. I .. is also not explained in the dictionary. Since mollo in (6) The cross-reference to brackets is quite appropriate since the latter is more frequently used. Also, due to considerations regarding frequency of use, no reference from brackets to bracket is necessary. However, in looking up brackets, the user does not get any additional information, e.g. in respect of types and use of brackets. On the contrary, he/she is misguided by the additional information given at the reference address namely that the lemma brackets is 26 RH. Gouws and D.J. Prinsloo translated in Afrikaans and Sepedi as necessarily square. Thus, in contrast to the singular bracket, the plural form excludes other types of brackets. It is important that the user should find more information at the reference address, otherwise the value of cross-referencing is devalued. Cross-reference or more specifically, the position of a cross-reference entry, indicates to the use; that this is the starting-point in the process of information retrieval. The usage frequency of the item which stands in the cross-reference position, is lower than the reference address. The lexicographer may never utilize the system of crossreferencing simply because he/ she does not want to give proper treatment to the items in question. If it is in the interest of the target user, a specific lemma should be entered and treated. Cross-references such as those attempted in (6) and (8), will have a negative effect on the target user. Once disappointed, it will discourage him/her from following up cross-references since it is impossible to distinguish between functional and nonfunctional references in the dictionary.
Consider also NTO's treatment of complainant versus plaintiff: It is unclear why no cross-reference from plaintiff to complainant is given. Such a reference is necessary because equivalents in both target languages are given under complainant. The addition of the translation equivalents eiser in the Afrikaans column and motlalei in the Sepedi column also raise a few questions. Firstly, it implies that eiser and motlalei are suitable equivalents for plaintiff but not for complainant. Secondly, to give eiser as the first translation equivalent for plaintiff suggests that it is the best option. However, although it is added in the case of motlalei to the translation equivalent paradigm in (9)(b) for the sake of eiser, it is given at the end. Thus the entire relationship between complainant and plaintiff becomes unclear. The user cannot determine in which relation they stand to each other. Central text-internal reference should strengthen the cohesion, as is correctly done in the case of molelo versus moHo in (6) above. In the case of (9) this cohesion is actually broken off. The user who wants to find translation equivalents in Afrikaans and Sepedi is referred to another word where the same treatment is given for no reason. An even more confusing example of cross-reference in NTO is its treatment of Brave West Wind versus anti-trade wind:

--. given from (lO)(a) to (lO)(b) but not vice versa. The compilers
Cross-refere~ct 1~ in the use of capital versus lower case letters in Anti-/ antiare not CO~SlS :~o. Ironically reference to the "West" in Sepedi, namely, tsa at\d Phefo. /p d~ed to the translation equivalent paradigm of (1O)(b) instead of bodike1a, is a (10)(% addition to implicit cross-referencing, two types of explicit cross-refer-.  The examples under (11) from TS, explicitly referring the user from kgarebe to lekgarebe is sensible since apart from the meaning "girl who can be married" which is similar to that given in sense 1 of lekgarebe, an extended meaning "a neatly, well-dressed girl" is given as sense 2. The fact that no explicit reference from lekgarebe to kgarebe is given, is also quite acceptable since the user who looks up lekgarebe will not find any new information under kgarebe. However kgarebe must be given as a synonym directly following the sense 1 definition. TS's treatment of kgarebe can also be improved in respect of the position allocated to the reference entry. The explicit cross-reference : lekgarebe should not be given in the focus position of the article. It can be regarded as an unnecessary or even nonfunctional cross-reference interfering with the user's information retrieval process. Formulated differently, the information primarily needed by the user who looks up kgarebe is that given in the definition. Once given the definition, he/she might be interested to consult the reference address for additional information. It can also be argued that the wrong reference symbol is used in the case of kgarebe versus lekgarebe. The relation is one of relatedness rather than difference -thus in terms of TS's conventions "BCNA" rather than ": ".
TSO's treatment of the same words are shown under (12). In the case of kgare-b@ only a definition is offered, while a synonym as well as a definition is given for lekgarebe. Since no cross-reference is made from kgarebe to lekgarebe, it , 28 R.H. Gouws and O.J. Prinsloo suggests that kgarebe is the entry with the higher usage frequency. However the user gets 'more information from lekgarebe, namely a synonym as well as ~ definition, than from kgarebe. This is confusing. In terms of cross-reference it can be said that the article of the lemma sign lekgarebe is a reference position of the reference entry kgarebe. Normally, for economical reasons, the same definition is not given in two places. Two definitions and the lack of cross-refer_ ence has a negative effect on cohesion. Here the user cannot establish which one is the more frequently used. The more frequently used word is the one likely to be treated. This in itself is an indication of higher frequency of use. It would thus be better to enter kgarebe with a definition, adding lekgarebe as a synonym. It is normal,practice to give a list of synonyms after the definition since they meet the criteria to be lemmatized themselves. Such synonyms can be listed in order of frequency of use if such criteria is available or otherwise alphabetically. Thus, since all synonyms have to be entered as lemmas, lekgarebe will be entered as a lemma sign but only with a cross-reference to kgarebe.
It is also not clear why in both TS and TSD the definitions differ in respect of the concept "grown up". In the case of lekgarebe "a girl who can be married" and in the case of kgarebe "a grown-up girl, one who is not yet taken/married". When comparing the two, the user can get the wrong impression that kgarebe implies an adult and lekgarebe not. ' Cross-references from the front matter, especially from the user's guidelines to the central text are crucial to the user for successful or optimal retrieval of information. Dead references, especially in the guidelines of a dictionary are defects which undermine the trust of the user in the dictionary as a reliable source of information, and in the value of the cross-referencing system as a whole. Such dead references often do not effect only one reference address, a key to a whole section can be lost. Consider the following example: In the guidelines to A Learner's Chichewa and English Dictionary (LCE), the compilers explain the policy not to lemmatize derived forms when the meaning is readily ascertainable from the root plus suffix combination. In support of this farreaching decision for lemmatization of an African language, they include crossreferences in the central text: "Thus, both -mva 'hear, understand' and its derived form -mvana 'get along together' are listed". However, the very examples that they quote to illustrate their policy, are not treated as such: -mva is listed but not -mvana. This dead reference to -mvana can cause the user to doubt the treatment policy not only in respect of a single entry but a :whole category of entries. A similar dead reference occurs in the next sentence: "the derived verb -mverana 'listen to each other' is not listed because its meaning is readily determined from the root -mvera 'listen to' plus an affix". However, again the root -mvera is not listed, clearly viola~g the claim "verbs are entered according to their root forms". The treatment of cross-references in the Dictionary of Northern Sotho Grammatical Terms (NGT) can now be considered. This dictionary is a pioneering first for Sepedi and very popular among its target users. In this article of tone explicit reference is made to Bantu languages and nucleus. At the reference address, Bantu languages, the user finds more useful information on tone in the African languages. Likewise, the user who consults the entry Bantu languages first will find, in addition to other useful informadon given there, "tone plays a distinctive role. See tone". This is good lexicographical practice since for the user who consults the entry Bantu languages, as well as for the user who looks up tone, the cross-references are useful. Both contain more information at the respective reference addresses with regard to two important and closely related issues such as African languages and tone. The same holds true for the explicit reference made to tone in the article of syllabic nasal. The treatment of syllable and tonal pattern as reference positions of explicit reference to the addresses nucleus and tone respectively, can however be improved.
(14) syllable (noko, sillabe/lettergreep) See nucleus. (15) tonal pattern ("patrone ya segal6, toonpatroon) See tone Firstly, syllable in (14) deserves full treatment, especially in a dictionary of grammatical terms. Apart from translation equivalents in Sepedi and Afrikaans, no definition is given, only an explicit reference to nucleus. In the article of nucleus, many references are once again made to syllable, such as "[a nu-cl~usJ is used to characterize the nature of a syllable". As for syllable, it is maintame? that "vowels form the nuclei of syllables", etc. However, syllable itself remams undefined. Thus syllable deserves a definition and treatment as for example in (16) from South African Student's Dictionary (SSD) and in (17) from New Student's Dictionary (NSD): ---(16) syllable ... noun: syllables A syllable is any of the parts, consisting of one or more sounds and usually including a vowel or a consonant acting like a vowel, that spoken word can be divided into: The word 'telephone' has three syllable a 'te ','Ie',and 'phone',and 'tiger' has two,'ti' and 'ger'. 5,(17) syllable ... syllables. N-C A syllable is a part of a word that contains a single vowel-sound and that is pronounced as a unit. For example 'book' has one syllable, and 'reading' has two syllables. ' (18) tone me ("segalwana, toneem/toonfoneem) See tone.
An important statement in the article of tone in (13) reads: "Tone is always associated with a particular syllable." The user of NGT consulting tone could easily perceive the italicized word syllable as an implicit reference entry but find it to be nonfunctional since in looking up syllable in (14), he/ she is referred to another address namely nucleus. Furthermore, although tone is one of the key issues discussed in the article of nucleus, no explicit reference is given to tone.
In the case of tonal pattern, the user is referred to tone but the distinction between tone and tonal pattern is unclear. From phrases such as "depending on its tone or tonal pattern", it is not clear whether or means "equal to" or "in contrast to". The user who wishes to know the mearting of tonal pattern is referred to tone but will not know for sure after having studied the treatment of tone whether tone and tonal pattern is synonymic or not. In the case of tone versus tone me in (13) and (18), or in the phrase "a tone or toneme" means tone is equal to toneme. It should rather be clearly stated that tonal pattern is a series of tones/tonemes. This could be explained by using mosadi as an example, where mo-has a low tone, -sa-a high tone and -di again a low tone. The tonal pattern of these three tones/tonemes is therefore low-high-Iow, often indicated as LHL. This suggests that tonal pattern deserves to be treated on its own. In Wiegand's (1996) terms, it means that if the user is referred from tonal pattern, which is the reference position, to the article of the lemma sign tone, the reference address, more information on/a fuller treatment of tonal pattem must be given. Thus the cross-reference from tonal pattern to tone is not observed in the sense that tonal pattern is not really treated within the article of tone. The purpose and value of the cross-reference is lost.
Finally, key terms used in the treatment of the lemma tone which are italicized such as pitch variations, tonemes, and especially syllable are not treated in the dictionary. The user expects a clearer distinction between implicit reference t? a different reference address, on the one hand, and mere instances of emphaSIS on the other. ' This does not mean that the lexicographer should solely utilize explic~t references to distinguish between emphasis and cross-referencing, since there IS -I lue difference between explicit and implicit reference sysno tundamenta ~a only more obvious than the latter. Thus it is suggested that teII'B. 1?'e form~r IS should utilize both as long as implicit references can be the lexlC~g.rap .e~ed from mere emphasis. The implicit cross-reference strategy clearly dlStIngu~: clearly apparent. This means that terms used within the arti-D\US t howe~er hich are themselves lemmatized and treated elsewhere in the Ie of entrIes w . c s. must stand out and be treated consIstently. dictionary, diostructure has not in all instances been employed to its full ~~ ~e a dictionary of grammatical terms, the mediostructure could be po~tia d as a powerful access structure by ensuring that at least all keywords elJ\~ oy'~ the treatment of a specific lemma which are themselves entered as lem- It can rightfully be argued that the lexicographer should guard against excessive text condensation. However, opportunities should be utilized to strengthen the cohesion of the dictionary by optimal organization of the mediostructure as an access structure. An excellent example in African language lexicography where mediostructure has been employed as a powerful access structure is the Lexicon Cilubil-Nederlands (LCN) compiled by De Schryver and Kabuta. This dictionary is highly successful in interconnecting the knowledge elements represented in different sectors of the dictionary on several levels of lexicographic description to form. a network. In contrast to GN, for example, the compilers of LCN are aware of the benefits of "keeping together what semantically and grammatically belong together" but also of the need (a) to avoid extremely long entries and (b) to ensure proper treatment of each derivation in terms of grammatical, tonal and lexical information. Compare the entries in LCN for -funda and its derivations: In contrast to cases such as dithekollibno above, the user can find even Com_ plicated derived words such as fundilangana firstly lemmatised separately in its proper alphabetical position and secondly fully treated. The user does not have to refer back to the stem entry fund a to find the necessary information.
An implicit cross-reference is nevertheless given to the root -funda where all the relevant derivations are listed. The compilers of LCN thus succeeded in harmonising lumping and splitting, capturing the advantages of both these . approaches. It can, of course, be argued that the listing of the different deriva_ tions occupies precious space in the dictionary. However, by substantially reducing the font size, this redundancy is diminished. Thus the compilers not only succeeded in linking stems and derivations and treating both stems and derivations satisfactorily, but they also employed a complex system of cross-referencing: De klassieke opdeling macro-vs. micro-structuur wordt helemaal Opengebroken door een doorgedreven netwerk van verwijzingen. Inderdaad l zowel vanuit een slot tussen rechte haken, als vanuit een slot van verta-lingenJomschrijvingen, a1s vanuit een slot van commentaar, als vanuit een slot van voorbeelden, als vanuit het slot van een samenstelling met lemmastatus, als vanuit een slot van versteende uitdrukkingen met lemmastatus, als vanuit het slot van de staart, kan men verwijzingen vinden naar of elementen uit de macro-, of elementen uit de micro-structuur van een ander artikel! De meeste van deze verwijzingen werden ofwel reeds impliciet behandeld, of zijn zo evident dat ze geen verdere uitweiding vereisen. Het zij voldoende te vermelden dat hiervoor de volgende "SYMBOLEN & VERWIJZINGEN' worden gebruikt: -, ... , ~, 0, 0, afk X, ant X, cf X, syn XI var X, vgl X; en ook Romeinse en/ of Arabische cijfers. Op enkele details nal zijn AL deze verwijzingen ook kruisverwijzingen! Cross-references, whether explicit or implicit, text-internal or text-external, are given from all possible slots of an article. See the following seven typical reference positions: Compare -ronda in the article of -fimdisha in (21)  The endeavour to achieve an optimal transfer of information Dictionaries are containers of knowledge (d. McArthur 1986). Although lexicographers have to take this int~ ac~ount, they should als~ be alert to the fact that a dictionary has to be compiled ill such a way that the mtended target user can employ it as a practical linguistic instrument. One of the components of a dictionary aimed at a better retrieval of information by the target user is the access structure. The access structure can be regarded as the search route of the user on his way to the lexicographic data needed. The internal access structure, _ that is the search route followed within the article, can display a variety of socalled structural markers. These markers signpost various microstructural data categories. Because reference markers indicate the reference entry, it can be argued that they are also part of the access structure, functioning as nontypographical structural markers. A reference marker does not only indicate the fact that a specific text segment relates to another text segment but it sometimes also explicates the type of relation that holds between the two segments. In the W9 dUferent strategies, elucidated in the explanatory notes in the front matter, are used to accomplish successful cross-referencing. One kind of cross-reference used is the reference to a variant of the lemma. The reference position in this mediostructural category accommodates the marker var. of In the article of the ~nuna-sign inclose a cross-reference "var. oj ENCLOSE, ENCLOSURE" explicates the kind of mediostructural relation between the lemma-sign and the reference address. This lexicographic procedure does not only constitute a valuable type of cro~s-referencing but it also assists in presenting the lexicon as a structured Collection characterised by a network of internal relations. Within a multilingual and multicultural society dictionaries have an important role to playas instruments to promote mutual understanding and comrr;,~cative ~ompetence. South African leXicographers should employ all availa e strategIes to create a dictionary culture and to enhance the dictionary-

Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2011)
using skills of their intended target users. This approach compels lexicographers to structure their dictionaries in such a way that the retrieval of information exceeds the traditional domains. It will always be important to find comments on specific lexical items in a dictionary and it will always be important to find information linking a specific lemma or a microstructural entry like a sense discrimination, to other text segments in the dictionary. However, dictionaries compiled for the South African linguistic environment should go further than this. Besides information regarding a specific lemma or article component as treatment unit, it is of vital importance that dictionaries should expose the underlying system by focusing on a lexical item as part of an overall linguistic or grammatical pattern. ' According to Jackson (1985: 53) grammar and dictionary are complementary parts of the overall description of language. However, the average member of a speech community uses a dictionary much more than a grammar. One of the assignments of the lexicographer in a multilingual society is to make his/ her target user aware of aspects regarding both the lexicon and the grammar of the specific language. The first step to achieve this goal is to include a minigrammar as a separate text in the front or back matter of the dictionary (d. Gouws 1989). The fact that the average dictionary user focuses his attention exclusively on the data presented in the central word-list, compels the lexicographer to employ innovative strategies to ensure a successful utilisation of the grammar as one of the other texts in the dictionary. The most obvious strategy would be the establishment of text-external mediostructural relations between the central list and the mini-grammar. Dictionaries compiled for use in South Africa should be text carriers that include, among others, separate texts in which the grammatical system of the treated language is explained. Lexicographers have'to employ an extended mediostructural application to guide their users from a variety of reference positions in the central list to specific reference addresses in the mini-grammar.
The front matter texts should also include a systematic exposition of other language-specific characteristics and these texts have to be addressed from the central list by means of a well-developed mediostructural network.

In conclusion
There is nothing as practical as a good theory. Therefore the success of a dictionary as a practical instrument depends on its theoretical basis. If a lexicographer cannot base his practical applications on sound theoretical principles, the dictionary is bound to be of a lesser quality. Knowledge of the structural components of a dictionary as a carrier', of texts equips the lexicographer with the expertise to produce a better dictionary. Understanding the importance of cross-referencing as a functional lexicographic device enables the lexicographer to compile a dictionary which offers the target user friendly access to participa-tion in the language game. This is desperately needed in the multilingual and multicultural South Africa.