Methods of Lexicographic Definition in the Oxford Advanced Learners Dictionary

Table of Contents

1. Introduction

2. Aim and Scope

3. Material
3.1 Corpuses
3.1.1 Word Selection
3.1.2 Part of Speech
3.1.3 Word Frequency
3.1.4 Concreteness
3.2 Dictionary

4 Method
4.1 Investigation
4.2 Defining 'Lexicographic Definition'
4.2.1 The Term 'Definition'
4.2.2 Lexicographic Definitions in Dictionary Entries

5 Classification of Lexicographic Definitions
5.1 Synonymous Definitions
5.2 Analytic Definitions
5.3 Synthetic Definitions
5.4 Implicative Definitions
5.5 Denotative Definitions
5.6 Ostensive Definitions
5.7 Regular Definitions

6. Analysis
6.1 Problems of Classification
6.1.1 Distinctiveness
6.1.2 Analytic and Synthetic Definitions
6.1.3. Synonymous Definitions
6.1.4 Remaining Difficulties
6.2 Results
6.2.1 Parts of Speech
6.2.2 Concreteness
6.2.3 Word Type Groups
6.2.3 Word Type Groups

7. Conclusion

References

Appendix: Complete Tables

1. Introduction

In this paper, I will classify and evaluate 80 lexicographic definitions from the Oxford Advanced Learner's Dictionary (henceforth ALD). There have been several attempts to classify different kinds of lexical definitions, as Weber (1996:19-32) shows. Yet it seems that there is no empirical study on the distribution of different methods of lexicographic definition. There seem to be certain traditional rules and assumptions on which words are to be defined in what way in lexicographers' manuals such as Landau (1989) or Benson et. al. (1986), for instance: "[s]imilar words should be defined in similar ways, and related words in related ways" (Benson et. al. 1986: 221). However, the manuals do not examine whether real lexicographers follow these rules.

Furthermore, this paper could be a contribution to a typology of lexical paraphrases or definitions.

Eine Untersuchung darüber, welche Lemmata mit welchen Methoden expliziert werden, wäre ein Beitrag zu der von Wiegand 1989:532 geforderten Explikationstypologie. (Weber 1996: 25)

Another starting point for this study was Jorgensen's (1990) work on "the efficacy of dictionary definitions in marking distinctions in sense" (Jorgensen 1990: 293). This paper discussed whether dictionary users could make sense of definitions by testing their encoding capabilities after reading definitions of nonsense words. It concluded that some of the underlying linguistic theories of dictionaries might not be very helpful or even correct. The idea that there is a linguistic theory inherent in each dictionary, however, is widely acknowledged.

All dictionaries necessarily adopt and transmit some points of view on language, even if the lexicographers are not aware of any. (Béjoint 1994: 173)

In this paper, I will investigate one aspect of the ALD as a theory of the lexicon by studying the interrelatedness of definitions and their headwords.

Günstigstenfalls (...) liegt qualitativ hochstehenden Wörterbüchern eine Beschreibungssemantik zugrunde, die wir rekonstruieren müssen und die, metatheoretisch beschrieben, so etwas wie eine Theorie der Bedeutungsexplikationen in diesen Wörterbüchern darstellen könnte. (Weber 1996: 9)

2. Aim and Scope

When I differentiate between different kinds of words, I will consider three aspects: a grammatical factor, i.e. part of speech, a pragmatic factor, namely word frequency, and concreteness as a cognitive factor. According to these properties I will compile eight groups of ten words each, each group representing a different word type. These word type groups will be the basis for my investigation of their respective definitions, questioning whether the lexicographers have followed Béjoint's (1994) point of view that

it is reasonable to argue that the techniques used for the explanation of meaning should vary with the type of word to be explained. (Béjoint 1994: 199)

For the purposes of this study, I will use a very narrow definition of the terms 'word meaning' and of 'lexicographic definition'. Except the pragmatic and cognitive factors my word type groups are based on, I will discuss denotation exclusively. Other aspects of semantics such as connotation, register, regional or social variation, polysemy, homonymy, collocations, idioms, etymology and aspects of syntax and grammar will be as much excluded as is possible.

Furthermore, this study will investigate one dictionary as a representation of one lexicological theory. It will not compare dictionaries, but rather create a basis for such a comparison in another paper.

3. Material

3.1 Corpuses

3.1.1 Word Selection

For my selection of words, I will use the combination of corpuses in the MRC Psycholinguistic Database on the Internet, which is an excellent tool for these purposes. The machine usable dictionary file of the MRC Psycholinguistic Database contains 150'837 words and provides information about 26 different linguistic properties, although it is not the case that information about every property is available for every word. Of this rich selection, I will choose three different properties to be able to specify different types of words and investigate whether their definitions vary according to these types.

3.1.2 Part of Speech

The obvious property to choose is part of speech. Apart from a word's denotation, it is certainly the most essential information a speaker must know about a word to be able to use it in a well-formed way. As research on malapropisms has shown, part of speech is not only fundamental in grammar books, but also in the cognitive rules for linguistic acceptability.

The finding that word selection errors preserve their part of speech suggests that the latter is an integral part of the word, and tightly attached to it. (Aitchison 1994: 100)

Lexicographers acknowledge the importance of part of speech by following the principle substitutability; i.e. each word should be defined by one or several words of the same part of speech, obviously modified by other words (Benson et. al. 1986: 205). For this paper, I will divide the different parts of speech into four classes: nouns, verbs, adjectives, and function words. The latter class is admittedly slightly fuzzy, for it includes determiners, pronouns, prepositions, conjunctions and adverbs, yet all of these parts of speech share the aspect of functionality, i.e. their use is more important than their referential meaning, which sometimes does not even exist or changes according to context (deixis). My results will confirm the idea that these words can be combined to one class.

3.1.3 Word Frequency

The second property I will use is word frequency, based on the Kucera/Francis (1967) word count. The number given in my charts refers to occurrences of the word per one million words of running written text. Jorgensen (1990) has shown that the frequency of a word has its influence on how difficult it is to write and understand a definition.

[D]efinitions for low-frequency words were better than definitions for high-frequency words, when they were based on an equivalent number of citations, perhaps because words of higher frequency tend to occur more often in depleted or noninformative uses. (Jorgensen 1990: 313)

Rather than create word type groups with words of different frequencies, I will eliminate the factor by selecting words which have more or less the same frequencies: between 20 and 30 per million for nouns and abstract verbs, between 10 and 60 per million for concrete verbs and adjectives.^[1] Words of low frequency have another advantage: they tend to be less polysemous than more frequent words (which, of course, is one reason for them being less frequent).

However, there is a problem with the frequency of function words. Since their frequency is generally much higher than those of nouns, verbs or adjectives, it is not possible to compare function words with other parts of speech in that aspect. Moreover, there are almost no function words with low frequencies - if 'low frequency' is not redefined for function words. Therefore, the factor of frequency will be neglected for function words, keeping in mind that this may influence my results.

3.1.4 Concreteness

Having eliminated, where possible, the factor of frequency, my investigation will be able to focus on a third lexical property, namely concreteness. The decision to include concreteness as an essential property of word meaning is based on Jorgensen's (1990) results as well.

[G]ood definitions seem to be somewhat easier to write for concrete words, at least for subjects who do not possess the lexicographer's reference tools (such as a thesaurus). (Jorgensen 1990: 314)

Therefore, concreteness is an important factor for the process of understanding word meaning, at least when a dictionary is used for that purpose. It also seems to be fundamental for the lexicographer's choice of definition methods.

Concrete things can be communicated far more effectively by analogy and comparison than can abstract relations. (Ayto 1984: 61)

The concreteness ratings of the MRC Psycholinguistic Database are derived from a merging of the norms of Pavio et al. (1968), Gilhooly/Logie (1980) and norms developed at the University of Colorado.^[2] The concreteness values are integer and range between 100 and 700. The actual mean of the words in the MRC corpus is 438. The MRC gives concreteness ratings for all parts of speech, but not all words of the corpus are rated.

3.2 Dictionary

The definitions of the chosen words to be classified in this paper have been taken from the Oxford Advanced Learner's Dictionary. There are several reasons for my choosing a learner's dictionary rather than a common purpose dictionary. First of all, the profile of the users of such a dictionary is predictable to a certain extent. They tend to be non-native speakers who have passed the beginners' status as students of English. They use dictionary definitions as explanations of words they do not know, in order to be able to specify its denotation (and its use, but that is not my subject). The learner's dictionary, like my definition of 'definition', is user-oriented. Secondly, definitions in learner's dictionaries tend to be short and to the point, which will make it easier for me to compare and classify them.

I will discuss the ALD rather than any other learner's dictionary, because, coming from the house of Oxford, it is still a rather traditional dictionary. While studies on the Collins Dictionary of Contemporary English or COBUILD would tell us more about the idiosyncratic methods of definition used exclusively in these volumes, the results of an investigation of the ALD as a traditional dictionary might shed some light on traditional methods of definition in general, or at least provide a basis for further research in that direction.

4 Method

4.1 Investigation

For my investigation, I will first assemble eight different word type groups, concrete and abstract nouns, adjectives, verbs and function words respectively. Concreteness values are chosen above and below average (438), the range of concreteness being 100 to 700. All 80 words will be classified according to the terms stated in chapter five. In the same chapter, the practical and linguistic value of each method of definition will be discussed. Finally, the distribution of different methods of definition according to parts of speech and concreteness of the sample words will be investigated. In conclusion, certain trends will be revealed, and the value of the ALD as a tool for dictionary users will be discussed.

4.2 Defining 'Lexicographic Definition'

4.2.1 The Term 'Definition'

Much has been written about the problems of the term 'definition' when used to refer to dictionary entries. The most important problem when using the word is the confusion of its meaning in philosophy and logic with its meaning in lexicography. The former is what Robinson (1950: 149ff) calls 'real definition', an explanation of a thing or concept, not an explanation of the meaning of a word. Wiegand (1985:60) is one of many to propose new terms for 'lexicographic definition', in his case 'lexikographische Bedeutungs­erläuterung'. While the emphasis on explanation in this term is important, it will suffice for my purposes to state that the terms 'definition' and 'lexicographic definition' in this paper will be used as synonyms, referring not to philosophical equation of terms, but to an explanation of a word by other words in a dictionary.

Lexical definition is our name for the enterprise of teaching some man the meaning actually borne by some word in some society. (Robinson 1950: 44)

This view of definitions is communicative in its nature, which is essential for my approach of the matter. The emphasis for lexicographic definitions, therefore, is on a successful explanation of the meaning of a word, and not on creating a logical and coherent system of words and their definitions.

Whereas philosophers are concerned with the internal coherence of their system of definition, lexicographers are concerned with explaining something their readers will understand. The methods each uses to achieve his goals only incidentally coincide. (Landau 1989: 121)

We are bound to make mistakes if we fail to distinguish the ideals and possibilities of logical systems from the enterprise of telling a person what some other persons mean by a word. (Robinson 1950: 44)

Moreover, I will be considering only one aspect of the meaning of a word being taught to a person, i.e. its referent or denotation.

We want to refer him [the reader of a definition] to the thing which is the meaning of this word, in order that he may henceforth connect the two and regard the word as a symbol of the thing. (Robinson 1950: 92)

Therefore, a lexicographic definition will be considered an act of establishing a relation between a word and its referent, rather than a specification of a word's meaning in the semiotic sense, i.e. its signified. As a small experiment by Wiegand (1985) has indicated, this view of lexicographic definition is actually assumed by the users of dictionaries.

Die Probanden neigen dazu, das Verhältnis von Lemmazeichen und sog. lexikographischer Definition eher im Sinne einer Feststellung über eine Relation von sprachlichen Ausdrücken zur nichtsprachlichen Welt aufzufassen. (Wiegand 1985: 47)

Wiegand concludes after this experiment that a strict separation of semantic and encyclopaedic information does not seem to be relevant for dictionary users (Wiegand 1985: 47). Furthermore, modern dictionaries have already abandoned this separation.

Both British and American dictionaries of practically any size [...] had also long given up the strict separation of linguistic and encyclopaedic knowledge and included all sorts of biographical, geographical, technical, and other information. (Lipka 1995: 382)

The relevant definition of 'lexicographic definition' for this paper is therefore: an explanation of a word, using other words, that enables a person to identify the word's referent or denotation.

4.2.2 Lexicographic Definitions in Dictionary Entries

Since dictionary entries consist not only of lexicographic definitions in the sense I defined above, it is necessary to specify which part of the entry text is taken to be the part to be classified.

As a rule, the text in standard small letters after the headword, its pronunciation in phonetic transcription, part of speech and different usage and grammar labels and the examples of different uses of the headword in italics will be taken to be the lexicographic definition. However, text in brackets, mostly referring to usage or syntax, will be ignored. Moreover, information after the label esp will be seen as not essential and therefore, for my purposes, not part of the definition. It is often the case that synonyms or further definitions are added to the first one after a semicolon. These could sometimes be seen as slightly different meanings of the headword, and in some cases, a second entry, indicated by a number or letter, would be justified. Therefore, the first definition will be the one being classified for my study.

If a headword is polysemous, i.e. if several definitions for one and the same word are given, indicated by numbers or small letters, only the first definition will be considered. It seems that the first definition usually specifies the most common use of the headword. However, the ALD commentary does not give any details on the dictionary's principles governing the hierarchy of polysemous word meanings.

[...]

^[1] The difference in range is caused by the fact that there are not as many concreteness ratings for words of the latter classes, and therefore a greater choice of words was necessary to be able to choose ten words of above- and below-average concreteness.

^[2] Unfortunately, I have not been able to retrieve either text yet. An online manual for the MRC with detailed information is not available at the moment. I contacted the programmer responsible for the MRC. He promised to send me the necessary information, but has not done so yet. So, I am afraid, essential information on how words were rated for their concreteness, especially non-nouns, is missing in this paper for the time being. However, it is clear from the output of the MRC that ratings for non-nouns make sense and are not just the ratings of morphologically related nouns, although this might be the case for some verbs and adjectives.