Knowledge of meaning and knowledge of facts
In Chapter 2 we mentioned the contrast between dictionary and thesaurus models of semantic organization (see the text box in 2.1.1). The sort of considerations we have just mentioned give rise to another contrast, that between dictionary and encyclopaedia aspects of meaning. This is the distinction between knowledge of a word’s meaning (dictionary knowledge), which would be conceived of as something fairly concise, perhaps like a dictionary definition, and encyclopaedic knowledge of facts about the objects to which the word refers. Dictionary knowledge is knowledge of the essential meaning of a word that all speakers must possess, and which dictionaries must accurately represent in order to allow the meaning to be acquired for the first time. Encyclopaedic knowledge, by contrast, is not essential to the meaning of the word and will vary significantly from speaker to speaker. Encyclopaedic knowledge is not linguistic in nature: that is, it does not determine any of a word’s linguistic behaviour. The question of which elements of the encyclopaedic information associated with a given word are relevant in any one situation is decided by general pragmatic principles, which have been described in a number of different ways (see Chapter 4).
The motivation for the distinction between dictionary and encyclopaedia is the fact that encyclopaedic knowledge seems to be quite independent of dictionary knowledge: thus, I need not know anything about fairy tales or the Australian water-holding frog in order to be able to use the word frog. Furthermore, it has been assumed that some such distinction must be psychologically realistic. If all of the encyclopaedic information associated with a word were part of its meaning, this would surely be too much for the brain to process. If, on the other hand, all that language processing involves is the retrieval of the concise dictionary-style representation associated with each word, then it appears as a much more streamlined and efficient process, much easier for the brain to accomplish – and much easier also for the computers on which we try to model the brain-processes involved in language (see Chapter 8).
The distinction between dictionary and encyclopaedia is not limited to referring expressions like frog. It also applies to predicating ones, like English verbs and adjectives. If we accept the distinction, it becomes important to be able to say exactly which pieces of information about a lexeme belong to the dictionary and which to the encyclopaedia. This is a particularly acute problem where it is necessary for practical reasons (for example lexicographical ones) to arrive at some precise description of a lexeme’s semantic content. In order to appreciate the descriptive issues involved here, we can consider the Warlpiri verb pinyi, usually glossed ‘hit’, which is often ambiguous between the meanings ‘hit’ and ‘kill’:
There are at least two possible ways of analysing this ambiguity. The first is that pinyi has two meanings, ‘hit’ and ‘kill’, which, in certain contexts, may be simultaneously present. The second is that there is one single, underspecified meaning, which we can only describe in English as ‘hit/ kill’. On this theory, it is the context which determines whether pinyi describes an act of hitting or of killing, just as context determined the reading of the English possessive morpheme in (1) above. This second solution would be favoured by many scholars. Whenever we are faced, says Levinson (2000: 20), ‘with a linguistic expression that is apparently systematically ambiguous, we should entertain the possibility that the correct analysis is in fact a simple, univocal, semantically broad sense with a defeasible set of generalized pragmatic restrictions’. (Defeasible means that the restrictions can be overcome by adding elements to the sentence which enforce one reading at the expense of the other. In (13), we could add an expression meaning ‘dead’ to the sentence which would eliminate the ambiguity.)
What would the details of this underspecified solution be? How does con text contribute to the contextual interpretation of pinyi? The English translations usually offered of sentences like (14) and (15) suggest that typically domestic animals like dogs usually provoke the ‘hit’ interpretation of pinyi, whereas typically wild, edible ones like kangaroos are associated with ‘kill’:
In contexts like (15), the ‘kill’ reading is quite strongly entrenched: as noted by the exclamation mark, the following statements appear bizarre to Warlpiri speakers:
On an underspecified view of the meaning of pinyi, it is the encyclopaedic knowledge which Warlpiri speakers have about their world which allows them to correctly understand what is meant in the sentences above. In contexts such as (13), where the object is equally likely to be hit or killed, there is no way of telling which interpretation is appropriate without further specification: the underspecification of the verb leaves no way of deciding. Further support for the underspecified solution comes from sentence (18):
Cats are neither traditional domestic nor wild animals for Warlpiri people; as a result (18) constitutes a ‘neutral’ context without established encyclopaedic expectations, where the verb may convey either sense. But once the verb is inserted into a grammatical context which specifies the object, encyclopaedic facts come into play to determine the intended reading.
Distinguishing between lexical dictionary knowledge and factual encyclopaedic knowledge thus enables an economical description of meaning. The lexical entry for pinyi does not need to detail the different contexts in which the two readings occur: this is not part of our knowledge of the meaning of the verb. Instead, the lexical entry simply contains the meaning ‘hit/kill’, and the choice of reading in any one context is reached on encyclopaedic grounds. The details of Warlpiri speakers’ representation of their encyclopaedic knowledge, and its interaction with linguistic structures, can then be legitimately neglected by linguists, as part of the explanatory job of psychology.
