Problems with the dictionary encyclopaedia distinction
In spite of its obvious labour-saving advantages in semantic description, the dictionary–encyclopaedia distinction is not accepted by many linguists. This is largely because the boundary between the two seems to be highly permeable, even non-existent. As any comparison of dictionaries will reveal, it is very hard to determine where information stops being part of a word’s dictionary meaning and becomes part of the encyclopaedic knowledge we have of its denotation. Which of the following pieces of information, for example, should be considered dictionary information about the meaning of the word cow, and which as facts about cows which form part of the encyclopaedic knowledge we have about them?
• they are mammals
• they moo
• they eat grass
• they are four-legged
• they have large eyes
• they live in fields and dairies
• they sometimes wear cow-bells
• they are often farmed for their milk
• they have several stomachs
• their young are called calves in English
• they incubate Mad Cow Disease for three to five years if infected
• they chew their food slowly
The difficulty of resolving this kind of question stems from the fact that, depending on the context, it would seem that any part of the supposedly encyclopaedic information associated with a word may become linguistically significant (see Katz and Fodor 1963: 178–179 for discussion). To return to the example of frog, note that I can use this in contexts where the dictionary meaning is not even present, and two pieces of encyclopaedic information are invoked. In the context of a discussion about a French person who is unlucky in love, for example, I might utter (19):
Someone would be able to understand this sentence without ever learning the dictionary meaning of the word frog, as long as they could make the connections with ‘French person’ and ‘enchanted prince’.
It might be argued that this sort of context simply shows that ‘Frenchman’ and ‘sleeping prince’ are, in fact, part of the dictionary entry for frog after all, and not just encyclopaedic facts about frogs. But this acknowledgment undermines the very reasons for drawing the dictionary encyclopaedia distinction in the first place. If we simply reassign a piece of previously encyclopaedic knowledge to the dictionary every time it becomes relevant to the linguistic behaviour of a word, the dictionary starts getting a lot bigger, and looking more and more like an encyclopaedia. The supposed processing benefits of concision in lexical representation thus disappear. Furthermore, we can assign an innovative piece of encyclopaedic knowledge to a word, which can then usurp the word’s former dictionary meaning. For example, some people do not know that tomatoes are, strictly speaking, fruit and not vegetables. This allows someone who has just been made aware of this to pedantically use sentences like (20):
(20) Get me some tomatoes and other fruit.
Here, a newly acquired piece of encyclopaedic knowledge has affected the co-occurrence possibilities of a lexical item: whereas tomato would typically be categorized as a vegetable, here it belongs to the incompatible category fruit. This sort of phenomenon suggests that there is no possible boundary between knowledge of the meaning of a word, and knowledge about the objects the word denotes. We know a variety of things about words and their denotation, and the greater the likelihood that a particular piece of this knowledge is shared between speaker and hearer, the greater the likelihood that it will determine the word’s linguistic proper ties. Sentence (20), for example, would be perfectly natural in the mouth of a botany student who was about to do an experiment on seeded fruit.
Langacker (1987: 159) sums up the case for the abandonment of any strict division between the dictionary and the encyclopaedia:
I do not specifically claim that all facets of our knowledge of an entity have equal status, linguistically or otherwise – quite the contrary. The multitude of specifications that figure in our encyclopedic conception of an entity clearly form a gradation in terms of their centrality. Some are so central that they can hardly be omitted from even the sketchiest characterization, whereas others are so peripheral that they hold little significance even for the most exhaustive description. Distinctions of this kind can perfectly well be made within the encyclopaedic approach. The thrust of this view is simply that no specific point along the gradation of centrality can be chosen nonarbitrarily to serve as a demarcation, such that all specifications on one side can uniformly be attributed linguistic significance and all those on the other side are linguistically irrelevant.
This sort of position has some significant methodological and theoretical consequences. Most importantly, it problematizes the notion of the meaning of a word. Since any fact known about a referent may become linguistically significant, the traditional linguistic semantic project of describing the lexical entry associated with each lexeme becomes an unending task, each lexical entry being, in principle, infinite.
QUESTION Consider the Arrernte (Pama-Nyungan, Central Australia) verb lyelye-ipeme, whose meaning is described as follows by Henderson and Dobson (1994: lyelye-ipeme): ‘push a stick or crowbar into creek sand, moving it around to make the hole bigger so as to force the stick further down. This is done to see if there is enough water there to dig out into a soakage.’ How might one go about deciding which parts of this definition were dictionary knowledge, and which were encyclopaedic? Are there any general criteria for deciding this question?
