Semantic Structures
Ray Jackendoff, 1990

I-Concepts Not Encoded as a List?
I disagree with Jackendoff's premises here. Just becuase I can (in theory) produce an infinite number of sentences does not mean that I have, at any given time, an internal representation of any significant number of them. I create them as needed. The situation is different for concepts. I agree that I create concepts as needed and that an infinite number of them is potentially conceivable. But this does not imply that I do not store my concepts in lists (or some equivalent mental structure). Unlike sentences, I do need to keep a good supply of concepts around.

This non-parallel structure is particularly evident for lexical concepts. A lexical concept is the concept expressed by a word. Jackendoff is not quite clear (at this point) whether this is the same as the semantic constituent of a lexical entry (known as an LSC) or some computed function of that LCS. In any case, I do not arbitrarily create such lexical concepts other than at the time when I learn a new word or a new sense of a word I already know. I cannot see any sense in which such concepts are infinite in number, even potentially.

The argument given here is that once I know what a dog is, then I can decide for each of an indefinitely large group of objects whether each is a dog. This argument confuses the categorization rules within the concept of dog with other rules used to create concepts, such as [TYPE DOG].

Jackendoff uses these infinity arguments as evidence that rules are used to create new concepts. I do not question that such rules exist. If I create one or more new concepts each time I understand a sentence, then clearly, I must be using rules to create those concepts. Similarly, for lexical concepts when I learn a new word. Thus, I agree that an infinite number of potential concepts must be constructable.

The summary statement is that "lexical concepts cannot [consist] of a list of instances, but must consist of finite schemas that can be creatively compared ... to novel inputs". I would argue that they consist of both, (non-exhaustive) lists of instances and creative rules.

Still, Jackendoff overworks the point. I cannot, during my lifetime, acquire an indefinite number of concepts. Even if I forget 99.9% of all I ever learned and you count those which I have forgotten, the number is still infinitely far from "an indefinite number".


Top of Page | SS Summary | Sort by Topic | Sort by Title | Sort by Author