Key Arguments

• He begins from the assumption that understanding sentences consists in 'retrieving stored information about the meaning of isolated words and phrases, and then combining and perhaps altering these retrieved word meanings to build up the meanings of sentences.'
• On this basis, he infers that 'one should be able to take a model of stored semantic knowledge, and formulate rules of combination that would describe how sentence meanings get built up from stored word meanings.'
• He further conjectures that if 'one could manage to get even a few word meanings adequately encoded and stored in a computer memory, and a workable set of combination rules formalized as a computer program,' then the system could 'bootstrap his store of encoded word meanings' by using itself to '"understand" sentences that he had written to constitute the definitions of other single words.'
• The bootstrapping mechanism is: whenever a new word is definable by 'a sentence using only words whose meanings had already been encoded,' then 'the representation of this sentence's meaning, which the machine could build by using its previous knowledge together with its combination rules, would be the appropriate representation to add to its memory as the meaning of the new word.'

Source Quotes

It seems reasonable to suppose that people must necessarily understand new sentences by retrieving stored information about the meaning of isolated words and phrases, and then combining and perhaps altering these retrieved word meanings to build up the meanings of sentences. Accordingly, one should be able to take a model of stored semantic knowledge, and formulate rules of combination that would describe how sentence meanings get built up from stored word meanings.44 Quillian also has great hopes for his system: It further seems likely that if one could manage to get even a few word meanings adequately encoded and stored in a computer memory, and a workable set of combination rules formalized as a computer program, he could then bootstrap his store of encoded word meanings by having the computer itself "understand" sentences that he had written to constitute the definitions of other single words.

It seems reasonable to suppose that people must necessarily understand new sentences by retrieving stored information about the meaning of isolated words and phrases, and then combining and perhaps altering these retrieved word meanings to build up the meanings of sentences. Accordingly, one should be able to take a model of stored semantic knowledge, and formulate rules of combination that would describe how sentence meanings get built up from stored word meanings.44 Quillian also has great hopes for his system: It further seems likely that if one could manage to get even a few word meanings adequately encoded and stored in a computer memory, and a workable set of combination rules formalized as a computer program, he could then bootstrap his store of encoded word meanings by having the computer itself "understand" sentences that he had written to constitute the definitions of other single words. That is, whenever a new, as yet uncoded, word could be defined by a sentence using only words whose meanings had already been encoded, then the representation of this sentence's meaning, which the machine could build by using its previous knowledge together with its combination rules, would be the appropriate representation to add to its memory as the meaning of the new word.45 But with a frankness, rare in the literature, Quillian also reports his disappointments: Unfortunately, two years of work on this problem led to the conclusion that the task is much too difficult to execute at our present stage of knowledge.

Accordingly, one should be able to take a model of stored semantic knowledge, and formulate rules of combination that would describe how sentence meanings get built up from stored word meanings.44 Quillian also has great hopes for his system: It further seems likely that if one could manage to get even a few word meanings adequately encoded and stored in a computer memory, and a workable set of combination rules formalized as a computer program, he could then bootstrap his store of encoded word meanings by having the computer itself "understand" sentences that he had written to constitute the definitions of other single words. That is, whenever a new, as yet uncoded, word could be defined by a sentence using only words whose meanings had already been encoded, then the representation of this sentence's meaning, which the machine could build by using its previous knowledge together with its combination rules, would be the appropriate representation to add to its memory as the meaning of the new word.45 But with a frankness, rare in the literature, Quillian also reports his disappointments: Unfortunately, two years of work on this problem led to the conclusion that the task is much too difficult to execute at our present stage of knowledge. The processing that goes on in a person's head when he ''understands" a sentence and incorporates its meaning into his memory is very large indeed, practically all of it being done without his conscious knowledge.

Key Concepts

• It seems reasonable to suppose that people must necessarily understand new sentences by retrieving stored information about the meaning of isolated words and phrases, and then combining and perhaps altering these retrieved word meanings to build up the meanings of sentences.
• Accordingly, one should be able to take a model of stored semantic knowledge, and formulate rules of combination that would describe how sentence meanings get built up from stored word meanings.
• It further seems likely that if one could manage to get even a few word meanings adequately encoded and stored in a computer memory, and a workable set of combination rules formalized as a computer program, he could then bootstrap his store of encoded word meanings by having the computer itself "understand" sentences that he had written to constitute the definitions of other single words.
• whenever a new, as yet uncoded, word could be defined by a sentence using only words whose meanings had already been encoded, then the representation of this sentence's meaning, which the machine could build by using its previous knowledge together with its combination rules, would be the appropriate representation to add to its memory as the meaning of the new word.

Context

Dreyfus is summarizing Quillian’s aims and motivating assumptions in designing a semantic‑memory program for natural‑language understanding, highlighting Quillian’s belief in compositional construction of sentence meanings and the possibility of bootstrapping a large lexicon from a small set of initially encoded words.