1

We define a word in a language using a set of other, agreed-upon words. In linear algebra, the set of basis vectors for a space are the minimum number of vectors needed to describe any other vector in the space. Consider each word in a language a "vector", which can be described using some "combination" of other words/"vectors". Is there a minimum set of words in a language that can be used to define all the words in a language?

Joseph Farah
  • 111
  • 2
  • What do you mean by "define"? I don't think you can define the word "dog", but you can say things about dogs that might clarify what you are referring to. Likewise "enset" (a plant species). – user6726 Feb 10 '21 at 00:22
  • 2
    Unfortunately, the analogy to vector spaces breaks down rather quickly. For example, you can multiply any vector in a vector space with a real number (or, more generally, a field), or you can add any vectors, and get another vector. But what is 3.5 times "unique" plus "dog" minus "artichoke"? – jick Feb 10 '21 at 01:19
  • The usual mathematical analogy is with prime numbers, leading to the theory of semantic primes. – curiousdannii Feb 10 '21 at 14:07
  • @jick I don't think that normal mathematical operations would carry over for this analogy, but "adding" words together could be granting attributes (i.e. "unique" plus "dog" minus "artichoke" could correspond to a new "vector" (a new word in this case) that means "a rare dog not holding an artichoke". I'm straining the analogy at best, but I don't think the analogy necessarily breaks. Since every word in the English language has a definition made from other words, the language technically does have a basis, which in this case is the entire language. It may not be the smallest basis though. – Joseph Farah Feb 10 '21 at 16:21
  • @curiousdannii actually, that does answer my question pretty well! I suppose those 1000 primitives would be a good starting point to approximate a basis. I'll flag to mark my question as a duplicate since I think that other question covers it pretty well. Thanks for the discussion folks – Joseph Farah Feb 10 '21 at 16:23
  • Take a look at Wierzbicka and Goddard's Natural Semantic Metalanguage. Note that this theory does not presume that all human languages use the same concepts. – jlawler Feb 12 '21 at 21:19

0 Answers0