Tokens, the oft-overlooked appetizer: Large language models, the distributional hypothesis, and meaning