There are like three different threads in here, which is totally my fault.
First off, sure, the piano and your vocal folds are tools. I might even be willing to say that language is a kind of technology we use to communicate (among other things). Another way to say that is, "We use words to communicate," where "words" basically means language. That ambiguity is unfortunate.
I wasn't looking for use in that broad sense, but in the sense that the meaning of a word is its use in sentences, the semantic contribution it makes to sentences it appears in. (When you learn how to use a word, you've learned what it means.) I was wondering if instead of there being a generic skill--using the word ____--maybe using the word "red" is a skill, using the word "crepuscular" is a skill, and so on, just as drawing a straight line is a skill, drawing a hand is a skill, etc. (I recall now that somewhere Dummett says the issue here isn't so much individual words, but word types: how to use color-words, number-words, mass nouns, proper names, etc.)
Looking back, I don't think I ever explicitly said I meant there to be an analogy between word/sentence and note/tune. (As the tune is made of certain notes in a certain special arrangement, the sentence is made of certain words in a certain special arrangement.) But that's why I end up unwilling to say that words are tools, even if I might be willing to say that language is a tool, because I was thinking of the use of a word in a sentence rather than the use of "words," i.e. language, to say something.
But now I need to say that these two uses of "use" are the same, or at least really closely related. A word is
used in a sentence precisely in the sense that a sentence is what we use to perform what we can vaguely call a "linguistic act," the sense in which we use language to do something. That still doesn't exactly make a word a tool. I'm not even sure I would say a sentence is a tool. Language is more like a shared technology that includes the producing and consuming sides of the transaction.
But I do have one more observation. I lean toward molecularism, which is why I was claiming that we don't learn an entire language in one go, and wanted to look at how we add words. But there may be a sense in which that's false. No one actually uses, or ever learns, more than a proper subset of a given natural language. This need not be an idiolect--I mean only that fewer than all the words in the language are used, but those are used in the standard way. We're surely not going to say that you don't speak English unless you speak all of it. Speaking a proper subset is the norm. There may be a
sorites here--how much of the language do you have to speak? Or we could just allow that there's a continuum. (I speak "ein bißchen Deutsch.") If we do that, then from the moment you learn how to use an English word, you're in the same position as the English speaker you learned the word from; the only difference is the cardinality of your subsets.
That allows us to connect language as a shared technology to the use of a word right from the start.