Things are what we want to know, not words, and so we should begin from things themselves, not from words. What we want to know are the natures of things, their properties, their causes, and so on. We can all feel a certain unease at just beginning with how we speak about something and not beginning with the thing itself.
The modern suspicion of “mere words” often comes from having seen language used to hide things, whether deliberately or not, by words that obscure some important reality. Slogans, jargon, technical terms, and official language that does not attend to what it is about can lead us to a certain way of being slipshod or careless in our words and that can make us think that a concern for words is somehow inappropriate, somehow “just semantics.”
It is also true that we can begin with trying to be precise about our words and then become lost in definitions and distinctions and never return to what we were trying to know. There is always a temptation to consider words rather than to study things themselves; this is especially true of modern historically-minded disciplines, where the concern can be primarily with the words that others in the past have used for the things in question.
We may also hesitate to begin with words because we think that words have no stable meaning, or perhaps we think that what we can express in words is merely subjective; we may even think that languages can be so different that some peoples can never understand each other. But if this were true of words, of expressions, or of whole languages, it would be equally true of any other symbolic system we might hope to use. Likewise, if the speakers of one language could not understand the speakers of another, it would be impossible for those peoples to learn from each other or to build anything together, since some languages would be completely or partially uninterpretable to others. But it is false that speakers of one language cannot learn from those of another, and it is a fact of experience that there are no uninterpretable languages.
It is true that the thing in question is what we are trying to know, and it is also true that considering speech as such will not by itself be sufficient for coming to know the thing in question. But the question here is this: Can we come to know things in the best way without attending to the meanings of the words by which we speak about them?