Polysemy and discourse.
If we observe the world, we will see that many if not most words are used in different ways. This can be easily observed by opening the dictionary and noticing that many words have multiple potential definitions.
One attitude about this can be seen in the great musical "My Fair Lady", which I happened to watch while I was eating dinner last night. In the opening scene Professor Henry Higgins says:
"Look at her a prisoner of the gutter
Condemned by every syllable she utters
By rights she should be taken out and hung
For the cold-blooded murder of the English tongue
Hear them down in Soho square
Dropping h's everywhere
Speaking English any way they like"
"Speaking any way they like," indeed. As if there were one precise correct definition for a given word. And only one correct way to speak.
This would not, I think, be a very productive view when engaging in academic discourse.
Obviously, if there is common agreement about a term, there is no real problem. But what if you want to use a term like "racism"? Wikipedia says "the term can ... have varying and hotly contested definitions." If you hold the Higgins view of language--that there is only one way to speak (or write)--then you are liable to get into some significant debate if you're using a term like "racism."
Instead one can look at a term for the way that different people use it, and compare that meaning with the way in which others have used it. You can place yourself within the discourse. You can recognize the different voices and align yourself with the meaning that is important to you. Of course this does not guarantee that no one will disagree with your usage, but at the least, even those who disagree will be forced to acknowledge that you are at least aware of the (empirically verifiable) debate on the usage of the term.
Because your writing is meant to express your own voice, there is no avoiding taking a position on issues. You must take positions unless your work is specifically designed to simply report observations (but even that is problematic, because you, the observer must have some idea of which data to observe--and that too requires taking a position concerning which data is important and which is not). By recognizing the different voices in the discourse that surrounds you, your work appears more sophisticated and better researched.